Lateral Load Tests on a Two-Story Unreinforced Masonry Building

Lateral Load Tests on a Two-Story Unreinforced Masonry Building Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Tianyi Yi1; Franklin L. Moon2; Roberto T. Leon3; and Lawrence F. Kahn4 Abstract: A full-scale two-story unreinforced masonry 共URM兲 building was tested in a quasistatic fashion to investigate the nonlinear properties of existing URM structures and to assess the efficiency of several common retrofit techniques. This paper presents the main experimental findings associated with the nonlinear properties of the original URM structure. The test structure exhibited large initial stiffness and its damage was characterized by large, discrete cracks that developed in masonry walls. Significant global behavior such as global rocking of an entire wall, and local responses such as rocking and sliding of each individual pier were observed in the masonry walls with different configurations. In addition, formation of flanges in perpendicular walls and overturning moments had significant effects on the behavior of the test structure. A comparison between the experimental observations and the predictions of FEMA 356 provisions shows that major improvements are needed for this latter methodology. DOI: 10.1061/共ASCE兲0733-9445共2006兲132:5共643兲 CE Database subject headings: Masonry; Seismic effects; Failure investigations; Full-scale tests; Retrofitting. Introduction Unreinforced masonry 共URM兲 construction has shown poor performance in past earthquakes. This type of construction, which has been widely used in the United States, therefore presents a large threat to life safety and regional economic development in seismic areas. While numerous experimental investigations have been conducted on URM elements, particularly piers 共Epperson and Abrams 1989; Abrams and Shah 1992; Magenes and Calvi 1992; Anthoine et al. 1995; Manzouri et al. 1995; ATC 1999 to name a few兲, relatively few have been carried out on complete URM structural systems 共Costley and Abrams 1996; Paquette and Bruneau 2003兲. Although these studies provided valuable insight into the nonlinear properties of URM structures and resulted in guidelines for the evaluation of existing structures 共FEMA 356, ATC 2000兲, much work is still needed to characterize the true three-dimensional 共3D兲 behavior of URM buildings. In order to develop strength evaluation and rehabilitation strategies for URM buildings in the mid America region, several coordinated research projects were conducted in the 1996–2002 period under the sponsorship of the Mid-America Earthquake 共MAE兲 center. Those projects included the characterization of 1 Structural Engineer, Stanley D. Lindsey and Associates, Ltd., Atlanta, GA 30339 共corresponding author兲. E-mail: tyi@sdl-atl.com 2 Assistant Professor, Dept. of Civil, Architectural and Environmental Engineering, Drexel Univ., Philadelphia, PA 19104-2816. 3 Professor, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0355. 4 Professor, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0355. Note. Associate Editor: Yan Xiao. Discussion open until October 1, 2006. Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on September 9, 2004; approved on February 17, 2005. This paper is part of the Journal of Structural Engineering, Vol. 132, No. 5, May 1, 2006. ©ASCE, ISSN 0733-9445/2006/ 5-643–652/$25.00. URM building inventory in mid America 共Project SE-1, French and Olshansky 2001兲, quasistatic in-plane strength and retrofit tests on URM piers 共Project ST-6, Franklin et al. 2003兲, and their analyses 共Project ST-4, Craig et al. 2002兲, shake table tests on URM out-of-plane walls 共Project ST-10, Simsir et al. 2002兲 and their analyses 共Project ST-9, Goodno et al. 2002兲, and testing of flexible wood diaphragms 共Project ST-8, Peralta et al. 2000兲 and their analyses 共Project ST-5, Kim and White 2002兲. As a capstone of those projects, a full-scale quasistatic test on a two-story URM structure was conducted at Georgia Tech 共Project ST-11兲. In parallel, a 1 / 2-scale shaking table test of the same structure was conducted at the US Construction Engineering Research Laboratory 共CERL兲 共Project ST-22兲. Based on the component behavior found from other related MAE center projects, the full-scale test was aimed at evaluating the global nonlinear properties of URM structures and investigating appropriate rehabilitation approaches at the structure rather than the component level. This paper is the first of a series presenting the findings of this experimental research, and it describes the findings related to the nonlinear properties of existing URM structures. Two other papers, including one on numerical simulation and one on design implications and recommendations for changes of current FEMA 356 provisions, have also been submitted for publication 共Moon et al. 2006; Yi et al. 2006兲. The outcome of the associated retrofit approaches can be found in Moon 共2004兲. Objectives Previous research has revealed that URM piers are the most important structural components of a URM building 共Bruneau 1994兲. The nonlinear behavior of a URM pier is dependent primarily on its material properties, aspect ratio 共the height divided by the width兲, and vertical stress. Based on a review of previous experimental research results, evaluation guidelines have been proposed 共ATC 2000兲. These guidelines provide strength evaluation equations for the basic failure mechanisms for a URM pier: JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 / 643 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 1. Unreinforced masonry test structure rocking, sliding, diagonal cracking, and toe crushing. These equations, if embedded into an appropriate model, could be used to evaluate the ultimate strength of a URM building. On the other hand, reliable predictors for deformation capacity of piers and mixed failure modes 共i.e., combinations of the basic mechanisms兲 are not available. An approach based primarily on component test data needs to be verified at the global structural level, since the behavior of a URM component in a structural context can be different from what has been learned from individual component tests. The main objective of the full-scale quasistatic test described herein was to address this knowledge gap. Specifically, this test was designed to: 共1兲 validate extrapolating from individual component behavior to the overall response in a URM building system; 共2兲 experimentally identify the critical components in order to develop a systematic method for evaluation and rehabilitation of URM structures; and 共3兲 experimentally validate current code provisions 共FEMA 356兲 as well as advanced analysis tools for URM structures subjected to seismic loading. Design of Test Structure The test structure was a two-story URM bearing wall structure with timber floor and roof diaphragms 共Fig. 1兲. It was intended to represent some structural characteristics of a typical existing URM critical facility 共such as fire station兲 in the mid America area. The dimensions of the building were 7.32 m 共24 ft兲 by 7.32 m 共24 ft兲 in plan with story heights of 3.6 m 共12 ft兲 for the first story and 3.54 m 共10 ft兲 for the second story. The building was composed of four URM masonry walls labeled Walls A, B, 1, and 2, respectively 共Fig. 2兲. The directions for the test structure and the nomenclature for the piers and spandrels in each wall, which are used throughout the rest of this paper, are also shown in Fig. 2. The wall were designed with different thickness and opening ratios to represent typical masonry walls. Walls 1 and 2 were composed of two-wythe brick masonry with a nominal thickness of 20 cm 共8 in. 兲. Wall 1 had relatively small openings, and represented a strong pier-weak spandrel type perforated wall. Wall 2 contained a large door opening 共indicative of the front of a firehouse兲, and represented a strong spandrel-weak pier type perforated wall. In addition, the large difference in stiffness between Walls 1 and 2 allowed the torsional behavior of the URM building to be investigated. Walls A and B were identical walls composed of three-wythe brick masonry giving a nominal thickness of 30 cm 共12 in. 兲. The moderate opening ratios in these two walls are representative of many existing URM buildings. The aspect ratios of piers in the test structure ranged from 0.4 to 4.0. This range was selected in order to allow both “shear” and “flexural” behavior, as had been observed in the individual piers tested by Franklin et al. 共2003兲 and others, to be investigated in the context of an entire, realistic structure. The four masonry walls were connected at the corners, a feature not always reproduced in the past URM tests. This allowed the contribution of transverse walls to the strength of the building to be investigated. Walls A and B employed URM arch lintels, while Walls 1 and 2 employed steel lintels. Both types of lintels are representative of typical lintels used for exiting URM structures 共Stoddard 1946兲. The masonry walls were constructed in standard American bond with a header course every sixth course. Timber roof and floor diaphragms were similar to the “MAE-2” diaphragm tested by Peralta et al. 共2000兲 and represented typical roof and floor diaphragms. The diaphragms were vertically supported on Walls A and B and a light bearing stud wall built through the center of the building. Detailed descriptions of the diaphragms and the rest of the structure can be found elsewhere 共Yi et al. 2002兲 Material Properties Both solid bricks and cored bricks were employed in the building. Nominal dimensions of both types of bricks were 20 cm 共7.75 in.兲 in length, 8.9 cm 共3.5 in.兲 in width, and 5.7 cm 共2.25 in.兲 in thickness. The cored bricks contained a longitudinal hole through the center, with a nominal diameter of 2.2 cm 共0.875 in.兲. The compressive strengths measured by ASTM C67 for the solid brick and the core brick were 41.6 MPa 共6,030 psi兲 and 36.5 MPa 共5,285 psi兲, respectively. The solid bricks were 644 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 2. Plan view and elevation of test structure 共dimensions in meter兲 prism 共ASTM E447兲 and four-bricks direct shear tests 共Yi 2004兲, and the results are given in Table 1. used for the lower 54 courses in the first story of the test structure 共to approximately the 3.8 m level兲, and the cored bricks were used for the remainder of the structure. Preliminary analyses predicted that most of the damage would concentrate on the first floor and thus the solid bricks were used in that area. Solid bricks were employed for all header courses. The materials for the mortar were based on a series of tests aimed at reproducing the strength of pre-1950 URM structures surveyed by the Southern Brick Institute and the fourth writer in the Memphis area. Analyses of mortar samples taken from those structures and conducted at Clemson Univ. indicated a very low amount of Portland cement 共Clemson University 2000兲. After many trial batches were tested, a