Two-stage make-to-order remanufacturing with service-level constraints

We consider a firm that meets demand for an order with remanufactured products, new products or a mix of both. There are constraints on the service level. We use a stylized two-stage GI/G/ 1 queuing network model to study the problem. The first stage is unique to each product, whereas remanufactured and new products share the second stage. The objective was to find the optimal, long run production mix that maximizes profit subject to a service-level constraint that restricts the average order lead-time. There is yield in the remanufacturing process, where yield is the per cent of returned used products that result in a good part after remanufacturing. In the analytic model, we make the simplifying assumption that the producer always gets enough return of used products to meet its remanufacturing needs in the production mix. However, we relax this assumption in a simulation. We find that the optimal solution is generally non-trivial, i.e. the firm generally uses a mix of remanufactured and new products to meet demand. When the new product is less profitable than the remanufactured product, then it is optimal to remanufacture 100%, provided that there is enough supply of used products. When the new product is more profitable, however, the proportion of remanufacturing increases as service level increases. We use simulation to test the robustness of the analytic model by including complexities such as stochastic product returns and stochastic production yield.