Ben's Website

Serious musings

Improving Freight Trains

Does US freight rail need to improve?

I've watched and read the coverage on the derailment in East Palestine, other recent mishaps, and the labor contract negotiations with interest and disappointment. For now I'll reserve my opinions on the overall challenge of rail transportation of people and freight and merely describe two improvements to freight rail that have been floating around my head for years and haven't found a better outlet.

  • enable multi-dimensional movement in railyards
  • increase car utilization

To be clear, I am not a rail professional, just an engineer and casual observer of the rail freight system. As such I'll certainly get some of the terminology wrong and present as new concepts that are known within the industry.

Backround observations on freight rail

Unlike semitrucks, where each trailer is independently moved by its own engine and driver over a 2D world, trains are inherently linear, each car connected end-to-end and confined to the current track. Adding, removing, or reordering the cars involves the entirety of the train: the half containing the car to be moved must be separated at one of its ends from the rest of the train, then both halves must be moved to isolate that car onto a side line, and then the train rejoined. Every operation on the train is then serial; all of the other cars are forced to wait for this operation to occur, and if the switchyard is limited, this operation has to occur on the main, through-going line, thereby preventing other trains from passing.

This switching constraint has not been solved; if anything, trains have grown in length as railroads have prioritized the movement of bulk commodities that are uniquely transportable on rail over other cargo that has more options. The prioritization of bulk commodities has led to cars designed for, and car-owning companies built around, long-term use cars. In fact, if you look closely at a freight train, you may be able to see that each wheel-axle set is not even connected to the car, rather the car truck is simply placed on top of the wheel-axle. (This video provides a ton of observations:

  • this is a field operation, even when performed in a yard
  • 3 person crew
  • operations are entirely manual, requiring significant movement of and about the components
  • only 1 specialized machine: the hydraulic car jacks
  • the truck assembly is lifted multiple times
  • there is no intelligence or sensors on the truck, just metal
  • large tolerances on most parts
  • 20+ minutes )

One takeaway is that the design incentives and standards for train cars are not very similar to those we are more familiar with, and these differences prevent freight rail from attaining the kinds of flexibilty we expect in road freight. Changing these incentives is the only way to actually improve rail freight (safety, efficiency, timeliness, maintainabilty, etc.); requiring additional safety devices or closer inspections are not going to put rail on a path to avoiding incidents like East Palestine.

Incenting rail freight innovation

Incentives can be changed in two broad ways. First: directly by the responsible parties through government regulation, commercial contracts, and financing practices. Second is by making improvements within the existing structure that promise even greater performance in some alternate, presently un-incented structure.

One encouragement of this latter form is in the operation of intermodal yards where train cars are un/loaded and trains are dis/assembled. COVID showed these yards to be significantly constrained by how quickly they could move goods between transportation modalities, eg from interstate train to truck for regional delivery. As many of these goods arrived on container ship, they are also transported by container-carrying train cars and can be unloaded by the same types of container forklifts and short-haul trucks as at the port. And both forklift and local truck are capable of moving a container about a 2D world, leaving the 1D train behind. These yards are being expanded as containerization spreads inward from the ports.

But why should the forklift be restricted to the container, why can't it operate on whole train cars and non-containerized cargo? After all, removing a container from a train car just leaves an empty car 'stored' in a moving train.

Enable multi-dimensional movement in railyards

Well, here the design of the train car with detachable axles and track-direction car couplers, not to mention the un-optimized mass of the car, make attempts to move whole train cars vastly more expensive. To my recollection the only examples of moving a train engine or car under some external power involve rotating or translating an entire section of track, for instance rotating a locomotive 180° to prepare for a return journey.

We can observe that all of the freedom of palletized and containerized loads comes from their having two independent means for supporting their load. Both pallets and containers can be placed on flat ground and they can be lifted by associated apparatus: the forks of a forklift or the gantry gripper of a container mover. Rail cars are only supported by their carriage wheels; when they are moved by crane, they typically use a frame that interfaces with the carriage wheels and then transfers the load above the car to the crane lift point. Unlike a pallet, this frame interferes with the car's normal operation, it's either rolling on track without the frame or being lifted by a crane and not touching a track.

To add a pallet-like interface to a train car, then, we must either modify some part of the wheel truck assembly or locate some alternate lifting point. Most freight cars appear to have wheel sets like those in the video embedded above, having a significant and exposed axle connecting the two wheels. While needing analysis, a lift entering under the car center might extend a cradle to engage both axles on the forward and rear carriages, so that all four axles can be simultaneously lifted, along with the car and its contents. The impact of the car's significant weight is lessened by the short lift distance, at minimum the distance to raise the wheel flanges above the rail. The broad outline is a squat vehicle having a perpendicularly forked lifting apparatus and widely set tires or treads of the style already used in large structure moving apparatus.

An alternate approach would be to temporarily install car adapters at some yard entry station or under a stopped train, the design being to again interface with the car axles while providing accessible interface points. The overriding point is that these preparatory operations need to occur quickly, as they directly impact the overall speed of operation.

Increase car utilization

With the ability to move a train car, where should it be moved to?

Well, lean manufacturing and warehousing was a large trend through the 2000s, focused on producing only what was needed and minimizing inventory sitting in warehouses. Lean is generally good because it makes businesses more responsive to their markets, linking the retail price more closely to the production cost. From this perspective it's pretty easy to see trains as slowly-moving warehouses, carrying goods with very long shelf lives and lengthy paths to market. I'm no expert on lean — though I am sure that its application to commodities has been explored — the point is that even commodities can benefit from faster market cycles, and faster market cycles respond to incentives more quickly, enabling faster improvement.


I'm dubious that technologies like positive traction control or more frequent track-side train sensors will appreciably improve rail freight simply because we should never prefer a remote, infrequent solution to a local one; that is one directly and continuously controlling the phenomena of interest. If the problem of derailments is in the operation of the cars, we need to find technologies and strategies that fix the cars and allow them to be operated without catastrophic failure.

As sketched above, this requires changing how freight cars are designed and used by increasing their utilization, which will allow cars to create more value and thereby enables them to include more technologies that allow them to be operated in safe, maintainable way. And all of this is made possible by the more flexible movement of freight cars in railyards, which may enable faster sorting and assembly of short, high-value, high-speed trains, and allow the better utilization of the rail for all cargoes.