Considerations for Predictive Logistics
“How will the use of Predictive Logistics change the way we support?” MG Mark Simerly posed this question in the January 2023 edition of the Combined Arms Support Command (CASCOM) innovation newsletter. The following offers a short response and explores the risk of the data analysis at scale – that data driven conclusions are only as valid as the accuracy of the first input.
Predictive logistics is a growth industry within the U.S. Army’s data revolution in pursuit of Secretary Wormuth’s data-centric Army. As a reference point, the Government Accountability Office (GAO) defines predictive maintenance as, “any effort that uses condition-monitoring technology or analysis of historical data to anticipate maintenance needs in a manner that reduces unscheduled reactive maintenance or overly prescriptive preventive maintenance.” This same definition can be applied more broadly across all the logistics functions including predictive transportation, predictive supply & resupply, and predictive personnel management.
It is hard to overstate the impact that predictive logistics will have on the support community.
The promise of predictive logistics is that decision makers will have access to accurate, real-time data to facilitate optimal decision making. In wartime, more accurate decision making will contribute to winning faster.
In peacetime, this same accuracy will reduce the overall resources required to maintain readiness for war. Predictive logistics will be a techno-manager’s dream. (Strategists everywhere are jumping out of their armchairs; this assessment is disconnected from the art of war. Better predictions only improve the science of war, but winning war will still require military objectives aligned against political outcomes.)
A fundamental risk inherent in the promise of predictive logistics is the validity of underlying data. No matter how perfect the algorithm; if it’s garbage in, the result is garbage out. This means the Army must drive to reduce the human factor in data input as much as possible. In daily life we interface and make assessments with data every day. Driving to work, a sensor in the gas tank tells us how many gallons we have left, a useful data point as we decide whether to stop and fill up now or risk being stranded. A more advanced car takes this sensor input and goes a step further, it calculates the vehicle’s average miles per gallon burn rate and tells the driver how much further it can go before the tank is empty. This succinctly illustrates the difference between data (10 gallons in the tank) and knowledge (I can drive 200 miles which is far enough to reach my destination). Predictive logistics would take this example a step further; the car would tell you that you have 10 gallons, which is a 200-mile driving range, and that your optimal refuel location is 173 miles away with a current pump wait time of 2 minutes and is the best price along your route. What happens if that sensor in the gas tank is broken? Or the satellite connection to the cloud is degraded by interference? The consequence in daily life is an annoyance, the consequence in war is human lives.
The Army relies on people: privates, sergeants, lieutenants, and captains to input and validate the data going into its systems. Soldiers and DA Civilians are the sensors for a wide variety of assessments on everything from the maintenance of equipment (input manually into Global Combat Support System-Army by a clerk), the training readiness of a unit (inputted manually into the Defense Training Management System (DTMS) by a training NCO), or the condition of a barracks (inputted manually into the Installation Status Report by a DA Civilian). The common thread between all three is the human factor. No matter how advanced, complex, or integrated the system – it still relies on a human to input the first data point, and this first input must be accurate.
A vignette most Soldiers will be familiar with and one that illustrates the risk of human input error is the maintenance documentation on a DA Form 5988-E. Traditionally maintenance activities occur on Mondays where operators will conduct a basic inspection of their equipment, called PMCS – preventative maintenance checks and services. Does my vehicle start or do I need a new battery? How much windshield wiper fluid do I have? Is there a slow leak and are my tires flat? Depending on the training schedule unit’s may or may not use their equipment frequently so weekly PMCS is the preferred method to maintain readiness. Any deficiencies are documented manually on a DA 5988-E by the operator and handed off to the company maintenance team for verification and to submit the request in the supply system. This process makes sense from a leadership perspective, leaders want operators to know their equipment and take ownership in its readiness. Secondarily most fleets are too large for a small maintenance team to do the required maintenance, so having operators perform the PMCS is essential. The shortfalls of this process is its reliance on a manual form and multiple human inputs before it reaches a data repository (GCSS-A). The operator annotates their needs on the form which then is handed to a maintenance soldier for fault verification before finally making its way to the clerk who creates the supply request. All too often a soldier will complain that they’ve been annotating the same shortfall for weeks without receiving the necessary part to fix the equipment. This is a tactical impact at the company level for data input error. The strategic impact is that GCSS-Army is an Army system of record which provides a snapshot of a unit’s equipment readiness and is reported on a monthly basis to the highest levels of Army Senior Leaders (ASLs). ASLs use this tactical readiness data to inform the Joint Staff, combatant commands, and Congress on the overall state of the Army.
In the drive towards predictive logistics, the Army is making strides to reduce this specific manual process because of its strategic impact. Recently, as part of Project Convergence the 2nd Brigade, 3rd Infantry Division executed a soldier touchpoint for Prognostic/Predictive Maintenance (PPMx). This is a handheld tablet based platform that completely eliminates the manual 5988 process. Now, when a Soldier documents a fault through the PPMx tablet it will instantaneously populate in GCSS-Army and be captured in a data repository. This improvement at the human input point is essential to achieving the promise of predictive logistics.
Returning to MG Simerly’s prompt and exploring how prediticitve logistics will change the way we support, there are two considerations worth noting. The first consideration is that with increased demand for data there will be an equivalent increase in demand for data accuracy and this will consume leaders time. Access to data reduces the degrees of separation between tactical and strategic leaders and flattens the Army’s command hierarchy. This isn’t necessarily a bad thing, but it is worth noting that the Army’s culture must recognize this fact. Battalion, Brigade, and Division Commanders will lose reaction time and decision space because the data they have access to is the same data available at higher and lower levels. A negative cultural response would be Commanders at all echelons spending more of their already limited time ensuring data accuracy to the detriment of their assigned mission. Commanders must recognize that there is increasing marginal opportunity cost across their formations to improve a data point from 96% to 99%. The second consideration is that just because the data is available, does not mean it is useful. This is the true promise of predictive logistics. Predictive logistics will tell leaders what is important to know without being bogged down in the ancillary data that is interesting, but not essential, to making a decision.
Brian Mathews is an active-duty Logistics Officer serving as an Army Staff Intern at the Pentagon.
Opinions expressed here are those of the author and do not represent those of the United States Army, the Department of Defense, or the United States Government.
References:
1 Wormuth, C. “Message from the Secretary of the Army to the Force.” February 8, 2022. https://www.army.mil/article/253814/message_from_the_secretary_of_the_army_to_the_force
2 GAO. Military Readiness: Actions Needed to Further Implement Predictive Maintenance on Weapon Systems. December 8, 2022. https://www.gao.gov/products/gao-23-105556