Introduction 1Background

Steel-concrete composite structures comprising the lateral load resisting systems in buildings in seismic areas have been widely explored and developed, with increasing recognition of the benefits obtained from several different types of the composite structures. In a composite steel-concrete lateral load resisting structural system, the steel and concrete components work integrally, such as the encased composite truss members in the Bank of China Tower in Hong Kong, composite shear walls in the First City Tower in Houston, and high-strength concrete infilled steel pipes in the Two Union Square in Seattle (Taranath, 1997). Among the different types of composite lateral-resistance systems, the steel frame-reinforced concrete (RC) infill wall system has been the subject of research within the National Science Foundation (NSF) U.S./Japan Cooperative Research Program on Composite and Hybrid Structures following the 1994 Northridge earthquake in California.

The 1994 Northridge earthquake revealed some unexpected problems regarding the seismic response of the steel moment-resisting frames, especially their connection regions (FEMA, 2000). Cracks were found to occur in the fully-restrained girder-flange-to-column-flange weld and even to propagate into the column flange and web, leading to expensive structural repairs. A series of solutions have been proposed to eliminate or reduce these connection problems directly, such as the removal of backing bars and the use of high quality welding materials (FEMA, 2000). However, a final consensus has not been reached in the earthquake engineering community regarding modifications to the design and construction procedure for moment-resisting frames. An alternative method is to reduce the seismic demands on steel frames by avoiding the use of fully-restrained connections. This philosophy has led to the development of the composite steel frame-RC wall system, utilizing the high shear stiffness of the RC infill wall to attract most of the lateral shear forces from earthquake motions.

Although initial design provisions for the composite steel frame-RC infill wall systems have been proposed in NEHRP (1997), this system is rarely used in practice in the United States due to lack of sufficient research. In Japan, design recommendations for earthquake-resistant design of this type of composite system has been implemented (AIJ, 1985). It is also well known that two buildings of this type behaved satisfactorily during the Great Kanto Earthquake of 1923 in Japan, while other buildings were badly damaged (Makino et al., 1980; Makino, 1984).

In the composite steel frame-RC infill system (Figure 1.2.1), the steel columns and girders serve as boundary members to resist gravity loads and most of the overturning moment, while the RC infill walls carry most of the shear forces. The RC infill walls have the benefiit of increasing the lateral stiffness dramatically, thus avoiding excessive drift and reducing the seismic demands on the steel frames. This provides the opportunity to use partially-restrained (PR) connections in conjunction with lighter steel framing instead of more expensive fully-restrained connections, which are required in the more common steel moment frames (without infill walls). Adequate interface connections, such as headed stud shear connectors, can be used to transfer shear forces uniformly between the

Fig. 1.2.1 The Composite Steel Frame-RC Infill Wall System

Historically, five different types of infill have been considered in the study of the behavior of infilled frames: brick, clay tile, concrete block, plain concrete, and reinforced concrete. The surrounding framing in these studies comprised either reinforced concrete or steel members. The following literature review focuses on the steel frame-RC infill wall system due to its distinctive behavior, with a brief summary of other infilled frames. According to Liauw et al. (1983a, 1983b), infilled frames may be divided into two categories: (1) those with connectors along the interfaces between the frames and the infill walls are called integral infilled frames; and (2) those without are called non-integral infilled frames. These definitions are used in the following literature review.

Research on infilled frames started with the investigation of their static behavior under monotonic lateral loading (e.g., Benjamin and Williams, 1957; Holmes, 1961) with the intention of developing an effective method to predict their ultimate lateral strength. With the recognition that the lateral loading imposed on infilled frames is induced by dynamic phenomena producing reversing load histories, such as earthquakes, wind, or explosions, researchers began to focus their efforts on the cyclic load behavior of infilled frames. During the last three decades, experiments with three different types of loading have been carried out to simulate dynamic phenomena, especially seismic forces: cyclic static and dynamic load tests, pseudo-dynamic tests, and shake-table tests.

