Current research group members:

Jack Baker (Assistant Professor)


Jack Baker's research focuses on the use of probabilistic and statistical tools for modeling of extreme loads on structures. He has investigated probabilistic modeling of seismic hazards, improved characterization of earthquake ground motions, dynamic analysis of structures, prediction of the spatial extent of soil failures from earthquakes, and tools for modeling loads on spatially distributed infrastructure systems.

Dr. Baker joined the Stanford University Department of Civil and Environmental Engineering from the Swiss Federal Institute of Technology (ETH Zurich), where he was a visiting researcher in the Department of Structural Engineering. He received his Ph.D. in Structural Engineering from Stanford University, where he also earned M.S. degrees in Statistics and Structural Engineering. During his PhD, he also spent time as a visiting researcher in Nagoya University. He earned his Bachelor of Arts degree in mathematics/physics from Whitman College. He has industry experience in seismic hazard assessment, ground motion selection, construction management, and modeling of catastrophe losses for insurance companies.

 

Nirmal Jayaram (PhD student)

Topic: Modeling of spatially distributed ground motions for risk assessment of distributed infrastructure systems.

Many seismic loss problems (such as damage of distributed infrastructure and losses to portfolios of structures) are dependent upon the regional distribution of ground motion intensity, rather than intensity at only a single site. This work extends traditional probabilistic seismic hazard analysis (which considers distributions of future event magnitudes, distances, attenuation variability, etc.), to consider spatial distribution of intensity. Spatial correlations of this intensity have been developed empirically based on data from well-recorded past earthquakes. This can then be used in forward simulations of ground motion intensity in future earthquakes. When implemented in a Monte Carlo simulation, one may obtain millions of simulations of intensity from future events, so methods have been developed to select a greatly reduced number of representative events from a large number of simulations. This will facilitate computationally-intensive risk analyses of systems such as transportation networks.

Ting Lin (PhD student)

Topic: Development of new ground motion selection approaches for structural analysis and building design codes.

Ground motion selection is the bridge between seismic hazard and structural response, the first two components in Performance-Based Earthquake Engineering. This research evaluates current practice and develops new tools for ground motion selection. A more rigorous ground motion selection methodology will carefully examine the aleatory uncertainties from ground motion parameters (Code-based ground motion selection), incorporate the epistemic uncertainties from multiple ground motion prediction models (Conditional mean spectrum (CMS) computation using multiple ground motion prediction models), make adaptive changes to ground motions at various intensity levels (Adaptive Incremental Dynamic Analysis (AIDA)), and potentially use CMS as the new target spectrum (impact of CMS on structural response).

Hyeuk Ryu (Postdoctoral researcher)

Topic: Methods for converting HAZUS capacity curves into fragility functions for risk assessment using PSHA.

 The traditional HAZUS methodology contains only a fragility function for the “complete” damage state, which converges to a collapse rate in the limit, but does not explicitly model collapse. The methodology used here is based on incremental dynamic analyses (IDA), a series of time history analyses with increasing ground motion intensity, for a nonlinear single degree of freedom (SDOF) system with properties chosen to represent a given HAZUS building type. The SDOF system has a multilinear capacity curve with negative stiffness after an ultimate (capping) point, which was recently proposed as an alternative to curvilinear model provided in HAZUS. IDA’s are performed to obtain a distribution of collapse capacities. Results are then combined with ground motion hazard curves to construct probabilistic seismic risk maps showing the probability of collapse of the building in a 50 year timespan, if it were to be located throughout the United States. .

Andrew Seifried (PhD student)

Topic: Techniques for analyzing the effect of response spectrum matching

Response spectrum matching involves non-uniform manipulation of a recorded acceleration time history to obtain a ground motion with a specified target response spectrum.  This procedure is popular in engineering design practice because the variance of the resulting structural response from dynamic analyses for various earthquake records is reduced, enabling an estimate of the mean response to be found with fewer analyses than with other techniques. A potential shortcoming of matching is bias in the mean response estimate.  This research aims to extend previous studies of spectrum matching, to consider a broader range of building types and matching techniques, and develop new criteria to analyze how appropriate the resulting manipulated ground motions are for structural engineering analysis.

Shrey Shahi (PhD student)

Topic: A comprehensive probabilistic approach for incorporating the effects of near-fault directivity into design criteria.

Ground motions with a pulse at the beginning of velocity time history are known to cause severe damage in structures. These ‘pulse-like’ ground motions are believed to be caused by seismological phenomenon like forward directivity and are mostly observed in near fault regions. Despite the recognized importance of these ‘pulse-like’ ground motions, methods to account for their effects in design are relatively ad-hoc. My research focuses on developing a scientific framework to account for ‘pulse-like’ ground motions in design by modeling their effects on both seismic hazard computations and structural response. The framework will combine simple models for different aspects of pulse like ground motions (e.g., the probability of observing them at a site, the orientations in which they are likely to be observed, their effect on response of simple structures and buildings). This framework is modular and thus will allow for changing of a particular model as new knowledge is gained, without affecting the rest of the framework.

Lynne Schleiffarth (MS student)

Topic: Effect of low-magnitude earthquakes on PSHA deaggregation and the Conditional Mean Spectrum

I am looking at the conditional mean spectrum as an alternate method of ground motion selection for dynamic analysis, as opposed to the uniform hazard spectrum which is overly conservative because it assumes that the response spectrum is a constant number of standard deviations above the predicted mean value at all periods. I am considering low magnitude earthquakes in PSHA and considering their effect on the conditional mean spectrum. My research will promote sustainable earthquake design by quantifying the risk posed by the more common low-magnitude earthquakes and encouraging engineers to consider the probabilities of occurrence of earthquakes of varying magnitudes when designing a lateral resisting system.

Yoshifumi Yamamoto (PhD student)

Topic: Wavelet analysis techniques for ground motion simulation and modification.

The goal of this research is to develop a stochastic method to generate artificial ground motions with time and frequency non-stationarities using wavelet analysis. These artificial ground motions are modeled by two types of probability distributions of wavelet packet coefficients, which correspond to large- and small-amplitude portions of the ground motions. The parameters of this model are then estimated using regression analysis as a function of magnitude, distance, and site conditions. Using this regression model, we can generate the artificial ground motions for arbitrary conditions. Further, because this simulation procedure is so computationally inexpensive, it can be combined with a seismic source model to provide a direct simulation method that circumvents traditional Probabilistic Seismic Hazard Analysis and selection of recorded ground motions.

 

© 2009 Jack Baker. All rights reserved.