PAC RSH ia! Title no. 96-821 RAL JOURNAI ECHNICAL PAPER Towards Cyclic Load Modeling of Reinforced Concrete by F. J. Vecchio Formulations are presented by which a secant stiffess-based finite ele. ‘ment algoritom can be modifted to provide analyse capability for arbi- trary loading conditions, including reersed cyclic loads. Plastic ofiet strains are defined for concrete and reinforcement, and these are incor porated into the analysis through the use of prestrain forces. The elastic ‘components of strain are then used to define efetive secant sifness factors. To proxide a record of plastic ofits, and of maczimwm stain ‘experienced during. previous loading, strain envelopes are defined using a Mohr's circle construction. Provisional constitutive models are presented for the concrete, although further work is required in this ‘area. An analysis ofa shearceal shes the procedure to be stable and compliant and to provide reasonably accurate simulations of behavior. ‘The results ofa prot series of panel tests are used to identify the aspects of concrete modeling that are in ned of further study and refinement Keywords: ycic lous; models reinforced concrete; shear tests INTRODUCTION In the analysis of reinforced conerete structures, a number of diverse approaches have been used for material modeling, “These include plasticty-based procedures, fracture mechanics procedures, and various nonlinear elastic models, In the latter ‘case, approaches range from discrete crack models to smeared crack models, and from fixed crack models to rotating crack ‘models. Researchers working in each of these areas generally hhave been successful in producing models that yield results of acceptable accuracy for conditions of monotonic loading, ‘These efforts are well-documented in the literature." Models that provide accurate simulations of behavior under general loading conditions, and specifically under reversed cy- clic loading, are somewhat less common, For such formulations, the smeared crack approach tends to be the most favored. Ob ‘mura and Maekawa* and Sittipunt and Wood,? among others, have documented models assuming fixed eracks directions and have demonstrated good correlation to experimental results. In some fixed crack formulations, separate models have been used to model the normal stress and shear stress hysteretic behaviors, ‘This is somewhat at odds both with the test observations and «with common elasticity approaches to constitutive modeling. Stevens et al* used a more comprehensive constitutive modeling approach for concrete in a rotating smeared crack content. How- ever, the tangent stiffness procedure used by Stevens resulted in some numerical difficulties under eyclic load conditions An alternative approach previously described was also based fon the smeared rotating. crack assumption.** The procedure was based on an iterative, secant stiffness formulation treat concrete as an orthotropic material modeled according to the constitutive relations of the modified compression field theory.” ‘The secant stiffness formulation resulted in a procedure marked by excellent convergence and numerical stability char- acteristics. Correlations to experimental data, for structures subjected to monotonic loading conditions, were generally very good. However, a criticism of the secant stiffness ap- proach has been that it cannot be effectively used to model re- sponse to general loading. Presented herein are formulations whereby a secant stiff: ness-based algorithm, employing the smeared rotating crack assumption, can be adapted to represent hysteretic material re- sponse under general and reserved eyclie loading. The compu- tational advantages of the secant stiffness approach are retained without any compromise to the material behavior models that can be used. The need to separate shear and nor- mal in-plane behavior is also eliminated, RESEARCH SIGNIFICANCE The need for accurate methods of analysis of reinforced con- crete structures under general loading conditions has been brought to the fore by structural failures sustained during the recent Northridge and Kobe earthquakes. Many ofthe collapses involved structural elements subjected to lateral shears, plac- ing high strength and ductility demands on the cracked rein- forced concrete. Current ability to model response under such conditions is not well-advanced, ‘This paper presents an alternative method by which finite clement analysis procedures can be made to provide accurate simulations of reinforced conerete subjected to reversed cyclic loads. Emphasis is placed on developing simple, numerically stable formulations. The analysis