Development of Non-Parametric Deterioration Models for Risk and Reliability Assessments of Concrete and Timber Bridges

Abstract

Distresses in concrete and timber bridge components occur due to physical, chemical and mechanical factors that manifest as observable symptoms of deterioration. Deterioration models relate the observable defects to the several factors affecting the deterioration. Only limited studies exist on the development of non-parametric (NP) models for bridge management, which have the unique advantage of not requiring any assumption on the functional form of the models. To address this research gap, this paper develops non-parametric deterioration models for risk and reliability assessments of concrete and timber bridges based on two methods: (1) average time in condition rating (ATICR), and (2) Kaplan-Meier (K-M) estimates. This study proposes relative deterioration rates based on the ATICR of various concrete and timber bridge components. The methodology for developing NP deterioration models based on K-M estimates for probabilistic prediction of bridge safety while accounting for the partial information from the incomplete bridge condition observations is explained. This study also investigates the effect of multiple factors affecting the bridge component condition, including aging, traffic loads, environmental conditions, and design characteristics, and estimates the median survival years for reinforced concrete decks.