seismic assessment and retrofit of existing …
TRANSCRIPT
POLITECNICO DI MILANO Dipartimento di Ingegneria Strutturale Dottorato di Ricerca in Ingegneria Sismica, Geotecnica e dell’Interazione Ambiente-Struttura
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
Georgios Tsionis
Supervisor Prof. Giandomenico Toniolo
June 2004
POLITECNICO DI MILANO Dipartimento di Ingegneria Strutturale Dottorato di Ricerca in Ingegneria Sismica, Geotecnica e dell’Interazione Ambiente-Struttura
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
PhD candidate: Georgios Tsionis
Tutor / Supervisor: Prof. Giandomenico Toniolo
PhD course coordinator: Prof. Alberto Castellani
i
TABLE OF CONTENTS
List of tables.......................................................................................................................vii
List of figures ......................................................................................................................ix
Aknowledgements............................................................................................................xvii
1. INTRODUCTION .......................................................................................................1
1.1. General – Motivation for the research .................................................................1
1.2. Objectives of the research....................................................................................3
1.3. Outline of the thesis .............................................................................................4
2. OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES - A
LITERATURE REVIEW ....................................................................................................7
2.1. Asynchronous excitation......................................................................................7
2.2. Displacement-based design and assessment ........................................................9
2.3. Soil-structure interaction....................................................................................11
2.4. Bridges with isolation and dissipation devices ..................................................12
2.5. Damage assessment ...........................................................................................15
2.6. Seismic retrofit of reinforced concrete columns ................................................16
2.6.1. Retrofit by jacketing ..................................................................................17
Concrete jackets .....................................................................................................17
Steel jackets ...........................................................................................................18
FRP jackets ............................................................................................................19
2.6.2. Retrofit for enhancement of lapped splices ...............................................21
2.7. Analysis and modelling......................................................................................23
3. EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH
SEISMIC DEFICIENCIES................................................................................................27
3.1. Introduction........................................................................................................27
3.2. Bibliographic research .......................................................................................29
3.2.1. Experimental assessment of piers with seismic design..............................29
3.2.2. Experimental assessment of piers without seismic design ........................32
3.3. Design of the test models ...................................................................................33
3.3.1. Scaling of the specimens............................................................................33
3.3.2. Geometry of the specimens........................................................................34
3.3.3. Test set-up and instrumentation .................................................................38
3.4. Cyclic test on a model of a short bridge pier .....................................................40
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
ii
3.4.1. Experimental results...................................................................................40
Force-displacement diagram and observed damage ..............................................40
On the definition of yield displacement.................................................................42
Flexural and shear deformation .............................................................................46
Equivalent plastic hinge length..............................................................................47
Distribution of curvature........................................................................................50
Damage assessment ...............................................................................................51
3.4.2. Comparison to empirical predictions .........................................................55
3.4.3. Comparison to a squat pier designed according to EC8 ............................58
General ...................................................................................................................58
Equivalent damping ...............................................................................................59
Damage assessment ...............................................................................................62
3.5. Cyclic test on a model of a tall bridge pier ........................................................63
3.5.1. Experimental results...................................................................................63
Force-displacement diagram and observed damage ..............................................63
Flexural and shear deformation .............................................................................67
Distribution of curvature........................................................................................67
Equivalent plastic hinge length..............................................................................69
Damage assessment ...............................................................................................71
3.5.2. Comparison to empirical predictions .........................................................71
3.6. Performance of hollow cross-section bridge piers.............................................72
3.6.1. Effect of mechanical and geometrical parameters .....................................72
3.6.2. Estimation of deformation limits ...............................................................75
3.7. Concluding remarks ...........................................................................................77
4. SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE ....................83
4.1. Introduction........................................................................................................83
4.2. The pseudodynamic testing method...................................................................85
4.2.1. The pseudodynamic testing method...........................................................85
4.2.2. The a-Operator Splitting scheme...............................................................86
4.2.3. The substructuring technique.....................................................................88
4.2.4. Substructuring in the case of asynchronous motion ..................................88
4.2.5. The continuous pseudodynamic testing with non-linear substructuring....89
4.2.6. Implementation for the Talübergang Warth Bridge tests ..........................92
4.3. Pre-test numerical simulation ............................................................................94
iii
4.3.1. Numerical models for the substructured piers ...........................................94
4.3.2. Numerical model of the bridge structure ...................................................96
Description of the model........................................................................................96
Damping matrix .....................................................................................................98
Modal analysis .......................................................................................................99
4.3.3. Input motion.............................................................................................101
4.3.4. Numerical simulation of the pseudodynamic tests ..................................101
4.4. Pseudodynamic testing of the bridge model ....................................................105
4.4.1. Testing programme ..................................................................................105
4.4.2. Low-level earthquake test ........................................................................106
4.4.3. Nominal earthquake test ..........................................................................108
4.4.4. High-level earthquake test .......................................................................111
4.4.5. Final collapse test.....................................................................................115
4.5. Seismic assessment of the bridge.....................................................................117
4.5.1. Deformation and curvature distribution in the physical piers..................117
4.5.2. Damage assessment .................................................................................119
4.5.3. Overall damage index ..............................................................................122
4.5.4. Vulnerability functions ............................................................................124
4.5.5. Effect of cycling.......................................................................................126
4.5.6. Irregularity issues.....................................................................................127
4.6. Application of simplified assessment methods................................................131
4.6.1. General .....................................................................................................131
4.6.2. HAZUS method .......................................................................................132
4.6.3. The substitute structure methods .............................................................135
4.6.4. Application to the Talübergang Warth Bridge tests ................................138
HAZUS ................................................................................................................138
Capacity Spectrum method ..................................................................................140
N2 method............................................................................................................145
4.7. Concluding remarks .........................................................................................150
5. NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE
PIERS...............................................................................................................................153
5.1. Introduction......................................................................................................153
5.2. Fibre/Timoshenko Beam modelling ................................................................154
5.2.1. The Fibre/Timoshenko Beam element in Cast3m ...................................154
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
iv
5.2.2. Constitutive laws......................................................................................155
Concrete ...............................................................................................................155
Steel......................................................................................................................158
5.2.3. Alternative configurations for the cross-section and the beam element ..160
5.2.4. Validation of the numerical models.........................................................164
Piers with flexure-dominated behaviour ..............................................................164
Piers with combined flexural and shear behaviour ..............................................167
Piers with tension shift.........................................................................................170
Numerical models for the Talübergang Warth Bridge PSD tests ........................174
5.3. FEM modelling ................................................................................................178
5.3.1. Constitutive laws......................................................................................179
Concrete ...............................................................................................................179
Steel-to-concrete interface ...................................................................................182
5.3.2. Validation of the numerical model ..........................................................184
Piers with tension shift.........................................................................................184
Piers with lapped splices......................................................................................189
5.4. Numerical modelling of piers with hollow cross-section and FRP jackets .....193
5.4.1. Experimental results.................................................................................193
5.4.2. FEM modelling of the confinement effect...............................................194
5.4.3. Global behaviour of retrofitted pier .........................................................199
5.5. Final remarks on the assessment of modelling tools .......................................203
6. SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS......................205
6.1. Introduction......................................................................................................205
6.2. Seismic retrofit of reinforced concrete bridge piers with hollow cross-section
206
6.2.1. Retrofit with steel jackets.........................................................................206
6.2.2. Retrofit with FRP jackets.........................................................................208
6.2.3. Performance of retrofitted bridge piers....................................................211
6.3. Design of retrofit of bridge piers with FRP .....................................................214
6.3.1. Global retrofit procedure..........................................................................214
6.3.2. Relocation of critical cross-section..........................................................216
Piers with curtailment of vertical reinforcement .................................................216
Piers with lapped splices......................................................................................216
6.3.3. Anchorage ................................................................................................217
v
Overlaying............................................................................................................218
Mechanical anchorage .........................................................................................219
U-anchor ..............................................................................................................221
6.3.4. Retrofit for flexural strength enhancement ..............................................223
6.3.5. Retrofit for shear strength enhancement ..................................................225
6.3.6. Retrofit for confinement ..........................................................................231
General .................................................................................................................231
FRP-confined concrete.........................................................................................231
Design based on ultimate curvature.....................................................................237
Design based on upgrade index ...........................................................................241
Detailing...............................................................................................................243
6.3.7. Retrofit for enhancement of lapped splices .............................................244
6.4. Design of FRP jackets for piers with rectangular hollow cross-section ..........248
6.4.1. General .....................................................................................................248
6.4.2. Numerical analysis - effect on concrete properties..................................248
Description of the numerical model.....................................................................248
Effect of confinement within the cross-section ...................................................251
Effect of confinement on the concrete properties ................................................253
6.4.3. Numerical analysis - effect on cross-section ductility .............................259
Description of the numerical model.....................................................................259
Curvature ductility capacity.................................................................................260
Effectiveness index ..............................................................................................268
Enhancement of moment capacity .......................................................................271
Cyclic behaviour and energy-dissipation capacity ..............................................272
6.4.4. Design equations and recommendations..................................................272
Curvature ductility ...............................................................................................272
Effectiveness index ..............................................................................................275
Alternative definition of curvature ductility ........................................................276
Recommendations................................................................................................277
6.5. Concluding remarks .........................................................................................279
7. CONCLUSIONS AND FUTURE RESEARCH .....................................................283
7.1. Summary and conclusions ...............................................................................283
7.1.1. Performance of existing bridge piers with hollow cross-section .............283
7.1.2. Performance of existing bridge structures ...............................................284
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
vi
7.1.3. Assessment of numerical tools for existing bridge piers .........................286
7.1.4. Retrofit of bridge piers with hollow cross-section using FRP strips .......287
7.2. Suggestions for future research........................................................................288
REFERENCES ................................................................................................................291
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS...............................309
APPENDIX B – CONSTRUCTION DRAWINGS.........................................................321
APPENDIX C – PHOTOGRAPHIC DOCUMENTATION ...........................................329
vii
LIST OF TABLES
Table 3.1. Similitude relationship between the full-scale prototype (P) and the constructed
model (M) .........................................................................................................34
Table 3.2. Material properties of the specimens (average values).....................................36
Table 3.3. Mechanical properties of the specimens and seismic code requirements.........37
Table 3.4. Experimental and empirical values of plastic hinge length for the short pier ..49
Table 3.5. Park & Ang Damage Index [Park & Ang, 1985]..............................................52
Table 3.6. Bridge damage and performance assessment [Hose et al., 2000].....................53
Table 3.7. Damage assessment of the short pier ................................................................55
Table 3.8. Experimental and empirical displacement for the short pier ............................58
Table 3.9. Comparison of the two piers (values at ultimate displacement).......................62
Table 3.10. Experimental and empirical values of plastic hinge length for the tall pier ...70
Table 3.11. Damage assessment of the tall pier.................................................................70
Table 3.12. Experimental and empirical displacement for the tall pier .............................70
Table 3.13. Drift capacity of piers with hollow cross-section ...........................................75
Table 4.1. Longitudinal reinforcement steel ratio and characteristic values of
displacement for the scaled bridge piers ...........................................................95
Table 4.2. Eigenfrequencies of Warth Bridge .................................................................100
Table 4.3. Damage of the bridge piers .............................................................................121
Table 4.4. Maximum drift and ductility demand for the piers.........................................121
Table 4.5. Dissipated energy and Damage Index ............................................................121
Table 4.6. Overall Park and Ang Damage Index .............................................................124
Table 4.7. Displacement ductility and drift capacities for the cyclic and PSD tests .......129
Table 4.9. Damage ratios for highway bridges [FEMA, 1999] .......................................134
Table 4.10. Discrete values of restoration functions for highway bridges [FEMA, 1999]
.........................................................................................................................134
Table 4.11. Parameters for estimation of damage probability .........................................139
Table 4.12. Assessment of the Talübergang Warth Bridge according to HAZUS ..........140
Table 4.13. Equivalent damping (%) ...............................................................................141
Table 4.14. Displacement of control point (m) for mean ADRS.....................................144
Table 4.15. Displacement of control point (m) for EC8 ADRS and triangular distribution
of lateral forces ...............................................................................................144
Table 4.16. Characteristic values of the equivalent bilinear structures ...........................144
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
viii
Table 5.1. Material properties used in the numerical models of the piers .......................162
Table 5.2. Characteristics of different models .................................................................162
Table 5.3. Longitudinal reinforcement steel ratio for the scaled bridge piers .................175
Table 5.4. Values of dissipated energy (kNm) for the bridge piers .................................176
Table 5.5. Characteristic values of force and displacement for the bridge piers .............178
Table 5.6. Material properties for the tall pier A40 .........................................................188
Table 5.7. Material properties for the short pier A70 ......................................................191
Table 5.8. Material properties for concrete and FRP.......................................................197
Table 5.9. Concrete properties for different zones of the cross-section ..........................201
Table 6.1. Bridge pier test database .................................................................................213
Table 6.2. FRP material safety factors f? [fib, 2001]......................................................227
Table 6.3. Parameters k, α , β , γ and η from regression analysis [Monti et al., 1998] 242
Table 6.4. Material properties..........................................................................................250
Table A.1. Seismic-deficient piers with solid cross-section: geometrical and mechanical
properties and deformation capacity...............................................................309
Table A.2. Code-designed piers with solid cross-section: geometrical and mechanical
properties and deformation capacity...............................................................311
Table A.3. Retrofitted piers with solid cross-section: geometrical and mechanical
properties and deformation capacity...............................................................315
Table A.4. Seismic-deficient piers with hollow cross-section: geometrical and mechanical
properties and deformation capacity...............................................................317
Table A.5. Code-designed piers with hollow cross-section: geometrical and mechanical
properties and deformation capacity...............................................................318
Table A.6. Retrofitted piers with hollow cross-section: geometrical and mechanical
properties and deformation capacity...............................................................320
ix
LIST OF FIGURES
Figure 3.1. Talübergang Warth Bridge, Austria ................................................................28
Figure 3.2. Geometry of the scaled models of the short (a) and the tall (b) pier and typical
cross-section (c) ............................................................................................35
Figure 3.3. Instrumentation of the short (a) and the tall (b) pier .......................................39
Figure 3.4. Cyclic test on the short pier: force-displacement curve ..................................41
Figure 3.5. Cyclic test on the short pier: crack pattern at the end of the test.....................41
Figure 3.6. Cyclic test on the short pier: experimental and envelop force-displacement
curves ............................................................................................................45
Figure 3.7. Cyclic test on the short pier: flexural and shear displacement ........................46
Figure 3.8. Cyclic test on the short pier: distribution of average curvature along the height
.......................................................................................................................51
Figure 3.9. Cyclic test on the short pier: evolution of vertical deformation ......................52
Figure 3.10. Cyclic test on the short pier: Park & Ang Damage Index .............................53
Figure 3.11. Relation between wall parameters and ultimate curvature [Wallace &
Moehle, 1992] ...............................................................................................57
Figure 3.12. Force-drift diagrams for the short, A70, and the squat, PREC8, piers..........59
Figure 3.13. Equivalent damping – drift for the PREC8 and A70 piers ............................61
Figure 3.14. Equivalent damping – displacement ductility: theoretical expressions and
experimental values for the PREC8 and A70 piers.......................................61
Figure 3.15. Cyclic test on the tall pier: force-displacement curve ...................................63
Figure 3.16. Cyclic test on the tall pier: crack pattern at the end of the test......................64
Figure 3.17. Equilibrium of internal forces in diagonally cracked element with shear
reinforcement ................................................................................................66
Figure 3.18. Cyclic test on the tall pier: flexural and shear displacement .........................66
Figure 3.19. Cyclic test on the tall pier: distribution of average curvature along the height
.......................................................................................................................68
Figure 3.20. Cyclic test on the tall pier: evolution of vertical deformation.......................69
Figure 3.21. Cyclic test on the tall pier: Park & Ang Damage Index ................................71
Figure 3.22. Performance of seismic-deficient (left column) and code-designed (right
column) bridge piers with hollow cross-section ...........................................74
Figure 4.1. Talübergang Warth Bridge, Austria ................................................................84
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
x
Figure 4.2. Parallel procedures: simple inter-field procedure (a), improved inter-field
procedure (b) and intra-field procedure (c) ...................................................91
Figure 4.3. PSD test with substructuring of the Warth Bridge at the ELSA laboratory....94
Figure 4.4. Cross-section of the deck.................................................................................97
Figure 4.5. Mode shapes of the bridge.............................................................................100
Figure 4.6. Input accelerograms for the abutments and the pier bases (see Figure 4.3 for
abutment and pier labels) ............................................................................102
Figure 4.7. Response spectra of the accelerograms for the nominal earthquake.............103
Figure 4.8. Displacement histories for the 0.4xNE test, pre-test numerical analysis ......104
Figure 4.9. Force-drift diagrams for the 0.4xNE test.......................................................107
Figure 4.10. Damage pattern of the tall pier for the 0.4xNE test.....................................108
Figure 4.11. Displacement histories for the 0.4xNE test, experimental (solid line) and
numerical (thin line) results ........................................................................109
Figure 4.12. Force-drift diagrams for the 1.0xNE test.....................................................110
Figure 4.13. Displacement histories for the 1.0xNE test, experimental (solid line) and
numerical (thin line) results ........................................................................111
Figure 4.14. Damage pattern of the tall pier for the 1.0xNE test.....................................112
Figure 4.15. Force-drift diagrams for the 2.0xNE earthquake test ..................................113
Figure 4.16. Displacement histories for the 2.0xNE test, experimental (solid line) and
numerical (thin line) results ........................................................................114
Figure 4.17. Damage pattern of the short pier for the 2.0xNE test..................................114
Figure 4.18. Final collapse test on the tall pier: force-displacement diagram .................116
Figure 4.19. Damage pattern of the tall pier for the final collapse test............................116
Figure 4.20. Flexural and shear deformation of the short pier.........................................118
Figure 4.21. Flexural and shear deformation of the tall pier ...........................................118
Figure 4.22. Distribution of average curvature along the height of the short pier...........120
Figure 4.23. Distribution of average curvature along the height of the tall pier..............120
Figure 4.24. Vulnerability functions: Park and Ang Damage Index (a), drift ratio (b),
displacement ductility (c) and overall Park and Ang Damage Index (d)....125
Figure 4.25. Distribution of ductility demand (a), drift demand (b), Park and Ang Damage
Index (c) and percentage of dissipated energy (d) among the piers............128
Figure 4.26. Fast Fourier Transforms of the pier top displacement.................................128
Figure 4.27. Change in stiffness between adjacent piers and distribution of stiffness ....130
Figure 4.28. Response spectrum used in HAZUS [FEMA, 1999]...................................132
xi
Figure 4.29. Suggested displacement shapes, adapted from [Fajfar et al., 1997]............136
Figure 4.30. Acceleration response spectra for 5% damping: mean experimental (a) and
EC8 Type 1, Soil Class B and PGA = 0.36 g (b) ........................................139
Figure 4.31. Fragility curves for the Talübergang Warth Bridge ....................................140
Figure 4.32. Distribution of lateral forces........................................................................141
Figure 4.33. Force-displacement curves from pushover analysis....................................141
Figure 4.34. Evaluation of performance point for mean experimental ADRS: trapezoidal
(a) and triangular distribution of forces (b).................................................142
Figure 4.35. Maximum displacement of the piers: experimental values and CS method for
control at top of pier A30 (left column) and control at top of pier A70 (right
column) .......................................................................................................143
Figure 4.36. Evaluation of performance point for EC8-compatible ADRS: trapezoidal (a)
and triangular distribution of forces (b) ......................................................143
Figure 4.37. Mean displacement spectra for use in N2 method: experiment (a) and EC8
Type 1 for Soil Class B (b) .........................................................................147
Figure 4.38. Maximum displacement of the piers: experimental values and N2 method for
control at top of pier A30 (left column) and control at top of pier A70 (right
column) .......................................................................................................147
Figure 4.39. Assessment of the bridge for the EC8 spectrum and N2 method................148
Figure 5.1. Monotonic constitutive law for in compression (a) and tension (b)..............156
Figure 5.2. Cyclic constitutive law for concrete in compression (a) and tension (b) ......157
Figure 5.3. Monotonic (a) and cyclic (b) constitutive law for steel.................................158
Figure 5.4. Discretisation of alternative models for the cross-section.............................161
Figure 5.5. Pier A20: force-displacement monotonic curves for different models .........162
Figure 5.6. Short pier A70: experimental and numerical force-displacement curves .....166
Figure 5.7. Short pier A70: experimental and numerical moment-curvature curves.......166
Figure 5.8. Short pier A70: experimental and numerical dissipated energy versus top
displacement................................................................................................167
Figure 5.9. Tall pier T250: experimental and numerical force-displacement curves ......169
Figure 5.10. Tall pier T250: experimental and numerical dissipated energy versus top
displacement................................................................................................169
Figure 5.11. Tall pier A40: distribution of average curvature along the height of the pier
for the original model..................................................................................170
Figure 5.12. Tall pier A40: force-displacement curves for alternative models ...............171
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
xii
Figure 5.13. Numerical model with elastic base for the tall pier A40: force-displacement
curves (a) and dissipated energy (b)............................................................172
Figure 5.14. Numerical model with increased steel at the base for the tall pier A40: force-
displacement curves (a) and dissipated energy (b) .....................................173
Figure 5.15. Numerical model with reduced steel at the critical cross-section for the tall
pier A40: force-displacement curves (a), dissipated energy (b) and
distribution of curvature along the height (c)..............................................173
Figure 5.16. Talübergang Warth Bridge piers: force-displacement curves for the fibre
model and the damage model......................................................................176
Figure 5.17. Talübergang Warth Bridge piers: dissipated energy versus lateral
displacement for the fibre model and the damage model ...........................177
Figure 5.18. Talübergang Warth Bridge piers: force-displacement curves from pushover
analysis and bilinear envelope ....................................................................179
Figure 5.19. Concrete constitutive law: loading (a) and unloading (b) of a crack ..........180
Figure 5.20. Steel-to-concrete bond constitutive law: Eligehausen- Balázs model.........184
Figure 5.21. Evolution of damage for pier A40: numerical (a) and experimental (b) results
.....................................................................................................................186
Figure 5.22. Deformed shape for pier A40, numerical results (displacements magnified
x15) .............................................................................................................187
Figure 5.23. Tall pier A40: experimental and numerical force-displacement curves......188
Figure 5.24. Mesh of the numerical model: concrete (a), longitudinal steel (b), and
transverse steel (c) elements........................................................................190
Figure 5.25. Close-up at the base of the mesh: concrete and steel elements ...................190
Figure 5.26. Short pier A70: experimental and numerical force-displacement curves ...191
Figure 5.27. Crack pattern of the short pier A70: numerical analysis for continuous
reinforcement (a), numerical analysis with joint elements (b) and
experimental data (c)...................................................................................193
Figure 5.28. Tall pier T250: force-displacement curves of the as-built and retrofitted
specimens [Peloso, 2003]............................................................................195
Figure 5.29. Numerical stress-strain curves for concrete under uniform compression ...196
Figure 5.30. Tall pier T250-FRP: numerical stress-strain curves for FRP-confined
concrete (a) and definition of zones (b) ......................................................198
Figure 5.31. Tall pier T250-FRP: distribution of maximum axial stress within the cross-
section .........................................................................................................199
xiii
Figure 5.32. Tall pier T250-FRP: different zones of concrete in the cross-section mesh200
Figure 5.33. Tall pier T250-FRP: numerical force-displacement monotonic curves ......200
Figure 5.34. Tall pier T250-FRP: experimental and numerical force-displacement curves
.....................................................................................................................202
Figure 5.35. Tall pier T250-FRP: experimental and numerical dissipated energy versus
lateral displacement.....................................................................................202
Figure 6.1. Drift capacity versus aspect ratio, transverse reinforcement ratio, splicing
length and normalised axial load for seismic-deficient (left) and code-
designed (right) piers...................................................................................212
Figure 6.2. Increase in drift capacity versus aspect ratio (a), transverse reinforcement ratio
(b), splicing length (c) and normalised axial load (d) .................................213
Figure 6.3. Global retrofit procedure for seismic-deficient bridge piers .........................215
Figure 6.4. Anchorage of flexural FRP reinforcement with external confinement
reinforcement ..............................................................................................220
Figure 6.5. Anchorage of shear FRP reinforcement in the compression zone by
embedment in concrete (a) and using bolts (b) ...........................................220
Figure 6.6. Anchorage of CFRP strips [Zehetmaier, 2000] .............................................222
Figure 6.7. Cross-section of an after-corner U-anchor [Khalifa et al., 1999]..................222
Figure 6.8. Cross-section analysis: (a) geometry, (b) strain distribution and (c) stress
distribution ..................................................................................................224
Figure 6.9. Bilinear constitutive model for FRP-confined concrete ................................232
Figure 6.10. Average absolute error: prediction of strength (a) and ultimate deformation
(b) of FRP-confined concrete [De Lorenzis, 2001] ....................................236
Figure 6.11. Moment – axial force interaction diagrams (a) and moment – curvature
monotonic curves (b) for different constitutive laws for FRP-confined
concrete [Yuan et al., 2001] ........................................................................236
Figure 6.12. Effectiveness of confinement for rectangular jacket ...................................237
Figure 6.13. Confinement of circular column with circular jacket ..................................239
Figure 6.14. Confinement of rectangular column with rectangular (a) and oval (b) jacket
.....................................................................................................................239
Figure 6.15. Failure of lapped splices ..............................................................................244
Figure 6.16. Confinement of circular column with circular jacket, equilibrium of internal
forces ...........................................................................................................246
Figure 6.17. Definition of cross-section geometry ..........................................................249
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
xiv
Figure 6.18. Concrete (blue) and FRP (red) jacket mesh ................................................250
Figure 6.19. Stress-strain curve of concrete for uniaxial compression............................250
Figure 6.20. Effect of jacket height and thickness on the compressive strength of concrete
.....................................................................................................................252
Figure 6.21. Stress-strain curves for confined concrete ( jh = 0.00 m) ............................254
Figure 6.22. Stress-strain curves for confined concrete ( jh = 0.05 m) ............................254
Figure 6.23. Stress-strain curves for confined concrete ( jh = 0.10 m) ............................255
Figure 6.24. Stress-strain curves for confined concrete ( jh = 0.15 m) ............................255
Figure 6.25. Stress-strain curves for confined concrete ( jh = 0.20 m) ............................256
Figure 6.26. Effect of jacket on the compressive strength of concrete............................258
Figure 6.27. Effect of jacket on the residual strength of concrete ...................................258
Figure 6.28. Effect of jacket on the softening stiffness of concrete ................................259
Figure 6.29. Concrete and steel mesh for the moment-curvature analysis ( jh = 0.05 m) 261
Figure 6.30. Material constitutive laws: concrete (a) and steel (b)..................................261
Figure 6.31. Definition of failure criteria and curvature ductility: at maximum moment (a)
and at 20% loss of strength (b)....................................................................261
Figure 6.32. Effect of axial load and amount of reinforcement on the curvature ductility
(as-built cross-section) ................................................................................262
Figure 6.33. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.00 m) ...............................................................................................263
Figure 6.34. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.05 m) ...............................................................................................263
Figure 6.35. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.10 m) ...............................................................................................264
Figure 6.36. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.15 m) ...............................................................................................264
Figure 6.37. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.20 m) ...............................................................................................265
Figure 6.38. Effect of reinforcement ratio ( jh = 0.10 m, jt = 5 mm, ν = 0.1) .................266
Figure 6.39. Effect of axial load ( jh = 0.10 m, jt = 5 mm, sρ = 1.02%)..........................266
xv
Figure 6.40. Effect of jacket thickness ( jh = 0.10 m, sρ = 1.02%, ν = 0.1).....................267
Figure 6.41. Effect of jacket height ( jt = 5 mm, sρ = 0.17%, ν = 0.1) ............................267
Figure 6.42. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.00 m) ...............................................................................................269
Figure 6.43. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.05 m) ...............................................................................................269
Figure 6.44. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.10 m) ...............................................................................................270
Figure 6.45. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.15 m) ...............................................................................................270
Figure 6.46. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.20 m) ...............................................................................................271
Figure 6.47. Cyclic behaviour: moment-curvature diagrams for the as-built and jacketed
cross-section ( ν =0.2, jt = 5 mm, sρ = 1.02%)............................................272
Figure 6.48. Empirical fit to the numerical values of curvature ductility: ν ≤ 0.1 (a) and
ν > 0.1 (b)....................................................................................................274
Figure 6.49. Comparison between empirical and numerical values of curvature ductility:
ν ≤ 0.1 (a) and ν > 0.1 (b) ..........................................................................275
Figure 6.50. Comparison between empirical and numerical values of effectiveness index:
ν ≤ 0.1 (a) and ν > 0.1 (b) ..........................................................................276
Figure 6.51. Empirical fit to the numerical values of curvature ductility........................278
Figure 6.52. Empirical (5% characteristic) and numerical values of the effectiveness
index............................................................................................................278
Figure B.1. Vertical reinforcement of pier A70 (side view)............................................321
Figure B.2. Horizontal reinforcement of pier A70 ..........................................................321
Figure B.3. Vertical reinforcement of pier A70 (sections A-A, B-B) .............................322
Figure B.4. Vertical reinforcement of pier A70 (sections C-C, D-D) .............................323
Figure B.5. Vertical reinforcement of pier A40 (side view)............................................324
Figure B.6. Horizontal reinforcement of pier A40 ..........................................................325
Figure B.7. Vertical reinforcement of pier A40 (sections A-A, B-B) .............................326
Figure B.8. Vertical reinforcement of pier A40 (sections C-C, D-D) .............................327
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
xvi
Figure C.1. Talübergang Warth Bridge in Austria (both independent lanes are shown).329
Figure C.2. General view of the tested piers A40 and A70 inside the laboratory ...........329
Figure C.3. Cyclic test on the short pier A70: crack pattern of the flange (a) and the web
(b) at the end of the test...............................................................................330
Figure C.4. Cyclic test on the short pier A60: crack pattern of the flange (a) and the web
(b) at the end of the test...............................................................................330
Figure C.5. 0.4xNE test: crack pattern of the flange (a) and the web (b) of pier A40.....331
Figure C.6. 2.0xNE test: crack pattern of the flange (a) and the web (b) of pier A70.....331
Figure C.7. Final collapse test on the tall pier A40: crack pattern of the web.................332
Figure C.8. Pier A40: buckling of longitudinal reinforcement at 3.5m...........................332
Figure C.9. 1.0xNE test: hysteresis loops for substructured piers A20 and A30 (a) and on-
line comparison of experimental and pre-test displacement histories (b)...333
xvii
AKNOWLEDGEMENTS
The present thesis is submitted to the Department of Structural Engineering of the
Technical University of Milan (Politecnico di Milano - Dipartimento di Ingegneria
Strutturale) for the fulfilment of the PhD degree within the course on Earthquake and
Geotechnical Engineering and Soil-Structure Interaction (Ingegneria Sismica, Geotecnica
e dell’Interazione Ambiente-Struttura).
The author is sincerely grateful to Professor Giandomenico Toniolo, tutor and supervisor
of the thesis, for the excellent collaboration and his useful comments throughout the
whole duration of the research.
The greatest part of the work was performed while the author was a grant holder at the
ELSA laboratory of the JRC, Contract No: 15775-2000-03 P1B20 ISP IT, financed by the
SAFERR research network, CEC Contract No: HPRN-CT-1999-00035. The experiments
were partly financed by the VAB research programme, Environment & Climate Project
ENV4-CT97-0574. The financial support of the European Commission is acknowledged.
To Dr Artur Vieira Pinto, scientific responsible of the author at the JRC, are due deeply-
felt thanks for his friendship and support, for the joy of working together and for his
continuous encouragement to improve the quality of the work.
The author expresses his gratitude to Dr Javier Molina for the meticulous preparation and
control of the tests, to Dr Pierre Pegon for his continuous support on numerical
modelling, to Professor Michel Géradin, ELSA Head of Unit, and to all ELSA staff for
the warm welcome and for providing all the means and support that allowed to
successfully perform the work in the laboratory.
To the, unexpectedly many, new friends from the JRC and DIS, as well as to the old ones
from Greece, the author is thankful for they rendered pleasant the few pauses.
I thank the above for their interest that provided a motivation to continue my work at the
times when I mostly needed it; more than that, for supporting me in pursuing a passion.
To my family I dedicate my most personal thoughts during the periods of questioning and
doubts.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
xviii
INTRODUCTION
1
1. INTRODUCTION
1.1. GENERAL – MOTIVATION FOR THE RESEARCH
This thesis deals with the seismic assessment and retrofit of existing reinforced concrete
bridges, with focus on piers with rectangular hollow cross-section. To what regards
assessment, an experimental campaign was followed and the test results were exploited
for the study of the seismic performance of single piers and complete bridge structures. A
numerical approach was adopted for the study of retrofitted hollow bridge piers. This first
chapter intends to introduce the subject by highlighting the importance of bridges, which,
combined with the seismic vulnerability of existing structures designed without
earthquake resistance, calls for the development of appropriate retrofit techniques.
Compared to buildings, bridges are less redundant and thus they have limited potential of
avoiding total collapse through the distribution of damage to a large number of plastic
zones within the structure. Indeed, collapse of a single beam/span and, even more, of a
column will most probably result in failure of the complete structure. In addition, most
bridges are valuable during the immediate post-earthquake emergency, since they are
required to ensure transport of heavy machinery, first-aid supplies and eventual victims
between earthquake-struck and surrounding areas. Being parts of complex
communication lifelines, bridges need to maintain a high level of occupancy, even in the
event of a strong earthquake. This is contrary to normal buildings for which significant,
but repairable, damage is accepted. Severe earthquake-induced damage on bridges results
in economic losses in the form of repair, or replacement, costs and disruption of traffic.
The above explain why particular attention and special studies are dedicated to bridges,
even though in most cases they can be considered as simple plane-frame structures.
The poor seismic behaviour of existing bridge structures has been verified during all the
recent important earthquakes, such as the 1987 Whittier Narrows earthquake [Gates et al.,
1988], the 1989 Loma Prieta earthquake [Housner & Thiels, 1990], the 1994 Northridge
earthquake [Housner & Thiels, 1995], the 1995 Hyogo-Ken Nambu (Kobe) earthquake
[Seible et al., 1995a; Kawashima & Unjoh, 1997] and the 1999 Kocaeli-Duzce
earthquakes [Imbsen et al., 2000]. Field and experimental observations allowed to
identify the main seismic deficiencies of existing bridges. They concern the abutments,
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
2
deck, columns, cap beams and foundation elements. Considering columns in particular,
the most common problems are the limited shear strength, presence of lapped splices in
the critical zones, limited ductility capacity and premature termination of longitudinal
reinforcement. It is argued that older bridge piers were designed with focus on strength
rather than deformation and without provisions to ensure stability of the response in the
post-elastic range [Pinto & Monti, 2000]. These observations support the need for retrofit
and also provide guidance on the targets to be sought.
Rectangular, octagonal, circular or wall-type solid cross-sections are often used for bridge
piers. In the case of tall piers, it is desirable to reduce the mass of the pier and
consequently the seismic loads it has to resist. In the USA the trend is to use solid
sections and to reduce the cross-sectional dimensions with height. In contrast, piers with
rectangular hollow cross-section are commonly used in Japan and Europe for highway
bridges that cross deep valleys [Hooks et al., 1997]. Despite the large population of
existing bridge piers with hollow cross-section, their seismic performance and appropriate
retrofit techniques have not been studied until recently. This provided the motivation to
focus the research presented herein to the assessment and retrofit of bridge piers with
rectangular hollow cross-section.
While various alternatives exist for the seismic upgrading of buildings [Fardis, 1998], the
practical solutions for bridges are rather limited. It is either desired to reduce the seismic
demand by modifying the structural response, or to increase the available strength and
deformation capacity. The use of isolation/dissipation devices for the seismic retrofit of
bridges often requires iterative procedures, or the study of alternative combinations of the
properties and position of the devices. Keeping in mind the code requirements for the
analysis of isolated bridges and the characterisation of the isolation/dissipation devices, it
seems that rather onerous studies are needed for each and every structure. Increasing the
capacity of the structure appears therefore as an appealing alternative, provided the
effectiveness of retrofit is verified and reliable design tools are available. The choice
between response modification and structural improvement should be supported by cost-
effectiveness studies and remains somehow at conceptual level. Without further
discussing this choice, the research described in this thesis follows the direction of
increasing the capacity of members.
INTRODUCTION
3
When designing the retrofit of a bridge structure, an important requirement is to consider
the cost of application, the large scale of the structure and the disruption of traffic. This is
the reason why fibre-reinforced polymer (FRP) materials have been extremely popular for
the seismic retrofit of bridges. Although the material is more expensive than reinforced
concrete and steel, the low transportation and application cost make it more attractive.
This lead to their use as a remedy to all problems and in some cases in over-designed
applications. Only recently they are seen with some scepticism and their absolute
effectiveness is questioned. While a sound background exists for the calculation of
flexural and shear strength [Seible et al., 1995b; fib, 2001], there is scarce confidence
with respect to design for confinement [De Lorenzis, 2001]. This provided the motivation
to study the effectiveness and eventual limitations of the use of FRP reinforcement for the
seismic retrofit of hollow piers with large dimensions.
The research was partly developed as contribution to two EC-funded research projects.
The VAB1 (Vulnerability Assessment of Bridges) project originated from the need to
assess the existing bridge stock, as a result of the inadequate seismic design of older
bridge structures, in combination with the recent revision of seismic hazard maps across
Europe. Particular attention was devoted to irregular highway bridges with rectangular
hollow piers. The evaluation of the deformation capacity of members with no seismic
detailing and the study of strengthening schemes and guidelines were among the tasks of
the SAFERR2 (Safety Assessment for Earthquake Risk Reduction) project.
1.2. OBJECTIVES OF THE RESEARCH
With the aim of assessing the cyclic behaviour of existing bridge piers with hollow cross-
section, cyclic tests on two large-scale (1:2.5) specimens of piers were performed. This is
seen as a contribution to the limited information available in the international literature on
bridge piers with this particular geometry. In fact, these are the first tests on large-scale
specimens of piers with rectangular hollow cross-section without earthquake detailing,
corresponding to an existing highway bridge situated in Austria. A second objective was
the calibration of non-linear numerical models that were employed for the substructured
piers during the pseudodynamic tests that followed.
1 www.arsenal.ac.at/vab 2 www.saferr.net
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
4
Aiming at the experimental assessment of a reinforced concrete bridge for increasing
intensity of the seismic input, a series of pseudodynamic tests on a complete bridge
structure were performed. A parallel objective was to implement the substructuring
technique in pseudodynamic testing with non-linear models for the numerical
substructure.
Having verified the seismic vulnerability of substandard bridges, a final objective related
to retrofitting was set. It is recognised that there is adequate knowledge and confidence on
the techniques and design tools for flexural and shear strength enhancement of seismic-
deficient bridge piers. In contrast, there is limited confidence in the effectiveness of
jacketing for increase of confinement for bridge piers with rectangular hollow cross-
section. Therefore, it was decided to undertake a numerical study on the effectiveness of
FRP jackets for confinement of rectangular hollow cross-sections.
1.3. OUTLINE OF THE THESIS
The thesis comprises seven chapters. Three appendices are given at the end, presenting
the main geometrical and mechanical characteristics of the examined bridge piers,
construction drawings for the tested specimens and a photographic documentation. A
short description of the contents of each chapter follows.
Following the present Chapter 1 which gives the motivation and objectives of the
research, Chapter 2 presents a bibliographic research on the effects of asynchronous input
motion and soil-structure interaction, innovative displacement-based methods, use of
isolation and dissipation devices in bridge engineering, procedures and indices for
damage assessment, methods of analysis and modelling and finally, techniques for the
seismic retrofit of bridge piers.
Chapter 3 deals with the cyclic tests on large-scale models of one short and one tall bridge
pier. The original piers belong to an existing highway bridge, situated in Austria and
designed in 1975. The experimental results are presented in terms of damage evolution,
force-displacement hysteretic curves, deformation and energy-dissipation capacity. A
literature review of experimental results on hollow bridge piers is presented and a
preliminary estimation of performance limits is based on the experimental results. This
INTRODUCTION
5
highlights the different capacities of piers designed before and after the introduction of
modern seismic codes and the need for retrofitting of seismic-deficient piers.
A series of pseudodynamic (PSD) tests performed on a large-scale (1:2.5) model of an
existing highway bridge are presented in Chapter 4. The research was focused on the
Talübergang Warth Bridge, which is considered representative of European highway
bridges designed before the modern seismic codes came into practice. The PSD testing
method is briefly presented, considering the particular cases of asynchronous input
motion and non-linear substructuring. The test results are presented in terms of hysteretic
curves, dissipated energy, deformation demands and distribution of damage. Standard
assessment procedures are checked against the experimental results. A simplified
procedure, appropriately corrected for the given structure, is applied for the assessment of
the bridge situated in a high-seismicity area: collapse is predicted for the design
earthquake (475 years return period) and therefore, retrofit of the bridge is imperative.
Numerical modelling is discussed in Chapter 5. Simplified fibre/beam models with
different configurations are calibrated on the results of the cyclic tests. Certain
modifications are introduced to surmount the limitations of these models, necessary for
the successful completion of the PSD tests. Refined FEM analyses allow overcoming the
restrictions of simplified models, at the expense of much larger computational demand.
Finally, a combination of these two approaches, namely FEM analysis for the material
properties and fibre modelling of the structural element, is validated against experimental
results for the study of hollow bridge piers retrofitted with FRP strips.
Chapter 6 deals with retrofitting and in particular with the problem of confinement for
rectangular hollow cross-sections with large dimensions. The effect of FRP wrapping on
the concrete properties in different parts of the cross-section is studied first and the
empirical constitutive laws for FRP-confined concrete are found inadequate for the
examined case. The results of these analyses are integrated in moment-curvature analyses
performed with the aim to study the effect of jacket dimensions, amount of reinforcement
and axial load on the ductility capacity of the cross-section. The effectiveness and
limitations of this retrofit method are discussed and finally, an empirical design equation
is formulated on the basis of more than 1000 numerical simulations.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
6
The main achievements are recapitulated and the principal conclusions are put forward in
Chapter 7. Some suggestions for future research are given, on the grounds of the
difficulties encountered during the development of the research and on the limitations of
the findings.
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
7
2. OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES -
A LITERATURE REVIEW
2.1. ASYNCHRONOUS EXCITATION
In design practice the effect of asynchronous input motion is not fully understood and due
to the complexity of the issue (representation of the seismic motion and time-consuming
non-linear dynamic analyses) it is usually ignored. According to Part 2 of Eurocode 8
(EC8-2) [CEN, 2002], the spatial variability of the input motion at the supports of bridges
must be taken into consideration for bridges in the presence of geological discontinuities
or marked topographical features and for bridges that are longer than 600 m. The effects
of the variability of the input motion on the response of the bridge are expected to be in
general small. Therefore, it is allowed to disregard them, or to use an idealised model.
Following the Italian seismic code [Ordinanza 3274, 2003], the total response of the
structure is obtained by adding the dynamic effects calculated from a response-spectrum
analysis and the pseudo-static effects for relative movement between the pier bases and
abutments.
Similar guidelines are provided by the Caltrans Seismic Design Criteria [Caltrans, 1999].
The Japanese seismic code [JSCE, 1996] requires the consideration of the epicentral
characteristics and amplification of the surface layer in defining the input motion for the
design earthquake with a rare probability of occurrence. The New Zealand Standards
[SNZ, 1995] recognise the need to increase the seismic loading due to site effects, but do
not directly take into consideration asynchronous motion of multi-support structures.
Following the Swiss code SIA 160 [SIA, 1989], a relative displacement at the base of
different piers or abutments due to travelling seismic waves is considered, in order to
prevent fall-down of the superstructure. Alternatively, it is recommended to provide the
structure with movement joints that subdivide the structure in simpler frames [Priestley et
al., 1996].
The main sources of the spatial variability of the ground motion are the finite dimensions
of the seismic source, the material heterogeneities and geometrical irregularities of the
earth surface and the wave-passage effects. The latter being easier to consider in defining
the asynchronous motion, more significant are the geological and topographical
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
8
irregularities, which are more complex and require detailed studies [Faccioli & Paolucci,
1990; Faccioli, 2002].
Dynamic analyses indicate that the effect of asynchronous motion is to decrease the
dynamic component and increase the pseudo-static component of the structural response.
From the structural point of view, it has been proved that designing for synchronous
motion, as is the common practice, provides a global upper bound of the response [Monti
et al., 1996]. Experimental results of an irregular bridge tested under asynchronous
earthquake input verify the above findings [Calvi & Pinto, 1996; Pinto et al., 1996].
On the other hand, an analytical study of a regular bridge subjected to non-stationary
multi-support random excitation suggests that the internal forces are in general reduced
for both the superstructure-deck and the substructure-piers, but in certain cases an
increase was observed [Perotti, 1990].
A theoretical study considering asynchronous motion due to wave-passage and
incoherence effects as well as local soil conditions proved that in certain cases, e.g. stiff
structures, the differential support motion may result in larger internal forces [Der
Kiureghian & Neuenhofer, 1992].
Dynamic analysis of an existing regular bridge showed that for small variability of the
ground motion the shear forces are decreased and the displacements increased, compared
to synchronous input, for both longitudinal and transverse excitation [Kahan et al., 1996].
Dynamic analysis of a regular bridge considering local soil amplification suggests
significant increase of shear forces and displacements in the case of substantial soil
amplification [Zembaty & Rutenberg, 1998].
A statistical approach has been adopted to study the importance of asynchronous motion
on the response of bridges [Lupoi et al., 2003]. Both regular and irregular bridges were
analysed, considering different combinations of the soil properties and levels of loss of
coherence and wave-passage effects. The results show that in presence of spatial
variability, the displacement ductility demands increase in the majority of the cases.
The coupling of spatial variability, soil-structure interaction and site effects was studied
through numerical parametric analyses [Sextos et al., 2003]. It was found that the effect
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
9
of spatial variability might increase up to 3.5 times the displacement demand and 50% the
bending moments, with respect to synchronous motion. In addition, the actual ductility
demand might be underestimated by a factor of 3, at the most unfavourable case, when
synchronous motion is considered in the design process.
From the above discussion it becomes clear that, although the importance of
asynchronous excitation is recognised for the design of bridges, there is absence of
general guidelines. This is due to the large scatter of the values that quantify the causes of
spatial variability of the ground motion. This in turn, makes their effects case-sensitive
and calls for relatively detailed analysis for each bridge and site.
2.2. DISPLACEMENT-BASED DESIGN AND ASSESSMENT
In recognition of the fact that earthquakes impose displacements, and not forces, on the
structures, several displacement-based procedures for the design and assessment of
structures have been proposed. At seismic code level, EC8-2 [CEN, 2002] introduces the
verification of deformation capacity of plastic hinges, in terms of rotation, for the case of
non-linear dynamic time-history analysis. In all other cases for new structures,
verification is performed for strength against applied forces. Part 3 of EC8 (EC8-3)
[CEN, 2003b], which deals with the strengthening and repair of buildings, makes the
distinction between ductile and brittle components and mechanisms and requires
verifications for three limit states, namely Near Collapse, Severe Damage and Damage
Limitation. Depending on the type of member and limit state, verifications are based on
either force or deformation. The Caltrans Seismic Design Criteria require that demand-
versus-capacity verifications of the members be performed for displacements and not
forces. The AASHTO Specifications [Buckle & Friedland, 1995] design approach is
based on forces, while displacements are mainly checked for bearing lengths.
Concerning design and assessment procedures, the basic concept is to substitute the real
structure with an equivalent linear single-degree-of-freedom (SDOF) one [Shibata &
Sozen, 1976]. Then, having estimated the equivalent damping, the effective period can be
derived for the target displacement from a displacement spectrum. Finally, the base shear
is defined from the stiffness corresponding to the effective period. This procedure is
called direct displacement-based design. In an early work [Iwan & Gates, 1979] that
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
10
compares different methods for the estimation of the properties of the equivalent linear
structure, the basic principles of the method are clarified. An overview of the recent
developments on the displacement-based design and assessment procedures can be found
elsewhere [Calvi, 1999; Priestley, 1998]. One drawback of the method is that damping
properties are estimated on the basis of semi-empirical expressions, although it is
understood that they play an important role in the procedure.
Displacement-based procedures have been applied for the design of single reinforced
concrete (RC) bridge columns and the validity of the method has been verified by
dynamic inelastic time history analyses [Kowalsky et al., 1995]. Concerning multi-
degree-of-freedom (MDOF) bridges, the concept works well for symmetrical structures,
but suffers some shortcomings for the case of irregular bridges [Calvi & Kingsley, 1995;
Fajfar et al., 1997]. This is due to the fact that a deflected shape, similar to the dominant
vibration mode, is assumed for the equivalent structure in order to achieve uniform
damage levels in the piers. Such assumption is true for regular bridges, but is not the case
for irregular structures, for which higher modes have a significant contribution to the
dynamic response.
To overcome these problems, a more elaborated procedure for the displacement-based
design of MDOF bridges, with either flexible or rigid superstructure, has been proposed
[Kowalsky, 2002]. In this, an effective mode shape is assumed for the equivalent
structure, as a combination of the modes calculated on the basis of the secant stiffness of
the piers and abutments and the elastic stiffness of the deck. The damping of the system is
a work-weighted sum of the values of damping of the components and then the standard
procedure for direct displacement-based design is followed. The accuracy of the proposed
method was verified for bridges with varying geometry and boundary conditions against
accurate dynamic analyses and consistency was observed.
An alternative procedure, supplementary to the current force-based ones, has been
proposed for RC buildings [Fardis & Panagiotakos, 1997]: the deformation demand is
estimated for a given lateral loading, corresponding to the serviceability earthquake, and
then the members are detailed to provide adequate deformation capacity. An expression
of the chord rotation capacity of members has been fit to a large databank of experimental
results on columns, beams and shear walls.
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
11
2.3. SOIL-STRUCTURE INTERACTION
Soil-structure interaction (SSI) and its effects on the response of structures are known for
well 30 years now. It is recognised that ignoring SSI may introduce errors on the unsafe
side when the spectral acceleration increases for periods higher than the fundamental
period of the structure [Newmark & Rosenblueth, 1971]. Apart from such extreme cases,
SSI is generally considered beneficial for structures.
At seismic code level, in EC8-5 [CEN, 2001] the lengthening of the fundamental period
of the structure, the increase in damping and the change of the eigenmodes and the modal
participation factors are recognised. Consequently, EC8 demands the consideration of SSI
for structures where P- δ effects are significant, for structures with massive or deep
foundations (e.g. bridge piers, offshore caissons and silos) or founded on very soft soil
and for very tall structures. The same criterion is adopted by the Italian seismic code
[Ordinanza 3274, 2003]. Following the New Zealand seismic code and the Caltrans
Seismic Design Criteria, SSI is always considered beneficial, as it reduces the seismic
design forces.
In order to quantify the effects of SSI, the lengthening of the fundamental period and the
increase in damping are typically estimated. Non-parametric methods using simple
transfer functions, e.g. [Paolucci, 1993], as well as parametric ones that study numerical
models for the transfer functions, e.g. [Stuart & Fenves, 1998], have been proposed for
the estimation of the dynamic properties of the structure based on the recordings of its
response during earthquakes. In addition, theoretical expressions, based on the soil and
structure properties as well as the geometry of the structure-foundation system, have been
proposed, e.g. [Stuart et al., 1999].
With the aim of studying the effects of SSI on inelastic structures, numerical
investigations are complementary to observations of data from instrumented structures.
Numerical analyses of 240 cases concerning a single bridge pier subjected to artificial
accelerograms matching the EC8 design spectrum, lead to the conclusion that the effects
of SSI are significant only in the extreme case of very stiff structures on soft soil
[Ciampoli & Pinto, 1995]. In terms of displacement ductility, the phenomenon seems to
have little effect, if not slightly beneficial. SSI results in increase of the displacement, but
only in terms of rocking motion due to the deformability of the soil and not in terms of
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
12
inelastic deformation of the pier. Similar results were obtained from non-linear pushover
analyses of bridge piers with various configurations [Elnashai & McClure, 1996].
In contrast, dynamic non-linear analyses of a similar model using accelerograms recorded
on soft-soil sites suggest significant increase of the ductility demand [Gazetas &
Mylonakis, 2001]. In fact, SSI effects are considered to have contributed to the collapse
of the Hanshin Expressway during the Kobe earthquake [Mylonakis & Gazetas, 2000].
The coupling between spatial variability of the ground motion, site effects and SSI was
studied considering different cases of symmetry, regularity and boundary conditions for a
bridge structure [Sextos et al., 2002]. It was found that the effects on the calculated
displacement ductility of the piers are important in the case of ground motions rich in low
frequencies, abrupt changes of soil stiffness, significant site response phenomena and
wave-passage effects.
2.4. BRIDGES WITH ISOLATION AND DISSIPATION DEVICES
Seismic isolation is an attractive alternative to design for ductility. In other words, the
effort on earthquake design of bridges is focused on minimising the forces to be resisted
by the piers. Isolation and dissipation devices can be used with the aim either to lengthen
the period of the structure (decrease the force but increase the displacement), or to
increase the damping (decrease both force and displacement), or for a combination of the
two. For the first case, particular attention should be paid for input motions with
significant amplification in the longer periods. For flexible bridges the reduction of forces
is insignificant, while displacements still increase linearly [Priestley et al, 1996].
The simplest device is the laminated rubber bearing consisting of horizontal steel plates
inserted in a block of rubber. This device offers small increase in damping. Inserting a
lead plug in the rubber bearing provides restoring force and damping [Robinson, 1982].
In addition, lead has good fatigue performance (actually, lead-rubber bearings retained
most of their dissipation capacity after five earthquake tests) and reliable mechanical
properties. Sliding bearings exhibit unreliable response, no centering force and should be
combined with dissipation devices.
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
13
Sliding bearings with a pendulum-type response constitute the principle of the friction
pendulum device (FPD). Experimental studies of a simple bridge structure isolated with
FPD verified the insensitivity of the device to the amplitude and frequency content of the
input motion, the stability of the response for a large number of cycles and also the
reduction of the shear forces and drift of the piers to almost half of the values for the non-
isolated bridge [Tsopelas et al., 1996]. Analytical studies of a regular bridge with FPD
between all piers and the deck showed that for the case of hard foundation soil, a
reduction in forces and displacements was obtained, with respect to the non-isolated
bridge, while a more complex distribution of the devices among the piers was needed for
efficient isolation in the case of soft foundation soil [Wang et al., 1998].
Steel hysteretic dampers (of cantilever, butterfly or crescent moon shape) are used to
dissipate energy and increase the damping of the structure in the desired directions.
Hydraulic dampers have been originally used to accommodate displacements at thermal
joints, but can be used as isolation/dissipation devices. Similar dissipation capacity
originates from the plastic deformation of lead in lead-extrusion dampers. The dampers
have no centering effect, but could be used as jacks to reposition the deck after the
earthquake. Shock transmitters, that allow slow displacements and prevent dynamic ones,
can be used between selected structural elements so that the structure remains isostatic for
normal use, while it becomes hyperstatic during an earthquake shacking [FIP, 2002].
Considering modelling and analysis methods, increasing degrees of sophistication,
namely: static linear, dynamic modal and time-history analysis, are proposed for
increasing importance of the bridge structure [Priestley et al., 1996]. It is recognised,
though, that safety factors that account for the reliability of the properties of the devices
have yet to be codified.
At seismic code level, EC8-2 [CEN, 2002] dedicates a section to the design of isolated
bridges and allows only fully isolated bridges, i.e. both the superstructure and
substructure should remain essentially elastic. Either elastic response spectrum
(fundamental mode or multimode) analysis or non-linear dynamic analysis can be
performed. Prototype tests of the devices are required for the characterisation of the
deformation and damping properties, as well as the effects of bilateral load, temperature,
aging and load history; a testing sequence and acceptance criteria are prescribed. The
isolating system is also demanded to provide a certain level of lateral restoring force.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
14
A comprehensive review of analysis methods required by seismic codes in the USA and
Japan can be found in [Hwang et al., 1994]. One approach is to define via semi-empirical
expressions the damping of the isolation devices and then enter the elastic design
spectrum with a combination of them. Either SDOF or MDOF structures are considered.
Alternatively, the effective damping and stiffness of the devices are determined and a
strain energy-based combination of the damping of all components is used to define the
overall structural damping. A tributary mass-based sum of the damping ratios has also
been proposed [Priestley et al., 1996].
Application to a sample bridge has shown that all the aforementioned methods provide
accurate results for the purposes of bridge design. Nevertheless, the accuracy of each
method seems to depend on the characteristics of the input earthquake. Linear time-
history analysis was found to be more accurate than response-spectrum analysis [Hwang,
1996]. Different simplified methods for the analysis of isolated bridges were compared
for three types of regular and irregular bridges [Franchin et al., 2001]. It was found that
the results of simplified response spectrum analyses were consistent with those of more
elaborated ones. The difficulties in identifying, in a comprehensive way, the damping
properties of the equivalent linear system were discussed and such procedure was found
as time-consuming as non-linear time-history analysis.
Considering the design of isolation for existing bridges, an iterative procedure is proposed
[Fardis & Calvi, 2001]. The strength of the devices is limited by the given strength of the
piers; the use of a safety factor is also recommended. Then, the devices should be
designed to limit the absolute and relative displacement of the deck. Assuming at first
stage a deformed shape for the deck, the combination of devices is checked, within an
iterative procedure, by means of a regularity index [Calvi & Pavese, 1996].
Seismic isolation can be applied to regularise the response of bridges with abrupt changes
in stiffness between adjacent piers. The combined use of bearings and hydraulic dampers
was proved successful in safely designing long viaducts, otherwise impossible to
construct [Isakovic et al., 2002]. Rubber or elastomeric bearings were used in order to
reduce the stiffness of the structure and the seismic forces, while dampers were used to
reduce the displacement of the deck and prevent impact with the abutments. Numerical
analyses performed with the aim to optimise the design and position of the devices,
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
15
highlighted the influence of the accelerogram characteristics, device properties, centering
capacity and SSI on the design forces and displacements.
The effects of various techniques on the seismic response of bridges were experimentally
and numerically investigated [DesRoches et al., 2003]. Simply supported and continuous
multi-span bridges with either steel or prestressed concrete girders, typical of Mid-
America were examined. It was found that continuity of the deck reduces the pounding
forces, while for simply supported bridges the critical elements are the columns.
Considering retrofit measures, numerical analyses pointed out that elastomeric bearings
result in about 50% decrease in ductility demand for the columns and in increase of the
displacement and pounding forces in the deck. Lead-rubber bearings had similar effect on
the ductility demand for columns and in addition decreased the displacements and forces
in the deck. Finally, when elastomeric bearings were combined with restrainer cables,
their results were reciprocally overpowered and the effectiveness was reduced.
Although various devices and techniques are available, and significant experience has
been gained from numerous past applications, there is still some uncertainty with regard
to the analysis method to be used and also concerning the damping characteristics of the
devices and the passage to the overall damping of the structure. For the above reasons, the
conceptual design of an isolated bridge is more art than science [Priestley et al., 1996].
2.5. DAMAGE ASSESSMENT
In the displacement-based procedures, demand-versus-capacity iterative verifications are
performed on the basis of different measures of deformation, such as drift [Calvi &
Kingsley, 1995], strain limit states related to top displacements [Priestley & Calvi, 1997;
Kowalsky, 2000; Paulay, 2002] and also chord rotation [Panagiotakos & Fardis, 1998].
For the quantification of structural damage, the most widely-used damage model is the
energy-based Park and Ang Damage Model [Park & Ang, 1985], that defines the damage
index by two parts, one taking into account the maximum deformations experienced and
the other accounting for the cycling effects. Concerning bridge structures and components
in particular, based on a database of experimental results, an attempt was made to
correlate different damage indices to performance levels and required repair [Hose et al.,
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
16
2000]. On the basis of cyclic tests on scaled circular bridge columns, a correlation of
damage limit states with visual observation has also been proposed [El-Bahy et al., 1999].
Cumulative damage, capacity and demand aspects and the influence of loading history are
to be taken into consideration [Krawinkler, 1996]. The effect of the loading history on the
failure mode and deformation capacity of members has been experimentally evaluated:
different failure modes and deformation capacities were observed, depending on the
loading history [Kunnath et al., 1997].
Assessment procedures exist in Europe, mainly for reinforced concrete and masonry
buildings [CNR, 1993; Grünthal, 1998] (for the USA see for example [FEMA, 1999]).
Alternative approaches have been proposed mainly in the USA, e.g. [Buckle & Friedland
1995; Taylor et al., 1997]. However, it has not yet been clarified which parameter is more
representative of deformations and which ultimate values correspond to each performance
level.
2.6. SEISMIC RETROFIT OF REINFORCED CONCRETE COLUMNS
Several techniques for seismic retrofit of structures exist, each one addressing a specific
seismic deficiency and requiring attention to details of the design and practical execution
[Fardis, 1998]. As far as bridges are concerned in particular, it is either desired to increase
the capacity, mainly by jacketing using concrete, steel or fibre-reinforced polymers (FRP)
or to decrease the seismic demand by isolation or dissipation devices. In certain cases it
might be desirable to improve the linkage between the bridge piers and the deck, e.g.
[Park et al., 1993]. It might be appropriate, instead, to add new members and rely on the
existing ones only for vertical load capacity, e.g. [Chua et al., 2001].
In this section the experimental verification of several retrofit techniques is presented.
The techniques comprise concrete, steel and FRP jacketing, which is aimed at increasing
the flexural and shear strength and the ductility capacity. The particular problem of
lapped splices is also discussed. Experimental results for retrofitted piers with hollow
cross-section are discussed in detail in a following chapter. The geometric and
mechanical properties of the specimens and the deformation capacities in terms of
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
17
displacement ductility and lateral drift are presented in Annex A and will be discussed in
the following chapter.
2.6.1. Retrofit by jacketing
Concrete jacketing is used to increase the ductility, as well as the shear and flexural
strength of components and is in general a low-cost intervention. Steel jacketing has the
same effects, but the cost varies depending on the amount of material used. FRP jackets
are effective in providing active or passive confinement, increasing the shear and flexural
strength and also the integrity of lapped splices. The materials are usually expensive, but
application is easier compared to the other techniques. The cost would increase for all
techniques in the case of low accessibility of the region in need of retrofit. An additional
advantage of FRP jackets is the limited change in stiffness, compared to concrete or steel
jackets [Priestley et al., 1996]. As a result, the dynamic characteristics of the structure are
not significantly altered and the retrofitted components do not attract higher seismic
loads.
Concrete jackets
Concrete jacketing was one of the first techniques used for seismic retrofit of columns. It
can be used to provide confinement, to increase the flexural or shear strength, or for a
combination of the above. Concrete jackets have been mainly used for rectangular
columns of buildings, but could also be applied on bridge piers.
Shotcrete jackets have been applied on repaired and strengthened short columns that
showed brittle failure due to shear [Bett et al., 1988]. Two different configurations were
considered for the jacket on the strengthened columns: additional vertical rebars only at
the corner, or distributed along the width of the pier. In all cases the vertical
reinforcement served only to support the additional horizontal reinforcement. The as-built
and repaired columns failed due to shear, whereas the strengthened columns failed due to
a combination of flexure and shear. The drift capacity was uδ = 2.0% and uδ = 2.5% for
the as-built and the retrofitted columns, respectively. The jackets did not have a
significant effect on the stiffness of the column, but almost doubled the strength. The
main effect of the jacket was to reduce the degradation of strength and stiffness with
cycling.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
18
The effectiveness of concrete jackets with additional vertical and horizontal
reinforcement was experimentally assessed [Ersoy et al., 1993]. The vertical rebars of the
jacket were welded on the existing ones and on plates at the top and bottom of the
column. As-built, repaired and strengthened specimens were tested. Failure of all
specimens was due to crushing of concrete and buckling of vertical reinforcement. The
stiffness of the strengthened columns was similar to the one of the as-built column, while
the stiffness of the repaired columns was equal to 75% of the stiffness of the as-built
column.
Concrete-jacketed columns with different amounts of horizontal reinforcement and
vertical reinforcement either distributed or concentrated in the corners were
experimentally tested [Rodriguez & Park, 1994]. The effect of the jackets was to increase
by almost three times the strength and stiffness of the as-built piers, as well as to reduce
the strength degradation with cycling. The two configurations for vertical reinforcement
did not show a different effect on the performance of the retrofitted elements. The jackets
with large amounts of horizontal reinforcement did not proportionally improve the
behaviour of the columns.
Steel jackets
Steel jackets have been extensively used for seismic upgrading of columns and bridge
piers. They are used mainly to increase the confinement and shear strength of existing
elements, without increasing the flexural strength. Confinement is also expected to
improve the behaviour of lapped splices. Steel jackets are placed around the existing pier
within the plastic hinge region and a gap is left between the existing pier and the jacket
and later filled with normal or expansive grout. A gap is left between the steel casing and
the foundation in order not to increase the flexural strength. The dimensions of the gap
above the foundation condition the length of the equivalent plastic hinge length.
Large-scale models of circular bridge piers retrofitted with steel jackets were
experimentally tested [Chai et al., 1991]. The retrofitted specimen showed stable
hysteretic response until large values of displacement ductility, uµ = 8, and drift, uδ = 6%,
while the as-built specimen experienced brittle failure for ductility uµ = 4 and lateral drift
uδ = 3%. A small increase in strength was observed, as well as a 10% increase in
stiffness.
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
19
An extensive experimental campaign on as-built and retrofitted rectangular and circular
bridge piers was performed [Priestley et al., 1994c]. Elliptical and circular steel jackets
were applied with the aim to increase the shear strength and confinement. The retrofitted
columns exhibited stable hysteretic response, with displacement ductility capacity uµ = 8
and drift capacity uδ = 4%. The pattern of inelastic deformation changed from
predominantly shear deformation to predominantly flexural deformation for the retrofitted
specimens. Steel jackets increased the stiffness of the columns by 30% and 64% for
circular and rectangular cross-sections, respectively, and the energy-dissipation capacity
by 150 times, in comparison to the as-built piers.
Small-scale rectangular bridge piers were tested until failure and then repaired with
circular steel jackets [Yang et al., 2000]. The retrofit increased the strength by 40% to
110%, depending on the axial load and concrete strength, and also the displacement
ductility capacity from uµ = 3.7 to uµ = 6. The increase in stiffness was found to depend
on the length of the jacket, namely full-height or limited in the plastic hinge length.
Analytical studies have shown that the stiffness increases with increasing thickness of the
jacket, aspect ratio of the pier and bond strength between jacket and column [Chai, 1996].
For an extreme case of good bond between the jacket and the column, large aspect ratio
and large jacket thickness, an increase of stiffness in the order of 150% was estimated.
FRP jackets
FRP jackets are used to increase strength and/or ductility. They can be applied to either
circular or rectangular columns. FRP jackets present the advantages of low weight, easy
application and low maintenance, in comparison to concrete and steel jackets. The
effectiveness of FRP jackets for wall-type bridge piers is also reported [Uemura, 2000].
The effects of strap thickness, spacing and type of fibre, namely glass and carbon, were
analytically studied [Saadatmanesh et al., 1994]. It was found that the jackets increase the
ductility, without significantly increasing the resistance. The improvement increases with
the thickness of the strap and decreases with the spacing of the strap and the concrete
strength. Piers with carbon-fibre reinforced polymer (CFRP) jackets were found to have
larger capacity of energy dissipation than piers with glass-fibre reinforced polymer
(GFRP) jackets. A drawback of the adopted analytical procedure is that constant, and not
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
20
increasing, confining stress with lateral dilation was considered for the FRP jacket. This
aspect will be further discussed in the following.
