shear testing of stack bonded masonry
Embed Size (px)
Construction and Building Materials 18 (2004) 125132
0950-0618/04/$ - see front matter 2003 Elsevier Ltd. All rights reserved.doi:10.1016/j.conbuildmat.2003.08.018
Shear testing of stack bonded masonry
P.B. Lourenco*, J.O. Barros, J.T. Oliveira
University of Minho, Azurem, Guimaraes P-4800-058, Portugal
Received 24 April 2003; received in revised form 24 April 2003; accepted 15 August 2003
Stack bonded masonry is scarcely used in practice, except for aesthetic reasons. Nevertheless, a regular array of units allowsplacing reinforcement in the joints, which can be of major importance for masonry shell roofs, as proposed by Eladio Dieste. Inorder to contribute to the knowledge of the behavior of stack bonded masonry under shear loading, which seems not to havebeen addressed before, an experimental research program using the triplet test was carried out. The specimens incorporate alignedjoints along two orthogonal axes, filled with micro-concrete. The main results of the experimental program are here presentedand discussed. 2003 Elsevier Ltd. All rights reserved.
Keywords: Masonry; Shear testing; Coulomb failure criterion; Dilatancy
Eladio Dieste w1x, (19172000), was a well-knownUruguayan engineer who designed a significant numberof innovative curved masonry shells for roofs and walls.The orthogonal array of facing ceramic bricks providedthe location for the placement of steel reinforcement,being simultaneously low-cost, aesthetically appealingand structurally efficient. The main difficulty of usingsimilar approaches in developed countries is related tocost of formwork and manpower. Currently, severalEuropean institutions and private companies areinvolved in the development of an industrialized solutionfor short and medium span shell roofs, under Europeancontract GROW-1999-70420 ISO-BRICK. The usageof stack bonded masonry, together with joints filled withmicro-concrete is uncommon and requires adequateexperimental testing. In this paper, the behavior of suchmasonry under shear loading is characterized.The failure behavior of masonry joints under shear
with moderate pre-compression levels can be representedby the Coulomb friction law, which establishes a linear
*Corresponding author. Tel.: q351-253-510-200; fax: q351-253-510-217.
E-mail addresses: [email protected] (P.B. Lourenco),[email protected] (J.O. Barros),[email protected] (J.T. Oliveira).
relationship between the shear stress t and the normalstress s, being given by
Here, c represents the cohesion and tanf is the tangentof the friction angle of the interface between unit andmicro-concrete. For higher normal compressive stresses,the validity of the Coulomb failure is lost and crushingyshearing of the units accompanied by cracking is found.In this case, a cap model can be adopted to representfailure of the combined jointunit ensemble, see, e.g.Ref. w2x.Another relevant feature of masonry joints is the so-
called dilatancy angle c, which measures the volumechange upon shearing. The ratio between the normaldisplacement u and the shear displacement u givesn stanc, which can assume positive or negative values.Usually, the dilatancy angle is positive, but tends to zeroupon increasing shear displacement and increasing nor-mal confining stress w3x.For the purpose of characterizing the behavior of
masonry under shear, different test methods to determinethe strength parameters c, tanf and the volume param-eter tanc have been adopted by researchers, e.g. Refs.w39x, see Fig. 1.All test methods fail to reproduce an absolutely
uniform distribution of the normal and shear stresses,
126 P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
Fig. 1. Different types of shear tests: (a) couplet test; (b) van der Pluijm test w3xand (c) triplet test.
Table 1Strength of masonry components, measured on the nominal unit size(in Nymm )2
Material Compressive strength Tensile strength
Concrete 30.3 1.73Unit (X direction) 71.8 3.50Unit (Y direction) 31.8 1.76
even if the triplet test has been adopted has the standardtest in Europe, prEN 1052-3Methods of test formasonryPart 3: Determination of initial shear strengthw10x. Here, the testing method adopted is the triplet testproposed in EN 1052-4Methods of test for masonryPart 4: Determination of shear strength including dampproof course w11x, given the stacked nature of themasonry panels and the novel use of micro-concrete. Intotal, nine masonry specimens have been tested in threeseries associated with three different normal pre-com-pression load levels. Additionally, the strength values ofthe masonry units and micro-concrete used for the jointshave also been characterized.
