Transport deadlines and common demand batches in regional generator spare stock allocation

When a massive generator fails in a remote location, the difference between a few hours of downtime and a catastrophic shutdown often comes down to a single spare part sitting in a warehouse. The challenge for engineers is deciding where to keep those parts. Should a company keep a massive, central warehouse with a huge stockpile, hoping to ship a part to any location when disaster strikes? Or should they scatter smaller stocks across different regions, keeping parts closer to the machines that need them? This question sits at the heart of inventory management, a field that tries to balance the cost of holding extra items against the risk of running out when they are needed most. A key complication is that machines do not always fail one by one. Sometimes, a single event—a storm, a power surge, or a coordinated maintenance check—can cause multiple generators to break at the exact same time, creating a sudden, simultaneous demand for parts in different places. Understanding how these "batch" demands interact with the time it takes to transport a part is crucial for keeping the lights on.

A recent study by Mahmoud Abdelfattah Abbas Mahmoud tackles this problem by building a precise mathematical model to compare these two strategies: one central pool versus three separate regional stocks. The researcher did not use real-world data from a specific company, but instead constructed a hypothetical scenario with three distinct regions, each with its own rate of random, single-unit failures. Crucially, the model also included a separate stream of "common events," where a single incident would trigger a request for one part in every region simultaneously. The study assumed that whenever a part was needed, a new order was immediately placed with a supplier, and that this new part would arrive exactly seven days later, no matter what. The goal was to see how well each system could meet a strict deadline for getting the part to the site, ranging from eight hours to forty-eight hours.

The study revealed that the answer depends entirely on how much time is allowed for delivery. When the deadline is tight, such as eight hours, the regional strategy wins decisively. In the specific scenario tested with nine total spare parts, the regional setup could fulfill about 91 percent of requests within the deadline. The central warehouse, however, managed only about 21 percent. This is not because the central warehouse lacked parts, but because the time it took to ship a part from the center to the furthest regions simply exceeded the eight-hour limit. Even if the central warehouse had infinite stock, it could not overcome the physical travel time to the most distant sites. As the deadline was extended to twenty-four hours, the central warehouse's performance improved significantly, reaching about 51 percent. By the time the deadline stretched to forty-eight hours, the central warehouse became the superior option, fulfilling nearly 98 percent of requests, while the regional stocks remained steady at their previous level.

The research also uncovered a subtle but important flaw in how some companies might estimate their needs. A common approach in inventory planning is to treat all demands as if they were independent, random events, ignoring the fact that they sometimes arrive in groups. The study showed that if you use this simplified "average" method to plan for a high service target, you will underestimate the number of parts you actually need. In one example where the goal was to have a part ready 98 percent of the time, the simplified method suggested sixteen parts were enough. The exact calculation, which accounted for the simultaneous three-part demands, showed that eighteen parts were actually required. Relying on the simpler math would have left the system short by two units, causing it to miss its target.

The study went a step further by looking at what happens when a part is not immediately available. In the real world, if a part is out of stock, the request waits in line until a new shipment arrives. The researcher calculated the total time from the moment a machine breaks to the moment the part is delivered, including this waiting period. When this waiting time was included, the performance of both systems improved, but the central warehouse gained the most. With a forty-eight-hour deadline, the central system's success rate jumped from about 98 percent to over 99 percent, while the regional system rose to about 95 percent. This suggests that while regional stocks are faster for immediate needs, a central warehouse is surprisingly effective at catching up and delivering parts on time, even if they have to wait for a new shipment to arrive first.

Ultimately, the paper provides a clear, exact comparison of these two architectures, showing that neither is universally better. The choice depends on the specific constraints of the operation: how far the parts need to travel, how long the supply chain takes to restock, and how often failures happen in groups. The study emphasizes that while the math can predict the likelihood of a part arriving on time, it does not guarantee the machine will be fixed, as that depends on many other factors like diagnosis and installation. However, by separating the logistics of moving the part from the mechanics of fixing the machine, this work offers a precise tool for managers to decide where to place their inventory. It highlights that practical application requires calibrated demand, replenishment, and transport data, and that the results identify the kinds of data needed to decide whether a central warehouse, regional stock, or a mixed architecture merits more detailed analysis.