ofa shearwall structure, and the results of a pilot test program, are also presented in a dis- cussion of the importance of proper constitutive modeling, PLASTIC OFFSET FORMULATION Consider the stress and strain conditions in a concrete ele- ment previously subjected to an arbitrary load history. Sup- ppose that at a particular point in time, in 2 principal strain direction, the stress and total strain in the concrete are f,, and €,. respectively, as shown in Fig. 1. The total strain can be con= sidered to consist ofan elastic strain component €/ and a plas tic strain component (or plastic offset) €?; thus c= ef set 0 ‘The elastic strain can then be used to compute an effective se- cant stiffness for the concrete @ However, the plastic strain must then be treated as a strain off set, similar to elastic offsets resulting from thermal expansion, shrinkage, or other prestrain effects Formulations were previously described in References 6 and 8, whereby elastic prestrains were rigorously incorporated into @ finite element algorithm based on a smeared, rotating crack model and a secant stiffness approach. The plastic offsets can 41 Scar og 6,802 Mach A 99 . pi is cnn Corre, Age sedge la Copies uns pemsin s ontaaed rm the opi oes. Petnan de fusion wl be puiie i he January Febuary 2080 ACY Src Jounal Fete by Sepeembe 1989FE Vecchio FAK cu pf of il ngiring othe Uni of ana Foo somber of Jt ACT ASCE Comes 4 Rnfoe Cer Chesed 47, Fre Elown tat of nfo Cnet Scher His ach tees ee one deg nln ean on reinationofraces be accommodated in an analogous manner. To do so, the plastic “offsets in the principal directions are first resolved into compo- nents relative to the reference axes, giving the vector Ce} ef, lef] =| 08, ° Yey From the prestrains, free joint displacements are determined as functions of the element geometry. Then, given the free dis- placements, the plastic prestrain riodal forces can be evaluated Using the effective element stiffness matrix due to the concrete component (sce Reference 6 for details) The plastic offsets developed in each of the reinforcement ‘components are handled in a similar manner. For a reinforce ‘ment component oriented in direction 0, having. a plastic off- set of €f, the prestrain matrix is f+ c0s24)/2 lef}, = Je? «(1 cos2a/2 @ sin2c, ’ ‘The corresponding plastic prestrain nodal forces are then de- termined accordingly. At the same time, the secant stiffness for the reinforcement must be calculated as ) where fis the current stress in the reinforcement and € is the clastic strain component (total strain net plastic strain). ‘The total nodal forces for the element, arising fro offsets, can then be calculated as the sum of the conerete inforcement contributions. These are added to prestrain forces arising from elastic prestrain effects and from nonlinear expan= sion effects (see Reference 8), The solution then proceeds ac- cording to the algorithm described in Reference 6 astic dre Envelope of concrete plastic offsets In the formulations being proposed, a critical requirement is to define and retain the plastic strains occurring in the con- crete. These strain offsets must be definable for any arbitrary direction in the concrete continuum, even as principal strain directions undergo rotation A Mohr's circle construction provides a simple means of tracking the plastic offsets [see Fig. 2b]. Given the concrete plastic strains relative to the x, yaxes (€f,, €f,, and ye?) the plastic strains in the directions corresponding, to the prin ‘axes are as follows ef tel, ef eh 2 eh = 4sin20 (6) Elastic and plastic components of concrete strain defined. ch seh of ef = Re cos20-L sind ‘The strains ef, and ef, are then used in the concrete consti- tutive models discussed later Avany particular load step, further plastie straining may' occur. Let Ae), and Ae? represent increments in the plastic in the principal directions; the inerements may be positive or negative, regardless of whether the total strains are tensile or compressive. The parameters defining the envelope of plastic strains can be updated as follows . bet, Aeh, 2 seh +S1(1 + 00526) +2 (1 ~ cos26) (8) elf = ef, +S + cos20) + S20 8) (8) ’ e deh eh 24 +52 c0s20) + S21 + c0528) (0) yh, = yh, +Aef + sin20-Ae’, -sin2@ (10) Note that the principal direction 8 is determined on the basis of the elastic strain components, not the total strains. Envelopes of maximum concrete strains In the hysteretic models for conerete stress that are typically used, stresses are calculated from sets of rules Hinked to a back= bone curve. The latter is usually a formulation describing the monotonic response and requires