Comparative Study of Learning Curve Models in Defense Airframe Cost Estimating.

摘要

The goal of this research was to identify which learning curve model is most accurate when applied to Defense acquisition programs. Wright's original learning curve model is widely accepted and used within Defense acquisitions, but the 75+ year old model may be outdated. This study compares Wright's model against three alternative learning curve models using total lot costs for the F-15 C/D & E programs: the Stanford-B model, the DeJong learning formula, and the S-Curve model. However, the results of the study are inconclusive. Two of the three alternative models, the DeJong and S-Curve, rely on the use of an incompressibility factor between 0 and 1 that represents the percentage of the production process that is automated. A Bureau of Labor Statistics report identifies that percentage as very low but does not give an exact number. Therefore assumptions about that parameter were made. When the factor falls between 0.0 and 0.1 the DeJong and S-Curve models appear to be more accurate; when the number is 0.1 or greater, Wright's model is still the most accurate. Further research should be targeted at the exact value of this factor to validate this, or future, comparative studies