mortar mix with a ratio of 0.5: 2: 9 共Portland cement: lime: sand兲 was chosen to represent a typical Type K mortar used prior to 1950. The average compressive strength of the mortar measured by ASTM C109 was 0.283 MPa 共41 psi兲. Three five-brick masonry prism specimens and three fourbrick direct shear specimens were constructed for approximately every 10 m2 of wall surface built to track the in situ material properties. Critical material properties were measured by both Loading History The initial test series consisted of low force-level tests on the roof diaphragm aimed at measuring the elastic stiffness of both the masonry walls and the flexible timber diaphragms. The main testing consisted of two series of in-plane wall tests: 共1兲 in-plane wall tests parallel to Walls 1 and 2; and 共2兲 in-plane wall tests parallel to Walls A and B. Walls 1 and 2 were repaired and tested before the in-plane tests on Walls A and B were conducted. The test setup consisted of two 1,000 kN 共220 kip兲 actuators located at the roof level, and two 450 kN 共100 kip兲 actuators located at the second floor level, as shown in Fig. 1. A slight post-tensioning 共350 kN兲 was used to connect the actuators to the masonry walls at the connection points. The test was conducted in displacement control, with a displacement profile based on the first vibration mode. Preliminary Table 1. Masonry Material Properties Strength parameters Masonry compressive strength 共solid brick兲 Masonry compressive strength 共hollow brick兲 Initial bed joint shear bondage strength Equivalent internal shear coefficient Shear sliding coefficient for cracked bed joint Elastic modulus Number of tests Mean COV R2 3 3 21 21 21 16 10.0 MPa 4.1 MPa 0.414 MPa 1.1 1.0 8.0 GPa 0.25 0.09 — — — — — — 0.75 0.75 0.75 — JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 / 645 J. Struct. Eng., 2006, 132(5): 643-652 Instrumentation Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 3. Loading history for test structure research had shown that the first vibration mode controlled the behavior of the very stiff URM walls 共Yi et al. 2003兲. The first vibration mode applied was updated according to the current secant stiffness of a simplified two degree of freedom 共2-DOF兲 structure throughout the tests. This procedure was aimed at capturing the nonuniformly accumulated damage in the test structure, and the resulting evolution of the first vibration mode shape of each masonry wall. The structure was loaded with increasing roof displacements and included two complete displacement cycles at each drift level, as shown in Fig. 3. The second cycle was employed to update the control displacements to achieve a better simulation of the targeted displacements and displacement profile. The applied maximum roof displacement for the tests parallel to Walls 1 and 2 was about 6.4 mm 共0.25 in.兲 or a final interstory drift of about 0.14% for the first story and about 0.05% for the second story. The applied maximum roof displacement for the tests parallel to Walls A and B was about 12.7 mm 共0.5 in.兲 or a final interstory drift of about 0.29% for the first story and about 0.075% for the second story. Each series of tests discussed above were duplicated on the structure after it was retrofitted. Detailed descriptions of these retrofits and the corresponding tests are given in Moon 共2004兲. A typical instrumentation scheme for an in-plane wall test is shown in Fig. 4 for the case of Wall 2. Displacement transducers 共LVDTs兲 GW2R and GW22 were employed to measure the lateral in-plane displacements of Wall 2 at the roof level and the second floor level, respectively. Potentiometers GV2LP and GV2RP were used to measure the global vertical movements. The lateral outof-plane displacements of the test structure were measured by LVDTs GOWAR and GOWBR at the roof level, and LVDT GOWB2 at the second floor level. A typical instrumentation setup for a first-story pier included two vertical LVDTs and two diagonal LVDTs. These LVDTS allowed the vertical, flexural, and shear deformation of the pier to be measured. Strain gages with a 30 mm 共1.2 in.兲 gage length were used at the bottom of each pier to measure the flexural strain. In addition, for some piers such as Pier 1-6, an additional potentiometer was employed to monitor possible sliding behavior. Deformation of the first floor spandrel was monitored by diagonal LVDTs or potentiometers for possible shear deformation. Test Results The initial test series revealed little coupling between parallel masonry walls because of the flexible diaphragms. Therefore, the response of the test structure is discussed separately for each wall. Wall 2, which showed a relatively simple component-dominated response, is presented first. It is followed by the discussions on the behavior of Walls l, A and B. Wall 2 The initial response of Wall 2 up to a roof displacement of 0.43 mm 共0.017 in.