Pentagon Harnessing Data for Predictive Logistics Planning

Pentagon Harnessing Data for Predictive Logistics Planning

Defense Dept. photo

OKLAHOMA CITY — “Logistics wins wars” is an age-old adage, but its application has evolved from logbooks and spreadsheets to data streams and analytics. The Defense Department now wants to harness these new practices to predict the future.

Christopher Lowman, assistant secretary of defense for sustainment, described predictive logistics as a method of utilizing data generated across various platforms — something the Defense Department has no shortage of — to better understand what tools need to be employed to turn data into actionable logistics decisions.

“It’s all about regenerating readiness and pushing the capability to satisfy demand closer to the point of need within the theaters of operation,” Lowman said at the National Defense Industrial Association’s recent National Logistics Forum.

Army Gen. Charles Hamilton, commanding general of Army Materiel Command, said precision and predictive sustainment means “not only knowing when, where and how much a unit is using something like ammo [or] water or maybe certain parts, it also means knowing precisely where and when, and how much of a commodity that you will need in the future. And the key is getting out in front.”

The ability to predict is also the ability to stay steps ahead of an adversary — one that requires “that we go into the next fight” with a “data-informed decision-making process,” Hamilton said.

The “crystal ball” of data analytics is taking all of the data gathered across the government and industry and analyzing and interpreting meaningful patterns. Hot button solutions like artificial intelligence and machine learning are tools that can help but need to be recognized as such — tools, and not the ultimate salvation, participants at a forum panel discussion suggested. The fundamental input is the data they rely on.

Kevin Gaudette, a retired Air Force colonel and senior vice president of integrated analytics and support at LinQuest, said AI and machine learning are, just like optimization and simulation, “tools — and they’re reliant on data. We throw these buzzwords out, everybody gets excited … and starts asking for things that they don’t even know what they’re asking.”

The question needs to be, “What is it that you’re trying to do?” he said. “Let’s start with your understanding of what it is. Because [AI and ML] may not be the answer. In a lot of cases, it’s not.”

But sometimes it is.

Marine Corps Maj. Gen. Keith Reventlow, commander of Marine Corps Logistics Command, called artificial intelligence and ChatGPT “awesome tools.” While they may be associated with cheating on term papers, the concept of “going through every source that’s on the internet on [a] subject, comparing it all and giving you an answer” could be revolutionary for logistics, he said.

“What if we could do something similar when we think of sustainment, and understanding our weapons systems, understanding predictability of conditions-based maintenance,” he said, such as predicting time between failures, “and give the commander the option of changing it out before it actually breaks. I think there are all kinds of opportunities we are working on to try to understand how we are going to leverage data using those tools.”

AI and machine learning are just a few tools across a broad spectrum of projects and research underway across the Defense Department to better understand how to leverage data.

One such effort is the Defense Logistics Agency’s Joint Additive Manufacturing Model Exchange, or JAMMEX — a tool that consolidates technical data packages and allows users to download and print models from multiple sources through a single system.

Adarryl Roberts, chief information officer for the Defense Logistics Agency’s information operations, highlighted a program building on JAMMEX called the Digital Sustainment Platform, “to not just be able to take the technical data packages, but also create a single platform where all our engineers … and digital twins can reside department-wide in order to get that efficiency” across combatant commands and the services.

Other DLA efforts include Digital-Business Transformation and a Warehouse Modernization Project, Roberts said. These efforts are buying down on what he called 20 years of technical debt, saying the department “got left behind in terms of … our ability to be flexible and have agility to move with technology.”

The Army last year stood up its Contested Logistics Cross-Functional Team, which Hamilton said “will never go away because it’s integrated with everything else that we’re doing.” Part of the team’s “very narrow problem set” is precision and predictive sustainment, he said.

Air Force Materiel Command launched its Digital Materiel Management initiative last year — an effort to integrate and employ digital methods across the entire capability lifecycle, from invention to retirement, leveraging digital tools, structured data and security.

