Investigation of Sensitivity of Building Loss Estimates to Major Uncertain Variables for the Van Nuys Testbed, PEER Report 2002-03


A major component of a building-specific seismic loss analysis is the estimation of repair costs in future earthquakes. A number of uncertain variables contribute to the uncertainty in these cost estimates. Among these are ground-shaking intensity, details of the ground motion, mass, damping, and
force-deformation behavior, component fragility, repair methods, contractor’s direct costs, and contractor’s overhead and profit, among others. This report addresses which of these significantly contribute to the overall uncertainty in future economic performance. We examine gross sensitivity by measuring the variation (or swing) of the economic performance when each variable is taken at its assumed median value and at its extremes, e.g., the 10th and 90th percentiles.

Such a study is undertaken for a 1960s nonductile reinforced-concrete moment-frame building located in Van Nuys, California, which is one of two buildings studied by the PEER testbeds program. Here, economic performance is measured in terms of the repair costs associated with the (uncertain) highest shaking intensity the site will experience in the next 50 years. Repair costs are estimated using the assembly-based vulnerability (ABV) method and the site’s seismic hazard.

We do not address all uncertainties. Notable among those excluded are the selection among competing models of hysteretic behavior of structural elements, the potential that fragility tests of structural and nonstructural elements do not accurately reflect actual field conditions, the selection among competing repair methods given a component damage state, the choice of nonunion versus union labor to perform repairs, and the potential for repair costs after an earthquake to be increased by demand-driven inflation (a phenomenon often called demand surge). Except for demand surge, all of these un certainties can be examined in future ABV analyses.

The study shows that among the parameters considered here, the top three contributors to uncertainty in earthquake repair cost, in decreasing order, are assembly capacity (i.e., for a building element, the relationship between physical damage and the relevant engineering demand parameter), shaking intensity (measured here in terms of damped elastic spectral acceleration, Sa), and details of the ground motion conditioned on Sa. Uncertainties in parameters of the structural model contribute modestly to overall uncertainty in economic performance, and are comparable in importance to the uncertainty in the unit costs that a contractor will experience in repairing the damage.

These observations are based only on the demonstration building, but they do offer intriguing implications for performance-based earthquake engineering. If duplicated for other buildings, it may be that much performance uncertainty could be reduced through more-detailed study and modeling of building component damageability. In addition, seismic loss analyses might reasonably neglect uncertainties in structural modeling parameters, without substantially underestimating overall uncertainty.

Full List of PEER Reports: click here.

Keith A. Porter
James L. Beck
Rustem V. Shaikhutdinov
Publication date: 
April 3, 2002
Publication type: 
Technical Report
Porter, K. A., Beck, J. L., & Shaikhutdinov, R. V. (2002). Investigation of Sensitivity of Building Loss Estimates to Major Uncertain Variables for the Van Nuys Testbed, PEER Report 2002-03. Pacific Earthquake Engineering Research Center, University of California, Berkeley, CA.