The effectiveness of GFRP jackets on circular bridge piers was experimentally
investigated [Saadatmanesh et al., 1996]. Both passive and active retrofit solutions were
examined. Passive retrofit consists in simply wrapping the FRP straps around the pier,
while for active retrofit, a gap is left between the pier and the jacket and then filled with
pressurised grout infill. The jackets increased the ductility from uµ = 4 to uµ = 6, as well
as the strength and the energy-dissipation capacity of the as-built piers. No significant
difference was observed in the behaviour of the piers with active or passive retrofit.
Flared columns with GFRP and CFRP jackets were experimentally assessed [Saiidi et al.,
2000]. The effect of the jackets was to increase the stiffness, ductility (from uµ = 5.4 to
uµ = 7.4 or uµ = 7.9 for CFRP and GFRP jackets, respectively) and strength (by almost
40%). No significant difference was observed between the responses of the specimens
retrofitted with GFRP or CFRP jackets.
Small-scale models of rectangular bridge piers were tested using the pseudodynamic
testing method [Chang, 2002]. As-built models were first tested and then repaired using
CFRP jackets. The jackets were successful in restoring and slightly increasing the original
strength, but managed to restore only 70% to 75% of the original stiffness.
The effectiveness of CFRP and GFRP jackets for the improvement of the behaviour of
RC columns subjected to accelerated corrosion was experimentally investigated [Bousias
et al., 2002]. The as-built specimens failed due to a combined flexural-shear mode and
showed low drift capacity, uδ = 2.8%, while the retrofitted specimens failed in a flexural
mode. The deformation capacity of the retrofitted specimens was significantly increased
until uδ = 5.1% and uδ = 7.5% in the strong and the weak directions, respectively. It was
observed that increasing the number of FRP layers above a certain limit, does not
improve the performance of the specimens and also that the improvement is not
proportional to the amount of FRP. This is due to the fact that other failure modes, e.g.
fracture of steel rebars, precede the failure of the FRP material.
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
21
2.6.2. Retrofit for enhancement of lapped splices
For construction convenience, starter bars were often spliced just above the foundation,
within the potential plastic hinge region. In addition, in structures designed prior to
modern seismic codes, the overlapping length is short and the amount of transverse
reinforcement low. This may cause loss of bond between spliced rebars and consequently
premature failure of members. Experimental studies suggest that bond failure results in
degradation of resistance and also that lapped splices above the base force the
deformation demand to concentrate at a very thin slice [Paulay, 1982; Chai et al., 1991;
Lynn et al., 1996]. Therefore, it has been proposed to rely on the strength of lapped
splices only until low levels of displacement ductility, µ = 3 [Priestley & Park, 1987].
One technique for retrofitting bridge piers with lapped splices within the plastic hinge
region is by applying external prestressed reinforcement. This technique was applied on
flexure-dominated circular columns and assessed experimentally [Coffman et al., 1993].
The retrofit resulted in no change of the column stiffness and in slight increase of the
strength and of the energy-dissipation capacity. It mainly resulted in a significant increase
of the total number of cycles before failure. Yielding of the longitudinal reinforcement
occurred within a small length adjacent to the footing.
It has been proposed to retrofit circular columns with lapped splices using steel jackets.
Experimental results showed that columns with lapped splices at the base exhibit small
ductility, uµ = 1.5, drift, uδ = 1.4%, and dissipation capacities and that their behaviour is
significantly improved by applying steel jacket within the overlapping length [Chai et al.,
1991]. The ductility capacity of the retrofitted pier was uµ = 7 and the drift capacity was
uδ = 5.2%. The curvature demand of the jacketed specimen was concentrated in the gap
between the jacket and the foundation. The retrofitted pier had almost twice the lateral
strength of the as-built specimen.
Cyclic tests on scaled models of grooved rectangular bridge columns retrofitted with steel
jackets showed similar performance of the specimens with circular or elliptical jackets as
well as normal or expansive grout between the jacket and column [Daudey & Filiatrault,
2000]. The jackets slightly increased the strength of the column. They significantly
increased the energy-dissipation capacity and the displacement ductility from uµ = 1.5 for
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
22
the as-built specimen to uµ = 6 for the retrofitted ones. The curvature demand in all tested
specimens was concentrated in the gap between the steel jacket and the foundation.
Larger gap between the jacket and the foundation was found to reduce by 10% the
stiffness increase and to reduce the concentration of stress on the vertical rebars.
An alternative approach is the relocation of the plastic hinge. This retrofit strategy
involves adding a reinforced concrete footing block in order to move the location of the
plastic hinge from the base of the column to the top of the lapped splices [Griezic et al.,
1996]. The moment capacity at the base must exceed the applied moment when yielding
occurs at the top of the lapped splices. When first yielding occurs, it is expected that it
would spread above and below the critical cross-section and thus result in a significant
plastic hinge length in the retrofitted column. Experimental assessment of the proposed
technique was performed, applying a steel jacket at the part above the added concrete
block in order to increase confinement. The retrofit increased the flexural strength by
about 75% and the displacement ductility from uµ = 2.6 to uµ = 6.6. It also increased the
capacity of energy dissipation.
GFRP jackets applied on circular columns with lapped splices were assessed, considering
both passive and active retrofit [Saadatmanesh et al., 1996]. The jackets were found to
increase the strength and energy-dissipation capacity of the column. While the as-built
specimen failed at low ductility, uµ = 1.5, the retrofitted ones failed at uµ = 6. No
significant difference was observed between the performance of the specimens retrofitted
with the active and passive jacket.
The effectiveness of prefabricated FRP jackets for enhancement of lapped splices was
experimentally investigated [Xiao & Ma, 1997]. The failure of as-built circular columns
was brittle, for low displacement ductility, uµ ≈ 1, before reaching the design flexural
resistance. For the repaired and strengthened specimens, a significant increase in
displacement ductility ( uµ = 4 and uµ = 6, respectively) was achieved with prefabricated
jackets, although loss of bond initiated at high levels of displacement. The repaired
specimen managed to develop the nominal flexural strength, while the strength of the
strengthened specimen almost doubled, compared to the as-built specimen.
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
23
2.7. ANALYSIS AND MODELLING
Concerning the analysis of RC bridges, a comprehensive review of the various methods
can be found in [Priestley et al., 1996]. The effects of stiffness, damping, ductility,
regularity, simplified models, response spectrum or time-history analysis for the design of
RC bridges are discussed in [Flesch & Klatzer, 1995]. Assumptions related to material
constitutive laws, definition of limit states, mesh refinement, inclusion of the foundation,
soil and deck have been proved to significantly affect the analytical results of interest in
the design of new and the assessment of existing bridges [Elnashai & McClure, 1996].
Different modelling possibilities for the seismic analysis of bridge structures, namely:
elastic single-mode spectral method, elastic multi-mode spectral method and inelastic
time-history analysis, depending on the seismic hazard, the importance of the bridge and
the structural regularity were found to result in significant differences in terms of design
forces and displacements [Fishinger et al., 1997].
The standard method of analysis for normal bridges according to EC8-2 [CEN, 2002] is
the response spectrum method. When the dynamic behaviour of the bridge can be
sufficiently approximated by a SDOF model, the equivalent static seismic forces can be
derived from the inertia forces corresponding to the fundamental natural period of the
structure. Power spectrum and time series analysis are also permitted. Non-linear time
history analysis may be used in combination with the standard response spectrum analysis
and in general, with the exception of isolated or irregular bridges, it may not be used to
relax the demands calculated with the standard method. Specific objectives and
requirements are demanded for non-linear time history analysis. For irregular bridges, a
combination of pushover and equivalent linear analysis may be performed.
Similarly, linear response spectrum analysis, both single-mode and multi-mode, and non-
linear time history analysis is designated in the Japanese [JSCE, 1996], New Zealand
[SNZ, 1995] and USA [Caltrans, 1999] seismic codes, depending on the structural
configuration and simplicity of the bridge. More refined analysis methods are prescribed
with increasing complexity and importance of the bridge structure.
Modern seismic codes contain detailing provisions that protect the members against
undesirable failure modes. Capacity design procedures dictate a controlled failure mode,
with inelastic deformation (plastic hinges) developing at desired parts of the structure, by
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
24
designing the remaining sections for resistance calculated on the basis of the flexural
over-strength of the plastic hinge regions. Transverse reinforcement is detailed so that it
provides sufficient confinement of the concrete core and support of vertical reinforcement
bars against buckling, whereas the overlapping length is enough to ensure transfer of
stresses between spliced rebars. Lack of such provisions may instigate phenomena, such
as significant shear deformation, tension shift, loss of bond between steel and surrounding
concrete and premature buckling, which are difficult to model. As a result, simplified
models for the structural components, e.g. fibre models, are not always successful in
representing the whole range of phenomena that affect the failure mode, deformation
capacity and hysteretic behaviour of elements. For this reason, it is often unavoidable to
resort to modelling with the finite element method using extremely elaborated meshes and
constitutive laws for the materials (concrete, reinforcement bars and steel-concrete
interface). The performance of the aforementioned numerical approaches for the
modelling of as-built and retrofitted bridge piers will be discussed in a following chapter.
Existing bridge piers often have lapped splices within the critical zones, which affect the
structural response both in terms of strength and deformation. Therefore, the behaviour of
the steel to concrete interface should be appropriately considered in the analysis of these
elements. Extensive theoretical and experimental research work has been performed in
the past, before formulating constitutive laws for steel-to-concrete bond. The factors that
affect the bond strength are the steel stress, concrete strength, side and bottom cover, bar
spacing, amount of transverse reinforcement and bond condition [Tassios, 1979;
Eligehausen et al., 1983; Eligehausen & Balázs, 1993]. The stress history and level of
force are also important for the bond behaviour [Bresler & Bertero, 1968; Balázs, 1991].
Experiments showed the effect of the loading rate [Chung & Shah, 1989] and the effect of
repeated and reversed load [Balázs, 1991] on steel-to-concrete bond. Finally,
experimental results on members with noncontact splices [Sagan et al., 1991] verified the
effect of cycling, transverse reinforcement and concrete compression on the lap splice
length.
Similar difficulties, often to a greater extent, exist for the numerical modelling of
elements retrofitted with FRP jackets. A major concern is the confining effect of jacketing
on the properties of concrete. The existing models proposed for confinement provided by
steel stirrups are not applicable to FRP-confined concrete. The complication arises from
OPEN ISSUES IN SEISMIC DESIGN AND ASSESSMENT OF BRIDGES – A LITERATURE REVIEW
25
the fact that, while steel jackets apply constant confinement stress after yielding, FRP
jackets, due to the linear behaviour of the material until failure, provide increasing lateral
pressure with increasing dilation of the concrete core. Various empirical constitutive laws
have been proposed, based on experimental results, e.g. [Fardis & Khalili, 1981; Samaan
et al., 1998; Kawashima et al., 2000a]. These empirical laws do not perform well in cases
where the materials are significantly different from the ones used for the calibration of
each of them. A recently-proposed theoretical constitutive law for composite-confined
concrete [Spoelstra & Monti, 1999] overcomes this limitation. The same important is the
bond between FRP strips and concrete. Loss of adhesion may initiate premature failure of
the retrofitted member, e.g. [Arduini & Nanni, 1997]. The proposed models, their
inconsistencies and the implications on the modelling of retrofitted elements will be
discussed in depth in a following chapter.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
26
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
27
3. EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH
SEISMIC DEFICIENCIES
3.1. INTRODUCTION
When designing a bridge pier, rectangular, octagonal, circular or wall-type solid cross-
sections are often used. In the case of tall piers, it is desirable to reduce the mass of the
pier and consequently the seismic loads it has to resist. In the USA the trend is to use
solid sections and to reduce the cross-sectional dimensions with height, whereas only a
small percentage (in the order of 3%) of the existing bridge piers have hollow cross-
sections [Poston et al., 1986]. In contrast, piers with rectangular hollow cross-section are
commonly used in Japan and Europe for highway bridges that cross deep valleys [Hooks
et al., 1997]. As most of the existing bridges were designed before the modern seismic
codes came into practice, there is the need to assess the seismic performance of bridge
piers with hollow cross-section.
Modern codes do not distinguish between piers with solid and hollow cross-section,
probably because of limited knowledge. On the other hand, it is recognised that the
rational assessment of existing structures might require the revision of limit states and
methods of analysis [Calvi et al., 2000]. Indeed, recent research [Rasulo et al., 2002;
Hines et al., 2002a; Hines et al., 2002b] originates from the questions concerning the
shear resistance and the estimation of the deformation capacity of piers with hollow
cross-section and very large dimensions.
This chapter deals with the experimental results from cyclic quasi-static tests on large-
scale models of one short and one tall bridge pier performed at the ELSA laboratory of
the Joint Research Centre. The original piers belong to an existing highway bridge,
schematically shown in Figure 3.1. A view of the real structure is given in Figure C.1.
The bridge is situated in Austria and was designed in 1975. One tall pier, termed A40,
and one short pier, termed A70, were chosen in order to study components with different
aspect ratios and detailing. The main characteristics are the presence of lapped splices
with short overlapping length within the potential plastic hinge region of the short pier
and the premature termination of longitudinal reinforcement in the tall pier.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
28
Figure 3.1. Talübergang Warth Bridge, Austria
The objective of the tests was twofold. The first scope was the assessment of the cyclic
behaviour of existing bridge piers with hollow cross-section. The second objective was
the calibration of non-linear numerical models for the substructured piers during the
pseudodynamic tests that followed. The pseudodynamic tests are presented in Chapter 4,
while the numerical modelling is further discussed in Chapter 5.
The design of the specimens is presented first, along with the instrumentation. The
experimental results are presented in terms of damage evolution, force-displacement
hysteretic curves and contribution of shear to the total displacement. The definition of
yield displacement is discussed in particular. A comparison to empirical expressions for
the prediction of yield and ultimate displacement highlights the importance of the
estimation of the equivalent plastic hinge length. The significance of seismic detailing is
evidenced through the comparison of the behaviour of the short pier to that of a squat pier
designed according to Eurocode 8 (EC8).
A literature review of experimental results on hollow bridge piers is also presented. An
attempt is made to qualitatively identify the effect of mechanical and geometrical
parameters on the deformation capacity of such piers. Finally, a preliminary estimation of
performance limits is based on the experimental results available in literature. This
highlights the differences between piers designed before and after the introduction of
modern seismic codes and also the effect of failure mode on the deformation capacity of
seismic-deficient piers.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
29
3.2. BIBLIOGRAPHIC RESEARCH
3.2.1. Experimental assessment of piers with seismic design
Experimental assessment of scaled (1:2.5) models of bridge piers with square cross-
section was performed with the axial load and the hoop spacing as the main variables
[Mander, 1984]. The specimens with closely spaced stirrups showed stable hysteretic
response until displacement ductility uµ = 8 and drift uδ = 3.5% for low axial load,
cN / Afν = = 0.1, and uµ = 6 and uδ = 2.4% for medium axial load, ν = 0.3. The specimen
with large hoop spacing and high axial load, ν = 0.5, showed brittle failure for uµ = 4 and
uδ = 1.4%, whereas the one with medium axial load, ν = 0.3, presented significant loss of
resistance only for uµ = 8 and uδ = 3.3%.
Six piers with circular cross-section, representative of offshore concrete platform legs,
were tested in order to examine different configurations of transverse reinforcement
[Whittaker et al., 1987]. Transverse steel was designed to resist shear demand and various
configurations for confinement reinforcement were studied. The specimens with inner
and outer spirals and closed hoops showed stable response until drift uδ = 4.2%. The
specimens with closed hoops and only outer spirals showed brittle behaviour and smaller
drift capacity, uδ = 1.6%. Finally, the specimens with larger amount of confinement
reinforcement showed ductile behaviour and failed at drift uδ = 2.4%. These results
highlight the importance of correct detailing for the ductile behaviour of piers with
hollow cross-sections.
Six models of circular hollow piers with one layer of longitudinal and spiral
reinforcement placed near the outside face, were tested under cyclic loading [Zahn et al.,
1990]. The examined parameters were the axial load and the thickness of the walls. The
specimens failed in a brittle manner (displacement ductility uµ = 2.4) when the
compression zone comprised the internal face, which was not properly confined. When
the neutral axis was near the internal face, or inside the wall, stable ductile behaviour was
observed until ductility uµ = 7.5. It was concluded that, for this geometry of the cross-
section and reinforcement, ductile behaviour can be achieved for low axial load and for
low amounts of longitudinal reinforcement.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
30
Within a large experimental campaign in support of EC8, seven large-scale (1:2.5)
models of rectangular hollow bridge piers were tested under cyclic and seismic loading
[Pinto et al., 1994; Pinto et al., 1996]. The piers were designed according to EC8-2 [CEN,
1994] and therefore conformed to seismic detailing. Piers with different aspect ratios and
tie configurations were tested. All the piers showed stable hysteretic behaviour until
ductility uµ > 5.5 and drift uδ > 2.3%, verifying the detailing criteria of EC8.
Three large-size circular hollow piers with vertical reinforcement only in the external face
were tested under cyclic loading in order to study flexural and shear failure modes and
also the effect of axial load [Ranzo & Priestley, 2000]. The pier with flexural failure
mode exhibited large capacities of displacement ductility, uµ = 6, and drift, uδ = 2.9%.
The specimen with shear-dominated response failed at ductility uµ = 3.5 and drift uδ =
2.5%, while the specimen with axial load ν = 0.15, failed after concrete spalled for
ductility uµ = 2.0 and drift uδ = 1.5%. A comparison of the experimental failure modes
and shear strength to theoretical expressions showed good agreement for the case of these
well-designed piers.
Small-scale models of hollow piers were tested in order to investigate the effect of the
aspect ratio and the amount of transverse reinforcement [Takahashi & Iemura, 2000]. The
tall specimens failed in flexure, whereas the short ones failed in shear. The tall specimen
with closely spaced horizontal reinforcement and crossties in the web, showed stable
behaviour and large drift capacity, uδ = 4.0%. The specimens with small amount of
transverse reinforcement showed rapid loss of resistance and drift uδ = 1.6% or uδ = 3.1%
for short (L/d = 2) and tall (L/d = 4) piers, respectively.
Two scaled specimens (1:3.5) of a rectangular hollow bridge pier were tested in order to
study the effect of lapped splices [Kim et al., 2001]. The specimen with lapped splices
showed stable response until uδ = 4.8%, when failure occurred due to loss of bond and
was associated with rapid loss of resistance. The specimen with continuous reinforcement
showed large deformation capacity, uδ = 7.8%. Failure was due to spalling of concrete
and buckling of steel.
A large number of square hollow bridge piers were tested with the aim of studying the
effect of the concrete compressive strength, axial load, aspect ratio and amount of shear
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
31
reinforcement [Mo et al., 2001; Mo & Nien; 2002; Yeh et al., 2002a; Yeh et al., 2002b].
All the specimens were detailed with crossties that provided protection against buckling
for all vertical rebars. The specimens with high-strength concrete had 20% higher
strength, but also 10% smaller deformation capacity. The effect of axial force was to
increase the strength and decrease the deformation capacity. On the other hand, larger
amount of horizontal reinforcement resulted in increase of both strength and ductility.
The maximum attained drift was uδ = 6.5% for a square pier with normal strength
concrete, low axial load, ν = 0.08, and adequate shear reinforcement. The smallest
attained drift was uδ = 2.1% for a pier with normal concrete, low axial load, ν = 0.08, and
inadequate shear reinforcement.
Three full-scale circular bridge piers with hollow cross-section were tested in order to
study the effect of shear and lapped splices [Yeh et al., 2001]. The piers were designed
according to modern codes, with the exception of one specimen that was provided with
insufficient horizontal reinforcement and a specimen that was detailed with lapped splices
at the base. The specimen with the lapped splices did not have crossties, in contrast to the
other two specimens. The specimen with seismic detailing showed stable hysteretic
curves until drift uδ = 5.9% and failed in flexure. The one with lapped splices failed
because of loss of bond at drift uδ = 1.6%. Only a few, relatively wide cracks were
observed and a plastic hinge did not fully develop. The pier with insufficient shear
reinforcement failed at uδ = 2.9% due to the combination of shear and flexure.
Eight specimens of rectangular hollow piers were tested with the aim to study the effect
of different tie configurations and concrete compressive strength [Mo et al., 2003]. All
specimens had sufficient shear reinforcement and failed after buckling or rupture of
longitudinal rebars at lateral drift ranging from uδ = 1.9% to uδ = 2.2%. The specimens
with different tie configurations showed similar performance. The spacing of stirrups
significantly affected the degradation after maximum load: specimens with double the tie
spacing had almost half the displacement ductility.
Scaled (1:4) bridge piers with hollow rectangular cross-section and highly confined
boundary element were tested [Hines et al., 2002a; b]. Based on the experimental data on
flexure-dominated specimens, an empirical expression for the calculation of the
equivalent plastic hinge length, that considers the effect of tension shift, was proposed.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
32
Tests on shear-dominated specimens allowed updating the existing shear behaviour
models in order to incorporate the effect of flexure, which results in non-parallel shear
cracks. An empirical relation between flexural and shear deformation of piers with
elongated hollow cross-section was also proposed. It was concluded that the force-
deflection behaviour of such piers could be modelled with reasonable accuracy based on
moment-curvature analyses, with assumed plastic hinge lengths, conservative steel strain
limit states and assumed shear displacements that are proportional to the flexural
displacements.
3.2.2. Experimental assessment of piers without seismic design
Concerning existing bridge piers with hollow cross-section, small-scale specimens have
been mainly tested. The differences between bridge piers designed according to early and
modern seismic codes lay mainly in the amount of vertical and horizontal reinforcement,
the presence of confinement reinforcement in the form of closed hoops or crossties, the
splicing of rebars within the potential plastic hinge zone and the premature termination of
longitudinal rebars with insufficient development length.
Scaled (1:5) models of a bridge pier were tested considering different aspect ratios and
development length for the curtailed vertical reinforcement [Kawashima et al., 1990]. All
the specimens with aspect ratio L/d = 5.4 failed in flexure at the base, with the exception
of the one with termination of vertical rebars at mid-height, that failed in flexure at the
cut-off. The specimen with aspect ratio L/d = 9.9 and cut-off at mid-height showed
flexural damage above the critical cross-section and in the end failed in shear. The
specimens with L/d = 9.9 and anchorage of the terminated rebars equal to either the pier
width or half the pier width, both failed in flexure at the base.
A scaled (1:8) model of an existing bridge pier was tested under cyclic load [Huang et al.,
1997]. The model exhibited stable behaviour until ductility uµ = 4 and drift uδ = 1.8%,
after which failure occurred. The failure mode was flexure-dominated, indicated by
spalling of concrete, buckling and rupture of longitudinal reinforcement at the base.
The seismic behaviour of scaled (1:4) hollow piers, representative of existing Italian
bridge structures, was studied [Calvi et al., 2000; Rasulo et al., 2002]. The variables were
the collapse mode, the axial load and the geometry of the specimens, namely vertical
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
33
reinforcement in one or both of the faces of the cross-section. Specimens with lapped
splices at the base and with 50% reduction of vertical reinforcement at 1/3 of the height
were also tested. Because of the significant effect of shear on the response of the piers, all
specimens failed for low levels of lateral drift, ranging from uδ = 0.8% to uδ = 1.9%.
Comparing the experimental results to theoretical expressions for the strength of
elements, it was concluded that while the response of piers with flexure-dominated
behaviour can be predicted, further study is needed for the case of interaction between
shear and flexure.
From the above discussion some major conclusions are drawn. The presence of closed
stirrups provides adequate confinement to the concrete core and protection of vertical
reinforcement against buckling, resulting in stable response and large deformation
capacity. In addition, the premature termination of vertical reinforcement combined with
the tension shift phenomenon can be the cause of undesirable failure modes. Concerning
modelling, the combination of flexure and shear, as well as the spread of plasticity,
should be appropriately considered.
The mechanical and geometric properties of all the specimens are presented in Appendix
A, along with the deformation capacity in terms of displacement ductility and drift ratio.
Although quite a large scatter in the values of deformation capacity is observed, the
difference between seismic-deficient and retrofitted or code-designed specimens should
be highlighted. The deformation capacity of piers with hollow cross-section is further
discussed in a separate section at the end of the chapter.
3.3. DESIGN OF THE TEST MODELS
3.3.1. Scaling of the specimens
The scaling factor, λ = 2.5, was chosen in order to allow for testing within the capacity of
the laboratory, to facilitate the construction and to use normal concrete and bar diameters.
The similitude law dictates that the stresses in the model, Mσ , are equal to the stresses of
the prototype, Pσ . Then, the similitude relations for other quantities of interest are
presented in Table 3.1. The longitudinal reinforcement of the shaft of the mock-up was
detailed to obtain the flexural capacity corresponding to the scaling of the prototype,
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
34
while representing as best as possible the interaction between concrete and steel
reinforcing bars. The last condition (overlapping lenght) requires to keep the steel rebar
diameters close to those of the prototype, while at the same time keeping the spacing,
overlapping and anchorage lengths in correspondence to the steel reinforcement diameters
used in both the model and the prototype. Splicing of steel rebars was kept in the model
as close as possible to the prototype: the overlapping lengths were scaled in proportion to
the rebar diameters and the location and distribution of splicing was maintained
throughout the height of the shaft of the model. Transverse reinforcement was detailed in
the model to comply with both the shear capacity and proper spacing in relation to the
longitudinal reinforcement. Concerning shear capacity, the same percentage of transverse
reinforcement was maintained for both the prototype and the model.
3.3.2. Geometry of the specimens
The scaled specimens had a rectangular hollow cross-section with external dimensions
2.74x1.02 m, see Figure 3.2c. The width of the flange and the web was 0.21 m and 0.17
m, respectively. The concrete cover was chosen equal to 0.015 m for easiness of
construction. A rigid steel cap was attached on top of the concrete shafts in order to apply
the horizontal and vertical loads. The total height of the short pier was 6.5 m, L/d = 2.4,
and the total height of the tall pier was 14.00 m, L/d = 5.1. Each model had a foundation
block with dimensions 5.5x2.5x1.2 m, see Figure 3.2. The complete series of construction
drawings for the two piers is given in Appendix B.
Table 3.1. Similitude relationship between the full-scale prototype (P) and the constructed
model (M)
Quantity Label Relationship Quantity Label Relationship
Length L P ML L= λ Acceleration a 1P Ma a−= λ
Area A 2P MA A= λ Force F 2
P MF F= λ
Volume V 2P MV V= λ Time t P Mt t= λ
Mass M 2P MM M= λ Strain ε P Mε = ε
Velocity v P Mv v= λ Stress σ P Mσ = σ
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
35
Figure 3.2. Geometry of the scaled models of the short (a) and the tall (b) pier and typical
cross-section (c)
Concrete class C35/45 (nominal characteristic cylinder strength ckf = 35 MPa) and steel
class S500 (nominal characteristic yield strength o,2kf = 500 MPa) as defined in Eurocode
2 (EC2) [CEN, 2002], were assumed in accordance to the materials specified for the
prototype pier. Resulting from compression tests on cubic specimens, the average
concrete strength in compression was cf = 38.9 MPa for the short pier and cf = 51.6 MPa
for the tall pier. Standard tests on specimens of reinforcement bars resulted in yield stress
yf = 540.2 MPa, yf = 543.1 MPa and yf = 546.4 MPa, ultimate stress tf = 595.8 MPa, tf =
632.6 MPa and tf = 660.2 MPa and strain at maximum stress uε = 0.149, uε = 0.115 and
uε = 0.065 for F 6, F 10 and F 12 rebars, respectively. The material properties are grouped
in Table 3.2.
The longitudinal reinforcement of the short pier consisted of F 10 deformed bars with
volumetric ratio sρ = 0.4% and the transverse reinforcement consisted of one F 6 bar at
each face of the flange and the web with volumetric ratio wρ = 0.09%. The starter bars
were terminated above the base block and the vertical rebars of the pier shaft were spliced
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
36
just above the base cross-section and within the potential plastic hinge region. The
overlapping length was 38F , 43F and 50F for different groups of rebars. As seen in
Figure 3.2c and also in Appendix B, no stirrups or closed hoops were placed, according to
the original design of the pier.
The longitudinal reinforcement of the tall pier consisted of F 12 and F 16 bars with
volumetric ratio sρ = 0.7% at the base. The transverse reinforcement consisted of one F 6
bar at each face of the flange and the web with volumetric ratio wρ = 0.09%. The starter
bars were terminated above the base block and the rebars of the pier shaft were spliced
over a length equal to 38F , 47F , 63F and 100F for different groups of rebars. Another
important characteristic of the longitudinal reinforcement, common to piers designed in
the same period, is the bar cut-off, as a result of the linear design and of the absence of
capacity design. At the height of 3.5 m (25% of the total height) from the base of the
scaled specimen (3.5x2.5 = 8.75 m for the prototype) the total amount of longitudinal
reinforcement is reduced by almost 50%; the reduction is 30% for the flange
reinforcement. As a consequence and in combination with the tension shift phenomenon,
discussed further on, the moment capacity of this cross-section is reached before the
flexural resistance of the base cross-section. This resulted in location of the critical cross-
section above the bar cut-off.
Table 3.3 compares the properties of the tested specimens to the provisions of seismic
codes for new bridges in Europe [CEN, 2000; CEN, 2002], the USA [AASHTO, 1995;
ATC, 1996; Caltrans, 1999], New Zealand [SNZ, 1995] and Italy [Ordinanza 3274, 2003]
for the design values of the material properties. Although all codes forbid the splicing of
rebars within the plastic hinge zone, the minimum overlapping length, o,minl , is presented
for comparison. According to EC2 [CEN, 2002], the design lap length, ol , is
Table 3.2. Material properties of the specimens (average values)
Steel Concrete
F (mm) yf (MPa) tf (MPa) uε (%) cf (cubic) (MPa)
6 540.2 595.8 14.9 Short, A70 38.9
10 543.1 632.6 11.5 Tall, A40 51.6
12 546.4 660.2 6.5
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
37
Table 3.3. Mechanical properties of the specimens and seismic code requirements
mins s,minρ (%) w,minρ (%) o,minl uµ
Short pier 20F 0.4 0.09 38F -50F 3.2
Tall pier 20F 0.7 0.09 38F -100F 2.3
Europe 6F 0.2 0.4 44F 3.51/32
U.S.A. 6F 1.0 0.4 43F 4
N. Zealand 6F 0.8 0.3 45F 63/34
Italy 6F 0.2 0.4 44F 3.51/2.42
1 A40 (L/d = 5.1), 2 A70 (L/d = 2.4), 3 plastic hinge above ground level, 4 not easily
accessible plastic hinge
o 1 2 3 4 5 b s,req s,provl a a a a a l A / A= (3.1)
where 1a , 2a , 3a , 4a and 5a are coefficients considering the effect of confinement by the
concrete cover, non-welded and welded transverse reinforcement, transverse pressure and
percentage of lapped rebars at the same cross-section, s,reqA and s,provA are respectively
the areas of required and provided vertical reinforcement and bl is the basic anchorage
length
b yd bdl ( / 4)f / f= Φ (3.2)
where Φ is the bar diameter, ydf is the design yield stress of steel and bdf is the design
value of the ultimate bond stress. According to the ATC-32 Report [ATC, 1996] the lap
length is
'b yd cl 0.04 f / f= Φ (3.3)
where ydf is the yield stress of steel (in psi) and 'cf is the concrete stress (in psi). In the
New Zealand Standards a similar expression is used for the basic development length of
flexural reinforcement in tension
'b a y cl 0.5 f / f= α Φ (3.4)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
38
where aα is a parameter considering the position of the rebar in the cross-section, yf and
'cf are measured in MPa. A different expression is used for diameters bigger than 32 mm.
All seismic codes demand the use of closed hoops or crossties for confinement of
concrete and protection of vertical reinforcement against buckling. Both piers fall short of
the requirements for seismic detailing. The most noteworthy difference concerns the
amount, wρ , and spacing, s, of transverse reinforcement that control the seismic
performance of members.
3.3.3. Test set-up and instrumentation
The horizontal displacement was applied by means of two hydraulic actuators of 1 MN
capacity each. The actuators were attached in one end to the laboratory reaction wall and
in the other to a rigid steel cap on top of the specimen. No torsion was allowed. Each
piston was equipped with a load cell and a displacement transducer. In addition,
displacement transducers mounted on an independent reference frame measured the
displacement of each horizontal actuator. The axial load was applied by means of 8 post-
tensioned rods anchored in the base block and attached to hydraulic actuators, positioned
on the steel cap, at the other end. The vertical actuators were controlled to apply constant
axial force during the test. A general view of the piers in the laboratory is given in Figure
C.2.
The instrumentation of the specimens consisted of a set of displacement transducers and
inclinometers, see Figure 3.3. The displacement transducers were arranged as a truss on
one face of the piers. Vertical displacement transducers placed in correspondence to the
flanges, shown in red in Figure 3.3, were used to measure the average slice rotation.
Diagonal displacement transducers, shown in green in Figure 3.3, were used to measure
the shear deformation of the slices. Additional horizontal displacement transducers were
used to form a statically indeterminate truss. Using the readings of the transducers,
relative displacements of the members of the truss, the absolute displacement of the nodes
of the truss can be calculated. More transducers were used in the lower part of the piers,
where significant deformation was expected, and less in the upper part, where elastic
behaviour was expected.
The vertical transducers were used to calculate the rotation, iΘ , of each slice
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
39
Figure 3.3. Instrumentation of the short (a) and the tall (b) pier
i l r i c( ) / LΘ = ∆ − ∆ (3.5)
where l∆ and r∆ are the vertical displacements measured at the ends of the slice and cL
is the horizontal distance between the transducers’ axes. Additional vertical displacement
transducers were placed within the first 1 m from the base of the pier on both flanges. The
average slice curvature was then computed by dividing the slice rotation by the slice
length, iL . The total rotation at the top of slice i was computed as the sum of the slice
rotations
i kθ = Θ∑ (3.6)
The flexural displacement at the top of each slice, ix , was then calculated as
i i 1 i ix x L−= + θ (3.7)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
40
For the short pier, A70, the total absolute displacement at the top of each slice was
obtained using the measurements of the horizontal displacement transducers connected to
a rigid bar hinged at both the base and the top; the top followed the pier top displacement.
The shear displacement was finally computed as the difference between the total and
flexural displacement. The truss of horizontal, vertical and diagonal displacement
transducers placed on the web throughout the height of the specimens was also used to
calculate the absolute horizontal and vertical displacement of its nodes. As both systems
gave consistent results, only the truss of transducers was used for the tall pier, A40.
A set of inclinometers, shown as black squares in Figure 3.3, was provided along the
height of the bridge pier. Additional inclinometers were placed along the length of the
second slice to measure the deformation in the web.
3.4. CYCLIC TEST ON A MODEL OF A SHORT BRIDGE PIER
3.4.1. Experimental results
A constant axial load, N = 3820 kN, corresponding to a normalised axial load ν = 0.09,
was imposed on the top of the specimen. The displacement history consisted of one cycle
of 2 mm, one cycle of 9 mm, two cycles of 27 mm, two cycles of 56 mm and one cycle of
100 mm in the strong direction.
Force-displacement diagram and observed damage
The force-displacement diagram is presented in Figure 3.4. As explained in the following,
a trilinear skeleton curve is used to approximate the experimental curve. Based on the
trilinear approximation, the yield displacement is identified as y,tu = 0.025 m and the
displacement ductility is u,tµ =3.2. The drift capacity was uδ = 1.5%.
For the first cycles of displacement 27 mm, flexural cracks appeared in the flanges within
the first 0.5 m of the pier height. With increasing displacement a crack appeared at the
interface with the foundation block and extended throughout the flange. A horizontal
crack appeared above the lapped splices at 0.5 m from the base. For the following cycles
some diagonal cracks appeared and the existing flexural ones extended. Spalling of
concrete was observed at the corners within the first 0.1m from the base, corresponding to
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
41
the spacing of the horizontal reinforcement. Finally, failure of the specimen was attained
when the vertical rebars at the base collapsed after buckling in previous cycles. The
failure mode was due to flexure, with some evidence of loss of bond between concrete
and steel. Figures 3.5 and C.3 present the crack pattern after the end of the test: green
colour corresponds to the cycles with amplitude 27 mm, whereas red colour corresponds
to the cycles with amplitude 56 mm and 100 mm. In is noted that the direction of testing
was in the East-West direction (strong direction of the cross-section).
-1500
-1000
-500
0
500
1000
1500
-0.10 -0.05 0.00 0.05 0.10
Displacement (m)
Forc
e (k
N)
Figure 3.4. Cyclic test on the short pier: force-displacement curve
Figure 3.5. Cyclic test on the short pier: crack pattern at the end of the test
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
42
On the definition of yield displacement
Displacement ductility is often used as a measure of the deformation capacity of
structures and elements. The value of ductility strongly depends on the yield and ultimate
displacements, which, in turn, are difficult to define. Indeed, the term ‘nominal’ yield
seems more appropriate. According to EC8-1 [CEN, 2003a], the global force-
displacement curve for buildings can be approximated by an idealised elasto-plastic
relationship: the yield force is equal to the force at the formation of the plastic mechanism
and the areas below the idealised and actual force-displacement curves are equal. For the
case of bridges, EC8-2 [CEN, 2002] assumes a bilinear equivalent curve, which best
approximates the actual force-displacement curve, and has an elastic stiffness equal to the
secant stiffness at the theoretical yield point.
Several definitions have been proposed for the yield displacement based on experimental
results, considering always a bilinear equivalent system [Park, 1989]. The yield
displacement can be the displacement when yielding first occurs at the system, the yield
displacement of an elasto-plastic system with the same elastic stiffness and ultimate load
as the real system, the yield displacement of an equivalent system with the same energy
dissipation as the real system or the yield displacement of an equivalent system with
reduced elastic stiffness. Following a similar proposal [Priestley et al., 1996], the yield
displacement is defined by extrapolating the elastic response up to the strength of the
structure or component.
An alternative approach is given in the ATC Report on Seismic Evaluation and Retrofit of
Concrete Buildings [ATC, 1996]. The value of the post-elastic stiffness, sK , is
considered to be an average stiffness in the range in which the structure strength has
levelled off. The yield force, yF , is defined by the intersection of the sK and eK lines,
where eK is the elastic stiffness. Then, the effective stiffness is a secant line passing
through the point of the experimental curve corresponding to 0.6 yF . This procedure
requires trial and error iterations.
An equivalent elastic-perfectly plastic system can be considered, where the initial
stiffness is calculated at the point of intersection of the line from the origin to 75% of the
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
43
ultimate load with the actual force-displacement curve [Elnashai & McClure, 1995]. The
plastic load of the equivalent system is the ultimate load of the inelastic system.
Using experimental data, the yield displacement can be calculated by extrapolating a
straight line from the origin through the moment-displacement point at 0.75 iM to the
theoretical flexural strength, iM [Priestley & Park, 1987; Sommer, 2000].
Alternatively, the bilinear curve is determined by considering the same post-yielding
stiffness and equal energy to failure [Reinhorn, 1997]. The post-elastic stiffness is again
approximated as a line representative of the actual force-displacement curve.
The ultimate displacement is defined as the post-peak displacement when the load
carrying capacity is reduced by more than 20%, or when buckling or rupture of steel
occurs, whichever is smaller [Park, 1989].
Although it is recognised that the initial stiffness, the yielding level and the post-yielding
stiffness are the important parameters that characterise the inelastic behaviour, no precise
guidelines are given for the estimation of these parameters. A unique definition for the
yield displacement does not exist: the post-yield stiffness is approximated by judgment
and the yield force varies from 60% to 100% of the member strength. Values ranging
from 67% to 78% have been proposed for reinforced concrete (RC) shear wall elements,
depending on the geometry of the cross-section [Paulay, 2002].
The standard definition of ductility was found not to accurately describe the performance
of brittle members (namely coupling beams) and in certain cases to be misleading, while
an energy-based ductility was more appropriate. The energy-based ductility is the ratio of
the ‘ultimate’ to the ‘yield’ energy of the system. The ‘ultimate’ energy is equal to the
area of the last force-displacement semi-cycle at failure and the ‘yield’ energy is equal to
the area of the first semi-cycle at yielding. It was then proposed to describe the
performance of elements through a combination of ductility criteria [Tassios et al., 1996].
It is important to notice that the aforementioned definitions consider a priori an equivalent
bilinear system. In addition, these procedures are design-oriented and therefore, the
strongly non-linear response of elements from the onset of cracking until the development
of a mechanism is of little interest for the calculation of the design forces and
displacements. Looking at experimental data, a bilinear equivalent system is realistic for
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
44
specimens with low height-to-depth ratio, often d/b = 1, and vertical reinforcement
concentrated near the most stressed fibre. For the piers presented herein, with elongated
cross-section, d/b = 2.7, and distributed reinforcement in the web, see Figure 3.2c, a
trilinear envelope is more realistic.
Consider an elasto-plastic equivalent system for the short pier, A70. A low post-yield
stiffness results from the experimental curve, consequently elastic-perfectly plastic
behaviour can be assumed. maxF being the maximum force and considering the secant
stiffness at the point of the experimental curve corresponding to 0.75 maxF , the yield
displacement is calculated as y,bu =0.01 m and the displacement ductility is u,bµ =8 (the
index b stands for bilinear). The bilinear curve is plotted along with the experimental one
in Figure 3.6: the bilinear curve is a poor approximation of the experimental envelope in
the region between cracking and maximum resistance. Alternatively, by extrapolating the
linear behaviour until the maximum strength, the yield displacement is calculated as
y,bu =0.008 m and the displacement ductility is then u,bµ =10. Such high values of
ductility capacity are unrealistic. Following the definition of ductility given in EC8, one
obtains y,EC8u = 0.015 m and then u,EC8µ = 5.3 (the index EC8 stands for the definition
according to EC8).
The vertical reinforcement of the short pier A70 is almost equally distributed throughout
the web, similarly to shear wall elements. Therefore, the rebars of the web, and not only
those of the flange, contribute to the strength. The sequence of physical phenomena that
occur with increasing displacement is: cracking, yielding of the external series of vertical
rebars of the flange (first yielding - y1u ), yielding of the internal series of vertical rebars
of the flange, progressive yielding of the vertical rebars of the web, fluctuation of the
neutral axis until stabilization (total yielding - yu ), crushing of concrete and buckling of
reinforcement, rupture of vertical rebars, failure - uu .
A trilinear envelope curve, also shown in Figure 3.6, is a better approximation of the
experimental curve. The first branch is defined by the secant stiffness at 0.75 maxF and
corresponds to essentially linear behaviour before first yielding and initiation of cracking.
The third branch corresponds to the part after the resistance has levelled off, identified by
the stabilization of the tangent stiffness. Then, the second branch, which corresponds to
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
45
the progressive yielding and cracking, is a linear approximation of the experimental
curve. Following these rules, the yield displacement is identified as y,tu =0.025 m and the
displacement ductility is u,tµ =3.2 (the index t stands for the trilinear envelope).
The apparent large ductility capacity, u,bµ =8, is also due to the small number of cycles
the tested specimens experienced. Following EC8-2 [CEN, 2002], ductility is defined for
the ultimate displacement at which the structure can undergo five cycles without initiation
of failure of the confining reinforcement or loss of strength more than 30%. As observed
from the results of numerical simulations, see Chapter 5, the short pier shows a loss of
strength of almost 30% for the fourth cycle at u = 0.04 m, therefore the ductility should be
u,bµ < 4 and u,tµ < 1.6. These values are consistent with the low deformation capacity of
the pier, as defined by lateral drift. It seems, then, that ductility alone cannot fully
describe the deformation capacity of elements. For this reason, in the following
comparisons will be made mainly on the basis of lateral drift.
-1500
-1000
-500
0
500
1000
1500
-0.10 -0.05 0.00 0.05 0.10
Displacement (m)
For
ce (k
N)
experimentalbilineartrilinear
Figure 3.6. Cyclic test on the short pier: experimental and envelop force-displacement
curves
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
46
0
1
2
3
4
5
6
7
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Hei
ght (
m)
shear flexure 0.1% drift
0
1
2
3
4
5
6
7
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Hei
ght (
m)
shear flexure 0.4% drift
0
1
2
3
4
5
6
7
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Hei
ght (
m)
shear flexure 0.9% drift
0
1
2
3
4
5
6
7
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Hei
ght (
m)
shear flexure 1.5% drift
Figure 3.7. Cyclic test on the short pier: flexural and shear displacement
Flexural and shear deformation
The flexural and shear deformation was computed from the measurements of the
displacement transducers, as explained before. Figure 3.7 presents the shear and flexural
deformation along the height of the pier model for increasing values of lateral drift.
Although the pier had a relatively small aspect ratio, L/d = 2.4, it showed a prevailing
flexural response. This is consistent with the observed damage (few diagonal cracks,
concentration of deformation demand at the base, failure due to rupture of vertical
reinforcement) and was due to the small amount of longitudinal reinforcement and to the
presence of lapped splices at the base of the pier. In fact, this resulted in a weak interface
between the pier and foundation, with most of the rotation concentrated there.
The ratio of shear to flexural displacement at the maximum positive and negative
displacement excursions ranges from 0.3 to 0.4. In agreement with the observed damage
and experimental plastic hinge length, the ratio is in the order of 0.5 within the first 1.0 m
from the base of the specimen and reduces to about 0.1 in the upper part of the specimen.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
47
The average value for all cycles is 0.35. This value is in quite good agreement with the
values 0.25 and 0.3 experimentally measured from tests on scaled models of rectangular
hollow bridge piers with highly confined boundary elements [Hines et al., 2002a]. It
points out that shear has a significant contribution to the total displacement of piers with
elongated hollow cross-section.
Equivalent plastic hinge length
Before a member reaches the yield deformation, the distribution of curvature is linear
along its length. With increasing displacement, plastic deformation is concentrated in a
small portion of the element, the plastic hinge, and the curvature follows a parabolic
distribution along the height of the element. For reasons of simplicity, it can be assumed
that the maximum curvature remains constant within an equivalent plastic hinge length,
and then the curvature follows a linear distribution. Among the expressions proposed for
the length of the equivalent plastic hinge, hL , EC8-2 [CEN, 2002] suggests
d)6.04.0(Lh −= (3.8)
and
h ykL 0.08L 0.022 f= + Φ (3.9)
Considering a cantilever of length L with a triangular distribution of moments, it writes
sysuhyu f/f)LL/(LM/M ≈−= , where uM and yM are, respectively, the maximum and
yield moments of the cross-section. For the commonly-used value sysu f/f = 1.15, it
follows that L13.0Lh = , which is similar to the first term at the right part of Equation
3.9.
A modification of Equation 3.9 is introduced in EC8-3 [CEN, 2003b] for the estimation
of the equivalent plastic hinge length of columns and beams in existing buildings
h sl y
1L 0.08L f
60= + α Φ (3.10)
where slα is equal to 1 if there is slippage of vertical reinforcement and 0 otherwise and
yf is the estimated yield strength of steel.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
48
A widely used expression for the estimation of the equivalent plastic hinge length is
[Pauley & Priestley, 1992]
hL 0.08L 6= + Φ (3.11)
Based on experimental observations of ductile columns with hollow cross-section, it has
been proposed to estimate the equivalent plastic hinge length as [Mander, 1984]
hL 0.4d= (3.12)
Piers with hollow cross-section and distributed reinforcement can be considered similar to
wall elements. For this case, the expressions
h wL 0.2l 0.044L= + (3.13)
and
Wh l)8.03.0(L −= (3.14)
have been proposed [Paulay & Priestley, 1992]. Also for wall-type elements, an
estimation of the plastic hinge length is [Wallace and Moehle, 1992]
Wh l)0.15.0(L −= (3.15)
In the above expressions wl is the length of the wall, L is the length of the pier, Φ is the
diameter of the longitudinal rebars, yf is the yield stress of steel and d is the section
depth.
Based on numerical parametric analyses, a formulation for the plastic hinge length has
been proposed, that considers the contribution of plastic rotation, IhL , and the
contribution of fixed end rotation, IIhL , according to the equations [Ceroni et al., 2003]
83.132/65.0yt
43.0Ih )1()1f/f()H/L(1.6L −− ν+ε−= (3.16)
II 0.2h b t yL 5d (f / f 1)= − (3.17)
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
49
where yf and tf are the yield and maximum stress of steel, e is the ultimate deformation
of steel, ? is the normalised axial load and db is the diameter of the longitudinal rebars.
These expressions were developed for rectangular columns and incorporate the effect of
plasticity spreading through the ratio of ultimate and yield stress of steel.
A modification of Equation 3.9 has been introduced [Hines et al., 2002b] to account for
the effect of tension shift in deep beams. The proposed equation was calibrated on
experimental results on large-scale specimens of bridge piers with rectangular hollow
cross-section and highly confined boundary elements. It takes the form
ykh f022.0d3.0L08.0L Φ++= (3.18)
Based on experimental results, the plastic hinge length, hL , can be calculated using the
expression [Paulay & Priestley, 1992]
p m u h hu ( )L (L 0.5L )= ϕ − ϕ − (3.19)
where pu is the measured plastic displacement, equal to the difference between the
maximum displacement and the yield displacement, mϕ and uϕ are respectively the
values of curvature measured at yield and at maximum displacement.
Table 3.4. Experimental and empirical values of plastic hinge length for the short pier Empirical Experimental
ykf022.0L08.0 Φ+ 0.65 uδ = 0.4% 0.05
(0.4~0.6)d 1.09~1.63 uδ = 0.9% 0.30
0.08L 6+ Φ 0.58 uδ = 1.5% 0.18
sl y
10.08L f
60+ α Φ 0.61
0.4 h 1.10
w0.2l 0.044L+ 0.83
(0.3~0.8) wl 0.80~2.20
I IIh hL L+ 4.71
ykf022.0d3.0L08.0 Φ++ 1.47
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
50
The values obtained from the empirical expressions are compared to those obtained from
Equation 3.19 for different levels of displacement ductility in Table 3.4. The yield
displacement was defined considering the trilinear equivalent curve. The experimental
values are smaller than the empirical ones for all levels of displacement ductility. The
extremely limited plastic hinge results from the presence of the lapped splices at the base.
The experimental equivalent plastic hinge length was smaller than the overlapping length
and smaller than the empirical values. Note the very large value predicted by Equations
3.16 and 3.17. It is recalled that the design plastic hinge length is used for detailing
purposes (e.g. extent of the critical region) and therefore should be larger than the
experimental value for reasons of safety.
Distribution of curvature
The evolution of average curvature, as measured from the instrumentation, along the
height of the pier for increasing values of drift is presented in Figure 3.8. Until 0.4%
lateral drift, the curvature demand was concentrated at the base of the pier, whereas for
higher values of drift, a significant demand was also observed above the lapped splices, in
accordance with the evolution of cracking pattern observed during the tests. In both cases,
the deformation demand was concentrated in a very thin slice with height in the order of
0.25 m. As stated before, such a small plastic hinge length was due to the presence of
lapped splices just above the base block that did not allow for the development of the
yield stress throughout the whole length of the overlapping bars.
Using the vertical displacement transducers placed on the flange of the specimen, one can
follow the average vertical deformation of the slices along the height of the pier model, as
shown in Figure 3.9 for the first four slices from the base. The first two slices are 0.25 m
high, whereas the next two are 0.5 m high. Note that the overlapping length is 0.50 m and
corresponds to the first two slices. In Figure 3.9 is plotted also the line at ε = 0.0025,
corresponding to the yield deformation of the steel rebars. The rebars near the external
face of the flange at the base reached the yield limit for displacement around y1u = 0.008
m during the first cycle with amplitude 0.009 m, whereas the ones of the third slice
reached the yield limit for displacement around u = 0.04 m during the first cycle with
amplitude 0.056 m. It should be highlighted that the rebars of the second and the fourth
slices did not seem to reach the yield strain. This leads to the conclusion that the
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
51
longitudinal reinforcement was not allowed to develop the yield stress through the whole
overlapping length.
Damage assessment
Several models have been proposed for the quantification of damage. One of the most
popular, the Park and Ang model [Park & Ang, 1985], defines the damage index by two
parts, one taking into account the maximum deformations experienced and the other
accounting for the cyclic effects. The damage index, DI, is defined as
m u,m y u,mDI u / u dE(F u )= + β∫ (3.20)
where mu is the maximum response deformation, u,mu is the ultimate deformation
capacity in monotonic loading, dE is the incremental dissipated hysteretic energy, yF is
the yield strength and β is a non-negative constant; β = 0.05 for reinforced concrete
components [Park et al., 1987].
0.1% drift
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Curvature (rad/m)
Hei
ght (
m)
0.4% drift
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Curvature (rad/m)
Hei
ght (
m)
0.9% drift
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Curvature (rad/m)
Hei
ght (
m)
1.5% drift
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Curvature (rad/m)
Hei
ght (
m)
Figure 3.8. Cyclic test on the short pier: distribution of average curvature along the height
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
52
h = 0.125 m
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0 500 1000 1500 2000Loading step
Stra
in
h = 0.375 m
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0 500 1000 1500 2000Loading step
Stra
in
h = 0.625
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0 500 1000 1500 2000Loading step
Stra
in
h = 1.125
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0 500 1000 1500 2000Loading step
Stra
in
Figure 3.9. Cyclic test on the short pier: evolution of vertical deformation
Table 3.5. Park & Ang Damage Index [Park & Ang, 1985]
DI Damage Physical appearance
> 1 Collapse Total or partial collapse
0.8 - 1 Severe Extensive crushing of concrete, disclosure of buckled reinforcement
0.3 - 0.5 Moderate Extensive large cracks, spalling of concrete
0.2 – 0.3 Minor Minor cracks, partial crushing of concrete
< 0.2 Slight Sporadic occurrence of cracking
Equation 3.20 yields unity for failure of the component or structure. Five degrees of
damage are associated with different values of DI, as seen in Table 3.5. The evolution of
the Park and Ang Damage Index for the short pier is shown in Figure 3.10, where the
observed damage is also indicated: good agreement between the structural damage and
the calculated values of the damage index is observed.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
53
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 500 1000 1500 2000
Loading step
Par
k &
Ang
Dam
age
Inde
x
cracking
rupture of 1st rebar
rupture of rebars
buckling, crushing
Figure 3.10. Cyclic test on the short pier: Park & Ang Damage Index
Table 3.6. Bridge damage and performance assessment [Hose et al., 2000]
Level Damage Repair Performance level Socio-economic
description
I No No Cracking Fully operational
II Minor Possible Yielding Operational
III Moderate Minimum Initialisation of local
mechanism Life safety
IV Major Repair Full development of
local mechanism Near collapse
V Local failure/collapse Replacement Strength degradation Collapse
An attempt was made [Hose et al., 2000] to correlate different values of several damage
indices to different damage levels, see Table 3.6. The relation depends on the behaviour
mode of the structure or component, namely brittle (sudden drop of resistance after the
maximum value), strength-degrading (gradual drop of resistance for increasing
displacement after yielding), and ductile (almost constant resistance for increasing
displacement after yielding). For the test presented herein the behaviour is identified as
ductile.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
54
The residual deformation index, RDI, is a function of the inelastic behaviour of the
structure and can be used to estimate damage. It is a non-dimensional index equal to the
ratio of the permanent residual displacement, pu , to the yield displacement, yu ,
according to the expression
p yRDI u / u= (3.21)
The normalised stiffness, kn , quantifies the stiffness degradation. It is equal to the ratio
of the effective stiffness, effK , to the initial stiffness at yielding, oK , according to the
expression
k eff on K / K= (3.22)
where the effective stiffness defines the slope of the equivalent elastic system (secant
stiffness) and can be calculated from the maximum force, F, and the corresponding
displacement, u,
effK F / u= (3.23)
and the initial stiffness can be calculated by the yield force and the yield displacement
o y yK F / u= (3.24)
The plastic rotation, pθ , is defined as the plastic displacement, at a given level divided by
the length of the member
p pu / Lθ = (3.25)
Equivalent damping, eqξ , can be seen as an index of the energy-dissipation capacity and
is defined as the ratio of the dissipated energy in one cycle, ED, to the strain energy of an
equivalent linearly elastic system. It can be calculated using the expression, e.g. [Clough
& Penzien, 1975]
Deq
max max
E2 F u
ξ =π
(3.26)
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
55
where maxF is the maximum force and maxu is the maximum displacement reached for the
cycle.
Table 3.7 presents the values of the previously defined damage indices for different levels
of displacement ductility. The stiffness degradation, as measured by the normalised
stiffness, increases rapidly for ductility µ > 2; the same applies for the residual
displacement index that measures the inelastic displacement. The values of equivalent
damping do not change significantly with increasing values of displacement ductility:
quite stable, but limited, resources of energy-dissipation capacity are available after
yielding. In general the different damage indices are able to follow the evolution of the
damage state of the specimen.
3.4.2. Comparison to empirical predictions
A comparison is made, in terms of yield displacement and displacement ductility
capacity, between the experimental data and the values predicted using empirical
expressions. Based on the theory of elasticity, we obtain [Paulay & Priestley, 1992]
2y yu L / 3= ϕ (3.27)
u y pu u u= + (3.28)
p u y h hu ( )L (L 0.5L )= ϕ − ϕ − (3.29)
where yϕ and uϕ is the curvature at yield and ultimate capacity respectively, yu and uu
are the yield and ultimate displacement respectively and hL is the plastic hinge length, as
calculated from Equation 3.9, are used to predict the values of displacement at yielding
and failure. The yield curvature can be estimated as [Park & Paulay, 1975]
Table 3.7. Damage assessment of the short pier
tµ bµ δ eqξ kn RDI DI pθ
1.1 2.7 0.4% 11.6% 0.30 0.12 0.30 0.0003
2.3 5.6 0.9% 13.1% 0.15 0.22 0.65 0.0048
3.2 8.0 1.2% 14.8% 0.10 0.42 1.15 0.0089
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
56
yy (1 k)d
εϕ =
− (3.30)
( ) ( ) ( )y
'tt
2
y
'tct2
y
2'tt 2
14
14
1k
αρ+ρ−
αρβ+ρ+
αρ+ρ= (3.31)
t yt
c
A f
bdfρ = ; c y'
tc
A f
bdfρ = ; y
yo
εα =
ε;
ddc
c =β (3.32)
where y y sf Eε = is the yield strain of steel, tρ and 'tρ are respectively the volumetric
ratios of the tension and the compression reinforcement, oε is the strain at maximum
strength of concrete and cd is the concrete cover. The ultimate curvature is
cmu
ucε
ϕ = (3.33)
where cmε is the ultimate strain of concrete in compression and uc is the height of the
neutral axis at the ultimate curvature. Hollow cross-sections with elongated web, d/b =
2.72/1.02 = 2.7, and longitudinal reinforcement almost equally distributed along the
length of the web can be considered similar to shear wall elements. EC8-1 [CEN, 2003a]
proposes the following expression for the depth of the neutral axis of wall elements at
ultimate curvature
u d v c c oc ( )h b / b= ν + ω (3.34)
d c c cdN / h b fν = ; v sv c c yd cd(A / h b )(f / f )ω = (3.35)
where the axial force, N, and the vertical reinforcement ratio are normalised to the area of
the flange, ch b .
For wall elements the following expression has been proposed for the yield curvature
[Paulay & Priestley, 1992]
y w0.0033 / lϕ = (3.36)
where wl is the length of the wall.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
57
Alternative expressions have been proposed for the yield curvature, yield displacement,
equivalent plastic hinge length, plastic displacement and ultimate curvature of shear wall
elements [Wallace & Moehle, 1992]
y w0.0025 / lϕ = (3.37)
2y yu 11 L / 40= ϕ (3.38)
p u y wu 0.5( )Ll= ϕ − ϕ (3.39)
y1 cm
cu w
y2
c w w c
f(0.85 3 '' )
fl
f N[1.50 ( '')-1.25 ']
f l t f
β + ρ εφ =
ρ + ρ ρ + (3.40)
where wt is the width of the wall and lb is a factor defined by ACI-318-89; Figure 3.11
plots the relation between the wall parameters and the available ultimate curvature. The
equivalent plastic hinge length is calculated by Equation 3.15, considering the lower
limit, because closer to the experimental value.
Figure 3.11. Relation between wall parameters and ultimate curvature [Wallace &
Moehle, 1992]
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
58
Table 3.8. Experimental and empirical displacement for the short pier
Empirical
Experimental Columns1 Walls1 Walls2
cu (mm) 3
y1u (mm) 8
yu (mm) 253/104 14 17 10
pl (m) 0.48 0.64 0.83 1.36
uu (mm) 82.5 207 263 60
µ 3.33/8.04 14.8 15.5 6
δ (%) 1.3 3.2 4.1 0.9
1 [Paulay & Priestley, 1996], 2 [Wallace & Moehle, 1992], 3 trilinear envelope, 4 bilinear
envelope
The results are compared to the experimental values in Table 3.8. While the empirical
expressions predict relatively well the yield displacement, they significantly overestimate
the plastic hinge length, which is conditioned by the lapped splices, and therefore also the
ultimate displacement. The empirical equations do not successfully predict the
deformation capacity of the pier, mainly in terms of the plastic displacement.
3.4.3. Comparison to a squat pier designed according to EC8
General
A comparison is made to a squat pier designed according to EC8 and tested within the
PREC8 research programme [Pinto et al., 1995; Calvi & Pinto, 1996]. The pier had a
rectangular hollow cross-section with external dimensions 1.60x0.80 m and thickness
0.16 m for both the flange and the web. The height of the scaled pier was 2.8 m, which
corresponds to aspect ratio L/d = 1.75. The longitudinal reinforcement was continuous
and the transverse reinforcement consisted of closed hoops. The reinforcement ratio was
sρ = 0.9% and wρ = 0.4% for the vertical and horizontal steel, respectively. Although
shear had a considerable contribution to the deformation of the pier, significant capacity
of ductility, uµ = 6, and drift, uδ = 2.6%, was observed.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
59
-1500
-1000
-500
0
500
1000
1500
-3.0 -2.0 -1.0 0.0 1.0 2.0 3.0
Drift (%)
For
ce (k
N)
A70 PREC8
Figure 3.12. Force-drift diagrams for the short, A70, and the squat, PREC8, piers
The force-drift diagrams for the short and the squat piers are presented in Figure 3.12.
The squat pier had larger drift, ductility and energy-dissipation capacity compared to the
short pier. This was due to the absence of lapped splices from the plastic hinge zone, the
larger reinforcement ratios and the seismic detailing of the transverse reinforcement.
Equivalent damping
The values of equivalent damping are plotted against drift in Figure 3.13 along with trend
lines of the form eq [(1 ( ) ] 5γξ = α + βδ + . For small levels of drift, both piers had similar
dissipation capacities, but the maximum value for the PREC8 pier, eq,PREC8ξ = 19%, was
higher than the one for the short pier A70, eq,A70ξ = 15%. Note that the capacity of pier
A70 was exhausted at drift u,A70δ = 1.5%, while the PREC8 pier exhibited stable capacity
of energy dissipation until almost twice this value, u,PREC8δ = 2.6%.
Several theoretical relations have been proposed for the estimation of equivalent damping
as a function of displacement ductility. These expressions correlate the equivalent
damping, eqξ , of a linear singe degree-of-freedom (SDOF) oscillator to the displacement
ductility, µ. Following an early proposal [Gulkan & Sozen, 1974]. the equivalent
damping, can be calculated as
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
60
eq 0.02 0.02( 1) /ξ = + µ − µ (3.41)
A large number of theoretical expressions for the estimation of the equivalent damping
can be found in [Iwan & Gates, 1979]. Two methods are used for the determination of the
equivalent damping, either based on harmonic response or based on random response. A
theoretical expression is
eq 2
1 ( -1)(1- )( - 0.5) 0.5 ( -1)
µ αξ =
π µ + α µ (3.42)
for a = 5%.
Based on analysis of column experimental data, the ratio of the dissipated hysteretic
energy to that of an elastic-perfectly plastic hysteretic loop was found to be 0.35
[Priestley, 1993]. By expressing this ratio as a function of displacement ductility, the
equivalent damping is
eq 0.2228(1 1/ )ξ = − µ (3.43) Considering the Takeda model for the
hysteretic behaviour of SDOF RC structures the equivalent damping can be calculated as
[Kowalsky et al., 1994]
eq
0.95(1- - 0.05 )
0.05µ
µξ = +
π (3.44)
A different set of empirical expressions is drawn from experimental data. An empirical
expression, resulting from test data and considering P-d effects for the calculation of the
equivalent damping is [Kowalsky et al., 1994]
eq 0.05 0.39372(1 1/ )ξ = + − µ (3.45)
Alternatively, the equivalent damping of reinforced concrete buildings can be estimated
as [Shimazaki, 2000]
eq 0.05 (1 1/ )ξ = + β − µ (3.46)
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
61
where the viscous damping index, β , is equal to 0.01 for shear-failure type RC structures,
0.1 for frame type RC structures with shear wall, 0.15 for frame type RC structures with
slippage of reinforcing bar at beam-column joint, 0.2 for frame type RC structures and
0.25 for frame type steel structure.
0
5
10
15
20
0 0.5 1 1.5 2 2.5 3
Drift (%)
Equ
ival
ent d
ampi
ng (%
)
A70 PREC8
Figure 3.13. Equivalent damping – drift for the PREC8 and A70 piers
0
5
10
15
20
25
30
35
0 2 4 6 8 10 12Displacement ductility
Equ
ival
ent d
ampi
ng (%
)
Gulkan & Sozen, 1974Iwan & Gates, 1979Priestley, 1993Kowalsky et al., 1994Kowalsky et al., 1994 Takeda modelShimazaki, 2000A70,experimentalPREC8, experimental
Figure 3.14. Equivalent damping – displacement ductility: theoretical expressions and
experimental values for the PREC8 and A70 piers
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
62
Figure 3.14 plots the experimental values of equivalent damping for the short, A70, and
the squat, PREC8, piers along with the empirical expressions described above. Ductility
is defined assuming the bilinear envelope. The empirical curves based on the Takeda
model [Kowalsky et al., 1994] and on the elastic-perfectly plastic behaviour [Priestley,
1993] fit well the experimental values for the PREC8 pier, which showed an almost
elastic-perfectly plastic response. On the other hand, the relations that consider pinched
hysteretic response [Shibata & Sozen, 1976; Shimazaki, 2000] are closer to the
experimental values for pier A70, which actually exhibited pinched hysteretic curves, as
seen Figure 3.4. This leads to the conclusion that it is possible to obtain a credible
estimation of the relation between equivalent damping and displacement ductility,
provided that the behaviour of the element is known. In assessment of existing structures
such behaviour is not known, unless detailed analyses are performed.
Damage assessment
A quantitative comparison of the behaviour of the two piers is made in Table 3.9, which
presents several indices of damage corresponding to the final cycles for both piers. The
energy-dissipation capacity, as quantified by the equivalent damping, of the squat pier
increased until eq,PREC8ξ = 18.6% for PREC8δ = 1.3% and remained quite stable until
u,PREC8δ = 2.6%. On the other hand, the energy-dissipation capacity of the short pier
increased until eq,A70ξ = 14.8% for lateral drift u,A70δ = 1.5%, where failure occurred.
Despite the larger stiffness degradation, partly due to the larger number of cycles
sustained by the specimen, the squat pier showed wider hysteretic cycles, as evinced by
the larger values of RDI and pϑ , and almost double the drift capacity of the short pier.
Table 3.9. Comparison of the two piers (values at ultimate displacement) sρ (%) wρ (%) uµ uδ (%) eqξ (%) kn RDI
A70 0.40 0.09 3.20a/8.0b 1.54 14.8c 0.10a/0.16b 2.2a/7.0b
PREC8 0.90 0.40 6.0 2.57 18.6d 0.16 4.4
a trilinear envelope, b bilinear envelope, c d = 1.5%, d d = 1.3%
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
63
-1000
-750
-500
-250
0
250
500
750
1000
-0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25
Displacement (m)
Forc
e (k
N)
Figure 3.15. Cyclic test on the tall pier: force-displacement curve
Observing the values of Table 3.9, better cyclic behaviour is indicated for the PREC8 pier
by all indices for ductility defined by the trilinear diagram. The damage indices for
ductility defined by the bilinear diagram imply similar or even better performance of pier
A70. This supports the idea that the bilinear envelope might guide to misleading results
and that the trilinear diagram describes better the performance of pier A70.
