Study On Scheduling Of Supply Chain With Batch Chemical Plant Under Market Demand Uncertainty

Many chemical plants use batch plant mode, since batch plant can adjust the production schedule according to the market demand changes. With an increasingly competitive market and the continuous increase in production costs, chemical plants must optimize the production process, reduce production costs, and improve market competitiveness. Market demand fluctuations have a significant impact on optimized results, so the traditional deterministic optimization method is no longer applicable. In addition, products from chemical plant are transported to the consumer markets through the distribution network of the supply chain, where products is saled out and acquire economic interests, and therefore only batch plant scheduling optimization or distribution network scheduling optimization, ignore the interaction between them, will lead to the lack of global optimization for the supply chain with a batch chemical plant scheduling problem, not effectively reduce production costs, storage costs and transportation costs, resulted in the loss of economic benefits. Therefore, the integration of batch plant scheduling and distribution network scheduling under the uncertain market demand has important theoretical and practical significances.The scheduling of supply chain with batch chemical plant under market demand uncertainty is studied in this dissertation. The main content of this paper is listed as follows:(1) The scenario tree is employed to describe the progressive realization of the uncertain market demand over the scheduling time horizon, which generates a called demand scenario tree. A concept called demand program set is proposed, which is used to describe the horizontal and lengthways relationships between the nodes of the scenario tree. Based on the demand scenario tree and demand program set, a multi-stage stochastic programming model is built.(2) The multi-stage stochastic programming model is large-scale, included many variables, discrete variables and constraint equations, which leads to considerable difficulties to solve. To overcome the great difficulty of the calculation, the periodic approximation solution (PAS) based on the aggregation and decomposition of the scenario tree is proposed. A set of deterministic models obtained from the aggregation of the scenario tree are used to search for potential candidate scheduling decisions for the first period of scheduling time horizon. The decomposition of scenario tree makes the original large-scale MILP problem be decomposed into several small-scale MILP sub-problems, which largely reduce the computation load. Then three-stage stochastic shrinking-horizon solutions can be applied to acquire the other periods scheduling decisions. PAS employs both the scheduling decisions and sales decisions recourse strategy, and can acquire the optimized complete scheduling decisions for all the periods of the scheduling time horizon.(3) From the case studies, the effectiveness and efficiency of the proposed periodic approximation solution is verified, which provides optimal scheduling decisions than other traditional approaches, and also suffers lower computation times than the rigorous multistage stochastic model.