the knowledge of the maxi- mum conerete strains previously attained, In a continuum, where the principal strain directions may be rotating, this pre= sents a complication, ‘Mohr's circle approach can again be used to approximately describe the maximum strains corresponding to an arbitrary direction. Consider first the maximum compressive strains in the concrete. Suppose, for the current load step, the principal elastic strainsare e, €5 oriented ina direction @ relative to the reference axis 4, y [see Fig. (a). The maximum compressive strains in the 1-and 2-directions, attained during previous loading, ae given by Eee Cony Yensy 0520 + Hs. sin20(11)o a % — Fins 4 1 Yon /2 Ymh . iL SA =] Eine © Fig. 2—Defining envelopes of mazimum concrete strains: (a) principal ‘ares in cracked element; (8) concrete plastic strains (c) Mohr’ circle for ‘maximum compressive strains and (d) Mohr’ circle fr mazimum ten- sil strain. ° 1X 0520 Vn. sin29 ( 9 — Es (12) where Ener Gone ANd Yoqay ARE Quantities defining: maximum compressive strains relative to the 1, j-axes and are used to de- fine Mohr’s circle in Fig. 2(¢). The principal strains eq and qo are then used in defining the backbone curve for the com= pression response (discussed later). If the current total compressive strains are greater than, those previously recorded, then the maximum strain envelope ‘must be updated. Thus, the strain increments Ae oq atid A Eger are defined as 0 ite>em Alin) = (13) fy Eom ife) ef ore.>0 file) =) (ee = — fel >e.>€,, (18) (Cay € ) Sic) ie, < eq ‘where €,» is the maximum compression strain attained during. previous loading, in the direction in question; fg is the stress corresponding to €,3 and fi(€,) is the stress calculated from the base curve for a strain &. (Note: €, and €,y are total strains, plastic strains included.) If the response falls on the envelope ‘curve (i €,€im where €, is the maximum censile strain attained during previ ‘us loading; fm is the stress corresponding t0 mg and fi(€) is the stress calculated from the base curve for a strain €, (On the first tension eyele, the plastic offset €f is held con- stant at the value determined as response crosses from the compression domain. The strain then used in the base curve calculation is net of the plastic strain (Le, the base curve is shifted such that its origin coincides with ef). In subsequent cycles, the plastic strain offset is redefined as follows (2s) (. if ef o “Thus, no positive offsets are currently considered due to lack of suitable model atthe moment. The implication is that the ten- sion response wll pas through the origin during both loading, and unloading: ‘When response falls on the envelope curve (ie, €,> Ey) the maximum tensile strain and corresponding stress are updated to€m and fy, respectively [sce Fig 8(b)). Stresses during un- loading can be calculated from Je(MPa) sa 000%) ig. +—Modeled response for element subjected to uniaxial strain fle) = Eygl€.-€2) () where unloading modulus E,, is defined as es) (€im — 8) ‘To better illustrate the loading and unloading response ob- tained from the previously mentioned formulations, consider the response shown in Fig. + It relates to a reinforced concrete clement subjected toa +2.5 x 10° strain excursion through the first eyele, and a #3.0 x 10" strain excursion through the see- cond cycle, The material properties are as shown; the loading is uniaxial, and it is assumed that the reinforcement does not yield across the cracks (ie, full tension stiffening is sustain- able). It is worth noting that, during the second loading cycle in compression, the reloading curve is not quite linear. This ‘occurs because the plastic offset is continually being redefined at each load step (a step size of 0.1 x 10° was used). The same effect leads to essentially zero tensile stresses during the second and subsequent cycles when €, < 0. It is recognized that the tension unloading and reloading curves passing through the origin and the absence of compressive stresses in the tensile strain domain are at odds with observed and accepted behavior, It is not a result of any limitations in the formulation approach, bbut due to lack of suitable models. Improved models can easily be adopted as they are developed.Pig. 5—Hlysteress model for reinforcement (adapted from Secki"”). REINFORCEMENT STRESS-STRAIN MODELING “The monotonic stress-strain response is assumed to be tri linear. The initial stifiness of the reinforcement is E, giving way to a flat-top yield plateau when the yield stress Jf is reached. Ata strain of €y, a strain-hardening response with a stiffness of Ey begins. ‘The reloading and hysteretic response of the reinforcement is modeled after Seckin!