3.5. CYCLIC TEST ON A MODEL OF A TALL BRIDGE PIER
3.5.1. Experimental results
A constant axial load, N = 4050 kN, corresponding to a normalised axial load ν = 0.10,
was imposed on the top of the specimen. The displacement history consisted of one cycle
of 6 mm, two cycles of 30 mm, two cycles of 70 mm, two cycles of 140 mm and one
cycle of 250 mm in the strong direction.
Force-displacement diagram and observed damage
The force displacement diagram is presented in Figure 3.15. Based on a trilinear envelop
curve, the yield displacement is identified as y,tu = 0.10 m and the displacement ductility
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
64
is u,tµ = 2.3. For the bilinear skeleton curve, the corresponding values are y,bu = 0.06 m
and u,bµ = 3.7. Following the EC8 definition, the yield displacement is identified as
y,EC8u = 0.07 m and the displacement ductility is then u,EC8µ = 3.3. The drift capacity was
δ = 1.6%, with respect to the base.
The crack pattern at the end of the test is shown in Figures 3.16 and C.4, where green
colour corresponds to the cracks developed during the cycles of 30 mm and 70 mm, while
red colour corresponds to the cracks developed during the cycles of 140 mm and 250 mm.
For the cycles of 30 mm flexural cracks appeared within the first 1.0 m from the base of
the pier. For the cycles of 70 mm the flexural cracking extended up to the height of 3.0 m
from the base. A large horizontal crack at the height of 1.5 m, corresponding to the top of
the lapped splices, was also observed. With increasing amplitude of displacement, cracks
appeared throughout the first 4.0 m of the pier. For the cycles of 140 mm a large
horizontal crack appeared at the critical cross-section at 3.5 m. With further cycling, the
flexural cracking above the critical cross-section increased and diagonal cracks appeared
in the lower part. For the final cycle of 250 mm spalling of concrete occurred at the
corners of the pier at 3.5 m and failure was caused by rupture of vertical rebars at the
flange after buckling. Failure was due to a combination of flexure and shear.
Figure 3.16. Cyclic test on the tall pier: crack pattern at the end of the test
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
65
The buckling length was equal to almost twice the transverse reinforcement spacing, see
Figure C.8, implying that the horizontal reinforcement, which was not properly anchored,
did not provide sufficient lateral support for the vertical rebars.
For the prototype pier the critical cross-section is at about 9.0 m from the base, not
conforming to the requirement of EC8-3 for accessibility of the plastic hinge region for
inspection and eventual repair and/or strengthening. Similar failure mode has been
reported for rectangular pier models with termination of the longitudinal reinforcement at
mid-height [Kawashima et al., 1990; Ogata & Osada, 2000; Calvi et al., 2000]. Premature
termination of vertical rebars was common practice in Japan before the introduction of
modern seismic codes. Actually, it is reported that failure of the Hanshin Expressway was
triggered by this deficiency [Kawashima, 2000a].
The location of failure can be explained with reference to Figure 3.17. Considering the
equilibrium of a diagonally cracked element with shear reinforcement, one can define the
following forces acting on a cross-section: cF is the compressive force on concrete, tF is
the tension force on longitudinal steel and in addition one can consider the uniformly
distributed inclined forces of the transverse steel (tension forces), wF , and the concrete
strut (compression forces), dF . From equilibrium of these forces and the external axial
load, N, shear force, V, and bending moment, M, we obtain
)cot(cotV5.0z/MN5.0z/)cot(cotVyz/MN5.0F ww1sc α−θ−+−≈α−θ−+−= (3.47)
and
)cot(cotV5.0z/MN5.0z/)cot(cotVyz/MN5.0F ww1st α−θ++≈α−θ++= (3.48)
where w wV F= ∑ is the contribution of the transverse reinforcement to the shear
strength, ϑ is the inclination of the cracks with respect to the element axis, α is the angle
of the transverse reinforcement, s1y is the distance between the reinforcement centreline
and the cross-section centroid and z is the internal lever arm. The last term in the previous
expressions is due to the diagonal shear cracking and is additive to the forces due to pure
bending. It is evident that after the development of diagonal cracks, the tension force
acting on the longitudinal reinforcement is greater than that required to resist the bending
moment alone. This phenomenon is termed tension shift. In EC2 tension shift is taken
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
66
into consideration by shifting the design moment curve in the unfavourable direction by a
distance 1a
1a z(cot cot ) / 2= ϑ − α (3.49)
which is in agreement with the previous theoretical expressions.
Figure 3.17. Equilibrium of internal forces in diagonally cracked element with shear
reinforcement
02468
10121416
-0.30 -0.20 -0.10 0.00 0.10 0.20 0.30Displacement (m)
Hei
ght (
m)
shear flexure 0.2% drift
02468
10121416
-0.30 -0.20 -0.10 0.00 0.10 0.20 0.30Displacement (m)
Hei
ght (
m)
shear flexure 0.5% drift
02468
10121416
-0.30 -0.20 -0.10 0.00 0.10 0.20 0.30Displacement (m)
Hei
ght (
m)
shear flexure 1.0% drift
02468
10121416
-0.30 -0.20 -0.10 0.00 0.10 0.20 0.30Displacement (m)
Hei
ght (
m)
shear flexure 1.6% drift
Figure 3.18. Cyclic test on the tall pier: flexural and shear displacement
α θ
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
67
Flexural and shear deformation
The flexural and shear deformations were computed from the measurements of the
displacement transducers, as explained before. Figure 3.18 presents the shear and flexural
deformation along the height of the scaled model for the maximum displacement of each
cycle. The contribution of shear to the total displacement was consistent with the
observed damage. Throughout the height of the pier and for all levels of lateral drift, the
displacement due to flexure was more than 80% of the total displacement. Note a
difference in the slice above the first lapped splices at 0.75 m from the base, where the
total displacement results from equally shared shear and flexural components. These
values support the need to further study the contribution of shear to the total deformation
of hollow piers, which is not included in simple models.
The mean values, over the height of the specimen, of the ratio of shear to flexural
displacement increase with the amplitude of imposed displacement, namely from 0.16
through 0.19 and 0.33 until 0.51. In accordance with the observed damage, for the cycles
with amplitude 3 mm and 7 mm, the ratio assumes large values, in the order of 0.5, within
the first 1.5 m from the base cross-section. For the cycles with larger amplitude, large
values of the ratio are observed in the part of the pier from the base until the height of 4.5
m. The mean value for all cycles with different amplitudes, 0.3, is in agreement with the
value experimentally measured for piers with highly confined boundary elements [Hines
et al., 2002a]. Nevertheless, it does not seem appropriate to use the mean value for a
magnitude that keeps increasing with displacement. Further studies are needed in order to
provide safe and comprehensive rules.
Distribution of curvature
The evolution of average curvature, as measured from the instrumentation, along the
height of the pier is presented in Figure 3.19 for increasing values of drift. Until 0.5%
drift, the curvature demand was evenly distributed within the first 3 m from the base of
the pier specimen. For increasing levels of lateral drift, the deformation demand was
shifted to the cross-section above the cut-off; a significant curvature demand was also
observed above the first lapped splices.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
68
0.2% drift
0
2
4
6
8
10
12
14
-0.03 -0.02 -0.01 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
0.5% drift
0
2
4
6
8
10
12
14
-0.03 -0.02 -0.01 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
1.0% drift
0
2
4
6
8
10
12
14
-0.03 -0.02 -0.01 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
1.6% drift
0
2
4
6
8
10
12
14
-0.03 -0.02 -0.01 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
Figure 3.19. Cyclic test on the tall pier: distribution of average curvature along the height
The evolution of vertical deformation of the slices of the tall pier is presented in Figure
3.20, along with the value ε = 0.0025 that corresponds to the yield deformation of steel.
The first two graphs correspond to the lower 0.5 m of the tall pier. The vertical rebars at
the base, close to the external face of the flange, exceeded the yield deformation for the
cycles with amplitude 70 mm. The third graph plots the vertical deformation of the slice
just above the first lapped splices. The vertical rebars reached the yield deformation for
the cycles with amplitude 140 mm, as observed by the deformation demand evident in
Figure 3.19. The last three graphs present the vertical deformation of the pier within a
distance of 1.5 m from the critical cross-section. The longitudinal rebars above the critical
cross-section exceeded the yield deformation for the cycles with amplitude 140 mm and
failed during the cycle of amplitude 250 mm, as they reached deformation above 20%. As
seen in the last graph of Figure 3.20, yielding occurred in the rebars within 1.0 m from the
critical cross-section. This is in agreement with the experimental value of the plastic
hinge length, hL = 0.90 m.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
69
h = 0.125 m
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1000 2000 3000Loading step
Stra
in
h = 0.375 m
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1000 2000 3000Loading step
Str
ain
h = 1.25 m
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1000 2000 3000Loading step
Str
ain
h = 3.75 m
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1000 2000 3000Loading step
Str
ain
h = 4.25 m
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1000 2000 3000Loading step
Str
ain
h = 4.75 m
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1000 2000 3000Loading step
Str
ain
Figure 3.20. Cyclic test on the tall pier: evolution of vertical deformation
Equivalent plastic hinge length
The empirical predictions, Equations 3.8-3.18, for the equivalent plastic hinge length are
compared to the values calculated using the instrumentation measurements, Equation
3.19, in Table 3.10. The values predicted by the empirical expressions are in fair
agreement with the experimental ones. Note that, contrary to what was observed on the
short pier, spread of plasticity was allowed above and below the critical cross-section of
the tall pier. Note again the unrealistic value predicted by Equations 3.16 and 3.17.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
70
Table 3.10. Experimental and empirical values of plastic hinge length for the tall pier
Empirical Experimental
ykf022.0L08.0 Φ+ 0.92 µ = 1.4 0.95
(0.4~0.6)d 1.09~1.63 µ = 2.3 0.90
0.08L 6+ Φ 0.91
y
10.08L f
60+ Φ 0.95
0.4 h 1.10
w0.2l 0.044L+ 1.16
(0.3~0.8) wl 0.80~2.20
I IIh hL L+ 6.44
ykf022.0d3.0L08.0 Φ++ 1.74
Table 3.11. Damage assessment of the tall pier
tµ bµ δ eqξ kn RDI DI pθ
1.4 2.2 1.0% 9.60% 0.76 0.070 0.6 <<
2.3 3.7 1.6% 13.7% 0.43 0.365 1.2 0.01
Table 3.12. Experimental and empirical displacement for the tall pier
Empirical
Experimental Columns1 Walls1 Walls2
cu (mm) 15
y1u (mm) 20
yu (mm) 1003/624 40 44 30
pl (m) 0.90 0.98 1.0 1.36
uu (mm) 230 520 532 110
uµ 2.33/3.74 13 12.1 3.7
uδ (%) 1.6 3.7 3.8 0.8
1 [Paulay & Priestley, 1996], 2 [Wallace & Moehle, 1992], 3 trilinear, 4 bilinear
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
71
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 500 1000 1500 2000 2500 3000 3500
Loading step
Par
k &
Ang
Dam
age
Inde
x
cracking
extended cracking
spallingrupture of rebars
Figure 3.21. Cyclic test on the tall pier: Park & Ang Damage Index
Damage assessment
The evolution of the Park and Ang Damage Index is presented in Figure 3.21, where the
observed damage is also indicated. As for the short pier, the Park and Ang Damage Index
follows well the evolution of damage: cracking was indicated at an early stage of the test,
then cracking extended without significant inelastic deformation and only near the end of
the test the element failed after spalling of concrete and rupture of vertical rebars.
The one-parameter damage indicators, defined previously, are presented in Table 3.11 for
the tall pier. All damage indicators follow the evolution of damage of the specimens, but
the limited amount of experimental data prohibits an insightful assessment of the
examined damage indicators.
3.5.2. Comparison to empirical predictions
The experimental results are compared to the results of the empirical expressions of
Equations 3.25-3.36 for the prediction of the yield and ultimate displacement. The
empirical values of displacement are calculated considering only the plastic hinge above
the bar cut-off. As seen in Figures 3.18 and 3.19, inelastic deformation of steel rebars is
indicated also at the base of the pier and at the cross-section above the first lapped splices.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
72
The empirical formulae underestimate the yield displacement also because they do not
take into consideration the deformation due to the rotation of the part below the critical
cross-section. The empirical formulae predict a smaller ultimate curvature than the one
developed during the test and consequently a lower ultimate displacement. As discussed
before, they predict fairly well the length of the equivalent plastic hinge. It is important to
remind that these empirical formulae refer mainly to elements with dominating flexural
response. When shear has a significant contribution to the total displacement, it becomes
difficult to estimate the displacement at collapse, as the interaction between flexure and
shear is not fully understood [Calvi & Pavese, 2003].
3.6. PERFORMANCE OF HOLLOW CROSS-SECTION BRIDGE PIERS
3.6.1. Effect of mechanical and geometrical parameters
The experimental data discussed in this chapter constitute a database that can be used for
parametric studies. The database consists of tests on 112 specimens, among which 75%
have rectangular cross-section and the remaining 25% have circular cross-section. 27% of
the specimens in the database are classified as seismic-deficient, while 73% of the
specimens are code-designed. Seismic-deficient piers reflect design procedures without
seismic provisions. Code-designed piers are detailed according to modern seismic codes,
or correspond to retrofitted piers. The complete list of the properties and the deformation
capacities of the piers is given in Appendix A. The population of the examined piers is
certainly not extensive, but the maximum effort was dedicated to collect all the published
material. The limited amount of available data, particularly for seismic-deficient piers,
adds to the importance of the experimental campaign described in this thesis and calls for
further investigation. Anyhow, the amount of available information is a restriction to the
generalisation of the observations made in the following.
The examined parameters are the aspect ratio, amount of transverse reinforcement,
overlapping length and axial load. The deformation capacity is quantified by the lateral
drift ratio, because of the problems related with the definition of displacement ductility.
Figure 3.22 presents the drift capacities of the piers versus the aforementioned parameters
for both seismic-deficient and code-designed bridge piers with hollow cross-section.
Although large scatter is observed in all cases, general trends can be identified.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
73
The deformation capacity of bridge piers with hollow cross-section increases with aspect
ratio and transverse steel ratio. As seen in Figures 3.22a and c, the increase is more
pronounced for seismic-deficient piers. This could be due to the fact that existing piers
are susceptible to shear failure, which is controlled by these two parameters. On the other
hand, code-designed piers are detailed in order to develop their full flexural capacity and
sustain large deformations before shear damage occurs. This is visually verified by linear
trend lines and lines defining a lower bound of drift capacity (bold and dotted lines,
respectively, in Figure 3.22).
It is interesting to note that in the presence of lapped splices, the drift capacity is
uδ < 2.0%, almost independently of the overlapping length, Figure 3.22e. As seen also in
Chapter 2 for piers with solid cross-section, it is possible to improve the deformation
capacity of hollow piers with lapped splices by jacketing, but failure will be always due to
loss of bond and the performance would only slightly improve, Figure 3.22f. Therefore,
alternative objectives, e.g. shifting of the critical cross-section, should be examined. As
expected, the deformation capacity decreases with axial load, for both groups of piers,
Figures 3.22g and h. The decrease seems to be more rapid for seismic-deficient piers,
probably because of the inadequate confinement.
To highlight the complexity of interaction between the parameters that affect the
deformation capacity of RC members, an empirical expression is recalled. This
expression was fitted to the results of more than 1000 experimental tests on beams,
columns and walls with flexure-dominated behaviour [Panagiotakos & Fardis, 2001]. The
mean value of the ultimate chord rotation, uθ , of an RC member under cyclic loading is
sx yw c
0.2 0.425f / fs
u st,w c
Lmax(0.01, ')0.3 f 25
max(0.01, ) hαρν ω θ = α ω
(3.50)
where st,wα is a coefficient for the steel of longitudinal rebars, ν is the normalized axial
load, ω and 'ω are, respectively, the mechanical reinforcement ratios of the tension and
compression longitudinal reinforcement, cf is the concrete strength, sL / h is the shear
span, α is a confinement effectiveness factor, sxρ is the transverse steel ratio and ywf is
the yield stress of steel. The coefficient of variation of the ratio of experimental values to
predictions of the empirical relation was 46%.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
74
0
2
4
6
8
0 5 10 15
Aspect ratio
Drif
t rat
io (
%)
(a)
0
2
4
6
8
0 5 10 15
Aspect ratio
Drif
t rat
io (
%)
(b)
0
2
4
6
8
0.0 0.5 1.0 1.5 2.0 2.5
Transverse reinforcement ratio (%)
Drif
t rat
io (
%)
(c)
0
2
4
6
8
0.0 0.5 1.0 1.5 2.0 2.5
Transverse reinforcement ratio (%)
Drif
t rat
io (
%)
(d)
0
2
4
6
8
0 10 20 30 40
Overlapping length (db)
Drif
t rat
io (
%)
(e)
0
2
4
6
8
0 10 20 30 40
Overlapping lenght (db)
Drif
t rat
io (
%)
(f)
0
2
4
6
8
0.0 0.2 0.4 0.6
Normalised axial load
Drif
t rat
io (
%)
(g)
0
2
4
6
8
0.0 0.2 0.4 0.6
Normalised axial load
Drif
t rat
io (
%)
(h)
Figure 3.22. Performance of seismic-deficient (left column) and code-designed (right
column) bridge piers with hollow cross-section
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
75
A general form of this expression is adopted in EC8-3 [CEN, 2003b] and takes the form
sx yw c d
0.2
slu st,w cyc wall c
0.425f / f 100s
max(0.01, ')(1 0.38 )(1 )(1 0.37 )0.3 f
1.7 max(0.01, )
L25 1.45
h
ν
αρ ρ
α ωθ = α − α + − α ⋅ ω
⋅
(3.51)
where cycα is equal to 0 for monotonic and 1 for cyclic loading, slα is equal to 1 if there
is slippage of the longitudinal rebars and 0 if there is not, wallα is equal to 1 for walls and
0 for beams and columns and dρ is the diagonal steel ratio.
3.6.2. Estimation of deformation limits
An attempt is made in this section to exploit the collected experimental data in order to
identify deformation limits for piers with hollow cross-section. The drift ratio is chosen as
the deformation indicator. The drift capacity is presented in Table 3.13 for different
failure modes and two groups of piers, namely seismic-deficient and code-designed.
Failure due to flexure is evidenced by concrete crushing, buckling or failure of
longitudinal rebars, fracture of longitudinal rebars or a combination of the above. Shear
failure is identified by diagonal concrete cracking, loss of bond at the region of lapped
splices and in extreme cases by failure of stirrups. Combined flexural/shear failure mode
corresponds to crushing of concrete and/or buckling of steel with significant shear
damage. The minimum, u,minδ , and mean, uδ , values of drift are presented along with the
number of piers in each group.
Table 3.13. Drift capacity of piers with hollow cross-section
Seismic-deficient Code-designed
No of
specimens u,minδ (%) uδ (%)
No of
specimens u,minδ (%) uδ (%)
Flexure 16 0.9 2.9 57 1.3 3.6
Shear 8 0.8 1.9 10 2.1 3.7
Flexure/Shear 6 1.0 2.1 4 2.4 3.7
All
specimens 30 0.8 2.5 71 1.3 3.6
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
76
Seismic-deficient hollow piers are shown to have limited deformation capacity, ud =
2.5%. Among them, piers with flexure-dominated failure mode have a somehow
acceptable deformation capacity, uδ = 2.9%, while shear-deficient piers and piers with
combined failure mode have very limited deformation capacity, uδ = 1.9% and uδ = 2.1%,
respectively. These differences are not the same pronounced for the case of code-
designed piers. The mean value of drift capacity is in the order of uδ = 3.6%,
independently of the failure mode. This is because code-designed and retrofitted piers are
able to develop their full flexural capacity and are detailed in order to sustain large
deformations before experiencing shear damage.
As a conclusion, failure of hollow bridge piers without seismic design is expected at uδ =
2.9% or uδ = 1.9% for flexural or shear failure mode, respectively. This demonstrates the
limited deformation capacity and the need to study appropriate retrofit solutions. On the
other hand, code-designed hollow piers are expected to have sufficient deformation
capacity in the order of uδ = 3.6%. This can be used as target value when designing the
retrofit of a deficient pier.
A number of experimental tests on bridge piers with solid cross-section were examined,
as described in Chapter 2. The geometrical and mechanical properties of these piers are
presented in Appendix A, along with their displacement ductility and drift capacities.
Seismic-deficient piers with solid cross-section show limited deformation capacity, uδ =
2.8%, similar to piers with hollow cross-section. Failure in seismic-deficient piers is most
of the times due to limited shear resistance and/or inadequate confinement. These
phenomena result in similar deformation capacities independently of the type of cross-
section (hollow or solid).
On the other hand, code-designed piers with solid cross-section have larger deformation
capacity, uδ = 4.8%. This value is more than 30% higher than the corresponding value for
code-designed piers with hollow cross-section. The database of piers with solid cross-
section contains a number of specimens with large drift capacity, uδ > 7, which contribute
to the higher mean value. These specimens correspond to piers with circular cross-section
designed according to American seismic codes. Apart from a few specimens with heavy
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
77
retrofit or special concrete, they have mechanical and geometrical properties that are
uniformly distributed in the whole range of values. On the contrary, most of the code-
designed piers with hollow cross-section have low values of aspect ratio and transverse
reinforcement ratio, which possibly explains the smaller deformation capacity.
In addition, confinement plays an important role. Circular hoops or spirals provide very
good confinement for the concrete core in solid circular piers. The effectiveness of
confinement is reduced for hollow piers with rectangular hoops or crossties and even
more for circular hollow piers with confinement reinforcement only on the external face.
In the latter case and when the position of the neutral axis is within the less confined part
of the cross-section, limited deformation capacity and brittle failure are expected. This
behaviour has been experimentally observed [Zahn et al., 1990; Ranzo & Priestley, 2000].
Other factors, such as the mechanical properties of steel, should be considered. Compared
to steel used in Europe, steel used in the USA results in better performance in terms of
plasticity spreading and resistance to buckling [Priestley et al., 1996].
In conclusion, seismic-deficient piers with solid and hollow cross-sections seem to have
similar deformation capacities because of the shear flexural mode. Solid piers with
seismic design seem to have slightly better deformation capacity than their hollow
counterparts. It is however possible, as proved by experimental testing, to obtain very
good seismic performance of hollow piers when sufficient confinement reinforcement is
provided for the whole compression zone in the form of closed hoops.
3.7. CONCLUDING REMARKS
The failure mode of the tested specimens deserves further discussion. Both piers were
designed with neither capacity design considerations nor seismic detailing. As a result,
they failed due to a complex combination of phenomena. Observing the distribution of
curvature demand along the height of the pier models, Figures 3.8 and 3.18, a collapse
mechanism with multiple hinges is evidenced. One plastic hinge forms at the base of the
piers, another initiates above the first lapped splices, and for the tall pier, a third plastic
hinge fully develops above the bar cut-off. Then, flexure, shear (tension shift) and steel-
to-concrete bond phenomena interact in different parts of the piers and at different
magnitudes of lateral displacement and all contribute to the failure mode. The failure
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
78
mode is flexure-dominated with limited deformation capacity, because of the small
amount of vertical reinforcement and the concentration of deformation in thin slices that
impose large strain on the rebars and lead them to premature failure. In addition, the lack
of protection against buckling results in collapse of the vertical rebars due to low-cycle
fatigue. Due to the large number of existing bridges with the same characteristics, the
response of bridges with multiple hinges has been recognised as a problem that has to be
addressed [Kawashima, 2000a].
The previous observations point to a complex behaviour that will probably obscure
numerical modelling, as well as eventual repair and/or strengthening interventions. With
reference to numerical modelling, the contribution of different phenomena to the failure
mode might prove simplified models to be inadequate, as will be discussed in a following
chapter. As far as retrofit is concerned, the initiation of different failure modes at distinct
levels of lateral displacement calls for the need for a global study that takes into account
possible failure modes that did not appear during the tests. For example, because of the
small amount and inadequate anchorage of horizontal reinforcement and should the
vertical reinforcement ratio have been higher, say equal to the minimum requirements of
modern seismic codes, a more pronounced contribution of shear would have been
expected. Retrofit considerations will also be further discussed in a dedicated chapter.
The comparison between experimental and empirical values for the yield and ultimate
displacement, as well as the equivalent plastic hinge length, is useful for the validation of
the empirical formulae for the case of elements with elongated hollow cross-sections,
distributed reinforcement and lapped splices within the potential plastic hinge region. As
far as yield displacement is concerned, the predictions are in fair agreement with the
experimental values. On the other hand, the ultimate displacement strongly depends on
the equivalent plastic hinge length. For the tall pier and considering the critical cross-
section above the cut-off, the empirical value of the equivalent plastic hinge length was
close enough to the experimental one. For the short pier, where an extremely limited
plastic hinge length was observed, due to the lapped splices that did not allow the yield
stress to develop in the whole overlapping length, the empirical formulae proved
inadequate. For piers with multiple hinges, the contribution of all of them (possibly not
fully developed) must be considered and this further complicates the estimation of the
deformation capacity.
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
79
These empirical formulae seem valid for hollow cross-sections, but not for members with
lapped splices: they failed to predict the plastic hinge length of the short pier not because
of some phenomenon related to the geometry of the cross-section, but because of the
presence of lapped splices. For members with lapped splices without sufficient
development length and/or confining reinforcement within the plastic hinge zone, the
equivalent plastic hinge length is much smaller than the predicted values.
The contribution of shear to deformation is not taken into consideration by simplified
empirical procedures for the calculation of strength and deformation capacity, with the
exception of a recent proposal [Hines et al., 2002a]. Shear displacement is actually
expected to be an important fraction of the total displacement for piers with elongated
hollow cross-section. For the short pier, A70, which showed a predominantly flexural
behaviour, the ratio of shear to flexural deformation was in the order of 0.3. In the case of
the tall pier, A40, for which shear phenomena were marked and appeared on various parts
of the element, the ratio increased with top displacement and therefore a constant value
cannot be considered.
Figure 3.14 shows that different theoretical or empirical expressions result in significantly
different estimations of damping for significant ductility demand, bµ > 2. The relations
that assume elastic-perfectly plastic response were accurate for the well-designed PREC8
pier, whereas the expressions that assumed pinched hysteretic curves were more suitable
for the short pier, A70, that did not have proper seismic detailing. In innovative
displacement-based design and assessment procedures and also in the design of isolation
systems for existing bridges, an overestimation of the equivalent damping would lead to
unsafe estimate of the spectral displacement, at least in the low period range. The
implication is that the hysteretic behaviour of the pier should be known a priori for the
assessment of existing structures, or should be guaranteed by proper detailing for the
design of new structures.
The importance of seismic detailing, principally of the minimum requirements for the
longitudinal and vertical reinforcement (see Appendix B) is verified by comparing the
behaviour of the existing short pier, A70, to that of the PREC8 pier, designed according
to EC8. As observed also in previous experimental campaigns, the amount of
confinement reinforcement plays a key role in the deformation and energy-dissipation
capacities of piers with hollow cross-sections: lack of adequate confinement results in
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
80
pinched hysteretic curves. The short pier, A70, had smaller ductility capacity and almost
half the drift capacity of the PREC8 pier.
Considering the definition of displacement ductility, it is concluded that the conventional
procedure based on a bilinear approximation of the envelope force-displacement diagram
leads to very high values of ductility capacity for pier A70, u,bµ = 8.0, which contrasts
with the value u,EC8µ = 5.3 and the limited ductility value, u,tµ = 3.2, calculated from the
yielding displacement corresponding to the stabilization of strength (‘total yielding’). It is
noted that the ductility capacity of the PREC8 pier was µu,PREC8 = 6.0, which, compared to
the values for the short pier, confirms that ductility per se can be misleading or
contradictory as a meaningful parameter of the deformation capacity of elements. On the
contrary, drift and energy-dissipation capacities can fully characterize the cyclic
behaviour of these bridge piers.
The quantification of damage is an important issue when assessing the available capacity
and when deciding on the target capacity for the retrofit of as-built elements. Depending
on the choice of the damage indicator, one might have an incomplete, or unrealistic,
estimation of the deformation capacity of the structure. Actually, the damage indicators
related to a single behaviour parameter, such as stiffness degradation or plastic
displacement, provide information only on the single parameter and not a complete
picture of the damage state of the element. Among the ones examined in this work, the
drift ratio better describes the performance of the piers. A combination of performance
parameters seems more appropriate. In fact, the Park and Ang Damage Index better
correlates with the observed damage, as it takes into consideration both maximum
displacement and dissipated energy. Nevertheless, either experimental or non-linear
numerical investigation is required in order to calculate this damage index.
A database of experimental results on specimens of bridge piers with hollow cross-section
has been built. The limited available information on piers with this geometry adds to the
significance of the research presented in this thesis. Based on the collected information, it
is concluded that hollow piers without seismic design are expected to have limited
deformation capacity, mean values of drift uδ = 2.9% for flexural failure mode and uδ =
1.9% for shear failure mode. These values show that existing bridge piers need upgrading.
On the other hand, hollow piers with sufficient confinement of the compression zone, as
EXPERIMENTAL ASSESSMENT OF HOLLOW BRIDGE PIERS WITH SEISMIC DEFICIENCIES
81
guaranteed by modern seismic design, are expected to have stable behaviour and large
deformation capacity, uδ > 3.6%.
These considerations give also a hint on the objectives of retrofit. The first seismic
deficiency that has to be addressed is the limited shear resistance. This can be easily
improved (by additional shear reinforcement) and then failure would be due to flexure
alone. Also in that case, a relatively limited deformation capacity is expected, uδ = 2.9%.
Better performance is expected for piers designed according to modern seismic codes,
due to the effective confinement of the compression zone. Therefore, confinement should
be a target when designing a retrofit intervention. The effectiveness and limitations of
confinement provided by fibre-reinforced polymer jackets will be studied through
extensive numerical studies in Chapter 6.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
82
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
83
4. SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
4.1. INTRODUCTION
Severe earthquake-induced damage on bridges, apart from the possible human victims,
results in economic losses in the form of significant repair or replacement costs and
disruption of traffic and transportation. For the above reasons, important bridges, which
are components of lifelines, are required to suffer only minor, repairable damage and to
maintain immediate functionality during the post-earthquake emergency. However, the
greatest part of existing bridges in Europe and other economically advanced areas has
been designed before their seismic response had been fully understood and modern codes
introduced. Consequently, they are expected to show unsatisfactory earthquake
performance because of limited deformation capacity, as well as poor hysteretic
behaviour and to represent a source of risk in earthquake-prone regions. This is verified
by field observations during all the recent destructive earthquakes, such as the 1987
Whittier Narrows earthquake [Gates et al., 1988], the 1989 Loma Prieta earthquake
[Housner & Thiels, 1990], the 1994 Northridge earthquake [Housner & Thiels, 1995] and
the 1995 Hyogo-Ken Nambu (Kobe) earthquake [Seible et al., 1995a; Kawashima &
Unjoh, 1997].
The need to assess the seismic capacity of existing bridge structures has been recognised
[Calvi & Pinto, 1996] and various research documents that contain overall considerations,
e.g. [Priestley et al., 1996; Pinto & Monti, 2000], or propose simplified methods, e.g.
[Fajfar & Gašperšic, 1996; Priestley & Calvi, 1997], have been produced. In the USA the
results of the research have been codified in normative documents, such as the FHWA
Seismic Retrofit Manual [Buckle & Friedland, 1995] and the HAZUS probabilistic
method for earthquake risk assessment [FEMA, 1999]. European codes, namely EC8,
lack such provisions, with the exception of EC8-3 [CEN, 2003b] that refers to building
structures. A possible reason for this is the number of parameters that should be
considered (the most important of which were discussed in Chapter 2) and the scarce
verification of simplified procedures for the assessment of complex bridge structures. For
these reasons, in practice each case is considered separately and general guidelines,
similar to those for the design of new structures, are followed.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
84
Figure 4.1. Talübergang Warth Bridge, Austria
As a contribution to the assessment of existing bridge structures, the results of an
experimental campaign are presented in this chapter. Within the VAB research
programme, a series of pseudodynamic (PSD) tests were performed on a large-scale
(1:2.5) model of an existing highway bridge. The project was focused on the Talübergang
Warth Bridge, schematically shown in Figure 4.1, which is considered representative of
European highway bridges. Three PSD tests were performed for input motions with
increasing amplitude.
The objective of the research was, on one hand, to develop and implement the
substructuring technique with non-linear models for the numerical substructure in PSD
testing, referred to in the following as non-linear substructuring. This advance in PSD
testing at the ELSA laboratory allowed for testing of the complete bridge system using
the existing laboratory capacity and reducing considerably the costs of the testing
campaign and set-up (two piers instead of six). On the other hand, the aim was the
seismic assessment of a reinforced concrete (RC) bridge, that presents characteristics
(such as hollow cross-section piers, lapped splices within the potential plastic hinge
region, bar cut-off with insufficient development length of the terminated reinforcement
at not easily accessible heights, low percentage of reinforcement, short overlapping
length, inadequate detailing of horizontal reinforcement and lack of appropriate
confinement reinforcement) commonly found in existing bridges in Europe and Japan
[Hooks et al., 1997]. The contribution of the author to the experimental campaign was
focused on the numerical modelling and exploitation of the test results, therefore these
aspects will be discussed in detail, while the implementation aspects of the testing method
will be briefly recalled.
The PSD testing method is presented first, considering the particular cases of
asynchronous input motion, non-linear substructuring and continuous PSD testing. The
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
85
pre-test numerical simulations of the PSD tests are presented next and the analytical
results are compared to the experimental ones, as well as to in-situ measurements of the
dynamic properties. The results of three earthquake tests are presented in terms of
hysteretic curves, dissipated energy and deformation demands. Then, the seismic
performance of the bridge is assessed for the different amplitudes of the earthquake
motion and a preliminary estimation of vulnerability curves for the bridge is performed.
The distribution of damage is discussed along with various definitions for irregularity of
bridges. In the last section simplified assessment procedures, both probabilistic and
deterministic, are checked against the experimental results. Introducing a correction for
the given structure, the simplified procedures are applied for the assessment of the bridge
for earthquake demand compatible with the EC8 spectrum for a high-seismicity zone.
4.2. THE PSEUDODYNAMIC TESTING METHOD
4.2.1. The pseudodynamic testing method
A PSD test [Shing & Mahin, 1985; Nakashima et al., 1992; Donéa et al., 1996] is one,
which, although carried out quasi-statically, uses on-line computer calculation and control
together with experimental measurement of the actual properties of the structure to
provide a realistic simulation of the dynamic response. For simulating the earthquake
response of a structure, a record of a real or artificially generated earthquake ground
acceleration history is given as input data to the computer running the PSD algorithm.
The horizontal displacements of the controlled degrees of freedom (DOFs), where the
mass of the structure can be considered to be concentrated, are calculated for a small time
step using a suitable time-integration algorithm. These displacements are then applied to
the tested structure by servo-controlled hydraulic actuators fixed to the reaction wall.
Load cells on the actuators measure the forces necessary to achieve the required
displacements and these structural restoring forces are returned to the computer for use in
the next time step calculation. Because the inertia forces are modelled, there is no need to
perform the test on the real time-scale, thus allowing very large models of structures to be
tested with only a relatively modest hydraulic power requirement.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
86
4.2.2. The a-Operator Splitting scheme
Consider the following system of semi-discrete differential equations
Ma Cv r(d) f+ + = (4.1)
which describe the motion of a structure, where a, v and d represent the acceleration,
velocity and displacement vectors, r and f the structural internal and external force
vectors, M and C the mass and damping matrices, respectively. In the case of
synchronous seismic loading, a, v and d represent the motion of the structure in a
reference frame which is relative to the uniform ground motion. The seismic action is
taken into account by means of an inertial contribution to the external force vector
baseMIaf −= (4.2)
where basea is the intensity of the base acceleration and I is a vector that accounts for the
direction of earthquake loading. To solve the system given by Equation 4.1, a numerical
step-by-step integration technique is adopted: in this work it is the so-called α method
implemented by means of an Operator Splitting (OS) technique. This scheme is
unconditionally stable and does not require iterations.
According to the α method [Hilber et al., 1977], the displacement and velocity vectors at
step n + 1 can be written in terms of both the acceleration vector and the previous step
values
n 1 n 1 2 n 1d d t a+ + += + ∆ β% (4.3)
n 1 n n 2 nd d tv 0.5 t (1 2 )a+ = + ∆ + ∆ − β% (4.4)
n 1 n 1 n 1v v t a+ + += + ∆ γ% (4.5)
n 1 n nv v t(1 )a+ = + ∆ − γ% (4.6)
with
2(1 ) / 4β = − α (4.7)
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
87
(1 2 ) / 2γ = − α (4.8)
for 1 3 0≤ α ≤ and then introduced into the following time discrete system of equilibrium
equations
n 1 n 2 n n 1 n n 1 nMa (1 )Cv Cv (1 )r r (1 )f f+ + + ++ + α − α + + α − α = + α − α (4.9)
This scheme is implicit since n 1d + depends on n 1a + , related to n 1r + , which is a function of n 1d + , and then it implies an iterative procedure. It is however possible to implement the
method without iterating by using an OS method [Combescure & Pegon, 1997], based on
the following approximation of the restoring force n 1r +
)d~
K)d~
(r~(dK)d(r 1nI1n1n1nI1n1n ++++++ −+≈ (4.10)
where IK is a stiffness matrix, generally chosen as close as possible to the elastic one, EK , and in any case, for stability reasons, higher or equal to the current tangent stiffness, TK (d) , of the structure. Note that the digital PSD experimental set-up clearly offers all
what is needed for an accurate measurement of the elastic characteristics of the structure
to be tested or of its current stiffness at the beginning of any test.
All useful quantities being known at time nt , the step-wise operations for reaching the
time n 1 nt t t+ = + ∆ are
(i) Compute (prediction phase) n 1d +% and n 1v +% .
(ii) Apply (control phase) the displacement n 1d +% to the tested (and eventually the
numerical) structure in order to get (measuring phase) the restoring force n 1r +% .
(iii) Solve for an+1 the system of linear equations
n 1 n 1ˆM̂a f+ + +α= (4.11)
where the pseudo-mass matrix M̂ and the pseudo-force vector n 1f̂ + +α are given by
2 IM̂ M t(1 )C t (1 )K= + γ∆ + α + β∆ + α (4.12)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
88
and
n 1 a n 1 n n 1 n 2 I nˆ ˆf (1 a)f af ar aCv (1 a)Cfa(?? tC ß? t K )a+ + + += + − + + − + +% % (4.13)
(iv) Compute (correction phase) n 1d + and n 1v + .
Note that the computation, and possibly the factorisation of M̂ , which usually does not
depend on the time, may be performed during the initialisation phase of the algorithm,
before entering the time stepping loop.
4.2.3. The substructuring technique
It is generally not feasible to test structures as large as bridges or offshore platforms.
However, earthquake loading often generates severe damage only in parts of the structure
and the rest of the structure could be modelled via finite elements. Therefore, it is useful
to combine PSD testing of only a part of the structure, the tested substructure, together
with an adequate time-integration of the equations of motion for the model of the rest of
the structure, the modelled substructure. To this purpose a substructuring technique is
proposed [Dermitzakis & Mahin, 1985], considering either synchronous or asynchronous
base motion. The modelled substructure must be approximated by an adequate numerical
model and the time-integration scheme must be applied to its spatially discrete equations
of motion.
The strategy adopted in the past [Pegon & Pinto, 2000] was to run two processes in
parallel. The one responsible for the PSD algorithm applied to the tested structure,
running in the master PSD computer and the other, responsible for the modelled structure,
running in a remote workstation. The communication between these two processes used
standard network capabilities. In the meantime, the experimental side was updated to run
the continuous PSD method, where the motion of the structure is controlled every 1 msec
(10-3 sec).
4.2.4. Substructuring in the case of asynchronous motion
For the PSD tests with asynchronous input motion [Pegon, 1996a], special attention has
been devoted to the mathematical and implementation aspects, which is the object of this
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
89
section. In fact, the PSD testing with substructuring for asynchronous motion is not a
trivial extension of the case with synchronous motion. The main difficulty at the
mathematical level comes out from the fact that the structure in the laboratory can only be
tested in a reference frame relative to the earthquake motion because its base is always
fixed to the floor. Then, in order to realize a meaningful test on a structure (tested and
modelled substructures) undergoing an asynchronous motion, only physically
unconnected parts of the tested structure can be submitted to different base accelerations.
This condition is easily verified for bridges: the tested substructure consists of a set of
different piers, which do not interact one with the others, apart through the modelled
deck.
Equation 4.1 may describe a relative or an absolute motion. The description with a
relative motion is the most widely used in earthquake engineering. The base of the
structure of interest is considered to be subjected to a uniform base acceleration field
basea . The basic principle of dynamics is expressed in a reference frame, which follows
this ground motion. The motion of the structure is originated by the inertial forces
considered as being part of the external force vector f, see Equation 4.2. The relative
motion description is quite natural since it expresses directly the contribution of the
structure response to the overall motion.
The description with an absolute motion is scarcely used, only when the other approach is
impossible. This is the case of an asynchronous ground motion where the base
acceleration field changes spatially from point to point. The motion of the structure is
now originated by the motion of some of its internal points. In consequence, provision
should be made while performing the discretisation of the structure not to eliminate the
ground connecting DOFs and to subject them to the convenient base acceleration basea .
Clearly, in the case of synchronous motion, both descriptions lead to the same results in
terms of intensive variables (internal stress state, for instance).
4.2.5. The continuous pseudodynamic testing with non-linear substructuring
In the conventional PSD test procedure, the actuator motion (S-ramp phase) is stopped
when the test specimen reaches the target displacement (hold period), so that the reaction
force can be measured and the next target displacement computed. On the contrary, in
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
90
continuous PSD testing, the servo-controller moves the actuator continuously. The
continuous PSD testing avoids the load relaxation problem and improves the quality of
the results. It allows a considerable reduction of the test duration.
The continuous PSD method is implemented by means of a synchronous process (in
electronics terms: communication to the outside is clock-governed) with short control
period and small time step. This introduces some difficulties for the implementation of
the substructuring technique. First, if the numerical part is complex, the analytical process
is unable to perform even an elastic computation during a control period of the
experimental process. In addition, it is not evident that using the same explicit scheme for
both the experimental and analytical processes would allow to obtain stable results.
Finally, the exchange of information between the two processes should not add a too
large overhead.
The experimental process of the continuous PSD method is synchronous and then unable
to wait for information coming from the analytical process. Thus, a parallel inter-field
procedure should be used. For the simple inter-field procedure, see Figure 4.2a, the
analytical part is advanced with a large time step, t∆ , using, possibly, at each new step
the acceleration, velocity and displacement of the connection points obtained through the
experimental process at the end of the previous large step. On the contrary, the
experimental part uses at each subcycle, tδ , as an additional external force, what was
generated in the analytical part at the end of the previous large time step.
The main drawback of this approach is that the force coming from the numerical part is
not well synchronized with the external loading of the physical model. This delay
introduces damping when the experimental and analytical processes are strongly coupled,
i.e. have similar mass. For the case of the numerical part having bigger mass than the
experimental (representative of bridges with experimental piers and numerical deck),
slowly increasing oscillations, associated with higher frequencies, were observed.
Introducing subcycling in the experimental part does not significantly improve the results
[Pegon & Magonette, 1999].
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
91
Figure 4.2. Parallel procedures: simple inter-field procedure (a), improved inter-field
procedure (b) and intra-field procedure (c)
An improved scheme is introduced, in which the modelled structure is integrated with a
time step 2? T, see Figure 4.2b. This allows to know the kinematics of the numerical
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
92
structure one large time step in advance with respect to the tested part. Then, an
approximation of the additional force n 1r + at time n 1t + is known before starting the
subcycling between nt and n 1t + . It is thus possible to drive the experimental structure
with more updated information than with the basic scheme. The force n m / Mr + that is used
in the experimental process at each subcycle level can be simply
n m / M n 1r r+ += (4.14)
However, since nr is also known, it is possible to use both quantities to improve the time-
integration accuracy. The additional force that enters the algorithm for the experimental
substructure can be interpolated as
n m / M n 1 n 1 nr r (1 m / M)( r r )+ + += − − − (4.15)
Using this scheme, improved results are obtained, compared to those obtained with the
simple scheme. However, the method is conditionally stable, depending on the
distribution of mass on the connecting DOFs and on the time step. The error is reduced
when reducing the time step, but convergence is slow. The number of subcycles does not
seem to significantly improve the results [Pegon & Magonette, 2002].
Considering the time-integration schemes for non-linear substructuring, it was found that
explicit schemes can be used for the analytical part only when a small number of DOFs is
involved, whereas implicit schemes depend strongly on the local nature of the problem
and could result in significant local deviations from the medium time step duration
[Pegon & Magonette, 2002].
4.2.6. Implementation for the Talübergang Warth Bridge tests
At the time of testing the inter-field procedures were not yet robust enough to be applied
in large-scale testing of bridge structures. In fact, there was evidence of the instability of
the global problem, which could not be solved within the time frame for the bridge tests.
The instability issue was subsequently tackled using an intra-field partitioned scheme
with different time steps and assuming a kinematic continuity between the subdomains
(numerical and experimental) expressed in terms of velocity [Gravouil & Combescure,
2001; Pegon & Magonette, 2002]. The problem was solved using an intra-field procedure
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
93
(a unique main process delegating all the tasks to numerical and experimental processes)
rather than an inter-field procedure (two or more processes running in parallel), see
Figure 4.2c. The experimental and numerical parts are advanced in time (subcycling is
introduced in both) and exchange information at the end of the time step, through the
Modelling Master workstation.
The implementation of the substructuring technique for the Talübergang Warth Bridge
PSD tests is schematically presented in Figure 4.3, which includes three main workstation
groups, namely: the Experimental Master, the Physical models and the Numerical models.
The Numerical models group includes the Modelling Master workstation (holding the
linear model of the deck and the lateral DOFs of the piers), which performs the time-
integration of the motion of the whole bridge, using the a-OS scheme. Each pier, either
modelled or tested, is condensed on two DOFs, namely the base and the top displacement.
Obviously, it is the difference of the displacement of these two DOFs that is transmitted
as target to the controller of the pier. The force, which is measured and transferred back to
the numerical process, is then associated to the two DOFs.
As soon as the predicted displacement, d% , see Equation 4.3, is computed, the
displacements to impose to the analytical piers are sent to two other computers
(Modelling Slaves) holding the non-linear model of the piers and using an iterative
process to equilibrate the internal nodes of the discrete model. The displacements to
impose to the experimental piers are sent to the Experimental Master, which in turns
generates the appropriate curve (S-shaped displacement ramp and elongated sustain-level
plateau) to be reached by two controllers, each of them responsible for one experimental
pier.
The communication is implemented in such a way that if the numerical piers require a
large number of iterations at the interior of one step, the controllers of the physical piers
can wait (further small time steps are performed) for the input of the next time step.
Actually, the time required to perform the numerical integration for the substructured
piers was always inferior to that value during the pre-test calculations. When the
analytical piers are equilibrated and the experimental piers attain the target displacement,
the force levels reached at the ends of each pier are transmitted back to the Modelling
Master in order to compute the next displacement. The communication between the
workstations was done via the laboratory local network.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
94
Figure 4.3. PSD test with substructuring of the Warth Bridge at the ELSA laboratory
4.3. PRE-TEST NUMERICAL SIMULATION
4.3.1. Numerical models for the substructured piers
In order to reduce the computational demand for the PSD tests and to increase the
robustness of the numerical models and procedures, a simple, yet accurate numerical
model should be used for the substructured piers. The most accurate choice is a refined
tri-dimensional finite element model with realistic constitutive laws for the materials and
the steel-to-concrete interface. Such a model, though, would demand exaggerated
computation time. For this reason, it was decided to use a fibre/Timoshenko beam
element [Guedes et al., 1994] implemented in the finite element code Cast3m [Millard,
1994], as a compromise between accuracy and simplicity. Taking into account the
symmetry of the geometry and the loading, a bi-dimensional (2D) beam element was
used. The constitutive law for concrete fibres was cyclic non-linear, while for the steel
fibres an elasto-plastic model with hardening was considered.
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
95
The vertical rebars in the specimen were uniformly distributed along the web and were
expected to be under tension gradually as the neutral axis fluctuates. Accordingly, in
order to enable a more realistic representation of the distribution of reinforcement, to
avoid eventual numerical problems and to reduce the computation time during the PSD
tests, the mesh for the cross-section was refined and new elements, with less integration
points, were implemented.
The experimental results discussed in the previous chapter, as well as the results of
numerical simulations using a 2D damage model [Faria et al., 2001; Pouca, 2001], were
used to calibrate the models for the substructured piers. The fibre model was adequate for
the case of the short pier, A70, that had a prevailing flexural behaviour; pier A20 was
expected to show similar behaviour. Concerning the tall pier, A40, the damage model
gave results similar to the experimental ones, while the fibre model was unable to
accurately represent the shear deformation in the lower part of the specimen. Diagonal
cracking caused tension shift that, combined with the bar cut-off and the inadequate
development length of the terminated rebars, dictated that the resisting moment of the
critical cross-section was surpassed before the base cross-section developed its full
strength. Piers A30, A50 and A60 had curtailment of the longitudinal reinforcement and
were therefore expected to show a similar failure mode. In the end, it was decided to use a
fibre model with reduced area of longitudinal reinforcement in the flange above the
critical cross-section for the substructured piers A30, A50 and A60. The amount of
vertical reinforcement for the piers is presented in Table 4.1.
Table 4.1. Longitudinal reinforcement steel ratio and characteristic values of
displacement for the scaled bridge piers
A-A B-B C-C
h (m) sρ (%) h (m) sρ (%) h (m) sρ (%) cu (m) yu (m) uu (m)
A20 2.76 0.9 7.04 0.6 2.12 0.3 0.006 0.065 0.187
A30 3.52 0.7 7.08 0.3 4.96 0.2 0.016 0.038 0.372
A40 1.20 1.0 2.30 0.7 10.5 0.2 0.013 0.081 0.230
A50 1.60 0.7 7.24 0.3 5.56 0.3 0.009 0.044 0.326
A60 1.48 0.8 2.68 0.4 7.84 0.3 0.007 0.043 0.179
A70 6.50 0.4 0.003 0.011 0.100
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
96
Table 4.1 presents also the displacement at characteristic points, namely cracking ( cu ),
yielding ( yu ) and ultimate capacity ( uu ), as resulting from numerical pushover analysis.
These values will be later used to identify the damage of the numerical piers. The yield
displacement, yu , is identified by assuming an elastic-perfectly plastic equivalent system,
where the initial stiffness is the tangent stiffness at 75% of the maximum force. The
ultimate displacement is the displacement at which the resistance drops by more than
20% or the steel fibres collapse [Park, 1989]. Numerical modelling considerations for the
bridge piers will be further discussed in the following chapter.
4.3.2. Numerical model of the bridge structure
Description of the model
Pre-test dynamic analyses were performed with the aim of predicting the earthquake
response of the bridge. The results of the dynamic analyses were compared to the results
of the numerical simulation of the PSD test in order to assess the ability of the latter to
represent a real earthquake test. In order to better represent the PSD test, the numerical
model of the bridge was accordingly simplified in comparison to the real structure.
The deck was modelled using a Fibre/Timoshenko beam element with the cross-section
schematically shown in Figure 4.4. The height of the scaled model was 2.00 m and the
width of the top and bottom flanges was 5.00 m and 2.48 m, respectively. The area of the
cross-section was A = 1.57 m2. The moments of inertia were yyI = 3.35 m4 and xxI = 3.18
m4 by the vertical and horizontal axes, respectively, and the rotational moment of inertia
was J = 2.72 m4. The deck was considered to remain elastic, in accordance with the
requirements of modern codes [CEN, 2002; Caltrans, 1999] for new bridges and
assumptions commonly used in the design and retrofit of bridges [Priestley et al., 1996].
All 6 DOFs by node have been kept in order to better describe the coupling between
translation, bending and torsion of the deck. Nine elements were used along each span.
A 2D beam model with 3 DOFs by node was adopted for the piers, as discussed in detail
in the next chapter. Each pier was divided into an adequate number of elements along its
height; more elements were used in the region where damage was expected and less
elements near the top of the pier, where the behaviour was expected to be essentially
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
97
elastic. The number and distribution of the elements was chosen such to obtain reliable
results with the minimum computational demand. The foundation blocks of the piers were
considered to remain elastic and to be fixed along all 3 DOFs. The assumption of
perfectly fixed base was exact for the physical models tested in the lab, but may not be
the case for the real structure due to soil-structure interaction. As stated before, all 6
DOFs per node were considered for the deck, whereas only 3 DOFs per node were
considered for the piers; an adequate connection between the corresponding DOFs of the
deck and the top of the piers was established.
Only the vertical and horizontal forces in the direction of testing, in the transverse bridge
direction, were transmitted between the piers and the deck. The abutments constrained the
displacement in all three axes and the rotation by the bridge axis.
During the PSD test only the DOF at the top of the piers (where the mass of the pier is
considered to be lumped) was controlled, therefore a concentrated external load was
applied on the top of the piers to take into consideration the vertical static forces
transmitted by the deck and the pier self-weight. During the test constant vertical forces
were applied on the physical models of the two piers.
Figure 4.4. Cross-section of the deck
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
98
The vertical loading can be considered comprising two parts. The first part results from
the mass of the deck. This is the reaction force at the top of the pier computed by an
elastic analysis. The second part results from the self-weight of the piers. Following the
similitude relations in Table 3.1, the stress at the base cross-section of the prototype and
the model should be equal. This means that the stress applied on the model should be
p p M MM g / A 2.5M g / A= =σ , where M denotes the mass of the pier and A the area of the
base cross-section. However, the mass of the pier, corresponding to a stress equal to
M MM g / A , is already present. Then, the vertical force on the physical piers is calculated
based on the mass of the model multiplied by 2.5 – 1 = 1.5.
Damping matrix
The damping matrix was formulated by considering only the contribution of the deck and
assuming a truncated modal damping [Pegon, 1996b] in order to avoid problems related
to rigid body motion in the case of asynchronous loading. This expression of damping is
based on the commonly used Rayleigh damping, see for example [Clough & Penzien,
1993], that consists of two parts, one proportional to the mass of the structure and the
other proportional to the stiffness
o 1C M K= +α α (4.16)
In the modal space, Rayleigh damping takes the form
R RM RK o 1 ii i i
i2 2α α ω
ξ = ξ + ξ = +ω
(4.17)
where iω is the angular frequency of the i-th mode. The proportionality
coefficients oα and 1α are calculated by setting known values of damping to two modes.
Then the values of modal damping for the significant eigenmodes can be calculated from
Equation 4.17. In fact, this simple expression for the damping matrix leads to different
results whether its computation is performed in the absolute or in the relative space,
because the mass matrix in not rank-deficient. In order to overcome this problem, it was
proposed [Pegon, 1996b] to reproduce the Rayleigh damping by means of a truncated
sum of the modal damping, expressed in the physical space using the stiffness matrix. In
this way, the same results are obtained independently from the absolute or relative frame.
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
99
For the truncated modal damping, the mass-proportional damping is distributed to the n
first modes and then added to the stiffness-proportional damping. In matrix form it writes
n
ii 1
C C−
= ∑ (4.18)
The mass-proportional damping matrix is
Ti i i iC K b K= ψ ψ (4.19)
where ψ contains the eigenvectors and the terms of the diagonal matrix Mb are
i i i ib 2 / k= ξ ω ; i = 1…n (4.20)
In the previous equations, iξ stands for the modal damping ratio, iω denotes the
eigenfrequency and ik represents the modal stiffness corresponding to the i-th mode.
Modal analysis
The numerical model was used to calculate the eigenfrequencies of the bridge structure.
Experimental measurement of the natural frequencies of the bridge structure has been
performed [Flesch et al., 1999; Flesch et al., 2002] by dynamic in-situ tests. The results
are reported in Table 4.2, along with the numerical values. It is reminded that the scaling
factor λ = 2.5 applies to the eigenfrequencies: the values for the model are obtained by
the values for the prototype multiplied by the scaling factor. Since the analytical values
were similar to the experimental ones, the numerical model for the bridge seemed capable
of representing the modal properties of the bridge structure. Figure 4.5 presents the first
eigenmodes of the bridge in the transverse direction.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
100
mode 1
mode 2
mode 3
mode 4
mode 5
mode 6
Figure 4.5. Mode shapes of the bridge
Table 4.2. Eigenfrequencies of Warth Bridge
Mode Measured (Hz) Measured (Hz) Calculated (Hz)
1 0.796 0.80 0.9988
2 1.095 1.10 1.2093
3 1.584 1.62 1.6978
4 2.194 2.23 2.2861
5 2.907 2.98 2.7052
6 3.709 3.77 3.1592
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
101
4.3.3. Input motion
According to modern seismic codes for bridge structures, e.g. EC8-2 [CEN, 2002] and the
Caltrans Seismic Design Criteria [Caltrans, 1999], the spatial variability of the ground
motion must be taken into consideration in the design of long bridges in the presence of
geological discontinuities or marked topographical features. As discussed in Chapter 2,
analytical studies of regular bridges subjected to asynchronous excitation suggest that the
internal forces are in general reduced, in comparison to synchronous input, but in certain
cases, e.g. stiff structures and significant soil amplification, an increase was observed.
Experimental results show the reduction of internal forces and increase in displacements
for an irregular bridge [Calvi & Pinto, 1996; Pinto et al., 1996].
Various models for asynchronous motion, considering the accelerations as samples of a
stochastic process, have been proposed. For the present study, the input motion was
defined by artificial accelerograms, shown in Figure 4.6. Based on the local soil
conditions, different focal depths, distance and source mechanisms, displacement,
velocity and acceleration time histories were calculated [Panza et al., 2001]. Different
accelerograms at the bases of the piers and the abutments were provided for the case of
asynchronous motion. Some treatment of these accelerograms was performed in order to
have a uniform distribution of the time step, null final displacement and velocity, null
average velocity and correct scaling. Loading in the bridge transverse direction alone was
considered. Figure 4.7 presents the elastic response spectra of the input accelerograms,
considering 5% of critical damping.
4.3.4. Numerical simulation of the pseudodynamic tests
The PSD test was numerically simulated in order to validate its representativeness of an
earthquake test. The same model and assumptions considered for the dynamic analysis
were used for the numerical simulation of the PSD tests, with the difference that only the
horizontal forces in the direction of loading were transmitted between the deck and the
piers, since the substructuring method would be applied. In addition, both an implicit
iterative and the a-OS explicit schemes were used for the integration of the equation of
motion. The results of the numerical simulation of the PSD tests were compared to the
results of the dynamic analysis. The same input motion used for the dynamic analysis was
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
102
used for the numerical simulation of the PSD tests for both the explicit and implicit
iteration schemes.
Abutment Wien
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Pier A50
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Pier A20
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Pier A60
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Pier A30
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Pier A70
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Pier A40
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Abutment Graz
-1.0
-0.5
0.0
0.5
1.0
0 1 2 3 4 5
Time (sec)
Acc
eler
atio
n (g
)
Figure 4.6. Input accelerograms for the abutments and the pier bases (see Figure 4.3 for
abutment and pier labels)
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
103
Abutment Wien
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)Pier A50
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Pier A20
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Pier A60
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Pier A30
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Pier A70
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Pier A40
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Abutment Graz
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3
Period (sec)
Acc
eler
atio
n (g
)
Figure 4.7. Response spectra of the accelerograms for the nominal earthquake
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
104
Pier A20
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Dynamic OS Implicit
Pier A50
-0.04-0.03-0.02-0.010.000.010.020.030.04
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Dynamic OS Implicit
Pier A30
-0.050
-0.025
0.000
0.025
0.050
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Dynamic OS Implicit
Pier A60
-0.02
-0.01
0.00
0.01
0.02
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Dynamic OS Implicit
Pier A40
-0.05
-0.03
-0.01
0.01
0.03
0.05
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Dynamic OS Implicit
Pier A70
-0.002
-0.001
0.000
0.001
0.002
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Dynamic OS Implicit
Figure 4.8. Displacement histories for the 0.4xNE test, pre-test numerical analysis
The results of the numerical simulation of the PSD test using the implicit and explicit
schemes for the numerical integration of the equation of motion were similar, see Figure
4.8. There was only a small difference in amplitude at certain time steps that can be
attributed to the numerical approximations of the residual in the a-OS method. In
addition, the results of these numerical simulations were in good agreement with the
results of the dynamic analysis, considered as the reference. Note, however, that the
results of the dynamic analysis were affected by the vertical modes. This was partly due
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
105
to the concrete constitutive model itself that does not represent perfectly the re-opening
and closing of cracks and to the assumption that the base is infinitely rigid.
The numerical simulation for the nominal earthquake indicated low level of damage for
the bridge piers. In detail, the short pier, A70, remained essentially in the elastic range
and the tall pier, A40, suffered only cracking. This was due to the amplitude and the
frequency content of the generated accelerograms that did not excite too much the
significant modes of the structure. For this reason, it was decided to perform a PSD test
with the given input motion, followed by a test with the input motion multiplied by a
factor equal to 2. The dynamic analyses showed that piers A40 and A70 would be more
heavily damaged. Therefore, it was decided to test physical models of these two and
numerically simulate the remaining piers.
4.4. PSEUDODYNAMIC TESTING OF THE BRIDGE MODEL
4.4.1. Testing programme
Before the main PSD test campaign, small-amplitude static and PSD tests were
performed. The objectives were to verify the communication between the different
workstations used for the control and data acquisition and to check the robustness of the
overall procedure. The amplitude of displacement was controlled not to produce inelastic
deformation of the structure. The check tests are described elsewhere [Pinto et al., 2002].
Three effective earthquake tests were performed. A low-level earthquake (0.4xNE) test
with amplitude 40% of the nominal earthquake was performed first. According to EC8-1,
the seismic action associated with the damage limitation criterion, in other words with the
Serviceability Limit State (SLS), is 0.4 or 0.5 of the design seismic action, depending on
the importance of the structure. Then followed a test for the nominal earthquake
(1.0xNE), corresponding to the Ultimate Limit State (ULS) and defined by the
accelerograms shown in Figure 4.6. Finally, a high-level earthquake (2.0xNE) test,
corresponding to twice the nominal earthquake was performed. After the completion of
the PSD tests a cyclic capacity test was performed on the model of the tall pier, A40; the
short pier, A70, collapsed during the 2.0xNE test. ULS and SLS are used in the sense of
EC8-2 [CEN, 2002], where the ULS is associated to the “non-collapse” requirement for
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
106
an earthquake with approximately 475 years return period and the SLS is associated to the
“minimisation of damage” requirement for a more frequent earthquake.
At the end of each PSD test the physical piers were brought at zero force; obviously a
residual displacement remained in the piers. The same was applied for the numerical
piers: at the end of the test they were brought at zero force and the residual displacement
was used as initial displacement for the following test. Since for the bilinear constitutive
law for steel no ultimate deformation was provided, at every time step the maximum
plastic deformation of the steel fibres at every cross-section of the numerical piers was
extracted from the results so that a warning would come into view in case failure of a
rebar in a numerical pier should occur. The bilinear constitutive law was adopted in order
to ensure the stability of the iterative processes; it was verified that this simplification did
not significantly affect the response of the structure. The calibration of the numerical
models is further discussed in the following chapter.
4.4.2. Low-level earthquake test
The first effective PSD test was performed for an earthquake input corresponding to the
given accelerograms scaled by 0.4. This scaling factor was chosen so that the effect of the
SLS earthquake could be studied. The force-drift curves are given in Figure 4.9.
For this amplitude of the input earthquake only minor damage, namely cracking, was
observed in the numerical piers. The damage of the numerical piers was estimated on the
basis of the displacement corresponding to cracking and yielding as results from the pre-
test numerical simulation, see Table 4.1. On the other hand, visual inspection of the
physical piers showed slight horizontal cracking within the first 2.5 m of the tall pier,
A40, and no damage in the short pier, A70. The damage pattern of the tall pier after the
end of the test is shown in Figures 4.10 and C.5. It is reminded that displacement was
imposed in the East-West direction.
The amount of energy dissipated by each pier was calculated: for the 0.4xNE test the
largest percentage corresponds to the physical tall pier and the smallest to the physical
short pier that remained essentially elastic during the test. The Park and Ang Damage
Index was calculated, based on the experimental data. The small values of the index
indicate slight damage for all the physical and numerical piers. In accordance with the
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
107
experimental observations, a very small Damage Index was calculated for the short
physical pier, A70.
The experimental displacement histories are compared to the ones obtained from the pre-
test calculations in Figure 4.11, where good agreement is observed. Only for the short
physical pier, A70, the experimental displacement was of inferior amplitude compared to
the numerical simulation. This was due to the fact that the stiffness of the physical piers
A40 and A70 used in the pre-test numerical simulation corresponded to the numerical
model of the piers and not to the physical models. After performing the first test, the
actual stiffness of the piers could be measured and resulted equal to about 2/3 of the
calculated stiffness. The numerical models were accordingly updated.
Pier A20
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A50
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A30
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A50
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A40
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A70
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Figure 4.9. Force-drift diagrams for the 0.4xNE test
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
108
Figure 4.10. Damage pattern of the tall pier for the 0.4xNE test
4.4.3. Nominal earthquake test
The following test was performed for the nominal earthquake, always considering
asynchronous motion for each pier and abutment. The residual displacement from the
0.4xNE test was considered as the initial condition for the numerical piers, in order to
represent the damage suffered in the previous earthquake test; residual displacement was
anyway present on the physical piers. The force-drift curves for the physical and
numerical piers are shown in Figure 4.12.
The experimental displacement histories are compared to those calculated from the pre-
test analysis in Figure 4.13: the difference due to the stiffness for this earthquake intensity
was less evident. The displacement histories are in agreement for the first and most
significant part of the accelerogram; near the end of the test, a phase difference is
observed in some piers.
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
109
Pier A20
-0.050
-0.025
0.000
0.025
0.050
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)Pier A50
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A30
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A60
-0.05
-0.03
0.00
0.03
0.05
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A40
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A70
-0.02
-0.01
0.00
0.01
0.02
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Figure 4.11. Displacement histories for the 0.4xNE test, experimental (solid line) and
numerical (thin line) results
The Park and Ang damage index for the physical and numerical piers was calculated for
the 1.0xNE test. Minor damage, described as cracking, was indicated for all piers. The
distribution of dissipated energy was similar to that for the 0.4xNE test: the largest
percentage corresponds to the tall physical pier, A40, whereas equal percentages are
allocated to the substructured piers. A small percentage was assigned to the short physical
pier, A70, that showed limited damage.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
110
Pier A20
-1500
-1000
-500
0
500
1000
1500
-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00
Drift (%)
For
ce (
kN)
Pier A50
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A30
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A60
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A40
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A70
-1500
-1000
-500
0
500
1000
1500
-1 -0.8 -0.5 -0.3 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Figure 4.12. Force-drift diagrams for the 1.0xNE test
For the 1.0xNE test, the existing cracks of the tall physical pier, A40, extended and new
cracks appeared up to the height of 4 m. A horizontal crack initiated at the height of 3.5 m
where the bar cut-off takes place, see Figure 4.14. For the short physical pier, A70, only
slight horizontal cracks near the base were observed. Based on the force-drift curves and
the pre-test numerical analysis, the damage of the numerical piers was identified as
extended cracking. Only the numerical piers A30 and A50 had maximum displacements
above the values that correspond to the conventional yielding displacement, yu .