Jim Sutton, senior director of strategy at Shipcom Wireless, said a recent kickoff event for the initiative involved conversations about the “idea of creating organizational constructs and aggregating data and identifying all the risks and setting up teams to reduce the risks,” but a near-complete absence of conversations around zero trust.

Zero trust is a Defense Department security framework that assumes a “never trust, always verify” mindset. Trust around data sharing is a well-documented hesitation between government and industry, and another speed bump on the way to effectively harnessing data analytics.

While there are many “underlying uncertainties” with data sharing as industry and government intersect on the delivery of capabilities, Sutton said the primary point of entry from a service leadership standpoint should be to press the implementation of the White House’s 2022 zero trust memorandum, which requires agencies to meet specific cybersecurity standards by the end of fiscal year 2024.

“If that were in place, essentially 90 percent of all of the discussion around risk evaporates,” he said. “And as a result of that, we would then focus on places where risks really matter in terms of getting to that actionable, decision-worthy information at the point of use.”

Aaron Jaffe, head of supply chain and logistics at Palantir Technologies, said data linkages between industry and the Defense Department need to be thought of in the broader community of the industrial base “as inputs that are limiting factors to how the DoD can operate as well as being able to integrate and work with our allies and partners in any decision that we’re making.”

The question becomes how to better understand the gaps and the seams, and the technology that will protect security while enabling seamless interoperability “across that much larger community than what we would historically look at,” he said.

The gaps present “tremendous challenges,” but not unprecedented ones, he said. The COVID-19 vaccine accelerator Operation Warp Speed and support to Ukraine are examples of industry and government collaborating on rapid timelines and “moving at the pace of conflict.”

A core enabler across those efforts — and one that industry needs now — is a clear sense of mission, he suggested — “a clear purpose and objective … is incredibly helpful and enabling for both us as an individual company but also for how we work with others and build on that approach that’s open and interoperable. You can leverage existing programs and new technologies that can help increase the pace of change across the DoD.”

While trusted data is crucial, the Defense Department also needs to get comfortable with dirty data, one industry representative said.
Justin Woulfe, chief technology officer at Systecon, said achieving perfect data is likely impossible, and while a noble and necessary pursuit, the department needs to accept some risk as well.

“What we really need to do is actually just start doing analytics,” he said. The idea of dirty data is “fascinating,” but not a hindrance, he added, suggesting no major acquisition program has perfect data, but can still use “data as dirty as it is today” to predict mission capable and full mission capable rates within 3 percent of what it set out to achieve.

Pristine data is ideal, but “it’s not an excuse to not do analysis,” Woulfe said. “So, the idea that we can’t start doing anything until we get perfect data is nonsense. When we think about contested logistics, we’re planning for things that are unknown. … So why do we expect to need to have a perfect viewpoint of our maintenance data or supply transaction history?”

Fuzziness is a certainty, he said, “but I think there’s this analysis paralysis that happens where the idea of, ‘We must know everything before we can start’ needs to be tossed to the side and just use the analysis, use the process of analytics, to learn more about our systems … and [accept] that [there are] some things that we don’t know perfectly and then move on from there and apply the risk and understand the risk and make decisions.”

Gaudette said for the first time “in a long time, we’ve got integrated strategy documents, and we’ve got integrated architectures, we just still don’t do data — it ends up going in a million different directions. So we have lots of pockets of excellence, but not an integrated solution still to this day.”

The ultimate solution may not exist, but the Pentagon is looking, and Sutton said that is the most important thing: “get started.”
Utilizing data to make informed, predictive decisions means putting tools in the hands of operators, he said, and “the more you do it, the better you are at it. … And the more we encourage our people to try using it in a space that is useful to their job, the more ubiquitous it will become — the easier it will be to adopt.” ND

Topics: Logistics, Logistics and Maintenance