رویکرد مبتنی بر بهینه‌سازی - شبیه‌سازی - بهینه‌سازی برای کاهش واریانس فعال در مونتاژ

This paper addresses the economic benefits of selectively assigning a batch of subassemblies to each other after inspecting and correcting them as needed. Our work is based on optimizing the collective cost of subassembly inspection, rework, scrap, final assembly failure, and the act of subassembly mating. The expected value for the cost is estimated using Monte Carlo Simulation and optimized using a metaheuristic. After each simulation replication where we simulate a batch of subassemblies, we assign the inspected subassembly parts so that the rolled yield throughput is maximized. The complexity of this work is attributed to the fact that we solve an optimization problem for an objective that is estimated using simulation, and in each simulation replication there is another optimization problem to be solved for selective assembly. Significant improvements in assembly lines are predicted to be accomplished when this work is integrated in a real production environment.

Often times, we hear the slogan of “doing things right the first time!” This slogan was taken into account during our work on this research. A successful manufacturing firm—in nowadays highly competitive manufacturing environments—is the one that implements agile production concepts, with cost minimization and variation reduction. We aim to achieve the cost reduction by minimizing the inspection and assembly costs. We also reduce the variation by integrating inspection planning with subassembly mating so that we can produce the least possible failed final assemblies. Finally, we seek to do-things-right-first-time by using simulated data when historical data are not available to develop the inspection plans.

We provide a solution to proactively reduce variation of newly introduced assembly designs without the need to do actual prototyping and physical tests. An optimization algorithm was used to solve the inspection planning problem and obtain corrective plan limits, frequency of inspection, and the decision whether to implement a particular type of selective assembly (DTM) or not. The challenge we faced here is that we were solving an optimization–simulation–optimization (OSO) problem since we were optimizing an expected value that was estimated using Monte Carlo Simulation, and in each simulation replication we were solving another optimization problem. In fact, even the combination of simulation and optimization is still considered a gray area, where not many papers were published on the topic. Our findings and conclusions can be summarized as follows: • We presented an approach in this work for reducing the inspection planning cost by integrating the inspection plan with subassembly mating in a dynamic-basis. As we expected, we verified by several numerical examples that the Total Cost (TC) can be substantially decreased when implementing the proposed hybrid variation strategy. • The accuracy of the solution depends on several parameters to be chosen. One of those parameters is the execution time for the simulated annealing algorithm to solve the Dynamic Throughput Maximization. The longer the execution of SA, the more likely possible savings would be discovered. The Dynamic Throughput Maximization (DTM) procedure would be more beneficial when the failure cost of the final assembly is high. The failure cost can be minimized by implementing rework/scrap procedure, DTM procedure or a combination of both.