In today's increasingly competitive global economy, firms are seeking any and every possible opportunity to differentiate themselves from competition, reduce costs, and add value to supply chains and end consumers. One option is to excel in reverse logistics and remanufacturing. Today, most companies have realized that reuse and remanufacturing activities offer opportunities to improve their supply chains by decreasing system-wide costs and improving customer service, besides improving sustainability. However, the reverse forms of material flow stemming from remanufacturing complicates supply chain planning activities due to uncertainty in timing, quantity, and place of returned product. Closed-loop supply chain (CLSC) management aims to effectively reuse and manage product returned at the end of the supply chain cycle.;In the literature, most operations management models dealing with reverse logistics and remanufacturing assume that product demand generally follows an independent and identically distributed (i.i.d.) probability distribution. However, in reality, factors such as customer behaviors, market conditions, and product life-cycle aspects often lead to different types of uncertainties, for example, periods of growth followed by maturity and decline. More importantly, market place events and economic conditions also lead to stochastic fluctuations in demand, for examples, periods of "high" demand followed by periods of "low" demand and vice versa. We deviate from the literature by modeling this non-stationary fluctuating demand process as a Markov chain, where demand information is maintained in the different states and transitions between the states are governed by a state-transition matrix. Our aim in particular is to support CLSC operations in tackling these demand variations and uncertainties in returns quantity and timing. Additional motivation for this research comes from the needs of our collaborator, Delphi Corporation, a leading global supplier of mobile electronics and transportation systems, offering several lines of remanufactured product for both OEM customers as well as the aftermarket. In remanufacturing operations planning within companies such as Delphi, efficient and accurate forecasting of demand and returns is crucial because of long product lead times and the need to balance the returns with demand. Poor demand and returns forecasting results in deteriorated planning, leading to excess or deficient inventory that impacts customer service and inventory costs.;Our first objective is to present the (re)manufacturer a modeling framework that facilitates accurate forecasting of demand and returns under non-stationary demand, in particular, demand following a Poisson Hidden Markov Model (PHMM) process. For cases where the underlying PHMM model is unknown, we offer a Bayesian scheme for learning and updating the model over time. Our second objective is to provide effective inventory control policies for balancing returns with demand. The computational burden of the proposed stochastic dynamic programming based inventory control policy increases rapidly with number of demand states, length of the planning horizon, and product return lead-times. Therefore, we also offer a number of different sub-optimal but efficient inventory control policies. Finally, we also offer insights into modeling and control of CLSCs with multiple players and stages.
展开▼
