I'm in the process of writing a ticket but I wanted to post my findings here for others to validate/criticize. TLDR: 1. RSN wagon weights are incorrect and too high. 2. The variable brake force on both DE and NA freight wagons is incorrect. 3. Average brake force is definitely wrong on DE wagons and probably wrong on NA wagons. 1. The wagon weights on Ruhr-Sieg Nord preserved is incorrect. For example the (loaded) coal train from Kreutzal to Letmathe has a weight that when subtracted the locomotive weight and divided by the number of wagons comes out to about 140 tonnes. This is massively incorrect! The correct loaded weights are 90 tonnes (as can be read from the printed text on the wagons). 2. In previous correspondence with DTG I was told that they simulate variable coefficient of friction between the brake shoes and wheels as a function of speed. This should mean that the brake force decreases at higher speeds for the same amount of brake cylinder pressure. However this curve they have used seems to be much too flat. If you search "brake shoe coeffiction of friction vs speed" you can find several studies on this subject but here is a link that shows the speed dependence for different types of brake shoes (from different studies): https://www.coalstonewcastle.com.au/images/physics/brakeshoe_friction.png From reading UIC documents the blue "westinghouse (1894) formula" seems to best correspond to European cast iron clasp braked wagons while the two "1982 Railway Engineering Book" are for North American freight wagons. The modern COBRA composite brake shoes has better brake force at higher speeds. You can test this out in TSW2 by braking a freight train using only air (release the E-brake handle in the BR185 for example) and time the intervals, there is very little change. For NA trains this is especially important because it changes how you have to drive the trains in descending grades. Taking a train consisting of 2 x AC4400 + 50 loaded coal wagons and drive it east to Sand Patch then begin descending. In reality train engineers will set some air (around 8-14 lb depending on grade and "Tons per operative brake") and then fine tune the speed using dynamic brakes. If the train has to stop you don't (usually) set more air but rather simply increase the dynamic brake effort. Here is where the variable friction coefficient comes into play: When the train decelerates the brake force will increase from the friction brakes resulting in higher deceleration but also this will allow the train to continue decelerating when the dynamic brakes fade. When the train can continue the train engineer will leave the brakes applied and throttle up. When nearing the previous speed again, the brake force from the friction brakes will have faded (from the variable dependence of the friction coefficient vs speed) enough so that the train begins to accelerate by itself (overcoming the brakes applied), so the engineer now has to throttle down and then apply dynamic brakes to hold the speed. This doesn't work/or works badly in TSW2! For example, when setting 13 lb of air and using about 30% dynamic brake the train is holding 25mph in the 1.5% descending grade but when I want to stop, the train will simply continue "forever" at around 0.5 mph where the dynamic brakes have faded. At 0.5 mph the brake force should be between 3 and 1.5 times as high according to the link I posted above. If 14lb is barely enough to hold speed then the train would've stopped with correct brake physics. 3. The brake force values are definitely incorrect for German freight trains and most likely incorrect for NA freight trains. 3.(a) German trains follow UIC rules which means that they will have a printed "Bremsgewicht" (brake weight) on the wagon. You can see it on the coal wagons for example. 58t for a fully loaded wagon. From the brake weight and the wagon mass you get a ratio (Bremshundertstel) or braking percentage. This braking percentage (denoted by lambda in UIC documents, lambda = brake weight/wagon mass) can then be converted into an average deceleration value from the following two equations: lambda = C/S - D 100 km/h: C = 52840, D = 10 120 km/h: C = 83634, D = 19 140 km/h: C = 119179, D = 19 160 km/h: C = 161280, D = 19 S = v^2/2a + 2v, this is the stopping distance from an emergency brake application with 4 seconds to achieve full brake force (the usual value). Some algebra gives the deceleration as a = v^2/2(C/(lambda + D) - 2v) For the freight wagons in TSW2 the 100km/h coefficients are used. This gives an average deceleration of 0.59m/s^2 for the loaded coal wagon. Note that this is the average value so the maximum value has to be inferred. The deceleration in TSW2 is much higher and incorrect in both maximum value and average value! 3 (b): For NA rolling stock there is a "Net braking ratio" (NBR) for each wagon. For freight wagons this is usually between 8.5% and 11% for when loaded. The maximum brake force can be calculated as: F_max = NBR * wagon_mass * g * µ(0) where µ(0) is the friction coefficient at zero speed and g is the gravitational acceleration (9.81m/s^2). In fact, the brake force (or deceleration) can be found for each speed as: F(v) = NBR * wagon_mass * g * µ(v) At 0 speed the acceleration for a wagon with 8.5% net braking ratio should then be a = 0.085 * 9.81 * 0.5 = 0.42m/s^2. I haven't tested this thoroughly yet in TSW2 but the brake force values seem wrong for the NA trains as well. If DTG already has implemented variable brake force then all these fixes should be relatively easy to implement since you would just have to adjust the curves and the max brake force values. I don't think I'm the only one that would like to see this better simulated.
