Before going into details, there's something that has to be understood about the way train car brakes are simulated - all of the advanced brake simulation for the train brakes, including the auxiliary reservoirs, train brake cylinders and the brake pipe itself, are done by the locomotive's script and blueprints, not in the individual cars that make up a consist.
To kick off, I'll try to respond to the points raised in Tyler's post. It would take me several pages to explain all the details, but I'll try to be succinct.
The DTM GP50 and GP30 do not have scripted (CMP) brakes. The DTG scripted brakes took a round-about way to achieve a degree of realism and failed IMO. I'm not sure if there is a train simulator available to the public that fully mimics the modern Westinghouse air-brake. There are several that get darn close, but I haven't seen any that recreate air signal propagation. I'm having difficulty in explaining the process, so I'll refer you to Al Krug's explanation: http://www.railway-technical.com/brake2.shtml
The SS SD40T-2 does indeed simulate air signal (pressure wave) propagation. In fact, so do my FEF-3 and ATN Consolidation, but in a different way. I based the SD40T-2 implementation on what I'd already done for the two steam locos, but then re-wrote some of it to make it even more realistic and accurate. That included the part that deals with air signal propagation (it's simulated to some extent in the steam locos, and the FEF-3 even has a HOT device where you can watch how changes in the tail-end brake pipe pressure, shown by the HOT's digital display, lag behind the pressure at the head end as shown by the analogue gauges in the cab).
Propagation is the idea that each freight car can "help" the next realize an application or release is in progress. I wish I could explain this part better.
When freight cars sense a rise in air pressure during an air-brake release, they fully release rather than releasing at the rate of the brake pipe psi increase. All it takes is an increase of a few psi to cause a modern triple-valve to proceed with the release process. It does take time for the release signal to be sent to the end of a long train, so the release may seem somewhat slow.
In the SS SD40T-2, the brakes start to release when the brake pipe pressure rises by a minimum of 1.5psi above the pressure being simulated in the auxiliary reservoirs. The triple valves then expose the brake cylinders to atmosphere and the air is exhausted from them, releasing the train brakes very rapidly. The script also simulates the speed of propagation of the pressure wave - it travels down the brake pipe at 150 m/s - and the fact that the pressure doesn't change on every car in a consist at exactly the same time. Therefore, the train brakes don't release on all cars in the consist simultaneously. Instead, the brakes on each car release when the pressure wave reaches them, and so the time it takes for that to happen on the entire consist depends on the length of the consist.
To see this in action, you have to understand that the gauges in the cab don't show the pressure in the brake cylinders of the cars in the consist (the train brake cylinder pressure). The locomotive (in real life) has no way of measuring it. However, in TS, you have the F5 HUD which does show a train brake cylinder pressure - it's an indication of the brake cylinder pressure averaged along the entire consist. So, as the brakes apply or release along the consist, that indication will go up and down, but remember that it's an average and what it's really giving you is a way to judge the overall brake force on the train. When it shows 0psi, it means all the brakes, on every car in the consist have released. Conversely, when it shows 65psi, it means that all the brakes are set fully. When it shows something in between and is changing downwards, it means that some brakes (those nearer the head end) are released but others (towards the tail end) are still set (because the pressure wave hasn't reached their triple valves yet). If the value is changing upwards, it means that some, but not all, brakes are set on the consist (the brakes set at the head end first and the tail end last).
When you move the automatic brake handle to the Release position, the brake pipe pressure (that you see on the gauge in the cab) rises slowly as air is fed in from the main reservoir via the equalising reservoir, but the train brakes don't release immediately. As soon as the brake pipe pressure at the head end is 1.5psi higher than the simulated pressure in the auxiliary reservoir on the head end car, the train brake release is triggered at the head end and travels down the brake pipe at 150m/s. So, on the F5 HUD, you see the brake cylinder pressure stay steady for a short while, until the release on the first car is triggered, then it rapidly goes to 0psi, but the longer the consist, the longer that takes.
Air pressure in a brake pipe can also mimic fluid dynamics. If I make a heavy set at the head end, air rushes forward from the end of the train to compensate for the lack of pressure at the head end. It then can "pile up" at the head end because of it's violent rush forward. Then, like a wave of water, the air pressure will head back to the rear, moving back and forth for a short period of time. In some cases, if you make a heavy set and close the trailing unit's angle ** before this sloshing back and forth stops, the sloshing can actually create enough of a psi change to tell the triple-valves that they should release where the air piles up. This is one reason why you should never "bottle" the air: closing the angle cocks on both ends of a consist to be dropped. A full release can be caused by the sloshing back and forth.
That's actually one of the main purposes of the equalizing reservoir. They're designed to absorb that shockwave. What you're describing used to be a much more serious problem before the equalsing reservoirs were invented. This sloshing is one thing that isn't simulated (it would have been a step too far).
