# Why are multiple locomotives attached only to the front for larger trains?

For long freight trains and those that will be climbing to stations at higher altitudes, an extra or two locomotives are attached to the front. I've always wondered why.

For argument's sake, if there are 30 bogeys each weighing 10 tons, the three engines are pulling a combined 300 tons. Each must be applying the exact same pull else if one is pulling harder than it is effectively taking on all the work with the other two idling. Plus, the load on the first coupling is 300 tons, on the second 290, and so on.

On the other hand, if locomotives are placed after every 10 bogeys, then each is pulling 100 tons only.

• faulty premise - en.wikipedia.org/wiki/Distributed_power Jan 24, 2022 at 21:07
• There's got to be an upper limit of how many cars you can push on before you'll derail the one in front of you, but the only limit on pulling is the coupler strength and HP? Definitely more than ten; public transit runs backwards all the time with about that. Jan 25, 2022 at 2:33
• Why is your question specific for freight trains? It also happens for passenger trains. Jan 26, 2022 at 8:44
• No particular reason other than I've mostly seen multiple engines in freight trains. Jan 26, 2022 at 9:33
• @gerrit Except that shown example isn't about load distribution,. These are two complete trainscouppled to save on crew time and simplify handling during a common leg. Jan 26, 2022 at 13:31

For long freight trains and those that will be climbing to stations at higher altitudes, an extra or two locomotives are attached to the front. I've always wondered why.

As usual there are multiple issues. Most important is landscape. If more power than a single locomotive can provide is not needed for the whole trip, but only a short stretch, like climbing or crossing a mountain range, then it would be a huge waste of resources to attach them the whole trip.

For example, take a 10 hour trip time, where all, except for a 1 hour stretch, can be pulled by a single engine. That stretch needs the power of 3 engines. Fitting a train with three engines would solve this, but occupy all 3 engines for the whole 10 hour trip time. More so two of them would only add weight to be pulled for most of the time.

By stationing two engines at the climb, a train can run on its sole engine up to the climb, get these two engines added in front, climb for one hour with three engines, after that, let the additional ones go and continue for the rest of the trip with just a single engine.

Sure, not a big saving when just looking at a single train per day, but what if there are 4 trains per day? To equip each with 3 engines the company would need 12 locomotives. In contrast, with stationing two engines at the climb, these 4 trains would only need one engine each, so with the two stationed ones, only half the number of engines need to be bought and maintained - that's quite a saving, isn't it?

And that brings us to adding to the front: it's simply the fastest and easiest way to add them. That can be done in minutes, while adding engines in the middle may easily take an hour of shunting.

Of course this only works with short-term power need and considerable train frequency. Routes that only feature very few trains (like some desert railroads in western Africa), or have their climbs not centralized will not benefit here. They have to run with multiple units all time.

Which brings reasons for placement:

On the other hand, if locomotives are placed after every 10 bogeys, then each is pulling 100 tons only.

• As already mentioned, it's least effort to put units in front, even if multiple units.

• Grouping units upfront simplifies coordination, as control (electronics) can be wired up with a simple cable between these.

• Units interspersed would need to be radio controlled (wiring along freight cars would be complex and expensive).

• With distributed engines the calculation isn't as easy as mentioned, as an intermediate engine should never push wagons in front. As soon as an engine starts to push, the couplers will start to oscillate, introducing heavy wear. Thus any engine will only be able to cover for a fraction of wagons behind, so all coupling sections will always stay under force.

• Distributed Power does reduce wear in curves, allows lighter (or less well maintained) ROW.

But most important of all, the pulling force a knuckle coupler can handle is way greater than anything a single locomotive can deliver. A Janney coupler, like used in heavy freight trains, is specified for 4,000 kN (Kilo Newton). That's way more than a reasonable sized locomotive can deliver. For example an EMD SD70 can deliver at maximum 850 kN. In other words four major engines at full power would still leave a safety margin >20%.

Bottom line, a train that can be pulled by two or three engines will be pulled upfront. Only if that limit gets exceeded will distributed power make sense. And it will be only inserted if used on the whole trip, or long parts thereof.

