COMPARISON OF DRILLED SHAFT STRUCTURAL RESPONSE FROM STRAIN GAGE AND SHAPEACCELARRAY (SAA) DATA

Andrew Z. Boeckmann, University of Missouri, Columbia, MO, USA, boeckmanna@missouri.edu Sarah G. Myers, Brierly Associates, Denver, CO, USA Minh Uong, University of Missouri, Columbia, MO, USA J. Erik Loehr, University of Missouri, Columbia, MO, USA

The University of Missouri performed a lateral load test program involving 32 drilled shaft load tests for the Missouri Department of Transportation (MoDOT). Tests were performed at two MoDOT geotechnical research sites, and all shafts were founded in shale. Shafts were instrumented to measure the displacement of the shaft head, and the structural response of each shaft was interpreted from strain gage data and from ShapeAccelArray (SAA) data. The SAA is a chain of rigid segments with sensors that use MEMS (microelectromechanical systems) technology to measure the tilt of each segment/joint. The measuring principle of the SAA is therefore similar to that of an inclinometer. The SAA device was placed in a casing installed in the center of each test shaft to provide a continuous record of deflection along the length of the shaft throughout each test.

Bending moment profiles were calculated from strain gage data and from SAA data. Interpretation of bending moments from measurements of strain (via strain gages) and displacement (via SAA) is a nontrivial exercise, primarily because the bending stiffness, EI, is nonlinear and greatly influenced by concrete cracking, which is difficult to predict. The procedure used for this work predicted values of bending stiffness along the length of the shaft as a function of the bending curvature, which was interpreted from the SAA data. Values for bending stiffness as a function of curvature were computed using Ensoft LPile v2012. Bending moment profiles from SAA data were typically similar to those from strain gage data. The SAA bending moment profiles were often “noisier” than the strain gage bending moment profiles, but the SAA bending moment profiles had more data points since the SAA chain was continuous from the top to the bottom of each shaft.

INTRODUCTION

A comprehensive drilled shaft load testing program for the Missouri Department of Transportation (MoDOT) was performed by the University of Missouri. Part of the program involved lateral load testing 25 drilled shafts. All shafts were instrumented with vibrating wire strain gages and ShapeArrayAccel devices, both of which can be used to interpret the structural response of the shaft. This paper describes interpretation of data from each device and presents a comparison of the observed results from each.

LOAD TESTING PROGRAM

The load testing program involved 25 drilled shafts constructed during the summer of 2010 at two MoDOT research sites, one near Frankford, which is northwest of St. Louis, and one near Warrensburg, which is southeast of Kansas City. The purpose of the research program was to develop specific design relations for MoDOT to

design drilled shafts in Missouri shale for axial and lateral loading and to establish appropriate resistance factors to achieve target reliability. Complete details regarding the drilled shafts and testing program can be found in Pierce et al. (2014) and Boeckmann et al. (2014).This paper only considers the lateral tests.

Subsurface Conditions

Both sites are founded on shale. The strength of the shale is rather variable, with uniaxial compressive strengths ranging from 3 to 100 ksf as shown in Figure 1 for the Frankford Load Test Site. The shale at Warrensburg is similarly variable, with UCS from 5 to 160 ksf and layer UCS coefficients of variation near 1.0. The shale layers at Warrensburg are 10 to 15 ft beneath the ground surface. To facilitate load transfer to the shale during the lateral load testing, some drilled shaft test pairs were re-tested after drilling several 3-in. diameter holes down to shale in front of the test shafts.

©2014 Deep Foundations Institute

141

Figure 1. Uniaxial compressive the Frankford Load Test Site.

Drilled Shafts

Test shafts at the Frankford Load T3-ft diameter and 20 to 35 ft longwere constructed using the “dry” meslurry and with a short section ocasing through the overburde(embedment of 5 ft or less). Test Warrensburg Load Test Site diameter through the clay overburdiameter in the shale. The total lWarrensburg test shafts ranged fromThe Warrensburg shafts were consimilar manner to the Frankfordmethod and with permanent casingstiff clay overburden). Casings wthrough the overburden at both sitestypical MoDOT practice. The casin

