Stochastic performance modeling considering maintenance effects for highway pavement management.

摘要

This dissertation presents concepts of transportation infrastructure performance modeling. The concepts are applied to pavement performance modeling and prediction involving various maintenance interventions. Methodologies, including regression analysis using explanatory variables and/or time as regressors, time series analysis and intervention analysis, are explored. Effects of six maintenance strategies on roughness are investigated based on actual field data obtained from a state Department of Transportation. It is found that, in general, maintenance improves ride quality and therefore should be used annually in practice to maintain the serviceability of highway transportation infrastructure. Models for predicting future roughness in terms of International Roughness Index (IRI) after a specific maintenance strategy are developed, which can be incorporated into existing pavement management systems to provide decision support for optimal benefit-cost maintenance strategy selection. A methodology is proposed to predict present serviceability index (PSI). Stationary and piecewise stationary renewal processes are used to model mixed traffic loading. Uncertainty in PSI modeling and prediction is formulated using rigorous statistical methods. The theory can be applied to other transportation infrastructure where regular condition data is taken for which the trend can be analyzed as a performance index, and where maintenance interventions are made in an effort to improve predicted performance.