Use of Soft Computing Applications to Model Pervious Concrete Pavement Condition in Cold Climates

Abstract

Development of an adequate performance model based on an appropriate condition index has been a major challenge for engineers particularly in case of a new type of design such as pervious concrete pavement structures (PCPSs) which suffer from limited long term performance data sets especially in colder climates. Soft computing techniques are significantly efficient at dealing with subjective, incomplete, and limited data. This paper proposes three most effective soft computing methods: fuzzy sets, the Latin Hypercube Simulation technique, and the Markov Chain process. A novel comprehensive condition index based on severity, density, and weighting factors of distresses occurring on PCPS has been developed incorporating fuzzy sets. A combination of homogeneous and nonhomogeneous Markov Chain has been applied to develop performance models. Transition probability matrices are presented using probability distribution functions rather than single values. A simulation technique is then used to incorporate the probability distribution function operations to compute the future condition of the pavements. The future performance of the pavements is expressed by both single expected values and suitable probability distribution functions. Ultimately, a probabilistic versus deterministic performance curve is presented.