Seismic Design Considerations for Precast Concrete Shear Wall Connections

The performance of precast shear wall panel structures subjected to earthquakes relies on the behaviour and integrity of the connections between the panels. The current lack of knowledge on the seismic behaviour of precast wall connections limits the use of such systems in earthquake areas. This paper presents results of an extensive experimental program undertaken at the University of Manitoba to study the behaviour of horizontal connections for precast wall panels subjected to large reversed cyclic inelastic deformations. The study included some of the current connections and new innovative configurations that are believed to enhance the ductility and/or energy dissipation capacity. Prototype specimens were tested under combined reversed cyclic flexure and shear with constant axial load normal to the connection to simulate gravity loads. The influence of mild steel reinforcement, post tensioning, shear keys and debonding of continuity reinforcement on the behaviour of the connections was investigated. The test results were used to identify the contribution of each component of the connection to the overall behaviour and to define the various limit states dynamic behaviour. The characteristics of the connection behaviour in terms of stiffness degradation versus number of cycles and load level, energy dissipation capacity and ductility were also examined. Based on the experimental results, design considerations are presented. INTRODUCTION Precast concrete shear wall panel system is popular in North America for low, medium and high rise residential construction due to its economical edge. At present, the use of such a system is very limited in seismically active regions due to the lack of knowledge of how this type of construction performs under seismic loading conditions. The seismic behaviour of the precast structure is significantly dependent on the connections between the precast elements. Current design codes do not specifically address the seismic design of precast wall connections. In addition, little or no information is available in existing literature regarding the cyclic behaviour of such connections. Therefore, in order for precast concrete shear wall panel systems to gain acceptance and be a competitive construction system in seismic regions, the cyclic behaviour of the connections between the precast members must be studied. An extensive multi-phase experimental program, undertaken at the University of Manitoba, investigates the behaviour of horizontal connections for precast concrete shear wall panels subjected to large reversed cyclic inelastic deformations. This program is a continuation of a five year study at the University of Manitoba of the monotonic shear behaviour of typical precast shear wall connections. The results of that study were published by the PCl Journal (Foerster, et.al., 1989, Serrette, eLal., 1989, Hutchinson, et.al., 1991). The objective of the current experimental program is to study the cyclic shear and cyclic flexural/shear behaviour of typical and new innovative connection configurations. This paper reports on the behaviour of the connection under the effect of fully reversed cyclic flexure and shear in the presence of constant axial load normal to the connection to simulate gravity loads. The cyclic pure shear behaviour of typical connection details is presented in a companion paper by the authors (West, et.al., 1993, published in these proceedings). The test results were used to determine stiffness, sU'ength, ductility capacity, energy dissipation, modes of failure, and contribution of the different mechanisms to the overall behaviour. Based on the experimental results, design recommendations for precast concrete wall panel connections are presented. EXPERIMENTAL PROGRAM The experimental program presented in this paper consists of a total of 11 full-scale wall panel/connection specimens were tested under the effect of shear and flexural acting concurrently with an axial load normal to the connection. Eight specimens were tested under reversed cyclic flexure/shear loading. Three specimens were tested to determine the static behaviour. In this study the influence of mild steel reinforcement, multiple shear keys and post-tensioning on the behaviour of the connection is investigated. The performance of new innovative connections, such as debonding of continuity bars and use of mechanical connections for continuity bars, are also examined. Table I gives details of the test program. Details of the different connection configurations investigated are shown in Figure 1. The wall subassemblage selected for testing represents typical horizontal connection near the base of a 10-story precast shear wall building. The test specimen consisted of two prototype precast concrete wall panels connected horizontally by drypack grout and vertical continuity elements (reinforcing or post-tensioning bars) across the joint region. Figure 2 shows the test setup used in this experimental program. The bottom wall panel of the specimen was fixed to the rigid floor of the structures lab by post-tensioning system against two abutments located at each end of the lower panel. The horizontal load was applied using a 1000 kN capacity MTS closed loop cyclic actuator by means of a push/pull-type loading yoke located across the top panel at 1800 mm above the joint region. This configuration provided a moment, M, to shear, V, ratio for the given connection length, L, (MN.L) of 1.5 for all test specimens. (MN.L ratio of 1.5 is typical value for connection near the base of a lO-story wall building). The vertical axial load normal to the joint was applied using an independent prestressing system designed to prevent any constraints in the direction of the applied shear and flexural loads. The vertical load was applied by means of a hydraulic jack and a set of dywidag bars fixed to swivel pin supports at the top and bottom ends of the system as shown in Figure 2. The specimen was instrumented to measure the applied load and to monitor different response mechanisms: 1) Overall panel to panel displacement; 2) Local deformations across joint including interface panel to panel slip and rocking (gap) behaviour i.e. opening and closing of the joint interface; and, 3.) Reinforcement strains at the connection region. At the beginning of each test, the specimen was loaded to a nominal axial stress of 2 MPa which was maintained constant for the entire test duration. The test proceeded by applying controlled series of quasi-static fully reversed cyclic loading pattern, three cycles at each level. Initially, the load was applied using 25% increments of connection yield strength. Subsequently, the specimen was subjected to displacement cycles at multiples of the yield displacement until failure. Between each of these sets of large amplitude cycles, a serviceamplitude cycle (60% of the yield) was inserted to evaluate possible degradation of the connection. The test was terminated when the load carrying capacity of the precast connection dropped below 80% of the maximum load reached. TEST RESULTS The overall connection behaviour for the eight cyclic specimens tested in this program is given by load versus displacement hysteresis loops in Figures 3(a) through 3(h). The results of the static specimens are shown in Figures 5(a) and 5(b). Figure 3 indicate that all the precast specimens exhibited "stable hysteresis" behaviour until onset of failure, followed by pinched loops during post-failure cycles. In general, the post-tensioned specimens exhibited narrow hysteresis loops in comparison to the connections reinforced with mild steel bars. The general response of the tested specimens could be characterized by the following: 1. Rocking: continuous opening and closing of joint region with initially limited slip. 2. Crushing of drypack and panel concrete in the compression zone at joint ends. 3. Yielding of continuity vertical reinforcement. 