® with some minor simplifications. The Bauschinger effect is represented by a Ramberg-Osgood for- ‘as illustrated in Fig. 5. The first loading cycle is as per the monotonic response. Unloading at any time's given by ME) = S,_, + BLe,-8) (26) where E, is the unloading modulus (given below}. The stresses upon reloading. are given by fle) (e-e)" (27) where (es) and if (@,-€,)4e, In a positive cycle, &, is the maximum positive strain attained during previous cycles fis the stress corresponding to €, as de- termined from the backbone curve, and Ey is the tangent stiff hess at Eq ‘The parameter , is the plastic offset strain correspondling to the zero stress point for the present cycle; itis redefined whenever the stress passes through zero, Note that the reloading curve is tangent to the unloading curve at €,= €,, and converges tangent to the backbone curve at €,=€,. The unload- 18101 Boundary element, 35 mm Wa wa, 1200 em ———+ ea Ey u Main Flex, ralntreament SECTION AA ” Fig. 6—Details of PCA Wall Ba, taken from Oestere etal? ing stiffness E, is dependent on the magnitude ofthe strain ex- ‘cursion relative to the yield strain e, Ina negative eycle, the same formulations apply except that qs the maximum negative strain previously attained. The stress fy and stiffness Bare evaluated accordingly. For the first reverse eycle (positive or negative), €q is taken as zero, with fy = rand “The model requires five parameters to be retained in memory for each reinforcement component at each integration point. These are the maximum positive and negative strains attained Eg. and e, ; the reinforcement stress and strain from the previ- ‘ous load step Jf, and €,,:and the zero point strain €,, ANALYSIS OF SHEARWALL To obtain a preliminary indication ofthe suitability of the proposed approach to cyclic response modeling of reinforced conerete, an analysis ofa large-scale shearwall structure was undertaken. The structure chosen for investigation was Wall Bo." The series of wall tests that included this specimen is widely regarded asa benchmark against which theoretical for~ imulations are calibrated. Wall Ba was a barbell-shaped wal, measuring 1910 mm (6 f 3 in) in total width and 4670 mim (15 fin height from the base (Fig. 6). The wall web was 102 tim (4rin) thick, and contained 063 percent horizontal rein= forcement and 0.29 percent vertical reinforcement. ‘The thick ened boundary elements were 05 mm (12 in) square and contained 3.67 percent vertical reinforcement. The wall was built integral with a heavy base structure and stiff top slab, ‘Theio wn 20 | Top slab 600] = 400] g 200) Zo @ a 2 ange = 40) x 4 800! +5 5 TOP DEFLECTION tm) * & wo CF Pomwatte2 600 Tension Stitfening Es) seme gL ed 3 Sa ® : L £ seal 2e1825 201825 600- a woltpititipii is a tape a conen [ostinato Src eeni ane |fmosomel [a ede 7 ton | ea | | [ton] ow ra Seo cool ROAWALL OR some |[ a | ear |erafen fae [amr | oe] ae scens|| vm | sr form an"| we" | "| oe'| gu “ioe can icaeome 3 al aT 2 Pig. 7—Pinite element mesh used to model Shearwall B2. 2 0) top slab was subjected to lateral cyclic displacements progres- solLtitiritiriiir sively increasing in magnitude, eventually leading to a concrete crushing failure nthe web, The sheat-dominated behavior of the ‘wall represents a stringent test of the analysis procedures, ‘Wall Be was modeled using the finite element mesh shown in Fig. 7, comprising 252 Tinear displacement rectangular ele- iments. The mesh is divided into three zones, representing the web portion, the flanges, and the top slab, respectively. Rein- forcement details and material properties for each 2one are also given in Fig. 7. The material properties, and other pertinent de- tails, used in modeling the structure Were as reported by Sit- tipunt and Wood,* Note that the wal is assumed fully fixed at the base, and no attempt is made to model bond slip. Loading history was imposed by specifying lateral displacements at the top center node of the top slab, An initial analysis was done assuming no further deterioration in the compression softening and tension stifening responses, due to cyclic loading, than is currently assumed by the MCFT for mono- ‘onic loading. ‘The imposed loading consisted of two eyces of top lateral deflection at eich amplitude level, beginning. at 75-mmi (3-in) amplitude and increasing by 25 mm unsil failure occurred TThe resulting predicted response is shown in Fig. (0), compare to the experimentally observed response shown in Fig. s(a) The correlation is reasonably good in the following: respeets: 1) the .N; 2) there is