Reliability-based life-cycle cost design of asphalt pavement using artificial neural networks

It is widely recognized that reliability-based optimization methodologies are rational and promising tools to perform the life-cycle management (LCM) of civil engineering structures. In order to realize the optimal design and maintenance of asphalt pavement, a comprehensive reliability-based optimization methodology considering the life-cycle cost (LCC) is proposed in this paper. Considering the powerful ability of Artificial Neural Networks (ANNs) to solve complex and nonlinear problems, ANNs are implemented to predict the performance of asphalt pavement based on the training data (i.e. structural, traffic, climatic, and performance parameters) selected from the Long-Term Pavement Performance (LTPP) program. Monte Carlo simulation (MCS) with Importance Sampling (IS) is conducted based on the obtained ANNs model to calculate the life-cycle reliability of asphalt pavement. Finally, the expected LCC including the initial construction cost, the preventive maintenance cost, inspection cost, users cost, and salvage value is minimized while maintaining a prescribed life-cycle reliability level for asphalt pavement. Several applications are presented to investigate the effects of traffic and climatic parameters on reliability-based optimum cost solution.