Learning Curve Analysis of Small Data Sets – Spacecraft Bus Cost Improvement Analysis

Methods and Models I Track



Recent initiatives to implement cost savings across DoD have produced a number of potential opportunities. Within the Space Portfolio one such opportunity is to acquire satellites more efficiently through block buys and limiting production breaks. To quantify these opportunities, we can analyze historical cost experience. This paper describes three approaches used to pool cost improvement data from multiple programs, allowing for regression of all data simultaneously. The approaches are (1) scaling based on weight, (2) normalizing to a common theoretical first unit, and (3) the use of program binary (“dummy”) variables. The advantages and disadvantages of each approach are described along with outcomes using spacecraft data. Additionally, considerable analysis was performed to normalize data across lots, identify breaks in production, and last production lots. The slopes of individual program cost improvement curves (using only quantity as an independent variable) varied widely, ranging from 87 percent to 176 percent using cumulative average unit cost. Connecting this data for multiple programs was challenging, given the limited data set. The identification of breaks in production and last production lots explained the wide ranges in improvement rates well. Both the Cumulative Average and Individual Unit Theory are calculated, and the results are presented. The resulting estimating coefficients for cost improvement, production break effect and last lot effect provides a strong quantitative basis for estimation of spacecraft hardware costs.


Brian A. Welsh
Technomics, Inc.
Mr. Brian A. Welsh works for Technomics, Inc. in Arlington, VA. Mr. Welsh has over 7 years of experience in the cost analysis arena working on projects for the Department of Defense, Department of Energy, NASA and the Intelligence Community. His current work focuses on developing learning curves of spacecraft for the Air Force Cost Analysis Agency. Mr. Welsh is also developing an acquisition cost database of Unmanned Aerial Systems and performing cost research for the Navy’s Visibility and Management of Operating and Support Cost database. He has also worked on Government Accountability Office engagements of FAA NextGen systems and Navy Ballistic Missile Defense cost estimates. Prior to joining Technomics he worked for TASC, Inc. While at TASC Mr. Welsh was a manager of cost and financial analysis programs supporting the IC. Mr. Welsh is a Certified Cost Estimator Analyst (CCE/A), has a Bachelor of Science in Systems Engineering from the University of Virginia and a Master of Science from George Mason University in Operations Research.

James L York
Technomics, Inc.
Mr. York is a Senior Cost Analyst with over 30 years experience in system cost analysis and system acquisition, with the Department of Navy and DOD support contractors. As a contract manager and lead analyst, Mr. York plans, directs and performs contract tasks, monitors progress, reports to the client, establishes task budgets and adjusts resources to resolve technical problems, and performs technical and contract cost reporting.