Research On Production Planning Of Multi-product Multi-stage Manufacturing System

With severe global competition and rapid changes in market, the mode of multi-product small-batch and short-period production is becoming more widely applied in the manufacturing industry, and in the meanwhile, the customer satisfactory and production economy are attached more attention. To tackle with these new changes, most of the manufacturing enterprises are building up multi-product multi-stage manufacturing systems with complex and flexible topology structure. Finding the optimal productin planning for such manufacturing systems will obviously improve the flexibility and efficiency, and provide an effective tool to deal with these challenges from increasingly complex cumstomization requirements and diversified production modes, which will contribute to high-efficient production resource allocation, high-quality orders completed on time, achieving enterprise strategic and tactic objectives, and so on. This thesis firstly gives a brief review of recent researches on manufacturing enterprise production planning problem, and points out some difficulties and challenges in production operations under the new situations. After that, using the mathematical programming method, the modelling and optimization of production planning for multi-product multi-stage manufacturing system is analyzed and discussed in depth. Several case studies are presented to verify the feasibility and effectiveness of the proposed strategies, which provide guidance for the manufacturing enterprises to optimize the production operations and enhance the efficiency to response to the market. The main contents and major contributions of this thesis are described as follows:1) A novel optimization model for production planning of multi-product multi-stage manufacturing system is developed according to the characteristics of multi-product small-batch production and operations conditions, which takes into account the lead times of raw material supply, each production stage and product delivery. The production route for each product is depicted by aggregating the sets, which construct the interactive constaints between the nodes and the super structure between raw materials and products. The model is to minimize the total cost within the planning horizon, and considers various conditions involving setup, raw material supply, inventory holding, backlogging, production and inventory capacity, production logic, and batch. A rolling horizon optimization strategy is proposed to solve the problem. As shown in the case study, the model is valid to decrease inventory and backlogging level; and the rolling horizon strategy ensures the efficiency and accuracy of the solutions.2) A novel approach to define the customer responses when stockout occurs is presented, which considers three scenarios including backlogging, lost sale and simultaneously allowing partial backlogging and lost sale. Three distinct mixed integer nonlinear programming (MINLP) models are formulated to address the production planning of a multi-product multi-stage manufacturing system under these scenarios. The model fully considers the influence of multi-period backlogging and lost sale on the production planning, and constructs piecewise function to depict the time critical point of partial backlogging and lost sale. A linerziation reformulation strategy is presented to transform the original MINLP models into mixed integer linear programming (MILP) models resulting in high solution efficiency. By analyzing the influence under different scenarios when stockout occurs, the decision-makers can obtain desirable trade-off between production and delivery.3)To gain an insight into the customization and flexible manufacturing for a steel rolling mill multi-product multi-stage manufacturing system, the influence of substitution in steel slabs and plates on the production and delivery is analyzed. Three feasible substitution schemes for steel plates are presented, by which the multi-route matching relationship between steel slabs and plates are constructed. A novel muti-period MINLP model is developed to address this production planning problem considering material substitutions. A linearization strategy and a decomposition iteration approach is proposed to solve the original MINLP model which results in better quality and high efficiency. The case study compares the results by different solution strategies, and verifies their effectiveness. The optimized production planning enhances the flexibility of production and delivery by material substitutions.4) A novel chance constrained programming approach is developed to model the production planning problem for a multi-product multi-stage manufacturing system under various uncertainty. The model addresses the uncertain conditions originating from internal and external factors including production efficiency, penalty cost of backlogging and safety stock deviations, raw material supply and market demand. Besides, the model additionally considers the constraints of production family and safety stock with penalty. By inducing confidence levels, the stochastic model is converted into an equivalent crisp MILP problem using a direct and an indirect approach. Compared with a mean method and a margin method, the result indicates that the proposed approach is flexible and that the trade-off between minimizing the cost and controlling the risk is efficient for production planning under uncertainty.5) A fuzzy multi-objective optimization model and approach is proposed to address a production planning problem with preference on the objectives in an uncertain multi-product multi-stage manufacturing environment. The model attempts to simultaneously minimize the relevant operations cost and maximize the average safety stock holding level and the average service level. The internal and external uncertainty is simultaneously considered and is formulated into a fuzzy multi-objective optimization model. The priority levels are constructed for the objective preference by the decision-makers. A weighted average method and a fuzzy ranking method are adopted, and a two-phase fuzzy optimization approach is developed to manage the preference extent and convert the fuzzy model into an auxiliary crisp one. Then a novel interactive solution approach is proposed to solve this problem. Based on the requirement of the objective preference extent, the decision-makers are capable to obtain satisfactory solutions through the interactive approach.