Hysteretic Behavior of Concrete Masonry Shear Walls with Unbonded Reinforcement Alok Madan', Andrei M. Reinhorn, John B. Mander* In reinforced concrete masonry shear walls with un routed cells, the longitudinal steel reinforcing bars are not bonded (unbonded) to the neighboring masonry. As a Te- sult, the ste! strains as well as forces are gavemed by the relative displacements ofthe end anchorages and ae, there fore, dependent on the deformations ofthe entire wall. Af- ter the development of the first flexural crack atthe base, the in-plane lateral load is resisted primarily by arch action (tie-strut mechanism) due to the absence of bonding in such walls, s opposed to beam action which is the domi- nant load resisting mechanism for shear walls with bonded reinforcement. Further, the flexural cracking causes a par- tial loss of connectivity atthe base, thus, introducing, in addition to the in-plane flexural and sheas deformations, an uplift or separation of the wall fromthe foundation at one end. Repeated reversals ofthe in-plane lateral load result in ‘consecutive opening and closing of the separation at the base, Since the unbonded longitudinal reinforcement can- not resist compression effectively due to absence of brac- ing along the wall height, cyclic load reversals generate a “rocking” type of motion in the wall in some cases, thus resulting in ‘pinching’ of force-displacement hysteresis ‘curves into ‘S” shaped loops which are characteristic of this class of structural walls. Because of these peculiari- ties, the presently available analytical models for reinforced ‘masonry elements are deficient for estimating the hyster- etic behavior of masonry shear walls with ungrouted longi tudinal (vertical) reinforcement under reversed cyclic load- ing. ‘An extensive review of literature on the state-of-the-art, ‘modeling of ungrouted post-tensioned and reinforced ma- sonry shear walls (with unbonded vertical reinforcement) is presented by Madan et al. (1996), An additional review of the subsequent literature indicates that there is a need for analytical modeling and design guidelines for evalua- tion of hysteretic performance of such walls, The existing analysis techniques for reinforced masonry or concrete structures are inadequate for predicting the hysteretic re~ ‘sponse of masonry shear walls with unbonded longitudi- nal reinforcement under cyclic load reversals. Moreover, the engineering design formulations based on the avail- "Assistant Professor, Department of Civil Engineering, Indian Institute of Technology, New Deli, India * Professor, Department of Clvil Engineering State University ‘of New York, Buffalo, NY 14260, > Associate Professor, Department of Civil Engineering, State University of New York, Buffalo, NY 14260 TMS Journal July 2000 able analytical modes for reinforced masonry and conerete ate deficient for reliable estimation ofthe lateral in-plane force-displacement response parameters ofthese walls in the nonlinear range of behavior (such as strength, ductility and post-yield stiffness). More rigorous analytical models need to be developed for this purpose which account for the absence of bond between the vertical steel reinforce- ‘ment and masonry using force equilibrium and displace- ‘ment compatibility [Madan (1996), Madan etal. (1996)]. ‘In unreinforced masonry shear walls subjected toan in- plane lateral load, subsequent othe formation of a flexural crack atthe base, a part ofthe base may separate from and Uft off the foundation if the applied overturning moment exceeds the restoring moment. In the case of reinforced masonry walls with bonded reinforcement, the bonding be- tween the steel and masonry restrains this uplift. The up- lift, which is governed by external forces and inertial ef fects, causes a local increase in stresses in the region of contact berween the wall base and foundation due to re- duction in contact area. As the cantact area reduces fur- ther with progressive uplif, the stresses may become clastic resulting in anon-linear fesponse, Atthe same time, ‘he external loads produce in-plane flexural and shear de- formations in the flexible wall body. The modeling of the uplift phenomenon may be simplified by assuming that the strain deformations ar limited tothe wall base-foundation contact surface, thus, considering the wall body tobe rigid [Priestley (1991)]. However, this assumption may be erro- neous in the range where the applied overturning moment is less than the overturning capacity of the walls. fn that case, the response will be dominated by the flexural and shear deformations of the entire wal. In case of masonry shear walls reinforced with unbonded reinforcement, the problem is complicated by the fact that the reinforcing bars ‘exer tie-down or restoring forces on the wall which depend ‘on the displacements of the entre wall [Madan etal, (1996). ‘A teview of state-of-the-art methods for analyzing the re- sponse of flexible strictures rocking on a flexible founda- tion indicates that the available