Dr. Henry V. Burton is an Associate Professor and the Presidential Chair in Structural Engineering in the Department of Civil and Environmental Engineering at the University of California, Los Angeles. His research is directed towards understanding and modeling the relationship between the performance of infrastructure systems within the built environment, and the ability of communities to minimize the extent of socioeconomic disruption following extreme events. Dr. Burton is a registered structural engineer in the state of California. Prior to obtaining his PhD in Civil and Environmental Engineering at Stanford University, he spent six years in practice at Degenkolb Engineers, where he worked on numerous projects involving design of new buildings and seismic evaluation and retrofit of existing buildings. He is a recipient of the National Science Foundation Next Generation of Disaster Researchers Fellowship (2014) and the National Science Foundation CAREER Award (2016).
The physical laboratory experiment is a primary tool in advancing our fundamental understanding of structural behavior under seismic loading. For a given project, the results from one or a small set of physical experiments are used to understand how different structural properties or design strategies affect behavior. The results from a given physical experiment can also be used to anecdotally validate a particular modeling strategy. The last few decades have accumulated hundreds of structural experiments for various types of components and sub-assemblies. Also, with the creation of cyberinfrastructure systems such as DesignSafe, these datasets are being widely used among structural engineers. A single dataset would typically include comprehensive information from experiments conducted on a particular type of component or sub- assembly.
The increase in the collection and curation of data from physical experiments has spurred the development of data-driven approaches for estimating the parameters that are used in structural analysis models. Such studies use data from largely disparate experiments (i.e., each performed with a different goal in mind) and various forms of regression to develop predictive models. A more recent trend has been to use ML algorithms as the primary engine for developing these predictive models. However, while useful for prediction, these data-driven models are unable to provide fundamental insights about the behavior of the component or system of interest.
This presentation will describe a framework that can be used to extract causal relationships from disparate structural experiments. The proposed approach leverages the language, tools, and models that have been (and continue to be) developed in the broad area of causal inference, which has been advanced in fields such as statistics, computer science and the social sciences. We demonstrate the proposed methodology by using a dataset comprised of more than 700 experiments on reinforced concrete shear walls (RCSWs). Specifically, the data is used to quantify, from a causal perspective, the performance implications of using overlapping hoops in RCSWs instead of single hoops with cross ties.