Combining a Multiattribute Value Function with an Optimization Model: An Application to Dynamic Resource Allocation for Infrastructure Maintenance

A key decision in infrastructure management is the allocation of resources to maintenance activities that consist of periodic rehabilitation actions and routine day-to-day operations. These activities improve the quality of different assets and operations. They also differ in terms of their objectives, costs, and life-cycle characteristics; yet they all impact the same infrastructure system and compete for resources from the same budget. To support the allocation of resources to these activities, we present a generic resource allocation model that we developed for the Finnish Road Administration (Finnra) by building and interlinking (i) a preference model, which yields the aggregate value of maintenance activities by applying multiattribute value functions to the quality distributions, of assets; (ii) a life-cycle model, which captures the deterioration-improvement dynamics associated with the maintenance activities; and (iii) an optimization model, which generates funding recommendations for maximizing the aggregate long-term value of maintenance investments. The optimization results were explored in facilitated workshops where on-the-fly computations gave senior managers insights into how the recommendations depended on preferences and budget levels. The case study was awarded for an outstanding achievement in Finnra's research program, and it was also recognized as a Finalist for the Decision Analysis Society Practice Award in 2007.