ABSTRACT
Decisions concerning bridge management are often taken based on engineering judgement and heuristics. Although experienced decision makers can identify the optimal action by intuition, we believe that an application of statistical decision theory would improve bridge management strategies and lead to a reduction of costs.
The bases of statistical decision-making have been studied since the formulation of axiomatic expected utility theory (von Neumann and Morgenstern 1944). According to expected utility theory, decisions are ranked in order of expected utility, which is calculated on the basis of the consequences and the probabilities of each possible scenario (Parmigiani and Inoue 2009). The rational decision-maker should regard as optimal, and therefore chooses, the action with the highest expected utility. Applications of expected utility theory have been considered in the field of civil engineering (Faber and Rackwitz 2004, Faber and Maes 2008); however, the previous research has been mainly focused on the evaluation of structural health monitoring systems (Zonta et al. 2014). The benefit of monitoring is usually quantified through the concept of value of information. Herein, we present an application of expected utility theory to a civil engineering decision problem in which the decision maker has to evaluate whether to perform an intervention of structural strengthening on a set of bridges before a defined overweight load is allowed to cross.
The case study we analyze concerns a set of bridges belonging to the strategic network of the Autonomous Province of Trento—a local administration in northern Italy. For each bridge, the design documentation and the construction cost are available and stored in a bridge management system database (University of Trento 2011). Before the decision maker chooses between performing the structural intervention or allowing the overweight loads to cross the bridge as it is, they can ask an estimator—a technician with experience in structural reliability—to investigate the condition of the bridge.
In this contribution, we present the assumptions and the procedures that can be carried out by the estimator in order to assess the structural condition. Then, we present the general framework of expected utility theory as proposed by the literature (Raiffa and Schlaifer 1961, Parmigiani and Inoue 2009). Next, we describe how the decision problem was modelled and how expected utility theory was implemented in order to obtain the solution. The risk profile of the Autonomous Province of Trento was taken into account to analyze the decision problem and to calibrate the numerical solution. The result is a set of thresholds defined in the space of the understrength, which quantify by how much the capacity of a bridge should be increased to equalize the overweight load effects. The calculated values enable the decision maker to select the optimal action at each stage of the decision problem, without further calculations.
