Supply Chain Optimization Tools Improve the Vehicle Distribution System at Volkswagen of America*

ABSTRACT In designing a tool to aid Volkswagen of America in its vehicle distribution system reengineering effort, it was necessary to compute performance measures in terms of both transportation cost and customer service that are functions of probabilistic and dynamic elements. These elements include inventory control policies, demand seasonality, customer demand, customer choice, and transportation delays. To consider all, a simulation approach appears appropriate, but the potential number of alternative designs that need to be evaluated is excessive. Discrete optimization models, on the other hand, help reduce the alternatives; however, it is difficult to incorporate those probabilistic and dynamic elements while keeping the model size manageable. With an innovative use of both types of models iteratively, we were able to solve the real problem that faced Volkswagen in analyzing vehicle distribution system alternatives. Introduction Volkswagen of America (VoA), a wholly owned subsidiary of Volkswagen AG, imports, markets and distributes Volkswagen and Audi vehicles in the United States. The vehicles are assembled in Mexico or Germany and distributed to a network of approximately 750 dealer sites across the United States. In 1995, VoA management appointed a project team with the mandate to review vehicle distribution process and develop new concepts to serve into the next century. This effort was part of an enterprise wide process of renewal that had begun in early 1994. The existing vehicle distribution process had served the organization over a number of years but was in need of review in light of changes occurring in automotive retailing. The existing system had been developed around two implicit assumptions. The first is that the dealer (retail operator) is the "customer" of the VoA distribution process, not the end user. The logic was that Volkswagen of America distributes to the dealer and the dealer distributes to the end user. The second assumption was that, significant vehicle inventory (60-90 days of sales) carried by the dealer operator creates an incentive (financing cost avoidance) to increase the rate of sales. Thus the distribution system operated as a "push" system. Efficiency was the key performance criteria. Neither of these assumptions were unique to the VoA distribution system, in fact they reflect the dominant logic of vehicle distribution the U.S. auto industry. The push system does have significant limitations. One is that not all customers are able to purchase their first choice configuration of vehicle color and optional equipment. Many manufacturers offer many more vehicle configurations than individual dealers are able to inventory (even carrying 6090 days of sales). Dealers, who are interested in completing sales transactions as quickly as possible, generally offer customers the selection vehicles from their current inventory. If the current inventory does not contain the customer's first choice configuration the dealer may offer a price discount as an