Multiobjective Optimization for Project Selection in Network-Level Bridge Management Incorporating Decision-Maker’s Preference Using the Concept of Holism

Abstract

Multiattribute utility functions are widely used for multiobjective optimization of project selection problems in network-level bridge management, based on their capability of incorporating multiple performance measures and decision-makers’ preferences. However, the existing way of using them needs to be improved. First, the addition of individual projects’ utility values to obtain a total utility score for a candidate project portfolio may lead to biased solutions because such additivity property is not always valid. In addition, there is an increasing need to express the network-level performance of bridges in terms of raw performance measures, which are most meaningful to decision makers, to reflect the impacts of systemwide investment. To address these issues, the paper presents a method that first evaluates the network performance of each candidate project portfolio using network-level performance measures prior to employing a multiattribute utility function and then identifies the optimal portfolio with the best network performance. This method effectively incorporates decision-makers’ preferences into decision making, avoids possible bias from nonassumption of additivity, and interprets investment performance directly in terms of raw performance measures.