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
111
Pier A20
-0.050
-0.025
0.000
0.025
0.050
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A50
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A30
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A60
-0.050
-0.025
0.000
0.025
0.050
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A40
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A70
-0.01
-0.01
0.00
0.01
0.01
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Figure 4.13. Displacement histories for the 1.0xNE test, experimental (solid line) and
numerical (thin line) results
4.4.4. High-level earthquake test
The final PSD test was performed for the input accelerograms times 2. The objective of
this test was to investigate the resistance of the bridge system subjected to a strong
earthquake after having suffered damage due to the ULS earthquake. The force-drift
experimental curves for the physical and numerical piers are presented in Figure 4.15. In
Figure 4.15 significant non-linear deformation and energy dissipation is observed mainly
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
112
in the physical piers, as predicted by the pre-test analyses. The limited energy dissipation
observed in the substructured piers is not due to some inadequacy of the numerical
models, which were calibrated to simulate as best as possible the cyclic behaviour, but to
the small amplitude of imposed displacement that induced limited inelastic deformation.
This will be further discussed in the following chapter.
The experimental displacement histories are presented in Figure 4.16 along with the
values resulting from the pre-test analysis. The fit in this case is not so good: there is a
phase difference after the first second of the input accelerograms. It is reminded that the
PSD tests were performed one after the other and at the beginning of each test the damage
suffered in the previous one was present. This was not considered in the pre-test
numerical analyses: for each level of the input earthquake the undamaged structure was
analysed.
Figure 4.14. Damage pattern of the tall pier for the 1.0xNE test
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
113
Pier A20
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A50
-1500
-1000
-500
0
500
1000
1500
-1 -0.8 -0.5 -0.3 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A30
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A60
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1Drift (%)
For
ce (
kN)
Pier A40
-1500
-1000
-500
0
500
1000
1500
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Pier A70
-1500
-1000
-500
0
500
1000
1500
-1 -0.8 -0.5 -0.3 0 0.25 0.5 0.75 1
Drift (%)
For
ce (
kN)
Figure 4.15. Force-drift diagrams for the 2.0xNE earthquake test
For the 2.0xNE test the numerical piers, as well as the tall physical pier, A40, were
already damaged and were able to further deform, without significant additional damage.
Regarding the tall physical pier, A40, the cracks at the lower part extended with a slight
inclination and remained open, while the crack at the height of 3.5 m extended through
the web and spalling of concrete was observed at that cross-section. All the numerical
piers were beyond the yield limit.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
114
Pier A20
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A50
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A30
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A60
-0.10
-0.05
0.00
0.05
0.10
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A40
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Pier A70
-0.08
-0.04
0.00
0.04
0.08
0 1 2 3 4 5Time (sec)
Dis
plac
emen
t (m
)
Figure 4.16. Displacement histories for the 2.0xNE test, experimental (solid line) and
numerical (thin line) results
Figure 4.17. Damage pattern of the short pier for the 2.0xNE test
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
115
The short physical pier, A70, was the most heavily strained for this earthquake amplitude:
it had to undergo a large number of cycles at significant levels of displacement. As a
result, a considerable loss of resistance was observed. A horizontal crack developed at the
base (interface with the foundation block) throughout the whole length of the pier. At the
corners the concrete crushed and two vertical rebars collapsed, showing evidence of
buckling. Vertical cracks were observed in the web in correspondence to the lapped
splices. The damage pattern of the short physical pier, A70, after the end of the 2.0xNE
test is presented in Figures 4.17 and C.6.
For the 2.0xNE test the short physical pier, A70, dissipated the largest percentage of
energy; this was consistent with the heavy damage suffered by this pier. Observing the
values of the Park and Ang Damage Index for the 2.0xNE test, the most heavily damaged
piers were the physical A40 and A70. Moderate damage was attributed to pier A40,
corresponding to extensive large cracks and spalling of concrete. For the numerical piers
the damage level was also moderate. For the short pier, A70, the damage level was
identified as severe, corresponding to crushing of concrete and disclosure of buckled
reinforcement. The calculated damage indices were in fair agreement with the observed
damage.
4.4.5. Final collapse test
After the end of the PSD tests, a cyclic test with increasing imposed displacement until
failure was performed on the tall physical pier, A40. The short pier, A70, also physically
present in the lab had already failed, as evidenced by the significant drop of resistance
and by the rupture of vertical reinforcement bars. The objective of the test was to assess
the ultimate capacity of the pier after a number of cycles at significant levels of
displacement.
The displacement history consisted of increasing monotonic displacement, in the presence
of vertical load, until failure. Then, the specimen was unloaded and as some residual
displacement was observed, it was pushed in the opposite direction in order to minimize
the residual displacement once the horizontal loads had been removed. The experimental
force-displacement curve for the final capacity test on the tall pier, A40, is presented in
Figure 4.18.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
116
-800
-600
-400
-200
0
200
400
600
800
1000
-0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20
Displacement (m)
For
ce (k
N)
Figure 4.18. Final collapse test on the tall pier: force-displacement diagram
Figure 4.19. Damage pattern of the tall pier for the final collapse test
The failure mode of the tall physical pier was as observed for the cyclic test that was
discussed in detail in the previous chapter. Figures 4.19 and C.7 present the crack pattern
after the end of the final capacity test. Green colour corresponds to the cracks observed
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
117
after the 1.0xNE test, red colour is for the cracks that appeared during the 2.0xNE test,
while black stands for the cracks that developed during the final capacity test. A
horizontal crack had already appeared at the critical cross-section at the bar cut-off during
the 2.0xNE test. During the final capacity test, the existing cracks extended with
significant inclination and a few new ones appeared in the region from 3.0 m until 4.0 m
from the base of the pier. Crushing of concrete and buckling of vertical reinforcement
bars was observed at the corners of the cross-section for both directions of imposed
displacement. Vertical cracking, indicating bar slippage, was observed in the flange above
and below the critical cross-section.
4.5. SEISMIC ASSESSMENT OF THE BRIDGE
4.5.1. Deformation and curvature distribution in the physical piers
The measurements of the instrumentation on the two physical piers were used to split the
flexural and shear deformations, using the procedure outlined in the previous chapter. The
results for the three earthquake tests are presented in Figures 4.20 and 4.21. For the short
pier and the 2.0xNE test, the ratio of shear to flexural displacement is 0.3, which is very
close to the value measured during the quasi-static cyclic test, see 3.4.1. The small
amplitude of displacement and limited damage, the piers remained essentially elastic or
suffered slight cracking, during the previous two tests does not allow a reliable
comparison of flexural and shear displacement.
To what concerns the tall pier, increasing values of the shear to flexural displacement
ratio were measured for the four tests. In detail the ratio was 0.2 for the 0.4xNE and
1.0xNE tests, 0.25 for the 2.0xNE test and 0.6 for the final cyclic capacity test. In
accordance with the damage observed during the PSD tests, shear deformation was a
significant portion of total deformation within the lower part of the pier. Only for the final
capacity test, in which severe damage was observed above the cut-off, shear deformation
was significant also above the critical cross-section.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
118
0.4xNE
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Displacement (m)
Hei
ght (
m)
shear flexure
1.0xNE
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Displacement (m)
Hei
ght (
m)
shear flexure
2.0xNE
0
1
2
3
4
5
6
7
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06Displacement (m)
Hei
ght (
m)
shear flexure
Figure 4.20. Flexural and shear deformation of the short pier
0.4xNE
02468
10121416
-0.2 -0.1 0.0 0.1 0.2Displacement (m)
Hei
ght (
m)
shear flexure 1.0xNE
02468
10121416
-0.2 -0.1 0.0 0.1 0.2Displacement (m)
Hei
ght (
m)
shear flexure
2.0xNE
02468
10121416
-0.2 -0.1 0.0 0.1 0.2Displacement (m)
Hei
ght (
m)
shear flexure Final cyclic
02468
10121416
-0.2 -0.1 0.0 0.1 0.2Displacement (m)
Hei
ght (
m)
shear flexure
Figure 4.21. Flexural and shear deformation of the tall pier
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
119
The evolution of average curvature along the height of the short physical pier, A70, is
presented in Figure 4.22 for the three earthquake tests. The curvature demand was
concentrated at the base, where failure was observed. For the 2.0xNE test, that caused
extended damage and collapse of the pier, a notable curvature demand was also observed
at the cross-section just above the first lap-splice. Similar distribution of deformation
demand (concentration at the base) was observed during the cyclic test performed on a
specimen of the same pier, discussed in the previous chapter, and also in previous
experiments on columns with lapped splices [Paulay, 1982; Chai et al., 1991; Park et al.,
1993; Lynn et al., 1996; Xiao & Ma, 1997].
Regarding the tall pier A40, the distribution of curvature varies for the different
amplitudes of the input motion, as seen in Figure 4.23. For the 0.4xNE and 1.0xNE tests,
the curvature demand was quite uniformly distributed within the region up to the height
of about 3.5 m, where the vertical reinforcement was curtailed. For the 2.0xNE and the
final cyclic test the demand was shifted to the critical cross-section above the bar cut-off.
Similar evolution of curvature with increasing displacement was observed during the
cyclic tests on the tall pier, presented in the previous chapter.
4.5.2. Damage assessment
Comparing the maximum attained displacement for each earthquake test to the
characteristic values of Table 4.1, the damage of the bridge piers can be identified, as
reported in Table 4.3. For the SLS earthquake, 0.4xNE test, the bridge suffered only
minor damage: the short pier remained elastic and only slight cracks were observed in the
rest of the piers. At this level of damage no repair is required. For the ULS earthquake,
1.0xNE test, the imposed displacement caused yielding, without significant inelastic
deformation, of the numerical piers A30 and A50. At this level of damage the cracks need
repair. For the high-level and less frequent earthquake, 2.0xNE test, significant damage
was observed in most of the piers that were beyond the yield limit. For the 2.0xNE test,
failure of the vertical reinforcement in the short pier occurred, without causing global
instability problems. The short pier reached its capacity: for consecutive cycles its
resistance dropped and in the end the vertical rebars in the flange failed. The cracks in the
numerical and tall physical piers need repair.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
120
0.4xNE
0
1
2
3
4
5
6
7
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
1.0xNE
0
1
2
3
4
5
6
7
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
2.0xNE
0
1
2
3
4
5
6
7
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03Curvature (rad/m)
Hei
ght (
m)
Figure 4.22. Distribution of average curvature along the height of the short pier
0.4xNE
0
2
4
6
8
10
12
14
-0.02 -0.01 0.00 0.01 0.02
Curvature (rad/m)
Hei
ght (
m)
1.0xNE
0
2
4
6
8
10
12
14
-0.02 -0.01 0.00 0.01 0.02
Curvature (rad/m)
Hei
ght (
m)
2.0xNE
0
2
4
6
8
10
12
14
-0.02 -0.01 0.00 0.01 0.02Curvature (rad/m)
Hei
ght (
m)
Final cyclic
0
2
4
6
8
10
12
14
-0.02 -0.01 0.00 0.01 0.02
Curvature (rad/m)
Hei
ght (
m)
Figure 4.23. Distribution of average curvature along the height of the tall pier
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
121
Table 4.3. Damage of the bridge piers
A20 A30 A40 A50 A60 A70
0.4xNE cracking cracking cracking cracking cracking -
1.0xNE cracking yielding cracking yielding cracking cracking
2.0xNE cracking yielding yielding yielding yielding failure
Table 4.4. Maximum drift and ductility demand for the piers
d (%) µ
A20 A30 A40 A50 A60 A70 A20 A30 A40 A50 A60 A70
0.4xNE 0.16 0.28 0.25 0.20 0.12 0.03 0.3 0.9 0.4 0.7 0.4 0.2 1.0xNE 0.37 0.57 0.44 0.41 0.30 0.13 0.7 1.9 0.8 1.4 0.9 0.7 2.0xNE 0.58 0.88 0.87 0.55 0.41 0.69 1.1 2.9 1.5 1.8 1.1 3.9 Capacity 1.57 2.39 1.43 2.26 1.49 0.69 2.9 9.8 2.5 7.4 4.1 3.9
Table 4.5. Dissipated energy and Damage Index
Dissipated energy (% of total) Park & Ang Damage Index
A20 A30 A40 A50 A60 A70 A20 A30 A40 A50 A60 A70
0.4xNE 7.0 8.2 62.2 13.3 7.4 1.8 0.11 0.10 0.16 0.09 0.09 0.02
1.0xNE 5.8 6.8 57.7 5.8 7.7 16.1 0.24 0.20 0.28 0.19 0.21 0.09
2.0xNE 1.2 1.3 26.2 4.8 2.7 63.8 0.39 0.31 0.56 0.25 0.29 1.00
Table 4.4 presents the maximum drift and ductility demands in the piers for the different
earthquake amplitudes. It is reminded that displacement ductility is defined assuming a
bilinear equivalent system. The first branch is defined by the tangent stiffness at 75% of
the maximum force and zero post-yield stiffness is assumed. For the numerical piers the
monotonic curves resulting from the numerical analyses are used. For the physical piers
the envelop curves of the 2.0xNE and the final collapse test for the short, A70, and the
tall, A40, piers, respectively are used to identify the yielding point and the displacement
capacity. It is interesting to note that for all the numerical piers and the physical tall pier,
the deformation demand/capacity ratio increases almost linearly with the earthquake
intensity, whereas for the short pier, A70, very low values correspond to the 0.4xNE and
1.0xNE tests and the capacity is reached for the 2.0xNE test.
The drift capacities of the piers are extremely limited and significantly lower that the
commonly adopted limit u,minδ = 3%. On the other hand, the values of displacement
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
122
ductility capacity range between uµ = 2.9 and uµ = 9.8 and then imply a desirable ductile
behaviour. This is inconsistent with the experimental observations and the drift capacities.
In addition, the ratio of maximum attained drift to the drift capacity correlates well with
the observed damage for increasing amplitude of the seismic input, while the same ratio
based on displacement ductility seems misleading. This originates from the conventional
definition of displacement ductility and supports the remark made in the previous chapter
that drift describes better the performance of the tested piers.
The distribution of dissipated energy is presented in Table 4.5. For the 0.4xNE test the
greatest part of the total energy was dissipated by the tall physical and the numerical
piers. The short pier remained almost elastic and dissipated only a small fraction of the
total energy. Similar distribution was observed also for the 1.0xNE test, with the
difference that the short physical pier dissipated a significant part of the total energy. The
distribution changed for the 2.0xNE test: the short pier dissipated the largest percentage
of energy. It is reminded that the short pier suffered the heaviest damage during the
2.0xNE test and in the end collapsed.
The Park and Ang Damage Index is also presented in Table 4.5 for all the piers and the
different earthquake amplitudes. For the 0.4xNE test slight damage (sporadic cracking) is
identified for all the piers. For the 1.0xNE test slight damage is predicted for piers A50
and A70 and minor damage (minor cracks and partial crushing of concrete) for the
remaining piers. Finally, for the 2.0xNE test minor damage corresponds to piers A50 and
A60, moderate damage (large cracks and concrete spalling) to piers A20 and A30, severe
damage (concrete crushing and steel buckling) to pier A40 and collapse to pier A70. The
calculated damage level is in general confirmed by the observations on the physical and
numerical piers after the end of each PSD test.
4.5.3. Overall damage index
Having established different damage indices for the substructures (bridge piers), an
extension to the whole structure (bridge) would provide a quantitative assessment of the
overall damage level. A number of alternative procedures for the estimation of overall
damage indices for buildings exist, e.g. [Powell & Allahabadi, 1988]. These procedures
apply provided that the distribution of damage is uniform throughout the structure; this
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
123
was not the case for the tested bridge. When the damage index is computed for the
substructures, an adequate combination can provide an overall damage index, tDI . The
latter is estimated as a weighted mean of the damage indices of the substructures
according to the expression
i it
i
w DIDI
w= ∑
∑ (4.21)
where iDI is the damage index and iw is the weight assigned to the i-th substructure.
The weight, iw , is a measure of the importance of the substructure for the structural
integrity. For the case of bridges, all piers are of the same importance for the structural
integrity. Due to the low statical indeterminacy, collapse of any single pier corresponds to
total or partial collapse of the bridge structure. Nevertheless, equal weight cannot be
attributed to the piers. The implication is that for the 2.0xNE test that caused collapse of
the short pier, and consequently failure of the bridge system, the overall damage index
would be smaller than the value that indicates collapse, as affected by the slightly
damaged remaining five piers. It seems, therefore, more realistic to use a weighting factor
proportional to the damage index of each substructure [Park et al., 1987]. The overall
damage index is then
2i
ti
DIDI
DI= ∑
∑ (4.22)
Alternatively, the Park and Ang damage index [Park & Ang, 1985] for buildings defines
the overall damage index as the weighted sum of the indices for each substructure
(storey), according to the expression
t i iDI w DI= ∑ (4.23)
where the weighting factor is calculated on the basis of the dissipated energy of each
substructure, iE , and is given by the expression
i i iw E E= ∑ (4.24)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
124
Table 4.6. Overall Park and Ang Damage Index
Damage-based weight Energy-based weight
0.4xNE 0.11 0.13
1.0xNE 0.22 0.23
2.0xNE 0.61 0.81
The same procedures can be applied using the experimental results for the bridge
structure and considering the piers as the substructures. The deck being considered
elastic, does not suffer any damage and therefore does not contribute to the estimation of
the damage of the whole structure. The values of the overall damage index according to
Equations 4.22-4.24 are presented in Table 4.6. The damage-weighted index
underestimates the overall damage of the structure, compared to the energy-weighted
sum. In fact, the energy-based weight is able to better reflect the distribution of damage
among the piers for each earthquake test.
4.5.4. Vulnerability functions
Vulnerability functions represent an important part of studies aimed at risk assessment
and various procedures are available, e.g. [Corsanego, 1991]. Vulnerability functions
correlate the damage (quantified using an appropriate damage index) observed in the
structure, to a parameter that defines the earthquake input (often macroseismic intensity
or ground acceleration). Data obtained from field investigations after a significant
earthquake are commonly used for the development of vulnerability functions. An
attempt is made in this section to perform a preliminary estimation of such functions on
the basis of the experimental results. The seismic motion was characterised by the peak
ground acceleration normalised to the value for the 1.0xNE test and different damage
indices were examined, namely the Park and Ang Damage Index, the drift demand and
the displacement ductility (defined by the bilinear envelope curve). The results are
presented in Figure 4.24. Concerning piers A20 to A60, damage increases almost linearly
with the earthquake intensity, but remains always at low levels. For the short physical
pier, A70, damage increases rapidly from the 1.0xNE level to the 2.0xNE level, for which
the pier failed. All damage indices give consistent results.
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
125
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 0.5 1 1.5 2 2.5a/anom
Par
k &
Ang
Dam
age
Inde
x A20 A30A40 A50A60 A70
(a)
0.0
0.2
0.4
0.6
0.8
1.0
0 0.5 1 1.5 2 2.5a/anom
Drif
t (%
)
A20 A30A40 A50A60 A70
(b)
0
1
2
3
4
5
6
0 0.5 1 1.5 2 2.5a/anom
Dis
plac
emen
t duc
tility
A20 A30A40 A50A60 A70
(c)
0.0
0.2
0.4
0.6
0.8
1.0
0 0.5 1 1.5 2 2.5
a/anom
Ove
rall
Dam
age
Inde
x
(d)
Figure 4.24. Vulnerability functions: Park and Ang Damage Index (a), drift ratio (b),
displacement ductility (c) and overall Park and Ang Damage Index (d)
Observing Figure 4.24, it is obvious that the vulnerability function cannot in general be
considered a linear one, as commonly assumed in the past. An exponential or higher-
order polynomial in the form suggested by [Powell and Allahabadi, 1988]
m
c ts
t
d dDI
d d −
= − (4.25)
seems more appropriate. An exponential or 3rd order polynomial fit the experimental
results better than a linear relation. It should be underlined, though, that the small number
of experimental results allows only a preliminary evaluation of the vulnerability
functions. Further numerical simulations could be used for a more reliable estimate of the
vulnerability functions.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
126
4.5.5. Effect of cycling
Seismic detailing, as required by modern seismic codes, guarantees ductility and
dissipation capacity of members and structures, in other words it allows for stable
response without significant loss of resistance and stiffness with cycling. Members
without seismic detailing are expected to be more sensitive to cycling, in terms of
strength and stiffness degradation. Cycling is also considered significant for the behaviour
of lapped splices [Balázs, 1991], which was a weak point of the tested bridge piers.
The relation between the ultimate displacement and the number of cycles has been
experimentally investigated for well-confined circular columns with flexural failure mode
[El-Bahy et al., 1999]. It was found that for a small number of cycles at significant levels
of displacement, similar to the static cyclic tests, failure is usually due to rupture of
reinforcement, whereas in the case of more cycles at smaller displacement levels, similar
to the PSD earthquake tests, failure is usually due to crushing of concrete. The cyclic
effects have been proved significant even in the case of well-confined circular columns:
for ductility demand in the order of µ = 2 the piers sustained a number of cycles without
significant structural damage, but for ductility close to µ = 4, moderate to severe damage
was more probable, depending on the number of cycles [Taylor et al., 1997]. Because of
cumulative damage, it has been suggested that in evaluating the seismic performance of
structures the capacity is dependent on the demand (in terms of number and amplitude of
inelastic excursions). This holds mainly for ductile steel structures and might not
completely apply to RC structures [Krawinkler, 1996].
The performance of the tall pier, A40, during the PSD tests is first compared to the
behaviour of the same pier subjected to cyclic loading until collapse. As seen in Table
4.7, the deformation capacity, as described by displacement ductility (defined by the
bilinear envelope) and lateral drift, was respectively 45% and 12% lower for the
earthquake tests, compared to the cyclic tests. The failure mode was in both cases due to
collapse of vertical reinforcement at the critical cross-section at 3.5 m from the base. The
smaller value of ultimate displacement was due to the structural degradation of the pier
during the large number of cycles of the three earthquake tests. The maximum drift
demand was ud = 1.5%, half of 3%, which is commonly set as the target value for new
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
127
bridges [Priestley et al., 1996]. No significant difference in the maximum strength was
observed between the cyclic and the PSD tests.
Also for the short physical pier, A70, the displacement ductility capacity was smaller for
the earthquake tests in comparison to the cyclic test, see Table 4.7. The maximum drift
demand was d= 0.7% for the PSD tests and d= 1.3% for the cyclic test. The
displacement ductility capacity was also reduced by 50%. This difference can be
attributed to cyclic effects, since for the PSD tests the pier had to undergo a significant
number of cycles at high levels of displacement. Again, no significant difference in the
maximum strength was observed.
4.5.6. Irregularity issues
The irregular distribution of damage among the bridge piers is evidenced in Figure 4.25,
which presents the distribution of displacement ductility, drift, Park and Ang damage
index and dissipated energy in the bridge piers for the different earthquake tests. The
displacement ductility and drift demands were calculated for the maximum displacement
of each pier that, in general, was reached at different time steps. For the 0.4xNE and the
1.0xNE tests, the ductility demand, drift demand, as well as the damage, are quite
regularly distributed among the piers. For the 2.0xNE test, though, the demand is shifted
to the short pier that remained practically undamaged for the previous tests. The
percentage of dissipated energy follows an irregular distribution for all three earthquake
tests.
This irregular distribution of damage is attributed to higher mode effects. For the 0.4xNE
and the 1.0xNE tests the bridge followed mainly the first two eigenmodes. Higher
eigenmodes (see Figure 4.5) impose on the short pier displacement similar to those
imposed on the other piers. Given the smaller height, larger drift and more severe damage
are expected for the short pier. The effect of higher modes is verified in Figure 4.26 that
plots the Fast Fourier Transforms of the displacement histories at the top of the piers for
the three earthquake tests. For the 0.4xNE test one peak at frequency 1.95 Hz is observed,
for the 1.0xNE tests there are two peaks at 1.75 Hz and 2.20 Hz, while for the 2.0xNE test
there are three peaks at 1.46 Hz, 2.20 Hz and 2.93 Hz.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
128
0
1
2
3
4
5
6
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t duc
tility
dem
and
0.4xNE NE 2.0xNE (a)
0.0
0.2
0.4
0.6
0.8
1.0
A20 A30 A40 A50 A60 A70
Drif
t dem
and
(%)
0.4xNE NE 2.0xNE(b)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
A20 A30 A40 A50 A60 A70
Par
k &
Ang
Dam
age
Inde
x
0.4xNE NE 2.0xNE(c)
0
20
40
60
80
100
A20 A30 A40 A50 A60 A70
Dis
sipa
ted
ener
gy (
% o
f tot
al)
0.4xNE NE 2.0xNE (d)
Figure 4.25. Distribution of ductility demand (a), drift demand (b), Park and Ang Damage
Index (c) and percentage of dissipated energy (d) among the piers
0.00
0.02
0.04
0.06
0.08
0.10
0 1 2 3 4Frequency (Hz)
Dis
plac
emen
t (m
)
A20A30A40A50A60A70
0.4xNE
0.00
0.02
0.04
0.06
0.08
0.10
0 1 2 3 4Frequency (Hz)
Dis
plac
emen
t (m
)
A20A30A40A50A60A70
1.0xNE
0.00
0.02
0.04
0.06
0.08
0.10
0 1 2 3 4Frequency (Hz)
Dis
plac
emen
t (m
)
A20A30A40A50A60A70
2.0xNE
Figure 4.26. Fast Fourier Transforms of the pier top displacement
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
129
Table 4.7. Displacement ductility and drift capacities for the cyclic and PSD tests
µ d (%)
Cyclic PSD Cyclic PSD
A40 3.7 2.4 1.7 1.5
A70 8.0 3.9 1.3 0.7
At seismic code level, regularity is related to the selection of analysis methods and of the
values of the behaviour factor. When a bridge is of regular geometry, equivalent static
analysis for lateral forces that follow a distribution similar to the first eigenmode is
allowed; otherwise more elaborate methods (usually multi-modal spectral analysis) are
required. Concerning the definition of regularity for bridge structures, several criteria
have been proposed. According to EC8-2 [CEN, 2002], a bridge is defined as regular
when
omin,d
max,d
ρ≤=ρ (4.26)
where
Ed,id,i
Rd,i
Mq q
M= (4.27)
In the above expressions q is the behaviour factor, EdM is the design moment for the
seismic action and RdM is the resisting moment. The value o? = 2 is recommended. One
or more piers can be excluded from the criterion if the sum of their shear force is inferior
to 20% of the total shear force.
The criterion of Equation 4.26 can be interpreted as a requirement that the maximum
value of ductility must not be higher than o? times the minimum value for a bridge to be
regular. For the tested bridge and the 0.4xNE test the maximum value of ductility was
d,maxq = 0.9 and the minimum d,mind = 0.2. Therefore, ? = 4.5 and the bridge is classified as
irregular. Also for the 1.0xNE test, ? = 2, and for the 2.0xNE test, ? = 3.6, the bridge is
classified as irregular.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
130
0
100
200
300
400
500
600
A20 A30 A40 A50 A60 A70
Stif
fnes
s ch
ange
(%)
Figure 4.27. Change in stiffness between adjacent piers and distribution of stiffness
According to the AASHTO Bridge Design Specifications [AASHTO, 1995], a bridge is
classified as irregular when changes in stiffness or mass from one pier to the adjacent
exceed 25%. Figure 4.27, plots the change of stiffness between adjacent piers, along with
the AASHTO limit. Following this criterion, the bridge is considered highly irregular.
The ratio of the maximum column stiffness, pk , to the deck stiffness, dk , has been also
proposed as a regularity index [Fishinger et al., 1997]. The stiffness ratio, kr , is defined as
k p dr k / k= (4.28)
where 3d d dk 48EI / L= is the stiffness of the deck and dL is the total length of the deck. A
bridge is regular if kr < 15. Qualitatively, a bridge is defined as regular when the columns
do not govern the response and consequently the influence of higher modes is not
important. For the Talübergang Warth Bridge the stiffness ratio is calculated as kr » 15:
this value indicates that the bridge is highly irregular.
The recent Italian seismic code [Ordinanza 3274, 2003] adopts the EC8 regularity
criterion for what concerns the selection of the behaviour factor. If a bridge is regular, the
nominal behaviour factor, q, is used for each pier. In the opposite case, it is multiplied by
2 / ? . For the selection of the analysis method, a different regularity criterion is adopted.
A bridge is regular in the longitudinal direction if the total effective mass of the piers,
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
131
corresponding to half the pier height, is less than 1/5 of the deck mass. In the transverse
direction a bridge is regular if the eccentricity, defined as the distance between the mass
centre and the stiffness centre, is less than 5% of the deck length. Following this criterion,
the Talübergang Warth Bridge is regular in the longitudinal direction and irregular in the
transverse direction.
4.6. APPLICATION OF SIMPLIFIED ASSESSMENT METHODS
4.6.1. General
After the serious damage suffered by bridges during recent destructive earthquakes and
the consequent losses, much research has been performed on the development of
simplified assessment methods. The simplest method follows the traditional force-based
design procedures: the structural forces of a linear model of the structure are determined
for an earthquake loading that is described by a response spectrum. Then, a
capacity/demand check is performed. In this area, much effort was dedicated towards the
assessment of the shear capacity of existing members, including the effect of splice
failure and different types of cross-section, e.g. [Ascheim & Moehle, 1992; Ascheim et
al., 1997; Priestley et al., 1994a; Kowalsky & Priestley, 2000; Rasulo et al., 2002]. The
force capacity/demand approach has several drawbacks [Priestley et al., 1996] and for this
reason, performance-based methods have been developed.
Performance-based methods can be divided in two categories. Probabilistic methods use
fragility curves to estimate the possibility of exceeding a certain damage level for a given
earthquake. Fragility curves result either from field observations or from numerical
analyses. The HAZUS methodology is briefly described in the following and then applied
for the assessment of the tested bridge.
For the detailed assessment of a single structure, deterministic methods are more
appropriate. Simplified methodologies make use of an equivalent linear structure and
displacement spectra. The structural response is obtained through non-linear incremental
analysis, which involves some sophistication without at the same time requiring much
effort and experience. The capacity spectrum (CS) and N2 methods, which belong to the
family of performance-based procedures, are discussed in the following.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
132
4.6.2. HAZUS method
HAZUS [FEMA, 1999] is a methodology for earthquake risk assessment developed by
the Federal Emergency Management Agency. Using information on the ground shaking
and structural type, loss estimates are performed. An integral part of the methodology is
the fragility curves for the various structural types. The method is probabilistic and many
uncertainties are inherent in the definition of the input parameters. It is anticipated that it
provides reliable results for damage estimates in extended regions and not for single
structures.
Elastic response spectra with 5% damping are used to characterize the ground-shaking
demand. These spectra have a standard format, see Figure 4.28, consisting of three parts:
a region of constant spectral acceleration, a region of constant spectral velocity and a
region of constant spectral displacement. The region of constant spectral acceleration is
defined by spectral acceleration at the period of 0.3 sec. The constant spectral velocity
region has spectral acceleration proportional to 1/T and is anchored to the spectral
acceleration at the period of 1 sec. The transition period, AVT , defines the intersection of
the regions of constant spectral acceleration and constant spectral velocity and varies
depending on the values of spectral acceleration that define these regions. The constant
spectral displacement region has spectral acceleration proportional to 21/ T and is
anchored to the spectral acceleration at period VDT , where constant spectral velocity
transitions to constant spectral displacement. VDT is considered a function of moment
magnitude.
Figure 4.28. Response spectrum used in HAZUS [FEMA, 1999]
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
133
Damage functions, or fragility curves, for bridges are modelled as log-normally
distributed functions that provide the probability of reaching or exceeding different
damage states for a given level of ground motion or ground failure. Each fragility curve is
characterized by a median value of ground acceleration or displacement and an associated
dispersion factor. Ground motion is quantified in terms of peak ground acceleration and
ground failure is quantified in terms of permanent ground displacement.
Bridges in particular are classified based on the following characteristics: seismic design,
number of spans (single or multi-span bridges), structure type (concrete, steel, others),
pier type (multiple column or single column bents and pier walls), abutment type
(monolithic or non-monolithic), bearing type and span continuity (continuous,
discontinuous or simply supported). Following this classification, 28 types are identified.
A total of five damage states are defined for highway system components. These are: No
damage, Slight, Moderate, Extensive and Complete Damage. No Damage is defined by
minor cracking and spalling of the abutments, cracks in shear keys at abutments, minor
spalling and cracks at hinges, minor spalling at columns (only cosmetic repair is required)
or minor cracking to the deck. Moderate Damage is defined by any column experiencing
moderate cracking (shear cracks) and spalling (the column is structurally still sound),
moderate movement of the abutments (< 5 cm), extensive cracking and spalling of shear
keys, any connection having cracked shear keys or bent bolts, keeper bar failure without
unseating, rocker bearing failure or moderate settlement of the approaches. Extensive
Damage is defined by any column degrading without collapse (shear failure, the column
is structurally unsafe), significant residual movement at connections, major settlement of
the approaches, vertical offset of the abutments, differential settlement at connections and
shear key failure at abutments. Complete Damage is defined by any column collapsing
and connection losing all bearing support, which may lead to imminent deck collapse or
tilting of the substructure due to foundation failure.
For each damage state damage ratios are provided, as shown in Table 4.9. Damage ratios
are defined as a fraction of the component replacement cost. Damage ratios are useful in
cost estimation and in the decision and prioritisation of action when a large population of
structures is screened.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
134
Table 4.9. Damage ratios for highway bridges [FEMA, 1999]
Damage ratio
Damage state Best estimate Range
Slight 0.03 0.01-0.03
Moderate 0.08 0.02-0.15
Extensive 0.25 0.10-0.40
Complete 1.00 0.30-1.00
Table 4.10. Discrete values of restoration functions for highway bridges [FEMA, 1999]
Functional percentage
Restoration period Slight Moderate Extensive Complete
1 day 70 30 2 0
3 days 100 60 5 2
7 days 100 95 6 2
30 days 100 100 15 4
90 days 100 100 65 10
In addition, restoration curves have been elaborated. Restoration curves give the
functional percentage at different periods of time after the occurrence of an earthquake
that produces a given damage state on the structure. They follow a normal distribution
characterised by a mean and a standard deviation. Table 4.10 gives values of the
restoration curves for highway bridges. A bridge with slight damage will be fully
functional (functional percentage 100%) in 3 days, while one with moderate damage in 30
days. After 90 days, bridges with extensive and complete damage will be only 65% and
10% functional, respectively.
The procedure for the estimation of shaking-related damage states for bridges is
summarized in the following. The first step comprises the classification of the bridge and
the identification of the geometric properties of interest (number of spans, skew angle,
span width, bridge length, and maximum span length). Then, the peak ground
acceleration and spectral accelerations are evaluated considering eventually the
amplification due to soil conditions. In the following step three modification factors
(taking into account the skew of the bridge, skewK , spectral ordinates, shapeK , and tri-
dimensional arch action in the deck, 3DK ) are evaluated and the medians for the standard
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
135
fragility curves are appropriately modified. Finally, the fragility curves are calculated for
the corrected medians and the given values of dispersion.
For a complete evaluation of the functionality of a bridge structure, the probability of
exceeding ground-shaking related damage states should be combined with the probability
of exceeding damage states related to ground failure. Since ground failure was not
considered in the experiments, only the former source of damage is examined in the
following.
4.6.3. The substitute structure methods
The basis of this family of methods is the substitution of the real MDOF structure with an
equivalent SDOF one [Shibata & Sozen, 1976], following the concepts developed for the
direct displacement-based design, e.g. [Priestley & Calvi, 1997]. The most difficult task is
the definition of the effective properties of the equivalent structure, namely: deformed
shape, stiffness/period and damping. Using the equivalent damping and effective period,
the displacement is estimated from a displacement spectrum.
Following the CS method [Freeman, 1998], the seismic demand is obtained through an
acceleration-displacement response spectrum (ADRS) that is a plot of spectral
acceleration versus spectral displacement. The capacity curve of the structure, obtained
by means of pushover analysis, is superimposed on the ADRS. Then, the performance
point is defined at the intersection of the capacity and demand curves. The performance
point gives the displacement that the substitute structure will experience for the given
earthquake. Iterations are necessary to define the performance point for the correct value
of structural damping, since damping is considered a function of displacement. Several
expressions of this relation have been proposed for different structural systems, e.g. [Iwan
& Gates, 1979]. The CS method is adopted in the FHWA Seismic Retrofitting Manual for
highway bridges [Buckle & Friedland, 1995].
Various alternatives have been proposed, such as the N2 method [Fajfar & Gašperšic,
1996], which was adopted by EC8-1 [CEN, 2003a] for the design of new structures. The
first step of the N2 method is the non-linear analysis of a MDOF model of the structure.
Pushover analysis is performed for a distribution of forces that corresponds to the
distribution of inertia forces due to an assumed time-invariant displacement shape. The
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
136
deformed shape can be the one corresponding to the first eigenmode. This choice is valid
for the case of single bridge piers [Kowalsky et al., 1995] and regular bridges [Calvi &
Kingsley, 1995; Fajfar et al., 1997]. For the case of irregular bridges the contribution of
higher modes is expected to be significant and therefore, a different deformed shape must
be adopted. It is believed that, within rational limits, different assumptions will produce
similar results [Fajfar et al., 1997]. Suggested displacement shapes for different types of
bridges are presented in Figure 4.29.
Based on the computed response, a bilinear force-displacement curve is determined for
the equivalent SDOF system. The force is the sum of the reactions at the pier bases, while
the displacement corresponds to the control point, which depends on the structural
system. It is proposed to consider the point of the deck where the maximum displacement
is expected [Fajfar et al., 1997], or the top of the critical pier where failure occurs first
[Panagiotakos et al., 2003]. Both definitions require an a priori knowledge of the dynamic
structural response. In the case of a regular bridge these points will probably coincide, but
it might not be so in an irregular bridge. The stiffness of the equivalent structure is
*y yK F / u= (4.29)
where yF is the yield force and yu is the yield displacement of the bilinear curve. EC8-2
[CEN, 2002] adopts a bilinear curve where the yield force is equal to the force at the
formation of the mechanism. The yield displacement is such that the area under the actual
curve, mE , is the same as the area under the bilinear curve
y m m yu 2(u E / F )= − (4.30)
Figure 4.29. Suggested displacement shapes, adapted from [Fajfar et al., 1997]
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
137
where mu is the displacement at the formation of the mechanism. The effective period is
then calculated according to the expression
* * *T 2 m / K= π (4.31)
The mass, *m , of the SDOF structure is [Fajfar & Gašperšic, 1996]
*i im m= Φ∑ (4.32)
where im is the mass of each DOF and iΦ is the assumed displacement.
The seismic demand for the SDOF structure is defined by inelastic displacement spectra.
The inelastic displacement spectrum, dS , is obtained from the elastic one, deS , following
the relation
d deS SRµ
µ= (4.33)
where µ is the displacement ductility factor and Rµ is a reduction factor due to ductility.
A simple expression for Rµ may be adopted [Fajfar, 2000]
c
TR ( 1) 1
Tµ = µ − + ; cT T< (4.34a)
Rµ = µ ; cT T≥ (4.34b)
where cT is the period that defines the transition from the constant acceleration to the
constant velocity part of the spectrum. From Equations 4.33 and 4.34b, it is shown that
the equal displacement rule holds for structures with long periods. The local demand is
obtained by imposing the spectral displacement on the control point of the MDOF system.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
138
4.6.4. Application to the Talübergang Warth Bridge tests
HAZUS
For the damage probability assessment according to HAZUS, the geometric data of the
bridge are first needed for the classification of the structural type. The bridge has single
column bents, the deck is made up by a prestressed continuous box girder and the design
is identified as conventional (as opposed to seismic). Then, the bridge is classified as type
HWB20. The seismic input is defined by the mean acceleration response spectrum for 5%
damping, shown in Figure 4.30a.
The correction factors can now be estimated for the structural type and spectral
accelerations, as reported in Table 4.11. For the given structural type, the median values
for each damage state are also given in Table 4.11. The modified medians and the
dispersion β = 0.4 are used for the estimation of the probability of exceeding each
damage state for the three earthquake tests. The fragility curves are plotted in Figure 4.31.
The control period and the spectral accelerations for the tested model of the bridge have
been modified following the similitude law. For the prototype bridge the control period
for the ground acceleration is PT = 1.0 sec, while for the model it is 1M PT T−= λ = 0.4 sec.
Accordingly, the spectral acceleration for the prototype is 1P Ma a−= λ . According to this,
the acceleration at MT = 0.4 sec is read from the spectra of Figure 4.30a, it is divided by
2.5 to obtain the acceleration of the prototype and this value is entered in the fragility
curves of Figure 4.31. As seen in Table 4.12, no damage is predicted for the 0.4xNE test,
which corresponds to minor cracking and need for cosmetic repair only. This is consistent
with the observed damage, as described in section 4.4.2. After the low-level earthquake
the bridge is expected to be fully functional and only a small repair cost (less than 10% of
the replacement cost) will be required. For the 1.0xNE test the probabilities of exceeding
all damage states are similar, therefore no judgement can be made. Finally, complete
damage is predicted for the 2.0xNE test. This is in agreement with the experimental
observations that indicate collapse of one pier and complete damage of the bridge.
Table 4.12 presents also the probabilities of exceeding each damage level for earthquake
input described by the EC8 response spectrum Type 1 for ground type B, damping 5% of
critical and PGA = 0.36 g, see Figure 4.30b. For the 0.4xNE test no damage or slight
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
139
damage is predicted. For the 1.0xNE similar probabilities of exceeding all damage states
are predicted, with slightly higher probabilities for moderate or more severe damage.
Finally, for the 2.0xNE test, complete damage is predicted, with higher probability
compared to the prediction for the PSD tests of the same amplitude. The expected repair
cost ranges between 30% and 100% percent of the replacement cost. The bridge would be
fully functional after more than 1.5 years from the date of the event.
The HAZUS methodology underestimates the damage observed during the 1.0xNE and
2.0xNE PSD test and predicts serious damage for the EC8 spectrum. This is due to the
characteristics of the accelerograms used in the PSD tests: the spectral accelerations at 1.0
sec are relatively low. On the contrary, the spectral accelerations at the same period for
the EC8 spectrum are higher by about 30 % (see Table 4.12). The vulnerability of the
bridge is verified, since the bridge would suffer severe damage for the EC8 design
earthquake and would most probably collapse for the 2.0xNE input.
0.0
0.5
1.0
1.5
2.0
2.5
0.0 0.5 1.0 1.5 2.0 2.5Period (sec)
Acc
eler
atio
n (g
)
0.4xNE
2.0xNE1.0xNE
(a)
0.0
0.5
1.0
1.5
2.0
2.5
0 1 2 3 4
Period (sec)
Acc
eler
atio
n (g
)
0.4xNE
2.0xNE1.0xNE
(b)
Figure 4.30. Acceleration response spectra for 5% damping: mean experimental (a) and
EC8 Type 1, Soil Class B and PGA = 0.36 g (b)
Table 4.11. Parameters for estimation of damage probability
Median aS (g)
Damage level Original Modified
skewK 1.00 Light 0.35 0.35
shapeK 2.19 Moderate 0.42 0.44
3DK 1.05 Extensive 0.50 0.53
shapeI 0.00 Complete 0.74 0.78
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
140
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
Spectral acceleration at 1.0 sec (g)
Pro
babi
lity
of e
xcee
ding
dam
age
stat
e
SlightModerateExtensiveComplete
Figure 4.31. Fragility curves for the Talübergang Warth Bridge
Table 4.12. Assessment of the Talübergang Warth Bridge according to HAZUS
Probability of exceeding damage level
Sa(1.0) No Slight Moderate Extensive Complete
Exp. 0.17 1.0 0.0 0.0 0.0 0.0 0.4xNE
EC8 0.22 0.9 0.1 0.0 0.0 0.0
Exp. 0.42 0.3 0.2 0.2 0.2 0.1 1.0xNE
EC8 0.54 0.1 0.2 0.2 0.3 0.2
Exp. 0.84 0.0 0.0 0.1 0.3 0.6 2.0xNE
EC8 1.08 0.0 0.0 0.0 0.2 0.8
Capacity Spectrum method
The performance-based assessment methods discussed before were applied to the tested
bridge structure. For the CS and N2 methods, a non-linear model of the complete
structure was subjected to incremental lateral loading. The forces followed either a
trapezoidal, or triangular distribution, shown in Figure 4.32. Two alternatives were
considered for the control point. The first is the top of pier A30, which had the maximum
displacement in all the PSD tests and the second is the top of pier A70, which collapsed
during the 2.0xNE test.
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
141
Figure 4.32. Distribution of lateral forces
0
500
1000
1500
2000
2500
3000
3500
0.00 0.05 0.10 0.15 0.20 0.25
Displacement (m)
For
ce (k
N)
Pier A30Pier A70
Figure 4.33. Force-displacement curves from pushover analysis
Table 4.13. Equivalent damping (%)
A20 A30 A40 A50 A60 A70 Total
0.4xNE 2.42 1.04 2.88 0.33 0.71 2.85 2.19
1.0xNE 1.04 0.96 4.46 0.69 0.06 2.85 3.20
2.0xNE 0.87 0.59 8.27 0.45 0.08 12.24 10.0
The force-displacement curves from the pushover analysis for trapezoidal distribution of
lateral forces are presented in Figure 4.33, along with the bilinear approximation
according to EC8. As expected, if the control point is the top of pier A70, a much stiffer
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
142
response is obtained. The curves are similar for triangular distribution of the lateral
forces.
The value of equivalent damping for the complete structure is calculated as a weighted
sum of the experimental values for each pier. The weight is the percentage of dissipated
energy for each earthquake test. The equivalent damping, eqξ , for each pier is estimated
for the cycle with maximum displacement for each test, according to the expression
Deq
max max
E2 F u
ξ =π
(4.35)
In the previous expression DE is the dissipated energy in the cycle, maxF and maxu are
respectively the mean values of maximum force and displacement in positive and
negative directions. The resulting values are reported in Table 4.13, where it is seen that
low values of damping correspond to the small and nominal-amplitude tests.
At this point, both the capacity and demand curves can be estimated. The capacity curve
is obtained by dividing the force of the bilinear curve by the mass of the equivalent
structure, see Equation 4.32. In Figure 4.34 the capacity curves are superimposed on the
mean ADRS of the accelerograms used in the PSD tests. For each earthquake intensity
the equivalent damping reported in Table 4.13 was used for the calculation of the ADRS
and hence, no iterations are required. The displacements of the control points for each
earthquake intensity are reported in Table 4.14. The displacements of the pier tops for
each earthquake intensity and the different control points and distributions of lateral
forces are compared to the maximum experimental displacements in Figure 4.35.
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08Displacement (m)
Acc
eler
atio
n (g
)
A30 A70 (a)
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08Displacement (m)
Acc
eler
atio
n (g
)
A30 A70 (b)
Figure 4.34. Evaluation of performance point for mean experimental ADRS: trapezoidal
(a) and triangular distribution of forces (b)
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
143
0.00
0.01
0.02
0.03
0.04
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
Triangular
Trapezoidal
Experimental
0.4xNE
0.00
0.01
0.02
0.03
0.04
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
0.4xNE
0.00
0.05
0.10
0.15
0.20
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
1.0xNE
0.00
0.05
0.10
0.15
0.20
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
1.0xNE
0.00
0.05
0.10
0.15
0.20
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
2.0xNE
0.00
0.05
0.10
0.15
0.20
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
2.0xNE
Figure 4.35. Maximum displacement of the piers: experimental values and CS method for
control at top of pier A30 (left column) and control at top of pier A70 (right column)
0
1
2
3
4
5
0.00 0.05 0.10 0.15 0.20Displacement (m)
Acc
eler
atio
n (g
)
A30 A70 (a)
0
1
2
3
4
5
0.00 0.05 0.10 0.15 0.20Displacement (m)
Acc
eler
atio
n (g
)
A30 A70 (b)
Figure 4.36. Evaluation of performance point for EC8-compatible ADRS: trapezoidal (a)
and triangular distribution of forces (b)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
144
Table 4.14. Displacement of control point (m) for mean ADRS
Trapezoidal distribution of forces
0.4xNE 1.0xNE 2.0xNE
CS N2 CS N2 CS N2
A30 0.019 0.008 0.031 0.020 0.050 0.035
A70 0.024 0.009 0.038 0.022 0.054 0.042
Triangular distribution of forces
0.4xNE 1.0xNE 2.0xNE
CS N2 CS N2 CS N2
A30 0.024 0.008 0.039 0.020 0.056 0.035
A70 0.003 0.010 0.049 0.025 0.061 0.047
Table 4.15. Displacement of control point (m) for EC8 ADRS and triangular distribution
of lateral forces
0.4xNE 1.0xNE 2.0xNE
CS N2 CS N2 CS N2
A30 0.035 0.042 0.119 0.105 0.175 0.210
A70 0.025 0.043 - 0.108 - 0.215
Table 4.16. Characteristic values of the equivalent bilinear structures
Trapezoidal Triangular
A30 3440 3406 yF (kN)
A70 3440 3406
A30 0.08 0.08 yu (m)
A70 0.02 0.02
A30 43000 42575 *K (kN/m)
A70 172000 170300
A30 2.66 2.03 *T (sec)
A70 1.33 1.02
The trapezoidal distribution of forces provides a very poor approximation of the
experimental response of the bridge. The triangular distribution of forces for control point
at the top of pier A30 gives better results for the 0.4xNE and the 1.0xNE, for which the
dynamic response followed mainly the first two eigenmodes, although in some cases the
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
145
displacements are underestimated by as much as 2.5 times. On the other hand, the
triangular distribution for control point at the top of pier A70 significantly overestimates,
by about two times, the experimental displacements. The 0.4xNE test is an exception
because of the small demand that is defined by the ADRS spectrum, see Figure 4.34b.
In Figure 4.36 the capacity curves are superimposed on the ADRS spectra according to
EC8. The ADRS spectra are calculated considering the values of damping in Table 4.13
for each test. It is reminded that the similitude law was applied in the construction of the
spectra of Figure 4.36 The acceleration, displacement and period of the prototype, given
in EC8, were appropriately scaled to obtain the values for the tested model. It is seen that
the bridge cannot meet the seismic demand for the 1.0xNE and 2.0xNE intensities when
the control point is considered at the top of pier A70.
The predicted displacements of the control points are presented in Table 4.15. Only the
results for triangular distribution of lateral forces are presented. The displacements
predicted for the EC8 spectrum are 1.5 times or 3 times, for the 0.4xNE or the 1.0xNE
and 2.0xNE tests respectively, the displacements predicted for the mean experimental
ADRS. These values correspond to moderate damage for the 0.4xNE test, severe damage
for the 1.0xNE test and imminent collapse for the 2.0xNE test. It is reminded that this
method underestimated the experimental displacements and therefore more damage
would be actually expected. Qualitatively, these results are in agreement with the damage
predicted by the HAZUS methodology and once again verify the high seismic
vulnerability of the bridge.
N2 method
For the N2 method, the mean displacement spectra for 5% damping are presented in
Figure 4.37a for the three earthquake tests. Following Equations 4.29 to 4.32, the
characteristic properties are estimated and reported in Table 4.16. As observed before,
stiffer response is obtained if the control point is the top of the critical pier, A70.
Comparing the trapezoidal and triangular distributions of forces, no major differences are
observed in terms of stiffness, but different periods are calculated. This is because the
effective mass is different for the two assumed displacement shapes, as shown in
Equation 4.32. Then, different periods correspond to different spectral displacements. As
shown in Figure 4.37a, the stiffer equivalent structure has higher displacement for the
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
146
accelerograms used in the PSD tests. For the EC8 displacement spectrum the periods of
the equivalent structure, appropriately scaled from the model to the prototype, fall within
the plateau of maximum displacement or within the descending branch, as seen in Figure
4.37b. All the periods are higher than CT , therefore the equal displacement rule applies
and the inelastic displacement is equal to the elastic displacement (see Equation 4.34).
Table 4.14 compares the displacement of the control point, as results from the two
alternative methods. The displacements predicted by the N2 method are 3, 2 and 1.6 times
smaller than the displacements predicted by the CS method, for the 0.4xNE, 1.0xNE and
the 2.0xNE tests respectively. This difference is of the same order independently of the
control point and the distribution of lateral forces.
The last step in the assessment procedure is the application of the target displacement,
shown in Table 4.14, to the control point, following the assumed deformed shape. The left
column of Figure 4.38 presents the results of the N2 method for control at the top of pier
A30 and the right column presents the results for control at the top of pier A70. The
trapezoidal distribution of forces produces results in poor agreement with the
experimental data, both in terms of absolute values of displacement and in terms of
deformed shape. It is concluded that this deformed shape cannot sufficiently describe the
dynamic response of the bridge. On the other hand, the triangular distribution of forces
better approximates the actual deformed shape of the bridge. Nevertheless, the predicted
displacements diverge from the experimental values. When considering the top of pier
A30 as the control point, the predictions underestimate the displacements by more than
4.5 times in the worst case. For the control point at the top of pier A70 the prediction
gives a better approximation of the distribution of the maximum experimental
displacements. Considering the absolute values, the prediction underestimates some
displacements by as much as 70 %.
Table 4.15 presents the values of displacement of the control points for the two methods,
following the triangular distribution of forces, and the EC8 (Type 1 and Soil Class B)
ADRS spectra. As for the CS method, the predicted displacements are significantly higher
than the displacements for the mean experimental displacement spectra. The N2 method
predicts moderate damage for the SLS earthquake, severe damage for the ULS earthquake
and imminent collapse for the 2.0xNE.
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
147
0.00
0.02
0.04
0.06
0.08
0.0 0.5 1.0 1.5 2.0 2.5Period (sec)
Dis
plac
emen
t (m
)
0.4xNE
2.0xNE
1.0xNE
(a)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0 2 4 6 8 10 12
Period (sec)
Dis
plac
emen
t (m
)
0.4xNE1.0xNE
2.0xNE
(b)
Figure 4.37. Mean displacement spectra for use in N2 method: experiment (a) and EC8
Type 1 for Soil Class B (b)
0.00
0.01
0.02
0.03
0.04
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
Triangular
Trapezoidal
Experimental
0.4xNE
0.00
0.01
0.02
0.03
0.04
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
0.4xNE
0.00
0.02
0.04
0.06
0.08
0.10
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
1.0xNE
0.00
0.02
0.04
0.06
0.08
0.10
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
1.0xNE
0.00
0.05
0.10
0.15
0.20
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
2.0xNE
0.00
0.05
0.10
0.15
0.20
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
TriangularTrapezoidalExperimental
2.0xNE
Figure 4.38. Maximum displacement of the piers: experimental values and N2 method for
control at top of pier A30 (left column) and control at top of pier A70 (right column)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
148
0.0
0.1
0.2
0.3
0.4
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)N2 methodN2 correctedYieldingUltimate
0.4xNE
0.0
0.1
0.2
0.3
0.4
A20 A30 A40 A50 A60 A70
Dis
plac
emen
t (m
)
N2 methodN2 correctedYieldingUltimate
1.0xNE
Figure 4.39. Assessment of the bridge for the EC8 spectrum and N2 method
Among the examined solutions, the N2 method for triangular distribution of forces and
control point at the top of pier A70 gives the best approximation of the experimental
behaviour. Figure 4.39 plots the results of the N2 method for the EC8 spectrum with peak
ground acceleration ga = 0.36 g and control point at the top of pier A70. The yield and
ultimate displacements of the piers are also plotted. It is expected that displacements (and
damage) might be underestimated in certain cases. For the 0.4xNE input, yielding of all
the piers with the exception of A20 is predicted. The highest ductility demand, µ = 2.4,
corresponds to pier A50. For the 1.0xNE input all piers are beyond the yield point and
piers A40, A50, A60 and A70 reach their capacity. The highest ductility demand, µ =
10.8, corresponds to the short pier A70. For this input, which corresponds to the SLS, the
bridge is considered to fail and then the no-collapse criterion is not met.
The predicted response shown in Figure 4.39 could be unrealistic due to the difference
between the assumed and the experimentally observed deformed shapes of the bridge.
Therefore, an attempt is made to correct the results of the N2 method in order to follow
the effective deformed shape. The pier top displacements are multiplied by a correction
factor, which is the ratio of the experimental displacement and the value predicted by the
N2 method for triangular distribution of forces and control at the top of pier A70. This is
an interpretation of the proposal [Fajfar & Gašperšic, 1996] to apply appropriate dynamic
amplification (or de-amplification) factors to selected design quantities when the effects
of higher modes are significant. It is reminded that the deformed shape changes for each
earthquake test and this can be attributed to the change of the dynamic properties of the
piers due to the damage suffered in the previous tests. Although cumulative damage is not
considered in the simplest version of the method, the correction seems more meaningful if
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
149
the damage pattern predicted for the EC8 spectrum is similar to the damage observed in
the PSD tests. This will be verified in the following.
The displacements of the piers according to the corrected N2 method are also plotted in
Figure 4.39. For the 0.4xNE input all the piers are beyond the yield displacement, while
only the short pier A70 is at incipient yielding. The distribution of damage is similar to
the pattern observed in the PSD tests, therefore the correction is considered reliable. The
maximum ductility demand, µ = 3.2, corresponds to pier A30. The values of drift and
ductility indicate that most of the piers develop almost half their deformation capacity,
which means that they suffer heavy damage and require significant repair. For the 1.0xNE
input all the piers have yielded and piers A20 and A40 have exceeded their deformation
capacity. The maximum ductility demand, µ = 7.5, corresponds again to pier A30. The
damage of all piers, as defined by the drift and ductility demands, is severe because they
either collapse, or develop more that 70% of their deformation capacity. At this level of
damage, repair might not be economically feasible and replacement of the bridge would
be required. This again verifies the seismic vulnerability of the bridge.
It should be kept in mind that the simple procedures used in the substitute structure
method do not account for cycling effects. In the case of poorly designed bridge piers
(lack of confinement reinforcement) this results in rapid strength and stiffness
degradation. Then, the displacement capacity of the monotonic curves is higher than the
actual capacity during an earthquake and the damage is underestimated. Certain
modifications have been proposed [Fajfar, 2000]. One possibility is to apply an equivalent
ductility factor that reduces the monotonic deformation capacity as a consequence of
cumulative damage. Alternatively, the seismic demand may be increased.
In addition, the method predicts the performance of the bridge subjected to an earthquake
without previous damage. This was not the case of the tested bridge, particularly for the
2.0xNE test. The previous tests had caused cracking and yielding of the piers and then the
stiffness and dynamic properties of the bridge were altered.
Comparing the examined methods, the HAZUS method results superior to the CS and the
N2 methods in terms of simplicity. Very few input data are required by the HAZUS
methodology and all the parameters are clearly defined and quantified. On the other hand,
the deterministic methods require more detailed information on the structure (not always
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
150
available) and the earthquake input and demand relatively refined analyses. In addition,
uncertainties related to the deformed shape, control point and the relation between
equivalent damping and lateral displacement require analyses of alternative cases. When
damage and changes in dynamic properties significantly influence the results of
simplified analysis methods, these results can only be used for qualitative comparisons.
This does not justify the demanded effort and in such case, probabilistic methods are
advantageous. It emerges from this application that, apart from the need for engineering
judgement and experience, non-linear dynamic analysis is the best available tool for the
reliable assessment of irregular structures.
4.7. CONCLUDING REMARKS
A few words should be spent on the implementation aspects of the PSD tests, and
rightfully so, because these tests were the first at world-level to have been performed
considering non-linear behaviour for the substructured part and asynchronous input
motion. The tests were performed with two physical piers tested in the lab and the rest of
the piers, abutments and the deck modelled in the computer. The implemented
substructuring technique was proved to be representative of an earthquake test: the
experimental results were in fair agreement with pre-test results of dynamic analyses,
applying two alternative time integration schemes. The simplified numerical models
guaranteed reasonably short computation time and reliable results. It is interesting to note
that the testing part has been completely controlled remotely and that the connection
between the various processes used standard internet features. Thus, this test campaign
showed that the tele-operation of experimental facilities, further combined with
sophisticated numerical algorithms running on decentralised hardware, is already a
working reality [Pinto et al., 2004].
From the engineering point of view, as far as seismic assessment of existing bridges is
concerned, the results from the tests represent a data set, which allowed to calibrate
numerical models and to assess the performance of a typical European bridge (highway
bridge with rectangular hollow cross-section and with many seismic deficiencies such as
short overlapping and development lengths, lack of transversal reinforcement to prevent
buckling, tension shift effect, absence of capacity design requirements, etc). The PSD
SEISMIC ASSESSMENT OF AN EXISTING HIGHWAY BRIDGE
151
tests demonstrated that these infrastructures represent a great source of risk in seismic
regions.
It is important to note that the damage patterns change with the intensity of the input
motion. In fact, the 0.4xNE PSD test, hypothetically corresponding to the SLS, caused
only minor damage (cracking) in almost all piers but the physical short pier that remained
elastic. Then, damage concentrated at the tall physical pier, A40, for the 1.0xNE test,
hypothetically corresponding to the ULS, whereas the numerical piers A30 and A50 were
also beyond yielding. However, the non collapse criterion was satisfied. Finally, collapse
was reached at the short pier, A70, for the 2.0xNE test. Such behaviour is termed
sequential yielding in EC8-2 [CEN, 2002], while the optimal behaviour is characterised
by simultaneous yielding of the piers. Sequential yielding of piers is expected to cause
deviations between the results of simplified linear analysis and the actual non-linear
response of the bridge. In particular, it can induce additional horizontal eccentricity and
also increase the transverse bending of the deck. For these reasons, non-linear time-
history analysis is demanded by EC8-2 in such cases.
Note also that for the tall physical pier, A40, and the numerical piers A30, A50 and A60
the most heavily damaged region was above the bar cut-off at about one third of the total
height of the scaled specimen (8.75 m for the real physical pier A40), further
complicating the eventual repair works. The above reflect the absence of design
strategies, which are presently included in the design codes for new structures.
The drift and displacement ductility capacities of the bridge piers do not meet the
requirements of modern codes for new bridge structures. A limited capacity of hysteretic
energy dissipation is also observed for all piers due to the lack of detailing for seismic
resistance.
Concerning the quantification of damage, the difficulty in defining an overall damage
index for the whole bridge structure was discussed. Because of the irregular distribution
of damage among the bridge components-substructures, a weighted sum of their damage
indices was used to estimate the overall damage index. Among the damage-weighted and
the energy-weighted sums, the latter described better the performance of the tested bridge
structure.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
152
The cyclic effects were evident since they resulted in a significant reduction of the
resistance of the bridge components and smaller displacement capacity, compared to the
same components tested under quasi-static conditions for a few cycles of increasing
displacement. For the short physical pier, A70, the larger number of cycles initiated
failure of the lapped splices, as indicated by the vertical cracks near the corner of the pier
in Figure C.6b.
The observed damage was compared to the predictions of simplified assessment methods.
The simplicity of the probabilistic methods was highlighted against the uncertainties
(concerning the deformed shape, control point, relation between equivalent damping and
lateral displacement, cycling effects and cumulative damage) encountered during the
application of deterministic methods that make use of a substitute structure. Due to the
differences between the assumed and the actual deformed shapes, the deterministic
methods were unable to predict the damage distribution observed during the tests and
their results could be used only for qualitative comparisons. Several improvements are
needed in these simplified methods in order to account for the aforementioned problems.
It was possible to introduce a correction of the N2 method in order to account for the
correct deformed shape. Nevertheless, this correction requires the knowledge of the actual
structural behaviour, in other words a complete non-linear dynamic analysis. Among the
available tools, dynamic non-linear analysis is the most appropriate for the detailed and
reliable assessment of existing structures with irregular configuration.
It is underlined that this bridge was considered in a low-seismicity zone (the prototype
was a viaduct in Austria) with near-field earthquakes, i.e. energy content only at high
frequency ranges. However, similar bridges exist in medium and high-seismicity zones in
Europe and with different earthquake scenarios, e.g. Italy, Greece, and Portugal. In this
last situation, it is felt that the structure would reach collapse for earthquake intensities
even below the nominal ones. This was qualitatively confirmed by the results of the
simplified assessment methods. Indeed, for the SLS earthquake significant inelastic
deformation demand, and consequently retrofit cost, is expected for all piers. For the ULS
earthquake collapse of two piers and severe damage (close to collapse) for the remaining
piers is expected. At the event of the ULS earthquake, repair might be economically
unfeasible and the solution would be the replacement of the bridge structure.
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
153
5. NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED
BRIDGE PIERS
5.1. INTRODUCTION
Numerical analysis is a procedure complementary to experimental testing. While test
results serve to validate numerical models, numerical analysis is used to generalise the
experimental findings. Considering reinforced concrete (RC) structures, either detailed or
simplified models may be employed. Detailed models make use of the Finite Element
Method (FEM) with appropriate mesh and constitutive laws for concrete, steel
reinforcement and contact interfaces. Reliable results are not easy to achieve and one has
to consider the computation time. Fibre-type models, which describe the behaviour at the
cross-section level, belong to this category. On the other hand, simplified models aim at
representing the global behaviour of structural elements such as columns, beams and
walls.
An attractive combination of the previous approaches is the use of fibre models for the
description at the cross-section level and the use of beam models for the description of the
flexural response at the element level. This combination offers a reliable and practical
solution for the non-linear analysis of RC elements, especially in the case of bending
combined with axial load. Non-linear cyclic constitutive laws can be used for both
concrete and steel and the results of the calculations include information on the stress and
strain distribution within the cross-section.
The use of macro-level models for reinforced concrete seen as a composite material has
been proposed also [CEB, 1996]. These models consider the smeared properties of the
composite material as an average of the effect of discrete rebars (contribution of steel)
and cracks (contribution of concrete). They have been mainly used for plate elements
with uniformly distributed reinforcement in plane stress conditions under both monotonic
and cyclic loading. Such models will not be examined in this work.
A fibre/Timoshenko beam model is described first, along with the cyclic constitutive laws
for concrete and steel. This modelling approach is adopted to simulate the response of the
two piers tested under static cyclic loading, as described in Chapter 3. The advantages and
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
154
limitations of this type of modelling are highlighted. In the following, appropriate models
are elaborated for the piers that were numerically simulated during the PSD tests
presented in Chapter 4. To overcome the problems related to shear deformation and
account for slippage between steel rebars and the surrounding concrete, the response of
the two piers is simulated using the FEM method. The constitutive laws for the materials
and interfaces are discussed and the results of the numerical simulation are compared to
the experimental data from the cyclic tests.
It is verified that the FEM is appropriate for studies where detailed information is needed
at local level, while fibre/Timoshenko beam (FTB) models are useful for parametric
studies where global behaviour is of interest. Both approaches are used for the numerical
analysis of piers with rectangular hollow cross-section retrofitted with fibre-reinforced
polymer (FRP) jackets. The FEM is applied for the study of the effect of confinement on
the concrete properties in piers with this particular geometry. The FTB model is then used
to study the global behaviour of retrofitted piers. The results of the FEM analyses are
used to define the material properties for concrete in the analyses with the FTB model.
This approach is validated against experimental data and will be used for parametric
analyses discussed in the following chapter.
5.2. FIBRE/TIMOSHENKO BEAM MODELLING
In support of the PSD tests on the bridge structure, presented in the previous chapter,
numerical pre-test analyses have been performed using a fibre/Timoshenko beam element
implemented in the finite element code Cast3m [Millard, 1993]. In this section a general
description of the model and constitutive laws for concrete and steel is given. Then,
alternative modelling configurations for the bridge piers are discussed and the results of
the FTB model are compared to the experimental data presented in Chapter 3 and to the
results of more refined analyses.
5.2.1. The Fibre/Timoshenko Beam element in Cast3m
A two level approach is adopted for this model: the first is the description of the section
(fibre elements) and the second is the description of the Timoshenko beam. At the section
level, the usual fibre modelling is adopted, which accounts only for interaction between
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
155
normal force and bending moment. The section is described by a model with various sub-
zones that correspond to different materials, namely steel and concrete. Linear bi-
dimensional (2D) elements, namely triangular with three nodes or quadrangular with four
nodes, may be used to construct the cross-section mesh. The components of interest at the
cross-section level are the normal, x xf ( ,...)σ = ε , and shear, xyτ and xzτ , stresses. Only
elastic shear is considered. At the element level, the Timoshenko beam theory is adopted,
assuming that plane sections remain plane after deformation, but not necessarily normal
to the beam axis.