An exception thereof are dividing trains, that is trains running over some stretch along the same route, to be split up at an intermediate station, heading for different destinations. Joining them for the common section can save the need for a second crew.

• Anyone who's looked at control systems will recognise the potential issues with trying to drive a long chain of cars that are all coupled together with a bit of slack in every coupling - it quickly becomes incredibly hard to share the load without constant "shunting" (oscillations due to overshoot/undershoot) as the couplings take up the slack. If you imagine each carriage connected by a length of rope, and that the loco drivers must somehow keep all the ropes taught all the time, you can see how difficult it becomes as the locos are placed away from each other. Jan 26, 2022 at 13:10
• @JohnU: in fact, what you describe is a popular reason for traffic jams on a straight road - not any singular event, but just people creeping up on one each other, and slight oscillations turning into a big "chaotic" event at the end. Also your rope example is great - anyone who has tried to pull another car with a static (non-stretchy) rope can attest. :)
– AnoE
Jan 27, 2022 at 10:10
• @AnoE - great example! However, even with humans out of the loop, as an automated PID control problem it's still very hard to solve without a lot of extra parts. Jan 27, 2022 at 12:31
• @JohnU The simple Solution, used for quite some years, is to stay with force delivered for each loco below the force needed for the wagons behind. Leave a good margin and all works out whithout much ad hoc calculation. Jan 27, 2022 at 13:05

Each must be applying the exact same pull else if one is pulling harder than it is effectively taking on all the work with the other 2 idling. Plus, the load on the first coupling is 300 tons, on the second 290, and so on.

The locomotives are controlled by power or throttle setting, not by speed setting. If power is set at 65% and there is a few percent difference between each locomotive then they will contribute in proportion to their power. The first coupling will have roughly 100 tons tension, the second 200 and the third 300. The coupling between the first and second car will have 290, etc.

For long freight trains and those that will be climbing to stations at higher altitudes, an extra or two locomotives are attached to the front. I've always wondered why.

I can think of several reasons:

• Ease of coupling at the origin and decoupling at the destination.
• Ease of re-fueling.
• Short walk to sort out technical problems and when starting the engines.
• Reliable cable connection between locomotives. There won't be control cables on each wagon to pass control signals through so that would force the use of radio control for the slave locomotives.

A bogey is one swiveling wheel set - typically four wheels on two axles. You mean wagons or rail cars.

• Instead, I've seen one engine in the front, and one behind the last railcar, especially in climbing condition.
– r13
Jan 24, 2022 at 14:43
• @r13. You might enjoy cartalk.com/radio/puzzler/all-aboard. Work it out before looking at the solution! Jan 24, 2022 at 15:28
• Watch the beginning and near the end of this video to see how the train was "pushed" by the trailing locomotive. youtube.com/watch?v=hs1XGX9veOg
– r13
Jan 24, 2022 at 16:16
• You mean you haven't seen a line of bogeys coupled together? Hint: logging railroad. Jan 25, 2022 at 1:44
• "There won't be control cables on each wagon to pass control signals through so that would force the use of radio control for the slave locomotives." -- this is regional; US passenger railways (both intercity and commuter) do push-pull operations and do run trainlines through the length of the train, allowing a locomotive to be fitted to one end, and then operated from a cab at the far end. In that case, loco on one end is simply operational convenience - conductors can move about the train and... Jan 25, 2022 at 2:40

Other answers have very valid points about speed of connection, cost effectiveness of using helpers only where necessary, communications, etc, but Mazura hit on a VERY KEY point in his comment that's been missed in all the answers:

### You cannot push a long string around a curve without it falling off the rails.

As consists have grown larger (mile plus long consists in the US West have been very common for 30+ years) pushing that around even the gentle mainline bends can simply push a car off the track causing an expensive, time consuming and dangerous derailment. Often, the steep climbs that require pusher engines have much sharper curves than are allowed on flat land mainline runs, meaning that pushing is even more likely to cause a derailment. Pulling from the front won't cause a derailment (all things working as they should).