only a ‘minor increase in lateral resistance with increasing displacement ‘walls lateral resistance is approximately 6 080408 ‘TOP DEFLECTION, (rm) Cy Fig. $—Response of Shearsall B2: (a) experimentally measured reo (0) predicted rons asoming mo dyn eto efi 1 (0) peed mip escamingl ay eso sng amplitude beyond nearly Oat-top; 8) failure occurs during: the £125 although in the predicted response, it occurred at the start of 75 mm, that is, the envelope of response in the second +125-mm cycle, whereas in the experiment failure it occurred towards the peak ofthe frst is characterized by crushing of the concrete web near the base region; 5) residual deflections upon unloading are significant and ‘cumulative, indicating a racheting effect accutving in the vertical reinforcement in the flanges; and 6) the lateral stiffness of the wall is progressively diminishing, The only notable discrepancy between calculated and observed behavior is the degeeeof pinching ‘evident in the load-delection hysteresis. This is likely related to the shape of the base hysteretic models used for the compression and tension responses, and to an overestimation in tension stiffening effects, Other mechanisms, such as eyclic shear slip long the rack. surfices due to deterioration in aggregate interlock also contribute to the pinched natureof the observed response mm eyeli; 4) failureI Fig. 9—Details of PDY series of panel tests (a) specimen layout; (8) test parameters (c) concrete cplinder response and (d) remforcement coupon response 1 et al abserved from panel tests that the tension stilf- ening effect diminishes significantly asa result of eyclic loading, To investigate the significance ofthis factor, the analysis of Wall, Be was repeated, assuming no tension stiffening effects (ie, no postcracking tensile stresses in the concrete) The resulting load~ deflection is shown in Fig. s(c). Note that the hysteretic response is somewhat more pinched than previously and isa closer match to the experimental behavior, Other aspects ofthe correlation to the experimental response remain strong, including the load ca- pacity, failure mode, stiffness response, and residual dellection, It should be noted that the analysis algorithms were highly. stable and quick to converge at each load stage. The indication from this analysis is that the proposed approach to modeling ey~ clic load effects isa viable one. PANEL TESTS [Analysis of structures such as the shearwall previously dis- ‘cussed are helpful in testing the ability of calculation procedure to model complex behaviors and interactions in_ structures. However, the data most valuable in assessing, the fundamental constitutive behavior of reinforced concrete, inder various con- ditions including eyelic loading, is that obtained from simple elements subjected to well-defined and well-controlled loads [As previously noted, Stevens et al subjected three large-scale panel elements to reversed eyelic load conditions. Among their findings were that “reversed cyclic loading under biaxial condi tions results in a reduction of the peak compressive stress in ex cess of that observed for monotonic loading.” They also observed reductions in the posteracking tensile stresses in the concrete as result of eyclic loading. Some constitutive relations were formu lated accordingly, although notin a form useful for this study. ‘To build on the work of Stevens, a more comprehensive series ‘of panel tests is planned. The objective of the work is to accu late the necessary data to more accurately define hysteretic mou cls for cracked reinforced conerete in compression and in tension, under general loading conditions. The pilot series for this pro- ‘gram involved three 890 x 890 x 70-mm orthogonally-reinforced panels, namely Panels PDV-1, PDV-2, and PDV-." The panels ‘were constructed of normal strength concrete and contained 1.82 percent reinforcement in one dirvetion and 0.91 percent rein= forcement in the perpendicular direction (.