Mallick and Severn (1968) performed half-cyclic dynamic load tests on small scale, two-story infilled steel frames, where the cyclic load was applied to the infilled frame in one direction only. The steel frames comprised 0.75 inch  0.75 inch square steel bars and the story dimension was 24 inch  24 inch. The dynamic characteristics, such as the damping ratio and the energy dissipation capacity, were compared between infilled steel frames with and without interface shear connectors. It was found that using a small number of shear connectors in loaded corners could prevent the rotation of the infill walls inside the steel frames and increase the stiffness of the system. However, they failed to see any strength increase with the use of shear connectors. The frequencies and mode shapes of multi-story infilled steel frames were obtained analytically using two models: a shear model in which the axial deformation of the steel components was ignored, and a cantilever model in which the bending deformation of the steel frame members was ignored. Test results showed that the cantilever model was better than the shear model for analysis of multi-story infilled steel frames, particularly for those with a height/span ratio greater than 2.

Liauw (1979) conducted both static and dynamic cyclic load tests on both integral and non-integral steel frames with RC infill walls. The four-story steel frame models comprised 22 mm  22 mm square steel bars, with the size of infill being either 305 mm  610 mm  22 mm or 305 mm  610 mm  22 mm (height  width  thickness). The reinforcement ratio for the infill wall was 0.56%. In contrast to the conclusion drawn by Mallick and Severn (1968), the presence of the interface connectors was shown to increase both frame stiffness and strength significantly. Furthermore, failure of the integral infilled steel frames was induced largely by shear between the steel frames and the infill walls, instead of by the diagonal compression failure of the infill walls in the non-integral infilled steel frames. The dynamic characteristics of these systems were studied further in another series of cyclic tests (Liauw and Kwan, 1985) on similar infilled steel frames having three different interface configurations: (1) no connectors; (2) connectors welded only along the infill wall/steel girder interface; and (3) connectors welded along the entire infill wall/steel frame interface. The tests showed that the infilled steel frames with the Type 3 interface configurations were the most reliable type of construction because they possessed the highest energy dissipation capacity, the greatest damping ratio in the nonlinear range of deformation and the slowest stiffness degradation.

In Japan, two sets of tests were conducted to investigate the behavior of portal steel frames infilled with reinforced concrete, subjected to combined action of constant gravity loading and static cyclic lateral loading (Makino et al., 1980). The portal steel frames were approximately one-third scale and comprised wide-flange sections. There were two specimens in each set, one of which had the strong axis of the steel columns oriented perpendicular to the plane of the infill wall, and the other having the weak axis of the steel columns oriented perpendicular to the plane of the infill wall. The portal steel frames in the first set of tests were cyclically loaded well into the inelastic range of behavior before casting of the infill walls. The two virgin portal steel frames used in the second set of tests had the same sizes as those in the first set, respectively. A few headed studs were employed with the objective of preventing out-of-plane failure of the infill walls. The composite action of system resulted in a uniformly distributed crack pattern in the RC infill walls. The infilled frames having columns bent about their strong axis were shown to have ductile behavior as good as those of typical bare steel frames. It was also concluded that an infilled steel frame having the steel frame pre-loaded into the inelastic range of behavior can have almost the same strength as a virgin steel frame with RC infill walls. A tentative design recommendation was proposed by Makino (1984) with the backing of more experimental research, in which the primary variable was the cross section area ratio between the concrete and the steel column. Also in Japan, another research group (Hayashi and Yoshinaga, 1985, 1986, 1987, 1994) performed a series of similar tests on portal infilled steel frames, in which the width-to-height ratio was chosen as the main variable, and interface connectors were either headed studs or deformed bars.