When steam ruled the rails, it was common for a helper engine to be a pusher added to the back of the consist. Here's a short article about communication between the lead and helper locos. In the late steam era, very powerful locomotives were built that could handle trains nearly as long as we see today. These longer consists, though, had to be broken up into sections when they got to the big grades, because a helper at the back was common, and they were simply too long to handle being pushed.

Here is a gratuitous video of a modern steam tourist excursion with 2 helpers up front and one bringing up the rear: https://www.youtube.com/watch?v=XPIFc-bdBsY

• Why is pushing more likely to derail than pulling? Won't pulling create a force to the inside of the curves? Jan 25, 2022 at 19:44
• @PaŭloEbermann Set up a series of pegs (as an analog for the rails) and pull a literal string around a curve. Now try pushing the string around that curve. The exact same thing happens with the rail cars. Because of the slack in the couplers and the free rotation of the bogies under the car bodies, they just don't push well except in fairly straight lines, and even then only at slow speeds. I don't understand all the physics of it, but having grown up in a railroad town, involved in model railroading, and my step-dad working for the Union Pacific, I know it's a fact. Jan 26, 2022 at 12:15
• @PaŭloEbermann If you're pulling and one of the car bodies goes out of alignment behind you, you get more force on the "outer" wheel to pull it back into line again. It's inherently stable and there's very little load on the wheel flanges. If you're pushing though, you get more force on the "inner" wheel which wants to push it more out of line (like reversing a trailer). The wheel flanges will keep it on the rails for smaller forces, but if you get get enough force to push the wheel flange over the rail then you're derailed. Jan 26, 2022 at 14:16
• Thanks for the help, @Graham! Jan 26, 2022 at 14:18
• Compare pushing a rope lying on the floor vs. pulling it as an extreme example... @PaŭloEbermann
– AnoE
Jan 27, 2022 at 10:12

Very large trains are not only powered from the front. The front is preferred, though, since it is the easiest to couple and uncouple extra locomotives and a wired connection is more reliable than radio. The crew travels in the cab, so there is no walking or backing the consist up to pick up the person who made the coupling. Many territories will have all their locomotives paired up back to back to aid in reversing - the crew just walks to the other cab - so double locomotives in the front are very common.

The number of power units that can be put at the beginning of the consist is governed by the knuckle/coupler strength. In the US, trains are run as "Distributed Power" via radio control for the units at the end or in the middle. Tough grades or very long trains may have extra locomotives put in both the middle and at the rear.

• "the person who made the coupling"; is physical presence still needed for modern automatic electric train coupling? The idea of someone physically coupling train wagons together sounds very 20th century. Surely that can be automated? Jan 26, 2022 at 8:39
• @gerrit: A human crew can also inspect equipment to look for things that might fail during the trip; something much harder to automate. I'm not at all surprised that humans are still involved, since it makes sense to have humans around anyway. Jan 27, 2022 at 18:16
• @PeterCordes Oh, I certainly I expect that humans are involved in driving a train! What I didn't expect is that a human would have to physically walk to the back of the train to couple one wagon to another, unless something goes wrong. Jan 28, 2022 at 8:06
• My point was, humans get bored easily. If you want them inspecting important things like locomotive to train couplings every hookup, it makes some sense to have them physically involved in making connections, so they don't skip inspections. (OTOH, lazy or under-manned train crews not setting enough manual brakes have led to disasters; en.wikipedia.org/wiki/Lac-M%C3%A9gantic_rail_disaster so leaving everything up to humans isn't always ideal.) Jan 28, 2022 at 8:11
• @gerrit, making up trains is still manual. The bar must be thrown manually to uncouple after the air line is unhooked manually. Cars will couple automatically, but the air line must be made by hand. These could certainly be automated if railroads wanted, but the million-car current inventory is a barrier to new tech. It took decades in the US to move from journal to roller bearings on freight cars. Jan 28, 2022 at 15:26

My intuitive attempt of reasoning, to be taken with a grain of salt since it's basically just an educated guess:

Couplings naturally have some amount of play. As long as all couplings are under constant pull force (all locomotives at the front), this play doesn't come into effect and the whole train moves continuously as a single unit.

