Determination of Optimum Post-earthquake Restoration Strategies for Highway Bridges by Markov Decision Process

Abstract

In this paper, a life-cycle cost (LCC) based Markovian framework for determining optimum post-earthquake restoration strategies of highway bridges is proposed. LCC analysis is performed for a highway transportation network as a whole considering seismic hazards on an infinite time horizon, and the optimum restoration strategy for each bridge is determined by a continuous-time Markov decision process (CTMDP) model. The indirect economic losses are calculated according to earthquake-induced traffic congestion which is estimated by the Frank-Wolfe algorithm. The state transition probabilities used in the CTMDP are derived from probabilistic seismic hazard analysis as well as fragility analysis. A 7-node, 12-link transportation network with three bridges is studied to demonstrate the application of the proposed model.