Figure 2. Loa

Reference Beams

Load Frame

F8

strength at

Test site were g. The shafts ethod without of permanent en material shafts at the were 3.5-ft

rden and 3-ft length of the m 30 to 50 ft.. nstructed in a d shafts (dry g through the

were included s since this is ngs were thin

(often 0.25-in.), and their effeduring the analysis of structura

Lateral Load Test Procedure

All tests were performed in shafts served as reactions foso that data from two shafts from a single test. A typical shown in Figure 2. Loads hydraulic center-hole jacks threadbar. Loads were transload frame consisting of steelon steel collars around the sha

In general, the lateral loadfollowed ASTM D3966 (2007)applied for each test followsequence of Procedure B fLoading. Initial readings instrumentation were recordeReadings from the dial gageevery minute. The SAtransducers, and strain gagecontinuously with data acqDetails of the instrumentaprovided in a separate section

Summary of Load Tests

Sixteen paired tests of 25 performed. Details of the testin Table 1. As explained abovthe Warrensburg Load Test Safter drilling through the overbtransfer of greater loads to the

ad test of test shafts F7 (foreground) and F8.

SAA Spools

s

F

ct was considered al data.

e

pairs so that test r one another and could be collected load test setup is were applied by on Grade 150

sferred through a H-beams bearing

aft casing stickup.

d testing program ). The loads were wing the loading for Static Excess for all of the

d prior to loading. es were recorded

AA, displacement es were recorded quisition devices.

ation system are below.

test shafts were ts are summarized ve, several tests at Site were repeated burden to facilitate e underlying shale.

F7

142

Table 1. Summary of Load Tests. “Soil removed” refers to shaft pairs that were re-tested after drilling several 3-in. holes in front of the shafts.

Test Pair Test

Date in 2012

Number of Load

Increments

Maximum Load (kips)

Maximum Displacement

(in.) Notes

F1-F2 8/23 11 306 6.0 Popping noise from F2 during 88-kip load increment.

F4-F6 8/29 11 321 7.8 Popping noise from F4 heard during 321-kip load increment.

F7-F8 8/27 13 365 7.3

W1-W2 8/1 16 321 2.7 Popping noise from W2 during 263-kip load increment.

W1-W2 Soil Removed 8/9 10 277 4.0 Popping noise from W1 during

263-kip load increment. W1-W5

Soil Removed 8/20 12 350 6.9

W3-W4 8/6 11 321 4.1 W3-W4

Soil Removal 8/7 10 292 4.3

W6-W7 7/30 8 173 1.1 Jacks readjusted several times during test. No SAA data.

W6-W7 Soil Removed 8/11 11 306 5.7

W8-W9 Soil Removed 8/15 12 350 6.5

W8-W9 7/24 20 115 0.8 Load and displacement were limited for first two tests as researchers became familiar with test equipment and procedure.

W10-W11 7/27 20 146 0.7

W10-W11 Soil Removed 8/13 9 262 5.1 Soil removal in front of W11 only

(limited rig access near W10). W12-W13

Soil Removed 8/18 8 233 8.0

W14-W15 Soil Removed 8/17 9 262 6.1

INSTRUMENTATION

All test shafts were instrumented with six levels of four strain gages and a SAA casing extending nearly the full depth of the shaft. In addition, vibrating wire displacement transducers and dial gages were used to measure the shaft head displacement, and the load applied to each shaft was measured by monitoring the hydraulic pressure supplied to the jacks. Data acquisition varied by device; strain gages and displacement transducers were read with a DataTaker 85g, which is equipped for vibrating wire measurements. Each SAA has its own data acquisition hardware that connects between the SAA device and a computer used for communication with the SAA and data storage.

Vibrating Wire Strain Gages

Each shaft was instrumented with six levels of vibrating wire strain gages. The strain gages were vibrating wire concrete embedment gages (Geokon Model 4200) attached to u-brackets welded inside the reinforcing cage. Most shafts were instrumented with four gages at each level (24 gages total per shaft), but six shafts at Warrensburg were post-grouted shafts and were instrumented with just two gages at each level (12 gages total per shaft).

ShapeAccelArray (SAA)

The SAA is a chain of rigid segments with sensors that use MEMS (microelectro-mechanical systems) technology to measure the

143

tilt of each segment/joint (MeasuThe SAAs used consist of a segments, each 19.7 in. (500 mmThe measuring principle of the SAAthat of inclinometers, and both provprecision and accuracy, but the SAcontinuous record of shaft deflectiotesting. This presents safety, reliabadvantages compared with takreadings with the conventional incthe end of each load step. Both devreuseable.

