Seismic capacity rating and reliability assessment of wood shear walls

A modified version of the BRANZ procedure for lateral capacity rating of bracing walls is used to determine the sustainable lateral mass of a 910-mm wide ‘2x4’ wood shear wall. The key modifications involve: (1) the use of a multicriteria system identification method to determine a structural model that simultaneously fits test data from both cyclic testing and pseudo-dynamic testing; and (2) probabilistic treatment of ground motions. Then the 50-year reliability index for the wall system that was rated according to the modified BRANZ procedure was estimated when subjected to a range of earthquake intensities in Tokyo. The reliability index (at the safety limit state) is found to be sensitive to the displacement capacity determined from the static cyclic test, and the suite of earthquakes from which the sustainable mass was calculated from. Thus, it is desirable to quantify and include the inherent uncertainty in displacement capacity and ground motions in the analysis. The method presented herein is general and can be applied to allow the direct use of laboratory data, from cyclic and/or pseudo-dynamic testing, for dynamic and seismic reliability analyses of lateral resisting systems with no distinct yield point. INTRODUCTION Current seismic design procedures for light-frame wood buildings have evolved through experience, field observations and a limited number of simple experiments. Design target performance and reliability are unknown or un-quantified. The 1994 Northridge earthquake and the 1995 Hyogo-ken Nanbu (Kobe) earthquake have not only caused extensive damage to wood buildings but they have also badly shaken the generally high level of confidence that most people have in the seismic performance of wood buildings. These earthquakes have inflicted severe emotional, social and economic difficulties on affected people and communities. Until recently, most programs in earthquake disaster mitigation do not address issues related to residential buildings. Whether the buildings are engineered or not, their reliability under earthquake loads is not known. We have learned from recent earthquakes that their overall impact in heavily populated areas cannot be effectively contained without serious efforts to improve the seismic performance of light-frame wood buildings. Many positive developments are expected in this area in the next several years with the recent federal and state funding of the CUREe-Caltech Woodframe Project in the US. The proceedings of this project’s first technical workshop provides a snapshot of current developments and identifies areas of needed research (Seible et al. 1999). One of the important topics that needs to be addressed is determining the seismic capacity of wood joints and components (e.g. shear walls) that form part of the overall lateral resisting system of a building. Following force-based seismic design procedures, and like the practice in other materials, the conventional approach in seismic design of wood buildings is based on the use of a ductility factor. But this is very difficult to determine for wood systems because there is no distinct yield point in a typical load-displacement curve. In New Zealand, the code of practice for ‘conventional’ timber construction (Standards New Zealand 1990) neatly avoided the problems associated with the force-based seismic design as long as the bracing wall was rated in terms of the mass that it could restrain for a particular earthquake rather than the force that it was capable of resisting (King and Deam 1998). A specific version of a procedure originally developed by the Building Research Association of New Zealand (BRANZ) is considered and extended in this paper. Another pressing topic is determining the reliability of wood buildings under earthquake loads. A simple method of estimating failure probability based on results of nonlinear random vibration analyses of hysteretic wood systems has Principal research scientist & Research engineer, respectively, CSIRO Building, Construction & Engineering, PO Box 56, Highett, 3190, Australia Senior researchers, Building Research Institute, 1 Tatehara, Tsukuba, Ibaraki 305 Japan been developed (Foliente et al. 1996a; 1996b). Ceccotti and Foschi (1998) used a response surface approach together with DRAIN-2DX, with a special hysteretic element, to determine the force reduction factor for buildings with wood shear walls that best satisfies the target performance specified in the National Building Code of Canada. Suzuki and Araki (1998) have presented a combined analytical-numerical method of reliability analysis of hysteretic systems with uncertain properties under deterministic and random excitations. They demonstrated the method for wood buildings with uncertain properties under specified seismic excitations (i.e., deterministic). Foliente et al. (1999) applied the reliability analysis method used in code calibration in Japan (Saito et al. 1998) to estimate the reliability of Japanese post-and-beam wall construction, modelled by an inelastic degrading and pinching single-degree-of-freedom (SDOF) system, subjected to ground motions in Tokyo and Osaka. These site-specific accelerograms were generated using a non-stationary stochastic process model and historical earthquake data. In this paper, we extend the application of the method by Saito et al. (1998) and Foliente et al. (1999) in determining the dependable seismic mass restrained by a ‘2x4’ wood shear wall using a modification of the BRANZ procedure. The reliability index for a wall system rated according to the modified BRANZ procedure, when subjected to a range of earthquake intensities in Tokyo, was then calculated. The method is general and can be applied to allow the direct use of laboratory data, from cyclic or PSD testing, for dynamic and seismic reliability analyses of lateral-resisting systems with no distinct yield point. STRUCTURAL MODEL Experimental basis Most light-frame wood buildings are of platform construction, also known as ‘2x4’ construction. Originally developed in the US, it has been adopted in many parts of the world. This form of construction has also been slowly gaining market share in Japan since its introduction there in the 1970’s. The Building Research Institute (BRI) in Japan has tested many different types and configurations of walls under static monotonic, static cyclic, PSD and shaketable loading. Figure 1a shows a schematic diagram of one such wall. In this paper, we have chosen a 910 x 2450 mm shear wall, which is typically found in Japanese houses and is commonly regarded as the minimum size shear wall in Japan. This wall was sheathed with a 9.5-mm thick JAS No.2 (conifer) plywood on one side. The framing members were 38 mm x 89 mm Spuce-Pine-Fir (S-P-F), JAS Standard grade. JIS CN50 nails, 50 mm in length and 2.87 mm in diameter, fastened the sheathing panel to the frame at a spacing of 100 mm along the four edges and 150 mm along the centre stud. At the end of the wall, the double studs were connected to the steel base with a hold down. Lateral load was applied to the top of the wall following the static cylic protocol shown in Fig. 1b. Similar walls were subjected to PSD and shaketable testing with the North-South component of the 1995 Kobe earthquake recorded at the Kobe Marine Meteorological Observatory, scaled to 60%, as input load. Details of the test program are given in Kawai (1998).