US Fright train Physics

[BNSF650]

I was just wondering does anybody think the RSC Team will fix these issues with the trains?

[Kali]

Couplers are an easy fix; everything else needs a core physics engine rewrite, and you tell me how likely that is.

[USRailFan]

I fear the physics of a fright train might be a bit too frightening to temper with...


(sorry, just had to)

[TrainMaster1]

Kali is spot on that to really get close to what an actual crew sees every day would require major work. You can get quite close though by having a team work at tweaking engines and cars and comparing them to real world performance as we did in our Cajon session. Took some doing but we had a BNSF G, X, S, Q and Z trains finish within a few minutes of what it would take for a 1:1 train cover the the same distance and with the same length and weight. So while a major overhaul would be needed (scary to see the slack action showing couplers pulled out of their pockets yet somehow the train was still together or cars pushed inside of other cars without derailing as well) there are somethings that are attainable.

Hoping that we do not have to go to such lengths every time we want to try a new route in a session as that will compromise the number of sessions we can do. Also hoping that other routes stay as close as Cajon does (albeit dated and not suitable for modern day). RSC finds itself in the unenviable position of trying to build a third floor onto a house whose foundation needs considerable work at this point. The tools are there to really make this an outstanding platform for true railroad action. Hope it gets the attention it deserves.

Nick

[Kali]

It'd be nice to have brake propagation - however anyone who understands software design might appreciate what seems like a little thing might actually be a fairly monumental change, going from working numbers out for a conglomerated whole to 80+ individual elements - and if you think coupler physics are slowing things down now, just imagine brakes on top!. What gives me some hope it's possible without major effort is that you can put the handbrake on a vehicle halfway down the train, and it visibly drags and stretches the couplers between it and the engine.

I am doing my best on engine brake systems, might be a little while before there's any visible progress on that though, I have a bit of a queue... similarily for power - if anyone has any TE graphs for any of the available ( US ) diesels, please share!

[styckx]

I mentioned this before and I know Kali gets it because he responded with what I already knew myself.

Let me just draw up a hypothetical scenario.

"Ok guys. I have 100% deciphered all TS 2012 physics and compiled it in a easy to understand spread sheet accompanied by lua templates to use in conjunction with my data"

Then what?

The problem and where we are at now goes all the way back to Rail Simulator. The acceptance of this platform by those with the brain power to fine tune the physics. If you really want to get into it, you can also add in the rumor (airing on the side of caution) that RSC either didn't know it themselves and/or was unwilling to divulge detailed information about the physics data in the sim to help developers and potential developers to... Develop a proper physics base for current stock or potential stock they would have built. So back to MSTS they went where they knew the physics inside and out.

Fast Forward to Railworks 1 and 2. The UK developers by this point really started getting a grasp of the physics engine. The unsung hero being that of Mr. David Brindley. They always had a head start (hence why UK stock overall isn't so fragmented) but then Oovee sent out the Class 57. In my opinion this was the turning point in acceptance. While David Brindleys models were the communities little known secret the Class 57 was really one of the first commercial products to shine an entire new light on what can be done with the physics engine.

Over here in the states, we're still having e-pride contests between message boards and "my sim is better than yours" 30 page threads. Slowly RSC started releasing moderately better product. Dash 9 being one, SW 1500, SD70 etc. Then finally, finally some traction and growth in the U.S. community started happening. The physics from RSC weren't perfect but by far leaps and bounds better then what we were getting.

So, here we are in 2011 and the acceptance of this sim among U.S. members is at its highest and still growing. The problem is there is roughly 5 years worth of rollingstock and repaints that suffered physics wise from the painfully slow acceptance of the sim here. It is entirely too much to simply "fix" and I honestly don't think it ever will be. The only place to go is forward and not look back on the mistakes made in the past. There is just too much of it to cope with.

[Kali]

Heh yes, I hadn't even got into the logistics of it. There are ways around that - not least just plain ignoring sections of existing blueprints in favour of new ones - but well, so much room for unwanted behaviour. At some point though you just have to bite the bullet, upgrade the core and attempt to minimise the impact, or there will never be any progress.

It's possible - Openrails is doing it for MSTS. Once you've got baseline physics for vehicles then they're all pretty similar, it's not like trying to make a submarine from an aircraft. Once someone wrote a simple porting tool where you could just pull down a template and fill a couple of boxes in for unique data like weight and so on, you could port wagons over very fast. Aside from doing cab interfaces and so on, engines are all very similar too, and if you got your physics model right you don't have to do much tweaking to get everything realistic, because the real world has already done it for you. That last reason is why I get so annoyed when people make random figures up to get one area of behaviour - you will get random results out again in others and screw other people's work up. There are reasons for standards.