Shake-table tests are believed to be the best approach for simulating the behavior of a structure during an earthquake. However, shake-table tests are expensive, especially for large scale models. Kwan and Xia (1995) reported a shake-table test on steel frames infilled with light reinforced concrete walls (non-integral). In their tests, the specimen comprised a pair of one-third scale infilled steel frames, connected through reinforced concrete floor slabs at each story, representing a one-bay, four-story structure. The steel frame comprised 40 mm  40 mm  2.5 mm high-yield tubes and story dimension is 1125 mm  1500 mm (height  width). Accelerograms from the EI Centro earthquake were used to excite the shake-table, with progressively increasing magnitudes of acceleration being applied until failure of the specimen occurred. It was observed that the infill walls separated from the steel frames during early loading stages and acted as a diagonal compressive strut. The specimen reached its maximum strength when the corners of the infill walls in the first story were crushed. The specimen did not fail even when the applied peak acceleration was increased to 1.50g, although the frames were badly damaged. However, it was concluded that the RC infill walls would have collapsed out-of-plane, had there not been steel plates attached to the outside of the RC walls to keep them in place. It was also found that the natural frequency of the specimen decreased rapidly as the applied peak acceleration of the simulated earthquake was increased, while the damping increased from 1.7% to 11.0%.

Masonry is a traditional infill material for both steel and RC frame structures. A number of experimental studies have been carried out to evaluate the effect of masonry infill on the seismic behavior of the surrounding frames. Although it was found that masonry infill can also significantly increase the stiffness and strength of frame structures (Klinger and Bertero, 1978; Mehrabi et al., 1996; Negro and Verzelett, 1996), designers in the United States are reluctant to treat unreinforced masonry infill as a structural element in seismic regions due to several unfavorable characteristics of the masonry infill. For example, Mander and Nair (1994) found that the shear strength of brick infill walls was greatly affected by cyclic loading; Mosalam et al. (1997) reported that there were significant pinching zones in the hysteretic load-displacement curves from his cyclic load tests on two-bay masonry infilled steel frames. Furthermore, shake-table tests indicated that it is often unavoidable for the masonry infill to collapse out-of-plane as a result of increased acceleration of ground motion (Dawe et al., 1989; Kwan and Xia, 1995).
1.3.2Analysis of Infilled Frames

During the last three decades, different approaches have been proposed for the prediction of the ultimate strength of infilled steel frames subjected to monotonic lateral load. Holmes (1961) proposed that the infill wall be replaced by an equivalent diagonal strut having a width equal to one-third of the diagonal length of the infill wall. In his elastic model, no axial deformation was included in the steel members and the infilled steel frame achieved its maximum lateral strength when the equivalent strut reached a limiting value of compressive strain. The corresponding deflection was calculated based on the shortening of the equivalent strut. Stafford Smith (1966) proposed an expression relating the width of the equivalent strut to the properties of the frame and infill wall. The width of the equivalent strut varied with value of the following non-dimensional factor:

where

 = slope of the infill diagonal relative to horizontal

A number of researchers have proposed analytical methods based on plastic analysis theory, because the equivalent strut method neglects the contribution of the steel frame and cannot fully represent the behavior of composite infilled steel frames up to their ultimate strength levels. Wood (1978) applied plastic analysis to non-integral steel frames, in which the collapse modes and loads depended on the bending strength of the steel frames and the crushing stress of the infill wall. However, an unrealistic assumption was made in his models that the whole infill wall would behave in a perfectly plastic manner. Therefore, an empirical penalty factor, _p, was needed to reduce the effective crushing stress of the infill wall and to account for the discrepancy between the theoretical predictions and experimental results. Four different failure modes were identified in Wood’s work: a composite shear mode, a shear rotation mode, a diagonal compression mode, and a corner crushing mode. Nadjai and Kirby (1998) extended the shear mode to non-integral infilled steel frames having different types of partially-restrained connections. The experimental and analytical results showed that the shear mode overestimated the shear capacity of the infill wall. Therefore, another panel reduction factor, , was used to further reduce the value of the penalty factor, _p. Based on the results of a nonlinear finite element analysis and experimental observations, Liauw and Kwan (1983a and 1983b) proposed plastic analysis models for both integral and non-integral infilled steel frames, in which the infill wall always failed through crushing in the corner regions. Furthermore, the formation of plastic hinges in the steel frames was determined based on the distribution of the compressive crushing stress in the corner regions of the infill walls. The contribution of interface friction was neglected in these models. Based on nonlinear finite element analysis and model tests, Saneinejad and Hobbs (1995) developed an inelastic method to calculate the ultimate lateral load and cracking load of non-integral infilled steel frames. This method is similar to that proposed by Liauw and Kwan (1983a, 1983b), except that the effect of the interface friction force was included and plastic hinges were assumed to form only in the loaded corners.

