Sequence-to-customer goal with stochastic demands for a mixed-model assembly line

Abstract

A mixed-model assembly line comprises a set of workstations in serial and a conveyor moving at a constant speed, which can assemble variety products in different models during a working shift or a working day. Initial-units that belong to different models are successively fed onto the conveyor at a given cycle time length to get into the assembling operations as semi-products. The conveyor moves semi-products to pass through the workstations gradually to complete the assembling operations for generating finished products. A set of warehouses stores finished products, and each model has a specified warehouse. Customers arrive at the warehouses to demand finished products with stochastic demand forms. A daily scheduling task is the determination of the sequence that specifies the feeding order of the models, which must be set out at the beginning of each day. This paper deals with a new goal, ‘sequence-to-customer’, with stochastic customer demands. An optimization problem is formulated with the objective of minimizing the system cost that includes the holding cost for finished products and the penalty cost for backordered customers during a decision horizon. A lower bound of the system cost is found, which is useful in verifying the optimality of any solution. A heuristic algorithm is proposed to solve the optimization problem, which can obtain optimal solutions or near-optimal solutions with almost ignorable relative errors to optimal solutions. By using the algorithm, the behaviour of the system cost with respect to the variation in customer demands is also investigated to provide insights into management of a mixed-model assembly line.

Keywords:

• Mixed-model assembly line
• Sequence
• Stochastic demand
• Inventory
• Algorithm

Acknowledgements

Research was supported by the NSF of China under Grant No. 70325004, and partially supported by the NSF of China under Grant No. 70532004.