2. Adopted materials and test set-up
2.1. Materials characterization
The clay units used in the masonry panels (producedand delivered in a single batch) have been produced
especially for the current research project. The unitdimensions are 215 mm (length), 100 mm (width) and65 mm (height). The unit is hollow with two holes of25=25 mm in order to reduce the weight of the masonryshells.The masonry joints have a thickness of 25 mm and
are to be filled with micro-concrete, made using smallaggregate size. Micro-concrete has been selected for thejoint, because in the final masonry roof shells (notconsidered here), a top screed of concrete will be usedfor structural purposes. The masonry roof shells are tobe reinforced with steel rebar in the joints, meaning thata special purpose concrete mix with a high slump isrequired. The adopted concrete composition consists of360 kg of cement, 615 kg of gravel, 1208 kg of sand,174 l of water and 3.60 l of Superplastifier Rebuilt1000ym , which returns a slump of 160 mm.3
The strength properties of the masonry componentsare given in Table 1.The compressive strength of the masonry units was
obtained according to w12x. Due to the anisotropy asso-ciated with the extrusion process and firing, the uniaxialcompression tests were carried out in two orthogonaldirections, namely along the length (X direction) andheight (Y direction) of the unit, as illustrated in Fig. 2.In order to limit the restraining effect of the machinesteel loading platens, full units have been used for
127P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
Fig. 2. Geometry of the (a) unit and (b) masonry specimens.
Fig. 3. Chronological history of the vertical load applied to panelsunder force control (s is the pre-compression level).
Fig. 4. Location of LVDTs in the panels with dimensions in mm.
testing in the X direction, while only half unit specimenswere used for testing in the Y direction. The surface ofall specimens was ground to ensure planarity of theloading faces.All tests were carried out in dry specimens. The
specimens were stored at constant temperature of 105"58C in stove, until constant mass was reached. A universaltesting machine with a maximum loading capacity of3000 kN was used in the tests. The values for thecompressive strength of the units represent the averageof eight specimens, indicating a 1:2.3 anisotropy due tothe holes in the units and the extrusion process.
The (direct) tensile strength of masonry units wasobtained carrying out direct tension tests with notchedspecimens; see Ref. w13x for details. The values for thetensile strength of the units represent, at least, theaverage of 10 specimens, indicating a 1:2.0 anisotropydue to the holes in the units and the extrusion process.The compressive strength of concrete at 28 days is
the average of four results obtained in the tests withcylindrical specimens (diameter of 150 mm and heightof 300 mm), according to the recommendations ofRILEM CPC4 w14x. The (flexural) tensile strength ofconcrete at 28 days is the average of four resultsobtained on three point bending tests with notchedconcrete beams, according to the recommendations ofRILEM FMC1 w14x.
128 P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
Fig. 5. Experimental results for series 1 in terms of: (a) shear loadvs. horizontal displacement; (b) relation between the horizontal dis-placements of the top and bottom joints.
