Date of Award

8-2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil Engineering

Committee Member

Dr. Weichiang Pang, Committee Chair

Committee Member

Dr. Scott Schiff

Committee Member

Dr. Nadarajah Ravichandran

Committee Member

Dr. Brandon Ross

Abstract

Light-frame wood is the most common construction for residential and low-rise commercial buildings in the North America. While widespread collapses were relatively rare for light-frame wood buildings, widespread damages were observed for light-frame wood buildings in the recent past earthquake events, for example, the 1989 Loma Prieta and the 1994 Northridge earthquakes. The main reason is that most of these buildings were constructed before the development of modern seismic codes. In San Francisco, about 75% of the buildings were designed and built before the modern seismic building codes. Many of the older wood buildings, in particular, those in the Bay Area in California, have a structural deficiency known as the "œsoft-story". These soft-story buildings often feature an open floor plan with first story for use as parking garage or commercial retail spaces. These soft-story buildings typically have large openings in the exterior walls and no or little interior walls. Accurate modeling of the nonlinear dynamic responses of light-frame wood buildings during the earthquake is very difficult. Light-frame wood buildings have many interconnected framing members and redundant elements, which render the load path less identifiable. In comparison, the seismic responses of the steel and concrete frame buildings with well-defined structural members can be modeled relatively easily by modeling only the main structural elements (i.e. the frames or beams and columns). In contrast, both structural elements (shear walls) and non-structural elements (e.g. drywalls) have significant influences on the seismic behavior of the wood buildings. Additionally, connections such as hold-downs and anchor bolts also have significant impacts on the overall building responses. This means that, in order to obtain accurate results, all of these critical aspects must be considered in the light-frame wood building analyses. In this study, a new numerical package for 3D analysis of light-frame building, called Timber3D, is developed. This new seismic analysis package is developed as part of the NSF funded NEES-Soft project. This new model addresses most of the deficiencies of existing simplified models. Since the Timber3D model is formulated based on large displacement theory, it can be used to model the light-frame wood building performance under seismic loadings from very small deformations all the way to the collapse condition. This package can also be used to perform analyses for both slow and real-time hybrid tests. There were three major phases of this doctoral study. In the first phase of this study, a series of numerical models for the soft-story wood-frame buildings were created for an NSF-funded project, called NEES-Soft. Specifically, four numerical models were created for various applications of the NEES-Soft project. In the first application, a 2D numerical model was created and utilized to perform real-time hybrid testing (RTHT) of a 20-ft. long wood shear wall with and without a toggle-braced damper as retrofit. In the second application, a 2D model was created for reversed cyclic analysis of a light wood-frame wall retrofitted with distributed knee-brace (DKB). In application number three, a series of 3D numerical models were created to perform slow pseudo-dynamic hybrid testing of a full-scale three-story wood-frame building to examine the effectiveness of various retrofits applied to the soft first story. The hybrid test was conducted at the NEES laboratory at the University at Buffalo. In the second phase, the modeling methodology was used to perform a parametric study to evaluate the influence of shear wall placement scheme on the seismic performance of mid-rise wood buildings. There is a recent trend for 4-story and 5-story wood-frame buildings to feature large window openings in the perimeters, leaving only narrow wall piers that are not effective for resisting seismic load. As a result, many of these newer buildings were designed with only shear walls located along the corridors in the longitudinal direction and without perimeter walls as oppose to conventional design where shear walls are distributed along the corridors and along the perimeter walls. The design with shear walls concentrated in the core of the building is referred to as the "œcore-only" shear wall placement scheme. This parameter study evaluated the seismic performances of core-only and conventional shear wall placement schemes. The analysis results show that there is only a marginal difference between the collapse risk of wood buildings constructed with core-only and conventional shear wall placement schemes. In phase three, the Timber3D package was utilized to perform incremental dynamic analysis (IDA) for over 130 wood-frame buildings using a suite of twenty-two FEMA P-695 ground motions. The study was conducted as part of the ATC-116 project funded by the Federal Emergency Management Agency (FEMA) via the Applied Technology Council (ATC). Three occupancy types for wood buildings were considered, namely, single-family dwellings, multi-family dwellings and commercial buildings. The objective of the study was to investigate the various modeling parameters that influence the seismic response and performance. Five parametric studies were carried out: (1) building configuration, (2) collapse displacement capacity, (3) non-structural interior and exterior wall finishes, (4) soil-structure interaction (SSI) and foundation flexibility, and (5) backbone curve shape. This study found that the performance of light-frame wood buildings is primarily related to the actual stiffness and strength contribution of both structural and non-structural elements. If the numerical models are developed without considering the interior and exterior wall finishes, then the predicted response and performance will not conform to the past observations during earthquakes and large-scale testing.

Share

COinS