Although the advantages of using interface connectors in the composite steel frame-RC infill wall system was recognized by Liauw (1979), Liauw and Kwan (1983a, 1983b, 1985), the interface connectors in their test were bent from wires and the shapes are very uncommon in today’s practice. Makino et al. (1980) and Makino (1984) did use the widely accepted headed studs, but only placed a few studs along the interfaces with the primary purpose of preventing out-of-plane collapse of the RC infill. Furthermore, little effort has been devoted to the improvement of the strength and ductility of the interface connectors used in this structural system. In addition, little work has been done on using partially-restrained connections to minimize the cost of structural system. All of the previous tests of integral RC-infilled steel frames were conducted on either very small-scale models with uncommon steel shapes or on single-story frames. Therefore, testing on a larger scale, including at least two-story specimens, is needed to obtain a better understanding of the cyclic behavior of this composite structural system. The objectives of the proposed research are:

1. 
   To investigate the cyclic behavior of composite steel frame-RC infill wall structural systems by performing cyclic static loading tests on a one-third scale, two-story, one-bay specimen. Strength and stiffness deterioration, energy dissipation capability, deformation capacity of the structural components, and proportions of the forces resisted by different members are the major concerns in the investigation.
   
2. 
   To determine the effect of the strength and ductility of the headed stud connectors on the seismic behavior of this composite system, and to explore a feasible construction method of using headed stud connectors in this composite system.
   
3. 
   To establish simple but rational design guidelines and suggestions for analysis modeling for this composite system based on the experimental results, as well as research results from other sources.
   
4. 
   To provide data for the validation and calibration of computational models of this structural system, such as nonlinear, cyclic finite element analysis, which are being developed in separate research effort at the University of Minnesota.
   

Normal strength steel and concrete are used in this work for all components.

Chapter 1 of this thesis proposes the objectives of this research based on the discussion of research background and review of the work done by previous researchers in related topics. Chapter 2 presents the analysis and design of prototype structures, so as to determine typical member sizes for this system as per the limited current design provisions. Chapter 3 discusses the experimental program, including the specimen design, the test setup, the instrumentation plan, and the loading history.

Chapter 4 to Chapter 7 present the experimental results and synthesis of the test data: Chapter 4 presents the general behavior of the specimen, such as the strength and stiffness of the specimen, the energy dissipation capacity of the specimen, and the observed phenomena of each component; Chapter 5 evaluates the performance of the steel frame and reinforced concrete infill wall in detail; Chapter 6 analyses the performance of the interface headed stud connectors based on measured axial and bending strains and slip and separation demands; Chapter 7 illustrates the force transfer mechanism of the specimen and estimates the percentage of the force resisted by each component.

Chapter 8 compares results from the finite element analysis with test data, so as to establish an appropriate finite element analysis method for design. Chapter 9 verifies the plastic mechanism analysis methods proposed by other researchers and establishes the plastic mechanism for the two-story specimen in the present research. Chapter 10 proposes the recommendations for future experimental study based on the performance of this specimen. Chapter 11 outlines the conclusions drawn from both experimental results and analytical results.