The goal of the proposed project is to systematically investigate the effects of uncertainty in geotechnical parameters on uncertainty in engineering demand parameters (EDPs). The work will be accomplished in a hierarchical manner, beginning with relatively simple geotechnical models and site conditions and progressing to more complicated models and conditions. The EDPs will be simple, and selected to be consistent with those being used in PEER structural engineering investigations. The results will provide insights into the level of uncertainty inherent in ground motion predictions with the tools most commonly used in contemporary geotechnical engineering practice, and with more advanced tools developed by PEER.

The effects of uncertainties in geotechnical parameters on EDPs will be investigated for a variety of sites and hazard levels. The site conditions will be represented by a series of idealized, but realistic, sites corresponding to the UBC site classification system (site classes SB – SF). A “base case” profile will be developed for each site class; where appropriate, the base case profiles will correspond to those of PEER testbed sites. Multiple realizations of the base case profiles, including spatially variability of soil properties about constant means, will be used. The median response of each set of spatially variable realizations will be considered as representative of the behavior of the profile.

Four different modeling approaches will be used for each site – one-dimensional equivalent linear, one-dimensional nonlinear, one-dimensional OpenSees, and two-dimensional OpenSees. The one-dimensional equivalent linear analyses will be performed using SHAKE91 or equivalent; this is the most commonly used model in current geotechnical earthquake engineering practice, and its use will therefore provide a good indication of the uncertainty in site response predictions inherent in current practice. The one-dimensional nonlinear analyses will be performed using a code such as DESRA or DMOD; these codes allow consideration of nonlinear effects in a relatively simple one-dimensional framework. These codes are among the most commonly used nonlinear site response codes in current geotechnical practice. The OpenSees analyses will allow consideration of more detailed aspects of soil nonlinearity using the more advanced constitutive models that have been implemented into that code. The one-dimensional OpenSees analyses will also provide validation, through comparison of results with those of the preceding one-dimensional analyses, of the OpenSees model. The two-dimensional OpenSees analyses will then illustrate the effects of geotechnical uncertainties on sloping ground sites, including permanent soil deformations (as an EDP) in addition to site response.

The profiles will be subjected to a series of input motions all scaled to consistent values of an intensity measure (IM). IMs corresponding to three hazard levels – 50% probability of exceedance in 50 yrs, 10% probability of exceedance in 50 yrs, and 2% probability of exceedance in 50 yrs – will be used. The number of input motions required will depend on the IM selected; the selection will be made in consultation with other PEER researchers involved in IM research. The uncertainties due to geotechnical parameters will be evaluated by comparing the mean (or median) response from the suite of input motions.

We have begun our initial efforts by considering the Van Nuys testbed, which is the testbed with which this project is formally associated. We have compiled site data, characterized uncertainties in geotechnical parameters, and prepared tornado diagrams to provide an initial view of the effects of uncertainties in seismological, geotechnical, and structural parameters on selected EDPs. A series of 20 ground motions were scaled to provide hazard-consistent mean spectra at the 475-yr level, propagated through the Van Nuys soil profile using an equivalent linear soil model, and input into a nonlinear, inelastic SDOF system. Peak acceleration, peak velocity, normalized peak range (maximum range of displacements normalized by yield displacement) and normalized hysteretic energy were computed for each case. The tornado diagrams shown at right illustrate the relative significant of a variety of input parameters for a structure with a fundamental period of 1.5 sec (which is the fundamental period of the Van Nuys Holiday Inn testbed structure).

This work is the only work of its type being conducted within PEER, although characterization and propagation of uncertainties in structural parameters is being pursued in Thrust Area 3. The characterization of uncertainties in geotechnical parameters, and evaluation of their effects on soils and foundations under static loading conditions, has been investigated by numerous researchers in different countries for a period of many years. Various investigators have also investigated the effects of uncertainties and spatial variability of soil properties on site response and liquefaction, particularly in recent years. The number of investigators involved in this previous research is to high to allow individual citation here. This project differs from those in the breadth of its scope (considering different soil profiles, using different types of nonlinear and equivalent linear soil models, considering different hazard levels) and in the fact that it is carrying the analyses through to the EDP level (previous studies have not directly investigated effects of geotechnical uncertainties on structural response).

• Presentation of Van Nuys testbed tornado diagrams at 2003 PEER Annual Meeting (March, 2003)
• Completion of FOSM analyses for equivalent linear and nonlinear one-dimensional cases (May, 2003)
• Completion of Monte Carlo simulations for equivalent linear and nonlinear one-dimensional cases (September, 2003)

• Report describing effects of geotechnical uncertainties on EDPs for different soil conditions and hazard levels, considering different methods of soil modeling.