Placement of the SAA on top of theduring testing is shown in Figure 3use the SAA, a 1-in. ID Schedule 40was grouted inside the inclinometehad been installed in the drilled sconstruction (during the planning research, it was anticipated that inclinometers would be used). A leof cement and water was used to fispace between the inclinometer PVC, and sand was poured in thannular space after the grout curedFigure 4. SAA Data were recordedusing the SAA software program.

Figure 3. SAA atop shaft during te

SAA Joint SASe

urand, 2012). chain of 30 m) in length. A is similar to vide excellent AA provides a on throughout bility, and time king manual clinometers at vices are also

e drilled shaft 3. In order to 0 PVC casing er casing that shafts during phase of the conventional

an neat grout ill the annular

casing and he remaining , as shown in continuously

esting.

Figure 4. SAA casing inscasing with annular space fisand.

INTERPRETATION OF DATA

Determining the structural reshafts from strain gage andapplication of bending mechais complicated by the bendwhich is nonlinear and greaconcrete cracking. The procwas to predict values of bendthe length of the shaft as bending curvature, which wathe SAA displacement data. Vstiffness as a function ofcomputed using Ensoft L-predicted values of bending sto calculate bending moment data and from strain gage dasteps is explained in more deta

Displacement, Cross-sectionBending Curvature Profiles f

Measurand software was usedata from the SAA accelerometers) to positionsegment. Details of this calcare described by MeaSubsequent calculations descperformed by the research tefrom the software is in threethe cross section of the shahorizontal and axes and tPosition data for the axis wafor all segments, an indicnegligible.

AA egments

side inclinometer illed by grout and

A

esponse of drilled SAA data is an

anics formulae that ding stiffness, , atly influenced by cedure used here ing stiffness along a function of the s interpreted from Values for bending f curvature were Pile v2012. The tiffness were used profiles from SAA

ata. Each of these ail below.

nal Rotation, and from SAA

ed to convert raw devices (MEMS data for each

culation procedure asurand (2013). cribed below were eam. Position data e dimensions, with aft defined on the the vertical axis. as nearly constant

cation noise was

144

Using Eq. 1, transverse displacemthe end of each load step were cadifferential movements in the and

where

= total transverse displacementsegment of the SAA

= position of the segment alo = initial (zero load) position of th

segment along -axis = position of the segment alo = initial (zero load) position of th

segment along the -axis

The displacement profiles interpreSAA data were differentiated accorto calculate profiles of cross-sectionthe shaft: tan

where

= cross-sectional rotation of theof the SAA, radians

= SAA segment length (500 mm

The rotation profiles interpreted frdata were differentiated in turn profiles of bending curvature of thEq. 3:

where

= bending curvature of the sthe SAA, radians per unit leng

= radius of curvature of the sthe SAA, consistent units of le

Bending Stiffness

L-Pile was used to estimate aloof the shaft from the values curvature. Bending curvature above from SAA data. It can also from strain gage data as the slopeprofile between gages along thesection. The routine employed documented in the program’s tech(Isenhower & Wang, 2011). In sumiterates the location of the neutral aequilibrium is satisfied, accountingcracking. Cracking of the concrete

ent values at alculated from

directions:

Eq. 1

t of the

ong -axis he

ong -axis he

eted from the rding to Eq. 2 nal rotation of

Eq. 2

e segment

m = 19.7 in.)

rom the SAA to calculate

he shaft with

Eq. 3

egment of gth segment of ength

ng the length of bending

was defined be estimated

e of the strain e shaft cross by L-Pile is

hnical manual mmary, L-Pile axis until force for concrete e is predicted

as a function of the compressconcrete, which was estimateshaft concrete compression te

An example of the shaft predicted by L-Pile as a functcurvature is shown in Figurestiffness decreases abruptly acurvature, which correspondsof the concrete. After the codecrease in stiffness is morsteel yields.

Figure 5. Example bendingfor shaft at Warrensburg Loa

Bending Moment

The estimated bending stiffnesused to calculate two promoment along the length of ththe SAA data and the otherdata using Eqs. 4 and 5. ·where

= bending moment of the = shaft bending stiffness a

segment ·where

= bending moment at the lstrain gage

= bending strain measured = shaft bending stiffness a

= distance from compressto neutral axis at straneutral axis is predicted

sive strength of the ed here from test sts.

bending stiffness tion of the bending e 5. The bending at small values of

s to initial cracking oncrete cracks, the re gradual as the

g stiffness curve ad Test Site (W7).

