An efficient simulation optimization methodology to solve a multi-objective problem in unreliable unbalanced production lines

摘要

This research develops an expert system to addresses a novel problem in the literature of buffer allocation and production lines. We investigate real-world unreliable unbalanced production lines where all time-based parameters are probabilistic including time between parts arrivals, processing times, time between failures, repairing times, and setup times. The main contributions of the paper are a twofold. First and foremost, the mean processing times of workstations and buffer capacities, unlike the existing literature, are considered as decision variables in a multi-objective optimization problem which maximizes the throughput rate and minimizes the total buffer capacities as well as the total cost of the mean process time reductions. Secondly, an efficient methodology is developed that can precisely reflect a real-world system without any unrealistic and/or restrictive assumptions on the probabilistic nature of the system, which are commonly assumed in the existing literature. One of the greatest challenges in this research is to estimate the throughput rate function since it highly depends on the random behavior of the system. Thus, a simulation optimization approach is developed based on the Design of Experiments and Response Surface Methodology to fit a regression model for throughput rate. Finally, Non-dominated Sorting Genetic Algorithm (NSGA-II) and Non-dominated Ranked Genetic Algorithm (NRGA) are used to generate high-quality solutions for the aforementioned problem. This methodology is run on a real numerical case. The experimental results confirm the advantages of the proposed methodology. This methodology is an innovative expert system with a knowledge-base developed through this simulation optimization approach. This expert system can be applied to complex production line problems in large or small scale with different types of decision variables and objective functions. The application of this expert system is transformative to other manufacturing systems. (C) 2019 Elsevier Ltd. All rights reserved.