But if we also put locomotives in the middle or at the back, small load variations will inevitably lead to some couplings switching back and forth between pushing and pulling, resulting in continuous coupling movement and wear, and possibly even some kind of resonance effect that further increases the oscillation.

• your intuitive reaoning is right "on track" (lol). This is known as slack, and managing slack is a key role of an engineer. Speeding up a train from stop can result in the last car getting a huge jolt resulting in damaged goods and broken couplers. Jan 25, 2022 at 16:25

To optimise efficiency with multiple locomotives, none of the driving wheels must slip. This means that coordination is necessary between the locomotives.

Passenger trains have been worked as multiple units for more than a century with driving motors (electric or diesel with mechanical transmission) distributed along the length of the trains, as you suggest. Coordination is by a train line which is a multi-wire connection communicating the throttle opening (and gear, where appropriate) between the units.

This is relatively easy for passenger trains because they are short (compared with freight trains) and all the cars are normally owned by the same company, so it's easy to ensure compatability between the units.

Freight trains are much longer, the wagons may be owned by different companies and are often of widely varying types. They are also frequently added and removed during a journey. All this makes it much simpler to put all the locomotives together at the front (as was done in steam days when each locomotive had its own crew and coordination was achieved by whistle signals) rather than try to run a train line to remote locomotives.

Now that there is only one crew, it's still convenient to have all the locomotives together for fault-finding and start-up / shutdown, which is normally not available remotely.

More recently (I think in the last 20-30 years), certain countries permit distributed power with a radio link between the head end and a locomotive in the middle of the train. It's important that the remote locomotive behaves appropriately when the radio link is lost (this often happens in long curves in rock cuttings).

In steam days, one or more banking locomotives would push trains up steep grades from the rear without coupling, then drop back at the summit. This is now considered dangerous because of the difficulty of coordinating the head-end and banking locomotives (for example, if it's necessary to make an emergency brake application, the banking locomotive will probably keep pushing).

• What happened at the end? The final sentence ended abruptly without being concluded.
– Fred
Jan 27, 2022 at 19:10
• @Fred Thanks for the comment. I'd forgotten to log in before posting and didn't notice that the post was truncated. Jan 27, 2022 at 19:33

Not really a true answer to the question as posed (and others have answered quite sufficiently that the premise is false). But, for additional background information, there are several other reasons beyond what's already been mentioned that have lead to distributed power (at least in the USA) being much more common now than in the past.

• Longer trains allow for more efficient use of resources and enhance track capacity, as one longer train with sufficient space behind it can take up a lot less space on the track than two smaller trains, each with that same sufficient space behind.
• Having distributed power with remote control can lead to safer and faster breaking in an emergency. Train brakes (at least in the USA) are controlled by varying air pressure in the brake line that runs the length of the train. With all engines in front, any brake application will need to have that pressure propagate through the brake pipe the whole length of the train, which can be quite a few seconds. Having distributed power allows all locomotives in the train (front/middle/back) apply brakes simultaneously, which allows the whole of the train's brakes to start applying more rapidly than if the brake signal needed to go car by car the whole length of the train. (in modern trains even without distributed power there's a rear end device (nicknamed a FRED) which can monitor and provide emergency braking functions, but they typically can't do anything but provide emergency braking control, so having an engine at the back can have that brake control logic that a FRED can't provide)
• Modern radio technology allows for better control and communication between the front engine, and the radio control units behind, even through tunnels and mountainous terrain, which historically have been challenging environments for radio signals.
• Finally, longer and fewer trains means fewer crew, and with distributed power all being remote controlled, it only takes one or two front end crew to run a really long train. In the past would require multiple trains, or crewed helper engines. After fuel, I believe employee costs are the second highest cost for most railroads, so reducing employee costs helps their bottom line.

In Siberian mountains, it is very usual to have full-time "double pull" due the steep inclines, and "extra push" loco attached at some complex segments of a route to make sure the train to stay on its schedule. Like 2 x 2TE10 (4 sections in sum) as pullers and trailing CHME3 as a pusher.