Ding‐Hsiang Huang
In modern interconnected infrastructures, ensuring the reliability of multicapacity networks (MCNs) is paramount, particularly as these networks accommodate components operating across multiple capacity states. This study introduces a novel approach that integrates Monte Carlo (MC) simulation with linear programming (LP) models to estimate the reliability of MCNs efficiently across all possible demand levels. This method leverages MC simulation to model the stochastic component capacities in the MCNs by generating numerous network configurations. LP models are then used to optimize the maximal flow for each simulated state. This methodology circumvents the computational intractability of exact analytical methods, especially in large-scale networks with diverse demand requirements. The algorithm’s effectiveness is validated through time complexity analysis and comprehensive numerical experiments on a benchmark MCN. For instance, our algorithm produced reliability estimates for the benchmark under the case K = 5 in approximately one second, whereas an exact algorithm failed to yield a solution in over three days. Further, a real-world application to the Taiwan Academic Network (TANet) underscores the algorithm’s practical utility in network design and improvement phases. With this approach, decision-makers can find optimal configurations and identify critical components to enhance. For instance, in the TANet case, our method determined that achieving a 95% global reliability threshold required a minimum of 37 components per arc or a component reliability of 0.980.