Collective Mind: Avoidance of Maintenance Cost Through Predictive Trending

NCMS Project #: 140498

Problem: Fleets of aerospace equipment are managed through carefully controlled supply chain processes.  When any of the planning assumptions fail, for example, due to a new mission, a batch of out-of-spec parts or an ill-conceived maintenance procedure, an unexpected demand on maintenance and supply can develop, leading to increases in operating costs and reductions of equipment availability.  Isolation of early warning signals of onset of such crises is critical to dealing with them proactively.  Complexities of aerospace fleets make it particularly difficult for managers to recognize emerging patterns of systematic failures before they escalate.  Often, only when availability is significantly affected, will attention be paid.

Systematic failures of components in man-made fleets bear an analogy to disease outbreaks among humans.  The spread of an out-of-spec parts from a delivery in some ways is like a new virus entering a community.  Often, healthcare providers remain unaware of the system-wide nature of a problem until relatively late in its progression.  Fortunately, data mining and machine learning technology has been demonstrated to reliably generate early warnings of the advent of human epidemics by observing the operation of the health care system over time.

Benefit: The DoD has organized an effort on Engineered Resilient Systems.  This effort focuses on agile and cost-effective design, development, testing, manufacturing, and fielding of trusted, assured, easily- modified systems.  Its products are engineering concepts, techniques, and design tools.  Its goal is to achieve a vitally-needed transformation in the contribution of Defense systems engineering throughout the systems lifecycle.  This is essential to address a geopolitical environment marked by rapidly changing threats, tactics, missions and technologies.  The pace of change renders current approaches unsustainable in both cost and time.

Meeting these challenges require engineering concepts and tools for decision making at every step in the systems lifecycle.  Choices in every phase continuously shape and redefine system capabilities, performance, cost, delivery schedule, and sustainability.  To be effective, decision makers must have a full understanding of interactions among candidate component elements of a system.  They must also understand changing implications for the system of potential missions across joint warfighting environments.  ERS engineering concepts and tools thus must enable engineering, warfighting and acquisition decision makers to manage activities with full and consistent information throughout the life of a system.  The concepts and tools must enable deeper consideration of design alternatives in a manner that facilitates keeping options open.  They must facilitate co-design of Concepts of Operation and of systems with concern for multiple alternative futures.  They must also facilitate redesign of components, maintenance processes, and/or ConOps to adapt to the realities of actual usage in operational context.

In order to fully develop ERS, it is essential to develop and conduct a number of analyses and demonstrations that provide data and proofs-of-concept.  These will illuminate the strengths and advantages of pursuing the effort.  The demonstrations in this project will convey technical services improving DoD capabilities for acquiring and improving the right system, at the appropriate cost, on the appropriate schedule.  They will produce evidence of substantial cost savings and return on investment opportunities.

We already know that many of the parts associated with U.S. Army aviation are not meeting the intended design life of platform requirements.  A successful program that quickly indentifies quality, maintenance practice and reliability problems would substantially help the aviation platforms meet the intended readiness requirements and decrease the system life cycle costs.  Given that the U.S. Army helicopter fleet numbers approximately 4,500 and is approximately 4.5 times as large as the USAF F-16 fleet, the savings per year in cost avoidance for the U.S. Army could exceed $80M.

The number of just the commercial jet aircraft in operation around the world is approximately 20,000 and hence is also 4.5 times the size of the U.S. Army helicopter fleet. Success­ful development of this technology could avoid several hundreds of millions of dollars of cost each year and make commercial maintenance and sustainment less costly; thus improving competitive edge for adopters of this technology.

Solution/Approach: This project will begin to explore the hypothesis that similar mathematics has value in alerting to unusual patterns when monitoring health of fleets of the U.S. Army’s helicopter fleets with the goal of seeing whether multi‐year cost avoidance can be achieved at the enterprise level.  This will be done by identifying high value analyses and relevant demonstrations with the current Army Aviation data that have the ability to scale across the DoD and enable explanation of variances in reliability calculations and demand forecasting so that proactive cost‐reducing actions may be taken across DoD.

The proposed approach is to transition existing Collective Mind technology to Army Aviation data enterprise, adapting it as necessary.  A relevant and representative sample of data will be provided to Carnegie Mellon University by ARMDEC.  It will be drawn from the U.S. Army data warehouses, including U.S. Army maintenance, logistics and operation data systems showing a part of the history of at least one of the largest U.S. Army helicopter fleets, as well as the source data for ASAP top driver analysis.  The U.S. Army will provide necessary guidance and identify current operating procedures and case scenarios for utility validation.  The Collective Mind team will perform appropriate analysis.  The U.S. Army and the Collective Mind team will jointly validate the results, attempt to quantify the resulting utility and formulate recommendations for transition to practice, including additional experiments and further research and development.

Impact on Warfighter:

  • Increase population Mean-Time-Between-Removals
  • Decrease frequency of overhauls per year.
  • Allow reduction of inventory.
  • Increase fleet availability by early detection of fleet-wide disturbances or downing events.
  • Exchange avoidance through earlier visibility.

DOD Participation:

  • U.S. Army Aviation and Missile Research Development and Engineering Center (AMRDEC)

Industry Participation:

  • Carnegie Mellon University
  • University of Massachusetts Amherst
  • NCMS