Optimization of the seru production system with demand fluctuation: A Mean-CVaR model

The increasingly diverse and non-deterministic customer demands have given rise to the seru production system. This study aims to address the multi-objective seru formation problem under demand fluctuations, with the mean and the CVaR or standard deviation of total completion time minimized. In particular, two multi-objective stochastic optimization models, i.e., Mean-CVaR and Mean-Std, are constructed to determine the number of seru, the allocation scheme of workers and batches. Then, the Local Search Fast Non-dominated Sorting Genetic Algorithm (NSGA-II-LS) is designed to solve the proposed models. Numerical experiments validate the method's effectiveness by demonstrating that, compared to the Mean-Std model, the Mean-CVaR model enhances the seru system's capability and stability in coping with stochastic demand fluctuations through an average reduction of 4% in expected value and an average reduction of 36% in standard deviation. The effects of confidence level, the product batch number, and the standard deviation on the results have been analyzed. The results show that the Mean-CVaR method helps achieve a more stable seru system in terms of reducing standard deviation. Moreover, product batch division is beneficial to the CVaR of total processing time but detrimental to the mean processing time.