Also, a triple-valve with an internal leak can feed air from it's reservoir back into the brake pipe, and can also raise the brake pipe's psi enough to trigger a full release.
Interesting point, but it hasn't been simulated.
Another interesting note: triple-valves react to the speed at which the brake pipe is vented as well. In any normal application, pressure is vented at a controlled pace and the triple-valve can sense this and trigger a normal application of the brakes on each car. However, if an emergency application is made, the pressure reduction is dramatically quicker. Triple-valves can sense the difference and trigger an emergency application rather than applying at the normal rate.
This is simulated. In fact, I've also simulated the way that the triple valves automatically connect the emergency reservoirs to supplement the air in the auxiliary reservoirs, so that you still get the emergency brake application even if you´ve pee'ed away the air in the auxiliary reservoirs.
The script also simulates the way that a break-in-two causes that sudden drop in brake pipe pressure, at a rate high enough to trigger the emergency brakes, and the way that the emergency brake application then propagates rapidly along the rest of the consist.
As an engineer, you watch the equalizing reservoir gauge to make an application. You decrease the equalizing reservoir to the pressure you want and the brake pipe pressure will follow and match it.
That's how you do it now with the advanced brakes. The time is takes for the brake pipe pressure to follow and match the pressure in the auxiliary reservoir depends on the length of the brake pipe (the length of the consist). Furthermore, the script simulates the pressure difference between the head end and tail end of the brake pipe, so the gauge in the cab can be showing the brake pipe and equalising reservoir pressures as equalised, but the tail end brake pipe pressure could still be catching up. You see the effect of that when looking at the train brake cylinder pressure shown in the F5 HUD - if the tail end is still catching up even though the head end has already caught up, the F5 HUD brake cylinder pressure will still be changing.
Note that the braking force on the consist depends on the train brake cylinder pressure shown in the F5 HUD. For example, if you release the brakes, and the gauge in the cab is showing the brake pipe pressure at 90psi, the tail end could still be catching up, and consequently the F5 HUD train brake cylinder pressure would still showing more than 0psi (but falling), so the train would still be braking for a short time. Similarly, when you apply the brakes, and the brake pipe pressure needle in the cab is showing that it has equalised with the pressure you set in the equalising reservoir, if the tail end is still catching up with the head end, it will be some time (depending on the length of the consist) before you actually feel that the brakes are all being applied.
Then watch the Air Flow Meter to determine when the "sloshing" has stopped. The AFM is also used during a release, as it can tell the engineer when the brake pipe is still charging even if the brake pipe gauge reads a full release pressure. The AFM will settle down when all of the train's auxiliary reservoirs have been recharged. Very few simulated locomotives include a working AFM.
Unfortunately, the SS SD40T-2 doesn't have a working AFM, but it would actually be very easy to script, because all the information it needs is already in there - the script simulates the recharging of the auxiliary reservoirs, which takes longer than just raising the brake pipe pressure (since the auxiliary reservoirs are recharged through a narrow feed groove in the triple valves, so it takes some time for the pressure in the auxiliary reservoir to equalise with the pressure in the brake branch pipe and close the feed valve).
Last fun fact for the night, as I'm tired and not really sure if I'm making any sense.
If you have a long train, let's say 120 cars, there will be small leaks throughout the train. A terminal air test should determine if these leaks are within an acceptable rate. A real mind-blower for me was learning that the EOT pressure at the rear of the train may never reach the full brake pipe pressure...ever. Your feed valve may be set for 90psi, but the rear of your train may only be able to reach 80psi. The fun thing is its all relative, and a 10 psi reduction on the head end will trigger a 10 psi reduction at the rear resulting in equal braking even though the actual numbers are different.
The terminal air test isn't simulated, and neither are the small leaks, although it could be done (maybe in the future, if there's enough interest). The equal braking is indeed simulated correctly.
Something else that's simulated accurately, and which is extremely important, is the effect of peeing away your air. If you do that 10psi set before the auxiliary reservoirs have recharged, the reduction from 90psi brake pipe pressure to the pressure currently in the auxiliary reservoirs, will have no effect. The only part of the set that does have an effect is the drop that takes the brake pipe pressure below the pressure in the auxiliary reservoirs. Then, with less than full pressure in the auxiliaries, you get less pressure in the brake cylinders. So it's a double whammy - it takes longer for the set to take effect, and when it does, the brakes apply with less force. You can eventually get to a situation where a full service application gives you no braking at all (in which case you'd have to use the emergency brakes).
By the way, the SD40T-2 is simulating the triple valves used on freight cars, so it isn't simulating the Schedule UC type, used on passenger cars, that have auxiliary reservoir depletion protection.
