Efficient optimization of a large-scale biorefinery system using a novel decomposition based approach

Abstract

Biorefinery design problems that consider processing and supply chain together are generally large-scale mixed integer linear programs (MILP) that are computationally difficult to solve. This work considers one such large-scale problem and proposes a solution method to solve it efficiently. The proposed method utilizes the Dantzig-Wolfe decomposition framework and a novel heuristic to simplify the original problem. This simplification is done using the sub-problem solutions obtained within the Dantzig-Wolfe decomposition iterations. The resulting simplified problem can be easily solved using standard procedures to obtain the final solution for the original problem. The proposed method is observed to be up to 92% faster than the standard CPLEX® MILP solver and can solve large-scale problems which are not solvable using standard approaches. A large-scale biorefinery design problem for the production of ethanol from lignocellulosic biomass in Maharashtra, India is solved. For an ethanol demand of 20 million kg/month, the minimum ethanol cost was determined to be INR 47.4 per litre. This method enables the formulation of more comprehensive biorefinery design models and can be employed on similarly structured large MILP problems in other fields.