兲 was essentially elastic, and no visual cracks were observed. The lateral displacements of the test structure and the corresponding lateral forces were input into a threedimensional elastic finite element model, and the elastic modulus Fig. 4. Instrumentation setup for Wall 2 646 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 5. Final crack pattern for test structure after in-plane wall tests parallel to Walls 1 and 2 of the masonry was backcalculated as approximately 7 GPa 共1,000 ksi兲. This value was consistent with data obtained from material tests on small prisms 共Table 1兲. With increasing lateral displacements, cracks gradually formed at the top and bottom of the three first-story piers, due to the flexural deformation and rocking of these piers. The final crack pattern for Wall 2 is shown in Fig. 5. At the maximum deformation 共roof displacement of 6.4 mm兲, the three first-story piers were clearly rocking. Although there was some evidence of slight rocking of the second-story piers, the damage to the second-story wall was minor. The entire wall above the first story piers behaved as a rigid body, displacing laterally and vertically on top of the first floor piers. The first noteworthy observation was that the majority of the cracks formed at the interfaces between bricks and masonry mortar. The interface between bricks and mortar was the weak link, confirming the strong unit-weak mortar behavior assumed in the experimental design. Second, the cracks at the top of the piers generally did not propagate along the horizontal bed joints as would be expected. Instead, the cracks propagated at an angle of about 45°, perpendicular to the direction of the maximum tensile stress at the corner of the opening. This diagonal type of crack pattern altered the effective length and, thus, the aspect ratios of the piers, resulting in different behavior than expected. Third, cracks developed not only in the in-plane walls, but also in the out-of-plane walls. As a result, the out-of-plane walls moved together with the in-plane walls and greatly affected the response of the test structure. This behavior, labeled as the flange effect, is discussed in later sections. These three phenomena were observed not only in Wall 2 but also in all the other three walls. The base shear-lateral roof displacement history for Wall 2 is shown in Fig. 6. When the lateral roof displacement reached 1.63 mm 共0.064 in.兲 in the positive direction 共southward兲, Wall 2 achieved its maximum lateral strength of 120 kN 共27.0 kip兲. This strength basically remained constant with increasing lateral roof displacement in the following cycles. Similar behavior occurred in the negative direction 共northward兲. When the lateral roof displacement reached −1.35 mm 共−0.053 in.兲, Wall 2 achieved its maximum lateral strength of −109 kN 共−24.4 kip兲. This strength exhibited little degradation with increasing lateral displacements. In addition, the energy dissipation area exhibited by the base shear-lateral roof displacement curves was rather small. This behavior indicated that Wall 2 was governed by rocking. The secant stiffness of Wall 2 decreased rapidly with increasing lateral displacements. Again, elastic 3D finite element 共FE兲 analysis was used to backcalculate the corresponding elastic modulus. The effective elastic modulus of the masonry decreased rapidly from the initial value of 7 GPa 共1,000 ksi兲 to about 0.9 GPa 共130 ksi兲 at a roof displacement of 6.4 mm 共0.25 in.兲. This rapid decrease of stiffness was observed not only in Wall 2 but also in all the other three walls. Wall 1 Significantly different behavior was observed in Wall 1 when compared to Wall 2. Wall 1 behaved essentially elastically when its roof lateral displacements were less than 0.76 mm 共0.03 in.兲. After that, cracks gradually developed in Wall 1 and adjacent Walls A and B. When Wall 1 was loaded in the positive direction 共southward兲, Fig. 6. Force–displacement relationship for Walls 1 and 2: 共a兲 Wall 2 and 共b兲 Wall 1 JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 / 647 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 7. Reading of horizontal potentiometer P1-6S at base of Pier 1-6 a flexural crack initiated in the bed joint at the north base of Pier 1-6 two courses above the first header course 共Fig. 5兲, stepped down to the top of foundation, and propagated farther along the structure–foundation interface. Meanwhile, cracks also developed above the door opening and Pier 1-7 at an angle of approximately 30°. These cracks propagated farther into Wall A, going up at an angle of approximate 45° to the west until it reached the top of Wall A. These diagonal cracks essentially separated Wall A and Pier 1-7 from the majority of Wall 1. As a result, the majority of Wall 1 except for Pier 1-7 rocked together with the triangular portion at the east top of Wall A about the south toe of Pier 1-6. Due to this motion, Pier 1-7 tended to be “left behind” and not to participate in resisting lateral loads. This is an important observation in the context of computing the strength of the entire systems. When Wall 1 was loaded in the negative direction 共northward兲, a horizontal crack initiated at the corner between Wall 1 and Wall B in the bed joint right above the first header course. This crack stepped down to the top of the foundation of Pier 1-6, and propagated farther to the north along this interface until it joined in the existing stair-step cracks at the north toe of Pier 1-6. Simultaneously, a large stair-step crack initiated at the north upper corner of Pier 1-6, and propagated as a stair-step crack upward and to the south. This stair-step crack separated Wall 1 into two big piers: Pier 1-6 underneath the stair-step crack and Pier 1-7 with Wall 1 at the second story above the stair-step crack. The two big piers rocked about the north toes of Piers 1-6 and 1-7, respectively. When Pier 1-6 rocked, the crack at the corner of Walls 1 and B also propagated into Wall B. It joined the horizontal cracks propagated from Wall 2, and completely separated Wall B into the portion below the cracks and that above the cracks. Wall B above the cracks was lifted with the rocking of Piers 1-6 and 2-9. A significant change was observed for the behavior of Wall 1 when Wall 1 was loaded in the negative direction 共northward兲 to a lateral roof displacement of −5.3 mm 共−0.21 in.