ABSTRACT
This chapter discusses the estimation and optimization of stochastic life-cycle performances of deteriorating engineering systems. To model the effects of deterioration processes, state-dependent stochastic models (SDSMs) are adopted. The SDSMs capture the effects of multiple deterioration processes and their interactions by modeling the changes in the system state variables. Then, capacity and demand models that take the time-variant state variables as input are adopted to convert the impact of the deterioration processes into the time-variant capacity, demand, and other system performance indicators. The SDSMs are then integrated into a renewal-theory life-cycle analysis (RTLCA) to efficiently evaluate life-cycle performance quantities such as availability, total cost, and benefits of operating the system. A stochastic simulation-based approach is adopted to estimate the time-variant performance indicators needed to inform intervention activities for the life-cycle analysis. To efficiently estimate and optimize the life-cycle performances (i.e., the selection of optimal value of intervention criterion), a novel sampling and kernel density estimation (KDE) based approach is presented for efficient approximation of the Probability Density Functions (PDFs) in RTLCA for any selected value of intervention criterion. The sampling and KDE based approach requires simulation of only one set of samples, which are used to establish the information for evaluating the life-cycle performance for any selected value of the intervention criterion, leading to a significant computational efficiency. As an illustration, the sampling and KDE based approach is used to evaluate and optimize the life-cycle performances of an example reinforced concrete (RC) bridge subject to deterioration due to corrosion and seismic loading.
