A STOCHASTIC MODEL FOR PAVEMENT PERFORMANCE AND MANAGEMENT

A methodology for the analysis and selection of pavement systems, given a set of goals and constraints is presented. A set of models at three different levels of analysis is discussed. 1) analysis of the physical behavior of the pavement system, 2) analysis of the distress and performance of the pavement system, and 3) analysis of the selection and optimization of maintenance activity given the initial pavement design configuration. The first involves a mechanical model of the pavement system, which is taken as a stochastic linearly elastic or viscoelastic layered halfspace subject to a static circular normal loading, which predicts the primary responses of the pavement system. These are the stress, strain, or deflection at any point in the halfspace. The second model uses the Boltzmann Superposition Principle to convert the primary responses into limiting responses under a realistic operating environment, with mechanical and phenomonological models to convert these limiting responses into cracking and permanent deformation. These distress parameters are then introduced into a probabilistic version of the AASHO serviceability equation, and the serviceability and reliability of the system are predicted. Changes in the state of the serviceability are modeled as a homogenous Markov process. The third level of analysis utilizes transfer functions to convert maintenance activities into improvement in system performance. Decision analysis, with traffic and maintenanced levels being the independent variables, is than used to rank and optimize various maintenance alternatives with or without budget constraints. In all of the levels of analysis, the system responses and performance are random variables, as is traffic, with environment maintenance level being deterministic variables. Sensitivity analyses performed at M.I.T. have shown that the trends predicted by the model are in reasonable agreement with the anticipated behavior of real world systems. /Author/