兲. By then, the base shear force in Pier 1-6 overcame its initial bed joint shear strength and the pier began to slide. Sliding took place along the existing cracks at the south toe of Pier 1-6 and a new crack between the north toe of Pier 1-6 and the foundation. This shifting of Pier 1-6 behavior from rocking to sliding occurred within a short period of time and was accompanied by a loud noise. The sliding of Pier 1-6 was captured by the reading of the potentiometer P1-6S mounted horizontally at the base of Pier 1-6, as shown in Fig. 7. After this drastic shift in performance, the behavior of Wall 1 stabilized in the following test cycles. When Wall 1 was loaded in the positive direction 共southward兲, Pier 1-6 rocked about its south toe and slid along the bed joint, while Pier 1-7 was “left behind.” When Wall 1 was loaded in the negative direction 共northward兲, Pier 1-6 rocked about its north toe and slid along the bed joint, while Pier 1-7 rocked about its north toe. The base shear-roof displacement history for Wall 1 is shown in Fig. 6. Wall 1 reached its maximum strength of 266 kN 共59.7 kip兲 with a lateral roof displacement of 1.07 mm 共0.042 in.兲 in the positive direction, and a maximum strength of −244 kN 共−54.9 kip兲 with a lateral roof displacement of −0.86 mm 共−0.034 in.兲 in the negative direction. Before sliding occurred, the base shear-roof displacement curves of Wall 1 exhibited small energy dissipation, typical of rocking behavior. The sharp drop of strength from −230 kN 共−51.6 kip兲 to −203 kN 共−45.7 kip兲 corresponding to a lateral roof displacement of −5.33 mm 共−0.21 in.兲 indicated the occurrence of the sliding of Pier 1-6. After that, the force–displacement curves of Wall 1 exhibited larger energy dissipation area as compared with the previous cycles. This indicated that the behavior of Wall 1 was a mixture of rocking and sliding. Walls A and B The tests of the in-plane Walls A and B were conducted after the previous URM and retrofit tests in the direction parallel to Walls 1 and 2 were finished. The previous tests had already caused several cracks to develop in Walls A and B, and thus the observed behavior of Walls A and B in the later tests could be considered to be those of predamaged masonry walls. The pre-existing cracks and the cracks that developed in the tests parallel to Walls A and B are shown in Fig. 8. Due to the identical configurations of Walls A and B, the observed crack patterns and the kinematic mechanisms were basically the same for the two walls. Similar to the behavior of Wall 2, the majority of the cracks developed in the first-story piers of Walls A and B. When the building was loaded in the positive direction 共westward兲, the interior piers and the exterior piers at the west side of the building Fig. 8. Pre-existing and final crack pattern for test structure for in-plane wall tests parallel to Walls A and B 共gray line: pre-existing cracks; black line, cracks formed during tests兲 648 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 9. Force–displacement relationship for Walls A and B 共Piers A-10 and B-7兲 rocked about their individual west toe. At the same time, diagonal cracks developed above the exterior piers at the east side of the building 共Piers A-7 and B-10兲. As a result, these two piers were separated from the other portion of the wall and were “left behind.” When the building was loaded in the negative direction 共eastward兲, similar behavior was observed. That is, the interior piers and the exterior piers at the east side of the building rocked, while the exterior piers at the west side of the building were “left behind” through the diagonal cracks developed at the top of these piers. One notable difference between the response of Walls A and B was that slight sliding deformation was observed in Wall B along the top of its first story piers, which was not observed in Wall A. Since the out-of-plane walls were asymmetric for Walls A and B, different crack patterns developed in the out-of-plane Walls 1 and 2. When the building was loaded in the positive direction 共westward兲, the diagonal crack on the top of Pier A-7 propagated upward into Wall 1. As a result, a triangular portion at the north top of Wall 1 moved together with Wall A. In contrast, the diagonal cracks on the top of Pier B-10 did not propagate into Wall 1. Instead, the entire large Pier 1-6 moved together with Wall B. When the building was loaded in the negative direction 共eastward兲 to a lateral roof displacement of −7.62 mm 共−0.3 in.兲, a horizontal crack developed at the midheight of Spandrel 2-6. As a result, the entire second-story wall of Wall 2 displaced upward when Walls A and B moved. However, when the building was displaced farther in the following test cycles, a diagonal crack developed at the south top of Wall 2. As a result, only the triangular portion at the south top of Wall 2 moved together with Wall B, while the entire north portion of the second-story Wall 2 moved together with Wall A. The different sizes of out-of-plane walls participating in the movements of the in-plane walls had a significant influence on the maximum lateral forces resisted by Walls A and B, as is discussed in more detail in a later section. The base shear-lateral roof displacement history for Walls A and B is shown in Fig. 9. Wall A reached its maximum lateral strength of 159 kN 共35.7 kip兲 at a displacement of 5.44 mm 共0.214 in.