The point I was responding to was that you cant start them when they are stacked, but you can push-start each car before it is stacked, using the inertia that train has already gained. I’m thinking “stacked” is the wrong word, because I can’t Google anyh hits using that expression to shed light on the subject.
Compressed maybe.
I started with ‘stacked’ earlier because I couldn’t think of anything better. Googling, compressed seems like it might get you some more data. I’m sure there’s a real word for it, I’m just not sure what it is yet.
There’s multiple wiki pages on how trains are coupled and the best I could come up with is compressed and tensioned, I’m surprised this isn’t easier to find.
The term railroaders use for the free play between cars is “slack”, the action of “stacking” is called run-in, stretching is run-out. Engineers also speak of having their train “stretched” or “bunched”.
SS
As mentioned, a pushing locomotive could draw out the slack by backing up first. But anyway the issue you describe was a lot more important for steam locomotives than it is for diesel-electric ones. The general rule of thumb was that a steam locomotive could pull any train it could start, but a diesel locomotive can start any train it can pull. Meaning that the limiting factor for a diesel is generally motor current and therefore temperature, and eventually damaging the motors, if it can’t get the train up to a reasonable speed (where the motor current then decreases), say 11-13mph for a lot of classic North American type dc motor diesel loco’s. The problem isn’t the initial start from zero, it’s when the loco can’t get the train up above the speed where continuous allowable motor current is exceeded before the short time current rating of the motors expires.
Anyway there’s no reason from a signalling and safety perspective to push a freight train just from behind. Unlike a commuter train it’s not going back and forth over the same route, nor is there is there a convenient place like a passenger car for the engineer to control the train when going in the locomotive trailing direction.
However the reason freight trains often have locomotives at the front and rear, or mid train, the latter ones being temporary ‘helpers’ used in mountainous sectors of the railroad, is the strength of the couplings. If you just keep adding locomotives to the front of the train to overcome steep upgrades, sooner or later the pulling force exceeds the strength of the coupler on the first freight car. If locomotives are added to the rear or mid train, the maximum pulling force on any one coupler is reduced.
Air [pneumatic] lines, not hydraulic.
On slack: I once saw a long freight start up. First the locomotive was used to back the train up as much as it could. Then it reverted to pull mode. I am pretty sure that what they were doing was creating slack to make it easier to start. This happened in Japan if that matters.
I remember as a kid standing by the CP line in north Whitby, watching the trains. A freight would start out by the cannery, and there would be a quick succession of bangs running down the train as the slack let out between each car in turn. Once all the slack was gone, the crew could really start to accelerate. (I knew what it was, even if I didn’t know what to call it…)
Only a few lines in North America use such equipment: A couple dozen serving New York or Philadelphia, and one Chicago line. All the rest are locomotive-hauled.
My commuter line in Montreal is the only electrified passenger service in Canada. Until 1995, it used locomotives, some going back to 1918 or so, but now they use permanently mated pairs consisting of a driving car and a drone. They typically run 10 car trains, with a powered car at each end, but a few of the drones have driving cabs so they could run 2 car trains, although AFAIK, they never have.
With the right control over the brakes so they all slow down almost perfectly equally, yeah. Assuming the cars are of varying weights, I would expect you to have to control each car’s brakes separately, as a given amount of braking may slow a lighter car faster than a heavier one, and mess up the whole thing. You’ll want the cars as far apart as possible if you’re going to be pushing them afterwards, so any lack of control while reversing just makes it harder to start up again. If you can’t control the brakes well enough that all cars stop simultaneously, I don’t think decreasing the rate of deceleration is going to help much.
ETA: How much can carriages even do this? Aren’t they generally coupled in such a way that there’s very little movement relative to one another? That’s obviously the case for passenger trains where it’s possible to walk directly between carriages.
There may be a regional language difference here. In the US, subways and similar urban rail are called “rapid transit”. These do use electric multiple-unit trains, like all subways in the world.
The term “commuter rail” specifically refers to routes longer than “rapid transit”. These trains are often push-pull trains consisting of bilevel coaches and diesel locomotives.
European (and probably Asian) railway experts are puzzled by the American practice of loco-hauled rather than multiple-unit suburban trains. In some places (New Jersey Transit comes to mind) MU equipment has been replaced by locomotive-hauled trailers even on lines with third rail. I think some of the difference is cultural, some is labor practices, some has to do with financing, and a big chunk of it is due to our rather curious brute-force approach to safety (namely, FRA buff-strength requirements).
NJT re-introduced DMU’s on one line a few years ago. I recall the original ones (in 70’s or 80’s?) on their main lines into Hoboken. One issue was that smoke at idle filled the whole semi enclosed station. A locomotive OTOH is always at the outer end of the platform and its smoke being vented up higher also. But of course today’s electronically controlled and/or common rail injection diesels smoke a lot less.
http://www.eastsiderailnow.org/dmu.html
Another vendor with recent success various places in US with DMU’s is Nippon Sharyo.
http://www.nipponsharyousa.com/products.htm
Despite the relatively paucity of DMU’s in the US, US based (though highly internationalized) Cummins is a major supplier of engines for DMU’s worldwide.
The relative unpopularity of EMUs and DMUs in the US is attributable to the FRA more directly and concretely: FRA considers every such car a locomotive for maintenance and inspection purposes, so that EMU cars are out-of-service for inspection more frequently than unpowered coaches. With that, it’s not surprising that NJT would go for electric locos but that Metra didn’t but stuck with EMU for the new Highliners. 
Freight cars can have up to a foot of slack at each coupler, so 2 feet of slack per car. With long trains, the locomotives and first cars will be moving minutes before the last car begins to move. The flashing rear-end device at the rear of the train transmits the brake pipe air pressure via radio to the locomotive, and also has an accelerometer that tells the engineer when the last car is moving or stopped.
Passenger cars don’t use slack action; additionally, they use ‘lock tight’ couplers with shelves on the top/bottom so that they can’t disconnect via vertical motion. Many times you’ll see a derailed passenger train with the cars on their sides, but still coupled.
Freight trains in the U.S. use ‘direct release’ air for the train brakes. This means that once the engineer makes a reduction and sets the brakes, he cannot gradually reduce them. He can make another brake pipe reduction and apply MORE pressure, but to release the brakes he must do a full release. As it takes some time for the air compressors in the locomotives to build air pressure back up through the train’s brake pipe, doing this several times in a row can lead to a runaway. Too many sets/releases is known in the industry as ‘pissing away your air.’
Passenger trains use ‘graduated release’ air for the train brakes. They are able to make a set, then gradually reduce the brake application.
Managing slack action is one of the biggest parts of learning to operate a freight train. Except when there’s distributed power in the middle or rear of the train, the engineer will use grade + the automatic (train) brake + either using or bailing off (disengaging) the independent, or engine, brakes. In general, on a level grade, if you make an application and bail off the engines, the train will stop stretched. Keeping the independent (or dynamics) engaged will, in general, bunch the slack.
A stretched train on an up grade is a problem–many times it will stall and have to be ‘doubled,’ where the train is broken in half and moved a piece at a time.
So, on switchbacks, the train would have had gone through repeated stretching/compressing cycles on the way up the mountain.