Abstract
The impact of various ground motion estimators on performance predictions was investigated within this study by performing a damage-level implementation of the PBEE framework for Lenihan Dam, an embankment dam located in California. Due to the importance associated with maintaining adequate crest height, this study developed a suite of PBEE frameworks around vertical crest deformation, or crest settlement, as an estimator for physical response. Ground motion parameters were estimated in terms of cumulative absolute velocity (CAV), peak ground acceleration (PGA), and peak ground velocity (PGV). Damage was estimated with the use of damage state indices corresponding to a damage classification system. A damage-level implementation of the PBEE framework was developed using the PEER methodology for a performance-based analysis of Lenihan Dam based on PSHA results of a generic site location. The PEER methodology was applied to develop three unique framing equations which ultimately reported performance in terms of a damage classification as an estimator for damage (i.e., DM). Each framing equation was developed around crest settlement as an estimator for physical response (i.e., EDP) using one of three IM’s to characterize ground motion: CAV, PGA, or PGV. A suite of response and damage models was developed using the probabilistic seismic hazard predictions determined through OpenSHA and probabilistic models relating each IM to SET. OpenSHA proved to be a useful tool for developing a probabilistic prediction of seismic hazard at Lenihan Dam. The NDA results from Armstrong (2018) were used to develop each of the probabilistic models relating SET to CAV, PGA, and PGV. The damage models were developed from the predictions of the response models as well as the probabilistic relationships relating a DC of 1, 2, 3, 4 to SET. The probabilistic relationships were developed from the seismic dam performance data collected in Pells and Fell (2002). The performance predictions from each of the IM-based PBEE frameworks suggest that the selection of the ground motion estimator does impact the predictions at each level of the framework but the magnitude of difference between each set of the IM-based predictions varies among and within the response-level and damage-level of the PBEE framework. The differences between the response-level performance predictions in terms of SET increased with the magnitude of SET and were deemed to be negligible. The differences between the damage-level performance predictions in terms of DC were observed to vary slightly but otherwise negligible among each DC index.