Case Studies: Definitive Screening Applied to a Simulation Study of the F100-229 Engine Repair Network

Abstract

ABSTRACTProblem: An adequate supply of F100-229 engines, which power the F-15 Eagle and late-model F-16 fighter jets, is vital to meet the operational demands of the Air Force. This supply is necessarily limited by government and inventory cost considerations. New requirement calculations indicate a pending shortage of available engines. At current supply levels, aircraft availability can be improved by reducing turnaround times across the F100-229 repair network. However, identifying slowdowns and bottlenecks in the repair network is a complex problem. To investigate process improvement opportunities, program managers desire parsimonious metamodels of the significant factors in the repair network.Approach: This article describes a statistical engineering approach to create metamodels of simulations for each Air Force base in an engine repair network that can be queried, as needed, to make managerial decisions. Headquarters Air Force Materiel Command Studies and Analyses Division (HQ AFMC/A9A), hereafter referred to as AFMC, described and captured the problem with a discrete-event simulation of the repair network built using real-world data. AFMC then applied design of experiments to identify those factors in the engine repair network that are driving the balance of available engines while staying within computational budget requirements for the project and then built statistical metamodels of the simulation.Results: The simulation model was verified and validated using actual data. The customer desired quick feedback on the model, resulting in a search for efficient designs to explore the simulation. After examining different experimental designs, we found that definitive screening designs provided the best combination of run-size efficiency and protection from model misspecification. The simulation data were used to create metamodels of the repair network, which were programmed into a spreadsheet-based decision support tool. The tool gives logistics engine managers the ability to make quick-turn decisions. Results from one of six bases of interest are reported.