Loss analysis and loss based seismic design for woodframe structures

Abstract

Light frame wood structures serve as the vast majority of construction type for residential structures throughout North America. This type of structure utilizes mainly wood shearwalls and a joist-sheathing floor system to resist the live and dead loads. Compared to concrete and steel structures, light frame wood structures typically have a relatively low construction cost, i.e. overall value. However, recent earthquake surveys and studies revealed that the costs involved in repairing these structures after earthquakes can be quite significant even when the structure survives the event without collapse. As a result of the large stock of residential construction in the U.S., earthquake-induced losses for this type of structure could have a detrimental financial impact on both the building owner and the community as a whole. During the 1994 Northridge earthquake in California, the economic loss directly connected with residential wood structures was more than $20 billion, and provides the impetus for this study. The limitation of current woodframe structural design philosophy is that it focuses only on life safety and does not address providing damage-limitations. The idea that financial losses for a structure during an earthquake should be addressed during the design stage leads to loss-based seismic design, which falls under a more comprehensive framework, performance-based seismic design (PBSD). The objective of this study is twofold. The first objective is to develop a method to establish a probabilistic model for earthquake-induced loss of existing or newly designed wood frame structures for a given period of time into the future; then this method is to be incorporated into a performance-based seismic design framework to conduct loss-based design/optimization for wood frame structures. The first objective involves the modeling of the structural behavior during earthquake loading, financial loss estimation of the structures, and probabilistic modeling/simulation of the seismic hazard. An improved non-linear structural model for light frame wood structures was proposed and used in the structural analysis. A vulnerability based loss estimation framework was developed to incorporate multiple uncertainty sources which contribute to financial losses. Bayesian techniques were used in the modeling of the framework elements in order to include as much usable information as possible. An alternative loss estimation procedure based on Monte-Carlo simulation was also proposed with the ability to consider more realistic situations such as damage accumulation and structural degradation. The results of the long-term earthquake-induced loss were obtained in the form of simulated sample pools conditional on time span. Based on the results from the first objective, simplified loss estimation in the design stage was introduced in order to obtain the vulnerability target for loss-based optimization. This optimization procedure was later used to implement loss-based design in the numerical examples. The most significant anticipated contribution of this study to the woodframe design and research communities will be the development of a long-term seismic induced loss estimation framework/procedure for woodframe structures that enables loss-based design and evaluations. Although this procedure will only be applied to woodframe structures in this dissertation, the general solution can also be used to analyze other structure types provided the necessary details are addressed. The automated analysis package developed in this study will essentially be a practical implementation of the performance-based seismic design (PBSD) concept for most commonly constructed residential structures (in North America). Other contributions will include a new formulation for the hysteretic models for wood shearwalls, a performance-based seismic design procedure for woodframe residential buildings, and a generalized user-friendly woodframe building seismic analysis tool package for the research community.