3. Subsequent buckling of veltical reinforcement through the joint (where applicable). 4. Spalling of concrete and dry pack adjacent to the continuity reinforcement. 5. Limited formation of flexural-shear cracks within the panel. 6. Hinge formation and deterioration of the drypack in the joint region at onset of failure. 7. Rupture or pull-out from the splice sleeve of the continuity bar or strand at the joint region at onset of failure. 8. Major slip with large displacement at the top of the panel at the onset of failure with the exception of the specimen with shear keys and post-tensioned bar specimen Specific values of the various force, displacement, slip and gap quantities at onset of failure of connection response are given in Table 2. Typical failure patterns of the various connections tested are shown in Figure 4. DISCUSSION OF TEST RESULTS Connection Response Based on observation and test results, the cyclic behaviour of the connection could be characterized by three distinct limit states depending on the degree of joint deterioration: Phase I: Phase II: Phase III: Pre-yield behaviour-initial linear elastic, stiff behaviour without any visible damage. Post-yield behaviour without significant joint deterioration behaviour is consistent with nearly stable hysteresis loops and minor damage without extensive crushing. Post-yield behaviour with significant joint deterioration onset of failure with significant reduction in the load carrying capacity under increased deformation. This stage was characterized by hinging at one of the joint with rupture or pull-out from sleeve of vertical joint reinforcement at the other end and by deterioration of drypack and panel concrete at both sides of the joint. The three phases of connection behaviour are illustrated in strength envelope plots of the load versus displacement in Figure Sea) through Sed). These plots also show the effect of the different connection configurations tested on the overall connection response. Figure Sea) and S(b) demonstrates the effect of reversed cyclic loading on the behaviour of bar and post-tensioned strand and bar connections as it compares two identical specimens tested under static and cyclic loading conditions. It is rather obvious the cyclic effects on the connection behaviour by lower maximum load capacity, reduced ductility as well as more severe damage at failure. The strength envelopes of specimen series with mild steel reinforcement with different continuity configurations are compared in Figure S(c). The behaviour could be summarized as: (i) Both welded and mechanically spliced connection detail have similar overall response characteristics; (ii) Shear key connection configuration had consistently limited slip response compared to the plain surface joint; (iii) Bolted RT (bar and tube) connection had a lower strength and ductility in comparison to the welded or spliced connection due to failure of the tube in shear; and (iv) the connection with partially debonded bar showed favourable performance in terms of strength, stiffness, ductility, and deform ability in comparison to the specimens with fully bonded reinforcements. Response of bonded and fully debonded post-tensioned connections with dywidag bars and bonded strands are compared in Figure 5(d). The behaviour indicates that both the posttensioned bar and strand connection had similar overall response characteristics. Fully debonded post-tensioned bar configuration showed the best performance in terms of strength, stiffness, ductility, and energy dissipation. Connection Stiffness The initial elastic stiffness of the connection for all tested specimens, was determined by regression analysis as the slope of the best-fit curve of the load-displacement before yield and is given in Table 2. The results indicate that these stiffness are of the same order of magnitude with maximum variability of 20%. This variation could be attributed to the differences in material properties used for each specimen. Stiffness degradation was monitored by means of service load cycles inserted between each ductility level. Typical measured stiffness degradation at failure was 85%. Table 2 lists the stiffness degradation at onset of failure for all cyclic specimens. Ductility The large connection deformability associated with rocking and slip allowed the panel system to withstand large cycles of deformation beyond first yield wi thou t reduction of the load up to failure. In this investigation, ductility capacity of the connection is defined as the ratio of the displacement at failure to the displacement at first yield. Table 2 gives the ultimate displacement ductilities for all specimens tested. Connection details of the rebar series had an ultimate ductility around 5. Post-tensioned connections had ductility capacity of 6. Debonding of continuity element across the joint had favourable influence on the ductility: Partial debonding of rebar produced ductility more than 6; and with full debonding of posttensioned bar, a ductility of 13 was achieved. Energy Dissipation In this investigation, energy dissipation per cycle is defined as the area enclosed by the loaddisplacement hysteresis curve calculated by integration using an irthouse computer program. Comparison of the cumulative energy dissipation for the specimens with rebar and posttensioning is given in Figures 6(a) and 6(b). As can be seen in Figure 6(a), the partially unbonded mild steel reinforcement specimen had the best per cycle and cumulative energy dissipation characteristics in comparison to all other specimens in the same category. The energy dissipation capability per cycle of specimens with and without shear keys, were nearly identical. The cumulative energy dissipation for the plain connection specimen was higher as it failed at a higher ductility level. The welded versus mechanically spliced configuration had a higher per cycle and cumulative energy dissipation. The bolted connection configuration had a lower cumulative energy dissipation per cycle due to failure of the tube element by shear. The post-tensioned bar detail had a higher per cycle and cumulative energy dissipation in comparison to the post-tensioned strand configuration, as shown in Figure 6(b). The unbonded post-tensioned bar detail had similar per cycle energy dissipation but by far the best cumulative energy dissipation in comparison to all specimens tested.