2.2. Description of the test set-up
The specimens consist of three masonry coursessubjected to a vertical pre-compression load, see Fig.2b. The top and bottom masonry courses are kept underconstant pressure, while a horizontal load is applied inthe middle masonry course. Eventually, this course slidesproviding the value of the shear strength of the joints.Therefore, two joints are tested simultaneously.In order to define the cohesion and the friction angle
of the joints, three different pre-compression stress levelswere adopted, namely 0.2, 0.6 and 1.0 Nymm . These2
stress levels were kept constant during the completetests duration. For each pre-compression stress level,three panels were tested, resulting in a total of ninetests. The specimens were divided in three series: series1 for a pre-compression level of 0.2 Nymm (panels B12
to B3), series 2 for a pre-compression level of 0.6 Nymm (panels B4 to B6) and series 3 for a pre-compres-2
sion level of 1.0 Nymm (panels B7 to B9).2
Two horizontal rigid supports restricted the movementof the top and bottom courses of the panel. Thesesupports are pinned and cover the full height of thecourse in order to minimize any bending effect of thepanel. The horizontal and vertical loading system con-sisted of two independent actuators. The horizontalactuator was applied directly on the middle course andthe vertical actuator was applied on a steel beam, sothat the load could be evenly distributed in the panel.Initially, the vertical compressive load was applied by
means of the vertical hydraulic actuator under forcecontrol at a rate of 10 kNymin. The maximum loadingcapacity of the vertical actuator is 50 kN. Subsequently,the vertical load was kept almost constant, as shown inFig. 3, which illustrates the time history of the verticalload for each panel series. It can be observed that thevertical load was kept approximately constant duringthe test duration, with the exception of some suddenload variations due to the geometrical irregularities ofthe failure surfaces of the joints. The irregularities haveimposed a minor ascending movement of the top steelbeam, inducing the temporary increase of the verticalload during testing.After the application of the selected pre-compression
level, the horizontal load was applied by imposing smalldisplacement steps with a hydraulic actuator of a maxi-mum loading capacity equal to 250 kN. The horizontalshear load measured with a load cell was applied witha velocity of 20 mmys.The displacements of the panel were recorded with
eight linear voltage displacement transducers (LVDTs).Four LVDTs were placed on the front of the specimensto measure the displacements in the horizontal directionand three LVDTs were placed on the back of thespecimen to measure the vertical displacements. Asstated above, one LVDT was located at the horizontalactuator so that the test could be carried out underdisplacement control. The positions of the LVDTs areillustrated in Fig. 4.
3. Obtained experimental results
3.1. Series 1panels B1, B2 and B3
Fig. 5a shows the relation between the shear load andthe horizontal displacement at the joints for series 1.Here, the horizontal displacement represents an average
129P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
Fig. 6. Failure mode of the panels for series 1: (a) Panel 1; (b) Panel 2; and (c) Panel 3.
of the measurements recorded using LVDTs 14 (theaverage of LVDTs 13 measurements was averagedwith the measurement recorded in LVDT 4), placed onthe horizontal direction. Fig. 5b illustrates the relationbetween the horizontal displacements of the two hori-zontal joints. Here, the bottom joint value is the averageof the recordings from LVDTs 13 and the top jointvalue is the recording of LVDT 4.Finally, Fig. 6 represents the failure mode of the
panels. It is noted that the variation of the recordingsamongst LVDTs 13, located at the bottom joint is
usually minimal for all specimens. The sole exceptionis the case of complex failure modes that include headjoints or diagonal cracks through the units, e.g. PanelB2.The post-peak response of panel B1 could not be
recorded but the post-peak response of panels B2 andB3 could be recorded until termination of the test.Panel B1 collapse exhibits a stepped crack at the
bottom joint and minor diffused cracked at the top joint.This indicates a probable displacement at the right topsupport that leads to minor sliding of the top joint. The
130 P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
Fig. 7. Experimental results for series 2 in terms of: (a) shear loadvs. horizontal displacement and (b) relation between the horizontaldisplacements of the top and bottom joints.
Fig. 8. Experimental results for series 3 in terms of: (a) shear loadvs. horizontal displacement and (b) relation between the horizontaldisplacements of the top and bottom joints.
collapse load of this test was similar to the other twotests, which indicates that both joints effectively slide.In panel B2, the cracks deviated from the horizontal
joints and progressed through the units and head joints.This can be due to the presence of firing cracks in thejoints and deficiently filled head joints. The existenceof gravel in the crossed joint can also preclude the crackpropagation along the interface. This peculiar shape ofcollapse is stressed in Fig. 5b, where the vertical linefor panel B2 indicates that no average horizontal dis-placement in the bottom joint could be recorded. Thisalso indicates that the use of an average horizontal
displacement in Fig. 5a is debatable for this particularpanel.The results for Panel B3 in terms of force-displace-
ment diagram are coherent and follow the expectedpattern. The residual plateau found in the response isassociated with the friction of the joint. Nevertheless, asudden increase of strength is observed at an averagedisplacement of 2.0 mm. This corresponds to the suddenjump in the diagram of Fig. 5b. It is believed that thedifferent displacements in the top and bottom joint arerelated to interlocking associated with the gravel sizeand to movements in the lateral supports. This clearly
131P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
Fig. 9. Relation between shear strength and normal stress.
