Condition-Dependent Maintenance Effectiveness in Dynamic Performance Models for Transportation Infrastructure

Abstract

Dynamic performance models that combine performance prediction and maintenance effectiveness are required for state-of-the-art optimization techniques such as optimal control. Because records of maintenance effectiveness depend on facility condition, nonlinear models are necessary to include interactions between variables to account for this dependence and estimate condition-dependent maintenance effectiveness. Therefore, this paper proposes a procedure for estimating nonlinear dynamic performance models that capture interactions between variables using panel data. The relationships between maintenance effectiveness and current facility condition and between structural design and traffic impact were found to be polynomial in a numerical example of highway pavements. It was also demonstrated that imposing physical constraints on the maintenance effectiveness based on external data sources generated more reasonable models when self-selected samples were used for estimation.