The evaluation of the stress resultant for each beam element proceeds as follows
(i) Evaluation of the generalized strain E at the integration point of each beam element,
from the nodal generalized displacement ( 1U , 1Θ , 2U , 2Θ ).
(ii) Use of the beam model in order to evaluate the strain tensor ε and in particular its
normal component, xε , at the level of each fibre, located at the Gauss integration points
of the elements describing the section.
(iii) Use of the constitutive relationship in order to evaluate the stress tensor σ at the
level of each fibre, in particular its normal component, xσ .
(iv) Integration over the section of the relevant stress components in order to compute the
generalized stress F for the section.
(v) Computation of the stress resultant ( 1F , 1M , 2F , 2M ) for the beam element.
5.2.2. Constitutive laws
Concrete
Concrete behaviour is represented by a parabolic curve up to the peak stress point
followed by a straight line in the softening zone. Confinement is taken into account by the
modification of the plain concrete curve and including an additional plateau zone at the
residual strength. Cyclic behaviour accounts for stiffness degradation and crack closing
phenomena. Tensile resistance is also considered.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
156
Strain
Str
ess
unconfinedconfined
(a)
Strain
Str
ess
(b)
Figure 5.1. Monotonic constitutive law for in compression (a) and tension (b)
The constitutive law for unconfined concrete in compression consists of two branches, as
shown in Figure 5.1a. The first branch, which is a parabolic function, describes the
ascending branch of the curve from zero stress to the maximum compression strength
co co co
2 σ ε ε
= − σ ε ε (5.1)
where coσ and coε are respectively the compressive strength and the corresponding strain.
The second branch is linear and describes the softening behaviour after maximum stress
and until failure
( )coco
1.0 Zσ
= + ε − εσ
(5.2)
where Z is the slope of the descending branch and depends on the degree of confinement.
The maximum stress and corresponding strain can be modified to account for
confinement, as shown in Figure 5.1a. For confined concrete, a third branch is considered
after the softening branch. This describes a compression plateau with constant value equal
to a residual strength, ptσ . The value pt co0.2σ = σ has been proposed [Park & Priestley,
1982].
In tension, a bilinear stress-strain curve has been adopted, as shown in Figure 5.1b. From
zero stress to the maximum tensile strength, tσ , the model presents a linear elastic
behaviour with a slope equal to the initial compression Young modulus, oE
ptσ
coσ
coε
tσ
tε
tmε
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
157
oEσ = ε (5.3)
The second branch describes the softening behaviour after cracking and follows a straight
line
tt
r ( / )r 1
− ε εσ = σ
− (5.4)
where
tm tr /= ε ε (5.4)
In the above expressions tσ is the tension strength and tmε is the deformation at zero
stress.
The compression monotonic curve is the envelope of the concrete behaviour under cyclic
compressive loading. Unloading from the envelope follows a straight line with a slope
depending on the maximum strain, maxε , ever reached during the loading history
( )
( ) ( )
2max co
d o 2max co max co
/E E 1
1 / /
ε ε= −
+ ε ε + ε ε (5.5)
The decrease of the slope with increasing maximum strain accounts for the degradation of
stiffness due to cycling. The reloading compression curve is also a straight line from zero
stress at the plastic strain, plε , until the last point reached on the envelop. No strength
degradation is considered.
Strain
Str
ess
Strain
Str
ess
Figure 5.2. Cyclic constitutive law for concrete in compression (a) and tension (b)
dE
maxεplε tmε
coε
tσ
plε
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
158
Strain
Str
ess
Strain
Str
ess
Figure 5.3. Monotonic (a) and cyclic (b) constitutive law for steel
Concerning tensile stresses, an envelope going from zero stress at coε to the tensile
strength is considered, as shown in Figure 5.2b. The envelope comprises a second linear
branch from the maximum tensile strength to the point of zero stress and strain tm trε = ε .
An appropriate value for tmε is the yield strain of the longitudinal reinforcing bars
[Barzegar-Jamshidi & Schnobrich, 1986]. Unloading after tr2 pl tmε = ε + ε follows a zero
stress path. When the maximum tensile strength is reached, no more tensile stresses can
be supported from the concrete in subsequent cycles. If no tensile stresses are considered
in the model, the plastic strain is determined by the zero stress point from the
compression unloading curve.
Steel
The monotonic constitutive law for steel is a five-parameter model. The first zone is
elastic and is defined by the Young modulus, E, and the yield strain, syε ,
syEσ = ε (5.6)
Then follows the yield plateau at the yield stress, syσ , until the hardening strain, shε ,
syσ = σ (5.7)
Finally, the hardening curve until the maximum stress, suσ , and corresponding strain, suε ,
follows a fourth-degree polynomial
suε
shε
syε
suσ syσ
oσ
rσ
rε
oε
o r( )ξ ε − ε
oE
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
159
4
susu su sy
su sh
( ) ε − ε
σ = σ − σ − σ ε − ε (5.8)
The same law is used also for compressive loading, if no buckling is considered for the
longitudinal rebars. Otherwise, some modifications are introduced.
A modified Menegotto-Pinto model is adopted for cyclic loading. If small unloading
cycles are imposed to the bars, no hysteretic effect is evidenced and a straight line with
slope equal to the initial Young modulus is adopted for both loading and unloading.
Otherwise, if the strain satisfies the condition
symax 3
εε − ε < (5.9)
where maxε is the maximum attained strain in the loading history, the behaviour of steel is
described by the relation
( )
* * *1 RR*
1 bb
1
−σ = ε + ε
+ ε
(5.10)
where
* s r
o r
σ − σσ =
σ − σ (5.11a)
* s r
o r
ε − εε =
ε − ε (5.11b)
1o
2
R Rα ξ
= −α + ξ
(5.11c)
In the previous expressions rσ and rε describe the last reversal point, b is the ratio
between the hardening slope, hE , and the initial slope, oE , and R is a parameter that
defines the shape of the transition branch of the curve. Parameters oσ , oε , and ξ are
defined in Figure 5.3b. Parameters oR , 1α and 2α should be obtained from experimental
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
160
data; the values oR = 20.0, 1α = 18.5 and 2α = 0.15 are proposed [Menegotto & Pinto,
1973].
Based on experimental results, buckling is considered important when the ratio between
the unsupported length, L, and diameter, D, of the rebars is greater than 5 [Monti & Nuti,
1992]. In that case, the unloading curve from tensile stresses and the reloading curve after
reversal from compression must be modified. The post-yield compression zone has a
softening behaviour and the b factor is accordingly modified. On the other hand, during
reloading after reversal from compression, also the Young modulus is reduced. Further
details on the implementation of the model can be found elsewhere [Guedes et al., 1994].
5.2.3. Alternative configurations for the cross-section and the beam element
In order to reduce the computational demand for the PSD tests and to increase the
robustness of the numerical models and procedures, a simple, yet accurate numerical
model should be used for the substructured piers. The most accurate choice is a refined
tri-dimensional (3D) finite element model with realistic constitutive laws for the materials
and the steel-to-concrete interface. It will be shown in the following that such a model
would demand exaggerated computation time. For this reason, it was decided to use a
fibre/Timoshenko beam element implemented in the finite element code Cast3m, as a
compromise between accuracy and simplicity.
Taking into account the symmetry of the geometry and the loading, a 2D beam element
was used. Different simplified configurations at the cross-section level have been
examined, as explained in detail in the following. Non-linear behaviour was assumed for
the concrete and elastic-perfectly plastic behaviour for the reinforcement steel. Numerical
analysis showed no significant difference between the results obtained using this bilinear
model for steel and those obtained with a, more realistic, modified Menegotto-Pinto
model. Table 5.1 presents the material properties used in the numerical model. Since no
closed stirrups or crossties were present in the tested piers, no confinement effect was
considered for the concrete.
Different configurations have been examined for the discretisation of the cross-section of
the piers. The results of each alternative configuration were compared and the adopted
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
161
numerical models were calibrated on the basis of the cyclic tests performed on scaled
models of the piers described in Chapter 3.
The bridge piers have a rectangular hollow cross-section. The corresponding fibre model
for the cross-section is presented in Figure 5.4a. The elements shown in blue correspond
to the (unconfined) concrete, whereas those shown in red correspond to the vertical
reinforcement steel. Each element used to model the reinforcement has four integration
points and an area equal to the area of the reinforcement of each face of the cross-section.
This configuration was initially chosen in order to be close to the geometry of the
specimens. Keeping in mind that the displacement was imposed in the strong direction of
the cross-section, the discretisation adopted for the steel elements was considered
adequate for the flanges, but not for the web. The vertical rebars of the specimen are
uniformly distributed along the two faces of the web and are expected to be under tension
gradually as the neutral axis moves along the web. This cannot be accurately modelled
with the initial configuration and could cause problems during the iterations for the PSD
tests with substructuring. The mesh for the cross-section was therefore modified in order
to avoid such eventual numerical problems.
Since the cross-section was symmetric and the loading would be applied along an axis of
symmetry, an alternative model at the cross-section level was examined, shown in Figure
5.4b. An equivalent I cross-section was considered with web width equal to the total
width of the two webs of the original cross-section. Rectangular elements with four
integration points each, better distributed along the web, were used for the steel
reinforcement. This configuration enabled a more realistic representation of the actual
distribution of vertical reinforcement along the web of the pier model, with similar
computational demand as for the original box cross-section.
Figure 5.4. Discretisation of alternative models for the cross-section
(a)
(d) (b)
(c)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
162
Table 5.1. Material properties used in the numerical models of the piers
Concrete Steel
cE (GPa) 33.5 cE (GPa) 200
Pν 0.2 Pν 0.3
cf (MPa) 43 yf (MPa) 545
tf (MPa) 3 uf (MPa) 611
coε (%) 0.257 syε (%) 0.5
Table 5.2. Characteristics of different models
Cross-section Concrete Steel Beam
A20 Hollow Section Section 3D
A20I I Section Section 3D
A20Ip I Section Point 3D
A20Ip_2D I Section Point 2D
A20s Line Segment Point 2D
0
200
400
600
800
1000
1200
1400
0.00 0.05 0.10 0.15 0.20
Displacement (m)
For
ce (k
N)
3D
3D
3D
2D
2D
Figure 5.5. Pier A20: force-displacement monotonic curves for different models
In an attempt to further simplify the model, a new element was implemented for the
reinforcement steel bars. The element originally used is a linear quadrilateral element
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
163
with four nodes. In order to reduce to minimum the integration points, an element with
one integration point was implemented. The geometrical support is a point, thus there is
one integration point instead of four. This enabled the use of more elements better
distributed to model the vertical rebars. The corresponding model of the cross-section is
shown in Figure 5.4c. The rebars of the flanges and the webs were lumped at elements
with equivalent area, positioned along the longitudinal axis of symmetry of the cross-
section. This configuration made it possible to reproduce the exact distribution of the steel
rebars of the tested model, without increasing, but actually decreasing, the computational
demand. In fact, the number of nodes for the steel elements was reduced from 40 (for the
discretisation shown in Figure 5.4b) to 25 (for the discretisation shown in Figure 5.4c).
Furthermore, a new element was implemented for the concrete fibres in order to meliorate
the speed of computation for the substructured models during the PSD tests. The new
element has only two integration points, since the geometrical support is a segment. The
corresponding model at the cross-section level is shown in Figure 5.4d. Also in this case
the number of nodes for the concrete elements was reduced from 112 (for rectangular
elements shown in Figure 5.4c) to 44 (for linear elements show in Figure 5.4d).
The original FTB model was used to perform 3D analysis, and then six degrees of
freedom (DOFs) were considered for each node of the beam element. The calculated
reactions were the forces along the three axes, namely xR , yR and zR , and the moments
by the three axes, namely xM , yM and zM . Considering the symmetry of the geometry
and the loading, only three DOFs were significant in the present case for the tested piers:
displacements along the vertical and longitudinal axes and rotation by the transversal
axis. A 2D beam model was used considering the cross-section shown in Figure 5.4d. The
results of all alternative models at the cross-section level are compared in Figure 5.5 for
pier A20. All alternative models gave consistent results. Therefore it was decided to
proceed with the simplest one.
Table 5.2 recapitulates the types of cross-sections and elements used for every alternative
model. Passing from the original hollow cross-section to the equivalent I cross-section, a
slightly larger resistance is observed, due to the larger lever arm of the steel fibres of the
web. The use of point elements for the steel fibres of the equivalent I cross-section results
in larger resistance, compared to the original configuration. This also can be attributed to
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
164
the discretisation of the vertical reinforcement fibres. Passing from the 3D beam element
to a 2D beam element, almost identical response is observed. Finally, for the 2D beam
element and using line elements for the concrete fibres the strength does not change
significantly.
5.2.4. Validation of the numerical models
Before the PSD tests, two scaled models of the piers A40 and A70 were tested under
cyclic loading. The results of the cyclic tests were used to calibrate the numerical models
for the substructured piers. The results of numerical simulations using a damage model
[Faria et al., 2001; Pouca, 2001] were also compared to the experimental values. The
main objective of the numerical simulation was the preparation of accurate models for the
numerical piers during the PSD tests. The fibre model was adequate for the case of the
short pier A70, whose failure mode was flexure-dominated. Pier A20 was expected to
have a similar failure mode. Concerning the tall pier A40, the fibre model resulted in
larger resistance and different failure location, whereas the damage model gave results
similar to the experimental values. The fibre model was unable to accurately represent the
shear effects in the lower part of the specimen. The same failure mode and overall
behaviour was expected also for piers A30, A50 and A60.
Piers with flexure-dominated behaviour
The experimental and numerical force-displacement curves for the short pier A70 are
compared in Figure 5.6. Very good agreement is observed in terms of strength and cyclic
behaviour. The numerical simulation was terminated at the displacement at which a
sudden drop of resistance is evidenced in the numerical curve. This was due to failure of a
steel fibre and occurred for a displacement similar to the experimental displacement that
corresponds to failure of the first vertical reinforcement bar. Although the effect of lapped
splices was not considered and perfect contact was assumed between concrete and steel
elements, the fibre model was successful in simulating the global behaviour of the short
pier.
Damage was observed at the lower part of the pier only. The numerical model indicated
yielding of steel within the first 1.5 m from the base cross-section. This is contrary to the
experimental observations that showed evidence of yielding at a limited zone just above
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
165
the foundation and at the end of the lapped splices. It is reminded that continuous
reinforcement and perfect bond between steel rebars and surrounding concrete was
assumed in the numerical analysis. This, combined with the assumption of elastic
behaviour of the base block (yield penetration in the foundation was not allowed in the
numerical model), may explain the aforementioned difference.
Figure 5.7 compares the experimental and numerical moment-curvature diagrams. The
experimental curvature was calculated from the instrumentation readings, while the
moment was obtained by dividing the measured force by the pier height. The numerical
moment-curvature curve was extracted from the results of the analysis. Also in this case
good agreement is observed, although the numerical model results in smaller dissipation
capacity and exhibits more pinched response, compared to the experimental behaviour.
The cumulative dissipated energy versus the lateral displacement is plotted in Figure 5.8.
The dissipated energy is equal to the area within the force-displacement loops, as results
from the numerical analysis and the experimental data. It is seen that for the cycles with
small amplitude (0.027 m), the numerical model results in smaller dissipation capacity,
compared to the experimental data. This shows the inability of the numerical model to
fully capture the damage (small horizontal cracks in the flanges within the lower part of
the pier) induced in the pier during the small-amplitude cycles. For the cycles with higher
amplitude the numerical model slightly overestimates the dissipated energy. The total
energy dissipated by the numerical model is only 6.4% higher than the experimental
value.
From the above considerations it is concluded the FTB formulation accurately reproduces
the experimental behaviour of the short pier in terms of resistance, damage location and
hysteretic response. It is reminded that certain simplifications were introduced, namely
elastic behaviour of the base, continuous reinforcement and no steel-concrete slip. It is
important to note that, as regards the tested specimen, predominant flexural response was
observed.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
166
-1500
-1000
-500
0
500
1000
1500
-0.10 -0.05 0.00 0.05 0.10
Displacement (m)
For
ce (k
N)
experimental
numerical
Figure 5.6. Short pier A70: experimental and numerical force-displacement curves
-10000
-8000
-6000
-4000
-2000
0
2000
4000
6000
8000
10000
-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05
Curvature (1/rad)
Mom
ent (
kNm
)
experimental
numerical
Figure 5.7. Short pier A70: experimental and numerical moment-curvature curves
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
167
0
50
100
150
200
250
-0.10 -0.05 0.00 0.05 0.10
Displacement (m)
Dis
sipa
ted
ener
gy (k
Nm
)
experimental
numerical
Figure 5.8. Short pier A70: experimental and numerical dissipated energy versus top
displacement
Piers with combined flexural and shear behaviour
In this section the results of a FTB model are compared to the experimental data from a
cyclic test on a scaled specimen of a bridge pier [Peloso, 2003]. The objective of this
simulation is to highlight the limitations of the classic FTB modelling when studying
elements with significant shear deformation and to validate the model for further studies
of a retrofitted pier (see section 5.4.3).
The scaled specimen had a rectangular hollow cross-section with external dimensions
0.45x0.45 m and internal dimensions 0.35x0.35 m. A foundation block 1.20x1.20 in plan
and 0.60 m high was used to anchor the specimen on the laboratory floor. The horizontal
load was applied at 1.35 m from the base cross-section of the specimen, corresponding to
aspect ratio L/d = 3. The longitudinal reinforcement consisted of 40 Φ 8 bars uniformly
distributed along the internal and external faces of the pier, corresponding to
reinforcement ratio sρ = 0.025. The horizontal reinforcement consisted of one rectangular
stirrup for each wall of the specimen, vertically spaced at 0.075 m. The stirrups were
made of Φ 3 smooth bars. The compressive strength of concrete was cf = 30.3 MPa. The
yield stress of steel was yf = 550 MPa, while the ultimate stress was uf = 660 MPa.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
168
Cyclic displacement with increasing amplitude was applied at the top of the pier in the
presence of constant axial load equal to 250 kN ( ν = 0.1). During the first cycles,
horizontal cracks appeared near the base and then within 1.0 m from the foundation. With
increasing displacement, diagonal cracks appeared and the existing ones opened further.
Finally, at drift δ = 3.6 % a sudden reduction of strength, associated with crushing of
concrete, was observed. The specimen was able to sustain further cycling with limited
resistance. The experimental force-displacement curve is shown in Figure 5.9. At the end
of the test, large diagonal cracks and disintegration of concrete at the webs and crushing
of concrete at the base of the flanges were observed. This points to a combined flexural-
shear response.
A FTB model of the tested pier was built. The constitutive laws for concrete and steel
presented in section 5.2.2 were adopted with properties as for the tested specimen.
Because of the configuration of the stirrups, no confinement effect was considered and
the residual strength of concrete was set to zero. At the cross-section level, rectangular
elements with four integration points each were used for the concrete fibres and point
elements were used for the steel fibres. In this way, a detailed discretisation of the section
was obtained and the exact position of all rebars was reproduced. For the beam element, 6
elements 0.15m-high were used for the lower part, where damage was observed during
the tests, and 2 elements were used for the remaining part. The foundation block was
considered to be fully fixed at the base and elastic behaviour was assumed.
The experimental and numerical force-displacement curves are compared in Figure 5.9.
The agreement is not as good as for the previous case. The numerical model reproduces
well the experimental behaviour for the cycles of small amplitude, but not for the cycles
of large amplitude. This can be attributed to the contribution of shear phenomena that are
not appropriately modelled in the numerical simulation. In fact, during the large cycles,
diagonal shear cracks developed in the tested specimen and are responsible for the
pinching observed in the experimental force-displacement curves for these cycles. The
numerical model accounts only for elastic shear and therefore does not succeed in
predicting the cyclic behaviour of the specimen at large deformations.
The experimental and numerical cumulative dissipated energy is plotted in Figure 5.10
against the top displacement. It is seen that for the cycles with small displacement, the
numerical model results in lower capacity of energy dissipation. This shows that it does
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
169
not capture the small damage observed in these cycles and some tuning of the material
parameters is needed to obtain better agreement. On the contrary, for the cycles with large
deformation, the numerical model results in larger dissipation capacity. This is a
consequence of the problems related to shear phenomena, as explained before.
-250
-200
-150
-100
-50
0
50
100
150
200
250
-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05
Displacement (m)
For
ce (k
N)
ExperimentalNumerical
Figure 5.9. Tall pier T250: experimental and numerical force-displacement curves
0
5
10
15
20
25
30
-0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04
Displacement (m)
Dis
sipa
ted
ener
gy (k
Nm
)
experimentalnumerical
Figure 5.10. Tall pier T250: experimental and numerical dissipated energy versus top
displacement
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
170
In conclusion, it is shown that the adopted fibre/Timoshenko beam model is fully valid
for the case of piers with pure flexural behaviour, while presents several limitations in the
case that shear phenomena have a significant contribution in the structural response.
Piers with tension shift
In this section the experimental data from the cyclic test on the tall bridge pier, A40, are
compared to the results of numerical analyses using the FTB modelling. The original
formulation of the FTB model for the tall pier resulted in higher resistance and ultimate
displacement, as well as different height-wise distribution of curvature, compared to the
experimental results. The experimental and numerical distributions of curvature at
maximum displacement are compared in Figure 5.11. This model predicted failure of the
pier above the cross-section at 1.20 m from the base, where the first reduction of
longitudinal reinforcement takes place. It is reminded that the reinforcement at the base
cross-section of the scaled specimen consists of 80 Φ 16 rebars, it is reduced to almost
half (76 Φ 12) at the height of 1.20 m and then to almost one quarter (40 Φ 12) at the
height of 3.5 m. Looking at bending capacity alone, the critical cross-section is identified
at 2.5 m from the base, as predicted by the original numerical model. Due to shear
cracking and to the tension shift phenomenon, as discussed in Chapter 3, the critical
cross-section was shifted above the bar cut-off at 3.5 m from the base.
0
2
4
6
8
10
12
14
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03
Curvature (rad/m)
Hei
ght (
m)
numericalexperimental
Figure 5.11. Tall pier A40: distribution of average curvature along the height of the pier
for the original model
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
171
As only elastic shear deformation is considered in the FTB model, this phenomenon could
not be captured. The same failure mode and overall behaviour was expected also for the
piers A30, A50 and A60. These piers were part of the numerical substructure of the PSD
tests and therefore an adequate numerical model that predicts correct failure mode,
resistance and hysteretic behaviour was needed in order to perform the PSD tests.
In order to obtain the correct failure location, and then the correct resistance and
deformation capacity, alternative configurations were considered. So as to artificially
increase the tension in the rebars above the cut-off, an alternative model with an external
moment, proportional to the applied force, applied at the critical cross-section was
studied; this model yielded results close to the experimental ones. Since for the PSD tests
the input at each time step would be the top displacement, while the restoring force would
be measured, or calculated, this configuration was complicated. An alternative solution
was to reduce the area of reinforcement in the flange above the critical cross-section. The
force-displacement monotonic curves for the two alternative solutions are compared to
the experimental envelope curve for the tall physical pier, A40, in Figure 5.12. Both
models yielded a good approximation of the experimental envelope. Because of the
complications related to the PSD testing procedures, it was decided to proceed along the
line of modifying the area of steel in the model, rather than apply an external moment.
0
200
400
600
800
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Displacement (m)
For
ce (k
N)
experimental
external moment
reduced steel area
Figure 5.12. Tall pier A40: force-displacement curves for alternative models
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
172
Alternative modifications were also considered. For the first alternative, elastic behaviour
was adopted for the part of the pier below the critical cross-section and thus, the correct
failure location was obtained. Looking at the cyclic behaviour, see Figure 5.13a, the post-
crack stiffness is not sufficiently reproduced. This is because cracking in the tested
specimen first occurred at the lower part, then extended along the height and only at
relatively large levels of top displacement damage concentrated above the bar cut-off,
while damage below the critical cross-section of the numerical model was not allowed.
For the same reasons, the energy-dissipation capacity is not correctly represented, see
Figure 5.13b. During the PSD tests, the post-crack behaviour of the numerical piers was
significant, because they were expected to respond mainly in this region for all three
earthquake amplitudes. Then, this configuration was not satisfactory.
Alternatively, the steel area below the critical cross-section was artificially increased.
This forced damage to concentrate above the critical cross-section and the objective of
getting the correct failure location was met. Nevertheless, the cyclic response, Figure
5.14a, indicates that also this model does not accurately follow the post-crack behaviour.
Looking at the dissipation capacity, it is seen that this model shows smaller capacity,
compared to the tested pier, see Figure 5.14b. This could be due to the fact that in the
lower part of the pier, the larger amount of steel results in smaller crack widths and
smaller strains in the rebars and consequently in smaller inelastic deformation and energy
dissipation. For the above reasons, also this solution was judged unsatisfactory.
-800
-400
0
400
800
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
ExperimentalNumerical
(a)
0
100
200
300
400
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
ExperimentalNumerical
(b)
Figure 5.13. Numerical model with elastic base for the tall pier A40: force-displacement
curves (a) and dissipated energy (b)
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
173
-800
-400
0
400
800
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
ExperimentalNumerical
(a)
0
100
200
300
400
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
Experimental
Numerical
(b)
Figure 5.14. Numerical model with increased steel at the base for the tall pier A40: force-
displacement curves (a) and dissipated energy (b)
-800
-400
0
400
800
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Forc
e (k
N)
ExperimentalNumerical
(a)
0
100
200
300
400
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)ExperimentalNumerical
(b)
0
2
4
6
8
10
12
14
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03
Curvature (rad/m)
Hei
ght (
m)
numericalexperimental
(c)
Figure 5.15. Numerical model with reduced steel at the critical cross-section for the tall
pier A40: force-displacement curves (a), dissipated energy (b) and distribution of
curvature along the height (c)
The last examined alternative was to reduce the amount of steel above the critical cross-
section. This solution enabled to approximate the damage observed within the lower part
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
174
of the pier during the cycles of small amplitude and also to approximate the post-crack
stiffness. The reduction was such that the correct resistance is obtained, as seen in Figure
5.15a. The dissipation capacity was fairly well estimated for the small-amplitude cycles.
Worse agreement is observed, though, for the large-amplitude cycles, see Figure 5.15b.
This can be attributed to the smaller amount of steel, compared to the actual amount
present in the physical model. The numerical model predicts the correct failure location,
as expected, but does not perfectly account for the damage in the lower part of the pier,
see Figure 5.15c. Despite the shortcomings, among the available solutions, this last one
was the one that better satisfied the requirements of correct failure location, correct
resistance and best possible approximation of cyclic behaviour and dissipated capacity.
Therefore, it was decided to proceed with this modification for the numerical models of
the substructured piers.
From the examples presented previously the FTB model can be validated for elements
with different behaviour modes, namely flexure-dominated, combined flexural and shear
and flexural response with tension shift. For the first case, the numerical model was able
to reproduce the experimental behaviour, even though continuous reinforcement and
perfect bond between concrete and steel was considered. For the second case, some
differences appeared in the cyclic behaviour because of the inability of the model to
account for inelastic shear damage. It is reminded that this was observed only at large
displacements. For the last case, the numerical model was able to predict the correct
failure location and resistance and to approximate the cyclic behaviour only after certain
modifications were introduced.
Numerical models for the Talübergang Warth Bridge PSD tests
In this section the numerical models for the substructured piers A20, A30, A50 and A60
are presented. It was decided to use fibre/Timoshenko beam elements with the cross-
section and material laws presented previously, appropriately modified to account for the
relocation of the critical cross-section due to the tension shift phenomenon. With this
scope, the results of refined FEM analysis using a damage model [Faria et al., 2001;
Pouca, 2001] were used as reference. It is reminded that this modification was necessary
for the numerical pier A30, A50 and A60. For these piers the area of reinforcement in the
flange above the bar cut-off was artificially reduced in order to obtain the correct
resistance. The “correct resistance” was taken from the results of the damage model
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
175
analyses. Table 5.3 shows the amount of longitudinal reinforcement for the scaled piers as
well as the values assumed in the numerical simulation to achieve the “correct
resistance”.
Figure 5.16 presents the force-displacement curves for all the piers and the two alternative
modelling approaches. For the short physical pier A70 also the experimental curve is
shown. Figure 5.17 presents the cumulative dissipated energy against the lateral
displacement for the two alternative models. It is seen that the difference between the
results lays mainly in the cyclic hysteretic behaviour. In Figure 5.16 it is observed that all
piers exhibit a pinched response for the damage model and much less for the fibre model.
The latter is in better agreement with the experimental evidence.
The difference in hysteretic response results in lower dissipation capacity for the damage
model with comparison to the fibre model, as seen in Figure 5.17 and Table 5.4. For the
physical piers A40 and A70 the numerical results can be compared to the experimental
results presented in Chapter 3. It is seen that the fibre model results in slightly higher total
dissipated energy than the experimental value, by 2.4% and 3.5% for the tall and short
pier, respectively. On the other hand, the damage model underestimates the experimental
total dissipated energy by 17% and 7.7% for the tall and short pier, respectively. In this
respect, the fibre model seems advantageous, compared to the damage model.
Table 5.3. Longitudinal reinforcement steel ratio for the scaled bridge piers
A-A B-B C-C
Height (m) sρ Height (m) sρ Height (m) sρ
A20 2.76 0.9 7.04 0.6 2.12 0.3
A30 3.52 0.7 7.08 0.4/0.3a 4.96 0.2
A40 1.20 1.0 2.30 0.7 9.75 0.4/03a
A50 1.60 0.7 7.24 0.4/0.3a 5.56 0.3
A60 1.48 0.8 2.68 0.5/0.4a 7.84 0.3
A70 6.50 0.4 a original/modified
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
176
Pier A20
-1500
-1000
-500
0
500
1000
1500
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Forc
e (k
N)
damage model
fibre model
Pier A50
-800
-400
0
400
800
-0.2 -0.1 0.0 0.1 0.2Displacement (m)
Forc
e (k
N)
damage modelfibre model
Pier A30
-800
-400
0
400
800
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Forc
e (k
N)
damage model
fibre model
Pier A60
-800
-400
0
400
800
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15Displacement (m)
Forc
e (k
N)
damage modelfibre model
Pier A40
-800
-400
0
400
800
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Forc
e (k
N)
damage model
fibre model
Pier A70
-1500
-1000
-500
0
500
1000
1500
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15Displacement (m)
Forc
e (k
N)
damage modelfibre modelexperimental
Figure 5.16. Talübergang Warth Bridge piers: force-displacement curves for the fibre
model and the damage model
Table 5.4. Values of dissipated energy (kNm) for the bridge piers
A20 A30 A40 A50 A60 A70
Damage model 27.7 126 227 97 79 296
Fibre model 34.4 223 280 162 145 332
Experimental - - 273.5 - - 320.8
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
177
0
20
40
60
80
100
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
fibre modeldamage model
Pier A20
0
50
100
150
200
250
-0.20 -0.10 0.00 0.10 0.20Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
fibre modeldamage model
Pier A50
0
50
100
150
200
250
300
350
-0.30 -0.20 -0.10 0.00 0.10 0.20 0.30Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
fibre modeldamage model
Pier A30
0
50
100
150
200
250
-0.20 -0.10 0.00 0.10 0.20Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
fibre modeldamage model
Pier A60
0
50
100
150
200
250
300
350
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
fibre modeldamage model
Pier A40
0
50
100
150
200
250
300
350
-0.10 -0.05 0.00 0.05 0.10Displacement (m)
Dis
sipa
ted
ener
gy (
kNm
)
fibre modeldamage model
Pier A70
Figure 5.17. Talübergang Warth Bridge piers: dissipated energy versus lateral
displacement for the fibre model and the damage model
In conclusion, the modified fibre models are considered to represent the actual behaviour
of the numerical piers with sufficient accuracy, for what concerns the failure location,
resistance, post-cracking stiffness and cyclic behaviour.
Finally, the numerical models developed in this section are used for monotonic analyses
in order to define the displacement ductility capacity of the piers. Ductility capacity, uµ ,
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
178
is defined as the ratio of the ultimate displacement, uu , to the yield displacement, yu . An
elastic-perfectly plastic approximation of the monotonic numerical curve is considered;
both curves are shown in Figure 5.18 for all the piers. The yield displacement is defined
at the intersection of the line from the origin passing from the numerical curve at max0.75F
and the horizontal branch of the bilinear curve. Ultimate displacement is defined at the
point after the attainment of maximum force, maxF , where the resistance is reduced to
max0.8F . Following this definition, the values of displacement and force at yielding and
ultimate are presented in Table 5.5, along with the values of displacement ductility. Table
5.5 presents also the values of displacement at cracking of the first concrete fibre, cu ,
which are identified at the attainment of the tensile concrete strength. These values of
cracking displacement, yield displacement and displacement ductility were used in the
previous chapter to define the damage suffered by the numerical piers during the PSD
tests and to evaluate the performance of the bridge structure, following the simplified
assessment procedures.
5.3. FEM MODELLING
To overcome the limitations of the FTB models, related to shear deformation and slippage
between steel rebars and concrete, the response of the two piers is simulated using the
FEM method. The constitutive laws for concrete and steel-concrete bond are presented
and the results of the numerical simulation are compared to the experimental data. It will
be shown that this modelling approach allows to obtain the correct evolution of damage,
failure location and resistance, at the expense of significant computational demand.
Table 5.5. Characteristic values of force and displacement for the bridge piers
A20 A30 A40 A50 A60 A70
cu (m) 0.006 0.016 0.013 0.011 0.007 0.003
yu (m) 0.065 0.038 0.081 0.044 0.043 0.011
uu (m) 0.187 0.372 0.230 0.326 0.179 0.100
uµ 2.9 9.8 2.8 7.4 4.1 9.1
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
179
Pier A20
0
200
400
600
800
1000
1200
0.00 0.05 0.10 0.15 0.20
Displacement (m)
Forc
e (k
N)
Pier A50
0
200
400
600
800
0.0 0.1 0.2 0.3 0.4
Displacement (m)
Forc
e (k
N)
Pier A30
0
200
400
600
800
0.0 0.1 0.2 0.3 0.4
Displacement (m)
Forc
e (k
N)
Pier A60
0
200
400
600
800
1000
0.00 0.05 0.10 0.15 0.20
Displacement (m)
Forc
e (k
N)
Pier A40
0
200
400
600
800
1000
0.00 0.05 0.10 0.15 0.20
Displacement (m)
Forc
e (k
N)
Pier A70
0
500
1000
1500
0.00 0.02 0.04 0.06
Displacement (m)
Forc
e (k
N)
Figure 5.18. Talübergang Warth Bridge piers: force-displacement curves from pushover
analysis and bilinear envelope
5.3.1. Constitutive laws
Concrete
A constitutive law that accounts for non-linear behaviour in tension and for linear elastic
behaviour in compression is presented in this section. To describe tension cracking, either
discrete or smeared cracks are considered. Discrete cracking allows cracks to form at the
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
180
boundaries of elements and for a more realistic representation, double joints have to be
declared from the beginning, or new elements have to be introduced after cracking. This
approach is best suited for structures where a single crack forms and offers detailed
results at the local level of the crack, at the expense of large memory and computational
demand. Following the smeared crack approach, cracks are assumed to be distributed
over a concrete element. This requires less computational demand and is best suited for
elements where distributed cracking is expected [CEB, 1996]. A smeared cracking model
[Dahlblom & Ottosen, 1990] is described in the following.
Considering an isotropic and homogeneous bar, it is assumed that two regions exist: one
region that exhibits strain softening and another region in elastic unloading. Following a
fictitious crack model [Hillerborg et al., 1976], the post-peak behaviour is described by an
undamaged, elastic, region, which occupies the whole length, L, of the bar, and an
infinitely thin damaged, softening, region, in which additional elongation, w, occurs in
the post-peak region. The fictitious crack model describes the behaviour of the cracked
zone in terms of normal stress, σ , and elongation normal to the crack plane, nw . For
simplicity, the stress-elongation behaviour of the second region, shown in Figure 5.19a, is
described by a straight line
n t
1w ( )
N= σ − σ (5.12)
where N is the slope of the curve and tσ is the uniaxial tensile strength. Then, the bar
elongation in the post-peak region is
Crack opening
Str
ess
Crack opening
Str
ess
(a) (b)
Figure 5.19. Concrete constitutive law: loading (a) and unloading (b) of a crack
0=β 1=β
5.0=β
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
181
t
1u L ( )
E Nσ
= + σ − σ (5.13)
where E is the Young modulus and the slope of the softening branch is obtained by
equating the fracture energy, fG , to the energy dissipated by the whole bar
f
2t
G2N
σ= (5.14)
At this point, the material characteristic length, λ , is introduced
2t
f EG2σ
=λ (5.15)
In order to consider the fact that the cracks are subjected to shear, it is assumed that the
tangential crack displacement, tw , occurs within the infinitely thin cracked zone and that
it depends on the shear stress, τ . A simple expression is adopted
τ=s
nt G
ww (5.16)
where sG is a material constant, called slip modulus.
In the general three-dimensional case it is assumed that three cracks may develop. The
crack planes are considered to be perpendicular to each other. The first crack initiates
when the maximum principle stress, 1σ , exceeds the tension strength, tσ , and the first
crack is taken normal to the principal direction of 1σ . A possible second crack develops if
the normal stress in some perpendicular direction exceeds the tensile strength. Finally, a
possible third crack develops if the stress perpendicular to the plane of the first two cracks
exceeds tσ . When a crack has been created, its direction is considered fixed.
For the particular case of plane stress formulation, and without repeating the intermediate
steps given in [Dahlblom & Ottosen, 1990], the constitutive relation in matrix form is
σ−
σσσ
++ν−
ν−+=
εεε
oJJ
K)G/1(000J)E/1(E/0E/J)E/1(
2y
x
t
xy
yy
xx
xy
yP
Px
xy
yy
xx
(5.17)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
182
x
x NL1
J = ; y
y NL1
J = ;
−=
y
yy
x
xx
sxy L
wLw
G1
K (5.18)
In the previous equations G is the shear modulus, xL and yL are respectively the
equivalent lengths related to the cracks in the x and y directions, while xxw and yyw are
the crack openings.
Introducing the aspect of mesh sensitivity, the equivalent length is in general the
maximum length of the cracked element region in the direction normal to the crack plane.
For example, for an element with three nodes the region of interest is the whole region of
the element, while for an element with eight nodes, the region of interest is the tributary
region of the Gauss point where cracking occurs. For general loading and with the
condition of keeping the slope of the stress-strain curve negative in the post-peak region,
the following restriction to the equivalent length applies
P
x 1L
ν+λ
< (5.19)
For normal concrete λ varies between 0.4 m and 0.8 m. The issue of mesh sensitivity is of
practical interest when elements are very lightly reinforced, while results will vary
insignificantly with mesh size for moderately and heavily reinforced elements [CEB,
1996].
For unloading-closing of a crack a linear relation between normal stress, xxσ , and crack
width, xxw , is adopted, as shown in Figure 5.19b. When the normal stress arrives at zero
value, a fraction β of the developed crack width is considered to remain open. The value
β = 0.2 provides results that are in good agreement with experimental data. If
compressive normal stress develops, no change of crack width occurs. Reloading follows
the unloading path until the previously obtained crack width, max,xxw , is regained.
Steel-to-concrete interface
When the global response of members is of interest, perfect contact is often considered
between steel and concrete elements and the interaction between the two materials
(namely slip and dowel action) is accounted for by appropriate modification of the
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
183
parameters of the constitutive relations of steel and/or concrete. It is however possible to
introduce contact elements between the nodes of the two materials. The simplest type is
the dimensionless link element, which connects the corresponding nodes of concrete and
steel. These elements are able to model the slip in the direction of the rebar by one spring
and the dowel effect by another spring in the direction normal to the rebar. Continuous
contact elements, which connect the corresponding faces of steel and concrete elements,
are also available. With these elements it is possible to model the effect of lateral pressure
on the behaviour of the interface [CEB, 1996].
Steel-to-concrete bond has been of interest for many years and extensive research has
been performed in this field. Among the aspects investigated with experimental testing
was the bond deterioration with repeated loads [Bresler & Bertero, 1968], crack pattern
[Goto, 1971], effect of loading rate [Chung & Shah, 1989], role of stirrups [Giuriani et
al., 1991; Soroushian et al., 1991] and fatigue [Balázs, 1991]. An early shear-slip relation
has been formulated [Tassios, 1979] and then several constitutive laws have been
proposed on these grounds [Eligehausen et al., 1983; Filippou, 1986; Hawkins et al.,
1987; Alsiwat & Saatcioglu, 1992].
The stress-slip constitutive model used in this work [Eligehausen & Balázs, 1993] is
shown in Figure 5.20. The model consists of an initial non-linear relationship
max 1(s / s )ατ = τ ; 10 s s≤ ≤ (5.20)
that corresponds to the mechanical interaction between the bar lugs and the surrounding
concrete. It is followed by a plateau
maxτ = τ ; 1 2s s s≤ ≤ (5.21)
that describes the advanced micro-cracking and the initiation of shearing of the concrete
between the lugs. Then comes a linearly decreasing branch
max max f 2 3 2( )(s s ) /(s s )τ = τ − τ − τ − − ; 2 3s s s≤ ≤ (5.22)
that refers to the reduction of bond resistance due to partly sheared off concrete between
the lugs. The final branch is a constant line
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
184
Slip
Bon
d st
ress
Figure 5.20. Steel-to-concrete bond constitutive law: Eligehausen- Balázs model
fτ = τ ; 3s s≥ (5.23)
that represents the residual bond capacity after the concrete between the lugs is
completely sheared off. The unloading branch is linear and valid for all parts of the
diagram. The same relationship is assumed regardless of whether the bar is pushed or
pulled. Values of 1s , 2s and 3s , as well as maxτ and fτ as a function of the concrete
compressive strength, have been proposed for confined and unconfined concrete, good
and bad bond conditions [Eligehausen & Balázs, 1993].
5.3.2. Validation of the numerical model
Piers with tension shift
To overcome the inadequacy of the fibre/beam model to simulate the experimental
behaviour of the tall pier, a 2D finite element model in plane stress formulation was
implemented. The pier shaft was modelled using quadrangular elements with four
integration points. Equivalent thickness was considered for the web and flanges. The
horizontal and vertical steel bars were modelled using linear elements with two
integration points.
Non-linear behaviour, based on the model described previously, was assumed for the
concrete elements of the pier, while the foundation block was considered elastic. Because
the model follows elastic behaviour in compression, it was necessary to reduce by 30 %
3s 2s 1s
maxτ
fτ
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
185
the nominal values of the elastic properties of concrete, so as to obtain the correct
stiffness and strength. In order to reduce the computation time, the part of the pier above
the height of 5.0 m, where damage was not observed during the quasi-static cyclic tests,
was considered to remain elastic. Elastoplastic behaviour with hardening was assumed for
the longitudinal steel elements. The horizontal steel elements were considered to have
linear elastic behaviour. The material properties are given in Table 5.6, where hE is the
hardening stiffness.
Figure 5.21 compares the evolution of damage resulting from the analysis to the
experimental observations. Figure 5.21a shows snapshots of the distribution of concrete
inelastic deformation at displacement that corresponds to the amplitude of the cycles
during the quasi-static tests. It is seen that damage is restricted to the lower part of the
pier shaft for small amplitude of the lateral displacement. Then, extended damage is
observed in that region and concentration of inelastic deformation at the critical cross-
section above the bar cut-off.
The crack pattern resulting from the numerical analysis shows distinct horizontal cracks
within the first 1.0 m from the base for top displacement 25 mm. The cracks extend
through approximately half the length of the web. For a lateral displacement of 30 mm,
corresponding to the first large cycle of the test, more horizontal cracks are observed at
the same zone and another appears at the critical cross-section at 3.5 m. With increasing
amplitude of displacement, namely 70 mm, cracking extends through three quarters of the
web length. Diagonal cracks appear above the height of 2.5 m. For top displacement
equal to 140 mm, cracking extends through the whole length of the web and new
horizontal cracks appear above the cross-section at 4.0 m. Finally, for top displacement
250 mm, diagonal cracks appear within the first 3.0 m from the foundation and new
horizontal cracks form above the critical cross-section.
Recalling the experimental observations, for the cycles of 30 mm horizontal cracks
appeared within the first 1.0 m from the base of the pier and extended up to the height of
3.0 m for the cycles of 70 mm. For the cycles of 140 mm a large horizontal crack
appeared at the critical cross-section at 3.5 m and with further cycling, the flexural
cracking above the critical cross-section increased and diagonal cracks appeared in the
lower part. The damage predicted by the numerical analysis is in fair agreement with
these experimental observations.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
186
(b)
u =
250
mm
u =
140
mm
u =
70 m
m
u =
30 m
m
u =
25 m
m
(a)
Figu
re 5
.21.
Evo
lutio
n of
dam
age
for p
ier A
40: n
umer
ical
(a) a
nd e
xper
imen
tal (
b) re
sults
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
187
u =
250
mm
u =
140
mm
u =
70 m
m
u =
30 m
m
u =
25 m
m
Figu
re 5
.22.
Def
orm
ed s
hape
for p
ier A
40, n
umer
ical
resu
lts (d
ispl
acem
ents
mag
nifi
ed x
15)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
188
Table 5.6. Material properties for the tall pier A40
Concrete Steel
cE (GPa) * 33.5 sE (GPa) 200
Pν * 0.2 Pν 0.3
ctf (MPa) 3.5 syf (MPa) 545
fG (MPa) 0.00014 sh E/E 0.003
sG (MPa) 4.2
β 0.2
*nominal values
0
200
400
600
800
1000
0.00 0.05 0.10 0.15 0.20 0.25
Displacement (m)
For
ce (k
N)
experimentalnumerical
Figure 5.23. Tall pier A40: experimental and numerical force-displacement curves
Figure 5.22 shows the deformed shape of the pier, as results from the numerical
simulation. The displacements are magnified to allow a better illustration of the results. It
is seen that until top displacement equal to 70 mm, the deformation is uniformly
distributed within the lower part of the pier shaft and shear seems to have an important
contribution. The behaviour changes for larger amplitude of lateral displacement, after the
concentration of deformation demand at the cross-section above the bar cut-off. While
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
189
shear continues to be significant in the lower part of the pier, the upper part seems to
rotate like a rigid body by the compressed flange of the critical cross-section.
Considering global results, it is verified in Figure 5.23 that the numerical force-
displacement curve provides a good approximation of the experimental envelope. Very
satisfactory fit is observed in the elastic region. The agreement is good also in the post-
cracking and yielding range.
Piers with lapped splices
For the numerical simulation of the cyclic test on the short pier with lapped splices, a
more refined 2D mesh was used. Quadrangular elements with four integration points were
used for the concrete of the foundation block and the pier shaft. In order to introduce the
joint elements, it was necessary to use quadrangular elements also for the longitudinal
steel. Then, it was decided to consider the vertical reinforcement concentrated at the
flanges. At each flange two rebars, with equivalent area, embedded in the foundation
block extended above the foundation until the height of 0.5 m (equal to the overlapping
length). Adjacent to each of them was positioned a steel bar that extends from the top face
of the foundation through the whole height of the pier. Uni-dimensional joint elements
were introduced between these two steel bars, as well as between the steel elements and
concrete both within the foundation block and the pier shaft. The horizontal
reinforcement was modelled with linear elements with two integration points.
Figure 5.24 presents the mesh of the numerical model. In Figure 5.24a the concrete
elements of the foundation block are shown in light blue. The concrete elements of the
flange are shown in blue and the concrete elements of the web are shown in green. Figure
5.24b presents the mesh for the longitudinal steel, where the starter bars are shown in blue
and the rebars of the pier shaft are shown in red. Finally, Figure 5.24c presents the mesh
for the horizontal steel bars. A close-up of the pier flange around the foundation is shown
in Figure 5.25, where the concrete elements are shown in blue and the steel elements are
shown in red. The overlapping length is also indicated.
Elastic behaviour was considered for the concrete elements of the foundation, while the
constitutive law described in 5.3.1 was used for the pier shaft. It is reminded that this
model considers non-linear behaviour only in tension, while it follows elastic behaviour
in compression. Because of this it was necessary to reduce by 50 % the nominal values of
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
190
the elastic properties of concrete so as to obtain the correct stiffness and strength. In order
to reduce the computational demand, elastic-perfectly plastic and elastic behaviour was
assumed for the longitudinal and horizontal steel elements, respectively. It was verified
that the strain in horizontal steel did not exceed the yield limit. The constitutive model
described in 5.3.1 was used for the joint elements. The material properties are listed in
Table 5.7, where sK denotes the elastic stiffness along the joint and nK stands for the
stiffness in the direction normal to the joint, for which a high value was considered.
(a) (b) (c)
Figure 5.24. Mesh of the numerical model: concrete (a), longitudinal steel (b), and
transverse steel (c) elements
Figure 5.25. Close-up at the base of the mesh: concrete and steel elements
Overlapping length
Starter bars
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
191
Table 5.7. Material properties for the short pier A70
Concrete Steel Joint
cE (GPa)* 33.5 sE (GPa) 200 sK ( 3m/N ) 710x2.2
Pν * 0.2 Pν 0.3 nK ( 3m/N ) 1010
ctf (MPa) 3.5 syf (MPa) 550 maxτ (MPa) 3.35
fG (MPa) 0.00014 fτ (MPa) 0.5
sG (MPa) 6.0 1s (mm) 0.6
β 0.2 2s (mm) 0.6
3s (mm) 1.0 *nominal values α 0.4
0
500
1000
1500
0.00 0.02 0.04 0.06 0.08 0.10
Displacement (m)
For
ce (k
N)
experimental
numerical
Figure 5.26. Short pier A70: experimental and numerical force-displacement curves
As seen in Figure 5.26, the numerical force-displacement curve is in good agreement with
the experimental envelope curve. After first cracking, the numerical curve shows a stiffer
response, compared to the experimental envelop. This is because of the concentration of
steel at the flanges and also because of the modification of the elastic properties of
concrete that results in higher strain at tensile cracking. Looking at the numerical results
at local level, it is observed that stresses are first concentrated on the starter bars around
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
192
the base cross-section and do not exceed the yield limit. For displacement equal to 6 mm,
the stress concentration is shifted to the steel elements of the pier shaft, around the cross-
section where the starter bars are terminated. At top displacement equal to 8 mm, the
stress on the outermost steel elements reaches the yield limit. This is reflected on the
global force-displacement curve that shows a stabilisation of resistance after this point. It
is reminded that the analysis was performed for monotonically increasing displacement
and therefore the degradation due to cycling was not included in the numerical curve.
The distribution of damage predicted by the numerical analysis is compared to the
experimental data in Figure 5.27 (displacement at the top is applied from left to right).
Figure 5.27a shows the crack pattern resulting from numerical analysis considering
continuous reinforcement and perfect contact between concrete and steel: distributed
horizontal cracks are predicted in the lower part of the pier until the height of 2.5 m.
Some diagonal cracks pointing to the compressed flange develop. The crack pattern
predicted by the numerical model that considers the lapped splices and joint elements at
the contact surfaces is shown in Figure 5.27b. It comprises two distinct horizontal cracks:
one appears at the base of the pier shaft and the second one just above the cross-section
where the starter bars are terminated. This is in agreement with the experimental damage
pattern, schematically shown in Figure 5.27c. Effectively, the experimental behaviour
was between the two cases considered in the numerical analysis: two large cracks
developed at the base and above the splices and secondary cracks formed in the lower
part of the pier.
The previous results show that is possible to obtain qualitative, at least, agreement with
experimental data in terms of damage evolution, failure mode and global force-
displacement curves. However, obtaining high accuracy at the local level for structural
elements with complicated configuration is not a trivial task. It is imperative to include in
the numerical model all materials and interfaces and to adopt a detailed discretisation in
order to avoid deviations from the experimental results. Furthermore, appropriate
constitutive laws, which consider cyclic non-linear behaviour both in tension and
compression (in the general case in three dimensions), require the calibration of a large
number of material parameters, which are not readily available.
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
193
(a) (b) (c) Figure 5.27. Crack pattern of the short pier A70: numerical analysis for continuous
reinforcement (a), numerical analysis with joint elements (b) and experimental data (c)
5.4. NUMERICAL MODELLING OF PIERS WITH HOLLOW CROSS-
SECTION AND FRP JACKETS
In this section a combination of the numerical tools described previously is employed
with the aim to study the behaviour of bridge piers with hollow cross-section wrapped
with FRP jackets. First, the experimental results concerning a small-scale specimen of a
bridge pier retrofitted with an FRP jacket are briefly presented. In the following, FEM
analysis of a hollow concrete cross-section wrapped with an FRP jacket is performed.
This study allows to investigate on the effect of confinement provided by the jacket.
Effectively, regions of the cross-section are identified, where the confinement effect is
qualitatively and quantitatively different. Finally, an FTB model is studied, in which the
effect of confinement is taken care of by modifying the concrete properties according to
the results of the FEM analysis.
5.4.1. Experimental results
The experimental results of a retrofitted pier, termed T250-FRP, recently tested at the
University of Pavia in Italy [Peloso, 2003] are briefly presented in the following. The
geometry of the specimen was identical to specimen T250, described in 5.2.4. The retrofit
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
194
intervention consisted in applying longitudinal and transversal GFRP strips. Two
longitudinal layers of 0.1 m-wide strips were applied on both faces orthogonal to the
loading direction. The longitudinal strips were anchored by means of a CFRP bar
embedded in the foundation block; additional steel collars were used to resist the large
tensile forces. The transversal strips had the same width and the spacing was 0.2 m from
centre to centre. The thickness of each strip was 0.23 mm. The nominal values, as
provided by the manufacturer, of the Young modulus and the ultimate stress in the fibre
direction were fE = 65000 MPa and ff = 1700 MPa, respectively.
Constant axial load of 250 kN ( ν = 0.1) was applied and cyclic displacement with
increasing amplitude was imposed on the top. At drift δ = 2.4% some problems with the
anchorage system were encountered. Concrete crushing was observed at δ = 3.6%, as for
the as-built pier, T250. Nevertheless, the specimen was able to sustain further cycling and
at δ = 4.8% crushing of concrete and buckling of steel was observed, without significant
loss of resistance. Finally, the pier failed at uδ = 6.0% by rupture of the transversal GFRP
strips at the base. The force-displacement curves of the as-built and retrofitted specimens
are plotted in Figure 5.28. Compared to the as-built specimen, the retrofitted specimen
had higher resistance, thanks to the additional longitudinal reinforcement. More
important, the retrofitted specimen showed larger deformation capacity, stable response
and larger capacity of energy dissipation. This clearly demonstrates the improvement
provided by the retrofit intervention.
5.4.2. FEM modelling of the confinement effect
As will be further discussed in the following chapter, the existing models for FRP-
confined concrete cannot be simply extrapolated to the case of hollow cross-section.
Apart from the inconsistencies concerning the predicted values of strength and ultimate
deformation of FRP-confined concrete, most of the empirical models have been calibrated
on the basis of tests on cylindrical specimens. In addition, recent experimental and
numerical studies suggest that differences exist between circular and rectangular cross-
sections [Tan, 2001; Karabinis & Rousakis, 2003; Monti, 2003]. Furthermore, no
verification of the confinement effect on rectangular hollow cross-sections is available.
For these reasons, it was decided to study the cross-section of pier T250-FRP by means
FEM analysis.
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
195
-300
-200
-100
0
100
200
300
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10
Displacement (m)
For
ce (k
N)
as-builtretrofitted
Figure 5.28. Tall pier T250: force-displacement curves of the as-built and retrofitted
specimens [Peloso, 2003]
A 3D plasticity-based constitutive law, implemented in Cast3m [Millard, 1993], was used
for the concrete elements. It follows a failure function )k,J,J,I(FF 321= that depends on
the three stress invariants, 1I , ,J2 and 3J , and on a hardening parameter, k. The total
strain pe ddd ε+ε=ε is the sum of an elastic, edε , and a plastic, pdε , component. The
elastic part is calculated following Hooke’s law, while the plastic deformation is
described by the flow rule
σ∂
∂λ=ε
Gdd p (5.24)
where )k,J,J,I(GG 321= is the plastic potential function.
To illustrate the ability of the model to account for confinement, stress-strain curves for
unconfined concrete and concrete confined with steel and FRP are plotted in Figure 5.29.
The results refer to a concrete cylinder of 0.15 m diameter under uniaxial compression
( cE = 33.5 GPa, Pν = 0.2, cf = 36 MPa for monotonic loading). Shell elements with
orthotropic behaviour were used for the steel and FRP jackets. Steel was considered to
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
196
follow an elastic-perfectly plastic constitutive law ( sE = 200 GPa, Pν = 0.3, yf = 500
MPa), while linear elastic behaviour was adopted for FRP ( fE = 52 GPa, Pν = 0.2). The
thickness of the FRP jacket was 2 mm, while the thickness of the continuous steel jacket
was 0.1 mm. The curve for unconfined concrete follows a parabola until maximum
strength and then comes a softening branch. The curve for concrete confined with a
continuous steel jacket shows larger resistance and smoother softening stiffness,
compared to plain concrete. Finally, FRP-confined concrete shows a bilinear response
with increasing strength for the second branch. These curves are in agreement with
experimental evidence and commonly adopted laws for concrete.
For the analysis of pier T250-FRP only one corner of the cross-section was modelled.
Cubic elements with eight integration points were used for concrete, while shell elements
with four integration points were used for the FRP strips. The FRP elements followed an
orthotropic linear stress-strain relation. The nominal value of Young modulus, 1fE , was
used in the fibre direction, while in the other direction, 2fE , the properties were estimated
considering only the contribution of the resin. The material properties are given in Table
5.8.
The thickness of the FRP strips was modified in order to take into consideration the effect
of partial wrapping. While continuous jackets exert a constant lateral pressure along the
height of the element, partial wrapping is less efficient as parts of the concrete remain
unconfined. A confinement effectiveness coefficient, ek , is introduced, based on the
assumption that the unconfined zone between two consecutive strips is enclosed by a
parabola with initial slope of 45o [fib, 2001]. The effectiveness coefficient takes the form
Strain
Str
ess
unconfinedsteel-confinedFRP-confined
Figure 5.29. Numerical stress-strain curves for concrete under uniform compression
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
197
Table 5.8. Material properties for concrete and FRP
Concrete FRP
E (GPa) 32.0 1fE (GPa) 65.0
Pν 0.2 2fE (GPa) 5.0
cf (MPa) 30.0 Pν 0.2
tf (MPa) 3.0
cmε (%) 0.2
cuε (%) 0.6
2
e
s 'k 1 1
2D ≈ − ≤
(5.25)
where s ' is the clear spacing between the FRP strips and D is the width of the cross-
section. Considering the values s ' = 0.1 m and D = 0.45 m, an effectiveness coefficient
ek = 0.8 is estimated. In the numerical simulation, the nominal thickness of the transverse
FRP strips is multiplied by this coefficient.
The results of the numerical model are presented in Figure 5.30 in terms of axial stress
and axial strain, where the curve for unconfined concrete is also plotted for comparison.
The axial stress is normalised to the nominal strength of unconfined concrete, cuf . The
global stress is obtained by dividing the axial resistance by the area of the cross-section.
The effect of wrapping is to increase the maximum strength, by no more than 10%, and
the corresponding deformation. The increase is expected to be more significant for larger
amount of FRP, as will be discussed in the following chapter.
The most important improvement concerns the post-peak behaviour. The proposed
empirical models for FRP-confined concrete contain a “hardening” branch after
attainment of the unconfined concrete strength. On the contrary, the numerical results
indicate that for the examined case a softening branch, with smoother slope compared to
unconfined concrete, follows the point of maximum compressive stress, until a residual
strength. While plain concrete is considered to have nil residual strength, the values for
FRP-confined concrete range from 0.3 cuf to 0.8 cuf in the different zones of the cross-
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
198
section. Confinement results also in increase of the concrete pseudo-ductility, as seen by
the deformation capacity of FRP-confined concrete, compared to unconfined concrete.
Figure 5.31 presents the distribution of maximum axial stress within the cross-section. It
is clearly seen that the effect of confinement is restricted to the corner of the cross-
section. Apart from a stress concentration at the external part of the corner, consistent
with experimental observations, the cross-section may be divided into three regions, as
shown in Figure 5.30a. One comprises the flange, termed Zone 1, where the maximum
strength is not increased, but the residual strength is r,1 cuf 0.2f= . The second part, termed
Zone 2, corresponds to the external part of the corner, where the maximum strength is
increased by 10% and the residual strength is r,2 cuf 0.75f= . In Zone 2 an ascending
branch initiates at strain levels of about 0.7%. This is reminiscent of the “hardening”
branch of the empirical constitutive laws for FRP-confined concrete. In the following this
branch will be ignored and constant residual stress, r,2f , will be considered. One reason is
related to the difficulties in incorporating this behaviour in the concrete constitutive law
used in the fibre modelling. More important, this behaviour is observed at a rather limited
zone and its effects on the global results are not visible, see Figure 5.30. Finally, Zone 3
comprises the remaining part of the corner, where the maximum strength remains
unchanged and the residual strength is r,3 cuf 0.5f= . The stress-strain curves for the three
zones are seen in Figure 5.30. These values hold for the particular geometry and
materials; more general considerations will be developed in the following chapter.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.2 0.4 0.6 0.8 1.0
Axial strain (%)
Nor
mal
ised
axi
al s
tress
unconfinedglobalzone 1zone 2zone 3
(a) (b) Figure 5.30. Tall pier T250-FRP: numerical stress-strain curves for FRP-confined
concrete (a) and definition of zones (b)
1
3
2
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
199
Figure 5.31. Tall pier T250-FRP: distribution of maximum axial stress within the cross-
section
5.4.3. Global behaviour of retrofitted pier
To study the global behaviour of the retrofitted pier, a fibre/beam model, similar to the
one used for the as-built specimen, was studied. The effect of confinement is not directly
taken into consideration. In lieu, the concrete properties are modified according to the
observations reported in the previous section. The concrete cross-section is shown in
Figure 5.32. Blue colour corresponds to zone 1, where only the residual strength is
increased. Green colour corresponds to zone 2, in which both the maximum strength and
residual strength are increased. Finally, zone 3, where an intermediate increase of the
residual strength is observed, is shown in red colour. The values of the concrete properties
used in the model are reported in Table 5.9. The steel rebars were modelled as for the as-
built specimen, see 5.2.4.
The longitudinal FRP strips were modelled using quadrangular elements with equivalent
area, positioned on the two faces of the cross-section. The constitutive law for the
longitudinal strips was considered elastic-perfectly plastic, although the actual behaviour
is elastic until failure. This behaviour was adopted in order to impose a limit,
corresponding to the nominal maximum stress ff = 1700 MPa, on the strength of the
longitudinal strips. Numerical simulation considering elastic behaviour of the longitudinal
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
200
strips showed that the strength constantly increased with displacement, even after many
concrete fibres had collapsed.
Figure 5.32. Tall pier T250-FRP: different zones of concrete in the cross-section mesh
0
100
200
300
0.00 0.02 0.04 0.06 0.08 0.10
Displacement (m)
For
ce (k
N)
As-built
Longitudinal FRP
FRP jacket
concrete crushing steel fracture
Figure 5.33. Tall pier T250-FRP: numerical force-displacement monotonic curves
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
201
Table 5.9. Concrete properties for different zones of the cross-section
cof (MPa) coε (%) rf (MPa) crε (%)
Unconfined 32.0 0.25 0.0 1.25
Zone 1 32.0 0.25 6.4 0.75
Zone 2 35.2 0.35 24.0 0.67
Zone 3 32.0 0.30 16.0 0.72
The results of monotonic numerical analyses are plotted in Figure 5.33. The numerical
force-displacement curve for the as-built specimen follows accurately the experimental
envelop. The loss of strength at lateral displacement 0.03 m is associated with crushing of
concrete, as observed during the test. Figure 5.33 plots also the monotonic curve of a pier
retrofitted only with the longitudinal FRP strips, i.e. without modifying the concrete
properties to account for confinement. Also in this case, crushing of concrete occurs at
lateral displacement around 0.03 m, after which some residual strength is provided by the
FRP strips and the undamaged concrete fibres, until rupture of the longitudinal steel
occurs for lateral displacement around 0.08 m. This shows that the improved performance
of the retrofitted pier cannot be attributed solely to the additional shear and flexural
resistance provided respectively by the horizontal and vertical strips. The numerical
force-displacement curve for the pier where both the longitudinal and transverse FRP
strips are considered, indicate that a small loss of strength occurs when some concrete
fibres (those of zone 1) reach their residual strength. Then, the pier can sustain further
deformation without loosing its load-carrying capacity, until failure occurs due to rupture
of steel rebars.
The complete model was used to simulate the cyclic test on the retrofitted specimen. The
numerical and experimental force-displacement curves are compared in Figure 5.34,
where good agreement is observed. The resistance and degradation predicted by the
numerical model are close to the experimental values. Some differences are observed in
the reloading branches of the cycles of large amplitude. As discussed before, the response
of the tested pier can be attributed to the opening and closing of cracks in the concrete
and also to inelastic shear deformation, both of which are not perfectly considered in the
numerical analysis. These differences result in higher energy dissipation in the numerical
model, as shown in Figure 5.35. The difference is localised on the cycles with large
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
202
displacement, δ = 4.8%, where relatively important shear deformation and big crack
openings are expected.
-300
-200
-100
0
100
200
300
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10
Displacement (m)
For
ce (k
N)
ExperimentalNumerical
Figure 5.34. Tall pier T250-FRP: experimental and numerical force-displacement curves
0
50
100
150
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10
Displacement (m)
For
ce (k
N)
experimentalnumerical
Figure 5.35. Tall pier T250-FRP: experimental and numerical dissipated energy versus
lateral displacement
NUMERICAL MODELLING OF AS-BUILT AND RETROFITTED BRIDGE PIERS
203
5.5. FINAL REMARKS ON THE ASSESSMENT OF MODELLING TOOLS
The fibre\beam model was successful in simulating the response of the short pier, whose
behaviour was flexure-dominated, but failed to capture the tension shift phenomenon, that
dictated the failure mode of the tall pier, and predicted different failure location and
resistance. Nevertheless, for the sake of simplicity and stability of the numerical
processes during the PSD tests, it was possible to obtain the correct failure location and
resistance, as well as similar dissipation capacity, by modifying the steel properties.
The 2D FEM models were able to simulate the physical phenomena that were significant
for the behaviour of the tested specimens and consequently predict the correct damage
evolution, resistance and failure location. This allows to conclude that this numerical
approach can be adopted for the analysis of structural elements similar to those studied in
this work, without necessarily performing experimental tests on large-scale specimens.
However, it was shown that it is not a trivial task to produce results of high accuracy at
the local level of complex elements, such as the ones examined in this study. The large
computational demand, due to the refined meshes used to describe the cross-section and
the distribution of horizontal and vertical reinforcement, as well as the joints (steel-
concrete interface) and also due to the constitutive laws for the materials and contact
surfaces, should be highlighted. Such onerous numerical tools appear useful for refined
studies on retrofit solutions, but would be prohibitive for use within the PSD testing
scheme.