~ and ydirections, respectively). The reinforcement was fabricated from deformed Ds bar, with a nominal diameter of 60 mm. Details of the speci- ‘men construction and material properties are provided in Fig. 9(2) an (b). Typical stress-strain response curves for the conerete and reinforcement are provided in Fig. 9) and (), respectively. ‘The test panels were loaded under conditions of biaxial compression and shear in the fixed proportion of fifi, = -0.4-0.4:1.0, Pane] PDV-1 was subjected to monotonically in- creasing load; Panel PDV-3 was subjected to cyclic loading (unidirectional, with unloading to zero between cycles) and Panel PDV-2 was subjected to reversed cyclic shear, Loads ‘were applied in equal increments of v= 0.5 MPa per load stage up until approximately 70 percent of wlimate capacity, then at increments of 0.25 MPa thereafter. All three panels filed by shear failure ofthe eonerete oceur= ring almost coincidently with yielding of the reinforcement in the z-direction. The reinforcement in the j-direction typically {yielded well before failure, The state ofthe panels at failure are Shown in Fig. 10; Fig. (1 gives the measured shear stress- strain response curves. In comparing the response of PDV-2 with that of control panel PDV-1, itis reasonable to surmise that the effec of eyelic load was to cause significant further de- terioration in the strength and stiffness of the concrete. Note that in comparing ultimate loads, it is important to consider that the strength of the conerete for PDV-3 was significantly higher than in the other two panels “The test panels were analyzed with the finite element proce- dure discussed previously. In all cases, it was found that the strength of the panels was governed by yielding of the rein- forcement in both directions. The large strains accompanying the yielding subsequently led to concrete shear failures. The predicted load-deformation responses are shown in Fig. 12. In the case of the monotonically loaded panel (PDV-1), the corre- lation between experimental and calculated response is good in all respeets. Inthe cyclic and reversed-eyelie panels, some dis- parities become evident. The experimental responses cemon= Strate a behavior more influenced by concrete shear failure, with substantially lower stiffness and significantly different re- sidual strains upon unloading, ‘The elear indication is that the damage to the concrete is more extensive than that assumed in the models. This becomes more clear when we examine the(a) {c) Fig. 10 rmanotnic food foading ond Test panels at failure: (a) Panel PDV-1 subjected to ng: (2) Panel PDV-2 subjected to reversed cyclic Panel PDV=3 subjeced ta cyclic loading. g 4 5 5 5 PANEL PDVI SHEAR STRANN, 7, (ri) a fe z 3 4 ob 5 ab gab gt pane. pov poyviuliyy SSS att SHEAR STRAIN, 7, (mmm) © 4 z7 = 6 2g ga Ey gy : PANEL POVS 5 | SHEAR STRAIN, y,,. (mm) ©) strain response of test panels (a) Pig. 11—Measured shear sre ‘and (¢) Panel PDV-s. Panel PDV-1; (6) Panel PDV- principal compressive and principal tensile stress-strain behav= iors. Shown in Fig, 1$ are the calculated and observed compres- sion responses of Panel PDV-2 for the first v= 5.0 MPa cycle, ‘The inadequacy of the hysteretic models is clearly evident, Thus, the nonlinear hysteric models for concrete subjected to cyclic loads are critical to properly modeling the response. Their influence may not be so apparent when examining mac- roscopie behavior in structures such as shearwalls, where be- havior is largely influenced by reinforcement yielding. However, it can be significant in correctly predicting localized damage, failure modes, and failure loads in shear or compres sion critical structures.s 2 e : i 2 a. z g é a 6 doeorved i ; i i : = Caius oo a ar SHEAR STRAIN, Y,, (rmvm) © 2s zt 2 L Fat E ab a4 La Pawel Pov2 pai 4 bud a SHEAR STRAIN, Y,, (mim) o * ts 15 3543 35 8 ey 10%) . ” i Pape Poi = for Panel PDP? during fret 6 = 3.0 MPa syle () prinpal eet Compresive sreses and) principal tne trees f 4A constitutive modeling approach based on a smeared ro= E tinge stn sats hee rete terete models for normal stress and shear stress Fesponse 2, anatwova | fem oy normal wes ad day res respons of 3 modeled 28 4 nonlinear orthotropic material in the conven- tional sense, with relations expressing the principal compres- FRc cer ete The loadeformation response of many reinforced con- SHEAR STRAIN, y,) (mv) crete structures, such as the shearwall examined, is dominated ) ig 12—Calelted stor srese-strain response of te panes 2) Piel PDI 8) Panel PDP and (0) Pane PDI CONCLUSIONS From the previous discussions, the following conclusions, can be drawn: 1. Employing strain offsets to model the plastic components of strain in the concrete and reinforcement leads to a viable means of using secant stiffness-based procedures for modeling, reversed cyclic load effects in reinforced concrete structures, 2, The resulting analysis procedure exhibits excellent con- vergence and numerical stability characteristics 3. A Mohr's circle approach is useful for constructing enve- lopes defining the plastic strain offset, the maximum compressive strain, and the maximum tensile strain sustained by the concrete, in any particular direction, as a result of previous loading, by the reinforcement behavior, even though ultimate load and failure mode may be governed by the concrete, Proper modeling of the Bauschinger effect in the reinforcement plays a large part in accurately simulating the structure's behavior. 