Recently, there has been a paradigm shift by certain energy and chemical companies towards modular manufacturing, whereby transportable modular production units can be relocated between production facilities to meet the spatial and temporal changes in the availabilities, demands, and prices of the underlying commodities. We refer to the optimal distribution, production, and storage of commodities, and the relocation and operation of the modular production units as the \textit{dynamic multiple commodity supply chain problem with modular production units}. To this end, we present a ``flow-based'' and a ``path-based'' mixed-integer linear programming formulation to model the problem. In an effort to solve large-scale instances of the problem, we propose an iterative three-stage matheuristic for the ``path-based'' formulation. In the first stage of the matheuristic, a feasible solution to the problem is generated by an integrated column generation and Lagrangian relaxation based heuristic. In the second stage of the matheuristic, a path-relinking procedure is utilized as a local search heuristic to further improve the solution. And in the final stage of the matheuristic, the Lagrangian multipliers are updated via a subgradient method. The effectiveness of the matheuristic is illustrated through numerical experiments with a set of randomly generated test instances. For the large-scale test instances, the results show that the matheuristic produces quality solutions orders of magnitude faster than a ``state-of-the-art'' mixed-integer linear programming solver.
Citation
Report Number: NA Institution Address: Texas A&M Energy Institute, 1617 Research Pkwy, College Station, TX 77843 Month/Year: April/2022