The Impact of Learning Curve Model Selection and Criteria for Cost Estimation Accuracy in the DoD

Abstract

The first part of this manuscript examines the impact of configuration changes to the learning curve when implemented during production. This research is a study on the impact to the learning curve slope when production is continuous but a configuration change occurs. Analysis discovered the learning curve slope after a configuration change is different from the stable learning curve slope pre-configuration change. The newly configured units were statistically different from previous units. This supports that the new configuration should be estimated with a new learning curve equation. The research also discovered the post-configuration slope is always steeper than the stable learning slope. Secondly, this research investigates flattening effects at tail of production. Analysis compares the conventional and contemporary learning curve models in order to determine if there is a more accurate learning model. Results in this are inconclusive. Examining models that incorporate automation was important, as technology and machinery play a larger role in production. Conventional models appear to be most accurate, although a trend for all models appeared. The trend supports that the conventional curve was accurate early in production and the contemporary models were more accurate later in production.