How MemComputing is keeping trains on track

Have you ever seen a train moving slowly or even coming to a stop on the track? Worse yet, been stuck at a railroad crossing waiting for what seems like hours for a freight train to pass?

You’re catching a glimpse of the fact that that train is part of a much larger, dynamic, and inter-dependent system. Keeping it all flowing in the real world is more interesting and complicated than many people may think.

Here, we’ll describe the reasons why that train may have stopped and how MemComputing can help keep everything on track, literally. 


For context, the rail network is the backbone of freight transportation in the US, shipping billions and billions of dollars of cargo annually. With four times the fuel efficiency of shipping by truck, railroad companies are a critical supply chain component to the U.S. economy. 

A fundamental challenge is that all the railway companies must share the sprawling network of tracks and other resources with each other. Therefore, trains operating in the U.S. follow highly detailed schedules and routes to efficiently move commodities, consumer goods, and passengers across the country. 

These schedules, often called timetables, define the desired order of trains as well as their departure and arrival times at junctions and platforms. In order to remain competitive, all railway companies try to optimize their timetables well in advance to maximize profits, minimize delays, and ensure customer satisfaction.

In some cases, that stopped train you see is part of the pre-planned schedule to accommodate the movement of other trains and constraints. However, even the most highly optimized schedules are disrupted by unexpected events such as:

  • Loading delays
  • Weather disruptions
  • Locomotive failures
  • Track incidents
  • Crew delays

These events force trains to deviate from the original timetable, thus affecting the running and departing times of not just one train (primary delay), but all the others in the network (secondary delays), turning the schedule effectively on its head if not rapidly adjusted. 

This is where train dispatchers come in. 

Dispatchers maintain a critically important job as they are responsible for ensuring trains move safely and reliably across the network to maximize efficiencies. 

They have the authority to assign which trains have priority to use the tracks in a specific location over a given time. This authority applies to their dispatching territory, which covers a specific length of track between two mileposts and can consist of single-track and multi-track segments. 

When a disruption occurs, dispatchers must quickly (within a few seconds) reschedule and re-route trains in their territory to minimize delays. Today, dispatchers rely on elementary decision support tools and manual efforts to get these trains back on track; a method that has proven to render sub-optimal solutions for this optimization problem at scale. 

As a result, delays can propagate excessively across the entire network, a problem that costs railroad companies millions annually due to increased fuel, labor, and equipment. 

Let’s take a closer look at the problem and current techniques to understand the complexity.

The Problem

Commonly referred to as the “Train Dispatching problem”, this problem has challenged dispatchers and researchers for decades. 

The goal is to optimally schedule the trains (in near real time) across the track network such that the trains don’t experience conflicts while adhering to operational constraints to minimize delays. 

Dispatchers must consider extensive amounts of information such as:

  • Different train lengths and speeds
  • Varying train priorities
  • Single and multi-track segments
  • Different local constraints
  • Crew availability
  • Run time

Due to time restrictions, thousands of variables and operational constraints, and lack of intuitive technology, dispatchers are only capable of generating feasible solutions for a small number of trains in a limited territory and cannot consider the effects on the entire network. 

At full scale, this problem is intractable to solve today using legacy systems. Heuristic algorithms and other techniques have been deployed to address this problem, but can only solve for small instances, or not at all. 

The larger the territory (and the more trains) that can be considered all at once, the greater the potential to optimize. Indeed, there is a need for new technologies to solve this scheduling problem more effectively and efficiently at scale. 

Enter MemComputing

MemComputing offers a revolutionary approach to solving complex optimization such as the dispatching problem. Due to its novel computational design, dispatching problems faced by today’s largest freight and passenger railroad companies can now be solved optimally at scale. 

Where existing optimization algorithms and heuristics fail to solve for the entire network and incorporate real-time information, MemComputing excels.

Solutions we have built for these problems run on our MEMCPU Platform™ and ingest live data across the network to rapidly calculate new routes and schedules for all trains in near real time. 

Advantages to the MEMCPU Platform solution include:

  1. Increased network capacity and velocity
  2. Improved on-time performance
  3. Enhanced asset utilization
  4. Significant cost savings
  5. Boosted rail customer satisfaction

MemComputing has demonstrated its solution to multiple large freight train companies, where the resulting solutions far exceeded the capabilities of current techniques. Our solution yields greater operational efficiencies and flexibility across a much larger section of the train network, while simultaneously reducing costs, carbon emissions, and uncertainty. 

When an unexpected delay occurs, our solution optimally reschedules all trains and gets everything back on “track” in near real time. 

Wrapping up

In order to remain competitive, railroad companies are constantly striving to improve their scheduling systems to improve reliability and the punctuality of their services to satisfy customers. 

This ultimately boils down to the quality of their planning and dispatching efforts, a process that is much easier said than done, especially when disruption occurs. 

MemComputing offers a novel solution to freight and passenger train companies interested in dramatically improving their operational efficiencies and resilience to disruption.

Remember that train that stopped on the tracks? If due to unforeseen congestion, MemComputing can help recompute and dispatch. If it was part of an original schedule, employing MemComputing’s solution would allow the rail company to include additional constraints regarding where, when and how long a train can stop when blocking traffic. 

MemComputing’s team is ready to advance your competitive edge with a tailored optimization solution. Explore a free account today or contact us directly at [email protected] to get started.