MPI Parallel Monte Carlo Framework for the Reliability Analysis of Highway Bridges

Abstract

Current emphasis on evaluating the resilience of civil communities to various natural and man-made hazards has required expanding the scale of structural analysis problems from performing deterministic analysis of isolated components to probabilistic analyses of large networks of interconnected structural systems and infrastructures. To control the enormous computational effort that the analysis of large-scale systems requires, researchers and engineers have often resorted to considerably simplifying the behavior of such systems. The drawback of the current approach is that extremely simplified models might not adequately predict the response of structures. Given the availability of more powerful computers at increasingly cheaper cost, it is now possible to create accurate models for large-scale systems and contain the computational requirements. As a consequence, probabilistic analyses of complex nonlinear models have now become possible. This paper describes a parametric object-oriented parallel message passing interface (MPI) framework for the structural reliability analysis of bridges for the effect of overweight traffic load. The framework consists of three modules that characterize bridge member capacities, develop bridge structural system models, and perform the reliability analysis using an efficient Monte Carlo strategy. The results of the analysis show that the hybrid approach considerably reduces the number of simulations when compared to a crude Monte Carlo simulation, running a full system reliability analysis in just few seconds and preserving good accuracy on the reliability index, on the order of 4–14%.