兲 in the positive 共westward兲 direction, and −182 kN 共−40.8 kip兲 at a displacement of −7.06 mm 共−0.278 in.兲 in the negative 共eastward兲 direction. Wall B reached a maximum lateral strength of 191 kN 共43.0 kip兲 at a displacement of 5.89 mm 共0.232 in.兲 in the positive 共westward兲 direction, and −164 kN 共−36.9 kip兲 at a displacement of −6.94 mm 共−0.273 in.兲 in the negative 共eastward兲 direction. Both walls maintained relatively constant strengths with increasing lateral displacements after the Fig. 10. Vertical stress distribution at base of wall 共with roof lateral displacement of 0.43 mm兲 peak point, and each exhibited relatively small energy dissipation areas. The cracking and displacement response indicated a rocking-dominated behavior for both Walls A and B. Flange Effects As discussed before, the in-plane wall tests caused cracks to develop not only in the in-plane walls but also in the out-of-plane walls. This indicated that a portion of the out-of-plane walls moved together with the in-plane walls. This phenomenon was termed flange effect. Flange effects were observed early in the elastic response of the test structure. Strain gages captured the vertical strain distribution at the base of the test building. Fig. 10 shows such a distribution corresponding to a lateral roof displacement of 0.43 mm 共0.017 in.兲 in the south direction. In addition to vertical tensile and compressive stresses introduced in the in-plane piers, a large vertical tensile stress was introduced in Wall A and a large vertical compressive stress was introduced in Wall B. When displacements were applied to the in-plane Walls 1 and 2, they had to overcome the tensile stresses in Wall A. This resulted in an increase of the lateral resistance of the test structure. Since Wall A was in tension, the contribution of Wall A was considered as a “tension flange effect.” On the other hand, the introduction of compressive stress in the other out-of-plane Wall B helped to even out the large compressive stress at the toe of the pier. This effect delayed a potential brittle toe crushing failure mode. Since Wall B was in compression, the contribution of Wall B was considered as a “compression flange effect.” More detailed discussions on the “tension flange” and “compression flange” can be found in Moon 共2004兲 and Yi 共2004兲. The contribution of flange effects was dependent on the sizes of the flanges. The sizes of tension flanges engaged by each inplane wall can be seen from the crack patterns developed in the out-of-plane walls, as shown in Figs. 5 and 8. The different flange sizes for in-plane Walls A and B explain the different maximum strengths of the two walls. As discussed previously, when the building was loaded in the positive 共westward兲 direction, the small triangular portion at the top north of Wall 1 was the tension flange for Wall A, while the entire left portion of Wall 1 worked as the tension flange for Wall B. As a result, the peak lateral load resistance of Wall B in the positive direction 共191 kN兲 was larger than that of Wall A 共159 kN兲. When the building was loaded in the negative 共eastward兲 direction, the small triangular portion at the top south of Wall 2 was lifted as the tension flange for Wall B, JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 / 649 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Fig. 11. Vertical movement of each Wall: 共a兲 Wall 2, 共b兲 Wall 1, and 共c兲 Wall A while the entire left second floor of Wall 2 was lifted as the tension flange for Wall A. This caused the peak lateral load resistance of Wall A in the negative direction 共182 kN兲 to be larger than that of Wall B 共164 kN兲. Effects of Overturning Moments The tests revealed that the overturning moment induced by the lateral forces had a significant influence on the behavior of the test structure. First, the overturning moment introduced additional vertical stress in the piers, as indicated by Fig. 10. The additional vertical stresses affected the nonlinear behavior of the piers, since lateral strengths and failure modes of a pier are dependent on its vertical stress, as discussed in FEMA 356 共ATC 2000兲. Second, the overturning moment introduced additional deformations. This effect, however, was dependent on the wall configuration, and indicated the formation of kinematic mechanisms in the walls. For example, when Wall 2 was laterally loaded, both the north and south sides of the top of Wall 2 moved upward 关Fig. 11共a兲兴, because Wall 2 was dominated by the local rocking of its first story piers. In contrast, when Wall 1 was laterally loaded, the tensile side of the wall moved upward, while the compressive side of the wall had only a small amount of vertical uplift 关Fig. 11共b兲兴. This was the result of Wall 1 sliding and rocking globally. When Walls A and B were loaded, the top of the wall at the tensile side moved up a considerable amount, and that at the compressive side moved up as well, but with a smaller value 关Fig. 11共c兲兴. This latter phenomenon indicated that the kinematic mechanisms for Walls A and B were a mixture of global overturning movement and local rocking. It should be recognized that the overturning moment results and the flange effects discussed herein are somewhat tied to the experimental design, However, by emphasizing the 3D effects, these test results represent substantial original data and point the way for future research in this area. Evaluation of Test Structure Based on FEMA 356 The FEMA 356 provisions 共ATC 2000兲 were employed to estimate the seismic resistances of the test structure. Since the test structure exhibited little coupling between parallel in-plane walls, each masonry wall 共Walls A, B, 1, and 2兲 was analyzed separately. In the analyses, each wall was modeled with rigid spandrels and nonlinear piers, and the effective height of each pier is taken as the height of adjacent openings. The strength corresponding to each possible failure mode of a URM pier 共rocking, sliding, toe crushing, and diagonal tension兲 was calculated by using FEMA 356 Eqs. 