Date of Award

Spring 4-24-2014

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

First Advisor

Abbie B. Liel

Second Advisor

Elwood J. Kenneth

Third Advisor

Keith A. Porter

Fourth Advisor

Guido Camata

Fifth Advisor

Yunping Xi

Abstract

Reinforced concrete frame buildings with masonry infill walls have been built all around the world, specifically in the high seismic regions in US. Observations from past earthquakes show that these buildings can endanger the life of their occupants and lead to significant damage and loss. Masonry infilled frames built before the development of new seismic regulations are more susceptible to collapse given an earthquake event. These vulnerable buildings are known as non-ductile concrete frames. Therefore, there is a need for a comprehensive collapse assessment of these buildings in order to limit the loss in regions with masonry infilled frame buildings.

The main component of this research involves assessing the collapse performance of masonry infilled, non-ductile, reinforced concrete frames in the Performance Based Earthquake Engineering (PBEE) framework. To pursue this goal, this study first develops a new multi-scale modeling approach to simulate the response of masonry infilled frames up to the point of collapse. In this approach, a macro (strut) model of the structure is developed from the response extracted from a micro (finite element) model specific to the infill and frame configuration of interest. The macro model takes advantage of the accuracy of the micro model, yet is computationally efficient for use in seismic performance assessments requiring repeated nonlinear dynamic analyses. The robustness of the proposed multi-scale modeling approach is examined through comparison with selected experimental results.

The proposed multi-scale modeling approach is implemented to assess the collapse performance of a set of archetypical buildings, representative of the 1920s era of construction in Los Angeles, California. The collapse performance assessment is conducted for buildings with varying height and infill configurations. Dynamic analyses are performed for the constructed nonlinear models. Results of this study capture the influence the infill panel has on the collapse performance of the frame. This assessment is also used to investigate the significant difference infill configurations have on the collapse performance of the frame. These results can be used to prioritize mitigation of the most vulnerable RC frames.

This research also examines the collapse performance of non-ductile RC frames without infill walls. One of the primary goals in the seismic assessment procedure used in this study is to identify the hazardous buildings that are in critical need of rehabilitation. These buildings are known as "killer buildings." In order to reduce the seismic hazard risk, we need a simple evaluation methodology for existing buildings that can quickly identify the killer buildings. In this evaluation methodology, the collapse safety of the buildings is defined as a function of a set of parameters that are known to significantly affect the risk of building collapse. These parameters are known as "collapse indicators." This research uses these collapse indicators to examine the trend between the collapse risk and variation of each indicator. In addition, this study investigates the relation between building collapse and the extent of deficiency. The extent of the deficiency is defined by the number or percentage of the deficient elements, for instance number of columns with wide transverse reinforcement spacing, in the story of interest. These results are used to investigate the appropriate definition of these collapse indicators in the evaluation methodology.

An important aspect of the seismic assessment procedure presented in this dissertation is to quantify the uncertainty embedded in the nonlinear model used in nonlinear dynamic analysis. In the last part of this study, a new methodology is proposed to quantify modeling uncertainty through a set of drift distributions derived from data submitted to a blind prediction contest conducted at UCSD (2007). In this contest, participants were asked

Share

COinS