Constraint Programming Based Production Planning in a Petrochemical Industry

Abstract

In this article, we provide a constraint programming based optimization model to determine the optimal production planning to maximize profit in a petrochemical industry. The proposed model despite its simplicity overcomes the limitations of the mixed integer linear programming model available in literature. The benefits of the proposed model are demonstrated on various instances of two case studies and it is shown that the proposed formulation enables an increase in the profit in the range of 0.58 to 3.68%. A petrochemical industry uses series of complex networks to convert feedstock such as oil and gas to produce primary petrochemicals such as methanol, ethylene, propylene, benzene, toluene, xylene, etc. These primary petrochemicals are subsequently converted into petrochemical intermediates and derivatives which are ultimately transformed into products used in the market. The petrochemical industry operates at very high production levels which is achieved either through the large scale of the individual equipment or through the large scale of the entire plant (1). A variety of optimization models have been proposed for efficiently operating the petrochemical plants. These include efficient production planning (2), mergers and acquisitions (1), integration of refineries and petrochemical plants (3), capacity expansion (4), efficient spatial organization of petrochemical plants (5), efficient job scheduling (6), and optimal supply chain management (7). In this article, we use Constraint Programming (CP), an optimization technique that has its origins in the Artificial Intelligence and Computer Science Community (8), for solving the production planning problem in an industry. A Mixed Integer Linear Programming (MILP) model has been proposed in literature (2) to determine efficient production plans to guide the petrochemical industry development in Saudi Arabia. However, this formulation restricts the productions to artificial levels thereby leading to suboptimal profit. In this article, we propose a CP based optimization formulation which overcomes the drawbacks of the MILP formulation in literature. The benefit of the proposed formulation has been demonstrated on eight different instances involving two distinct requirements. In the next section, we provide a brief description of CP and this is followed by the description of the problem statement. In the next section, we review the MILP formulation in literature and describe its limitations. In the subsequent section, we present a CP based optimization formulation to solve the production planning problem and subsequently demonstrate the applicability of this formulation on various instances of two scenarios. We conclude the article by discussing the developments in this article and present possible future work in this direction.