共7-1兲 and 共7-3兲–共7-6兲 共ATC 2000兲. However, In order to estimate true behavior of the test structure, the penalty factor of 1.6 specified in FEMA 356 Eq. 共7-6兲 for the toe crushing strength was not used. The force–deformation relationship for Table 2. Failure Modes and Maximum Strengths of Test Structure 共Calculated versus Experimental Observation兲 Maximum strengths 共kN兲 Failure modes Test results: V Ea Analytical method: VA % difference VA − VE VE 共%兲 1 266 254 −5 2 120 64 −46 A 182 147 −19 B 191 147 −23 Wall a Experimental observations Positive direction: Wall 1 global rocked and slid, Pier 1-7 left behind; negative direction: Pier 1-6 slid and rocked, Pier 1-7 rocked All three first story piers rocked First story piers rocked; pier at tension side left behind First story piers rocked; slightly sliding; pier at tension side left behind Maximum strengths are the larger strengths in the positive and negative directions. 650 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 J. Struct. Eng., 2006, 132(5): 643-652 Analytical method Pier 1-6 slid, Pier 1- 7 rocked All three first story piers rocked All first story piers rocked All first story piers rocked Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. each failure mode also followed Fig. 7-1 and Tables 7.4 of FEMA 356 共ATC 2000兲. The actual failure mode and the corresponding strength of a URM pier were assumed to be controlled by the failure mode with the lowest strength. Overturning moment and flange effects were not considered in the analyses. The material properties employed were based on Table 1. A detailed description of this method can be found in Moon et al. 共2006兲. The predicted maximum strengths and failure modes of the test structure are compared with the experimental results in Table 2. The analyses underestimated the maximum strengths of Walls 2, A, and B up to 46%, clearly due to the neglect of flange effects and overturning moments. The predicted maximum strength of Wall 1 is close to the test observation 共−5 % 兲. However, note that FEMA Eq. 共7-1兲 employed in the analyses used both friction and the masonry shear strength to calculate the bed joint sliding resistance. This method overestimated the shear strength if the bed joint precracked due to rocking as observed in the test. On the other hand, ignoring flange effects underestimated the maximum strength. Therefore, these two factors counteracted each other and resulted in a seemingly accurate prediction. The analyses predicted the failure modes of all walls fairly well. However, the pier-left-behind phenomena observed in Walls 1, A, and B were not captured by the analyses, again due to the neglect of overturning moments. In addition, if the penalty factor of 1.6 was used for toe crushing strengths, the analyses predicted that Walls A and B would fail due to toe crushing, which was not consistent with the experimental observations. Conclusions A full-scale two-story URM building was tested in a quasistatic fashion to investigate its seismic resistances. The structure was tested only up to 0.14% first-story drift in the direction parallel to Walls 1 and 2 and 0.29% first-story drift in the direction parallel to Walls A and B to allow the retrofit of this structure with fiberreinforced polymer 共FRP兲 and post-tensioning system 共Moon 2004兲. The main findings from the URM test structure are as follows. 1. The test URM structure exhibited large initial stiffness, but this stiffness decreased rapidly with small increasing lateral drift. 2. Damage was characterized by large cracks developing at the interfaces between brick and masonry mortar. The failure mechanisms of the test structure were dominated by rocking and sliding of the first-story piers. 3. The observed kinematic mechanisms of the four masonry walls in the test structure were different. The behavior of Wall 2 with large opening ratios was controlled by local rocking of its first-story piers; the behavior of Wall 1 with small opening ratios was controlled by global rocking and sliding; and the behavior of Walls A and B with moderate opening ratios was somewhere between those of Walls 1 and 2. 4. The flange effects due to the movement of out-of-plane walls had a considerable effect on the behavior of the test structure, often significantly increasing the lateral load resistance. 5. Overturning moments affected the distribution of vertical stresses among the piers and contributed to the deformation of the entire structure. 