Fig. 10. Relation between horizontal and normal displacement forpanel B3.
demonstrates the complexity of the phenomena involvedin the failure of the triplet test.Finally, the failure modes illustrated in Fig. 6 show a
clear trend for the occurrence of stepped crack in theright side of the joints in the intersection of horizontaland vertical joints. It is believed that this stepped crackis due to the triplet test set-up, which is known toinduce rotation of the principal stresses in the joint nearthe left and right edges.
3.2. Series 2panels B4, B5 and B6
Fig. 7a shows the relation between the shear load andthe horizontal displacement at the joints for series 2.Due to the higher pre-compression level, the maximumshear load was increased. Fig. 7b illustrates the relationbetween the horizontal displacements of the two hori-zontal joints.The results for Panel B4 in terms of forcedisplace-
ment diagram are coherent and follow the expectedpattern. This panel failed simultaneously in the top and
bottom joints. The joints in Panels B5 and B6 havefailed in a similar way. Initially, only one joint has failedbut, upon reloading, a second crack developed at thenon-cracked interface. This is clearly in agreement withthe diagrams in Fig. 7b, where it can be seen that slidingin the bottom joint is not accompanied with sliding inthe top joint, until a significant relative displacement isfound. In Panel B6, it can be observed that globalsoftening occurs only once, both top and bottom cracksare formed at an average horizontal displacement equalto circa 1.3 mm (or a displacement in the bottom jointof circa 2.6 mm).
3.3. Series 3panels B7, B8 and B9
Fig. 8a shows the relation between the shear load andthe horizontal displacement at the joints for series 3.Due to the higher pre-compression level, again themaximum shear load was increased. Fig. 8b illustratesthe relation between the horizontal displacements of thetwo horizontal joints.The post-peak response of panel B7 could not be
recorded but the post-peak response of panels B8 andB9 could be recorded until termination of the test. PanelB7 exhibited a full crack of the bottom joint at a loadof 120 kN. This crack resulted in a rigid body rotationbetween the top and bottom part. Panel B8 and PanelB9 exhibited different loaddisplacement diagrams. Theresponse of Panel B9 follows the expected patternwhereas a very ductile response was found for PanelB8. The first crack occurred in the top joint as it can beseen in Fig. 8b.
3.4. Definition of the joint strength parameters
Fig. 9 shows the relation between the normal stressand the shear strength for all tests, as well as a linearregression carried out with the shear strength averagefor each series of tests. The correlation coefficient r of2
the linear regression is 0.997, which indicates an excel-lent correlation.The linear regression indicates a cohesion value c
equal to 1.39 Nymm and a tangent of the friction2
tangent tanf equal to 1.03, see equation Eq. (1). Instandard masonry, the value of the tangent of frictionangle seems to range between 0.7 and 1.2, according todifferent combinations of units and mortars w3x. Thevalue obtained can, therefore, be considered acceptable.It is also stressed that according to the European NormEN 1052-4 w11x, the characteristic value of the initialshear strength or cohesion is only 80% of the experi-mental value or 1.11 Nymm in this case.2
3.5. Evaluation of the dilatancy
The relation between the vertical and the horizontaldisplacement is termed dilatancy. This quantity measures
132 P.B. Lourenco et al. / Construction and Building Materials 18 (2004) 125132
the uplift of one unit over the other upon shearing. It isknown that the dilatancy decreases under increasing pre-compression levels. Additionally, dilatancy decreases tozero under increasing shearing displacement due to thesmoothing of the sheared surfaces.Fig. 10 illustrates the measured dilatancy for panel
B3, which is the first panel with recorded significanthorizontal displacements. The figure seems to demon-strate that a zero dilatancy is also retrieved for themicro-concrete joints adopted in this case. The completeresults for all tests are given in Ref. w15x.It is further noted that the given displacements were
calculated using the average of displacements recordedby LVDTs 1, 3 and 4 (horizontal displacement) andLVDTs 5, 6 and 7 (normal displacement). The selectedhorizontal LVDTs (1, 3 and 4) were located on theopposite side of the panel associated with verticalLVDTs (5, 6 and 7). In such a way, the horizontal andvertical displacements were measured in the samelocation.
