Research On Batch Scheduling Problems In Petrochemical Industry

China is one of the most important countries in manufacturing and consuming petrochemical products. China’s annual yields of ethylene, synthetic resin products, refined oil, chemical fertilizers and pesticides rank the top level in the world. During the rapid developing procedure, the long-term accumulated contradictions and issues are becoming apparent increasingly. The contradictions and issues mainly lie in areas such as low production efficiency, bad intensive, large resource consumption, and high production costs. To overcome these issues, petrochemical industries need advanced ideologies in the aspect of operation management in order to optimize and organize production process. Production scheduling is an important part of the petrochemical operations management. Reasonable and effective production scheduling can increase productivity, reduce production costs and improve product quality so as to ensure that production systems work within a good condition. Therefore, modelling and optimization methods that suit petrochemical production scheduling problems have become one of hot research topics in both industrial and academic areas.In general, petrochemical products are produced in batch mode. The so-called batch is a set of objects that are produced on the same facility using the same production recipes and same operational conditions. This dissertation takes the intermittent batch process of petrochemical industry as research background, and study on modeling techniques and approximate solution methods for such batch scheduling problems under single unit, flowshop, flexible flowshop and job shop environments. The main contents include:1) The single unit batch scheduling problem with precedence constraints is studied, and its objective is to minimize the total weighted tardiness. After descript and analyse precedence constraints, the principle is proposed to judge whether a feasible solution exist. In addition, concidering tardiness is decided by completion time and deadline, the computed formula is proposed to give the upper and lower bounds of completion time in order to reduce the search space and the scale of the problem. Based on the above analysis, a discrete time model with precedence relationship is established and an improved particle swarm optimization algorithm is proposed. In the particle swarm optimization algorithm, large-scale neighborhood search is introduced to further improve the performance of the whole algorithm. A continuous coding method with the selection probability of batches is designed to map the coding of the particle swarm optimization algorithm with the solution of the problem. In order to avoid invalid search infeasible solutions, such strategies as initial population generation, scheduling generated strategy and boundary repair strategy are proposed.2) The integrated batching and scheduling problem for flowshop with the conversion rate between input and output of materials is stuieded, and its objective is to minimize the maximum completion time. This paper has taken two modeling strategies. As for the first modeling strategy, the conditions are given to meet the material and inventory constraints. And then determine precedence relation of batches and establish continuous-time model. Because the model has nonlinear constrain, the linear conversion method is proposed to change the model into a mixed integer linear model. The model is solved by standard optimization software. As for second modeling strategy, a series of events in each unit is given out. And then analyze the characters of the problems and establish a continuous-time model. For the number of event is greater than the number of batches and some events are not assigned to batches, the principle to judge invalid events is proposed to reduce the scale of the problem. An improved particle swarm optimization algorithm is proposed. In the particle swarm optimization algorithm, two-dimensional coding represent two information of the batch size and conversion rate to enhance the ability of handling the constraints. Repair, drift and scheduling generation strategies are proposed to obtain feasible solutions. The operators of contraction and divergence are introduced to maintain the diversity and convergence of the particle.3) The integrated batching and scheduling problem for flexible flow shop in each stage is studied, and its objective is to minimize the maximum completion time. For batch relation includes material and inventory limitations in the longitudinal direction and production capacity allocation and product task switching in the horizonal direction, the problem is very complex. To simplify the problem, analyse on the relationship of the above constraints and the beginning time to process batches. And divide the relationship into the minimum time span and the maximum time span which is called as the time window of batches. Then all the constraints are converted into the time window. Batches in the units can be seen as operation in a project. A resource-constrained project scheduling method with time windows of batches is proposed to solve the problem. An improved particle swarm optimization algorithm is proposed. In the particle swarm optimization algorithm, simplex method is introduced to further improve the performance of the algorithm, which can speed up closer to the optimal solution and jump out of the local scope to increase possibilities of the optimal solution by reflection, expansion, compression and contraction. In the aspect of parameters, the weight fells with nonlinear reduction strategies to enhance the convergence of the particle. In the construction of the algorithm, continuous coding represent the total quantityof each product in each unit and the particle recovery strategies ensure to meet the demand and the maximum capacity constraints.4) The batch scheduling problem for jobshop with decomposition, synthesis, cross and recycle process is studied, and its objective is to minimize the maximum completion time. The character of this problem is the complexity of the material flow. The relation between the tasks is too much. The following relation has to be considered:As for decomposition process, a task produces a variety of intermediate products to feed the several tasks at the next stages. As for synthesis process, a variety of intermediate products from pre-tasks feed a task. As for cross process, a product supplys to multiple tasks, or comes from multiple tasks. As for recycle process, a material or intermediate product considers the input not only from the first stage but also from the later stage. In order to deal with these relationships, a separation strategy of batches is proposed to depart a batch into two operations of begin and end. And then, the time relation of batches are converted into the minimum time span in order to obtain a feasible batch sequence which can meet material and inventory limitations. A timing and repair strategy is proposed to solve the problem. An improved particle swarm optimization algorithm is proposed. In the particle swarm optimization algorithm, quantum evolutionary algorithm is introduced to further improve the performance of the algorithm.