Considering the case of piers retrofitted with FRP strips, the comparison between the
numerical and experimental results serves as a validation procedure for the numerical
tools and their combination. The adopted approach consists of FEM analyses used to
study the confinement effect provided by FRP strips, combined with simplified modelling
of the global behaviour. The numerical simulation yields rational results, which are able
to interpret the experimental behaviour of the tested specimen. The agreement between
the numerical and experimental global results provides confidence in the numerical tools
and procedure. This combined approach will be used in the following chapter for
parametric analyses in order to study the effect of various parameters (namely:
dimensions of the cross-section and of the jacket, amount of longitudinal reinforcement
and axial load) on the effectiveness of this retrofit technique and to identify eventual
limitations.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
204
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
205
6. SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
6.1. INTRODUCTION
It has become clear from the material presented in the previous chapters that existing
bridge piers are vulnerable to earthquakes and therefore appropriate retrofit solutions
should be studied. This need is further accentuated by the significant economic loss
related to collapse, or even serious damage, of a major bridge and by the importance of
bridge structures within complex transportation and communication systems. Particular
attention is devoted in this work to bridge piers with hollow cross-section. Although this
structural type is common in highway bridges across Europe and other seismic-prone
regions, it has been the object of research only recently.
While various alternatives exist for the seismic upgrading of buildings [Fardis, 1998], the
practical solutions for bridges are rather limited. It is either desired to reduce the seismic
demand by modifying the structural response using isolation/dissipation devices, or to
increase the available strength and deformation capacity of the substructure. Considering
the last objective, jacketing has seen numerous practical applications. Concrete jackets
have not been used extensively and in recent years fibre reinforced polymers (FRP) are
becoming much more attractive than steel. Apart from the undisputed advantages of high
strength-to-weight ratio, fast application and corrosion resistance, this preference follows
the general popularity of FRP materials. This trend to indiscriminately use FRP
reinforcement is not justified and only very recently it has been recognised that this
material is not a panacea. This provided the motivation to investigate on the performance
and eventual limitations of the use of FRP reinforcement for the seismic retrofit of hollow
piers with large dimensions.
The present chapter deals with the design of retrofit solutions for seismic-deficient bridge
piers with hollow cross-section using external FRP reinforcement. The first part of the
chapter is dedicated to a literature review on the experimental assessment of bridge piers
with hollow cross-section retrofitted with FRP reinforcement. The experimental results on
one hand provide confidence in the effectiveness of various retrofit techniques,
considering in particular FRP jackets for confinement of hollow piers [Peloso, 2003]. On
the other hand, they highlight the need for rational design rules, as some times FRP
reinforcement is over-designed [Ogata & Osada, 2000]. Then follows a review of the
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
206
available design procedures and equations for retrofit of bridge piers using FRP strips.
This is all examined within the framework of a global retrofit procedure that accounts for
all possible failure modes and for the desired sequence of them. While a sound
background and experimental verification exist for the calculation of flexural and shear
strength [Seible et al., 1995b; fib, 2001], there is scarce confidence with respect to design
for confinement [De Lorenzis, 2001].
The particular problem of confinement for rectangular hollow cross-sections with large
dimensions constitutes the core of the chapter and is addressed following a two-level
numerical analysis approach. The effect on the concrete properties in different parts of the
cross-section is studied with Finite Element Method (FEM) analyses. The empirical
constitutive laws for FRP-confined concrete are found inadequate for the case of
rectangular hollow cross-sections with large dimensions. The results of the previous
analyses are integrated in moment-curvature analyses performed with the aim to study the
effect of jacket geometry, amount of reinforcement and axial load on the ductility
capacity of the cross-section. The values of the previous parameters, above which there is
no further improvement, are identified. Finally, a preliminary design equation is
formulated on the basis of more than 1000 analyses.
6.2. SEISMIC RETROFIT OF REINFORCED CONCRETE BRIDGE PIERS
WITH HOLLOW CROSS-SECTION
6.2.1. Retrofit with steel jackets
The effectiveness of steel jackets for the seismic retrofit of bridge piers with hollow
cross-section has been experimentally investigated in two cases. One experimental
campaign dealt with the problem of premature termination of the longitudinal
reinforcement [Kawashima et al., 1990], while the second dealt with the problem of
limited deformation capacity [Huang et al., 1997].
Four scaled specimens of bridge piers with square hollow cross-section were tested in
order to study the effectiveness of steel jackets [Kawashima et al., 1990]. The specimens
had high aspect ratio, L/d = 5.2, and axial load ν = 0.3. The longitudinal reinforcement,
sρ = 1.3%, was reduced to almost half at 35% of the pier height. Steel jackets were
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
207
applied in that region. The examined parameters were the jacket length, equal to 1.0 or
1.5 times the cross-section width, and the material injected between the jacket and the
pier, namely concrete mortar or epoxy resin.
The retrofitted piers showed stable response until displacement ductility uµ = 7. For the
jackets with length equal to the pier width, damage was observed at the base of the pier
and failure was due to flexure. A small crack was observed at the bar cut-off cross-section
after removing the steel jacket. For the jacket with length 1.5 times the pier width and
epoxy resin, similar performance and damage pattern was observed. For the specimen
with mortar injected between the jacket and the pier, shear cracks developed at the lower
part of the pier and failure was due to a combination of flexure and shear. From this
experimental campaign it was shown that steel jackets were effective in addressing the
problem of rebar curtailment. No major difference was observed for the examined lengths
of the jacket and the injection materials.
Two small-scaled (1:8) specimens of bridge piers with rectangular cross-section
retrofitted with steel jackets were tested [Huang et al., 1997]. The specimens had high
aspect ratio, L/d = 5.9, medium axial load, ν = 0.2, and longitudinal reinforcement
sρ =1.3%. Normal concrete and steel were used. The as-built specimen showed limited
deformation capacity, uµ = 4 and uδ = 1.9%. Steel jackets were applied at the base of the
retrofitted specimens. In the first specimen a gap was left between the jacket and the pier
foundation, while in the second specimen the jacket was anchored in the foundation. Both
retrofitted specimens showed improved stability and larger energy dissipation and
deformation capacities, uµ = 6 and uδ = 2.8%. Nevertheless, the inelastic deformation
demand in the retrofitted specimens was concentrated at the gap between the jacket and
the foundation and failure was because of buckling and rupture of the steel rebars in that
region.
In conclusion, this experimental campaign verified the effectiveness of steel jackets for
the enhancement of the deformation and energy-dissipation capacities of piers with
hollow cross-section. It is noted that the effectiveness was limited, compared to steel-
jacketed piers with full circular cross-section. Finally, the small scale of the specimens
does not allow a reliable verification of the effectiveness of steel jackets for the full-scale
piers, which have much larger dimensions.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
208
6.2.2. Retrofit with FRP jackets
Limited experimental data from piers with hollow cross-sections retrofitted with FRP
jackets are available in literature. A few recent experimental campaigns performed in
Japan, Taiwan and Italy are discussed in detail in the following.
Five small-scale (1:20) models of existing bridge piers with hollow circular cross-section
were tested as-built and retrofitted with FRP strips [Ogata & Osada, 2000]. The internal
diameter of the cross-section increased with height and the amount of longitudinal
reinforcement decreased. The former characteristic resulted in shear failure at the top part
of an as-built specimen with small width: at that region the contribution of concrete to
shear strength was small and the amount of transverse reinforcement was not sufficient to
resist the shear demand. The decrease of longitudinal reinforcement caused yielding of
longitudinal rebars at all cross-sections above cut-off, although failure was due to shear.
The drift capacities for the first and second specimens were uδ = 2.3% and uδ = 3.1%,
respectively. The concrete strength of the second specimen was 1.8 times the strength of
the first.
Three retrofit solutions were experimentally evaluated. The first solution (CF) consisted
of longitudinal FRP strips in order to shift the critical cross-section to the base. A small
amount of transverse FRP strips was also used, mainly to ensure the transfer of stresses
between the longitudinal strips. For the second alternative (CF2), the amount of
longitudinal FRP was reduced to half and the amount of transverse FRP was increased 5
times. The third solution (CF3) comprised only transverse strips. Pier CF presented
flexural damage only above the foundation, while failure was due to shear at drift uδ =
3.0%. Pier CF2 presented stable response with flexural damage at the base, without
collapse, until drift uδ = 5.4%. Finally, pier CF3, without longitudinal FRP, showed stable
response until uδ = 5.3%. Horizontal flexural cracks at all cut-off cross-sections were
observed at the end of the test after the jacket was removed.
The failure mode of the retrofitted specimens calls for a comprehensive and rational
retrofit procedure. External FRP reinforcement is often over-designed: half of the FRP
longitudinal reinforcement used in pier CF was successful in shifting the critical cross-
section at the base of pier CF2. The overall retrofit procedure must also consider the fact
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
209
that upgrading for the most critical failure mode (e.g. flexure at rebar cut-off) might
initiate the second most critical one (e.g. shear). The above observations are particularly
important in cases of real structures that present a variety of seismic deficiencies that
come into light for different levels of loading, as was observed during the experimental
tests described in Chapters 3 and 4.
Scaled specimens of bridge piers with rectangular and circular cross-section were tested
until failure, repaired and strengthened with FRP jackets and tested again [Cheng et al.,
2003]. Different collapse modes were observed, namely: flexural failure (specimen PS1-
C), loss of bond (PI1-C), shear failure (PI2) and combination of flexure and shear (PI2-
C). After testing, the cracks were repaired and FRP jackets, following design equations
that are presented in a following section [Seible et al., 1995b], were applied. All the
retrofitted specimens showed the same failure mode as the as-built ones. Drift capacities
were moderate, ranging from uδ = 2.1% to uδ = 4.5%. Drift capacities were higher than
the corresponding values for the as-built specimens, with the exception of pier PS1-C that
was heavily damaged before retrofit. Note that the comparison of the seismic
performance of the specimens based in displacement ductility is misleading. Indeed, the
reported values of ductility for all the retrofitted specimens are lower than those for the
as-built piers, which would lead to the conclusion that retrofit was detrimental for the
deformation capacity of the specimen.
The experimental data verify the effectiveness of FRP jackets for shear retrofitting of
seismic-deficient bridge piers with hollow cross-sections, either rectangular or circular.
The proposed design equations were successfully applied to change shear-dominated
failure mode to flexure-dominated (specimens PI2 and PI2-C). Due to the fracture and
repair of longitudinal rebars, it is not clear whether, and more important how much, the
jacket was effective in improving the behaviour of lapped splices (specimen PI1-C). Also,
no indication can be drawn on the confinement effect alone: the deformation and energy-
dissipation capacities were improved, but it is believed that this was mainly due to
shifting to a ductile flexural failure mode. Unfortunately, the specimen that failed due to
flexure at the as-build condition was severely damaged and the consequent poor
behaviour after repair does not allow to validate the confinement by the jacket.
Five small-scale specimens of square hollow piers were tested as-built and retrofitted with
FRP strips [Peloso, 2003]. The two short specimens failed in shear and were retrofitted
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
210
with horizontal aramid-FRP (AFRP) strips. For the two tall specimens that failed due to a
combination of flexure and shear, it was decided to apply longitudinal GFRP strips in
order to increase the flexural capacity. Horizontal GFRP strips were also applied in order
to meet the increased shear demands. For the specimen with lapped splices at the base
(overlapping length equal to 20 rebar diameters) it was decided to relocate the critical
cross-section above the lapped splices and for this, longitudinal CFRP strips were applied
near the base. Due to problems with the anchorage system for the longitudinal strips, the
retrofit technique was not successful in relocating the plastic hinge and therefore, the
experimental results for this specimen are not further discussed.
The as-built shear-deficient short piers failed at low values of drift, uδ = 2.4%,
independently of the axial load. For these piers the retrofit was successful in changing the
failure mode to flexure-dominated. This resulted in increased energy-dissipation capacity.
The drift capacities of the retrofitted piers were uδ = 4.8%, which is double the capacity
of the as-built piers.
The as-built tall piers failed due to a combination of flexure and shear at uδ = 2.4%. For
the retrofitted specimens, shear damage was avoided and failure was due to buckling of
longitudinal reinforcement. Compared to the as-built specimens, higher strength and
deformation capacity, uδ = 6.0% or uδ = 3.6% for low and high axial load respectively,
were achieved. A problem concerning anchorage was encountered. The anchorage system
comprised an FRP bar, around which the longitudinal strips were wrapped, embedded in
the foundation and steel angles positioned above the FRP bar and bolted at the foundation
block. This system was not sufficient to resist the forces from the longitudinal strips and
uplifting of the steel angles was observed after a certain level of displacement. In
addition, buckling of longitudinal rebars was observed above the steel angles and not
directly over the base cross-section.
A final remark concerns the effectiveness of FRP strips for the retrofit of piers with
hollow cross-section. Although they do not completely preclude buckling of longitudinal
rebars, they can delay it and consequently increase the deformation and energy-
dissipation capacities. FRP jackets seem actually to have confining action. Therefore, the
improved performance, in terms of strength and deformation, can be attributed partly to
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
211
the longitudinal FRP reinforcement and partly to the confinement offered by the
horizontal strips.
6.2.3. Performance of retrofitted bridge piers
A total of 289 experimental tests on as-built and retrofitted bridge piers are discussed in
this section. Both as-built and retrofitted specimens are examined. The main parameters
of the test campaigns were the cross-section type (full or hollow; circular, oval,
rectangular or wall-type), concrete properties (normal-strength or high-strength), presence
of lapped splices, premature termination of longitudinal reinforcement, amount and
detailing of transverse reinforcement and retrofit technique. The retrofit techniques
comprise mainly jacketing (rectangular, oval or circular; steel or FRP) and application of
external stirrups. The tested piers can be divided in two categories based on seismic
detailing. Seismic-deficient piers have inadequate detailing and are expected to have poor
performance. Retrofitted piers and piers designed according to seismic codes have proper
detailing and are expected to perform in a desired manner. The latter two types of piers
are referred to in the following as code-designed piers. Table 6.1 shows the general
characteristics of the cross-sections and the percentage of each pier type within the
database. The geometric and mechanical properties, as well as the deformation capacities
of all specimens are presented in Appendix A.
Figure 6.1 plots the drift ratio versus different geometrical and mechanical properties of
the piers, namely: aspect ratio, transverse reinforcement ratio, splicing length and
normalised axial load. The first column presents the results for seismic-deficient piers,
while the second presents the results for code-designed piers. Although large scatter is
observed with respect to all examined parameters, general trends may be identified. For
seismic-deficient piers, drift capacity increases with aspect ratio and transverse
reinforcement ratio (squat piers with low amount of horizontal reinforcement exhibit
brittle failure at low levels of drift) and decreases with axial load. For code-designed piers
similar trends are observed. Considering in particular piers with lapped splices, drift
capacity remains in the range uδ = 1% ~ 4%, independently of the overlapping length.
Retrofitting for enhancement of the behaviour of lapped splices can be successful, as
specimens can develop values of lateral drift up to uδ = 8%.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
212
02468
10121416
0 2 4 6 8 10 12
Aspect ratio
Drif
t rat
io (
%)
(a)
02468
10121416
0 5 10 15
Aspect ratio
Drif
t rat
io (
%)
(b)
02468
10121416
0 1 2 3 4 5
Transverse reinforcement ratio (%)
Drif
t rat
io (
%)
(c)
02468
10121416
0.0 1.0 2.0 3.0 4.0 5.0
Transverse reinforcement ratio (%)
Drif
t rat
io (
%)
(d)
02468
10121416
0 10 20 30 40
Splicing length (db)
Drif
t rat
io (
%)
(e)
02468
10121416
0 10 20 30 40
Splicing length (db)
Drif
t rat
io (
%)
(f)
02468
10121416
0.0 0.2 0.4 0.6
Normalised axial load
Drif
t rat
io (
%)
(g)
02468
10121416
0.0 0.2 0.4 0.6
Normalised axial load
Drif
t rat
io (
%)
(h)
Figure 6.1. Drift capacity versus aspect ratio, transverse reinforcement ratio, splicing
length and normalised axial load for seismic-deficient (left) and code-designed (right)
piers
)(Φ )(Φ
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
213
0
2
4
6
8
10
0 2 4 6 8
Aspect ratio
Drif
t rat
io (
retr
ofitt
ed/a
s-bu
ilt)
(a)
0
2
4
6
8
10
0.0 0.2 0.4 0.6 0.8 1.0
Transverse reinforcement ratio (%)
Drif
t rat
io (
retr
ofitt
ed/a
s-bu
ilt)
(b)
0
2
4
6
8
10
0 10 20 30 40
Splicing length (db)
Drif
t rat
io (
retr
ofitt
ed/a
s-bu
ilt)
(c)
0
2
4
6
8
10
0.0 0.1 0.2 0.3 0.4
Normalised axial load
Drif
t rat
io (
retr
ofitt
ed/a
s-bu
ilt)
(d)
Figure 6.2. Increase in drift capacity versus aspect ratio (a), transverse reinforcement ratio
(b), splicing length (c) and normalised axial load (d)
Table 6.1. Bridge pier test database
Cross-section
geometry No %
Cross-section
type No % Detailing No %
Rectangular 157 54 Full 187 65 Seismic-
deficient 77 27
Circular 119 41 Hollow 102 35 Code-
designed 212 73
Oval 10 3
Wall-type 3 1
The drift capacity of the retrofitted specimen divided by the drift capacity of the as-built
specimen can be considered as an index of the effectiveness of the retrofit technique. This
ratio is plotted in Figure 6.2 for all the parameters considered. Although large scatter is
observed, qualitative observations may be made. The effectiveness decreases with
increasing aspect ratio and increasing transverse reinforcement ratio. This is due to the
fact that most retrofit interventions aim at improving the shear performance of existing
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
214
bridge piers, which is more critical for squat piers and/or for piers with low amount of
transverse reinforcement. The effectiveness of the retrofit decreases with the axial load
and with the splicing length of longitudinal rebars. The same trends are observed when
one considers displacement ductility capacity.
6.3. DESIGN OF RETROFIT OF BRIDGE PIERS WITH FRP
6.3.1. Global retrofit procedure
A global retrofit procedure is proposed in this section and the existing design guidelines
and equations are discussed. The term global refers to the consideration of all possible
failure modes. The first step of the retrofit procedure is the assessment of the as-built
structure and the identification of all possible failure modes. Considering RC bridge piers,
the main seismic deficiencies, which have been reported from field observations and
laboratory testing, are
• premature termination of vertical reinforcement
• lapped splices within the potential plastic hinge region
• inadequate confinement (small drift, ductility and dissipation capacities)
• inadequate shear capacity
• small amount of longitudinal reinforcement
Existing bridge piers are usually found to possess adequate flexural strength. However,
they might not be able to develop this nominal strength, because of shear failure or failure
of lapped splices.
Quite often, existing bridge piers present a combination of the aforementioned seismic
deficiencies. Nevertheless, only one of them, the weakest, is the main cause of failure
during an earthquake or a laboratory test, while the contribution of others is difficult to
quantify. It is obvious that if the objective of the retrofit intervention is to provide
resistance against the weakest mechanism only, then most probably failure will be due to
the second weakest mode. For this reason, a global retrofit procedure, that takes into
consideration all possible failure modes, is proposed in Figure 6.3.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
215
Seismic Deficiency Objective of Retrofit Method of Retrofit
• tension shift
• lapped splices in the
plastic hinge region
relocation of critical cross-
section
additional longitudinal
reinforcement
?
• small length of
plastic hinge
• development of full
plastic hinge
• increase concrete
strength and deformation
jacketing
?
• limited shear
capacity • increase shear capacity
additional transverse
reinforcement
Capacity Design
Figure 6.3. Global retrofit procedure for seismic-deficient bridge piers
The procedure outlined in Figure 6.3 presents three retrofit methods, namely additional
longitudinal reinforcement, jacketing and additional transverse reinforcement. Each
method is related to specific seismic deficiencies and retrofit objectives. Apparently,
certain steps can be omitted if the single deficiency does not exist. The arrows indicate
the sequence of the objectives to be sought: the critical cross-section is chosen and
provided with sufficient confinement and then the remaining sections are protected
against shear failure, following Capacity Design philosophy.
Jacketing is not proposed as a means to avoid failure of lapped splices, because it is felt
that, in the case of cross-sections with large dimensions and even more for hollow piers, it
cannot provide the required confinement for enhancement of lapped splices. However,
jackets can provide some degree of confinement that is effective for the increase of
concrete strength and deformation.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
216
6.3.2. Relocation of critical cross-section
Piers with curtailment of vertical reinforcement
Many existing bridge piers present premature termination of vertical reinforcement and
insufficient development length, resulting from elastic design methods and relaxed
detailing rules. This, combined with shear cracking and tension shift, can cause failure
above the bar cut-off, which results in low resistance and deformation capacity. This has
been observed during the tests described in Chapters 3 and 4 and also during previous
experimental campaigns [Kawashima et al., 1990; Calvi et al., 2000; Ogata & Osada,
2000]. Premature termination of vertical rebars was common practice in Japan. Actually,
it is reported that failure of the Hanshin Expressway was triggered by this deficiency
[Kawashima, 2000]. Failure above the cut-off might not conform to the requirements of
EC8-2 [CEN, 2002] for accessibility of the plastic hinge region for inspection and
eventual repair after an earthquake. This problem can be solved relatively easy by
increasing the flexural strength of the cross-section at the cut-off and thus relocating the
critical cross-section at the base. External FRP reinforcement, designed according to the
procedure outlined in a following section, can be applied. The effectiveness of this
method was experimentally verified [Ogata & Osada, 2000].
Piers with lapped splices
In case of bridge piers with lapped splices within the potential plastic hinge zone, two
solutions exist. Either a jacket is applied to provide confinement, or it is decided to
relocate the critical cross-section. Sufficient confinement of large rectangular cross-
sections might require jackets with extremely large dimensions. This would prove
impractical and the alternative of relocation of the critical cross-section should be
examined.
Considering the relocation of the critical cross-section, the solution will depend on the
location and detailing of the longitudinal reinforcement. If all vertical rebars are spliced at
the base, the critical cross-section can be shifted just above the lapped splices. In the case
of a more complex scheme of vertical reinforcement, the choice of the new critical cross-
section might not be straightforward.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
217
Attention must be paid not to force failure to happen at a cross-section far from the base.
This would decrease the shear span of the pier and consequently increase the shear force
it has to resist. Since existing bridge piers are often found to have low shear strength,
increase of the shear demand would further complicate the design of the retrofit and
increase the cost. Decrease of the shear span would also increase the demand on the
foundation. It is felt that the actual solution will result as a compromise between the
above restrictions.
Once the new critical cross-section is chosen, its moment capacity can be estimated from
moment-curvature analysis with assumed, or, if possible, measured, values of the material
properties. Then, the resisting moment of the critical cross-section of the as-built pier
should be increased to the moment that corresponds to failure of the new critical cross-
section, applying an appropriate safety factor to account for material overstrength.
Capacity Design considerations apply for the verification in shear. The design shear force,
SdV , of the retrofitted pier can be estimated based on the resisting moment, RdM , as
Sd CD RdV M / L= γ (6.1)
where L is the height of the pier and CDγ is a Capacity Design safety factor. Values of
CDγ = 1.35 and CDγ = 1.2 are proposed for bridges [CEN, 2002] and buildings [CEN,
2003a], respectively. The recent Italian seismic code [PCM, 2003] provides a relation
between the safety factor and behaviour factor, CD 0.7 0.2qγ = + , which yields CDγ = 1.4
for q = 3.5. The eventual additional shear reinforcement can be dimensioned following
the procedure described in a following section. The effectiveness of this method has been
verified for circular columns retrofitted with concrete jackets [Griezic et al., 1996.] for the
case of square hollow piers retrofitted with FRP strips [Peloso, 2003] failure of the
anchorage conditioned the effectiveness of the retrofit system.
6.3.3. Anchorage
A major concern is the anchorage of FRP strips or rods in order to guarantee safe transfer
of forces from the FRP to concrete. Debonding of FRP is recognised as a possible failure
mode that can condition the design of FRP reinforcement and limit its effectiveness. In
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
218
the following, three different methods of anchorage, namely overlaying, mechanical
anchorage and U-anchors, are presented.
Overlaying
The most common way of anchoring FRP strips is by overlaying. In the fib Bulletin 14 -
Externally Bonded FRP Reinforcement for RC Structures (fib Bulletin) [fib, 2001] three
alternative procedures are proposed for the design of anchorage. The simplest one,
presented in the following, consists in anchorage verification and FRP strain limitation. A
second procedure consists in the calculation of the envelope line of tensile stress, while
for the third procedure the force transfer between FRP and concrete is verified.
According to the first approach, to prevent peeling-off, the ultimate tensile strain εfu
(ranging from 0.005 to 0.015) at Ultimate Limit State (ULS) is restricted to a certain
value. In addition to this, the end anchorage has to be verified using methods mainly
based on fracture mechanics and bond stress-slip relationships. Recent test results have
demonstrated that the FRP tensile strain when peeling-off occurs depends on a broad
range of parameters, such as the properties of FRP and concrete, the loading pattern, the
crack spacing, etc.
The model proposed by [Holzenkämpfer, 1994], as modified by [Neubauer & Rostásy,
1997], is adopted. It gives the maximum FRP force that can be anchored, fa,maxN
fa,max 1 c b f f ctN c k k b E t f= α (6.2)
and the maximum anchorage length, b,maxl , equal to
f fb,max
2 ct
E tl
c f= (6.3)
In the above expressions α is a reduction factor, approximately equal to 0.9, to account
for the influence of inclined cracks on the bond strength, ck is a factor accounting for the
state of compaction of concrete (for FRP bonded to concrete faces with low compaction,
e.g. faces not in contact with the formwork during casting, ck = 0.67, otherwise ck = 1.0)
and bk is a geometry factor
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
219
fb
f
2 b / bk 1.06 1.0
1 b / 400−
= ≥+
(6.4)
with fb / b ≥ 0.33. In the above expressions b is the width of the element, fE , ft and fb
are respectively the Young modulus, thickness and width of the FRP strip and ctf is the
tensile strength of concrete. Note that b, fb and ft are measured in mm, and fE , ctf are
measured in MPa. c1 and c2 in Equations 6.2 and 6.3 may be obtained through calibration
with test results; for CFRP strips they are equal to 0.64 and 2.
For bond length b b,maxl l< , the ultimate bond force can be calculated as [Holzenkämpfer,
1994]
b bfa fa,max
b,max b,max
l lN N 2
l l
= −
(6.5)
Mechanical anchorage
For FRP strips used for increase in flexural strength, it is recommended to improve the
anchorage by mechanical means. One technique is to externally confine the ends of the
FRP strip using additional reinforcement in the orthogonal direction (Figure 6.4). This
type of anchorage is considered effective and avoids plate-end shear failure.
FRP strips used for shear strengthening can be anchored following the scheme shown in
Figure 6.5. It is desirable to anchor the FRP in the compression zone by fully wrapping it
around the member in order to guarantee sufficient anchorage. When this is not feasible,
U-shaped strips can be anchored using bolts (Figure 6.5b). This solution is easier to apply
and satisfies the requirement for anchorage in the compression zone. For the case of
insufficient anchorage in the compression zone, the usable height (inner lever arm) of the
cross-section has to be reduced, so that the member has a fictitious lower ultimate
bending resistance [fib, 2001].
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
220
Figure 6.4. Anchorage of flexural FRP reinforcement with external confinement
reinforcement
Figure 6.5. Anchorage of shear FRP reinforcement in the compression zone by
embedment in concrete (a) and using bolts (b)
With respect to bolted systems, it is not adequate to drill through the strengthening strip
omitting special provisions and merely fixing with a bolt, as drilling holes through
unsupported composites severs the unidirectional fibres. As compressive forces can
further weaken the strip and as it is not possible for the forces in the strip to be
transmitted into the bolt, the end tabs should be designed to take the full force to be
anchored. Bolted systems should be positioned at suitable spacing and anchored in the
concrete to a depth beyond the steel reinforcement. The bolts should be supplied with
large washers and tightened up to a specified torque to prevent crushing of the composite
materials. At holes that are necessary when bolts are applied, interlaminar shear failure or
splitting of the strip may initiate. Moreover, holes reduce the cross-section of the strip. In
general, anchoring devices that may influence the integrity of the strengthening system
are not recommended [fib, 2001].
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
221
A special mechanical anchorage system is shown in Figure 6.6 [Zehetmaier, 2000]. With
this mechanical anchorage a significant increase in anchored tensile force can be
obtained. This system can be applied in case where no wrapping is possible, for local
strengthening and as an anchorage for prestressed strips. A minimum concrete cover of
about 20 mm is required for practical reasons.
Special anchorage heads exist for pre-stressed FRP rods. The systems vary with each
single manufacturer.
U-anchor
The U-anchor [Khalifa et al., 1999] can be used with FRP strips and pre-cured laminates
that are unbonded or fully bonded to concrete. It is based on the idea of embedding a bent
portion of the end of the FRP reinforcement into a groove in the concrete. The U-anchor
is expected to avoid high stress concentration and durability concerns. Three possible
uses of the U-anchor are proposed, namely surface anchorage of strips, after-corner and
before-corner U-anchors.
Anchoring of the FRP reinforcement is attained by grooving the concrete near the end of
the FRP strip, perpendicular to the fibre direction. In the case where the groove is before a
corner or on a plane surface, the edges of the groove should be rounded off (for CFRP a
minimum radius of 10 mm is suggested). This is intended to reduce stress concentration
and prevent premature failure of the FRP strip. The FRP reinforcement is bonded to the
concrete surface and to the walls of the groove. After the saturant has set, the groove is
filled half way with a high-viscosity binder. An FRP bar is then placed into the groove
and is pressed in place. The bar can be held in place using wedges at appropriate spacing.
The groove is then filled with the same paste and the surface is levelled. A cross-section
showing details of an after corner type U-anchor is shown in Figure 6.7. The system can
be applied with or without an FRP bar and can be used to anchor continuous, as well as
discontinuous strips. If the U-anchor is not specifically designed for the intended
application, a recommended FRP bar diameter is 10 mm and groove dimensions can be
taken as 1.5 times the bar diameter.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
222
CFRP-strip
Adhesive
40 mm
L-section length: 100 - 300 mm
Concrete
slits
30 mm
Figure 6.6. Anchorage of CFRP strips [Zehetmaier, 2000]
Figure 6.7. Cross-section of an after-corner U-anchor [Khalifa et al., 1999]
Experimental verification of the method was performed. For a beam strengthened in shear
with CFRP strips without U-anchor, the strength increased by 72% in comparison to the
control beam, but failure was governed by debonding of the CFRP. The maximum strain
in the CFRP strip was 28% of the ultimate value. In the specimen where the U-anchor
was used, the shear capacity increased by 145% and no FRP debonding was observed at
failure. The maximum strain in the CFRP strip was 40% of the ultimate value. The failure
mode was flexure-dominated. This resulted in a ductile failure mode, uµ ≈ 3, compared
to the brittle failure due to shear observed in the control beam and the one reinforced with
CFPR without U-anchor. Based on experimental observations [Peloso, 2003], there seem
to be some limitations on the effectiveness of this technique, when relatively large tensile
forces develop in the FRP strips.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
223
6.3.4. Retrofit for flexural strength enhancement
Reinforced concrete elements can be strengthened in flexure using FRP reinforcement
externally bonded to the tension zones, with the direction of the fibres parallel to that of
tensile stresses. In design for bending, the following assumptions are made: Bernoulli’s
hypothesis is valid, which means that the strain is linearly distributed over the cross-
section and implies complete composite action between the materials; cracked concrete
has no tensile strength; the FRP is linear elastic until failure. Concerning concrete and
steel, the stress-strain curves used in design of RC members apply, e.g. from Eurocode 2
(EC2) [CEN, 2002].
The possible failure modes can be divided in two groups: failure where full composite
action is maintained until concrete crushes or steel yields and failure where composite
action is lost before failure of concrete or steel. In general six primary failure modes are
considered, namely concrete crushing, yielding of reinforcement, laminate failure,
anchorage failure, peeling-off and delamination [Täljsten, 2003]. Yielding of the tensile
reinforcement in the ULS is assumed, while yielding of the reinforcement in the
Serviceability Limit State (SLS) is not permitted. In the following, only the ULS is
considered, as the SLS is not expected to be critical for seismic retrofitting of structures.
The SLS can be critical in case a small amount of FRP is enough to increase the strength
to the desired level the strength for ULS, but might not be enough for deformation control
at SLS.
For the design in bending the actual strain distribution over the member is calculated,
following the previous assumptions. According to the fib Bulletin [fib, 2001], the initial
stress state of the element, usually cracked under service loads, must be considered for the
design of the strengthening. The failure criterion for concrete is c cue e< = 3.5‰ and for
the FRP f fue 0.6e< .
The stress and strain distributions of a RC beam considering full composite action and
failure due to yielding of steel and crushing of concrete, which is the most desirable
failure mode, is shown in Figure 6.8. The neutral axis depth, ? , can be calculated from
equilibrium of internal forces
cd s2 s s2 s1 yd f f f0.85? f b? A E e A f A E e+ = + (6.6)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
224
where ? = 0.8 and
2s2 cu
? de e
?−
= ; s s2 ydE e f≤ (6.7)
f cu o
h ?e e e
?−
= − (6.8)
In the above expressions cdf is the design value of the concrete compressive strength, sE
is the Young modulus of steel, s1A and s2A are respectively the areas of tension and
compression steel, s2e is the strain of the compression steel, oe is the initial deformation
of concrete in the tension zone before the application of the FRP strip, cue and fue are the
ultimate deformations of concrete and FRP, respectively, fA is the area of the FRP, d and
2d are respectively the distance of the tension and compression reinforcement from the
most compressed concrete fibre.
The design bending moment capacity is
Rd s1 yd G f f f G s2 s s2 G 2M A f (d d ?) A E e (h d ?) A E e (d ? d )= − + − + − (6.9)
where Gd = 0.4. The required amount of FRP, fA , can be then calculated from Equations
6.6 to 6.9.
Figure 6.8. Cross-section analysis: (a) geometry, (b) strain distribution and (c) stress
distribution
s1e
c cue e=
cd0.85? f
s1 ydA f
f f fA E e
s2 s s2A E e
oe fe
s2e
d
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
225
For the equations given above to be valid, the following assumptions should be checked:
yielding of tensile steel reinforcement and FRP strain is limited to the ultimate maximum
strain, fude
yds1 cu
s
fdE
− χε = ε ≥
χ (6.10)
f cu o fud
h − χε = ε − ε ≤ ε
χ (6.11)
Other failure modes, due to loss of composite action should be considered. These are
peeling-off caused by shear cracks, peeling-off at the end anchorage and at flexural
cracks, end shear failure and peeling-off caused by the unevenness of the concrete surface
[fib, 2001]. Several parameters, such as the crack opening displacement, flexural and
shear rigidity of the FRP and tensile strength of concrete influence the resistance against
peeling-off. Design for these failure modes requires verification of the anchorage and
might result in additional shear reinforcement.
6.3.5. Retrofit for shear strength enhancement
According to the fib Bulletin [fib, 2001], the external FRP reinforcement may be treated
in analogy to the internal steel (assuming that the FRP carries only normal stresses in the
principal FRP material direction), assuming that at the ULS in shear the FRP develops an
effective strain in the principal material direction, fee . The effective strain, fee , is in
general less than the tensile failure strain, fue . Hence, the design shear resistance, RdV , of
a strengthened element may be calculated according to the EC2 [CEN, 2002] format as
Rd cd wd fdV V V V= + + (6.12)
where cdV is the shear strength of concrete and wdV is the shear strength provided by the
transverse steel reinforcement. The FRP contribution to shear capacity, fdV , can be
written in the following form
fd fd,e f f wV 0.9e E ? b h(cot ? cot a ) sin a= + (6.13)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
226
where fd,ee is the design value of effective FRP strain, wb is the minimum width of the
cross-section over the effective depth, h is the depth of the cross-section, f? is the FRP
reinforcement ratio, ? is the angle of diagonal cracks with respect to the member axis
(usually assumed equal to 45o) and a is the angle between the principal fibre orientation
and the longitudinal member axis. The same equation with ? = 45o is adopted in EC8-3
[CEN, 2003b].
The design value of the effective FRP strain, fd,ee , equals the characteristic value, fk,ee ,
divided by the partial safety factor, f? . Given the lack of sufficient data, fk,ee may be
approximated by multiplying the mean value of the effective FRP strain, f ,ee , by a
reduction factor. Yet, fk,ee should be limited to fue , in order to ensure that the shear
integrity of concrete is maintained sufficiently, so that other mechanisms, such as
aggregate interlock, may be activated. In other words
fk,e f ,e fue ke e= ≤ (6.14)
where k = 0.8 and fue ≈ 0.006.
A more elaborated expression for the reduction factor, k, is available [Duthinh &Starnes,
2001] in the form
f fnE tk 1 0.9
428= − ≤ ; f fnE t ≤ 210 kN/mm (6.15a)
f f
105k 0.9
nE t= ≤ ; f fnE t > 210 kN/mm (6.15b)
where n is the number of plies applied.
The partial safety factor is taken from Table 6.2 if failure involves FRP fracture
(combined with, or following diagonal tension), or f? = fd? = 1.3 if debonding leading to
peeling-off dominates. Application types A and B in Table 6.2 refer to the quality control
of the system and the application procedure. Type A stands for good quality control,
while type B stands for bad quality control.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
227
Table 6.2. FRP material safety factors f? [fib, 2001]
Application type A Application type B
CFRP 1.20 1.35
AFRP 1.25 1.45
GFRP 1.30 1.50
It is recommended to consider a design strain limit of fde = 0.004 for a CFRP jacket
[Seible et al., 1995b]. This value is below the ultimate strain of CFRP, fue = 0.01, but
higher than the yield strain of steel reinforcement. This in turn, ensures that the
contribution of steel reinforcement, wdV , can be fully activated.
EC8-3 [CEN, 2003b] provides design formulae for the estimation of the effective FRP
strain depending on the material. For well-anchored CFRP jackets it writes
0.302 3
cfd,e fu
f f
fe 0.17 e 0.006
E ?
= ≤
(6.16)
For U-shaped CFRP jackets the ultimate strain is
0.56 0.302 3 2 3
3 c cfd,e fu
f f f f
f fe min 0.65 10 ;0.17 e 0.006
E ? E ?−
= × ≤
(6.17)
Finally, for properly anchored AFRP jackets
0.472 3
cfd,e fu
f f
fe 0.048 e 0.006
E ?
= ≤
(6.18)
In the previous equations cf (in MPa) is the estimated value of concrete compressive
strength.
A similar approach is adopted by the Japanese Society of Civil Engineering. The
contribution of FRP to the shear strength is added to that of concrete and steel
reinforcement. The FRP shear strength is estimated as [JSCE, 2001]
fd f f fV KA f z(sin a cosa) / s= + (6.19)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
228
where K is an effectiveness coefficient, ff is the tensile strength, z is the internal lever-
arm, a is the angle of fibres to the longitudinal direction and fs is the spacing of the
strips. Based on experimental results, the effectiveness coefficient is estimated as
0.4 K 1.68 0.67R 0.8≤ = − ≤ (6.20)
where
( )12
331fud4
f f 'f cd
f 1R E
E f
= ρ
; 0.5 R 2.0≤ ≤ (6.21)
The material presented above refers to RC members of rectangular cross-section. If the
cross-section is circular, the contribution of FRP (wrapped around the column) to shear
capacity is controlled by the tensile strength of the FRP jacket, but is limited to a
maximum value corresponding to excessive dilation of the concrete due to aggregate
interlock at inclined cracks [fib, 2001]. By limiting the concrete dilation, that is the radial
strain, to a maximum value, maxe , one may easily show that for inclined cracks forming
an angle ? with the column axis, the FRP contribution to shear capacity is
2
maxfd f f
f
DV E 0.5 cot
4ε π
= ρ ϑγ
(6.22)
where D is the column diameter. The derivation of Equation 6.22 is easily understood if
one assumes that at shear failure all the FRP material crossing an inclined crack is
strained uniformly at fue .
The above discussion concerns retrofit by applying externally bonded FRP strips. In case
an FRP jacket is applied, its contribution to shear strength, fdV , can be calculated for
circular jackets as
fd f fd fV 0.5 E t h cot= π ε ϑ (6.23)
and for rectangular jackets as
fd f fd fV 2E t h cot= ε ϑ (6.24)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
229
where h is the length of the cross-section in the direction of loading and ϑ is the shear
crack inclination (conservatively it can be considered ϑ = 45o).
The Japan Building Disaster Prevention Association Guidelines provides a different
expression for the estimation of shear strength of elements retrofitted with FRP strips
[JBDPA, 1999]
0.23t c
Sd w wy o
0.053 (17.6 f )V 0.845 f 0.1 bj
M / Vd 0.12 ρ +
= + ρ + σ + ∑ (6.25)
where
w wy ws wys wf fdf f f 10ρ = ρ + ρ ≤∑ MPa (6.26)
oσ is the axial stress ( oσ < 7.84 MPa) and j is the internal lever-arm (j = 0.8d for
columns). Equation 6.25 expresses the shear strength as the square root of the summation
of the steel and FRP contribution and is similar to the one used for the design of steel
shear reinforcement for new buildings. The effective ultimate strain of FRP is fue = 0.007.
Experimental studies suggest that elements strengthened with FRP sheets fail in shear
mainly in one of the two modes: tensile rupture of the FRP and debonding of the FRP
from the sides of the RC element. Common methods of strengthening include side
bonding, U-jacketing and wrapping. Available experimental data indicate that almost all
beams strengthened by wrapping failed due to FRP rupture. In contrast, almost all beams
strengthened by side bonding only, and most strengthened by U jacketing, failed due to
FRP debonding. For the case of debonding failure mode, it has been proposed to limit the
strain in the FRP to an effective strain, which is obtained from regression of experimental
data [Triantafillou & Antonopoulos, 2000]. A reduction factor 0.8 is proposed [fib, 2001].
When failure is due to debonding, the contribution of FRP to the shear strength can be
calculated as [Chen & Teng, 2003]
f ffd f f
b f
f h (sin cos )V 2 t w
sβ + β
=γ
(6.27)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
230
where bγ = 1.25 is the partial safety factor, fw , fh and fs are the width, effective height
and spacing of the FRP strips, respectively, and β is the angle of fibre orientation. The
design effective FRP stress, ff, is defined as
f f f ,max,df D= σ (6.28)
The maximum design stress in FRP, f ,max,dσ , can be obtained by using the 95-percentile
characteristic value of the bond strength
ff ,max,d w L c f
f
E0.315 f f
tσ = β β ≤ (6.29)
where wβ and Lβ are coefficients that reflect the effect of FRP-to-concrete width ratio
and the effect of bond length, respectively. They are calculated as
Lβ = 1; λ ≥ 1 (6.30a)
L sin( / 2)β = πλ ; λ < 1 (6.30b)
max eL / Lλ = (6.30c)
f fe
c
E tL
f= (6.30d)
f fw
f f
2 w /(s sin )1 w /(s sin )
− ββ =
+ β (6.30e)
where maxL is the bond length.
In the design of a shear strengthening scheme using U jacketing or side-bonding with
strips, an iterative procedure is required because the coefficient wβ is related to the ratio
of strip width to strip spacing f fw /(s sin )β . An initial value of wβ = 1 may be used. The
iterative process will converge very quickly, with three iterations being usually sufficient
[Chen & Teng, 2003].
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
231
6.3.6. Retrofit for confinement
General
The objective of confinement is to provide lateral support of the longitudinal
reinforcement, to enhance the ultimate compression strain of concrete and consequently
the deformation capacity of the member and to prevent premature spalling of concrete. In
general, confinement by circular hoops or circular jackets would be most beneficial. In
rectangular columns either circular or oval jackets can provide confinement along the
entire column perimeter, while rectangular jackets provide only inward forces at the
corners and therefore significant jacket thickness needs to be provided between corners to
restrain the lateral dilation of concrete and the buckling of vertical rebars. In the presence
of lapped splices, jackets with elliptical shape are more advantageous to provide
continuous lateral pressure along the entire perimeter of the column.
Considering different FRP materials, it seems more appropriate to use GFRP jackets that
also increase deformation capacity, while maintaining the same effectiveness of CFRP
jackets with respect to strength enhancement. CFRP should be used to provide concrete
with higher strength increase and moderate deformation capacity, whereas GFRP should
be used to provide higher deformation capacity and moderate strength increase [Pinto &
Monti, 2000].
FRP-confined concrete
The effect of confinement due to FRP jacketing on the ultimate stress and strain of
concrete is different from the effect due to steel jacketing. The difference originates from
the elastic behaviour of FPR jackets that apply increasing pressure on the concrete with
lateral dilation. On the contrary, a steel jacket applies constant pressure after yielding.
Based on experimental evidence, a number of bilinear constitutive models, as seen in
Figure 6.9, for FRP-confined concrete have been proposed. FRP-confined concrete
displays a distinct bilinear response with a sharp softening and a transition zone at the
level of the unconfined strength, cmf , after which the tangent stiffness stabilises at a
constant value, until reaching the ultimate strength, cuf .
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
232
Strain
Str
ess
Figure 6.9. Bilinear constitutive model for FRP-confined concrete
Following the proposal of [Samaan et al., 1998], the stress-strain relation of concrete is
written as
1 2 cc 2 c1/
1 2 c
o
(E E )f E
(E E )1
f
ηη
− ε= + ε
− ε +
(6.31)
The parameters to define the model (Figure 6.9) are defined as
1 coE 3950 f= (6.32a)
0.22 co f fE 245.61f 1.3456E t / D= + (6.32b)
o co lf 245.61f 0.371f 6.258= + + (6.32c)
In the previous expressions cof is the compressive strength of unconfined concrete and lf
is the pressure provided by the jacket. The first slope of the model, 1E , depends solely on
concrete, while the second slope, 2E , is a function of the stiffness of the confining jacket
and of the strength of unconfined concrete. η is a parameter that controls the curvature of
the transition zone; a constant value η= 1.5 is recommended [Samaan et al., 1998].
1E
2E cof
cuf
cuε
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
233
The strength of confined concrete, cuf , can be estimated as
0.2cu ocu co l
2
f ff f 6f
E−
= + (6.33)
The ultimate concrete strain, cuε , is defined by the geometry of the bilinear curve
cu ocu
2
f fE−
ε = (6.34)
An alternative empirical bilinear stress-strain relationship for FRP-confined concrete,
considering the effect of both steel stirrups and FRP jackets has been proposed
[Kawashima et al., 2000]. For the first branch the proposed law takes the form
n 1
c2c c c
c o
E1f E 1 1
n E
− ε = ε − − ε
(6.35a)
and for the second
c t 2 c of f E ( )= ε − ε (6.35b)
The strain at the intersection of the two branches, oε , is defined as
o f f f co sh yh co0.003 0.00939 E / f 0.0107 f / fε = + ρ ε + ρ (6.36)
The stress at the intersection of the two branches, tf , is calculated as
t co f f f sh yhf f 1.93 E 2.2 f= + ρ ε + ρ (6.37)
and the stiffness of the second branch, 2E , is defined as
2co
2 f ff f f sh yh
fE 0.658 0.078 E
E 0.098 f= − + ρ
ρ ε + ρ (6.38)
where shρ and yhf are the volumetric ratio and yield stress of steel stirrups, respectively.
Finally, η is calculated as
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
234
C 2 co
c co t
(E E )E f
− εη =
ε − (6.39)
The empirical bilinear models have been fitted on experimental results performed mainly
on cylindrical specimens applying a relatively large number of FRP layers. Experimental
results from specimens wrapped with a few layers and/or specimens with rectangular
cross-section, indicate that the stress-strain relation for FRP-confined concrete comprises
a softening post-peak branch with higher ultimate strain and residual strength, compared
to plain concrete. The bilinear empirical models cannot capture this behaviour and the
need for a theoretical approach emerges. According a proposed theoretical model
[Spoelstra & Monti, 1999], the lateral strain of concrete is related to the axial strain and
then an iterative procedure is followed until attainment of a given ultimate deformation of
the FRP. Based on regression analysis, the following simplified formulae have been
proposed for the maximum strength of confined concrete
( )cu co lf f 0.2 3 f= + (6.40)
and for the ultimate strain
( )cu co c fu l2 1.25E fε = ε + ε (6.41)
where coε is the ultimate strain for unconfined concrete, c c coE E / f= is the concrete
elastic modulus and l l cof f / f= is the maximum confining stress, both normalised to the
maximum strength of unconfined concrete, cof .
Exact formulae for the estimation of strength and strain of FRP-confined concrete have
also been proposed. The ultimate strength of confined concrete is
l lcu co
co co
f ff f 2.254 1 7.94 2 1.254
f f
= + − −
(6.42)
and the ultimate stress is
cucu co
co
f1 5 1
f
ε = ε + −
(6.43)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
235
A large number of constitutive models for FRP-confined concrete have been collected
and examined with the objective to systematically assess their performance [De Lorenzis,
2001]. The performance is compared on the basis of the predicted values of the properties
of interest, namely compressive strength and ultimate deformation of concrete and
ultimate strain in the FRP. Figure 6.10 plots the average absolute errors for the ultimate
strength and deformation of concrete. Considering the compressive strength, errors
ranging from 10% to 60% are observed, often on the unconservative side. The predictions
of the ultimate deformation overestimate the experimental data and the errors are higher
(ranging from 20% to 140%) than for the compressive strength.
Experimental data on the ultimate FRP strain were also examined. For FRP-encased
specimens the experimental values were very close to the theoretical values. On the
contrary, for wrapped specimens, the effective ultimate strain was significantly lower
than the theoretical value. The most important parameters that affect this reduction were
identified as the quality of execution, size effects and the radius of curvature. A tentative
equation was proposed and takes the form
1.8271
f f8.7375D E ntR (1.25 0.0063D)e−−= − (6.44)
where R is the ratio of the measured hoop strain in the confining FRP at tensile failure to
the FRP ultimate strain in uniaxial tension, D is the diameter of the specimen (in mm), ft
is the thickness of the FRP (in mm) and n is the number of plies.
The discrepancies of constitutive laws for FRP-confined concrete are reflected on the
global response of elements, as highlighted in a comparative study [Yuan et al., 2001]. A
circular cross-section (diameter 1.5 m and longitudinal reinforcement ratio sρ = 2.3%)
wrapped with ten layers of CFRP ( ff = 3483 MPa) was analysed. Figure 6.11a presents
the moment – axial force interaction diagrams following different constitutive laws for
FRP-confined concrete. All the models show similar trend of moment with axial load, but
a variation of the actual values is observed. Considering moment-curvature analysis of the
same cross-section (for axial load ν = 0.4), the results in Figure 6.11b indicate significant
differences in the predicted values of strength and ultimate curvature. While moment
capacity is influenced mainly by concrete strength, curvature capacity directly depends on
the ultimate deformation of concrete. This explains why larger scatter is observed in the
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
236
predicted values of maximum curvature, than in the values of maximum moment.
Concrete ultimate strain and cross-section curvature are of great importance in seismic
retrofit and hence it is unfortunate that no reliable model is yet available.
The material presented above refers mainly to FRP-wrapped columns with circular cross-
section. In an experimental campaign, the effectiveness of confinement of a rectangular
jacket on a rectangular column was found to decrease with increase in the aspect ratio,
b/h, of the cross-section [Cole & Belarbi, 2001]. This will be discussed with reference to
Figure 6.12, where a rectangular cross-section with dimensions b and d is shown. A
rectangular jacket is applied after rounding-of the corners at a radius r. The hatched area
shows the part of the cross-section that is not effectively confined.
Ave
rage
abs
olut
e er
ror
Ave
rage
abs
olut
e er
ror
Theoretical model Theoretical model (a) (b)
Figure 6.10. Average absolute error: prediction of strength (a) and ultimate deformation
(b) of FRP-confined concrete [De Lorenzis, 2001]
(a)
(b)
Figure 6.11. Moment – axial force interaction diagrams (a) and moment – curvature
monotonic curves (b) for different constitutive laws for FRP-confined concrete [Yuan et
al., 2001]
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
237
Figure 6.12. Effectiveness of confinement for rectangular jacket
Considering a parabolic shape for the unconfined areas, the total unconfined area is
2 2
u
(b 2r) (d 2r)A
3− + −
= (6.45)
A confinement effectiveness coefficient, ek , is introduced [fib, 2001]
2 2
es
(b 2r) (d 2r)k 1
3bd(1 )− + −
= −− ρ
(6.46)
where sρ is the longitudinal reinforcement ratio. The effectiveness coefficient multiplies
the lateral pressure calculated for a fully confined cross-section, i.e. circular cross-section
with circular jacket.
Design based on ultimate curvature
The basis for the design of a FRP jacket for confinement is the volumetric reinforcement
ratio, fρ , required to satisfy a performance criterion, quite often a target value of
curvature ductility. Ultimate curvature is defined at the maximum strain of concrete. A
conservative expression for the concrete ultimate compression strain, cuε , in circular piers
confined with circular jackets is [Priestley et al., 1996]
f fu fucu
cd
2.5 f0.004
fρ ε
ε = + (6.47)
For a circular column retrofitted with a circular jacket of diameter D (Figure 6.13), the
volumetric ratio is
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
238
f f4t / Dρ = (6.48)
From Equations 6.44 and 6.45 the required thickness of the jacket, ft , can be calculated
cu cdf
fu fu
0.1( 0.004)Dft
fε −
=ε
(6.49)
A similar approach is adopted in the technique proposed in the FHWA Seismic Retrofit
Manual for Highway Bridge [Buckle & Friedland, 1995]. A layer of thickness at stressed
to provide an active confinement stress, af , is combined with a layer of thickness pt that
provides passive confinement. Assuming that the ultimate stress of FRP is fuε = 0.004 and
on the condition that lf ≥ 2 MPa, the minimum required jacket thickness is
a a p p at E t E 125D(2 f )+ = − (6.50)
where aE and pE are respectively the elastic moduli of the active and passive layers.
Experimental evidence suggests that, under the previous limitations, piers retrofitted
according to Equation 6.50 will be capable to sustain drift uδ = 4% with an adequate
reserve of displacement capacity. The jacket needs to extend until the cross-section where
the moment decreases to 75% of the maximum moment and not less that half the column
diameter.
For columns with rectangular cross-section, either rectangular or elliptical jackets can be
considered. For the case of rectangular columns with rectangular jackets (Figure 6.14a), a
modification of the aforementioned design equations is proposed in order to consider the
reduced effectiveness of the jacket [Priestley et al., 1996]. According to this proposal,
Equation 6.47 becomes
f fu fucu
cd
1.25 f0.004
fρ ε
ε = + (6.51)
Considering that the volumetric ratio of confinement, fρ , can be written as
f f2t (b h) / bhρ = + (6.52)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
239
where b and h are the dimensions of the cross-section, the required thickness of the
jacket, ft , is calculated as
It is reminded that experimental evidence shows that the circumferential failure mostly
occurs at strains lower than the ultimate strain, fuε , obtained by standard tensile testing of
the FRP strip. This reduction is due to the triaxial state of stress of the wrapping
reinforcement, the quality of the execution, the curved shape of the wrapping
reinforcement and the size effects when applying multiple layers. Proper design values
for the effective ultimate circumferential strain, fuε , justified by experimental evidence,
should be taken into account.
cu cdf
fu fu
( 0.004)f bht 1.6
f (b h)ε −
=ε +
(6.53)
Figure 6.13. Confinement of circular column with circular jacket
Figure 6.14. Confinement of rectangular column with rectangular (a) and oval (b) jacket
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
240
For rectangular columns, elliptical jackets are expected to be more effective, as they
provide more uniform compression stresses. Design guidelines for rectangular jackets
applied on rectangular columns have been proposed [Seible et al., 1995b]. The rules are
limited to cross-sections with aspect ratio h/b < 1.5, otherwise it is recommended to
design a circular or oval jacket. Following these rules, the corners of the cross-section
should be rounded to a radius r > 50 mm and the thickness of the jacket should be twice
the one calculated for a column with the equivalent diameter eD , as defined in the
following. Some limitations for the application of rectangular FRP jackets on rectangular
columns are also proposed [Priestley et al., 1996]. These limitations concern the axial
load ( maxν = 0.15), longitudinal reinforcement ratio ( l,maxρ = 3%) and aspect ratio
( maxh / b = 3).
For the design of oval FRP jackets applied on rectangular columns, an equivalent column
diameter, eD , is defined as
e 1 2D R R= + (6.54)
where the jacket radii are
21R b / a= ; 2
2R a / b= (6.55)
In the above expressions a and b are the dimensions of the oval jacket (Figure 6.14b).
The jacket dimensions that minimise the total length of the principal axes are
a = k b (6.56a)
2 2A B
b2k 2
= +
(6.56b)
23A
kB
=
(6.56c)
where A and B are the dimensions of the original rectangular cross-section.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
241
Then, the equivalent diameter, eD , can be used in Equation 6.49 to calculate the required
thickness of the oval jacket. It is expected that for rectangular cross-sections with high
aspect ratio the radius of the jacket will be extremely large and will result in ineffectual
confinement.
Design based on upgrade index
An alternative design procedure, developed for circular columns and based on an upgrade
index has been proposed [Monti et al., 1998]. The upgrade index, I, is defined as the ratio
of target and available values of strength, F, and displacement ductility, uµ , of the
member
tar u,tar
av u,av
FI
Fµ
=µ
(6.57)
The available strength and ductility are estimated from the geometry of the cross-section
and the material properties. The target values are calculated for a given loading. If a
strength-only intervention is decided, then u,tar u,avµ = µ . Similarly, if the objective is to
increase ductility only, then tar avF F= .
Based on regression analysis, the upgrade index, I, is written as
cu cul
co co
fI 1 k(1 ) 1
f
γ
α β ε= + + ην ρ − ε
(6.58)
where lρ is the percentage of longitudinal reinforcement.
The confined concrete strength, cuf , is estimated as
f f fu f f fucu co
co co
0.5 E 0.5 Ef f 2.254 1 7.94 2 1.254
f f
ρ ε ρ ε= + − −
(6.59)
and the ultimate strain of concrete is
c f f fucu co fu
co co
E 0.5 E2 1.25
f f
ρ εε = ε + ε
(6.60)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
242
The maximum strain, fuε , for GFRP and CFRP is respectively 0.02 and 0.01. The
parameters k, α , β , γ and η are estimated through parametric analyses and are
presented in Table 6.3.
The same authors proposed a simplified upgrade index for circular columns, considering
only the increase in curvature ductility, ϕµ , in the form
,tar
,av
I ϕ
ϕ
µ=
µ (6.61)
Following some simplifying assumptions [Monti et al., 2001] the required thickness of
the jacket is
2
cu,av cu,av2f 3
fu fu
ft 0.175DI
f
ε=
ε (6.62)
This design equation has been adopted in EC8-3 [CEN, 2003b] for the design of
strengthening for building elements. The design procedure begins with the estimation of
the available curvature ductility, based on the analysis of the cross-section with known
geometry and material properties. For given loads, the target curvature ductility is defined
and then the upgrade index is calculated from Equation 6.61. Finally, the jacket of the
thickness is calculated from Equation 6.62 for a given FRP material. No specific
limitations are given for the maximum strain of the FRP, fuε .
Table 6.3. Parameters k, α , β , γ and η from regression analysis [Monti et al., 1998]
GFRP CFRP k 1.0/0.7* 0.8/0.5* α 0.6 0.6 β 0.2 0.2 γ 0.8 0.7 η 125 125
*optimal value/minimum value
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
243
Detailing
To prevent buckling of longitudinal reinforcement for rectangular columns with shear
span L/D > 4, it is recommended [Seible et al., 1995b] to apply a jacket with minimum
thickness
f ft 6.9nD / E= (6.63)
where n is the number of longitudinal rebars.
The jacket must be extended beyond the expected plastic hinge region. For bridge
columns with typical axial loads ν < 0.3, the confinement length, fl , should be [Seible et
al., 1995b]
fl L / 8≥ (6.64a)
fl 0.5D≥ (6.64b)
measured from the cross-section of maximum moment. In the previous expressions L is
the length of the member and D is the (equivalent) diameter of the cross-section.
Moreover, a jacket with reduced thickness, 0.5 ft , should be extended for a length for a
distance equal to 2 fl from the cross-section of maximum moment.
When the jacket increases significantly the cross-section, a similar increase in strength
and stiffness of adjacent members should be avoided. For that purpose, a gap should be
provided between that jacket and adjacent members. A minimum gap length of 50 mm is
recommended [Seible et al., 1995b], while more exact calculations can be performed
considering the maximum expected plastic hinge rotation.
When applying an FRP jacket it is important to round the corners of the as-built cross-
section. The effect of curvature radius was experimentally investigated [Yang et al.,
2001]. Depending on the curvature radius, two different failure positions were identified
for the FRP strips: at the corner (for radii smaller than 19 mm) and at the flat portion (for
larger radii). When using large radius, a smaller difference in the strains measured at
different positions along the perimeter was observed. The test results indicate that only
67% of the nominal ultimate strength can be developed when the FRP is wrapped around
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
244
a circular section. As the corner radius decreases, the efficiency of the wrapping
decreases.
6.3.7. Retrofit for enhancement of lapped splices
It was common practice in bridge piers designed according to early seismic codes to
splice the longitudinal rebars just after the foundation, i.e. within the potential plastic
hinge zone. This can be the cause of brittle failure, before reaching the nominal flexural
strength, associated with limited deformation capacity. The use of FRP jackets has been
proposed for enhancement of lapped splices through confinement.
Before presenting the design guidelines, a simple model for failure of lapped splices
[Priestley et al., 1996] is introduced. Failure of lapped splices requires the formation of a
series of fracture surfaces orthogonal to the column surface to allow the rebars to slide
relative one to the other and a further fracture surface parallel to the column surface to
allow the radial cracks to dilate and to permit the bars to slide relative to the concrete core
(Figure 6.15).
A characteristic block of concrete can be identified, whose length is equal to the
overlapping length, bld , and the perimeter, p, for rectangular columns is
bl bl
sp 2(d c) 2 2(d c)
2= + + ≤ + (6.65a)
and for circular columns
bl bl
Dp 2(d c) 2 2(d c)
2nπ
= + + ≤ + (6.65b)
Figure 6.15. Failure of lapped splices
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
245
In the above equations s is the spacing between pairs of spliced rebars, c is the concrete
cover and n is the number of longitudinal rebars evenly distributed in the cross-section.
A jacket can be designed to provide adequate lateral compression that will not allow the
development of the aforementioned crack surfaces. The confining stress required to
inhibit failure of lapped splices can be estimated using the expression [Priestley et al.,
1996]
2bl s
ls
0.8d ff
pl= (6.66)
where sf is the stress to be transferred between the spliced rebars, sl is the overlapping
length and p is the perimeter of the characteristic block of concrete. For jackets within the
plastic hinge region, the stress to be transferred between the lapped rebars should
correspond to hardening and consider material overstrength. It is recommended [Priestley
et al., 1996] to consider s yf 1.7f= , where yf is the nominal yield stress of steel. More
relaxed criteria, suggest s yf 1.4f= [Seible et al., 1995b]. The more conservative
overstrength factor, s yf 1.7f= , can be used on the safe side. Then, Equation 6.66 becomes
2bl y
ls
1.21d ff
pl= (6.67)
Based on the simplified model of a circular jacket shown in Figure 6.16 [Seible et al.,
1995b], the jacket tensile force, fT , developed by the jacket stress, ff , acting over the
jacket thickness, ft , is calculated as
f f fT t f= (6.68)
From equilibrium of forces (Figure 6.16) we obtain
f f l2t f f D= (6.69)
and the required jacket thickness is
lf
f
Dft
2f= (6.70)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
246
Figure 6.16. Confinement of circular column with circular jacket, equilibrium of internal
forces
where the lateral pressure, lf , to avoid failure of lapped splices is obtained from Equation
6.67.
Following the approach proposed by the FHWA Seismic Retrofit Manual for Highway
Bridges [Buckle & Friedland, 1995], the minimum required jacket thickness of active, at ,
and passive, pt , FRP reinforcement for enhancement of lapped splices is obtained from
the expression
a a p p at E t E 500D(2 f )+ = − (6.71)
The jacket needs to be provided only over the length of the lapped splices.
Based on experimental results, it is suggested to limit the strain of the FRP below fuε =
0.001 [Seible et al., 1995b]. Then, the stress on the jacket is f ff 0.001E= . However, due
to the low modulus of elasticity of certain composite materials, particularly GFRP
composites, the stress developed at that level of strain, fuε = 0.001, will be low and a very
large jacket thickness will be required. For that reason, it has been proposed to provide
active confinement of the column, either by winding the FRP under tension, or by
prestressing the jacket by pressure grouting between the jacket and column [Priestley et
al., 1996]. In that case, the required jacket thickness is estimated as
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
247
2bl f
as
ff
0.56d ff
plt 0.5D
0.0015E
−= (6.72)
where af is the active confinement stress after losses, and the maximum FRP strain is
considered equal to fuε = 0.0015.
The contribution of existing horizontal reinforcement can be considered in the design of
the jacket for enhancement of lap splice behaviour. When the horizontal reinforcement
ratio is low, it can be omitted, but the contribution of circular hoops or spirals can be
estimated considering the same limit for lateral dilation, suε = 0.001, as [Seible et al.,
1995b]
sh shh
0.002A Ef
Ds= (6.73)
where shA , shE and s are the area, elastic modulus and spacing of horizontal
reinforcement. The jacket should extend in the region recommended for the jacket for
confinement according to Equation 6.64 [Seible et al., 1995b].
The above are valid for circular jackets applied on circular cross-section. Since the lateral
pressure to prevent failure of lapped splices in piers with rectangular cross-section can be
quite high, a curved jacket is required to provide such force. Therefore, no rectangular
jackets are recommended. However, if controlled debonding is allowed, rectangular
jackets can prevent cover concrete from spalling and preserve the vertical load carrying
capacity of the column. A design rule for twice the jacket thickness of a circular jacket is
recommended, with the same restrictions for the geometry and dimensions as before
[Seible et al., 1995b]. The concept of equivalent diameter can be applied for the design of
circular or oval jackets for rectangular cross-sections.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
248
6.4. DESIGN OF FRP JACKETS FOR PIERS WITH RECTANGULAR
HOLLOW CROSS-SECTION
6.4.1. General
Up to this point, the seismic deficiencies of existing bridge piers have been identified, as
discussed in Chapters 3 and 4. In addition, in the first part of the present chapter the
available retrofit techniques have been reviewed. It emerges that while design guidelines
exist to remedy most seismic deficiencies, there is a lack for the particular problem of
confinement of piers with hollow cross-section. Experimental evidence [Peloso, 2003]
suggests that FRP jackets provide effective confinement for piers with this particular
geometry. Nevertheless, because of the small scale of the specimens and the limited
number of tests, further studies are needed to verify the effectiveness of this technique for
hollow piers with large dimensions.
Towards this direction, numerical analyses have been performed in order to identify the
effect of various parameters. A two-level approach was adopted. First, the effect of the
FRP jacket on the properties of concrete was examined by FEM analyses. This was
necessary because of the inconsistencies of existing constitutive laws for FRP-confined
concrete (see 6.3.6) and of the limited confidence in their applicability to large hollow
cross-sections. Then, moment-curvature analysis of a hollow cross-section was performed
considering different configurations. The effect of confinement was modelled by
appropriately modifying the properties of concrete. The proposed approach was verified
against experimental results in the previous chapter and only the parametric analyses are
discussed in the following.
6.4.2. Numerical analysis - effect on concrete properties
Description of the numerical model
In order to verify the actual effect of concrete jacketing and wrapping with FRP on the
properties of concrete, FEM analyses were performed using the computer code Cast3m
[Millard, 1993]. The basic dimensions of the cross-section are defined in Figure 6.17,
while Figure 6.18 shows the concrete (in blue) and FRP (in red) elements for a parabolic
jacket with height jh = 0.10 m. The width of the pier cross-section was b = 1.0 m and the
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
249
depth d = 3.0b. The thickness of both the web and the flange was pt = 0.2b. For reasons of
symmetry, only a fourth of the cross-section was analysed. Cubic elements with 8 or 6
integration points were used for the concrete cross-section, while shell elements with 4
integration points were employed for the FRP jacket. Five values were considered for the
height of the concrete jacket, namely jh = 0.0 (no concrete jacket), jh = 0.05 m, jh = 0.10
m, jh = 0.15 m and jh = 0.20 m. The thickness of the FRP jacket was jt = 1 mm, jt = 3
mm and jt = 5 mm. In this way a total number of 15 different cases were studied. The
corner of the concrete cross-section was rounded at a radius in the order of 5 cm, as
recommended for practical applications.