The triplet test was used successfully to assess theshear behavior of stack bonded masonry with micro-concrete joints. Standard masonry bond is the runningbond, which results in discontinuous vertical joints. It isimportant to stress that the masonry panels studied inthis paper had continuous vertical joints, just becausethey will be used to build reinforced masonry shells.Typical failure modes have been obtained and the shearstrength seems to adequately follow Coulomb frictionlaw. Therefore, both the use of a stacked configurationand the use of micro-concrete for the joints areacceptable.The mechanical strength parameters that characterize
the interface of the joints is a cohesion c of 1.39 Nymm and a tangent of the friction angle tanf of 1.03.2
According to Ref. w11x, the characteristic value of thecohesion c is 1.11 Nymm .2
It was also found that the dilatancy of the masonrymicro-concrete joints in the stack bond configuration issimilar to standard masonry. In particular, dilatancy tendsto zero upon progressive shearing.
The present work was partially supported byGROWTH project GROW-1999-70420 ISO-BRICKfunded by European Commission.
w1x Dihel KN. Eladio Dieste reinforced brick structures. Ziegelin-dustrie Int 1991;44(10):55663.
w2x Lourenco PB, Rots JG. A multi-surface interface model for theanalysis of masonry structures. J Eng Mech ASCE1997;123(7):6608.
w3x Pluijm Rvd. Out of plane bending of masonry behaviour. Ph.D. Thesis. Eindhoven University of Technology, The Nether-lands, 1999.
w4x Copeland RE, Saxer EL. Tests on structural bond of masonrymortars to concrete block. J ACI 1964;61(11):141151.
w5x Smith BS, Carter C. Hypothesis for shear failure of brickwork.J Struct Div ASCE 1971;97(4):105562.
w6x Sinha BP, Hendry AW. Tensile strength of brick specimens.Proc Br Ceram Soc 1975;24:91100.
w7x Hamid AA, Drysdale RG, Heidebrecht AC. Shear strength ofconcrete masonry joints. J Struct Div ASCE1979;105(5):122740.
w8x Hofmann P, Stockl S. Tests on the shear-bond behaviour in thebed-joints of masonry. Mas Int 1986;9:115.
w9x Atkinson RH, Amadei BP, Saeb S, Sture S. Response ofmasonry and joints in direct shear. J Struct Eng1989;115(9):227696.
w10x CEN. European norm for methods of test for masonry Part3: Determination of initial shear strength. prEN 1052-3, 1996.
w11x CEN. European norm for methods of test for masonry Part4: Determination of shear strength including damp proofcourse. EN 1052-4, 2000.
w12x CEN. European norm for methods of test for masonry units Part 1: Determination of compressive strength. EN 772-1,2000.
w13x Almeida JC, Lourenco PB, Barros, JA. Characterization ofbrick and brickmortar interface under uniaxial tension. In:Santos FA et al. (Eds.) Proceedings of 7th InternationalSeminar on Structural Masonry, Brazil: CEFET-MG, 2002. pp.6776.
w14x RILEM. Technical recommendations for the testing and use ofconstruction materials, 1994.
w15x Oliveira JT, Lourenco PB, Barros JO. Shear testing of stackbonded masonry. Report no. 02-DECyE-10. Universidade doMinho, Portugal, 2002. p. 33.
Shear testing of stack bonded masonryIntroductionAdopted materials and test set-upMaterials characterizationDescription of the test set-up
Obtained experimental resultsSeries 1-panels B1, B2 and B3Series 2-panels B4, B5 and B6Series 3-panels B7, B8 and B9Definition of the joint strength parametersEvaluation of the dilatancy