ss from L-Pile was ofiles of bending he shaft, one from

from strain gage

Eq. 4

SAA segment at the SAA

Eq. 5

level of the

d by the gage at the gage ive edge of shaft

ain gage, where by L-Pile analysis

145

For all calculations, the bending stiffness at a particular depth was limited to its minimum historic value if the L-Pile analysis indicated concrete cracking had ever occurred at that depth. For example, if the curvature at a depth of 10 ft was great enough to initiate concrete cracking for an applied lateral load of 100 kips, the bending stiffness used to calculate the bending moment for subsequent loads would be limited to a maximum value of the stiffness calculated for the 100-kip load, even if the bending curvature at a subsequent load was less than it had been at a load of 100 kips. This reasoning was extended to the shafts at Warrensburg that were tested twice, once before drilling out soil between shafts and once after.

RESULTS

The data interpretation procedures explained in the previous section were applied to data from all 25 test shafts to produce displacement and bending moment profiles (i.e. with depth). Example results are presented and discussed below. Complete results are presented in Boeckmann (2014).

SAA Displacement Profiles

Profiles with depth of transverse displacement from SAA data were calculated for all test shafts for all load steps according to the explanation presented above. An example result is included in Figure 6, showing a pair of shafts from one test side-by-side and including profiles for each load step. Similar plots for all tests can be found in Boeckmann (2014).

The shapes of the displacement profiles in Figure 6 are consistent for both shafts as well as for all load steps. Each profile indicates the respective shaft is essentially fixed around elevation 765, which is 2 to 3 ft below permanent casing. The bottom of permanent casing is approximately at the depth of overburden, so a point of fixity could be interpreted 2 to 3 ft into the shale layer. Above the shale, the shaft deflects without much curvature; the rotation in the overburden is relatively uniform.

Bending Moment Profiles

Profiles with depth of bending moment from strain gage and SAA data were calculated for all test shafts for all load steps according to the explanations presented above. Example results

are included in Figure 7 (Frankford), Figure 8 (Warrensburg), and Figure 9 (Warrensburg after Soil Removed), with each including profiles for each load step from strain gages and SAA. Similar plots for all tests can be found in Boeckmann (2014), and Table 2 includes a summary of ultimate results for each test.

Both sets of bending moment profiles for Frankford test shaft F4 (Figure 7) show similar shapes and magnitudes below the permanent casing and in the first shale layer, where the data suggest the maximum moment occurs. The bending moment profile from the SAA data shows considerably more “noise” than the strain gage profile. This could perhaps be explained by there being more SAA segments than strain gages, but the noise is better explained by having to differentiate the SAA data twice to calculate bending moment from displacement, whereas the strain gages provide a direct measurement of the bending strain.

Large negative bending moments shown near the ground surface for SAA data are difficult to explain. They are perhaps a result wind rattling the top segments, which could have been slightly loose in some casings. The negative moments do not show up in strain gage data, but there are no corresponding measurements. (Topmost moments shown are calculated from the known height of load above the ground.)

The same observations apply to the Warrensburg data before (Figure 8) and after (Figure 9) soil removal. For both tests shown, the shape and magnitudes from SAA and strain gage data are similar, but interpretation of the SAA data produces considerably more noise and unreasonable values toward the top of the shaft. Two main observations arise from comparing the results before and after soil removal: (1) the ultimate load was decreased after soil removal by about 30 kips, and (2) after soil removal, the bending moment in the soil was decreased considerably while the bending moment in the rock increased, producing a flatter bending moment profile that extends deeper than before soil removal. This indicates the drilling operation was successful for increasing the depth of load transfer.

The results in Table 2 indicate the maximum bending moment observed in each shaft was typically greater for SAA data than strain gage data.

146

Figure 6. Displacement profilesr

Figure 7. Comparison of bending

s from SAA data for test of shafts (a) W10 and (b)removal at Warrensburg test site.

g moments in test shaft F4 at Frankford test site fand (b) strain gage data.

) W11 after soil

from (a) SAA data

147

Figure 8. Comparison of bendinremoval f

Figure 9. Comparison of bendiremoval f

ng moments in test shaft W3 at Warrensburg testfrom (a) SAA data and (b) strain gage data.

ing moments in test shaft W3 at Warrensburg tesfrom (a) SAA data and (b) strain gage data.

t site before soil

st site after soil

148

Table 2. Summary of ultimate drilled shaft responses.

