Integrated Order Selection And Production Scheduling In A Two-machine Flow Shop

Rapid progress in science and technology leads to an obviouslyincreased productivity and a significantly reduced production cycle. As aresult, the product categories are increasing in a large scale and thecustomer demand is in a trend of diversification.In order to improve their own competitiveness, many manufacturingenterprises are transforming their production strategies from the productionmode of MTS (Make-To-Stock) to MTO (Make-To-Order).Under the MTO strategy, the planning horizon is very limited. As aresult, manufacturers cannot select all but to choose parts of the whole setof potential orders. Besides, the make-span of all the selected orders isrelevant to the processing sequence of the order. Thus, manufacturers arefacing a problem about how to make integrated decisions of order selectionand production scheduling. This thesis focuses on the problem in atwo-machine flow-shop and separates the research content into two partsand provides with different solutions depending on whether the potentialselective orders belong to a single class or multiple classes.For the research on single class orders, it elaborates a detailedanalysis of the research on how to select and sequence the orders undervery limited planning horizon with no consideration of setup timesbetween the consecutive processing orders. In this part, firstly it formulatesthe problem as a MIP (mixed integer programming) model and examinesthe model with a concrete example. Then, it puts forward an independentdecision making scheme on how to select and sequence the ordersseparately. Through a data experiment, it compares the value of integrateddecision-making (i.e. result from running the model algorithm) and that ofthe independent decision making scheme, finding the model serves a much better solution. But with the increase of the order size, model algorithmwould no long applicable to the situation with a larger size of orders due toa significant increase on model algorithm complexity. In order toefficiently solve the problem the larger size of orders, this thesis designs aheuristic algorithm on selecting and sequencing orders based on the timevalue rate of the orders and tests its validity through a data experiment.The result shows that the profit gap between the heuristic algorithm andthe model solution is no more than4.52%.For the research on multiple classes of orders, by analyzing thedetailed difference from the single class order problem, it formulatesanother MIP model, combined with a data example to examine the modellogic. In addition, in order to clear the complexity of the algorithm, itdesigns a detailed data experiment and the results are analyzed. Althoughthe model can get optimal solution to a small scale problem, thecalculation time increases significantly as the order size increases.Therefore, this thesis also puts forward a heuristic algorithm to solve thelarger scale problems. Besides, the heuristic algorithm is proved efficientthrough a validity check.