A pragmatic decision model for inventory management with heterogeneous suppliers

ABSTRACT

For enterprises, it is imperative that the trade-off between the cost of inventory and risk implications is managed in the most efficient manner. To explore this, we use the common example of a wholesaler operating in an environment where suppliers demonstrate heterogeneous reliability. The wholesaler has partial orders with dual suppliers and uses lateral transshipments. While supplier reliability is a key concern in inventory management, reliable suppliers are more expensive and investment in strategic approaches that improve supplier performance carries a high cost. Here we consider the operational strategy of dual sourcing with reliable and unreliable suppliers and model the total inventory cost where the likely scenario lead-time of the unreliable suppliers extends beyond the scheduling period. We then develop a Customized Integer Programming Optimization Model to determine the optimum size of partial orders with multiple suppliers. In addition to the objective of total cost optimization, this study takes into account the volatility of the cost associated with the uncertainty of an inventory system.

KEYWORDS:

• Decision support system
• dual sourcing
• lateral transshipment
• supplier heterogeneity
• cost volatility

Acknowledgement

The authors wish to thank the Australian Research Council for funding support under the ARC Discovery grant of DP130101114 and National Social Science Foundation of China for funding support under grant of 16BJY119.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Holding cost, ordering cost and shortage cost are the most common in inventory management, other costs may include costs data acquisition, training et al. (Axsäter 2006).

2. The purchasing cost includes the ordering cost and procurement cost. Usually the ordering cost is assumed to be charged on batches while procurement cost is proportional to the size of order. However, considering the small share of ordering cost in total purchasing cost, we assume purchasing cost is proportional to the sizes of order.

3. Since backordering cost is determined by the supply lead time, expected demand and size of lateral transshipment, which supply arrives first, does not matter. In other words, the receipt of the scheduled delivery of supplies from or outstanding delivery from the supplier does not change the result of the analysis because and are both certain variables. In the beginning of the scheduling period the delivery time of the outstanding delivery from the order made with the unreliable supplier is known. The analysis based on the situation that reliable supplier arrived first is the same as the situation when an outstanding order from supplier arrives first. The latter condition has been explicitly used in this study.

4. The customer arrival is considered as a Poisson process, which is a memoryless stochastic process with a given intensity. In addition, the size of customer demand is assumed to have a geometric distribution which means the size of customer demand is also a stochastic variable that is independent of other customer demands and the distribution of the customer arrivals (Axsater, 2000, pp. 78). Therefore, the random demand process is an independent and identical distribution.

5. Appendix 1 provides the details of this calculation.

6. Please note that in the numerical study, the monetary value can be adjusted by scaling, therefore does not necessarily need to be the same as real value in the market.

Additional information

Funding

This work was supported by the Australian Research Council [ARC Discovery grant of DP130101114];National Social Science Foundation of China [16BJY119];