Article ID: | iaor2001803 |
Country: | United States |
Volume: | 47 |
Issue: | 2 |
Start Page Number: | 175 |
End Page Number: | 182 |
Publication Date: | Mar 1999 |
Journal: | Operations Research |
Authors: | Murty Katta G., Djang Philipp A. |
Keywords: | location, facilities, programming: integer |
For training National Guard units, the U.S. Army National Guard will field 21 combat vehicle training simulators called mobile trainers. Each National Guard unit must train at a station that is not farther than a specified maximum travel distance from its armory. We address the problem of finding: the optimum locations for the home bases for the mobile trainers; the locations of secondary training sites to which the mobile trainers will travel to provide training, and; the actual routes that the mobile trainers will take to cover all these secondary training sites. The aim is to allocate each National Guard unit to a training site within the maximum travel distance from its armory, while simultaneously minimizing the mobile trainer fleet mileage and the total distance traveled by all units. The problem is too large and complex to solve as a single model. We apply a heuristic decomposition strategy to break the overall problem into manageable stages, developing suitable substitute objective functions for each. This approach led to a solution in which the mobile trainer fleet mileage is 72,850 miles per year: about 70% smaller than the 231,000 miles per year in the original Army's procurement plan. Our solution has been implemented (with minor modifications) by the Army.