Pavement project optimization and analysis

To meet the increasing travel demand and reduce the huge amount of cost in pavement construction and rehabilitation, there is a need to improve pavement project management in three aspects. First, the design models should be able to predict the performance, condition and structural capability of design alternatives. Second, several new factors in the revised America Association of State Highway and Transportation Official (AASHTO) Guide, such as reliability, environment, drainage, resilient moduli and soil support, should be included in a design working system. Third, the modern optimization methods could be applied in life-cycle design and cost analysis to find the optimal solution, in terms of minimum cost, while satisfying the design criteria. This dissertation focuses on the three aspects and developed a PC Windows-based computer system: "Optimization for Pavement Analysis" (OPA), for flexible pavements. This system contains both the AASHTO model, and the mechanistic multi-layer elastic model and is capable of predicting pavement distresses of roughness, fatigue and rutting. In this research, a numerical algorithm based on the Newton-Raphson algorithm is developed for the solution of the AASHTO design model of pavement structures. The ELSYM5 program is re-written in C++ and interfaced with the system to predict pavement fatigue cracking and rutting. A non-linear dynamic programming optimization algorithm is developed for a pavement life cycle design and cost analysis, which considers highway agency cost and/or highway user cost. The dynamic depth-first search based on Artificial Intelligence is developed to solve the non-linear dynamic optimization problem. The results of the case studies indicate that optimal design strategies vary with traffic and materials during an analysis period.