6. A comparison between the experimental observations and the predictions of FEMA 356 provisions showed that current FEMA 356 provisions did not consider the effects of flanges and overturning moments on the behavior of the test structure, and often lead to erroneous predictions. Major improvements are needed in the FEMA 356 methodology. Acknowledgments This research was primarily supported by the National Science Foundation through the Mid-America Earthquake Center 共Award No. EEC-9701785兲. Additional financial support was provided by the Market Development Alliance of the FRP Composites Industry 共MDA兲. Material donations by Cherokee Brick and Tile, Lafarge Cement, MDA, Dur-O-Wal Inc., and Simpson StrongTie, are also gratefully acknowledged. References Abrams, D. P., and Shah, N. 共1992兲. “Cyclic load testing of unreinforced masonry walls.” Advanced Construction Technology Center Rep. No. 92-26-10, College of Engineering, Univ. of Illinois at UrbanaChampaign, Urbana, Ill. Anthoine, A., Magonette, G., and Magenes, G. 共1995兲. “Shearcompression testing and analysis of brick masonry walls.” Proc., 10th European Conf. on Earthquake Engineering, Taylor and Francis, Rotterdam, The Netherlands, 1657-1662. Applied Technology Council 共ATC兲. 共1999兲. “Evaluation of earthquake damaged concrete and masonry wall buildings, technical resources.” FEMA-307, Federal Emergency Management Agency, Washington, D.C. Applied Technology Council 共ATC兲. 共2000兲. “Prestandard and commentary for the seismic rehabilitation of buildings.” FEMA-356, Federal Emergency Management Agency, Washington, D.C. Bruneau, M. 共1994兲. “State-of-the-art report on seismic performance of unreinforced masonry buildings.” J. Struct. Eng. 120共1兲 230–251. Clemson University. 共2000兲. “Test results for bricks and mortar samples.” The National Brick Research Center, Clemson Univ., Clemson, S.C. Costley, A. C., and Abrams, D. P. 共1996兲. “Dynamic response of unreinforced masonry buildings with flexible diaphragms.” NCEER-960001, Univ. of Buffalo, Buffalo, N.Y. Craig, J., Goodno, B., Towashiraporn, P., and Park, J. 共2002兲. “Fragility reduction estimations for URM buildings using response modification.” Proc., 12th European Conf. on Earthquake Engineering Research, London, Paper No. 805. Epperson, G. S., and Abrams, D. P. 共1989兲. “Nondestructive evaluation of masonry buildings.” Rep. No. 89-26-03, College of Engineering, Univ. of Illinois at Urbana-Champaign, Advanced Construction Technology Center, Urbana, Ill. Franklin, S., Lynch, J., and Abrams, D. P. 共2003兲. “Performance of rehabilitated URM shear walls: Flexural behavior of piers.” 共CD-ROM兲 Mid-America Earthquake Center, Univ. of Illinois at UrbanaChampaign, Urbana, Ill. French, S., and Olshansky, R. 共2001兲. “Inventory of essential facilities in Mid-America.” 共CD ROM兲 Mid-America Earthquake Center, Univ. of Illinois at Urbana-Champaign, Urbana, Ill Goodno, B., Craig, J., and Losiriluk, T. 共2002兲. “Seismic vulnerability of partial height URM partitions.” Proc., 12th European Conf. on Earthquake Engineering Research, Paper No. 806, London. Kim, S. C., and White, D. 共2002兲. “Three-dimensional nonlinear time history analysis of lowrise shear wall buildings with flexible diaphragms.” Proc., 7th U.S. National Conf. on Earthquake Engineering 共CD-ROM兲, EERI, Boston. Magenes, G., and Calvi, G. M. 共1992兲. “Cyclic behavior of brick masonry walls.” Proc., 10th World Conf. on Earthquake Engineering, A.A. JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 / 651 J. Struct. Eng., 2006, 132(5): 643-652 Downloaded from ascelibrary.org by Virginia Poly Inst & St Univ on 01/12/16. Copyright ASCE. For personal use only; all rights reserved. Balkema, Madrid, Spain, 3517–3522. Manzouri, T., Shing, P. B., Amadei, B., Schuller, M., and Atkinson, R. 共1995兲. “Repair and retrofit of unreinforced masonry walls: Experimental evaluation and finite element analysis.” Rep. No. CU/SR-95/2, Dept. of Civil, Environmental and Architectural Engineering, Univ. of Colorado, Boulder, Colo. Moon, F. 共2004兲. “Seismic strengthening of low-rise unreinforced masonry buildings.” Ph.D. thesis, Georgia Institute of Technology, Atlanta. Moon, F., Yi, T., Leon, R.T., and Kahn, L.F. 共2006兲. “Recommendations for seismic evaluation and retrofit of low-rise URM structures.,” J. Struct. Eng. 132共5兲, 663–672. Paquette, J., and Bruneau, M. 共2003兲. “Pseudo-dynamic testing of unreinforced masonry building with flexible diaphragm.,” J. Struct. Eng. 129共6兲, 708–716. Peralta, D. F., Bracci, J. M., and Hueste, M. B. D., 共2000兲. “Seismic performance of rehabilitated floor and roof diaphragms.” ST-8 Project Final Rep., Mid-America Earthquake Center, Texas A&M Univ., College Station, Tex. Simsir, C., Aschheim, M., and Abrams, D. 共2002兲. “Influence of diaphragm flexibility on the out-of-plane response of unreinforced masonry bearing walls.” Proc., 9th North American Masonry Conf. 共CD-ROM兲, Clemson, S.C. Stoddard, R. P. 共1946兲. Brick structures, How to build them, McGraw– Hill, New York. Yi, T. 共2004兲. “Experimental investigation and numerical simulation of an unreinforced masonry structure with flexible diaphragms.” Ph.D. thesis, Georgia Institute of Technology, Atlanta. Yi, T., Moon, F., Leon, R., and Kahn, L. 共2002兲. “Performance characteristics of unreinforced masonry low-rise structure before and after rehabilitation.” Proc., 7th U.S. National Conf. on Earthquake Engineering 共CD-ROM兲, EERI, Boston. Yi, T., Moon, F., Leon, R., and Kahn, L. 共2003兲. “Structural analysis of a prototype unreinforced masonry low-rise building.” Proc., 9th North American Masonry Conf. 共CD-ROM兲, Clemson, S.C. Yi, T., Moon, F. L., Leon, R. T., and Kahn, L. F. 共2006兲. “Analyses of a two-story unreinforced masonry building,” J. Struct. Eng. 132共5兲, 653–662. 652 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2006 J. Struct. Eng., 2006, 132(5): 643-652