Exception handling for demand management in a collaborative supply network
Supply networks are becoming increasingly collaborative, with more and more data being shared and decisions taken through intense deliberations. With increased collaboration comes a greater risk of exceptions, which can undermine the collaborative advantages. Handling the exceptions is essential to make such a collaboration system scalable. One such instance that occurs in the operative layer of the supply network is when two entities in a supply network make separate forecasts for orders to satisfy demand for a specific period of time, and run into a exception in the form of a forecast mismatch. In this work, an exception handling mechanism is introduced to resolve such exceptions, by making decisions dynamically based on past performances of each entity's forecasts, making this system a co-system. The main contribution from this work will be a new measure, defined as Forecast Viability, to determine the usefulness of the forecast methods used in the co-system. The performance, in terms of costs, of this mechanism is calculated and then compared with the retailer system and with the performance of a system using a simple average of forecasts as an exception resolution mechanism. In the experiments, when comparing the costs from using the Forecast Viability measure against the retailer system, the new measure yileded reduction in costs of upto 10%. Recommendations are then made based on the experimental results. When adding an additional random demand component about which only one of the entities will have knowledge at that given time, the co-system performance yielded reduction in costs of upto 25% when compared to a retailer system and significant savings when compared to the system using a simple average of forecasts.
Nof, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our