analytical models for pre- dicting the force-displacement envelope of such structures undergoing strain deformations along with uplift are based on simplifying idealizations such as assuming a rigid wall body on an elastic plastic fourdation (Priestley (1991)], representing the flexible foundation by distributed Winkler springs [Badawi (1989), Housner (1957)] which are effec- ‘ive only in compression (for modeling the non-linear con- tact problem) or adding a rigid body rotational degree of freedom about the center of the base [Housner (1957), 31Haroun (1980), Haroun et al. (1981), Haroun etal. (1985), ‘Natsiavas( 1988), Natsiavas etal. (1988), Yi et al. (1992)] which may be unrestrained or restrained by appropriate rotational springs or dampers or both. There is a need for rigorous analytical formulations that realistically model the non-linear stress distribution atthe ground-structure inter- face resulting from uplift of the base. ‘A micro element model is presented in this paper for predicting the nonlinear hysteretic behavior of masonry shear walls with ungrouted longitudinal reinforcement ac- ‘counting forthe interaction of flexure and uplift under lat- eral in-plane cyclic loading. An analysis procedure is de- veloped for evaluating the hystereti in-plane lateral force- displacement response of masonry shear walls with uunbonded longitudinal reinforcement subjected o reversed cyclic loading in the inelastic range of behavior. The formu- lation of the analytical procedure is completed assuming cantilever boundary conditions for the shear wall and a concentrated lateral load atthe free end. As a result, the ‘model is applicable in the present form only to such canti- fever walls. However, the model can be extended for other boundary conditions by altering the analytical formulation suitably, The proposed analytical model was used to pre- dict the experimental force-displacement hysteresis behav- jor of wall specimens tested under lateral in-plane cyclic loading, as partof the experimental evaluation in aresearch program based on investigation of in-plane behavior of ‘masonry shear walls with unbonded vertical reinforcement with oF without prestress. The present paper is part of a series based on the results ofthe research program. BACKGROUND ‘An analysis technique was presented by Madan et al. (1996) for predicting the monotonic inelastic flexural mo- ment-curvature response of masonry shear walls with unbonded vertical reinforcement. The technique was based ‘ona modified fiber element model (micro element model) in which the masonry wail is discretized into a finite number of sections along the height and each of these finite sec- tions is discretized into a finite number of fibers along the section fengtt, The modified fiber modet proposed by Madan et al. (1996) was extended in this development for analyzing the eyclic in-plane force-displacement behavior ‘of masonry shear was with unbonded vertical reinforce- ment. A brief review of the previously proposed modified fiber model for flexural behavior of masonry shear walls with unbonded vertical reinforcement (Madan etal. (1996}} is included in this section with the purpose of providing, the necessary background for further development and completeness. Consider the inelastic flexural response analysis of a reinforced concrete masonry shear wall [Figure Xa} of height 1, width b and length L under the action of an in- plane lateral load ¥, applied at the top ofthe wall with 7, number of reinforcing bars of cross-sectional areas, 10- cated at coordinates from the center of wall section [Fig- ure I(d)], where the subscript, refers tothe, ba. The axial force P(x) and the bending moment M(x) at any discrete section fat height x, from the base can be statically deter- ‘mined and their variations along the wall height are shown in Figures 1(b) and I(c) respectively. The inelastic flexural response ofthe masonry shear wall may be analyzed using common assumptions such as: (@) The strain distribution in masonry at any given section varies linearly along the depth ie. plane sections remain plane, which is a realistic repre sentation of physical behavior for shear walls with large aspect ratios (ratio of height # to length L) (b) Theuniaxial stress-strain relationships for masonry and steel are known. (©) The widths of discrete cracks in masonry are con- lered smeared along the height ofthe wall ie.a “smeared crack model” of the wall is used, The assumption does not allow the determination of local conditions (e.g. local curvatures) atthe dis- crete crack locations. However, the smeared crack ‘modeling is a widely accepted pragmatic approach that is efficient yet accurate for the purpose of evaluating average moment-curvature relationship ‘over a farge span (height in this case) and, there- fore, the global or overall flexural response of the wall, Assumption (a) implies thatthe longitudinal masonry strain ¢(x) in any section iat coordinate from the center may be caiculated as: 5, (4.9) =e) + ¥O(%) o