However of course - most people would have to wait for stock to become re-available, and people who do have the ability to port will get monstrously fed up with porting everything in sight. I would suggest not releasing something like that until there are easy to use tools, but we still don't have any dev docs for RW2, let alone RW3. I see stuff in the post RW3 content that didn't used to do anything at all, which now might or might not be necessary... and I see one or two things I'm not sure of the function of. And I'm sick of having to work it all out.

[styckx]

It has to start with RSC. Unfortunately the mistakes I mentioned continue to this day. Even a child could tell you the Dash 9 and SD40 in Donner Pass are tuned from two completely different ends of the spectrum. Sure they are different engines but overall it is like Ying and Yang sitting on rails.

[Kali]

Someone is sitting there with some bunch of notes about setting stock up which hasn't changed for years... and just doing the same thing over and over. I'm almost starting to be able to tell who's set something up by reading the blueprint. Nobody is going to get it right anywhere without gutting the core physics though - I can get you proper power response for the little that is generally worth ( given most drivers seem to either use full power or coast over here, I assume that's somewhat the case everywhere ) but ... can't do a thing about making a train stop properly. And well... there isn't anything else to do! it's a train, starting and stopping is all it does.

And that's why it's actually quite important to have detailed behaviour; many people might say they don't really care if it's that accurate, but: if every new DLC drives exactly like the last one then eventually people are going to wonder what the point of buying a new one is; the experience of the whole thing is what matters, not just a pretty face and some makeup. Driving trains isn't the most involving thing to do anyway - you're doing no favours if you ditch any resource management and necessary forethought. And the other aspect - operations - is held down by the dispatcher and AI issues at the moment.

One can hope that someone might recognise some of this and not be going for the short term milking of the DLC market ( I'm not sure how long that's sustainable either, it's already frightening to a new customer ) but where's the money coming from for the coding team to write a new core? and... can you imagine trying to recruit an experienced game architecture designer to design a physics engine for a train simulator? how many people who can do all of that are there in the world!

Anyway, if anyone can produce a TE graph for a Geep ( preferably one we have a model of ) I'll be quite grateful, and attempt to do something fun with it. Otherwise I'll probably just use one from a UK 37 and you'll have to put up with it :P

[styckx]

I get exactly what you mean with the old go to performance figures. Great example is the new Class 47. It acts exactly the same as the old, which is nearly the same as the Class 37. Nice slow build to speed out of the gates then out of no where it is like someone pressed a Turbo boost button and all of a sudden you're accellerating like half your train was just uncoupled. It is always funny watching the amps meter jump up randomly like you're shifting gears.

[Kali]

Yeah, that's because of something broken from the start, the field diverts. What it's meant to do is knock the ammeter down until the divert happens, knock a couple of % off the output TE, and carry on - if you look at the output curve from a real engine it's just a normal curve with a couple of tiny stairs. Whoever wrote it managed to reduce output TE instead of just the ammeter, so it makes the TE curve look like a shark's taken chunks out of it; that's why you get stuck going up hills at 22mph in the 37. There's a lot of bad code in the physics due to misuderstandings and whoever writing it not actually knowing anything about school physics, but I think that is the only part that is absolutely non-functional ( the diesel hydraulic/mechanical code comes close ). Only thing you can do is open the simulation blueprint and just remove all the diverts, can't fix it.

Actually I have a pic from something:


Speed across the bottom, amount of tractive effort upwards. Just delete 'em :p

[Kali]

Was messing with blueprints today and created a new coupler blueprint to check something after I went messing with suspension damping; it seems according to the BPE the spring units we've been messing with are in Newtons/Meter. Not kN!. I'm boggling at the idea that 750N/m was thought enough...

Then again the BPE has been known to be wrong, like telling you to use HP instead of kW.

[johnmckenzie]

I chuckled when I saw the thread title was "US fright train physics". Thinking about it, the word fright is absolutely correct when you see flying goods wagons!

[Chacal]

Maybe it is a part of the zombie pack that came out on Halloween?

[kevhead]

I'm sure a payware will fix it. send me money send me green I'll fix your broken trainsimulator so you won't scream.