Big data analytics is the process of examining large and often multiple data sets to reveal trends, patterns, correlations, and other information that can be utilized to assist organizations in making more-informed business decisions. Big Data Analytics can be a very effective tool in support of materiel readiness, mission assurance and sustainment cost reduction in the Department of Defense. This forum will consist of a series of presentations which will describe analytical capabilities and solution sets that assist DoD maintenance managers and leaders with making decisions on what to maintain, when to maintain, and what maintenance actions have the biggest potential impacts on equipment availability and efficiency.
Agenda
1300-1309: Welcome and Overview – Greg Kilchenstein (OSD-MPP)
1309-1310: Administrative Notes – Debbie Lilu (NCMS)
1310-1330: Big Data at Marine Corps Logistics Command – Major Brian Bagley and Major Amanda Coleman
1330-1350: OPNAV/NAVAIR DWO Big Data Pilot – Jeff Woell (NAVAIR)
1350-1410: “One Network – Data Fusion” which uses existing data sources – Brandon Coffey (WPAFB) and Jim Dubberly (OneNetwork)
1410-1430: “Machines Don’t Have to Break” – Trey Clark (Uptake)
1430-1450: MADW to identify maintenance cost & readiness drivers – Eric Herzberg (LMI), Greg Kilchenstein (OSD), and Janice Bryant (NAVSEA)
1450-1500: Wrap-Up
Minutes
Event: On 28 November 2017, the Joint Technology Exchange Group (JTEG), in coordination with the National Center for Manufacturing Sciences (NCMS), hosted a virtual forum on “Big Data Analytics”.
Purpose: The purpose of this forum was to present analytical capabilities and solution sets that assist DoD maintenance managers and leaders with making decisions on what to maintain, when to maintain, and what maintenance actions have the biggest potential impacts on equipment availability and efficiency.
Welcome: Greg Kilchenstein (OSD) welcomed everyone to the forum, thanked the presenters and all the listeners for their attendance, and briefly previewed the agenda.
Administrative: This was an open forum. The presentations, along with questions and answers, were conducted through Adobe Connect. A separate audio line was used. Approximately 60 participants from across DOD and industry joined in the forum.
Predictive Analysis for NAVSEA CBM+ – Samir Mehta (GE Aviation) & Patrick Cotton (NAVSEA) provided an overview of the NAVSEA 05Z APM / Digital Twin Pilot. Findings from phase included 45 cases opened for anomalies detected by SmartSignal with 13 closed as actioned. The estimated cost avoidance is $1.04M. Phase 2 will continue the investigation of using GE SmartSignal APM technology to provide predictive analytics.
OPNAV/NAVAIR DWO Big Data Pilot – Jeff Woell (NAVAIR) provided an outbrief on the Super Hornet Readiness Digital Pilot conducted from April through September 2017. The pilot applied data science expertise from the Center for Naval Analyses and physical/domain expertise from the Naval Aviation Enterprise to produce two kinds of outcomes:
- Mission Outcome: Leverage big data & analytics to address Super Hornet readiness challenges in Naval aviation. Recommendations centered on BIT data and bad actors.
- Digital Outcome: Improve how Navy uses data as a strategic asset. Recommendations included identify the requirements, resources, and timeline to provide a cloud-based aviation readiness data environment, and to develop a strategic communication plan to emphasize the value of data and the importance of data stewardship.
Big Data at Marine Corps Logistics Command (MCLC) – Major Brian Bagley and Major Amanda Coleman first discussed the LOGCOM mission and provided an overview of the Studies, Analysis & Innovation Division, then described how the MCLC uses big data to include the Price and Performance Model, Impact to Deferred Maintenance, Unit Repair Cost insights, and Future uses of big data.
“Machines Don’t Have to Break” – Trey Clark (Uptake) provided a briefing and software demonstration on Uptake. With the phrase “Machines don’t have to break if they’re powered by Uptake,” Trey discussed how Uptake uses predictive analytics software to improve efficiency and help people and machines work better, smarter and faster. There proprietary data models have generated alerts to make trains run more safely, wind farms perform more efficiently and heavy machinery operate more reliably.
Maintenance & Availability Data Warehouse (MADW) to Identify Maintenance Cost & Readiness Drivers – Eric Herzberg (LMI), Greg Kilchenstein (OSD), and Janice Bryant (NAVSEA) discussed the MADW tool and how it has been used in DoD to date. MADW collects data obtained from DoD maintenance records, reviews and verifies the data, and provides the user with the ability to conduct multiple types of analysis to include costs and availability. Examples include determining the annual maintenance cost of electronic components for which there is “no fault found”, and what are the 10 best candidates for improvement? A cost per day of availability metric can transform how success is measured in maintenance. Additionally, a predictive capability in development will be a powerful enhancement to the MADW.
Closing Comments: Greg Kilchenstein thanked the presenters for their contributions and the audience for their participation. He suggested continuing the information exchange beyond the forum and the importance of teaming/partnership to the DoD maintenance community.
Action Items:
- Obtain “public release” versions of the presentations and post to the JTEG website. These meeting minutes, the Q&A, and those briefing slides approved for public release, will be posted on the JTEG website at https://jteg.ncms.org/ . (All presenters, LMI, NCMS)
Next JTEG Meeting: The next JTEG virtual forum is 19 December, 1:00 – 3:00 pm EST. The topic is “Improved Business Processes”.
POC this action is Ray Langlais, rlanglais@lmi.org , (571) 633-8019
Forum Q&A
Predictive Analysis for NAVSEA CBM+ – Samir Mehta (GE Aviation) & Patrick Cotton (NAVSEA)
———————————————————————————————————————————–
Q1. How did you quantify the $1m cost avoidance from this pilot?
A1. Good question. We monitor many items and have a series of estimates including conducting a cost avoidance study.
Q2. Are you able to determine the readiness impact from smart signal?
A2. Yes, its purpose is for mission readiness. Prognostics is the first step. There is a lot more data and systems that are looked at, such as costs and availability.
Q3. Can you talk to how big the data is? 2. How is the Data retrieval/data compression of the big data from ship? 3. What Algorithms or deep learning/neural networks are used?
A3. Kilobytes. When we switch to ERM it will be megabyte data. We have experienced problems with bandwidth “dehydration” which we are trying to reduce. No neural networks are being used. We use a smart signal suite that looks for anomalies.
OPNAV/NAVAIR DWO Big Data Pilot – Jeff Woell (NAVAIR)
————————–
Q1. What was the average level of parts immediately available for issue for repair (e.g.,, when 10 parts are required, all 10 are available at the start of the repair)?
A1. I do not know. That is specific to the FRC. It is definitely a consideration.
Q2. Any idea of whether supply or maintenance is driving the majority of availability loss? Do you know what the ratio is of NMC due to supply vs maintenance?
A2. I do not know. Roy Lancaster is the subject matter expert who could answer that.
Q3. With the sensor data integration pilot success, what is the next step for NAVAIR to move beyond a pilot and roll out this capability across the F-18 fleet?
A3. Recorder unable to capture the answer.
Q4. Can you explain the quantum jumps in the improvement of overhaul success (slide 12)?
A4. Bad actors out of the gate.
Q5. Is NAVAIR using IUID now to serialize components critical to F-18 readiness? If not, what is being done to “serially manage” these components?
A5. Yes
Q6. Where does PMA-260 involvement come into this? Did this project involve Support Equipment at all? Will SE be involved in future efforts?
A6. Absolutely. We are married at the hip to develop the maintenance IT infrastructure – Test Program Set Structure.