where isthe strain ie masonry at the center of tke ith section and (x,) is the angle of inclination of the linear strain profile at the i section also known as the curvature at that section (Figure (()] Evidently, at any section 4, there are 2 unknown ma- sonry strain Variables (the masonry strain at the center of wall section ¢(x,) and the curvature &x,) of the section) and n, unknown steel strain variables (the strains & (x, y,) in each ofthe steel rebars) To solve for these unknown strain variables, two equations are available at each section from the force and moment equilibrium of the section. The force and moment equi- librium of the 7 section may be written as: Bai}= Csi) Bei) @ Mz(31) = M(x)) ~ M(x) ® where Céx;) are the internal compressive forces in masonry and T,(x) are the internal tensile forces in stel in the axialCO) tases eeeeeeee aah © foun ; Say Figure 1—Conerete Masonry Shear Wall with Distributed Vertical Reinforcement direction which are calculated as: Als) = fax). o(4))= ‘ffaed+ v0(s)]ealae) +70 (0)]oa U2 (4a) Tha)= ¥ esleaddy eatin) a) and Mx) and M(x) ar the internal in-plane resisting mo- ‘ments due to masonry and steel respectively at the section and, are calculated as: Mx) = MUgx,), OCR )J= Pista) votsdleles) +» 0G co) Miea= ¥ eulendAyeslsvys) ey ; in which, E,(€,,) and E,(£,) are the masonry and steel stress-strain relationships, respectively. In ungrouted reinforced conerete masonry shear walls (Ge. with bonded longitudinal reinforcement), the absence cof bonding between longitudinal stee! and surrounding "TMS Journal July 2000, masonry does not permit the assumption of local strain compatibility between steel and masonry at any section. In any case, however, the displacement of the anchored end of any unbonded reinforcing bar at top or bottom of the ‘wall must equal the displacement of the masonry section at that location. The lack of bond implies that the strain in the rebar remains constant along the height of the wall, i. €., o Additionally, displacement compatibility at the wall boundary requires that the total extension of the rebar over the height H between the anchorages must equal the total elongation of the masonry fiber at that location (Park and Paulay (1975)]. Thus, the component of strain in the /* rebar due to flexural deformations &,4),) may be calculated from the aforementioned compatibility criteria as follows: 4 fests y;}ae a © vi a o J[sl2)+ », o€s)]ee e+ + —__ 7 A‘where & is the original prestressing strain in the einfarc- ing bars (if such prestressing is applied) “The foregoing review makes it evident that the flexural response of concrete masonry shear walls with unbonded reinforcement is indeterminate atthe sectional level Madan etal. (1996)]. The displacement compatibility between ma- sony and reinforcing steel atthe location of end anchor- ages on the wall boundary (Equation 7] provides the nec= essary and sufficient condition for unique solution of the unknown flexural strain variables. Detailed formulation of the numerical solution technique is presented by Madan et al (1996) for obtaining the flexural moment-curvature e- sponse at each of the discrete sections along the height in a masonry shear wall with unbonded reinforcement under the action of monotonic in-plane lateral loading. The in- plane lateral force and displacement corresponding to the flexural curvatures ofthe masottry shear wall may be caleu- lated by idealizing the top ofthe wall asa free end and ihe base as a fixed end (cantilever end conditions). Thus, the lateral force ¥, applied at height H above the base of the wall equals the shear force at any section i along the height ofthe wall and may be calculated as: Vy =P (x1) =V(x3) =.= Va, Vsq) = Mela) (=x) ‘The lateral displacement at the top of the wall due to flex- tural deformations may be obtained by integrating the cur- vatures along the height as: (8) i f= [ete (H aha ® The authors wish to note here that Equation 8 is valid ‘nly fora cantilever wall witha concentrated lateral load at the free end ofthe wal. Further, Equation 9s inadmissible for any other set of boundary conditions. Thus, the ana- Iytical procedure presented in this paper is applicable only to cantilever walls with a lateral load atthe free end. For other boundary and loading conditions, the formulation will have to be modified by replacing Equations & and 9 With the appropriate statical and kinematic relationships respectively. ANALYTICAL MODELING OF HYSTERETIC FORCE-DISPLACEMENT RESPONSE 4m reinforced concrete masonry shear walls with un- routed vertical reinforcement, the formation of a flexural ctack at the base causes loss of connectivity, thus, result- ing ina separation or uplift ofthe wall base from the foun- dation atone end. In walls with grouted reinforcement, this Uplift is restrained by the bond between the vertical forcing bars (anchored into the foundation) andthe a ing concrete. The uplift ofthe base introduces yet another unknown degree of freedom to the wall in addition to the | ‘exural stain variables. Since the modified fiber modely ‘originally propased for flexural moment-curvature anal under monotonic loading [Madan eal (1996), the anah cal model needs to be expanded to account for uplif¢ cyclic load reversals forthe purpose of developing an ang sis procedure for predicting the in-plane hysteretic fr displacement behavior ofthe wall. The modified fiber ma was complemented inthe present development with ath retical model for estimating the uplift displacements, Ad tionally, cyclic constitutive models of the component me rials i, masonry and steel were implemented for analy of hysteretic response under reversed cyclic leading. Sit the proposed analysis procedure is based on a model flexural behavior, the shear deformations of the wall; ‘inherently neglected by the analysis. Modeling of Uplift In masonry walls with unbonded reinforcement, flex: ‘racking atthe base resus inaparial loss of fixity atthe thus causing a part ofthe base to uplift off the founda Under these conditions, a rigid body mation coexists witht flexural and shear response (strain deformations). Since| strains in the unbonded reinforcing bars depend on there tive displacements of the end anchorages {Park and Pal (1975), the see! trains are functions of strain deformatg as well as uplit displacements of the wall. The degree ‘which the wall undergoes rigid Body motion in comparist the strain deformations at any time of loading depends| ‘various parameters such as the overtuming moment at ‘M.(x), time rate of change of the overturning moment, aspectratio or height to length 1/L ratio, axial load P| the base, presress in the vertical unbonded rears) wall weight 7. The exact dependence of the magn rigid body motion on these parameters isa complex ki phenomenon. Intuitively, however, progressive uplift ‘wall base causes a reduction in the area of contact bet the base and foundation thus producing a stress tion in the contact region. Therefore, there exists a com ibility berween the magnitude of base uplift or rigid motion and the strains atthe base. The following as tions were made in this formulation to model the upl rocking response of the masonry shear walls with vertical reinforcement: |. Subsequent to the formation of flexural crack at base, rigid body rotation ofthe shear wall 2) occurs along with the strain deformations as of lateral loading. The instantaneous center of rigid body rotation was assumed at the point ofa of the resultant of compressive stress distribution the point of reaction C) at the base. 2 The incremental strain displacement 2, atthe wall (extreme compression fiber) can be related tothe ‘mental angle of uplift 39, between the uplifted and foundation atthe point of separation (Figure: 3. The incremental strain displacement 2, at the‘compression fiber (toe) can be averaged over a finite height H, to obtain the incremental compressive nor- ‘mal strain at that location de (x) L/2). The hei will be referred tin this study as the average height of the stressed zone and its value depends on the param- eters that govern rigid body motion which have been ‘mentioned earlier. Using the foregoing assumptions, a compatibility con- ‘is derived between the rigid body rotation @ of the wall and curvature #(3,) atthe base. The proposed com- patibility relation is based on the wall length L, contact length d between the base and foundation, distance d, of ‘centroid of the compressive stress distribution (point of reaction) at base from the extreme compression fiber (toe of | the wal), the average height, ofthe stressed zone. Refer- ring to Figuce 2, the following relation may be obtained at any stage of loading from. geometry 5 (30) Be ad (L-d) From assumption 3 above, the curvature at base ®(x,)may be related to the strain displacement at the extreme com- ay pression fiber as follow: Ske 12) 4, 2m(x;)= a ipa ‘Thus, from the above assumptions, using Equations 10, 1 and {2, the following compatibility condition ean be writ- ten between the rigid body rotation @ of the wal and base curvature ®(x,): 28-0; if ey(m.-L/2)2eq (day Hy(L-d) 29= TAEAD sala jit eglay—L/2)< eer (138) aay CMI ley, £12) < tor C30) in which, eis the cracking strain for masonry andd isthe instantaneous contact length atthe base which may be estimated a: , Se), 2° of) The instantaneous location ofthe resultant C ofthe com- pressive stress distribution at the base may be calculated from mechanics as: a aay M(xi) «sy Compressive Sires ‘Diseaution Led Figure 2: Displacement Compati ‘TMS Jounal July 2000 35in which y i.» coordinate of the point of action of com- pressive stress resultant C. Thus, the distance 2. of the Compressive stress resultant from the extreme compres- sion fiber (te ofthe wall) is obtained as: L L_ M(x) Sr F a 16) 2 2 Cy) ‘The lateral in-plane displacement (x,) due to uplift or rocking response may be estimated using small angle ap- proximation as: v(x, y) =x a7) Using assumption 1, the vertical in-plane displacement 146.3) due to uplift may be approximated as: Mex) aa) The later in-plane displacement atthe tp leeward edge (e= H,y= L/2) due to uplift or rocking response may bbe obtained from Equation 17 as: 4, =¥,(H,L!2)=6H ay) de (x9) = Ove 9) = | as) ‘The tota) /ateral in-plane displacement J is obtained by adding the component due to flexure from Equation 9 as: b= 4; +4, = [0()(H=x)de+ 0H 20) 0 ‘The vertical in-plane displacement U, (),) of the nt sec~ tion at the location of the reinforcing bar due to wall uplift may be calculated using Equation 18 as: M(x) U(y,)=(y.