A Methodology for Developing a Performance Model for Pervious Concrete Pavement

Pavement performance models have been mainly developed based on long term pavement condition data. Developing performance models is a challenging task, especially, in case of a new type of pavement such as Pervious Concrete Pavement (PCP) which has been recently investigated and limited relative data is available. This paper proposes a methodology to develop performance models for the PCP incorporating limited data. To deal with incomplete data, soft computing techniques can provide some assistance. Firstly, this paper proposes fuzzy sets to quantify severity, density, and weighting factors of pavement distresses to define a novel condition index for the PCP. Secondly, a combination of homogeneous and non-homogeneous Markov model has been proposed to develop a realistic performance model incorporating the condition index. To build up Markov model, transition probability matrices are presented utilizing normal distribution functions. Thirdly, the Latin Hypercube Simulation (LHS) technique is employed to incorporate the probability distribution function operations to compute the future condition of the PCP. The future condition of the PCP is expressed by single expected values (deterministic model) and compared with probability distribution functions (probabilistic model). Finally, a semi-Markov model has been proposed as an efficient alternative tool to improve the Markov process by reducing a large number of transition probabilities. Overall, this paper presents four modeling techniques that are applied in an interactive manner to provide a methodology to predict PCP condition.