6. Localized behavior, load capacity, and failure mode may, in some cases, be strongly influenced by the highly nonlinear hys- teretic response of the concrete. Simple linear unloading and re loading rules do not adequately represent response andl prevent the analysis procedure fom capturing important subtleties in bbehavior. In particular, the inability of cracks to completely close before compression response begins isan important factor. Experimental work is currently underway to define addi- tional damage factors for concrete subjected to repeated cyclic loads, both for further softening of the concrete in compression and for the shake-out of stress in tension. In addition, the ex- perimental work will ead to the formulation of more realistic uunloading-reloading models useful in a smeared rotating crack context. These refined models can then be easily implemented using the plastic offset method described.uy on Tony NOTATION inal stfioess modulus of concrete secant sifness modulus of concrete tnloading modula fr cance compression unloading modulus for reinforcement intial ties mols fr reinforcement secant sihers modulus for eeinforcement strain hardening modulus fr reinforcement unloading modulus for concrete in tension concrete stress calculate from base curve normal stress in concrete concrete ylinder compressive strength concrete stress corresponding to maximum compressive tra stress in enforcement concrete stress corresponding to maximum tensile strain ‘ied stres of reinforcement, ‘orientation of reinforcement, clockwise frm reference ais first principal serainin concrete second principal trainin conerete ‘otal norma strain in coneret in given direction clastic component of conerete strain in tiection Plastic component af coerete strain in nection maximum compression seain in riection strain a peak compressive stress in concrete cylinder strain corresponding to peak stress in concrete bse curve elastic component af trainin reinforcement, plastic component of strain in reinforcement Strain in enforcement at which strain hardening begins ‘maximum tensile strain in concrete in nection plastic shear strain in conerete relative to.ranes shear strain associated with maximams compressive stains in shear strain associated with maximus tensile stains in conerete orientation of principal strain dretions, relative to axis REFERENCES 1. "Finite Element Analysis of Reinforond Concrete International Workshop, ASCE, New York, 1993,717 pp. 2 Okamura, Hl, and Maekawa, K, Noincar Anais and Comsitative Modes of Rsnfored Concrete, Gibo-do Press, University of Tokyo, Tokyo, 1801, 182 pp. 5. Sittipant, Cand Wood, SL, “nflence of Web Reinforcement on the Celie Response of Steutural Walls” ACI Structural Journal N82, No.6, Now-Des. 1085, pp. 745-756, 4 Stevens N Jet al, “Analytical Modeling af Renforeed Concrete Sub- jected to Monotonic and Reversed. Loedings” Publication No. 87 Department of Civil Engineering, University of Faron, Toronto, 1987 5 Vecchin FJ, "Nonlinear Finite Bement Analysis of Reifored Concrete Menbranes” ACI Stractural Journal 85, No. tin-Fe, 1989 pp. 2-35. 16. Vecchio F J, "Reinforced Concrete Membrane Element Formulations” urna of Stace Engnarng, ASCE. 118, No 3, Ma. 1990, pp. 730-750 1. Vecchio, FJ, and Collin M. P, “The Modified Compression Fila “Theory for Reinforced Concrete Elements Subjected to Shear” ACI JOUR Wat, Prcadings W 83, No.2, Mar-Apr 196, pp. 219-281, ' Vecchio, F J, “Finite Element Modeling of Coneete Expansion and Confinement,” Jourval of Strutural Enginerng, ASCE, V. 118, No.9, Spt. 1998, pp. 2360.20, 9, Veechio,F 1, and Collins, M.P, “Compression Response of Cracked Reinforeed Concrete Journal of Suctural Engaeerng, ASCE, V 119, No 12, Dee. 1983, pp. 8890-8610. 10. Seckin, M, "Hysteretic Behaviour of Castin-Place Exterior Beam. CColume-SiabSubasserblies” PHD thesis, Department of Gil Engineer ing, Universiy of Toronto, 198 TL Oesterle, KG. et al, “Earthquake-Resistant Structural Walle— Tests of Isolated Walls” Report to the National Science Foundation, Con struction Technology Laboratories, Portland Cement Association, Skokis TL, Now 1976, 314 pp. 12: Orsterl,R. Get al, “Web Crushing of Reinforced Coneste Structural Wall,” ACI OUANAL, PrandingrV 1, No 3, May-June 1984, pp 231-241 14 Villani, D. Ry "Reinforced Concrete Subject to Cyelc Loads. A Pilot Study” BASe thesis, Department of Civil Engineering, University of “Toronto, 1995, 147 pp. Proc