A plasticity-based 3D constitutive model was used for the concrete elements; the stress-
strain curves for uniaxial compression are given in Figure 6.19. Elastic behaviour was
considered for the FPR jacket. Orthotropic material behaviour was used to account for the
presence of fibres only in one direction (at the plane of the cross-section). The material
properties in the other direction (normal to the cross-section plane) were estimated
considering the contribution of the matrix resin only. The material properties for concrete
and FRP are listed in Table 6.4. It was decided to study a GFRP jacket, which is
considered more effective for the enhancement of deformation capacity [Pinto & Monti,
2000]. Perfect bond was considered between the two materials.
Figure 6.17. Definition of cross-section geometry
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
250
Figure 6.18. Concrete (blue) and FRP (red) jacket mesh
Strain (%)
Stre
ss (M
Pa)
unconfinedFRP-confined
Figure 6.19. Stress-strain curve of concrete for uniaxial compression
Table 6.4. Material properties
cE (GPa) Pν cf (MPa) oε
33.5 0.2 36.3 0.00254
f1E (GPa) f 2E (GPa) Pν G (GPa)
52 5.0 0.2 2.0
The analyses were performed for increasing compressive axial load. At this stage, the
vertical steel rebars were omitted. In the following the results of the analyses are
discussed in terms of general stress-strain behaviour, increase in compressive strength and
distribution of the confinement effect along the cross-section.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
251
Effect of confinement within the cross-section
The distribution of the effect of confinement within the cross-section is discussed with
reference to Figure 6.20 in which the areas with the same compressive strength have the
same colour. Compressive strength is normalised to the compressive strength of
unconfined concrete. Blue colour corresponds to the smallest increase and changes to
green, yellow and red for the areas with the highest increase. The first row presents the
results for a rectangular FRP jacket applied directly on the as-built cross-section. The
second, third, fourth and fifth rows correspond to the analyses for concrete jacket with
height jh = 0.05 m, jh = 0.10 m, jh = 0.15 m and jh = 0.20 m, respectively. The thickness
of the FRP jacket increases when moving from left to right in Figure 6.20 from jt = 1 mm
through jt = 3 mm until jt = 5 mm for the last line.
A general observation from Figure 6.20 is that the effect of confinement is not
homogeneous throughout the cross-section. Zones with different extent of confinement
may be identified within the cross-section; in certain cases the boundaries between the
zones are not clearly defined. In order to avoid the analysis of a prohibitive number of
cases, the cross-section is divided as follows. Zone 1 comprises the corner of the cross-
section, without the external rounded part. As will be seen in the following, a moderate
increase in compressive and residual strength is observed in this zone. The external part
of the corner, where the biggest increase is observed is termed Zone 2. Zone 3 coincides
with the flange, where a small increase in compressive and residual strength is observed.
Finally, Zone 4 encompasses the additional concrete for the parabolic jacket. In this zone
a small increase in strength is observed. The web is not considered to benefit from the
jacket.
A similar stress pattern was identified for full rectangular sections wrapped with FRP
strips [Monti, 2003]. Resulting from FEM analysis of a rectangular concrete cross-section
under uniaxial compression, the circular core of the cross-section shows a bilinear stress-
strain curve, in accordance to the empirical constitutive models discussed previously. A
strut originating from the corner and expanding towards the core shows a smaller
enhancement of concrete properties, while the remaining parts of the cross-section are
practically unaffected.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
252
jt = 1 mm jt = 3 mm jt = 5 mm
h =
0.00
m
h =
0.05
m
h =
0.10
m
h =
0.15
m
h =0
.20
m
Figure 6.20. Effect of jacket height and thickness on the compressive strength of concrete
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
253
Effect of confinement on the concrete properties
The effect of the FRP jacket on the concrete properties is discussed in this section.
Figures 6.21 to 6.25 plot the stress-strain curves of concrete in the four zones of the cross-
section for the examined values of jacket height and thickness. The stress is obtained as
the sum of the nodal reactions at each zone, divided by the area of the zone. The stress is
normalised to the compressive strength of unconfined concrete. The stress-strain curve for
unconfined concrete is included in all graphs for comparison.
It is clearly seen in Figures 6.21 to 6.25 that the behaviour of concrete (softening after
peak strength) does not follow the empirical constitutive relations proposed for FRP-
confined concrete (bilinear relation with increasing strength for the second branch). In
fact, the effect of confinement is to increase the compressive and residual strength of
concrete. This is reminiscent of the behaviour of full cross-sections confined with steel
stirrups or jackets. A bilinear stress-strain curve with increasing stiffness for the second
branch was observed only for extremely high values of the FRP jacket thickness, jt ≈ 20
mm, that do not have practical application. Such behaviour was also observed in a few
elements close to the rounded corner. Nevertheless, it is not reflected in the global stress-
strain curves for each zone. Similar behaviour was shown in experimental tests on
rectangular concrete specimens wrapped with small amounts of FRP ( sρ =0.23% and
sρ =0.83% for CFRP and GFRP, respectively) [Karabinis & Rousakis, 2003]. The
specimens showed softening after maximum strength, while increasing strength in the
second branch was observed for higher amounts of FRP.
In some of the examined cases, namely cross-sections with small jacket thickness and/or
large jacket height, the numerical analysis was terminated before reaching a residual
strength. This was because of collapse of some elements in zone 5, where no confinement
effect was observed. The softening stiffness was calculated from the global stress-strain
curves for each zone and the residual strength was estimated by analysing a similar model
with higher ultimate deformation for the elements in zone 5.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
254
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
stj = 1 mm tj = 3 mm tj = 5 mm
zone 1
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 2
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 3
Figure 6.21. Stress-strain curves for confined concrete ( jh = 0.00 m)
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 1
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mm
zone 2
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mm
zone 3
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.22. Stress-strain curves for confined concrete ( jh = 0.05 m)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
255
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
stj = 1 mm tj = 3 mm tj = 5 mm
zone 1
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mm
zone 2
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mm
zone 3
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.23. Stress-strain curves for confined concrete ( jh = 0.10 m)
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 1
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 2
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 3
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.24. Stress-strain curves for confined concrete ( jh = 0.15 m)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
256
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 1
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 2
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0
Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 3
0.0
0.5
1.0
1.5
2.0
0.0 0.2 0.4 0.6 0.8 1.0
Axial strain (%)
Nor
mal
ised
axi
al s
tres
s
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.25. Stress-strain curves for confined concrete ( jh = 0.20 m)
Elastic behaviour was considered for the FRP elements and thus it was not possible to
introduce an ultimate deformation. This is a limitation of the analyses, as failure due to
tensile fracture of the FRP strips (and “explosive” collapse of concrete) was not
considered. Looking at the global behaviour of the cross-section in bending, the above
phenomena will happen at large deformations and close to collapse. Then, the inadequacy
of the model to capture the failure of the FRP strips is not expected to affect the global
results at failure. In all the examined cases, the FRP strain was not greater than 1.2%,
which is below the typical values of ultimate strain obtained from tensile tests. Typical
values of ultimate strain can be even higher than uε =3% [fib, 2003]. Experimental results
indicate failure of FRP strips for smaller strains and then a safety factor in the order of 0.6
has been proposed. This design value was not surpassed in the analyses.
The analytical results are further discussed with focus on characteristic values of the
stress-strain curves. The properties of interest are the compressive strength, the residual
strength and the softening stiffness. These parameters, on one hand, give a qualitative
view of the improvement of the concrete properties due to confinement. On the other
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
257
hand, they are the material parameters used for the moment-curvature analysis described
in the following section.
The effect of the jacket on the concrete compressive strength is discussed in more detail
with reference to Figure 6.26. For all cases the concrete compressive strength increases
with the jacket thickness, jt . In Zones 1 and 3 the compressive strength is increased by
40% and 20%, respectively, for jt = 5 mm. In Zone 2 the compressive strength is
increased by more than 80% for jt = 5 mm. The compressive strength enhancement
increases until jh = 0.10 m and then remains constant. Finally, it is interesting to note that
concerning Zone 4, the compressive strength decreases with the jacket height. The above
observations suggest that a parabolic jacket of limited height is beneficial for the concrete
properties, but when the jacket becomes relatively large, no further beneficial effect is
obtained, or even a detrimental effect is observed. The value jh = 0.10 m can be
considered as an upper limit for the examined case.
Figure 6.27 plots the change of the concrete residual strength for different values of the
jacket height and thickness. The residual strength is normalised to the compressive
strength of unconfined concrete. The residual strength increases with the jacket thickness.
The area that benefits more is again the region near the rounded corner, where in most
cases a bilinear response is obtained. Moderate improvement is observed in Zones 1 and
4, while in Zone 3 the residual strength remains at less that 0.6 cf . An interesting feature is
that for small amount of FRP, jt = 1 mm, the improvement does not increase with the
jacket height. For what concerns the residual strength, no further improvement is obtained
for jackets with height larger than jh = 0.10 m.
The softening stiffness of concrete, normalised to the compressive strength of unconfined
concrete, is presented in Figure 6.28 for the examined values of jacket thickness and
height. The softening stiffness is obtained as the difference of the maximum and residual
strength, divided by the difference of the corresponding strains. It is observed in Figure
6.28 that the softening stiffness (absolute values) decreases with the jacket height. The
decrease is faster until jh = 0.10 m and then no change, or slow decrease is observed. The
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
258
jacket height does not seem to significantly affect the softening stiffness, with the
exception of Zone 4, where smaller stiffness corresponds to larger jacket height, jh .
1.0
1.2
1.4
1.6
1.8
2.0
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Com
pres
sive
str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mm
zone 1
1.0
1.2
1.4
1.6
1.8
2.0
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Com
pres
sive
str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 2
1.0
1.2
1.4
1.6
1.8
2.0
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Com
pres
sive
str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 3
1.0
1.2
1.4
1.6
1.8
2.0
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Com
pres
sive
str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.26. Effect of jacket on the compressive strength of concrete
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Res
idua
l str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 1
0.0
0.5
1.0
1.5
2.0
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Res
idua
l str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Res
idua
l str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 3
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Res
idua
l str
engt
h
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.27. Effect of jacket on the residual strength of concrete
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
259
0
50
100
150
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Sof
teni
ng s
tiffn
ess
tj = 1 mm tj = 3 mm tj = 5 mmzone 1
0
50
100
150
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Sof
teni
ng s
tiffn
ess
tj = 1 mm tj = 3 mm tj = 5 mmzone 2
0
50
100
150
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Sof
teni
ng s
tiffn
ess
tj = 1 mm tj = 3 mm tj = 5 mmzone 3
0
50
100
150
0.00 0.05 0.10 0.15 0.20Jacket height (m)
Sof
teni
ng s
tiffn
ess
tj = 1 mm tj = 3 mm tj = 5 mmzone 4
Figure 6.28. Effect of jacket on the softening stiffness of concrete
6.4.3. Numerical analysis - effect on cross-section ductility
Description of the numerical model
Moment-curvature analyses were performed using the finite element code Cast3m
[Millard, 1993] with the aim to study the effect of axial load and reinforcement ratio on
the effectiveness of the retrofit. The cross-section of the numerical model is shown in
Figure 6.29. An equivalent I cross-section was analysed, instead of the original
rectangular hollow cross-section. This geometry of the cross-section was proved to
provide results in agreement with experimental data, as discussed in Chapter 5. The steel
elements (shown as red points in Figure 6.29) were distributed at the external and internal
faces of the flanges, as well as through the web: 20% of vertical reinforcement was
concentrated at the external face of the flange, 10% was concentrated at the internal face
of the flanges and the remaining 70% was distributed along the web.
Non-linear cyclic behaviour was considered for the concrete fibres and a modified
Menegotto-Pinto constitutive law for the steel fibres [Guedes et al., 1994]. The
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
260
constitutive laws were described in the previous chapter and only the monotonic stress-
strain curves for concrete and steel are presented in Figure 6.30 for reasons of
completence. The effect of confinement was considered by modifying the concrete
properties in accordance to the results of the previous FEM analyses. The cross-section
was divided in the five distinct zones described in the previous section and shown with
different colours in Figure 6.29.
Moment-curvature analysis of the cross-section described above was performed.
Monotonically increasing curvature was imposed at the presence of constant axial load.
Different values of axial load (ranging from ν = 0.0 to ν = 0.3) and longitudinal
reinforcement ratios (namely: sρ = 0.17%, sρ = 0.34%, sρ = 0.68% and sρ = 1.02%) were
considered. The normalised axial load, c cP / A fν = , and the ratio of longitudinal
reinforcement, s s cA / Aρ = , are defined for the original rectangular cross-section without
the jacket, cA . P is the axial force, cf is the nominal compressive strength of concrete
( cf = 35 MPa) and sA is the area of steel rebars.
Curvature ductility capacity
Curvature ductility is defined considering a bilinear envelop of the moment-curvature
diagram (Figure 6.31a). The yield moment is considered equal to the maximum moment,
maxM . The yield curvature, yϕ , is defined at the intersection of a line from the origin
passing through the numerical curve at 0.75 maxM and a horizontal line at maxM . The
ultimate curvature, uϕ , is defined at the point where the resistance reduces to uM =
0.8 maxM , or at the point where there is a sudden drop of resisting moment, whichever
occurs first. Finally, curvature ductility, ϕµ , is calculated as
u y/ϕµ = ϕ ϕ (6.74)
This definition follows the most common procedures for the estimation of ductility from
experimental data [Park, 1989] and the standard definitions given in seismic codes for the
design of structures, e.g. EC8 [CEN, 2002].
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
261
Figure 6.29. Concrete and steel mesh for the moment-curvature analysis ( jh = 0.05 m)
Strain
Str
ess
unconfinedconfined
(a)
Strain
Str
ess
(b)
Figure 6.30. Material constitutive laws: concrete (a) and steel (b)
Curvature
Mom
ent
(a)
Curvature
Mom
ent
(b)
Figure 6.31. Definition of failure criteria and curvature ductility: at maximum moment (a)
and at 20% loss of strength (b)
The results in terms of curvature ductility are presented in Figures 6.32 to 6.37. The
parameter that mostly affects the curvature ductility capacity is the axial load. In
qualitative terms, axial load dictates the failure mode, either due to tensile fracture of steel
or due to crashing of concrete in compression. For both the as-built and retrofitted cross-
sections, increase of axial load initially increases the curvature ductility. This range of
max0.75M
maxM
max0.2M
yϕ yϕuϕuϕ
maxM
max0.75M
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
262
axial load corresponds to failure of the cross-section due to collapse of steel, as indicated
by rapid loss of resistance in the moment-curvature diagrams. After a certain value,
further increase of axial load causes decrease of curvature ductility. This range of axial
load corresponds to failure of the cross-section due to crushing of concrete, as indicated
by smooth decrease of resistance after the peak in the moment-curvature diagrams. The
limit value of axial load ranges from ν = 0 to ν = 0.1 for the as-built cross-section and
from ν = 0.05 to ν = 0.20 for the jacketed cross-section. This limit value decreases with
the reinforcement ratio. This point of transition is also responsible for the peculiar shape
and discontinuities of some curves shown in the following figures, e.g. Figure 6.35 for
jt = 3 mm and jt = 5 mm.
Axial load somehow influences also the effect of the jacket thickness, jt , on the curvature
ductility of the cross-section. Keeping all other parameters constant, curvature ductility
does not significantly increase with jacket thickness. As seen in the following figures, the
actual effect of jacket thickness is to move the transition point of the curves to higher
values of axial load. On the other hand, the effect of jacket height, jh , seems to be
independent of the level of axial load. Similarly to what was observed previously for the
enhancement of the properties of concrete, the curvature ductility capacity increases with
the jacket height until the value jh = 0.10 m, after which no further improvement is
obtained. As expected, curvature ductility decreases with increasing amount of
longitudinal reinforcement.
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
Figure 6.32. Effect of axial load and amount of reinforcement on the curvature ductility
(as-built cross-section)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
263
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
adp hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.33. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.00 m)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.34. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.05 m)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
264
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
adp hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.35. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.10 m)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.36. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.15 m)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
265
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70
Curvature ductility
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.37. Effect of axial load and amount of reinforcement on the curvature ductility
( jh = 0.20 m)
Selected moment-curvature diagrams are presented in Figures 6.38 to 6.41 in order to
illustrate the effect of the examined geometrical and mechanical characteristics. The
curves shown in Figure 6.38 highlight the effect of reinforcement ratio. The curves
correspond to jacket height jh = 0.10 m, axial load ν = 0.1, and jacket thickness jt = 5
mm. As expected, higher resistance corresponds to larger amount of longitudinal
reinforcement. This increase in strength affects the conventional yield curvature and
although all cross-sections fail at almost the same curvature (which corresponds to
attainment of the ultimate strain in the outermost steel element), smaller ductility
corresponds to the cross-sections with higher axial load.
The effect of axial load is shown in Figure 6.39 that plots the moment-curvature diagrams
for jacket height jh = 0.10 m, jacket thickness jt = 5 mm and longitudinal reinforcement
ratio sρ = 1.02%. The resisting moment increases and the ultimate curvature decreases
with increasing values of axial load. This in turn leads to drastic decrease of curvature
ductility for high axial load. Figure 6.39 is very illustrative also on what concerns the
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
266
effect of axial load on the failure mode of the cross-section: collapse of reinforcement for
ν = 0.1 and concrete crushing for ν = 0.2 and ν = 0.3.
0
5000
10000
15000
20000
0.000 0.010 0.020 0.030 0.040
Curvature (1/m)
Mom
ent (
kNm
)
Figure 6.38. Effect of reinforcement ratio ( jh = 0.10 m, jt = 5 mm, ν = 0.1)
0
5000
10000
15000
20000
25000
0.00 0.01 0.02 0.03 0.04
Curvature (1/m)
Mom
ent (
kNm
)
Figure 6.39. Effect of axial load ( jh = 0.10 m, jt = 5 mm, sρ = 1.02%)
8.17,1.0 =µ=ν ϕ
0.9%, 21.3ρ = µ =
6.14,2.0 =µ=ν ϕ
ρ = µ =
8.5,3.0 =µ=ν ϕ
0.9%, 21.3ρ = µ =
8.17%,02.1s =µ=ρ ϕ
6.19%,68.0s =µ=ρ ϕ
4.24%,34.0s =µ=ρ ϕ
7.39%,17.0s =µ=ρ ϕ
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
267
0
5000
10000
15000
20000
0.00 0.01 0.02 0.03 0.04 0.05
Curvature (1/m)
Mom
ent (
kNm
)
Figure 6.40. Effect of jacket thickness ( jh = 0.10 m, sρ = 1.02%, ν = 0.1)
0
2000
4000
6000
8000
10000
0.00 0.01 0.02 0.03 0.04
Curvature (1/m)
Mom
ent (
kNm
)
Figure 6.41. Effect of jacket height ( jt = 5 mm, sρ = 0.17%, ν = 0.1)
The effect of jacket thickness for a cross-section with axial load ν = 0.1, longitudinal
reinforcement sρ = 1.02% and jacket thickness jt = 5 mm is shown in Figure 6.40. The
resisting moment is not significantly affected by the jacket thickness. The ultimate
curvature and curvature ductility increase with jt . Note that in this example and for the
highest value of jacket thickness, jt = 5 mm, curvature ductility decreases. This is due to
9.12,mm1t j =µ= ϕ
2.20,mm3t j =µ= ϕ
0.9%, 21.3ρ = µ =
8.17,mm5t j =µ= φ
0.39,m15.0h j =µ= ϕ 7.39,m10.0h j =µ= ϕ 6.37,m05.0h j =µ= ϕ 4.35,m00.0h j =µ= ϕ
4.38,m20.0h j =µ= ϕ
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
268
the change of failure mode, which is due to steel, while failure was determined by
concrete for the smaller values of jacket height.
Finally, the effect of jacket height is discussed with reference to Figure 6.41, in which the
curves for ν = 0.1, jt = 5mm and sρ = 0.045 are plotted. The effect of the jacket height is
slightly beneficial until = 0.10 m, but after this value, a slow decrease of curvature
ductility is observed. This is similar to the observation made on the effect of the jacket
height on the concrete properties and supports the suggestion to consider an upper limit in
the jacket height for practical applications.
Effectiveness index
The ratio of the curvature ductility of the jacketed cross-section to the curvature ductility
of the as-built cross-section can be used as an effectiveness index. The effectiveness
index is plotted in Figures 6.42 to 6.46 for different values of the examined parameters.
The effect of axial load is to initially increase the effectiveness of the retrofit (increase of
the effectiveness index until ≈µµ −φφ builtas,dretrofitte, / 7) and then to decrease the
effectiveness, similarly to the effect on the ductility capacity of the cross-section.
Furthermore, the effectiveness index increases with the jacket thickness and jacket height.
The effect of longitudinal reinforcement ratio is related to the axial load. In fact, as seen
in the following figures, the effectiveness index increases with axial load before the limit
value and then decreases. This inversion of the trend is again related to the change in
failure mode, which depends on the applied axial load.
The limit value of the axial load ranges from ν = 0.05 to ν = 0.25, depending on the
amount of longitudinal reinforcement, jacket thickness and jacket height. For low to
medium amounts of vertical reinforcement (from sρ = 0.17% to sρ = 0.68%) jacketing is
not effective for piers with axial load ν < 0.10. In these cases, large tension strains
develop in the steel before significant compression on the concrete and failure is due to
collapse of steel. For higher amounts of longitudinal reinforcement jacketing is effective
even for low values of axial load. The effectiveness, though, is reduced, compared to the
cases of lower reinforcement ratios.
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
269
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
adp hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.42. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.00 m)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.43. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.05 m)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
270
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
adp hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.44. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.10 m)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.45. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.15 m)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
271
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 1 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 3 mm
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1 2 3 4 5 6 7
Curvature ductility (confined/as-built)
Nor
mal
ised
axi
al lo
ad
p hor = 0.17%p hor = 0.34%p hor = 0.68%p hor = 1.02%
tj = 5 mm
Figure 6.46. Effect of axial load and amount of reinforcement on the effectiveness index
( jh = 0.20 m)
Enhancement of moment capacity
For the given range of axial load, the ratio of moment capacity of the retrofitted cross-
section to the moment capacity of the as-built cross-section is constantly increasing. As
expected, the increase is higher for higher axial load. The maximum increase of moment
capacity, for the cases examined in this study, is about 20% and in few cases reaches
30%. This corresponds to an increase in shear demand that might exhaust the as-built
shear capacity of the pier. In that case, retrofit for shear strength enhancement will be
required in order to avoid brittle failure, following Capacity Design philosophy. Also,
retrofit of the foundation might be needed. When increasing the dimensions of the cross-
section an increase of stiffness is expected. A member with increased stiffness will attract
higher seismic forces and this fact has to be taken into consideration when designing the
global retrofit. This increase will also affect the dynamic properties, which are significant
for bridge structures. Jacketing will decrease the usually high periods of bridge piers and
then most probably will increase the spectral ordinates and accordingly the seismic
demand.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
272
-25000
-20000
-15000
-10000
-5000
0
5000
10000
15000
20000
25000
-0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04
Curvature (1/m)
Mom
ent (
kNm
)
as-builtretrofitted (h = 0.2 m)retrofitted (h = 0.0 m)
Figure 6.47. Cyclic behaviour: moment-curvature diagrams for the as-built and jacketed
cross-section ( ν =0.2, jt = 5 mm, sρ = 1.02%).
Cyclic behaviour and energy-dissipation capacity
The effect of the addition of concrete and the wrapping with FRP on the energy-
dissipation capacity is verified in Figure 6.47 that plots the moment-curvature numerical
curves for the favourable case of ν = 0.2, sρ = 1.02% and jt = 5 mm and the two extreme
values of the jacket height. For no concrete jacket and the same values of curvature, no
difference is observed between the as-built and the retrofitted cross-sections.
Nevertheless, the retrofitted cross-section has a much higher curvature ductility, which
can be exploited also under cyclic loading. For higher values of the concrete jacket, both
strength and ultimate curvature increase and then a more pronounced dissipation capacity
is evidenced by the wider cycles shown in Figure 6.47.
6.4.4. Design equations and recommendations
Curvature ductility
The objective of this section is to derive empirical design equations based on the
numerical results presented previously. First, an expression of curvature ductility as a
function of the examined geometrical and mechanical parameters (namely axial load,
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
273
longitudinal reinforcement ratio, height and thickness of the jacket) is sought. These
parameters are grouped in a single design parameter, termed j j sS f ( ,h , t , )= ν ρ in the
following, and then an expression in the form f (S)ϕµ = is fitted to the numerical results.
Observing Figures 6.32 to 6.37, there seems to be a limit value of normalised axial load
until which the values of curvature ductility increase and after which they decrease.
Although in general this limit value is not an independent parameter, the value ν = 0.1
was chosen. The results of the numerical analyses were divided in two groups according
to the limit value of axial load. This choice is arbitrary and based on a visual
interpretation of the obtained results, but simplifies the derivation of the empirical
formulae. In fact, it was not possible to obtain an empirical fit with high correlation,
without introducing this grouping. For each group different forms of the functions
j j sS f ( ,h , t , )= ν ρ and f (S)ϕµ = were studied. Figure 6.48 plots the values of curvature
ductility versus the design parameter, S, for the two groups. A linear relation fits the
numerical results for ν ≤ 0.1 (Figure 6.48a), while a power expression fits the results for
ν > 0.1 (Figure 6.48b).
The empirical formulae for ν ≤ 0.1 are
63.10S42.57 1m, +=µϕ (6.75a)
23.7S94.45 105.0, +=µϕ (6.75b)
8 0.03
j j1
s
(1 ) (1 h )(1 t )S
10.045
+ ν + +=
ρ+
(6.75c)
and for ν > 0.1
78.12m, S63.0 −
ϕ =µ (6.76a)
74.1205.0, S35.0 −
ϕ =µ (6.76b)
0.7 0.08 0.3
j s2 0.3
j
(1 h )S
(1 t )
ν + ρ=
+ (6.76b)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
274
In the previous equations ν is the normalised axial load, jh is the jacket height (in m), jt
is the jacket thickness (in mm) and sρ is the longitudinal reinforcement ratio (in %). m,ϕµ
is the mean value of curvature ductility, while 05.0,ϕµ is the 5% characteristic value (95% of the
empirical values are lower than the corresponding numerical values). The numerical values are
plotted in Figure 6.48, along with Equations 6.75 and 6.76. The correlation factor for Equation
6.75a is 2R = 0.82 and for Equation 6.76a it is 2R = 0.70.
0
10
20
30
40
50
60
70
0.0 0.1 0.2 0.3 0.4 0.5 0.6
S1
Cur
vatu
re d
uctil
ity
Equation 6.75aEquation 6.75b
(a)
0
10
20
30
40
50
60
70
0.0 0.1 0.2 0.3 0.4 0.5 0.6
S2
Cur
vatu
re d
uctil
ity
Equation 6.76aEquation 6.76b
(b)
Figure 6.48. Empirical fit to the numerical values of curvature ductility: ν ≤ 0.1 (a) and
ν > 0.1 (b)
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
275
Figure 6.49 plots the empirical values according to Equations 6.75b and 6.76b versus the
numerical values. A straight line passing from the origin with a 45o slope is also plotted in
Figure 6.49. It is seen that conservative results are obtained, as most of the empirical
values are lower than the numerical values. It should be highlighted that the large scatter
imposed a large safety factor and then the jacket may be over-designed in many cases.
Effectiveness index
The ratio of curvature ductility of the retrofitted cross-section to the ductility of the as-
built cross-section is used as an effectiveness index. With the aim to derive an empirical
expression for the design of FRP jackets, the numerical results are studied in the two
groups defined previously. Using Equation 6.75b, the effectiveness index for ≤ν 0.1
takes the form
045.023.7)1(94.45
045.023.7)t1)(h1()1(94.45
81.0I 3.0s8
3.0s03.0
jj8
05.0,builtas,
05.0,dretrofitte,05.0 ρ
+ν+
ρ+++ν+
=µµ
=−ϕ
ϕ (6.77)
For ν > 0.1 and using Equation 6.76b, the effectiveness index is written as
14.0j
52.0j
05.0,builtas,
05.0,dretrofitte,05.0 )h1(
)t1(6.0I
++
=µµ
=−ϕ
ϕ (6.78)
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70
Curvature ductility (empirical)
Cur
vatu
re d
uctil
ity (
num
eric
al) (a)
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70
Curvature ductility (empiricall)
Cur
vatu
re d
uctil
ity (
num
eric
al) (b)
Figure 6.49. Comparison between empirical and numerical values of curvature ductility:
ν ≤ 0.1 (a) and ν > 0.1 (b)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
276
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7
Effectiveness index (empirical)
Effe
ctiv
enes
s in
dex
(num
eric
al) (a)
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7
Effectiveness index (empiricall)
Effe
ctiv
enes
s in
dex
(num
eric
al) (b)
Figure 6.50. Comparison between empirical and numerical values of effectiveness index:
ν ≤ 0.1 (a) and ν > 0.1 (b)
Because of the scatter in the numerical results, a safety factor is introduced in Equations
6.77 and 6.78, so that 95% of the empirical values are higher than the numerical values.
The comparison between the empirical and numerical values is given in Figure 6.50,
where a line with 45o inclination is also shown.
Alternative definition of curvature ductility
It is seen in the previous section that a complex relation holds between the geometric and
mechanical characteristics of the cross-section and the curvature ductility capacity. This
made it necessary to introduce an artificial grouping of the numerical results and even in
this way an unsatisfactory correlation between the empirical and numerical results was
obtained. To make things easier, an alternative definition of curvature ductility is
introduced in this section. Moment and curvature at yield are defined as before and
ultimate curvature is conservatively defined at maxM , see Figure 6.31b. Although this
might lead in over-dimensioning at certain cases, the results remain on the safe side. It is
reminded that values of the effectiveness index as high as builtas,dretrofitte, / −ϕϕ µµ = 6.8 are
obtained if ultimate curvature is defined at the point of the moment-curvature diagram
where there is a 20% loss of strength. This definition is more realistic and demonstrates
the actual effectiveness of the retrofit method.
Following the alternative definition of ductility introduced in this section, a single
expression fits all the numerical results and takes the form
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
277
+ρ+ν+−
=µϕ 2.0j
3.0s
1.0j
7.0
m, )t1()h1()1.0(86.6
exp94.52 (6.79a)
+ρ+ν+−
=µϕ 2.0j
3.0s
1.0j
7.0
05.0, )t1()h1()1.0(10.8
exp35.42 (6.79b)
2.0j
3.0s
1.0j
7.0
)t1()h1()1.0(
S+
ρ+ν+= (6.79c)
The numerical values are plotted in Figure 6.51, along with Equations 6.77a and 6.77b.
The correlation factor for Equation 6.77a is 2R = 0.81.
Using Equation 6.77b it is possible to obtain also a single empirical expression of the
effectiveness index as a function of the examined geometrical and mechanical parameters.
This expression takes the form
++
−ρν+=µµ
=−ϕ
ϕ2.0
j
1.0j3.0
s7.0
05.0,builtas,
05.0,dretrofitte,05.0 )t1(
)h1(1)1.0(1.8exp6.0I (6.80)
Due to the relatively poor correlation, a modification is included in Equation 6.78 to
obtain the 5% characteristic value of the effectiveness index (95% of the empirical values
are lower than the numerical values). The empirical and numerical values are compared in
Figure 6.52, where a straight line with 45o inclination is also shown.
Recommendations
In Equations 6.75 to 6.80 there are two unknown design parameters, namely jacket
height, jh , and jacket thickness, jt . This means that the designer has to select one of
them and then enter the design formulae, or the graphs presented previously, with this
value and the desired value of curvature ductility, or effectiveness index, in order to
calculate the other. It might be preferable to limit the jacket height so as not to increase
the strength and stiffness of the retrofitted member. The limit jh = 0.10 m, identified in
this study, may be considered. Alternatively, it might be desired to limit the jacket
thickness in order to avoid practical problems related to the superposition of many FRP
layers. The above imply an iterative procedure and require some engineering judgement.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
278
0
10
20
30
40
0.0 0.1 0.2 0.3 0.4 0.5 0.6
S = f(v, hj,tj,p)
Cur
vatu
re d
uctil
ity
Equation 6.79aEquation 6.79b
Figure 6.51. Empirical fit to the numerical values of curvature ductility
0
1
2
3
4
0 1 2 3 4
Effectiveness index (empirical)
Effe
ctiv
enes
s in
dex
(num
eric
al)
Figure 6.52. Empirical (5% characteristic) and numerical values of the effectiveness
index
The equations derived in this section can be used when the geometrical and mechanical
characteristics of the piers fall within the limits examined in the present study.
Considering axial load, longitudinal reinforcement ratio, jacket height and thickness it is
believed that the whole range of interest has been examined. For piers with dimensions
smaller than those examined in this study, the jacket dimensions may be scaled down and
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
279
conservative results will be obtained. The formulae should not be used for piers with
larger dimensions because it is expected that confinement will not be the same effective
and the proposed empirical equations might provide unsafe results.
Two groups of design formulae have been elaborated, namely Equations 6.75 to 6.78 and
Equations 6.79 and 6.80. For the first group a failure criterion, considering failure due to
concrete crushing or rupture of steel, was used to define ultimate curvature. Then, a limit
value of the normalised axial load was set and different formulae were elaborated for the
two sub-groups of the numerical results. To avoid this artificial grouping, an alternative
definition of ultimate curvature was considered and a single equation was found to fit the
numerical values. The alternative definition of ultimate curvature simplifies the
procedure, because the failure mode does not influence the ultimate curvature.
Conservative results are obtained because maximum moment occurs always at curvature
lower or equal to the curvature that corresponds to failure of concrete or steel fibres. For
reasons of simplicity, it seems preferable to use Equations 6.79 and 6.80.
6.5. CONCLUDING REMARKS
Several methods have been proposed for the retrofit of seismic-deficient bridge piers
using FRP reinforcement. External reinforcement is applied for the enhancement of
flexural or shear strength, while jackets are wrapped around the piers for the improvement
of ductility capacity and behaviour of lapped splices. Design equations, which take into
consideration all possible failure modes, have been proposed for flexural and shear FRP
reinforcement. Regarding FRP jackets, noteworthy deviations are observed in the
predictions of different models for the properties of confined concrete. The semi-
empirical rules proposed for the design of FRP jackets for rectangular or circular solid
cross-sections cannot be simply extrapolated to the case of hollow cross-sections with
large dimensions.
Numerical analyses of a concrete cross-section wrapped with an FRP jacket suggest that
for the case of hollow piers with large dimensions, the existing empirical laws for FRP-
confined concrete are not suitable. The effect of lateral pressure provided by the jacket is
to increase the maximum and residual stress, as well as the ultimate deformation of
concrete, similarly to concrete confined by steel stirrups or jackets. A softening branch,
with a slope less steep than the one for unconfined concrete, follows the peak strength. A
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
280
bilinear stress-strain curve with increasing stress in the second branch, as proposed for
full cross-sections, is only observed for unrealistic values of the jacket thickness.
The effect of confinement is mainly concentrated in the corners and its extent depends on
the dimensions of the jacket. For rectangular jackets, the confined zone is limited to the
corners, even for large amounts of FRP. For oval jackets, the confined area comprises
also part of the flange and added concrete. In this case, the confined area increases with
increasing thickness of the FRP jacket. A limit value of the concrete jacket, after which
no further enhancement of the concrete properties was obtained, was identified.
The effectiveness of jacketing for the enhancement of the deformation capacity of cross-
sections is conditioned by the axial load, amount of longitudinal reinforcement and
dimensions of the concrete jacket. The jackets are most effective in the range of ν = 0.1 to
ν = 0.3, which, luckily, corresponds to the axial loads bridge piers usually carry. The
effectiveness reduces with increasing amount of longitudinal reinforcement. Parabolic
jackets are found to be more effective for enhancement of ductility capacity than
rectangular jackets. Indicatively, it is noted that the curvature ductility of the jacketed
cross-section can be up to 7 times the ductility of the as-built cross-section, provided a
significant amount of FRP and the corresponding oval concrete jacket are provided. For
the simplest solution of a rectangular jacket, i.e. application of the FRP strips directly
around the original cross-section without adding new concrete, the curvature ductility can
be increased up to 3 times, as seen in Figure 6.42. This demonstrates that this fast and
economic intervention can effectively improve the deformation capacity of seismic-
deficient bridge piers with rectangular hollow cross-section.
Parabolic jackets increase the stiffness of the cross-section. This might result in higher
design shear forces. On the other hand, the moment capacity of the cross-section is
slightly increased (about 20%). Upgrade of adjacent members might then be required.
Based on the results of numerical analyses, FRP jackets are effective in improving the
cyclic behaviour and the energy-dissipation capacity of retrofitted cross-sections,
particularly for the case of cross-sections with medium to high amount of longitudinal
reinforcement. In case a very small amount of longitudinal reinforcement is present,
failure will be due to steel rupture and then the enhancement of the concrete properties
will not affect the global behaviour of the cross-section. Therefore, more intrusive
SEISMIC RETROFIT OF EXISTING HOLLOW BRIDGE PIERS
281
interventions (e.g. additional steel reinforcement) will be required in order to increase the
dissipation capacity.
On the grounds of limited experimental observations and extended parametric numerical
analyses, it is concluded that parabolic FRP jackets constitute an effective method for
improving the seismic response of poorly-detailed hollow bridge piers, keeping in mind
the limitations in terms of axial load and reinforcement ratio, highlighted in this study. To
assist in the design procedure, empirical design equations have been elaborated on the
basis of more than 1000 numerical analyses.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
282
CONCLUSIONS AND FUTURE RESEARCH
283
7. CONCLUSIONS AND FUTURE RESEARCH
In this final chapter the salient aspects of the research presented in the thesis are
summarised and the principal conclusions are put forward. The chapter is organised in
sections that recapitulate the main findings of each chapter of the thesis. In closure, a
number of suggestions for future research are given.
7.1. SUMMARY AND CONCLUSIONS
7.1.1. Performance of existing bridge piers with hollow cross-section
Cyclic tests were performed on large-scale (1:2.5) specimens of existing bridge piers with
rectangular hollow cross-section. A short pier with lapped splices at the base and a tall
pier with premature termination of the longitudinal reinforcement were tested. Given the
scale and geometry of the specimens, the test results constitute an important contribution
to the assessment of the seismic performance of existing highway bridge piers.
In both piers a collapse mechanism with multiple hinges (above the base, lapped splices
and bar cut-off) was observed. The failure mode of the short pier was flexure-dominated
with limited spread of plasticity, while a combination of flexure and shear dictated the
failure mode of the tall pier. Limited deformation capacity, because of the small amount
of horizontal and vertical reinforcement, was observed in both specimens. The lack of
protection against buckling resulted in collapse of the vertical rebars due to low-cycle
fatigue. The initiation of different failure mechanisms at distinct levels of lateral
displacement and various regions of the piers calls for a global retrofit procedure that
takes into account and addresses the whole range of possible failure modes.
The importance of seismic detailing, principally of the minimum requirements for
horizontal reinforcement, was verified by comparing the short pier to one designed
according to EC8. The first pier had smaller ductility and energy-dissipation capacity and
almost half the drift capacity of the second.
The conventional definition of ductility, based on a bilinear approximation of the force-
displacement diagram, can be misleading or contradictory as a meaningful parameter of
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
284
the deformation capacity. On the contrary, drift and energy-dissipation capacities can
fully characterize the cyclic behaviour of these bridge piers.
The predictions of empirical formulae were in fair agreement with the experimental
values, as far as yield displacement is concerned. On the other hand, ultimate
displacement strongly depends on the equivalent plastic hinge length. For piers with
multiple hinges, the contribution of all hinges (possibly not fully developed) must be
considered. The empirical formulae were found to be valid for hollow cross-sections, but
not for members with lapped splices, for which the equivalent plastic hinge length was
much smaller than the predicted values.
For these piers with elongated hollow cross-section deformation due to shear was found
to constitute a significant fraction of the total displacement. For the short pier the ratio of
shear to flexural displacement was 0.3, while for the tall pier this ratio increased with
imposed top displacement from 0.2 to 0.5.
A database of experimental results on specimens of bridge piers with hollow cross-section
has been compiled. On this basis, it is concluded that hollow piers without seismic design
are expected to have limited deformation capacity with mean values of drift uδ = 2.9% for
flexural failure mode and uδ = 1.9% for shear failure mode. These values show that
existing bridge piers need upgrading. Hollow piers with sufficient confinement of the
compression zone are expected to have stable behaviour and large deformation capacity,
uδ > 3.6%. The previous considerations give also a hint on the objectives of retrofit: the
main seismic deficiencies that have to be addressed are the limited shear resistance and
poor confinement.
7.1.2. Performance of existing bridge structures
A series of pseudodynamic tests on a large-scale (1:2.5) model of a highway bridge were
successfully performed. The substructuring technique, considering non-linear behaviour
for parts of the numerical substructure, was implemented for the first time at world level.
The representativeness of the testing method was checked against the results of dynamic
non-linear numerical analyses. This represents an advance in experimental techniques and
demonstrates the effectiveness of this hybrid method. It constitutes a step forward in the
CONCLUSIONS AND FUTURE RESEARCH
285
tele-operation of experimental facilities and distributed testing, as the tests were carried
out with a distributed computer and testing system with communication via standard
internet connection.
As far as seismic assessment of existing bridges is concerned, the results from the tests
allowed to assess the performance of a typical European bridge (highway bridge with
rectangular hollow cross-section piers and with many seismic deficiencies such as short
overlapping length, lack of transversal reinforcement, tension shift, absence of capacity
design, etc). The PSD tests demonstrated that these infrastructures represent a source of
risk in seismic regions. In fact, the test corresponding to the SLS caused minor damage.
Damage concentrated at the tall physical pier for the test that corresponds to the ULS,
whereas two numerical piers were beyond yielding. This corresponds to significant
damage, although the non collapse criterion was satisfied. Collapse was reached for the
2.0xNE test. Note that damage increased disproportionally to the seismic intensity and
that the damage pattern changed for each earthquake test.
The cycling effects resulted in a marked reduction of the resistance and displacement
capacity of the bridge piers, compared to the same components tested under a few cycles
of increasing displacement. The larger number of cycles initiated failure of the lapped
splices at the base of the short pier. It is reminded that cycling effects are not expected to
significantly influence the performance of bridge piers with proper seismic detailing.
The observed damage was compared to the predictions of simplified assessment methods.
The simplicity of probabilistic methods was highlighted against the uncertainties
encountered during the application of deterministic methods that make use of a substitute
structure. Due to the differences between the assumed and the actual deformed shapes,
the deterministic methods were unable to predict the damage distribution observed during
the tests and their results could be used only for qualitative comparisons. It was possible
to introduce a correction of the N2 method in order to account for the correct deformed
shape. Nevertheless, this correction requires the a priori knowledge of the structural
behaviour. It is concluded that among the available tools, dynamic non-linear analysis
remains the most appropriate for the detailed and reliable assessment of existing
structures with irregular configuration.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
286
The tested bridge was considered in a low-seismicity zone. The higher expected damage
in medium and high-seismicity zones was confirmed by the results of the modified N2
method. For the SLS earthquake at a high-seismicity region, significant inelastic
deformation demand was estimated for all piers, while collapse of two piers and severe
damage (close to collapse) for the remaining piers was predicted for the ULS earthquake.
At this level of damage, repair might be economically unfeasible and the solution would
be the replacement of the bridge structure. Therefore, appropriate retrofit is a necessity.
7.1.3. Assessment of numerical tools for existing bridge piers
The simple fibre/beam model was successful in simulating the response of the short pier,
whose behaviour was flexure-dominated, but failed to capture the tension shift
phenomenon, that dictated the failure mode of the tall pier, and predicted different failure
location and resistance. For the sake of simplicity and stability of the numerical processes
during the PSD tests, it was possible to obtain the correct failure location and similar
resistance and dissipation capacity, by modifying the properties of steel in the numerical
model. This modelling approach provided reliable and detailed results both at global and
local level. Keeping in mind the small computational demand, it seems appropriate for
extensive parametric studies that require a large number of analyses.
The 2D FEM analyses were able to simulate the whole range of physical phenomena that
were significant for the response of the tested specimens. However, the large
computational demand, because of the refined meshes used to describe the cross-section
and the distribution of horizontal and vertical reinforcement, as well as the steel to
concrete interface, and also because of the constitutive laws for the materials and contact
surfaces, should be highlighted. Such numerical tools appear useful for refined studies on
retrofit solutions.
A combination of the previous numerical approaches was used to study the behaviour of
bridge piers with hollow cross-section wrapped with FRP strips. The finite element
method was used to study the effect of the jacket on the concrete properties. The
observations of these analyses were then used to modify the material properties for
concrete in a fibre/beam model of a retrofitted pier. Based on the comparison between
experimental and numerical results, it is concluded that this combination of modelling
tools is valid and therefore it was used for further parametric analyses.
CONCLUSIONS AND FUTURE RESEARCH
287
7.1.4. Retrofit of bridge piers with hollow cross-section using FRP strips
Several methods have been proposed for the retrofit of seismic-deficient bridge piers
using FRP strips. Design equations, which take into consideration all possible failure
modes, have been proposed for flexural and shear FRP reinforcement. Regarding FRP
jackets, inconsistencies and limitations are observed in the predictions of different models
for the properties of confined concrete. Experimental evidence calls for a global retrofit
procedure that considers all possible failure modes (not only the weakest one) and also
rational design rules, as sometimes retrofit interventions are over-designed.
FEM analyses of a concrete cross-section wrapped with an FRP jacket showed that the
existing empirical laws for FRP-confined concrete are not suitable for hollow piers with
large dimensions. The effect of lateral containment provided by the jacket is to increase
the maximum and residual stress of concrete. A softening branch, with a slope less steep
than the one for unconfined concrete, follows the peak strength. A bilinear stress-strain
curve with increasing stress in the second branch, as proposed for full cross-sections, was
only observed for unrealistic values of the jacket thickness.
The effect of confinement is mainly concentrated in the corners and its extent depends on
the dimensions of the jacket. For rectangular jackets, the confined zone is limited to the
corners, while for oval jackets it comprises also part of the flange and added concrete. A
limit value of the jacket height, jh = 0.10 m, after which no further enhancement of the
concrete properties is obtained, was identified. The limit value corresponds to 10% of the
flange width.
The effectiveness of jacketing for the enhancement of the deformation capacity of cross-
sections is conditioned by the axial load and amount of longitudinal reinforcement. The
jackets are most effective in the range of ν = 0.1 to ν = 0.3, which, luckily, corresponds
to the axial loads bridge piers usually carry. The effectiveness is reduced with the amount
of longitudinal reinforcement. Indicatively, it is reminded that the curvature ductility of
the jacketed cross-section can be up to 7 times the ductility of the as-built cross-section,
provided a significant amount of FRP and the corresponding oval concrete jacket are
provided. For the simplest solution of a rectangular jacket, the curvature ductility can be
increased up to 3 times. This demonstrates that this fast and economic intervention can
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
288
effectively improve the deformation capacity of seismic-deficient bridge piers with
rectangular hollow cross-section.
Parabolic jackets increase the stiffness of the cross-section. The moment capacity of the
cross-section is increased by not more than 20%. Therefore, upgrade of adjacent members
might be required.
On the grounds of limited experimental observations and extended numerical analyses, it
is concluded that, keeping in mind the aforementioned limitations, FRP jackets constitute
an effective method for improving the seismic performance of poorly detailed hollow
bridge piers. To assist in the design procedure, empirical design equations were
elaborated on the basis of more than 1000 numerical analyses.
7.2. SUGGESTIONS FOR FUTURE RESEARCH
The work performed within the framework of this thesis offered an overview of the issues
that are important in seismic design, assessment and retrofit of bridges. These issues were
discussed in Chapter 2 and several research needs were identified. It is felt that the
statement “in the whole field of assessing the safety of existing bridges, designing
strengthening intervention and re-assessing the safety of strengthened bridges, more
research is needed” [Calvi & Pinto, 1996] still holds. With respect to the problems that
were the object of this work, namely assessment of as-built piers with hollow cross-
section, assessment of existing highway bridges and retrofit of hollow bridge piers using
FRP strips, some aspects that require further investigation are suggested in the following.
Considering the assessment of bridge piers with hollow cross-section, it is reminded that
only 30 specimens of seismic-deficient hollow bridge piers have been tested world-wide
(the first 4 with elongated rectangular cross-section are presented in this work), while
experimental results for a much larger population of solid bridge piers are available.
Therefore, further experimental testing involving large-scale, and hopefully full-scale,
specimens would be most interesting. The interpretation of additional experimental
results should aim at the systematic assessment of the deformation capacity of these
structures and at the identification of a relation between damage indicators and
performance levels. Furthermore, it will be possible to further validate the empirical
CONCLUSIONS AND FUTURE RESEARCH
289
formulae used for the assessment of their strength and deformation capacities, considering
the significant contribution of shear to the total displacement. With reference to the last
issue, an interesting experimental campaign, involving large-scale specimens of hollow
piers with modern seismic design, has been recently concluded [Hines & Seible, 2003].
The above will also contribute to the calibration of numerical models for these bridge
piers.
Laboratory testing of complete bridge structures is a most challenging task. The
successful completion of the pseudodynamic tests with non-linear substructuring provides
confidence in advanced experimental techniques. A further improvement is the practical
application of continuous pseudodynamic testing combined with the non-linear
substructuring technique, already sufficiently studied at theoretical level [Pegon &
Magonette, 2002]. With the same objective, an ambitious and promising project is
undertaken in the USA [Elnashai, 2002].
A large number of highway bridges designed and constructed without provisions for
seismic resistance, whose seismic vulnerability is verified by field observations as well as
experimental and numerical studies, are present in seismic-prone regions around the
world. As retrofit of all is not economically feasible, prioritization of interventions
requires fast and at the same time reliable assessment tools. It was shown in this study
that the existing simlpified assessment procedures need further refinement with attention
on bridges with irregular configuration. An interesting proposal is the estimation of
dynamic correction factors [Fajfar & Gašperšic, 1996].
While FRP strips have been used successfully for the seismic retrofit of many structural
types in the past years, their application on large bridge piers with elongated hollow
cross-section has not been studied to a satisfactory degree. This problem is of particular
importance in Europe and other seismic region around the world, where many existing
bridges comprise piers with this geometry. The most important issue is the effect of
confinement on the concrete properties and the distribution of this effect within the cross-
section. Dedicated studies should aim at the development of an appropriate mechanical
model. For full rectangular cross-sections, such a model has been recently developed on
the basis of numerical studies [Monti, 2003].
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
290
Considering design tools, the empirical formulae elaborated in Chapter 6 need further
refinement. The effect of concrete, steel and FRP properties needs to be studied, although
it is expected that no significant corrections will be needed. On the other hand, the
consideration of bond between FRP and concrete and larger cross-sectional dimensions
will provide safer design and extend the applicability of the proposed equations.
Finally, it is necessary to experimentally verify the effectiveness of the proposed
technique and design tools for the case of large and full-scale piers. The most important
parameters that should be examined are the cross-section and jacket dimensions.
REFERENCES
291
REFERENCES
1. AASHTO. Standard specifications for highway bridges, 16th edn, American Association of State Highway and Transportation Officials, Washington DC; 1995.
2. J.M. Alsiwat & M. Saatcioglu. Reinforcement anchorage slip under monotonic loading, ASCE Journal of Structural Engineering 118(9): 2421-2438; 1992.
3. M. Arduini & A. Nanni. Behavior of precracked RC beams strengthened with carbon FRP sheets, ASCE Journal of Composites for Construction 1(2): 63-70; 1997.
4. M. Aschheim & J.P. Moehle. Shear strength and deformability of RC bridge columns subjected to inelastic cyclic displacements, UCB/EERC-92/04, University of California, Berkeley; 1992.
5. M. Aschheim, J.P. Moehle & S.A. Mahin. Design and evaluation of reinforced concrete bridges for seismic resistance, UCB/EERC-97/04, University of California, Berkeley; 1997.
6. ATC. Improved seismic design criteria for California bridges: provisional recommendations, bridge design specification, Report No. ATC-32-1, Applied Technology Council, Redwood City; 1996.
7. ATC. Seismic evaluation and retrofit of concrete buildings, California Seismic Safety Commission, Report SSC 96-01, Applied Technology Council, Redwood City; 1996.
8. G.L. Balázs. Fatigue of bond, ACI Materials Journal 88(6): 620-629; 1991.
9. F. Barzegar-Jamshidi & W. Schnobridh. Nonlinear finite element analysis of reinforced concrete under short term monotonic loading, Civil Engineering Studies, SRS 530, University of Illinois at Urbana-Champaign; 1986.
10. B.J. Bett, R.E. Klinger & J.O. Jirsa. Lateral load response of strengthened and repaired reinforced concrete columns, ACI Structural Journal 85(5): 499-508; 1988.
11. S.N. Bousias, L.A. Spathis, T.C. Triantafillou & M.N. Fardis. Seismic retrofitting of corrosion-damaged RC columns, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
12. B. Bresler & V. Bertero. Behaviour of reinforced concrete under repeated load, ASCE Journal of the Structural Division 94(ST6): 567-1589; 1968.
13. I.G. Buckle & I.M. Friedland (eds). Seismic retrofitting manual for highway bridges, Federal Highway Administration; 1995.
14. A.J. Calderone, D.E. Lehman & J.P. Moehle. Behavior of reinforced concrete bridge columns having varying aspect ratios and varying lengths of confinement, Pacific Earthquake Engineering Research Center Report 2000/08; 2000.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
292
15. Caltrans. Seismic design criteria, California Department of Transportation, Sacramento, California; 1999.
16. G.M. Calvi & G.R. Kingsley. Displacement-based design of multi-degree-of-freedom bridge structures, Earthquake Engineering and Structural Dynamics 24(9): 1247-1266; 1995.
17. G.M. Calvi & P.E. Pinto (eds). Experimental and numerical investigation on the seismic response of bridges and recommendations for code provisions, ECOEST – PREC8 Report No.4, LNEC, Lisboa; 1996.
18. G.M. Calvi & A. Pavese. Conceptual design of isolation systems for bridge structures, Journal of Earthquake Engineering 1(1): 193-218; 1997.
19. G.M. Calvi. A displacement-based approach for vulnerability evaluation of classes of buildings, Journal of Earthquake Engineering 3(3): 411-438; 1999.
20. G.M. Calvi, A. Pavese & A. Rasulo. Experimental studies on hollow bridge piers, Proceedings of the 3rd UK-Japan Workshop on Implications of Recent Earthquakes on Seismic Risk, London; 2000.
21. G.M. Calvi & A. Pavese. Experimental and numerical studies on the seismic response of reinforced concrete hollow bridge piers, CAFEEL-ECOEST2/ICONS Report 2, Seismic assessment, strengthening and repair of structures, LNEC, Lisbon, Portugal; 2003.
22. CEB. RC elements under cyclic loading, State of the art report, Comité Euro-International du Béton, Thomas Telford; 1996.
23. CEN. Eurocode 8 – Structures in Seismic Regions - Design - Part 2: Bridges, Comité Européen de Normalisation, Brussels; 1994.
24. CEN. Eurocode 2: Design of concrete structures – Part 1: general rules and rules for buildings, Comité Européen de Normalisation, Brussels; 1999.
25. CEN. Eurocode 8: Design of structures for earthquake resistance – Part 5: foundations, retaining structures and geotechnical aspects (Final Project Team Draft), Comité Européen de Normalisation, Brussels; 2001.
26. CEN. Eurocode 8: Design of structures for earthquake resistance – Part 2: bridges (Project Team Draft - Draft No 2), Comité Européen de Normalisation, Brussels; 2002.
27. CEN. Eurocode 8: Design of structures for earthquake resistance – Part 1: general rules, seismic actions and rules for buildings (Draft No 6 – Version for translation), Comité Européen de Normalisation, Brussels; 2003a.
28. CEN. Eurocode 8: Design of structures for earthquake resistance – Part 3: strengthening and repair of buildings (Final Project Team Draft - Draft No 3), Comité Européen de Normalisation, Brussels; 2003b.
REFERENCES
293
29. F. Ceroni, G. Manfredi & M.R. Pecce. A formulation of plastic hinge length in R.C. columns, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
30. Y.H. Chai, M.J.N. Priestley & F. Seible. Seismic retrofit of circular bridge columns for enhanced flexural performance, ACI Structural Journal 88(5): 572-584; 1991.
31. Y.H. Chai. An analysis of the seismic characteristics of steel-jacketed circular bridge columns, Earthquake Engineering and Structural Dynamics 25(2): 149-161; 1996.
32. S.Y. Chang. Pseudodynamic responses of bridge columns subjected to earthquake ground motions, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
33. J.F. Chen & J.G. Teng. Shear capacity of FRP-strengthened RC beams: FRP debonding, Construction and Building Materials 17(1): 2741; 2003.
34. C.T. Cheng, J.C. Yang, Y.K. Yeh & S.E. Chen. Seismic performance of repaired hollow-bridge piers, Construction and Building Materials 17(5): 339-351; 2003.
35. H.Y. Chua, H.Y. Chang & J.L. Wu. Retrofit/repair of a damaged bridge for seismic loading, Proceedings of the 7th International Conference on Inspection, Appraisal, Repairs & Maintenance of Buildings and Structures; 2001.
36. L. Chung & S.P. Shah. Effect of loading rate on anchorage bond and beam-column joints, ACI Structural Journal 86(2): 132-142; 1989.
37. Y. Chung, C. Park & J. Park. Seismic ductility evaluation of large-scaled bridge piers by the quasi static and the pseudo-dynamic test, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
38. M. Ciampoli & P.E. Pinto. Effects of soil-structure interaction on inelastic response of bridge piers, ASCE Journal of Structural Engineering 121(5): 806-814; 1995.
39. R.W. Clough & J. Penzien. Dynamics of structures, 2nd edition, McGraw-Hill, New York; 1993.
40. CNR. Rilevamento della vulnerabilità sismica degli edifici, Istruzioni per la compilazione della scheda di I livello, Appendice N. 1 alla publicazione “Rischio sismico di edifici pubblici”, Parte I – Aspetti metodologici, Consiglio Nazionale delle Ricerche, Gruppo per la Difesa dai Terremoti, Roma; 1993.
41. H.L. Coffman, M.L. Marsch & C.B. Brown. Seismic durability of retrofitted reinforced-concrete columns, ASCE Journal of Structural Engineering 119(5): 1643-1661; 1993.
42. C. Cole & A. Belarbi. FRP jacketed reinforced concrete columns, CIES 99-13, University of Missouri – Rolla, Department of Civil Engineering, 2001.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
294
43. D. Combescure & P. Pegon. a–Operator Splitting technique for pseudo-dynamic testing, error propagation analysis, Soil Dynamics and Earthquake Engineering 16(7): 427-443; 1997.
44. A. Corsanego. Seismic vulnerability evaluations for risk assessment in Europe, Proceedings of the fourth International Conference in Seismic Zonation, EERI, Stanford, California; 1991.
45. O. Dahlblom & N.S. Ottosen. Smeared crack analysis using generalized fictitious crack model, ASCE Journal of Engineering Mechanics 116(1): 55-76; 1990.
46. X. Daudey & A. Filiatrault. Seismic evaluation and retrofit with steel jackets of reinforced concrete bridge piers detailed with lap-splices, Canadian Journal of Civil Engineering 27(1): 1-16; 2000.
47. L. De Lorenzis. A comparative study of models on confinement of concrete cylinders with FRP composites, Work No 46, Publication 01:04, Chalmers University of Technology, Göteborg; 2001.
48. S.N. Dermitzakis & S.A. Mahin. Development of substructuring techniques for on-line computer controlled seismic performance testing, Report No. UCB/EERC-85/04, Earthquake Engineering Research Center, Berkeley; 1985.
49. R. DesRoches, R.T. Leon & S. Dyke. Response modification of bridges, MAE Center Project ST-12, Mid-America Earthquake Center; 2003 (in preparation).
50. J. Donéa, G. Magonette, P. Negro, P. Pegon, A. Pinto & G. Verzeletti. Pseudodynamic capabilities of the ELSA laboratory for earthquake testing of large structures, Earthquake Spectra 12(1): 163-180; 1996.
51. D. Duthinh & M. Starnes. Strength and ductility of concrete beams reinforced with carbon FRP and steel, NISTIR 6830, National Institute for Standards and Technology; 2001.
52. A. El-Bahy, S. Kunnath, W. Stone & A. Taylor. Cumulative seismic damage of circular bridge columns: benchmark and low-cycle fatigue tests, ACI Structural Journal 96(4): 633-641; 1999a.
53. A. El-Bahy, S. Kunnath, W. Stone & A. Taylor. Cumulative seismic damage of circular bridge columns: variable amplitude tests, ACI Structural Journal 96(5): 711-719; 1999b.
54. R. Eligehausen, E.P. Popov & V.V. Bertero. Local bond stress-slip relations of deformed bars under generalized excitations, Report No. UCB/EERC-83/23, Earthquake Engineering Research Centre, University of California, Berkeley, California; 1983.
55. R. Eligehausen & G.L. Balázs. Bond and detailing, Bulletin d’Information CEB No. 217, Comité Euro-International du Béton; 1993.
REFERENCES
295
56. A.S. Elnashai & D.C. McClure. Effect of modelling assumptions and input motion characteristics on seismic design parameters of RC bridge piers, Earthquake Engineering and Structural Dynamics 25(5): 435-463; 1996.
57. A.S. Elnashai. Integration of earthquake testing, analysis and field observations for seismic performance evaluation, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
58. U. Ersoy, A.T. Tankut & R. Suleiman. Behaviour of jacketed columns, ACI Structural Journal 90(3): 288-293; 1993.
59. E. Faccioli & R. Paolucci. Engineering seismology studies for the design and analysis of bridges, European Earthquake Engineering 3: 17-28; 1990.
60. E. Faccioli. Complex site effects in earthquake ground motion, including topography, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
61. P. Fajfar & P. Gašperšic. The N2 method for the seismic damage analysis of RC buildings, Earthquake Engineering and Structural Dynamics 25(1): 31-46; 1996.
62. P. Fajfar & H. Krawinkler (eds). Conclusions and Recommendations of the Working Group on Bridge Design Issues, Proceedings of the Workshop on Seismic Design Methodologies for the Next Generation of Codes, Balkema, Rotterdam; 1997.
63. P. Fajfar, P. Gašperšic & D. Drobnic. A simplified nonlinear method for seismic damage analysis of structures, Proceedings of the Workshop on Seismic Design Methodologies for the Next Generation of Codes, Balkema, Rotterdam; 1997.
64. P. Fajfar. A nonlinear analysis method for performance based seismic design, Earthquake Spectra 16(3): 573-592; 2000.
65. M.N. Fardis & H. Khalili. Concrete encased in fibreglass-reinforced plastic, ACI Journal Proceedings 78(6): 440-446; 1981.
66. M.N. Fardis & T. Panagiotakos. Displacement-based design of RC buildings: Proposed approach and application, Proceedings of the Workshop on Seismic Design Methodologies for the Next Generation of Codes, Balkema, Rotterdam; 1997.
67. M.N. Fardis. Seismic assessment and retrofit of RC structures, Proceedings of the 11th European Conference on Earthquake Engineering; 1998.
68. M.N. Fardis & G.M. Calvi (eds). Innovative design concepts, ECOEST2 – ICONS Report No.3, LNEC, Lisboa; 2001.
69. R. Faria, N.V. Pouca & R.M. Delgado. Numerical models to predict the nonlinear behaviour of bridge piers under severe earthquake loading, Preliminary Final Report, Deliverable N. 6, Task 3, Report N. 3/2 (VAB Project), University of Porto: Porto; 2001.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
296
70. FEMA. Evaluation of earthquake damaged concrete and masonry wall buildings – basic procedures manual, FEMA 306, Federal Emergency Management Agency; 1999a.
71. FEMA. Evaluation of earthquake damaged concrete and masonry wall buildings – technical resources, FEMA 307, Federal Emergency Management Agency; 1999b.
72. FEMA. Earthquake loss estimation methodology – HAZUS®99 Technical manual, Federal Emergency Management Agency; 1999.
73. fib. Design and use of externally bonded FRP reinforcement (FRP EBR) for reinforced concrete structures, Fédération Internationale du Béton, Lausanne; 2001.
74. F. Filippou. A simple model for reinforcing bar anchorages under cyclic excitations, ASCE Journal of Structural Engineering 112(7): 1639-1659; 1986.
75. FIP. Seismic devices, Technical brochure, FIP Industriale; 2002.
76. M. Fishinger, T. Isakovic & P. Fajfar. Seismic analysis of viaduct structures – Which method to choose? Proceedings of the Workshop on Seismic Design Methodologies for the Next Generation of Codes, Balkema, Rotterdam; 1997.
77. R.G. Flesch & A. Klatzer. Earthquake resistant design of R/C bridges: State-of-the-art report, Proceedings of the 10th European Conference on Earthquake Engineering; 1995.
78. R.G. Flesch, P. Palumbo, R. Delgado, A. Pinto, F. Romanelli, A. Barbat & F Legeron. European Project ENV4-574: Advanced methods for assessing the seismic vulnerability of existing motorway bridges, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
79. P. Franchin, G. Monti & P.E. Pinto. On the accuracy of simplified methods for the analysis of isolated bridges, Earthquake Engineering and Structural Dynamics 30(3): 363-382; 2001.
80. S.A. Freeman. Development and use of capacity spectrum method, Proceedings of the 6th US National Conference on Earthquake Engineering; 1998.
81. S. Fujikura, K. Kawashima, G. Shoji, J. Zhang & H Takemura. Effect of the interlocking ties and cross ties on the dynamic strength and ductility of rectangular reinforced concrete bridge columns, JSCE Journal of Structural Mechanics and Earthquake Engineering 640(I-50): 71-88; 2000.
82. J.H. Gates, S. Mellon & G. Klein. The Whittier Narrows, California Earthquake of October 1, 1987 – damage to state highway bridges, Earthquake Spectra 4(2): 377-388; 1988.
83. G. Gazetas & G. Mylonakis. Soil-structure interaction effects on elastic and inelastic structures, Proceedings of the 4th International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics; 2001.
REFERENCES
297
84. E. Giuriani, G. Plizzari & C. Schumm. Role of stirrups and residual tensile strength of cracked concrete on bond, ASCE Journal of Structural Engineering 117(1): 1-18; 1991.
85. Y. Goto. Cracks formed in concrete around deformed tension bars, ACI Journal 68(4): 244-251; 1971.
86. A. Gravouil & A. Combescure. Multi-time-step explicit-implicit method for non-linear structural dynamics, Numerical Methods in Engineering 50(1): 199-225; 2001.
87. A. Griezic, W.D. Cook & D. Mitchel. Seismic retrofit of bridge column-footing connections, Proceedings of the 11th World Conference on Earthquake Engineering; 1996.
88. G. Grünthal (ed). European Macroseismic Scale 1998, Conseil de l’Europe, Cahiers du Centre Européen de Géodynamique et de Séismologie 15, Luxembourg; 1998.
89. J. Guedes, P. Pegon & A.V. Pinto. A Fibre/Timoshenko beam element in CASTEM 2000, JRC Special publication I.94.31; 1994.
90. P. Gulkan & M. Sozen. Inelastic response of reinforced concrete and masonry buildings, ACI Journal 71; 1974.
91. N.M. Hawkins, I. Lin & T. Ueda. Anchorage of reinforcing bars for seismic forces, ACI Structural Journal 84(5): 407-418; 1987.
