The research objective of this award is to formulate and solve a tractable closed-loop supply chain network design problem that includes: (a) facility configuration that is robust to variations in the cost and constraint implications of possible carbon emission regulations; (b) transportation capacities of various modes that accommodate random variations in demands and returns; and (c) a product return policy that balances the tradeoffs in revenues, transportation costs, and emission-related costs. The problem's difficulty arises from the discrete character of facility investment and return policy choices, the nonlinearity introduced by seeking a robust design, and the large number of scenarios required to accurately represent stochastic demand and return quantities. By restricting how the transportation capacities depend on the uncertain carbon policy parameters, an affinely adjustable robust counterpart is obtained for the design problem with stochastic subproblems. The research tasks are to approximate the discrete solution to this robust counterpart and to efficiently reduce the number of stochastic scenarios to be considered.
If successful, the results of this research will help manufacturing firms to maintain global competitiveness by proactively designing their closed loop supply chains in anticipation of possible market mechanisms to control carbon emissions while mitigating the uncertainties associated with transportation demands. The innovative combination of robust and stochastic optimization will be widely applicable to capacity investment problems under uncertainties that differ fundamentally in character, depending on the existence of historical precedent and the repeatability of conditions. Results and insights derived from this work will enhance the training of graduate students in network and stochastic optimization as well as undergraduate students in engineering economic analysis. A related complex, ill-structured problem-solving experience will be developed for engineering economy courses and, to ease its adoption in large-enrollment settings, will be disseminated via open-source software tools.
