Performance of partially and fully anchored wood-frame shear walls under earthquake loads.

The U.S. Census Bureau (2004a) estimated that $538 billion of new housing would be constructed in the United States in 2004. A large percentage of the 1.3 to 1.5 million homes built annually (U.S. Census Bureau 2004b) is wood-frame construction. Typically, wood- frame construction has been thought to perform well in earthquakes, but an estimated $10 billion of damage to residential structures caused during the 1994 Northridge earthquake (EERI 1996) stirred a renewed interest in improving wood-frame construction codes and methods. Projects such as the CUREE-Caltech Wood-frame project (Seible et al. 1999) were launched to further understand behavior during earthquakes.

The San Andreas Fault, which runs most of the length of California, is commonly recognized as a highly active seismic zone. Since the 1994 Northridge earthquake, much energy has been focused on understanding and correcting problems caused by earthquakes in the area surrounding the San Andreas Fault. The Pacific Northwest has traditionally been considered an area of relatively low seismic risk. In the 1980s, however, it was discovered that the Cascadia subduction zone between southern Oregon and Vancouver, British Columbia has the potential to generate very large magnitude earthquakes (Atwater 1987). This seismic hazard is caused by the action of the Pacific Plate and smaller Gorda, Juan de Fuca, and Explorer Plates subducting below the North American Plate.

In North America, subduction zone faults tend to build strain energy over very long periods of time and the earthquakes tend to be less frequent, but are often larger magnitude and longer duration than for strike slip faults. Similar types of faults in Alaska and Chile have generated earthquakes with magnitudes between 8.0 and 9.0 on the Richter scale. Despite the damage potential that these types of faults pose, very little research for wood-frame shear walls has focused on subduction zone earthquakes. This project examines the performance of wood shear walls under both subduction zone and strike-slip earthquakes.

Literature review

Several studies have used shake table testing to analyze the performance of wood shear walls. Falk and Itani (1987) measured the natural period and damping of 2.44 by 7.32 m perforated shear walls. Dean et al. (1988) analyzed several walls subjected to sinusoidal vibration and 1940 El Centro shake table tests. They observed an increase in strength and stiffness of walls tested dynamically over walls tested statically. Primary failure modes observed were nail pull out from framing members or nails pulling through the plywood. Dolan (1989) analyzed walls subjected to the 1952 Kern County and 1971 San Fernando earthquakes to verify the accuracy of a numerical model. Several sheathing-nailing combinations were subjected to various amplitudes of each earthquake. Filiatrault and Foschi (1991) compared the performance of walls constructed with nails and construction adhesive to conventionally nailed walls. They loaded walls with 1971 San Fernando, 1940 E1 Centro, and 1977 Romania earthquake time histories and static pushover tests and found the walls behaved almost elastically to failure and experienced almost no damage under moderate level events. Kamiya et al. (1996) tested plywood and gypsum shear walls subjected to 1940 E1 Centro and 1952 Taft pseudo-dynamic shake table tests. They concluded the gypsum board shear walls failed even with a seismic mass allowed by then current design shear values in Japan. Kawai (1998) performed pseudo-dynamic tests on various shear walls using the 1995 Kobe earthquake to compare actual performance with bilinear hysteresis models. Yamaguchi and Minowa (1998) compared shake table tests of perforated shear walls under a Kobe type earthquake to static tests. Common failure modes they observed were nail pullout from the framing and nail pull-through in the sheathing. They also found the shake table tests showed an ultimate strength 14 percent greater than quasi-static tests, but only about 50 percent of the displacement at peak load. Karacabeyli and Ceccotti (1998) performed several different pseudo-dynamic tests using various earthquake time histories and compared the results to several cyclic test protocols. They found that on average the maximum load-carrying capacity of walls tested pseudo-dynamically was 15 percent greater than for walls tested using cyclic protocols. Failure modes observed were nail pull through, nail withdrawal from the framing, nail edge tear out, and occasionally nail fatigue. Yamaguchi et al. (2000) compared the results of an E1 Centro shake table test to monotonic, cyclic, and pseudo-dynamic tests at various loading rates. They found the maximum strength of the wall tested dynamically was approximately 125 percent of the value from the pseudo-dynamic test. Results of the slow cyclic test were similar to the pseudo-dynamic test, however, the fast cyclic test was similar to the shake table test. Duram et al. (2001) tested 12 walls with oversized (2.44 by 2.44 m) and conventional (1.22 by 2.44 m) oriented strandboard (OSB) panels. They compared results of shake table tests with the 1992 Landers earthquake time history to monotonic and cyclic tests. They found walls sheathed with oversized OSB panels were stiffer and thus experienced lower drift than conventionally sheathed walls. Ceccotti and Karacabeyli (2002) conducted a shake table test of a two-story shear wall system to compare with design values from the Canadian design code. They found the actual period of the structure was much longer than estimated by the code equation.

