Methodology for Risk-Based Maintenance Optimization
The proposed methodology for network-level bridge deck maintenance optimization is based on the simultaneous satisfaction of three relevant and competing criteria, namely: (i) maximization of condition rating; (ii) minimization of maintenance costs; and (iii) minimization of user costs. The development of such a methodology requires the integration of three simple and practical decision support models, namely: (i) qualitative condition assessment model; (ii) qualitative deterioration prediction model; and (iii) multi-objective optimization model to determine the optimal maintenance strategy for a network of bridge decks. The proposed methodology is symbolically outlined in Fig.1. The qualitative condition assessment model enables to take full advantage of the available bridge inspection data. The deterioration of bridge decks is modeled using an appropriate stochastic process that captures the time-dependence and uncertainty of the deterioration mechanism. The deterioration prediction model is compatible with the existing condition assessment procedure, and is developed based on the historical field performance data collected during bridge inspections.
As mentioned earlier, the governing failure modes for bridge decks are the loss of serviceability and loss of functionality due to corrosion-induced damage. The consequences of bridge deck deterioration can range form a simple riding discomfort to a loss of life as a result of a traffic accident on the deteriorated deck or on a detour route (with a poor condition) due to the closure of one lane or the entire bridge during its maintenance or due to its excessive deterioration. As opposed to life cycle cost or cost-benefit criteria used in most bridge maintenance optimization studies, the use of a risk of failure as a criterion for maintenance optimization is more rational and relevant, however its
Fig. 1. Schematic of multi-objective maintenance optimization approach for bridge decks
implementation is not easy given the complexity of assessing the consequences of failure in monetary terms This means that monetary values need to be assigned for fatalities, injuries, and social costs which are not easily quantified, and various methods have been developed.
Given the difficulty of accepting the notion of placing any sort of value on human life, Starr (1969) evaluated the risk of death from various causes and identified two general categories for risk of death: (i) risk associated with voluntary activities in which the individual evaluates and adjusts his exposure to risk; and (ii) risk associated with involuntary activities, which are determined by regulations from governmental agencies. Starr (1969) indicated that the public typically was willing to accept voluntary risks 1,000 times greater than involuntary risks. Pate-Cornell (1994) proposed different ranges of acceptable levels of risks for the public and workers ranging from 10~8 to 10~3 per year.
The consideration of these three objectives within the multi-objective optimization framework is a practical approach for the solution of the risk minimization problem through the minimization of the probability of failure (by maximizing the condition rating) and minimization of the consequences of failure (by minimizing the maintenance costs and user costs), as indicated in Fig.1. The proposed approach overcomes the difficulties of assessing quantitatively the probability of failure and the consequences of failure for a large network of bridge decks. A detailed description of the models is given in the next sections.