-y,)=9 Med ivi) =Ove- 94) [43 »| an Assuming that lower anchorage ofthe wall is stationary, the relative longitudinal displacement of the upper anchor- age with respect to the lower one duc t uplift can be ap- Proximated as the vertical in-plane displacement U, ()) above. Thus, the uniaxial longitudinal strain component £ (inthe rebar due of uplift may be obtained as: /] 2) M(x) a H ‘The total longitudinal strain inj rebar ©, (9; is obtained by adding the strain component due to flexure and pre- stressing from Equation 7 as follows: bulyy £41) =u (0s) * eels) tr M(x1) @3) . [sd Ane Fey % H 7 H in which &, is the strain due to prestress in the rebars Cyclic Constitutive Models ‘The analysis of lateral in-plane hysteretic foree~displace- ‘ment response of the masonry shear wall under cyclic load reversals requires specification of eyclic stress-strain rules for masonry and steel. A piece-wise linear stress-strain en- ‘velope shown in Figure 3(a) was specified for masonry in uniaxial compression, The compressive stress-strain enve~ lope consists of abitinear ascending part to model the de- creasing stiffness as the stress f, increases to the peak compressive strength ',, After attaining the peak value, the stress was assumed to drop linearly o a control stress- strain point. At higher strains, the stress was assumed to remain constant ata specified value untilthe crushing strain is reached. At this point, the masonry stress goes to zero and the masonry fiber is assumed to fail in compression. The tensile stress-strain envelope was assamed to be lin- carly elastic for stresses untt the cracking strength (/,) is reached. For higher tensile strains, the stress inthe fiber is assumed to be zero and tensile failure is assumed, The elastic stiffness of the tensile stress-strain envelope was assumed the same as the initial slope of the compressive stress-strain envelope. A linear stress-strain relationship ‘was assumed for the masonry fibers during unloading and reloading. The unloading branches of the stress-strain curve ‘were assumed to converge t0 a specified contro! point, on the tensile strength envelope. In any case, the stiffness or slope of the linear unloading branch was constrained to be less than the initial stiffness of the stress-strain enve~ ‘ope. The slope of the reloading brarich was assumed equal to the initial slope ofthe strength envelope. A bilinear elastic-perfectly plastic stress-strain enve- lope was defined forthe longitudinal see! reinforcing bars. “The slopes of both the unloading and reloading branches were assumed equal to the initial elastic stiffness of the envelope. Thus, acyclic elasticperfedy plate constitu tive law iusrated in Figure 3(0) was specified for steel, A differen value of yield stress may be specified for tension (j)and compression f,) forthe sake of generalization In realty, reinforcing steel exhibits significant strain hard ing, however, ste longitudinal steel yields over its entire length, the plastic strains tend tobe modest- generally less than 3 percent. Therefore, an elastic-plastic model forthe steel stress-strain behavior is justified. In any case, che proposed analysis method is siffcient}y exible to allow the implementation of more general stress-strain models. Numerical Solution of Hysteretic Response ‘A numerical solution technique was proposed by Madan ‘etal (196) based on the modified fiber model for analyzing the inelastic flexural response of concrete masonry shear walls with unbonded reinforcement under the action of a ‘monotonic in-plane lateral load. The proposed method it volves discretization of the wall into a finite number of‘COMPRESSION Strength Envelope Strain (¢,.) TENSION Strength Envelope Pl Strain (©) COMPRESSION. | by Steel Figure3—Cyclie Mate 1 Constitutive Models for Masonry and Concrete sections m along the height [Figure 1(a)]. Since the strains in the unbonded rebars do not vary along height of the wall, the internal steel force 7(x,) and moment M(x) also remain constant forall then sections, In mathematical terms, Msi =M,= SEC6s}4y¥) 65(y)) 5) The numerical technique is based on the idealization that internal action of the unbonded steel reinforcing bars from Equations 4(b), 5(b) and (6): is statically equivalent to a structural force 7, given by Tai)= t= $ésles)Ay e404) (a4) austin (24) and a structura! moment M, given by Equa 2 sl tion (25) that are unknown and depend an kinematics of the im structural masonry wall [Madan etal, (1996)] ‘TMS Journal July 2000 37