Sensitivity of optimal hazmat routes to limited preference specification

The authors show that a multi-attribute utility (MAU) formulation of the single shipment, cost-exposure, hazardous materials optimal routing problem can conceivably be solved with traditional shortest path algorithms. They argue, however, that the parameter scaling and single attribute cost disutility function into the two attribute cost-exposure disutility function could not be estimated to more than an order of magnitude. In a numerical study using the Columbus, Ohio highway network, the authors see that such an estimate is not precise enough either to specify a unique optimal route or to screen out a substantial number of routes as suboptimal when the optimal route is to be other than the minimum cost or minimum risk of exposure route. The number of routes obtained when varying the parameter across its permissible range is small enough, however, that the MAU formulation can be thought of as one that generates ‘MAU noninferior routes’ before interacting with the decision maker. The authors speculate on the impacts of extending the problem to multiple shipments and more attributes.