Linking product quality and customer behavior for performance analysis and optimization of make-to-order manufacturing systems

• Published in: International journal of advanced manufacturing technology Vol. 95; no. 1-4; pp. 587 - 596
• Main Authors: Konstantas, Dimitris, Ioannidis, Stratos, Kouikoglou, Vassilis S., Grigoroudis, Evangelos
• Format: Journal Article
• Language: English
• Published: London Springer London 01.03.2018; Springer Nature B.V
• Subjects: CAE) and Design; Computer-Aided Engineering (CAD; Consumer behavior; Customers; Dynamic programming; Engineering; Evaluation; Industrial and Production Engineering; Inspection; Mathematical models; Mechanical Engineering; Media Management; Optimization; Original Article; Performance evaluation; Produce to order; Product quality; Profitability; Queues; Queuing theory; Markov decision processes; Closed queueing networks; Production control; Quality control
• ISSN: 0268-3768; 1433-3015
• DOI: 10.1007/s00170-017-1225-x

A queueing network model is developed for understanding the way product quality may affect the profitability of production systems, when the consumers base their future demand patterns on the quality of the products they have recently purchased. We examine a multistage make-to-order system which receives orders from regular and occasional customers, the former having a higher mean demand rate. Each outgoing item undergoes inspection and quality grading to decide whether it will be discarded as nonconforming or shipped to a customer. In the latter case, the customer who purchases the item will subsequently become a regular or occasional customer with complementary probabilities which depend on the quality level of that item. The solutions of simple test cases with dynamic programming show that the optimal policy is state-dependent, complex, and computationally intensive. A much simpler, threshold-type policy is proposed, whose performance evaluation and optimization uses closed queueing network formulas and has minimal computational requirements. Numerical results indicate that the proposed policy performs almost as well as the optimal policy.