Risk Vs Reward With Vehicle Unit Deployment Lists

By 1LT Patrick Donovan

Article published on: September 1, 2025 in the Fall 2025 edition of Army Sustainment

Read Time: < 5 mins

The leadup and outload for a major training rotation is a stressful time. Condensed timelines and a large number of moving parts make it a whirlwind few weeks. Individual packing lists must be gathered, tough boxes packed, and mission planning completed. A massive piece of the outload process is the maintenance and upload of vehicles that are on the unit deployment list (UDL).

From my perspective as an engineer company executive officer during the outload for the Joint Pacific Multinational Readiness Center ( JPMRC) rotation in Alaska this past winter, a ton of effort from mechanics and vehicle operators was put into getting UDL vehicles to a full-mission-capable (FMC) status. Many UDL vehicles were not fixed in time for outload, and some that were fixed broke down in the preliminary stages of the exercise. Was the juice worth the squeeze? Was the amount of effort expended to fix vehicles that gave us no value during the exercise justified? Is there a more proactive way to determine what to focus on during the outload process?

Getting the Most Value Out of Your Time and Effort

Each vehicle on the UDL is critical to the mission in some capacity, whether it be as a troop mover, an asset hauler, or a piece of engineer equipment. Scratching any vehicle on that list introduces risk to the unit’s ability to effectively complete the mission. Alaska’s harsh climate, dynamic operational tempo, and the hard miles already put on these vehicles make keeping them FMC throughout the year a substantial challenge.

In my company’s case, only six out of the 15 total vehicles on the UDL were FMC once the outload truly began for JPMRC, which was about a month before the exercise started. That month consisted of long hours at the motor pool trying to fix as many vehicles as possible and to get them uploaded via rail, commercial linehaul, or convoy. We fixed 10 out of the 15 in time for upload, but three of those 10 broke down before even entering the box. Another three vehicles broke down during the box. Other companies had similar numbers.

Despite the huge amount of time and effort invested into getting certain vehicles into the fight, many served little or no purpose toward the mission and created more of a headache trying to get them back to home station because they were broken. It is difficult to cut ties early and make the decision to scratch a piece of equipment you have already spent a lot of time trying to fix. At the company and possibly battalion level it is simply not an option to tell your higher headquarters you are scratching a vehicle, except at the very last minute. But at the end of the day, if the goal during the outload process is to be as efficient as possible with our time, we must spend the most time on the vehicles that are going to give us the most value. In the next section we look at ways to determine which vehicles will do that.

Machine Learning to Aid Decision Making

A more proactive approach to determining which vehicles to focus maintenance efforts on may be found using machine learning, which is a subfield of artificial intelligence that uses results from historical data to make predictions on the outcome of input data. In LTG Christopher O. Mohan’s recent article in the Army Sustainment Professional Bulletin titled “Predictive Logistics is the Way of the Future,” he writes, “Predictive logistics represents a shift from traditional, reactive sustainment models to a proactive, data-driven approach that allows us to position supplies to ensure the right resources are available at the right time and place.” Just as predictive logistics is the next step for allocating supplies on the battlefield, predictive maintenance is the next step for allocating resources into fixing vehicles and equipment.

For this to work, we must first understand what problem we want to solve. Should we try and fix this vehicle to bring it to the fight, or should we scratch it because it is not worth the effort? Second, we must collect data that helps determine when vehicles may break down. Readings from vibration sensors on wheel bearings and temperature gauges on engines are examples of the data that paints a picture of when a machine is in good shape and when it is about to break. We must train the complex machine learning network by feeding it that sensor data and whether the associated machine broke down. We must also feed it important information like outside temperature and the type of roads the vehicles were driving. Lastly, we must attach some sort of weight to the importance of each vehicle, like what the Army does when they say that certain vehicles are pacers and others are not. For example, a High Mobility Engineer Excavator is going to be more important to the mission than a Humvee. The system places vehicles into two categories. Vehicles in the first category will receive precious time and resources for repair, while vehicles in the second category will not because the risk that they will break down is too great compared to their potential value.

Conclusion

Organizations traditionally use predictive maintenance to decide when to do maintenance on an engine, not whether to do maintenance. This is a worthwhile goal to achieve eventually. For example, when a piece of important engineer equipment is conducting obstacle emplacement during the defensive phase of an operation, a trained machine learning network can determine the remaining use of the vehicle from its sensor data. Just as LTG Mohan points out, this represents a more proactive approach that allows maintainers to fix important equipment before having a critical shortage of it at a very inopportune time. As for recommendations to implement predictive maintenance initially, the vibration sensors and temperature gauges mentioned earlier must be included in all newly procured Army equipment. The data from those sensors must go into a centralized system such as Global Combat Support System-Army so that we can start gathering data that can train a machine learning system with measurements that are associated with vehicles breaking down.