Markov decision process approach for multiple objective hazardous material transportation route selection problem

We consider a route selection problem for hazardous materials (hazmats) transportation. The objective is to select the route that minimises the consequence (cost or population) of undesirable events. We propose a Markov decision process (MDP) approach for solving this problem. First, an MDP model with single objective is described and the existence of an optimal policy for hazmat route selection problem is presented. Later, a generalised framework is presented for solving this problem with multiple objectives. We introduce a methodology to estimate the input parameter for the MDP models using the incident data and variables found in the Department of Transportation website. Finally, numerical examples are discussed for the parameter estimation and the multiple objective MDP model using a test network.