MIT Forum for Supply Chain Innovation Conference

Demand Forecasting

Keynote Presentations
Demand Forecasting When there is No Point-of-Sales Data
Prof. David Simchi-Levi, Massachusetts Institute of Technology

Point of sale data has been used to improve performance in many consumer product supply chains. There are, however, many supply chains without point of sale data. Many emerging markets, for example, function without this technology. David Simchi-Levi will discuss alternative approaches to improving supply chain performance in this situation and review several cases illustrating these strategies.

Achieving Supply Availability in the Face of Highly Uncertain Demand – A Case Study of Military Aviation Spare Parts
Dr. Bill Killingsworth, Executive Director, MIT Forum for Supply Chain Innovation

Demand for military aviation spare parts is notoriously difficult to forecast. These parts have very long production lead-times, true demand and inventory data is often suspect, and operations are heavily influenced by politics. In many ways, this forecasting challenge is similar to predicting unknown unknowns two years out. One approach to dealing with forecasting difficulties and the inevitable forecast errors is to develop a push-pull boundary in the manufacturing process. This boundary establishes inventories at critical points in the production process such that surges in demand can be met at a reasonable cost. Detailed supply chain maps have been developed for a number of major aviation parts. Inventory optimization software has been used in conjunction with these maps to develop inventory strategies in the manufacturing supply chain and to establish the push-pull boundary. A dynamic simulation model has been developed that incorporates ten channels (representing ten critical components of the major part) with each channel having a three tier supply network. This model has been parameterized for the studied parts and used to assess the benefits of the push-pull boundary under a number of demand scenarios and forecast errors. These manufacturing push-pull strategies are shown to be effective, low-cost methods for dealing with highly uncertain demand over long time periods.

Reception: May 14, 7-9PM
Meeting: May 15
Both Events Will Be Held at the MIT Faculty Club

AGENDA