Test Pair (Shaft 1-Shaft 2)

Maximum Load (kips)

Shaft 1 Shaft 2

Displacement at Head (in.)

Max Moment (k-ft) Displacement at Head (in.)

Max Moment (k-ft) Strain Gage SAA Strain

Gage SAA

F1-F2 306 5.4 2194 4983 6.0 1454 6399 F4-F6 321 3.5 1503 2041 7.8 2591 6249 F7-F8 365 4.8 2166 3622 7.3 3842 8463

W1-W2 321 2.0 1793 2792 2.7 3004 3371 W1-W2 Soil Removed 277 2.5 1467 2290 4.0 2655 4609

W1-W5 Soil Removed 350 5.3 2116 2657 6.9 7764 4024

W3-W4 321 3.8 3080 3699 4.1 2798 2421 W3-W4 Soil

Removal 292 4.3 1469 2997 3.0 1708 2558

W6-W7 173 0.9 1856 N/A 1.1 1295 N/A W6-W7 Soil Removed 306 5.7 800 5129 4.0 3048 3031

W8-W9 115 0.5 2245 1441 0.8 1317 1748 W8-W9 Soil Removed 350 5.5 3060 3308 6.5 3309 4732

W10-W11 146 0.7 1780 1633 0.6 455 1429 W10-W11 Soil

Removed 262 3.4 1696 2911 5.1 2047 2691

W12-W13 Soil Removed 233 3.6 N/A 2246 8.0 2550 2471

W14-W15 Soil Removed 262 4.4 2763 2749 6.1 1779 2549

CONCLUSIONS

The drilled shaft lateral load test program documented in this paper and in Boeckmann (2014) produced an extensive dataset and valuable experience comparing drilled shaft responses interpreted from strain gage data and from SAA data. Among the most important observations:

• Interpretation of data from strain gages and from SAA both involve relatively straightforward bending mechanics formulae, but application of the formulae is complicated by the need to define bending stiffness of concrete and to compute second-order derivatives.

• Ensoft L-Pile includes a convenient method for computing the bending stiffness of reinforced concrete sections as a function of bending curvature, which can be estimated from the slope of the strain profile between gages along the shaft cross section or from

the second derivative of the SAA displacement profile.

• Strain gages provide a direct estimate of bending strain (i.e., without computing derivatives), but many strain gages implementations will be limited by quantity in their ability to produce a well-defined bending moment profile. The incidence of strain gages not being installed correctly, being damaged during installation, or shorting out after installation can exacerbate this problem.

• SAA data is well-defined with depth, but interpretation of SAA data to calculate bending moments requires computing second-order derivatives that typically result in “noisier” profiles.

• Results from the load test program are largely consistent with the interpretation observations: bending moments from strain gage data and SAA data were often similar in shape and magnitude, but profiles based on SAA were noisier, especially near the top of the shaft.

149

• Maximum bending moments interpreted from SAA data were typically greater than those interpreted from strain gage data.

Both the SAA and strain gages are useful devices for measuring drilled shaft structural response during lateral load testing. Selection of SAA versus strain gage is of course dependent on needs that will vary from project to project. Cost of the SAA is “front loaded” (i.e. significant for the first shaft), but may be less than strain gages on a per shaft basis in many cases (since the device can be reused). Strain gages are sacrificial but with little upfront cost. An additional advantage of strain gages is they provide measurement of axial load whereas the SAA does not.

REFERENCES

ASTM Standard D3966, 2007. Standard Test Methods for Deep Foundations Under Lateral Load, ASTM International, West Conshohocken, PA, 2007, DOI: 10.1520/D3966-07, www.astm.org, 18 p.

Boeckmann, A.Z., Myers, S.G., Uong, M., and Loehr, J.E., 2014. Load and Resistance Factor Design of Drilled Shafts in Shale for Lateral Loading, Report to Missouri Department of Transportation, 300 p.

Geokon (2012), Model 4200 Series Vibrating Wire Strain Gages Instruction Manual, Document Rev. P, 26 p.

Isenhower, W.M. and Wang, S.-T., 2011. Technical Manual for L-Pile, Ensoft, Inc., Version 6, 214 p.

Measurand Inc., 2013. ShapeAccelArray Manual, Rev. 0, 111 p.

Pierce, M.D., Loehr, J.E., and Rosenblad, B.L., 2014. Load and Resistance Factor Design of Drilled Shafts in Shale Using SPT and TCP Measurements, Report to Missouri Department of Transportation, 102 p.

150