A limitation of the research just discussed to this project is that all of the studies investigated earthquakes from crustal strike-slip fault mechanisms. Duration, magnitude, and frequency of subduction zone earthquakes may cause a different structural response than crustal earthquakes. Secondly, although limited monotonic and cyclic testing has been done on prescriptive walls (without hold-downs) (see discussion in Seaders 2004), all of the dynamic testing has been on engineered shear walls (with hold-downs). Since residential construction is often designed according to prescriptive codes such as the International Residential Code (IRC) (ICC 2006a), there is a need to quantify the performance of these structures under actual loading conditions.

In recent years there has been some controversy over whether design values based on monotonic testing should be reduced, as the City of Los Angeles/UC Irvine shear wall test program (CoLA/UCI 2001) recommended a reduction in allowable unit shear values. Similarly, Dinehart and Shenton (1998) recommended a 25-percent reduction in allowable unit shears based on monotonic tests due to a reduction in load between the first and fourth cycles of repeated cycles with equal peak displacement, observed using the sequential phased displacement (SPD) cyclic protocol (SEAOSC 1996). CUREE design recommendations (Cobeen et al. 2004), however, contradicted the City of Los Angeles/UC Irvine report and the research by Dinehart and Shenton (1998), stating there is no evidence to support a reduction in design loads at this time. Research results in this paper hope to contribute to this question by comparing shake table testing presented in this paper to monotonic and cyclic testing conducted in another portion of this project (Seaders 2004).

Objectives

This paper presents a portion of a two-phase project to investigate the performance of walls under monotonic, cyclic, and earthquake loading protocols. Overall objectives are:

1. To understand the behavior (load-deflection response, strength, failure mode, ductility, and energy dissipation characteristics) of shear walls under various dynamic loadings: a) subduction zone, long duration earthquakes from Oregon/Washington and b) earthquakes from specific sites in California.

2. To compare the behavior of shear walls under standard static (ASTM E564) and cyclic (CUREE) test protocols to the behavior of walls subject to various actual earthquake records.

Phase I of the research included monotonic and cyclic testing and preliminary subduction zone earthquake testing. Phase II included additional earthquake testing.

This paper presents the results of earthquake testing conducted in Phase I of the project. Specific objectives for this paper are to:

1. Compare the performance of walls under earthquake tests to results obtained under monotonic and cyclic loading protocols,

2. Evaluate the effects of anchorage on wall performance, and

3. Evaluate the performance of the walls qualitatively and quantitatively with respect to code-defined performance measures.

Materials and methods

Load frame and test equipment

All of the tests were conducted in the Department of Wood Science and Engineering at Oregon State University.

The loading frame used for earthquake testing is shown in Figure 1. The bottom rail was a 102 by 152 by 10 mm steel beam providing a moveable foundation for the wall. The beam was supported on pin joints connected to linear bearings at both ends. Linear bearings traveled on 51 mm steel shafts mounted to the strong floor. This ensured the foundation moves with a minimal amount of friction and without internal stresses due to flexure in the foundation beam. A 44.5 kN servo controlled hydraulic actuator with 153 mm total stroke was attached to the foundation beam with a 90.0 kN dynamically rated load cell in-line.

Inertial mass was placed on a cart and coupled to the wall by means of a lever assembly. Two 25.4 by 914 by 914 mm steel plates fastened to a four-wheeled cart provided the mass. A laterally braced tower supported a 102 by 152 by 10 mm steel beam acting as a vertical moment arm to couple the mass cart and a steel channel (top rail) attached to the top of the wall. The moment arm had two alternate pivot points for attachment to the support tower. A 51-mm shaft and bearing attached at the midpoint or quarter point of the moment arm, allowing either a 1:1 or a 1:9 equivalent mass ratio at the top of the wall. Horizontal struts connected the mass cart and top rail to the moment arm. The total equivalent inertial mass applied to the top of the wall was 4545 kg, which represents the tributary mass on a 2440 mm wide brace panel in a one-story house.

Print Get the Mag Weekly Updates