92. L. Henry & S.A. Mahin. Study of buckling of longitudinal bars in reinforced concrete bridge columns, Report to the California Department of Transportation; 1999.
93. H.M. Hilber, T.J.R. Hughes & R.L. Taylor. Improved numerical dissipation for time integration algorithms in structural dynamics, Earthquake Engineering and Structural Dynamics 5(3): 283-292; 1977.
94. A. Hillerborg, M. Modéer & P.E. Petersson. Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements, Cement and Concrete Research 6(6): 773-782; 1976.
95. E.M. Hines, A. Dazio & F. Seible. Seismic performance of hollow rectangular reinforced concrete piers with highly-confined boundary elements. Phase III: Web crushing tests, Report SSRP 2002/27, Department of Structural Engineering, University of California, San Diego, California; 2002a.
96. E.M. Hines, F. Seible & M.J.N. Priestley. Seismic performance of hollow rectangular reinforced concrete piers with highly-confined boundary elements. Phase I: Flexural tests, Phase II: Shear tests, Report SSRP 99/15, Department of Structural Engineering, University of California, San Diego, California; 2002b.
97. E.M. Hines & F. Seible. Seismic performance limits of the skyway piers for the new East Bay spans of the San Francisco – Oakland Bay Bridge, Proceedings of the fib Symposium: Concrete Structures in Seismic Regions; 2003.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
298
98. P. Holzenkämpfer. Ingenieurmodelle des verbundes geklebter bewehrung für betonbauteile, Dissertation, TU Braunschweig, 1994.
99. J. Hooks, J. Siebels, F. Seible, J. Busel, M. Cress, C. Hu, F. Isley, G. Ma, E. Munley, J. Potter, J. Roberts, B. Tang & A.H. Zureick. FHWA study tour for advanced composites in bridges in Europe and Japan, U.S. Department of Transportation, Federal Highway Administration, Washington; 1997.
100. Y.D. Hose, F. Seible & M.J.N. Priestley. Strategic relocation of plastic hinges in bridge columns, Structural Systems Research Project, 97/05, University of California, San Diego, La Jolla; 1997.
101. Y. Hose, P. Silva & F. Seible. Development of a performance evaluation database for concrete bridge components and systems under simulated seismic loads, Earthquake Spectra 16(2): 413-442; 2000.
102. G.W. Housner & C.C. Thiels. Competing against time: report of the Governor’s Board of Inquiry on the 1989 Loma Prieta Earthquake, Earthquake Spectra 6(4): 681-711; 1990.
103. G.W. Housner & C.C. Thiels. The continuing challenge: report on the performance of state bridges in the Northridge Earthquake, Earthquake Spectra 11(4): 607-636; 1995.
104. L. Huang, H. Hikosaka, T. Soma & M. Satake. Seismic evaluation & retrofit of reinforced concrete tall bridge piers with hollow cross-section, Proceedings of the 7th International Conference on Structural Faults and Repair; 1997.
105. J.S. Hwang, L.H. Sheng & J.H. Gates. Practical analysis of bridges on isolation bearings with bi-linear hysteresis characteristics, Earthquake Spectra 10(4): 705-727; 1994.
106. J.S. Hwang. Evaluation of equivalent linear analysis methods of bridge isolation, ASCE Journal of Structural Engineering 122(8): 972-976; 1996.
107. R.A. Imbsen, C J. Roblee, M. Yashinsky, M.M. Berilgen & S. Toprak. Impact on highway structures, Earthquake Spectra 16(S1): 411-435; 2000.
108. T. Isakovic, M Fishinger & R. Medeot. Seismic isolation of viaducts in Slovenia, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
109. W.D. Iwan & N.C. Gates. Estimating earthquake response of simple hysteretic structures, ASCE Journal of Engineering Mechanics 105: 391-405; 1979.
110. JBDPA. Seismic retrofit design and construction guidelines for existing reinforced concrete (RC) and steel-encased reinforced concrete (SRC) buildings with FRP materials, Japan Building Disaster Prevention Association; 1999 (in Japanese).
111. JSCE. Standard specifications for design and construction of concrete structures, Japanese Society of Civil Engineering; 2001.
REFERENCES
299
112. JSCE. Recommendations for upgrading of concrete structures with use of CFRP sheet, Japanese Society of Civil Engineering – Concrete Committee; 2001.
113. M. Kahan, R.J. Gibert & P.Y. Bard. Influence of seismic waves spatial variability on bridges: a sensitivity analysis, Earthquake Engineering and Structural Dynamics 25(8): 795-814; 1996.
114. A.I. Karabinis & T.C. Rousakis. Behaviour of rectangular FRP confined concrete elements subjected to monotonic and cyclic compressive load, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
115. K. Kawashima, S. Unjoh & H. Iida. Seismic inspection and seismic strengthening of reinforced concrete bridge piers with termination of main reinforcement at mid-height, Proceedings of the 1st US Japan Workshop on Seismic Retrofit of Bridges; 1990.
116. K. Kawashima & S. Unjoh. The damage of highway bridges in the 1995 Hyogo-Ken Nambu earthquake and its impact on Japanese seismic design, Journal of Earthquake Engineering 1(3): 505-541; 1997.
117. K. Kawashima. Seismic design and retrofit of bridges, Proceedings of the 12th World Conference on Earthquake Engineering; 2000.
118. K. Kawashima, H. Une & J. Sakai. Seismic performance of hollow reinforced concrete arch ribs subjected to cyclic lateral force under varying axial load, Report No. TIT/ EERG 01-1, Tokyo Institute of Technology; 2001.
119. K. Kawashima. Kawashima Laboratory, Department of Civil Engineering Tokyo Institute of Technology, http://seismic.cv.titech.ac.jp/menu.html; 2003.
120. A. Khalifa, T. Alkhrdahi, A. Nanni & S. Lansburg. Anchorage of surface mounted FRP reinforcement, Concrete International 21(10): 49-54; 1999.
121. J. K. Kim, I. H. Kim, H. W. Lim & J.-H. Lee. Cyclic loading test of bridge pier models without seismic detailing, Proceedings of the 8th East Asia-Pacific Conference on Structural Engineering and Construction; 2001.
122. A. Der Kiureghian & A. Neuenhofer. Response spectrum method for multi-support seismic excitations, Earthquake Engineering and Structural Dynamics 21(8): 713-740; 1992.
123. M.J. Kowalsky, M.J.N. Priestley & G.A. MacRay, Displacement-based design, A methodology for seismic design applied to single degree of freedom reinforced concrete structures, Report No. SSRP – 94/16, Structural Systems Research, University of California, San Diego, La Jolla, California, October 1994.
124. M.J. Kowalsky, M.J.N. Priestley & G.A. Macrae. Displacement-based design of RC bridge columns in seismic regions, Earthquake Engineering and Structural Dynamics 24(12): 1623-1643; 1995.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
300
125. M.J. Kowalsky, M.J.N. Priestley & F. Seible. Shear and flexural behavior of lightweight concrete bridge columns in seismic regions, ACI Structural Journal 96(1): 136-148; 1999.
126. M.J. Kowalsky. Deformation limit states and implications on design of circular RC bridge columns, ASCE Journal of Structural Engineering 126(8): 869-878; 2000.
127. M.J. Kowalsky & M.J.N. Priestley. Improved analytical model for shear strength of circular reinforced concrete columns in seismic regions, ACI Structural Journal 97(3); 2000.
128. M.J. Kowalsky. A displacement-based approach for the seismic design of continuous concrete bridges, Earthquake Engineering and Structural Dynamics 31(3): 719-747; 2002.
129. H. Krawinkler. Cyclic loading histories for seismic experimentation, Earthquake Spectra 12(1): 1-12; 1996.
130. S.K. Kunnath, A. El-Bahy, A.W. Taylor & W.C. Stone. Cumulative seismic damage of reinforced concrete bridge piers, NISTIR 6075, National Institute of Standards and Technology, Gaithersburg; 1997.
131. A. Lupoi, P. Franchin, G. Monti & P.E. Pinto. Relevance of spatial variability of ground motion on seismic design of bridges, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
132. A. Lynn, J.P. Moehle, S A. Mahin & W.T. Holmes. Seismic evaluation of existing reinforced concrete building columns, Earthquake Spectra 12(4), 715-739; 1996.
133. J.B. Mander. Experimental behaviour of ductile hollow reinforced concrete columns, Proceedings of the 8th World Conference on Earthquake Engineering; 1984.
134. M. Menegotto & P.E. Pinto. Method of analysis for cyclically loaded reinforced concrete plane frames including changes in geometry and nonelastic behaviour of elements under combined normal force and bending, IABSE Symposium on Resistance and Ultimate Deformability of Structures Acted On by Well-defined Repeated Loads, Final Report, Lisbon; 1973.
135. A. Millard. Castem 2000, Guide d’ utilisation, Rapport CEA 93/007, Saclay; 1993.
136. Y.L. Mo, C.H. Jeng & S.F. Perng. Seismic shear behavior of rectangular hollow bridge columns, Structural Engineering and Mechanics 12(4): 429-448; 2001.
137. Y.L. Mo & I.C. Nien. Seismic performance of hollow high-strength concrete bridge columns, ASCE Journal of Bridge Engineering 7(6): 338-349; 2002.
138. Y.L. Mo, D.C. Wong & K. Maekawa. Seismic performance of hollow bridge columns, ACI Structural Journal 100(3): 337-348; 2003.
REFERENCES
301
139. G. Monti. Confinement: behaviour, axial capacity, ductility, lap splices, rebar buckling, Notes from fib course: Strengthening with externally bonded FRP reinforcement – behaviour, design and application; 2003.
140. G. Monti & C. Nuti. Nonlinear cyclic behaviour of reinforcing bars including buckling, ASCE Journal of Structural Engineering 118(12): 3268-3284; 1992.
141. G. Monti, C. Nuti & P.E. Pinto. Nonlinear response of bridges under multisupport excitation. ASCE Journal of Structural Engineering 122(10): 1147-1159; 1996.
142. G. Monti, N. Nisticò & S. Santini. Design equations for FRP wrapping and steel jacketing of bridge piers, Proceedings of the 11th European Conference on Earthquake Engineering; 1998.
143. G. Monti, N. Nisticò & S. Santini. Design of FRP jackets for upgrade of circular bridge piers, ASCE Journal of Composites for Construction 5(2): 94-101; 2001.
144. M.J. Moyer & M.J. Kowalsky. Influence of tension strain on buckling of reinforcement in RC bridge columns, ACI Structural Journal 100(1): 75-85; 2002.
145. I.R.M. Munro, R. Park & M.J.N. Priestley. Seismic behaviour of reinforced concrete bridge piers, Report 76-9, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand; 1976.
146. G. Mylonakis & G. Gazetas. Seismic soil-structure interaction: beneficial or detrimental? Journal of Earthquake Engineering 4(3): 277-301; 2000.
147. K. Nagaya & K. Kawashima. Effect of aspect ratio and longitudinal reinforcement diameter on seismic performance of reinforced concrete bridge columns, Report No. TIT/EERG 01-*, Tokyo Institute of Technology; 2001.
148. N. Nakashima, H. Kato & E. Takaoka. Development of real-time pseudo dynamic testing, Earthquake Engineering and Structural Dynamics; 21(1): 79-92; 1992.
149. U. Neubauer & F.S. Rostásy. Design aspects of concrete structures strengthened with externally bonded CFRP-plates, Proceedings of the 7th International Conference on Structural Faults and Repair, 2, 109-118; 1997.
150. N.M. Newmark & W.J. Rosenblueth. Fundamentals of earthquake engineering, Prentice-Hall Inc., Englewood Cliffs, N.J; 1971.
151. K.H. Ng, M.J.N. Priestley & R. Park. Seismic behaviour of circular reinforced concrete bridge piers, Report 78-14, Department of Civil Engineering, University of Canterbury, Christchurch; 1978.
152. T. Ogata & K. Osada. Seismic retrofitting of expressway bridges in Japan, Cement and Concrete Composites 22(1); 17-27: 2000.
153. Ordinanza 3274. Primi elementi in materia di criteri generali per la classificazione sismica del territorio nazionale e normative tecniche per le costruzioni in zona sismica, Ordinanza n. 3274 della Presidenza del Consiglio dei Ministri, supplemento ordinario 72 alla Gazzetta Ufficiale n° 105; 2003.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
302
154. T.B. Panagiotakos & M.N. Fardis. Deformation-controlled seismic design of RC structures, Proceedings of the 11th European Conference on Earthquake Engineering; 1998.
155. T. Panagiotakos & M.N. Fardis. Deformations of R.C. members at yielding and ultimate, ACI Structural Journal 98(2): 135-148; 2001.
156. T.B. Panagiotakos, B. Kolias & A. Plakas. Performance-based seismic design and assessment of irregular bridges, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
157. G.F. Panza, F. Romanelli & F. Vaccari. Effect on bridge seismic response of asynchronous motion at the base of bridge piers, Deliverable N. 10-11-12 Task N. 5, Report 5/1,2,3F (VAB Project), ICTP, Trieste; 2001.
158. R. Paolucci. Soil-Structure interaction effects on an instrumented building in Mexico City, European Earthquake Engineering VII(3): 33-44; 1993.
159. R. Park & T. Paulay. Reinforced concrete structures, John Wiley & Sons; 1975.
160. R. Park & M. Priestley. Ductility of square-confined concrete columns, ASCE Journal of Structural Division 108(ST4): 929-950; 1982.
161. Y.J. Park & A. H.S. Ang. Mechanistic seismic damage model for reinforced concrete, ASCE Journal of Structural Engineering 111(4): 722–739; 1985.
162. Y.J. Park, A.H.S. Ang & Y.K. Wen. Damage-limiting aseismic design of buildings, Earthquake Spectra 3(1): 1–26; 1987.
163. R. Park. Evaluation of ductility of structures and structural assemblages from laboratory testing, Bulletin of the New Zealand National Society of Earthquake Engineering 22(3): 155-166; 1989.
164. R. Park & T. Paulay. Use of interlocking spirals for transverse reinforcement in bridge columns, Strength and ductility of concrete substructures of bridges, RRU Bulletin 84(1): 77-92; 1990.
165. R. Park, M. Rodriguez & D.R. Dekker. Assessment and retrofit of a reinforced concrete bridge pier for seismic resistance, Earthquake Spectra 9(4): 781-800; 1993.
166. T. Paulay. Lapped splices in earthquake-resisting columns, ACI Journal 79(6): 458-469; 1982.
167. T. Paulay & M.J.N. Priestley. Seismic design of reinforced concrete and masonry buildings, John Wiley & Sons; 1992.
168. T. Paulay. Displacement limits for ductile systems, Earthquake Engineering and Structural Dynamics 31(3): 583-599; 2002.
169. P. Pegon. PSD testing with substructuring: the case of asynchronous motion, Special Publication No.I.96.42, EC, Joint Research Centre, ISIS; 1996a.
REFERENCES
303
170. P. Pegon. Derivation of consistent proportional viscous damping matrices, Special Publication No.I.96.49, EC, Joint Research Centre, ISIS; 1996b.
171. P. Pegon & G. Magonette. Continuous PsD testing with non-linear substructuring: recent developments for the VAB Project, Special Publication N.99.142, EC, Joint Research Centre, ISIS; 1999.
172. P. Pegon & A.V. Pinto. Pseudo-dynamic testing with substructuring at the ELSA laboratory, Earthquake Engineering and Structural Dynamics 29(7): 905-925; 2000.
173. P. Pegon & G. Magonette. Continuous PSD testing with non-linear substructuring: presentation of a stable parallel inter-field procedure, Special Publication SPI.02.167, EC, Joint Research Centre, IPSC; 2002.
174. S. Peloso. FRP seismic retrofit of square hollow section bridge piers, MSc dissertation, ROSE School, Università degli Studi di Pavia; 2003.
175. F. Perotti. Structural response to non-stationary multiple-support random excitation, Earthquake Engineering and Structural Dynamics 19(4): 513-527; 1990.
176. J. Petrovski & D. Ristic. Reversed cyclic loading test of bridge column models, Report IZIIZ 84-164, Institute of Earthquake Engineering and Engineering Seismology; 1984.
177. A.V. Pinto, G. Verzeletti, P. Negro & J. Guedes. Cyclic testing of a squat bridge pier, EUR 16247 EN, EC, Joint Research Centre, ISIS; 1994.
178. A.V. Pinto, G. Verzeletti, P. Pegon, G. Magonette, P. Negro & J. Guedes. Pseudo-dynamic testing of large-scale R/C bridges, EUR 16378 EN, EC, Joint Research Centre, ISIS; 1996.
179. A.V. Pinto, J. Molina & G. Tsionis. Cyclic test on a large-scale model of an existing short bridge pier (Warth Bridge - Pier A70), EUR 19901 EN, EC, Joint Research Centre, ISIS; 2001a.
180. A.V. Pinto, J. Molina & G. Tsionis. Cyclic test on a large-scale model of an existing tall bridge pier (Warth Bridge - Pier A40), EUR 19907 EN, EC, Joint Research Centre, ISIS; 2001b.
181. A.V. Pinto, P. Pegon, G. Magonette, J. Molina, P. Buchet & G. Tsionis. Pseudodynamic tests with non-linear substructuring and asynchronous motion of an existing RC bridge (Warth Bridge), EUR 20525 EN, EC, Joint Research Centre, IPSC; 2002a.
182. A. Pinto, P. Pegon, J. Molina & G. Tsionis. Laboratory tests on large-scale bridges, Proceedings of the 12th European Conference on Earthquake Engineering; 2002b.
183. A. Pinto, J. Molina & G. Tsionis. Cyclic tests on large-scale models of existing bridge piers with hollow cross-section, Earthquake Engineering & Structural Dynamics 32(13): 1995-2012; 2003.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
304
184. A.V. Pinto & G. Tsionis. Experimental and numerical assessment of existing RC bridges, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
185. A.V. Pinto, P. Pegon, G. Magonette & G. Tsionis. Pseudo-dynamic testing of bridges using non-linear substructuring, Earthquake Engineering & Structural Dynamics; 2004 (accepted for publication).
186. P.E. Pinto & G. Monti (eds). Vulnerability evaluation and strengthening criteria for R.C. bridges, CNR – Gruppo Nazionale per la Difesa dai Terremoti, Roma; 2000.
187. R.W. Poston, M. Diaz & J.E. Breen. Design trends for concrete bridges, ACI Structural Journal 83(1): 14-20; 1986.
188. N.V. Pouca. Simulação numérica da resposta sísmica de elementos laminares em betão armado, PhD thesis, Universidade do Porto; 2001 (in Portuguese).
189. G.H. Powel & R. Allahabadi. Seismic damage prediction by deterministic methods: concepts and procedures, Earthquake Engineering and Structural Dynamics 16(5): 719 – 734; 1988.
190. M.J.N. Priestley & R. Park. Strength and ductility of concrete bridge columns under seismic loading, ACI Structural Journal 84(1): 61-76; 1987.
191. M.J.N. Priestley. Myths and fallacies in earthquake engineering – Conflicts between design and reality, Proceedings of the Tom Paulay Symposium – Recent Developments in Lateral Force Transfer in Buildings; 1993.
192. M.J.N. Priestley, R. Verma & Y. Xiao. Seismic shear strength of reinforced concrete columns, ASCE Journal of Structural Engineering 120(8): 2310-2329; 1994a.
193. M.J.N. Priestley, F. Seible, Y. Xiao & R. Verma. Steel jacket retrofitting of reinforced concrete bridge columns for enhanced shear strength – Part 2: test results and comparison with theory, ACI Structural Journal 91(5): 537-551; 1994b.
194. M.J.N. Priestley, F. Seible & G.M. Calvi. Seismic design and retrofit of bridges, John Wiley & Sons; 1996.
195. M.J.N. Priestley & G.M. Calvi. Concepts and procedures for direct displacement-based design and assessment, Proceedings of the Workshop on Seismic Design Methodologies for the Next Generation of Codes, Balkema, Rotterdam; 1997.
196. M.J.N. Priestley. Displacement-based approaches to rational limit states design of new structures, Proceedings of the 11th European Conference on Earthquake Engineering; 1998.
197. G. Ranzo & M.J.N. Priestley. Seismic performance of large RC circular hollow collumns, Proceedings of the 12th World Conference on Earthquake Engineering; 2000.
REFERENCES
305
198. A. Rasulo, D. Bolognini, A. Pavese & G.M. Calvi. Shear behaviour of as built RC hollow bridge piers, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
199. A. M. Reinhorn. Inelastic analysis techniques in seismic evaluations, Proceedings of the Workshop on Seismic Design Methodologies for the Next Generation of Codes, Balkema, Rotterdam; 1997.
200. W.H. Robinson. Lead-rubber hysteretic bearings suitable for protecting structures during an earthquake, Earthquake Engineering and Structural Dynamics 10(4): 593-604; 1982.
201. M. Rodriguez & R. Park. Seismic load tests on reinforced concrete columns strengthened by jacketing, ACI Structural Journal 91(2): 150-159; 1994.
202. H. Saadatmanesh, M.R. Ehsani & M.W. Li. Strength and ductility of concrete columns externally reinforced with fiber composite straps, ACI Structural Journal 91(4): 434-447; 1994.
203. H. Saadatmanesh, M.R. Ehsani & L. Jin. Seismic strengthening of circular bridge piers with fiber composites, ACI Structural Journal 93(6): 639-647; 1996.
204. H. Saadatmanesh, M.R. Ehsani & M.W. Li. Seismic retrofitting of rectangular bridge columns with composite straps, Earthquake Spectra 13(2): 281-304; 1997.
205. V.E. Sagan, P. Gergely & R.N. White. Behaviour and design of noncontact lap splices subjected to repeated inelastic tensile loading, ACI Structural Journal 88(4): 420-431; 1991.
206. M.S. Saiidi, D.H. Sanders, F. Gordaninejad, F.M. Martinovic & B.A. McElhaney. Seismic retrofit of non-prismatic RC bridge columns with fibrous composites, Proceedings of the 12th World Conference on Earthquake Engineering; 2000.
207. J. Sakai & K. Kawashima. Effect of varying axial loads including a constant tension on seismic performance of reinforced concrete bridge columns, Report No. TIT/EERG 00-2, Tokyo Institute of Technology; 2000.
208. M. Samaan, A. Mirmiran & M. Shahawy. Model of concrete confined by fiber composites, ASCE Journal of Structural Engineering 124(9): 1025-1031; 1998.
209. F. Seible, M.J.N. Priestley & D. Innamorato. Earthquake retrofit of bridge columns with continuous carbon fiber jackets – Volume II, Design guidelines, Report No. ACTT-95/08, Structural Engineering, University of California, San Diego, La Jolla, California; 1995a.
210. F. Seible, M.J.N. Priestley & G. McRae. The Kobe earthquake of January 17; 1995 – Initial impressions from a quick reconnaissance, Report No. SSRP – 95/03, Structural Systems Research, University of California, San Diego, La Jolla, California; 1995b.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
306
211. A. Sextos, A. Kappos & K. Pitilakis. Effect of analysis complexity on the calculated ductility demand of R/C bridge piers, Proceedings of the 12th European Conference on Earthquake Engineering; 2002.
212. A. Sextos, A. Kappos & K. Pitilakis. Seismic response of long concrete bridges: effect of coupled ground motion variability and soil-foundation interaction, Proceedings of the fib 2003 Symposium: Concrete Structures in Seismic Regions; 2003.
213. A. Shibata & M.A. Sozen. Substitute structure method for seismic design in reinforced concrete, ASCE Journal of the Structural Division ST1: 1-18; 1976.
214. K. Shimazaki. Evaluation of structural coefficient by displacement response estimation using the equivalent linear method, Proceedings of the 12th World Conference on Earthquake Engineering; 2000.
215. P.B. Shing & S. Mahin. Cumulative experimental errors in pseudodynamic tests, Earthquake Engineering and Structural Dynamics 15(4): 409-424; 1987.
216. SIA. Actions sur les structures porteuses, Code SIA 160, Schweizer Ingenieur-und Architekten-Verein, Zurich; 1989.
217. SNZ. The design of concrete structures, NZS 3101:1995, Standards New Zealand, Wellington; 1995.
218. A. Sommer. Torsion und duktilitätsbedarf bei hochbauten unter erdbebeinwirkung, IBK Bericht Nr. 252, Birkhäuser Verlag; 2000.
219. P. Soroushian, K Obaseki, M. Rojas & H.S. Najm. Behavior of bars in dowel action against concrete cover, ACI Structural Journal 84(2): 170-176; 1987.
220. M.R. Spoelstra & G. Monti. FRP-confined concrete model, ASCE Journal of Composites for Construction 3(3): 143-150; 1999.
221. W.C. Stone & G.S. Cheok. Inelastic behavior of full-scale bridge columns subjected to cyclic loading, NIST BSS 166, Building Science Series, Center for Building Technology, National Engineering Laboratory, National Institute of Standards and Technology; 1989.
222. J.P. Stuart & G.L. Fenves. System Identification for evaluating soil-structure interaction effects in buildings from motion recordings, Earthquake Engineering and Structural Dynamics 27: 869-885; 1998.
223. J.P. Stuart, R.B. Seed & G.L. Fenves. Seismic soil-structure interaction in buildings, I: analytical methods, ASCE Journal of Geotechnical and Geoenvironmental Engineering 125(1): 26-37; 1999.
224. Y. Takahashi & H. Iemura. Inelastic seismic performance of RC tall piers with hollow section, Proceedings of the 12th World Conference on Earthquake Engineering; 2000.
REFERENCES
307
225. H. Takemura & K. Kawashima. Effect of loading hysteresis on ductility capacity of reinforced concrete bridge piers, Journal of Structural Engineering 43(A): 849-858; 1997.
226. B. Täljisten. Strengthening concrete beams for shear with CFRP sheets, Construction and Building Materials 17(1): 1526; 2003.
227. K.H. Tan. Strengthening of rectangular RC columns using FRP systems, Proceedings of the International Conference on FRP Composites in Civil Engineering; 2001.
228. T. Tassios. Properties of bond between concrete and steel under load cycles idealizing seismic actions, Bulletin d’Information CEB No. 131 Structures en Béton sous Actions Sismiques, Volume 1 – State of the Art Reports, Comité Euro-International du Béton; 1979.
229. T.P. Tassios, M. Moretti & A. Bezas. On the behaviour and ductility of reinforced concrete coupling beams of shear walls, ACI Structural Journal 93(6): 711-720; 1996.
230. A.W. Taylor, W.C. Stone & S.K. Kunnath. Performance-based specifications for the seismic design, retrofit and repair of RC bridge columns, MCEER Bulletin 11(1); 1997.
231. T. C. Triantafillou & C. P. Antonopoulos. Design of concrete flexural members strengthened in shear with FRP, ASCE Journal of Composites for Construction 4(4): 198-205; 2000.
232. P. Tsopelas, M.C. Constantinou, Y.S. Kim & S. Okamoto. Experimental study of FPS system in bridge seismic isolation, Earthquake Engineering and Structural Dynamics 25(1): 65-78; 1996.
233. M. Uemura. Reinforcement of concrete columns with CFRP, Comprehensive Composite Materials 6: 529-539; 2000.
234. F. Wang, P. Sollogoub, F. Gantenbein & M. Nedelec. Hualien experiment: seismic analysis and evaluation of the current French design practice, Proceedings of the 11th European Conference on Earthquake Engineering; 1998.
235. N. Wehbe, M.S. Saiidi & D. Sanders. Confinement of rectangular bridge columns for moderate seismic areas, National Center for Earthquake Engineering Research (NCEER) Bulletin 12(1); 1998.
236. D. Whittaker, R. Park & M.J.N. Priestley. Experimental tests on hollow circular columns for use in offshore concrete platforms, Proceedings of the 3rd Pacific Conference on Earthquake Engineering; 1987.
237. Y.L. Wong, T. Paulay & M.J.N. Priestley. Squat circular bridge piers under multi-directional seismic attack, Report 90-4, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand; 1990.
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
308
238. Y. Xiao & R. Ma. Seismic retrofit of RC circular columns using prefabricated composite jacketing, ASCE Journal of Structural Engineering 123(10): 1357-1364; 1997.
239. X.B. Yang W. Yuan, G. Chen, & L. Fan. Seismic performance assessment and retrofit of rectangular bridge piers with externally encased circular steel jackets, Proceedings of the 12th World Conference on Earthquake Engineering; 2000.
240. X. Yang, A. Nanni & G. Chen. Effect of corner radius on performance of externally bonded FRP reinforcement, Proceedings of the 5th Conference on Fibre-reinforced Plastics for Reinforced Concrete Structures; 2001.
241. Y.K. Yeh, Y.L. Mo & C.Y. Yang. Seismic performance of hollow circular bridge piers, ACI Structural Journal 98(6): 862-871; 2001.
242. Y.K. Yeh, Y.L. Mo & C.Y. Yang. Seismic performance of rectangular hollow bridge columns, ASCE Journal of Structural Engineering 128(1): 60-68; 2002a.
243. Y.K. Yeh, Y.L. Mo & C.Y. Yang. Full-scale tests on rectangular hollow bridge piers, RILEM Materials and Structures 35(246): 117-125; 2002b.
244. X.F. Yuan, L. Lam, J.G. Teng & S.T. Smith. FRP-confined RC columns under combined bending and compression: a comparative study of concrete stress-strain models, Proceedings of the International Conference on FRP Composites in Civil Engineering; 2001.
245. J.W. Wallace & J.P. Moehle. Ductility and detailing requirements of bearing wall buildings, ASCE Journal of Structural Engineering 118(6): 1625-1644; 1992.
246. F.A. Zahn, R. Park & M.J.N. Priestley. Design of reinforced bridge columns for strength and ductility, Report 86-7, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand; 1986.
247. F.A. Zahn, R. Park & M. J. N Priestley. Flexural strength and ductility of circular hollow reinforced concrete columns without confinement on inside face, ACI Structural Journal 87(2): 156-166; 1990.
248. G. Zehetmaier. Entwicklung mechanischer Endverankerungen für aufgeklebte CFK-Lamellen, Massivbau 2000: Forschung, Entwicklungen und Anwendungen, Münchner Massivbau-Seminar 2000217-232, Springer VDI Verlag, Düsseldorf; 2000.
249. Zembaty & A. Rutenberg. On the sensitivity of bridge seismic response with local soil amplification, Earthquake Engineering and Structural Dynamics 27(10): 1095-1099; 1998.
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
309
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
Table A.1. Seismic-deficient piers with solid cross-section: geometrical and mechanical
properties and deformation capacity
Reference b (m)
d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%)
uµ
Chai et al., 1991 C 0.61 0.61 6.00 2.53 0.17 20 0.177 1.4 1.5 C 0.61 0.61 6.00 2.53 0.17 0.177 3.8 5.0 Coffman et al., 1993 R 0.46 0.46 6.6 1.20 0.10 35 0.150 2.0 Park et al., 1993 R 0.46 0.46 6.5 1.30 0.20 0.085 3.0 7.0 Priestley et al., 1994 C 0.61 0.61 2.0 2.50 0.08 0.065 1.1 2.5 C 0.61 0.61 2.0 2.50 0.08 0.177 0.9 3.0 C 0.61 0.61 2.0 2.50 0.08 0.056 0.7 1.0 C 0.61 0.61 1.5 2.50 0.08 0.066 0.8 1.0 R 0.41 0.41 2.0 2.50 0.08 0.054 1.4 3.0 R 0.41 0.41 2.0 2.50 0.08 0.059 1.0 1.4 R 0.41 0.41 1.5 2.50 0.08 0.063 0.7 0.8 Saadatmanesh et al., 1996 C 0.31 5.6 2.48 0.07 20 0.167 2.4 1.5 C 0.31 5.6 2.48 0.07 0.166 5.2 4.0 Saadatmanesh et al., 1997 R 0.24 0.24 4.9 2.70 0.11 16 0.150 1.5 R 0.24 0.24 4.9 5.45 0.11 0.150 2.0 Xiao & Ma, 1997 C 0.61 4.0 2.00 0.08 20 0.050 1.8 Daudey & Filiatrault, 2000 R 0.36 0.36 2.4 3.20 0.42 26 0.040 2.0 Fujikura et al., 2000 O 0.40 0.9 2.8 0.74 0.56 0.022 2.4 O 0.40 0.9 2.8 0.74 0.91 0.023 3.2 O 0.40 0.9 2.8 0.83 0.90 0.023 4.1 Saiidi et al., 2000 R 0.14 0.14 2.5 0.80 5.4 Sakai & Kawashima, 2000 R 0.40 0.4 3.4 1.58 0.79 0.131 2.4 R 0.40 0.4 3.4 1.58 0.79 0.043 4.0 R 0.40 0.4 3.4 1.58 0.79 0.087 4.0 R 0.40 0.4 3.4 1.58 0.79 0.117 3.0 Yang et al., 2000 R 0.20 0.2 5.0 2.40 1.13 0.360 5.5 R 0.20 0.2 5.0 3.40 0.81 0.150 4.0 R 0.20 0.2 5.0 2.40 1.22 0.150 4.0 R 0.20 0.2 5.0 2.40 0.81 0.150 3.7 Kim et al., 2001 C 1.00 5.0 0.9 0.2 0.05 0.8 1.5 Bousias et al., 2002 R 0.25 0.5 6.4 0.8 0.2 0.38 2.5 R 0.25 0.5 6.4 0.8 0.2 0.38 2.8 R 0.50 0.25 3.2 0.8 0.1 0.38 4.1 R 0.50 0.25 3.2 0.8 0.1 0.35 4.4 Chang, 2002 R 0.60 0.6 5.4 2.00 0.50 0.100 2.1 R 0.60 0.6 5.4 2.00 0.50 0.100 2.2 Chung et al., 2003 C 1.20 4.0 1.0 0.1 31.6 0.1 1.7 2.3 C 1.20 4.0 1.0 0.2 31.6 0.1 2.2 3.0 C 1.20 4.0 1.0 0.1 31.6 0.1 1.1 1.9 C 1.20 4.0 1.0 0.2 31.6 0.1 2.4 3.2 Kawashima, 2003 (webpage) C 0.40 3.4 2.02 0.75 0.067 5.0
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
310
Reference b (m)
d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%)
uµ
Kawashima, 2003 (webpage) C 0.40 3.4 2.02 0.75 0.067 5.0 C 0.40 3.4 2.02 0.75 0.064 5.1 C 0.40 3.4 2.02 0.75 0.065 1.9 C 0.40 3.4 2.02 1.49 0.057 3.5 C 0.40 3.4 2.02 0.75 0.057 C 0.40 3.4 2.02 0.75 0.054 4.5 C 0.40 3.4 2.02 1.49 0.054 4.5 Kawashima, 2003 (webpage) R 0.40 0.4 4.0 0.95 0.77 15 0.044 3.3 R 0.40 0.4 4.0 0.95 0.77 30 0.043 3.3 R 0.40 0.4 4.0 0.95 0.77 15 0.044 3.7 Kawashima, 2003 (webpage) C 0.40 3.4 1.89 0.26 0.049 3.4 C 0.40 3.4 1.89 0.13 0.049 3.0
R: Rectangular cross-section, C: circular cross-section, O: oval cross-section
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
311
Table A.2. Code-designed piers with solid cross-section: geometrical and mechanical
properties and deformation capacity
Reference b (m)
d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%)
uµ
Munro et al., 1976 O 0.50 5.5 3.28 1.26 0.003 4.7 Ng et al., 1978 O 0.25 5.4 3.28 1.87 0.009 4.8 O 0.25 3.7 3.28 2.48 0.333 4.4 Ang et al., 1981 R 0.40 0.4 4.0 1.51 2.83 0.380 0.3 R 0.40 0.4 4.0 1.51 2.22 0.210 3.6 Petrovski & Ristic, 1984 C 0.31 6.2 1.83 0.63 0.051 2.6 C 0.31 6.2 1.83 0.63 0.095 1.8 C 0.31 2.9 1.83 0.63 0.055 3.5 C 0.31 2.9 1.83 0.63 0.100 2.5 Zahn et al., 1986 R 0.40 0.4 4.0 1.51 1.56 0.223 4.4 R 0.40 0.4 4.0 1.51 1.99 0.390 0.3 Stone &Cheok, 1989 C 1.52 3.0 1.99 1.49 0.071 6.2 C 0.25 3.0 1.98 1.41 0.102 1.3 C 0.25 3.0 1.98 1.41 0.212 6.0 C 0.25 6.0 1.98 0.68 0.097 5.1 C 0.25 3.0 1.98 1.41 0.101 7.2 C 0.25 3.0 1.98 1.41 0.202 6.4 C 0.25 6.0 1.98 0.68 0.106 4.5 Kawashima et al., 1990 W 1.60 0.40 6.5 1.31 0.08 0.028 2.7 5.0 W 1.60 0.40 6.5 1.31 0.08 0.030 3.5 7.0 W 1.60 0.40 6.5 1.31 0.08 0.023 2.3 7.0 Lim et al., 1990 C 0.15 7.5 5.57 1.45 0.240 7.9 C 0.15 3.8 5.57 1.45 0.240 8.0 C 0.15 3.8 5.57 1.45 0.350 8.0 Park and Paulay, 1990 R 0.40 0.6 3.0 1.88 2.17 0.100 4.8 Wong et al., 1990 C 0.40 2.0 3.20 1.42 0.190 5.0 C 0.40 2.0 3.20 0.47 0.390 2.0 C 0.40 2.0 3.20 1.42 0.390 3.2 Chai et al., 1991 C 0.61 0.61 6.00 2.53 0.17 20 0.177 2.5 3.0 C 0.61 0.61 6.00 2.53 0.17 0.177 6.0 8.0 C 0.61 0.61 6.00 2.53 0.17 20 0.177 C 0.61 0.61 6.00 2.53 0.17 20 0.177 5.3 7.0 Coffman et al., 1993 R 0.46 0.46 6.6 1.20 0.10 35 0.150 4.0 R 0.46 0.46 6.6 1.20 0.10 35 0.150 4.0 R 0.46 0.46 6.6 1.20 0.10 35 0.150 4.0 Park et al., 1993 0.46 0.46 6.5 1.30 0.20 0.085 3.0 7.0 0.46 0.46 6.5 1.30 0.20 0.085 4.0 10.0 Priestley et al., 1994a 0.61 0.61 2.0 2.50 0.08 0.059 4.4 10.0 0.61 0.61 2.0 2.50 0.08 0.173 4.1 10.0 0.61 0.61 2.0 2.50 0.08 0.051 5.5 10.0 0.61 0.61 1.5 2.50 0.08 0.065 5.2 8.0 Priestley et al., 1994b R 0.41 0.41 2.0 2.50 0.08 0.053 3.6 11.0
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
312
Reference b (m)
d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%)
uµ
Priestley et al., 1994b R 0.41 0.41 2.0 2.50 0.08 0.057 3.8 8.0 R 0.41 0.41 1.5 2.50 0.08 0.062 3.7 7.0 Saadatmanesh et al., 1996 C 0.31 5.6 2.48 0.07 20 0.159 8.2 7.0 C 0.31 5.6 2.48 0.07 20 0.158 7.1 6.0 C 0.31 5.6 2.48 0.07 0.167 7.1 6.0 Hose et al., 1997 C 0.61 6.0 2.66 0.89 0.148 8.7 Kunnath et al., 1997 C 0.31 4.5 2.04 0.94 0.094 5.0 C 0.31 4.5 2.04 0.94 0.094 11.5 C 0.31 4.5 2.04 0.94 0.086 4.2 C 0.31 4.5 2.04 0.94 0.086 5.5 C 0.31 4.5 2.04 0.94 0.086 5.5 C 0.31 4.5 2.04 0.94 0.093 5.8 C 0.31 4.5 2.04 0.94 0.093 5.8 C 0.31 4.5 2.04 0.94 0.093 4.6 C 0.31 4.5 2.04 0.94 0.101 6.0 C 0.31 4.5 2.04 0.94 0.101 7.7 C 0.31 4.5 2.04 0.94 0.101 5.9 Saadatmanesh et al., 1997 R 0.24 0.24 4.9 2.70 0.11 16 0.150 6.0 R 0.24 0.24 4.9 5.45 0.11 0.150 6.0 R 0.24 0.24 4.9 5.45 0.11 0.150 6.0 Takemura & Kawashima, 1997 R 0.40 0.4 3.5 1.58 0.57 0.027 1.5 R 0.40 0.4 3.5 1.58 0.57 0.027 1.4 R 0.40 0.4 3.5 1.58 0.57 0.029 1.9 R 0.40 0.4 3.5 1.58 0.57 0.030 3.1 R 0.40 0.4 3.5 1.58 0.57 0.027 2.4 R 0.40 0.4 3.5 1.58 0.57 0.027 3.1 Xiao & Ma, 1997 C 0.61 4.0 2.00 0.08 20 0.050 4.0 C 0.61 4.0 2.00 0.08 20 0.050 6.0 C 0.61 4.0 2.00 0.08 20 0.050 6.0 Wehbe et al., 1998 R 0.38 0.61 3.8 2.22 0.37 0.098 0.5 R 0.38 0.61 3.8 2.22 0.37 0.239 5.9 R 0.38 0.61 3.8 2.22 0.48 0.092 5.5 R 0.38 0.61 3.8 2.22 0.48 0.232 0.1 Henry & Mahin, 1999 C 0.61 4.0 1.49 0.70 0.120 5.3 C 0.61 4.0 1.49 0.35 0.060 5.3 Kowalsky et al., 1999 C 0.46 8.0 3.62 0.92 0.296 9.9 C 0.46 8.0 3.62 0.60 0.271 5.7 C 0.46 8.0 3.62 0.92 0.281 9.3 Calderone et al., 2000 C 0.61 3.0 2.73 0.89 0.091 6.8 C 0.61 8.0 2.73 0.89 0.091 12.3 C 0.61 10.0 2.73 0.89 0.091 14.6 Daudey & Filiatrault, 2000 R 0.36 0.36 2.4 3.20 0.42 26 0.040 6.0 R 0.36 0.36 2.4 3.20 0.42 26 0.040 7.0 R 0.36 0.36 2.4 3.20 0.42 26 0.040 8.0 R 0.36 0.36 2.4 3.20 0.42 26 0.040 6.0 Kawashima et al., 2000 R 0.40 0.4 4.0 1.07 0.77 0.050 3.1
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
313
Reference b (m)
d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%)
uµ
Kawashima et al., 2000 R 0.40 0.4 4.0 0.95 0.77 0.051 2.7 R 0.40 0.4 4.0 0.99 0.77 0.052 3.1 R 0.40 0.4 4.0 0.95 0.77 0.050 3.3 Lehman & Moehle, 2000 C 0.61 4.0 1.49 0.70 0.072 5.3 C 0.61 8.0 1.49 0.70 0.072 9.1 C 0.61 10.0 1.49 0.70 0.072 1.4 C 0.61 4.0 0.75 0.70 0.072 5.3 C 0.61 4.0 2.98 0.70 0.072 7.3 Saiidi et al., 2000 R 0.14 0.14 2.5 0.80 7.4 R 0.14 0.14 2.5 0.80 7.9 Yang et al., 2000 R 0.20 0.2 5.0 2.40 1.13 0.360 6.0 R 0.20 0.2 5.0 3.40 0.81 0.150 6.0 R 0.20 0.2 5.0 2.40 1.22 0.150 6.0 R 0.20 0.2 5.0 2.40 0.81 0.150 6.0 Kim et al., 2001 C 1.00 5.0 0.9 0.2 0.05 3.5 4.5 Nagaya & Kawashima, 2001 R 0.40 0.4 2.4 1.58 0.79 0.033 4.5 R 0.40 0.4 2.9 1.58 0.79 0.032 4.5 R 0.40 0.4 3.4 1.58 0.79 0.034 4.0 R 0.40 0.4 3.4 1.49 0.79 0.032 4.0 Chang, 2002 R 0.60 0.6 5.4 2.00 0.50 0.100 4.6 R 0.60 0.6 5.4 2.00 0.50 0.100 4.6 Moyer & Kowalsky, 2002 O 0.46 5.3 2.08 0.92 0.043 6.1 O 0.46 5.3 2.08 0.92 0.041 5.8 O 0.46 5.3 2.08 0.92 0.044 7.6 O 0.46 5.3 2.08 0.92 0.042 6.2 Bousias et al., 2002 R 0.25 0.5 6.4 0.8 0.2 0.37 4.1 R 0.25 0.5 6.4 0.8 0.2 0.38 5.1 R 0.25 0.5 6.4 0.8 0.2 0.34 3.8 R 0.25 0.5 6.4 0.8 0.2 0.37 4.1 R 0.50 0.25 3.2 0.8 0.1 0.35 7.2 R 0.50 0.25 3.2 0.8 0.1 0.37 7.2 R 0.50 0.25 3.2 0.8 0.1 0.37 7.5 R 0.50 0.25 3.2 0.8 0.1 0.34 7.5 Chung et al., 2003 C 1.20 4.0 1.0 0.2 0.1 5.1 6.1 C 1.20 4.0 1.0 0.2 0.1 4.7 6.6 Kawashima, 2003 (webpage) R 0.40 0.4 3.4 1.35 0.99 0.041 4.5 R 0.40 0.4 3.4 1.35 0.99 0.041 2.8 R 0.40 0.4 3.4 1.35 0.99 0.037 3.0 R 0.40 0.4 3.4 1.90 1.19 0.042 3.8 R 0.40 0.4 3.4 1.90 1.19 0.040 2.2 R 0.40 0.4 3.4 1.90 1.19 0.039 2.5 Kawashima, 2003 (webpage) R 0.40 0.4 3.4 1.27 0.79 0.034 4.0 R 0.40 0.4 3.4 1.27 0.79 0.033 3.9 R 0.40 0.4 3.4 1.27 0.79 0.034 3.5 R 0.40 0.4 3.4 1.27 0.79 0.032 3.5 R 0.40 0.4 3.4 1.27 0.79 0.038 2.8
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
314
Reference b (m)
d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%)
uµ
Kawashima, 2003 (webpage) R 0.40 0.4 3.4 1.27 0.79 0.037 3.5 Kawashima, 2003 (webpage) C 0.40 3.4 1.89 0.26 0.049 5.0 C 0.40 3.4 1.89 0.26 0.053 3.6 C 0.40 3.4 1.89 0.13 0.049 4.5 C 0.40 3.4 1.89 0.13 0.053 4.5
R: Rectangular cross-section, C: circular cross-section, O: oval cross-section, W: wall-type cross-section
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
315
Table A.3. Retrofitted piers with solid cross-section: geometrical and mechanical
properties and deformation capacity
Reference b (m) d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν uδ 1 uµ 1
Chai et al., 1991 C 0.61 0.61 0.61 2.53 0.17 20 0.177 C 0.61 0.61 0.61 2.53 0.17 20 0.177 1.8 2.0 C 0.61 0.61 0.61 2.53 0.17 0.177 C 0.61 0.61 0.61 2.53 0.17 0.177 1.7 1.6 C 0.61 0.61 0.61 2.53 0.17 20 0.177 C 0.61 0.61 0.61 2.53 0.17 20 0.177 1.4 1.4 Coffman et al., 1993 R 0.46 0.46 6.6 1.20 0.10 35 0.150 R 0.46 0.46 6.6 1.20 0.10 35 0.150 2.0 R 0.46 0.46 6.6 1.20 0.10 35 0.150 2.0 R 0.46 0.46 6.6 1.20 0.10 35 0.150 2.0 Park et al., 1993 R 0.46 0.46 6.5 1.30 0.20 0.085 R 0.46 0.46 6.5 1.30 0.20 0.085 1.0 1.0 R 0.46 0.46 6.5 1.30 0.20 0.085 1.3 Priestley et al., 1994 C 0.61 0.61 2.0 2.50 0.08 0.065 C 0.61 0.61 2.0 2.50 0.08 0.059 4.0 C 0.61 0.61 2.0 2.50 0.08 0.177 C 0.61 0.61 2.0 2.50 0.08 0.173 4.6 C 0.61 0.61 2.0 2.50 0.08 0.056 C 0.61 0.61 2.0 2.50 0.08 0.051 7.9 1.0 C 0.61 0.61 1.5 2.50 0.08 0.066 C 0.61 0.61 1.5 2.50 0.08 0.065 6.5 8.0 R 0.41 0.41 2.0 2.50 0.08 0.054 R 0.41 0.41 2.0 2.50 0.08 0.053 2.6 3.7 R 0.41 0.41 2.0 2.50 0.08 0.059 R 0.41 0.41 2.0 2.50 0.08 0.057 3.8 5.7 R 0.41 0.41 1.5 2.50 0.08 0.063 R 0.41 0.41 1.5 2.50 0.08 0.062 5.3 8.8 Saadatmanesh et al., 1996 C 0.31 5.6 2.48 0.07 20 0.167 C 0.31 5.6 2.48 0.07 20 0.159 3.5 4.7 C 0.31 5.6 2.48 0.07 20 0.158 3.2 4.0 C 0.31 5.6 2.48 0.07 0.166 C 0.31 5.6 2.48 0.07 0.167 1.5 1.5 Saadatmanesh et al., 1997 R 0.24 0.24 4.9 2.70 0.11 16 0.150 R 0.24 0.24 4.9 2.70 0.11 16 0.150 4.0 R 0.24 0.24 4.9 5.45 0.11 0.150 R 0.24 0.24 4.9 5.45 0.11 0.150 3.0 R 0.24 0.24 4.9 5.45 0.11 0.150 3.0 Xiao & Ma, 1997 C 0.61 4.0 2.00 0.08 20 0.050 C 0.61 4.0 2.00 0.08 20 0.050 2.2 C 0.61 4.0 2.00 0.08 20 0.050 3.3 C 0.61 4.0 2.00 0.08 20 0.050 3.3 Daudey & Filiatrault, 2000 R 0.36 0.36 2.4 3.20 0.42 26 0.040 R 0.36 0.36 2.4 3.20 0.42 26 0.040 3.0
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
316
Reference b (m) d (m) L/b sρ
(%) wρ
(%) sl
( bd ) ν uδ 1 uµ 1
Daudey & Filiatrault, 2000 R 0.36 0.36 2.4 3.20 0.42 26 0.040 3.5 R 0.36 0.36 2.4 3.20 0.42 26 0.040 4.0 R 0.36 0.36 2.4 3.20 0.42 26 0.040 3.0 Saiidi et al., 2000 R 0.14 0.14 2.5 0.80 R 0.14 0.14 2.5 0.80 1.4 R 0.14 0.14 2.5 0.80 1.5 Yang et al., 2000 R 0.20 0.2 5.0 2.40 1.13 0.360 R 0.20 0.2 5.0 3.40 0.81 0.150 R 0.20 0.2 5.0 2.40 1.22 0.150 R 0.20 0.2 5.0 2.40 0.81 0.150 R 0.20 0.2 5.0 2.40 1.13 0.360 1.1 R 0.20 0.2 5.0 3.40 0.81 0.150 1.5 R 0.20 0.2 5.0 2.40 1.22 0.150 1.5 R 0.20 0.2 5.0 2.40 0.81 0.150 1.6 Bousias et al., 2002 R 0.25 0.5 6.4 0.8 0.2 0.38 R 0.25 0.5 6.4 0.8 0.2 0.38 R 0.25 0.5 6.4 0.8 0.2 0.37 1.5 R 0.25 0.5 6.4 0.8 0.2 0.38 1.8 R 0.25 0.5 6.4 0.8 0.2 0.34 1.3 R 0.25 0.5 6.4 0.8 0.2 0.37 1.5 R 0.50 0.25 3.2 0.8 0.1 0.38 R 0.50 0.25 3.2 0.8 0.1 0.35 R 0.50 0.25 3.2 0.8 0.1 0.35 1.6 R 0.50 0.25 3.2 0.8 0.1 0.37 1.6 R 0.50 0.25 3.2 0.8 0.1 0.37 1.7 R 0.50 0.25 3.2 0.8 0.1 0.34 1.7 Chang, 2002 R 0.60 0.6 5.4 2.00 0.50 0.100 R 0.60 0.6 5.4 2.00 0.50 0.100 2.2 R 0.60 0.6 5.4 2.00 0.50 0.100 R 0.60 0.6 5.4 2.00 0.50 0.100 2.1 Kawashima, 2003 (webpage) C 0.40 3.4 1.89 0.26 0.049 C 0.40 3.4 1.89 0.26 0.049 1.4 C 0.40 3.4 1.89 0.26 0.053 1.1 C 0.40 3.4 1.89 0.13 0.049 C 0.40 3.4 1.89 0.13 0.049 C 0.40 3.4 1.89 0.13 0.053
R: Rectangular cross-section, C: circular cross-section, 1 retrofitted / as-built
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
317
Table A.4. Seismic-deficient piers with hollow cross-section: geometrical and mechanical
properties and deformation capacity
Reference b (m)
d (m)
ft (m)
wt (m)
L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%) uµ b
(m)
Kawashima et al., 1990 RH 0.50 0.50 5.0 2.03 0.10 0.005 2.4 8.0 0.50 0.50 5.0 2.03 0.10 0.005 2.0 6.5 0.50 0.50 5.0 2.03 0.10 0.005 2.8 8.5 0.50 0.50 5.0 2.03 0.10 0.004 2.8 8.0 0.50 0.50 9.2 2.03 0.10 0.023 3.9 4.0 0.50 0.50 9.2 2.03 0.10 0.024 2.6 6.0 0.50 0.50 9.2 2.03 0.10 0.026 2.6 6.0 Huang et al., 1997 RH 0.94 0.64 0.139 0.125 5.9 1.30 0.15 0.019 1.3 4.0 Calvi et al., 2000 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 20 0.07 1.2 0.45 0.45 0.075 0.075 3.0 1.8 0.3 20 0.15 1.2 Takahashi & Iemura, 2000 RH 0.32 0.32 0.085 4.0 1.96 0.34 0.000 5.5 0.32 0.32 0.085 4.0 1.96 0.17 0.100 3.1 0.32 0.32 0.085 4.0 1.96 0.17 0.100 3.1 0.32 0.32 0.085 4.0 1.96 0.34 0.100 3.9 0.32 0.32 0.085 4.0 1.96 0.34 0.100 3.9 0.32 0.32 0.085 2.0 1.96 0.17 0.100 1.6 0.32 0.32 0.085 2.0 1.96 0.34 0.100 3.1 Kim et al., 2001 RH 0.60 1.04 0.150 0.150 8.3 1.1 0.5 0.05 4.8 6.0 Yeh et al., 2001 CH 1.50 0.300 3.7 2.15 0.29 23 0.094 1.6 2.8 Pinto et al., 2001a R 1.02 2.74 0.21 0.170 2.4 0.40 0.09 38 0.090 1.3 3.2 Pinto et al., 2001b R 1.02 2.74 0.21 0.170 5.1 0.70 0.09 38 0.100 1.6 2.3 Pinto et al., 2002a R 1.02 2.74 0.210 0.170 2.4 0.4 0.1 38 0.09 0.9 2.3 Pinto et al., 2002b R 1.02 2.74 0.210 0.170 5.1 0.7 0.1 38 0.10 1.5 2.0 Pinto et al. 2003 Pinto & Tsionis, 2003 Rasulo et al., 2002 RH 0.45 0.45 0.075 2.0 1.10 0.13 0.150 0.8 0.45 0.45 0.075 3.0 1.80 0.25 0.150 1.0 0.45 0.45 0.075 3.0 1.10 0.13 0.150 1.1 0.45 0.45 0.075 3.0 1.10 0.13 0.075 1.9 0.45 0.45 0.075 3.0 1.10 0.13 0.150 1.2 Cheng et al., 2003 CH 1.50 0.300 3.7 2.2 0.9 0.10 6.5 9.7 CH 1.50 0.300 2.3 2.2 0.3 0.10 2.9 4.9
RH: Rectangular hollow cross-section, CH: circular hollow cross-section
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
318
Table A.5. Code-designed piers with hollow cross-section: geometrical and mechanical
properties and deformation capacity
Reference b (m)
d (m)
ft (m)
wt (m)
L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%) uµ
Mander, 1984 RH 0.75 0.75 0.120 0.120 4.3 1.6 0.8 0.10 3.5 8.0 RH 0.75 0.75 0.120 0.120 4.3 1.6 1.6 0.50 1.4 4.0 RH 0.75 0.75 0.120 0.120 4.3 1.6 1.2 0.30 3.3 8.0 RH 0.75 0.75 0.120 0.120 4.3 1.6 0.8 0.30 2.4 6.0 Whittaker et al., 1987 CH 0.80 0.100 4.0 2.3 1.1 0.13 4.2 8.0 CH 0.80 0.100 4.0 2.3 1.4 0.30 4.5 12.0 CH 0.80 0.050 4.0 2.9 1.0 0.30 1.6 3.0 CH 0.80 0.050 4.0 2.9 0.8 0.30 1.3 3.0 CH 0.80 0.050 4.0 2.9 1.0 0.30 2.0 5.0 CH 0.80 0.050 4.0 2.9 0.6 0.30 2.4 6.0 Kawashima et al., 1990 RH 0.50 0.50 5.2 1.3 0.1 0.03 7.0 RH 0.50 0.50 5.2 1.3 0.1 0.03 7.0 RH 0.50 0.50 5.2 1.3 0.1 0.03 8.0 RH 0.50 0.50 5.2 1.3 0.1 0.03 7.0 RH 0.50 0.50 5.2 1.3 0.1 0.03 8.0 Zahn et al., 1990 CH 0.40 0.094 4.5 3.6 1.1 0.08 12.4 CH 0.40 0.094 4.5 3.6 1.3 0.40 2.4 CH 0.40 0.075 4.5 4.2 1.4 0.10 7.5 CH 0.40 0.075 4.5 4.2 1.7 0.22 4.9 CH 0.40 0.055 4.5 5.4 1.9 0.12 6.6 CH 0.40 0.055 4.5 5.4 2.3 0.12 3.0 Pinto et al., 1996 RH 0.80 1.6 0.160 0.160 7.0 1.2 0.4 0.03 3.8 6.6 RH 0.80 1.6 0.160 0.160 10.5 0.6 0.5 0.02 2.5 6.0 RH 0.80 1.6 0.160 0.160 3.5 0.9 0.4 0.03 3.3 7.6 RH 0.80 1.6 0.160 0.160 3.5 1.7 0.5 0.03 5.0 9.4 RH 0.80 1.6 0.160 0.160 7.0 1.2 0.4 0.03 3.5 6.3 RH 0.80 1.6 0.160 0.160 3.5 0.5 0.5 0.03 2.3 13.1 RH 0.80 1.6 0.160 0.160 6.8 1.2 0.4 0.02 6.2 5.5 Huang et al., 1997 RH 0.94 0.64 0.139 0.125 5.9 1.3 0.2 0.02 1.8 6.0 RH 0.94 0.64 0.139 0.125 5.9 1.3 0.2 0.02 1.8 6.0 Ogata & Osada, 2000 CH 0.35 0.100 5.1 4.3 0.7 0.03 2.3 2.0 CH 0.35 0.100 5.1 4.3 0.7 0.02 3.1 3.0 CH 0.35 0.100 5.1 4.3 0.7 0.03 3.0 3.0 CH 0.35 0.100 5.1 4.3 0.7 0.03 5.4 5.0 CH 0.35 0.100 5.1 4.3 0.7 0.03 5.3 5.0 Ranzo & Priestley, 2000 CH 1.60 0.152 2.5 1.4 0.3 0.05 2.9 6.0 CH 1.52 0.139 2.5 2.2 0.3 0.05 2.5 3.5 CH 1.52 0.136 2.5 2.2 0.3 0.15 1.5 2.0 Mo et al., 2001 RH 0.50 0.50 0.120 3.6 1.9 0.3 0.09 4.9 5.3 RH 0.50 0.50 0.120 3.6 1.9 0.3 0.18 4.7 4.9 RH 0.50 0.50 0.120 3.6 1.9 0.1 0.09 4.7 4.6 RH 0.50 0.50 0.120 3.6 1.9 0.1 0.19 4.3 4.4 RH 0.50 0.50 0.120 3.0 1.9 0.1 0.14 4.2 4.2
APPENDIX A – EXPERIMENTAL DATA FOR BRIDGE PIERS
319
RH 0.50 0.50 0.120 3.0 1.9 0.1 0.26 3.4 3.7
Reference b (m)
d (m)
ft (m)
wt (m)
L/b sρ
(%) wρ
(%) sl
( bd ) ν
uδ (%) uµ
Mo et al., 2001 RH 0.50 0.50 0.120 3.6 1.9 0.3 0.09 4.6 4.6 RH 0.50 0.50 0.120 3.6 1.9 0.3 0.13 4.3 4.3 RH 0.50 0.50 0.120 3.6 1.9 0.2 0.08 4.8 4.5 RH 0.50 0.50 0.120 3.6 1.9 0.2 0.13 4.1 3.9 RH 0.50 0.50 0.120 3.0 1.9 0.2 0.11 4.6 4.4 RH 0.50 0.50 0.120 3.0 1.9 0.2 0.05 4.8 4.7 Kawashima et al., 2001 RH 0.40 0.4 0.100 0.100 3.38 2.5 1.1 0.08 4.1 RH 0.40 0.4 0.100 0.100 3.38 2.5 1.2 0.07 4.5 RH 0.40 0.4 0.100 0.100 3.38 2.5 1.1 0.15 3.5 RH 0.40 0.4 0.100 0.100 3.38 2.5 1.2 0.14 3.5 Kim et al., 2001 RH 0.60 1.04 0.150 0.150 8.3 1.1 0.5 0.05 7.5 8.5 Yeh et al., 2001 CH 1.50 0.300 3.7 2.2 0.9 0.10 5.9 9.0 CH 1.50 0.300 2.3 2.2 0.3 0.10 2.9 4.9 Yeh et al., 2002a RH 1.50 1.5 0.300 2.3 1.7 1.1 0.08 6.5 1.3 RH 1.50 1.5 0.300 3.0 1.7 0.4 0.08 4.4 8.7 RH 1.50 1.50 0.300 4.3 1.7 0.3 0.08 2.1 4.1 Yeh et al., 2002b RH 1.50 1.50 0.300 4.3 1.7 0.3 0.08 6.5 11.1 RH 1.50 1.50 0.300 3.0 1.7 0.1 0.08 4.4 8.6 RH 0.50 0.50 0.120 10.8 1.9 0.4 0.09 5.1 5.5 RH 0.50 0.50 0.120 10.8 1.9 0.4 0.18 4.9 5.3 RH 0.50 0.50 0.120 10.8 1.9 0.2 0.09 4.7 4.3 RH 0.50 0.50 0.120 10.8 1.9 0.2 0.19 4.3 3.5 Cheng et al., 2003 CH 1.50 0.300 3.7 2.2 0.9 0.10 2.7 2.1 CH 1.50 0.300 3.7 2.2 0.3 22.7 0.09 2.1 2.4 RH 1.50 0.300 2.3 1.7 0.3 0.08 2.5 4.9 CH 1.50 0.300 2.3 2.2 0.3 0.10 4.5 3.6 Mo et al., 2003 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.6 0.19 2.0 3.7 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.6 0.09 2.2 6.3 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.3 0.11 2.2 6.6 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.3 0.11 2.0 6.3 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.6 0.11 1.9 5.6 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.6 0.06 2.0 7.1 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.3 0.08 2.0 7.2 RH 0.50 0.5 0.100 0.100 4.0 5.5 0.3 0.08 2.1 7.1 Peloso, 2003 RH 0.45 0.45 0.075 0.075 2.0 1.1 0.1 0.10 4.8 8.0 RH 0.45 0.45 0.075 0.075 2.0 1.1 0.1 0.19 4.8 8.0 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 0.08 6.0 6.0 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 0.15 3.6 4.0
RH: Rectangular hollow cross-section, CH: circular hollow cross-section
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
320
Table A.6. Retrofitted piers with hollow cross-section: geometrical and mechanical
properties and deformation capacity
Reference b (m)
d (m)
ft (m)
wt (m)
L/b sρ
(%) wρ
(%) sl
( bd ) ν uδ 1 uµ 1
Huang et al., 1997 RH 0.94 0.64 0.139 0.125 5.9 1.30 0.15 0.019 RH 0.94 0.64 0.139 0.125 5.9 1.30 0.15 0.018 1.4 1.5 RH 0.94 0.64 0.139 0.125 5.9 1.30 0.15 0.016 1.4 1.5 Cheng et al., 2003 CH 1.50 0.300 3.7 2.2 0.9 0.10 CH 1.50 0.300 3.7 2.2 0.3 22.7 0.09 RH 1.50 0.300 2.3 1.7 0.3 0.08 CH 1.50 0.300 2.3 2.2 0.3 0.10 CH 1.50 0.300 3.7 2.2 0.9 0.10 0.4 0.2 CH 1.50 0.300 3.7 2.2 0.3 22.7 0.09 1.3 0.9 RH 1.50 0.300 2.3 1.7 0.3 0.08 1.3 1.4 CH 1.50 0.300 2.3 2.2 0.3 0.10 1.6 0.7 Peloso, 2003 RH 0.45 0.45 0.075 0.075 2.0 1.1 0.1 0.10 RH 0.45 0.45 0.075 0.075 2.0 1.1 0.1 0.10 2.0 RH 0.45 0.45 0.075 0.075 2.0 1.1 0.1 0.19 RH 0.45 0.45 0.075 0.075 2.0 1.1 0.1 0.19 4.4 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 0.08 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 0.08 2.5 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 0.15 RH 0.45 0.45 0.075 0.075 3.0 1.8 0.3 0.15 1.5
RH: Rectangular hollow cross-section, CH: circular hollow cross-section, 1 retrofitted / as-built
APPENDIX B – CONSTRUCTION DRAWINGS
321
APPENDIX B – CONSTRUCTION DRAWINGS
Figure B.1. Vertical reinforcement of pier A70 (side view)
Figure B.2. Horizontal reinforcement of pier A70
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
322
Figure B.3. Vertical reinforcement of pier A70 (sections A-A, B-B)
APPENDIX B – CONSTRUCTION DRAWINGS
323
Figure B.4. Vertical reinforcement of pier A70 (sections C-C, D-D)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
324
Figure B.5. Vertical reinforcement of pier A40 (side view)
APPENDIX B – CONSTRUCTION DRAWINGS
325
Figure B.6. Horizontal reinforcement of pier A40
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
326
Figure B.7. Vertical reinforcement of pier A40 (sections A-A, B-B)
APPENDIX B – CONSTRUCTION DRAWINGS
327
Figure B.8. Vertical reinforcement of pier A40 (sections C-C, D-D)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
328
APPENDIX C – PHOTOGRAPHIC DOCUMENTATION
329
APPENDIX C – PHOTOGRAPHIC DOCUMENTATION
Figure C.1. Talübergang Warth Bridge in Austria (both independent lanes are shown)
Figure C.2. General view of the tested piers A40 and A70 inside the laboratory
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
330
(a)
(b) Figure C.3. Cyclic test on the short pier A70: crack pattern of the flange (a) and the web
(b) at the end of the test
(a)
(b) Figure C.4. Cyclic test on the short pier A60: crack pattern of the flange (a) and the web
(b) at the end of the test
APPENDIX C – PHOTOGRAPHIC DOCUMENTATION
331
(a)
(b)
Figure C.5. 0.4xNE test: crack pattern of the flange (a) and the web (b) of pier A40
(a)
(b) Figure C.6. 2.0xNE test: crack pattern of the flange (a) and the web (b) of pier A70
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
332
Figure C.7. Final collapse test on the tall pier A40: crack pattern of the web
Figure C.8. Pier A40: buckling of longitudinal reinforcement at 3.5m
APPENDIX C – PHOTOGRAPHIC DOCUMENTATION
333
(a)
(b)
Figure C.9. 1.0xNE test: hysteresis loops for substructured piers A20 and A30 (a) and on-
line comparison of experimental and pre-test displacement histories (b)
SEISMIC ASSESSMENT AND RETROFIT OF EXISTING REINFORCED CONCRETE BRIDGES
334