I ask without a trace of malice; I'm not suggesting anything; sheer curiosity Inquisitiveness, hobby-speculation, or maybe a technical education? It looks quite solid! GJ
Hmm, that's a tricky question to answer hehe. I would say that I derive the most enjoyment out of trying to drive the trains in as realistic fashion as possible. Combine that with having done a significant amount of reading of the official documents used for railway standards in mainly Germany but also the UK and a little bit in the USA, and also a technical education in physics/data and lastly fact checking my theoretical/document assumptions/conclusions with train drivers, that's probably how I ended up with this thread for example. I have tried to keep the speculation to a minimum and back up my conclusions with sources. Here is the source for the UIC BrH to retardation and vice versa equations: https://www.rssb.co.uk/rgs/standards/GMRT2045 Iss 4.pdf page 41. Source for typical EU cast iron brake shoe: https://rail21.pagesperso-orange.fr/Freinage_Organes_EN.pdf http://runeberg.org/tektid/1937m/0020.html (in Swedish but you can read the graph easily) Source for North American net braking ratios: http://www.standardsteel.com/rdpapers/freightcar.pdf Explanation by railway employees in USA of how NBR is calculated: https://www.trainorders.com/discussion/read.php?10,3307865 BNSF train cars actual net braking ratios: https://www.bnsf.com/ship-with-bnsf/ways-of-shipping/equipment/ Click each train car and then open the pdf-files. Left column has NBR and right column has brake shoe type (composite = COBRA = green curve in my initial post link). Hopefully that will satiate the curious for a while. Lastly I did some brake testing of the loaded coal train on RSN. From 4 braking tests I timed the intervals of braking (10km/h steps) and made a coefficient of friction braking dependence graph and compared it to the EU cast iron brake curve and the NA COBRA brake curve: PICTURE UPLOAD DOESN'T WORK - CLICK LINK: https://ibb.co/G7SS4nh
You should condense it somehow, catalog it. I once dreamed of a website bordering on hardcore-science and fun, e.g. the TSW series, but not only. For people for whom "it works" is not enough. Good job! I guess even that's an understatement...
Yeah, I agree with that. I might make a small .pdf with graphs comparing TSW2 and what it should be and also give them the exact tables that they can just plug into SimuGraph for each vehicle and then simply send the pdf to DTG every week until they respond Right now I feel like they've tagged my account/email and forward it to recyclebin@dovetailgames.com
I’ve only ever really noticed it on the Class 101 DMU where you can tell the braking force increases as you come close to stopping, so a there is a definitely a nice curve on that train.
View attachment 32719 I agree, Stujoy, that the Class 101 feels like it really bites at low speed, I'll need to test it to make sure it's not just the same old linear function but with a higher scaling. If you look at the green curve which is cast iron brakes you can see the huge difference to what TSW2 models. The North American COBRA brakes are also quite different, the important part is the curving around the low speeds a that are common for mountain grades. You can think of the area under the curves as the braking power. TSW2 braking power is massively larger than cast iron brakes, it looks almost like disc brakes.
Update: I tested the Class 101 from the Tees Valley preserved collection DLC: As can be seen from the graph the Class 101 has more realistic braking simulation but it closely resembles what you'd expect to see for composite brake shoes. I think the Class 101 has cast iron brakes though? EDIT: Basically if the Sand Patch grade wagons/locomotives can be modified to have the same dependence as the Class 101 they will be realistic at least in that regard.
I really do hope someone from DTG is keeping an eye on this, as you really did put in a lot of effort on this matter!
If the calculations are being done for the braking on every wagon on a train individually and constantly in real time, and the devs were looking for optimisation options, how much would taking out the braking curve save in CPU usage? I ask because it is there in the Class 101 but not in other trains where the physics are more intensive in other areas, and it is a feature that not many people would even notice if it was removed during development, so maybe that is an explanation.
Yes, I tried to find areas where the gradient was flat or 0.1%. Although, as long as the gradient doesn't change you would see the same forms of the curves for different gradients because they would just be raised up or down by a constant factor. I tried to do as many tests as my patience allowed (lol) to minimize the random errors from me mistiming (reaction time etc). Actually, the last data point (0 speed) for the Class 101 was tested in a falling 0.5% gradient (I was at first only going to test 10,20 etc) by measuring the speed it took the stop from 5 mph (brakes already fully developed). Because of the linearity above I simply had to add 0.005*9.81 to the acceleration to get the correct value. This is an interesting question. But I do not think that the framerate will be noticeably affected by more accurate curves. I have implemented brake shoe physics on the Class 66 and Class 150 by Armstrong Powerhouse for TS2020 (I've not released these patches because of copyright reasons etc) and there was no change in FPS. What I presume is happening is that DTG has tabulated values for x = speed, y = brakeforce percentage of max, and then SimuGraph or whatever interpolates (Lagrange interp I think) the value for each speed at each "tick". For modern computers this is extremely quick (tens of nanoseconds probably). And remember that in later TSW/TSW2 releases physics calculations were reduced in accuracy for AI to improve performance. You only need these accurate curves for the player loco. EDIT: Actually I think it might even be just a linear interpolation (first order Taylor polynomial). Basically the operation can be taken to be instant for anything other than pocket-calculators hehe.
Just a quick thing, if you want a quick and flat testing ground just set up a scenario in RSN starting out from Hagen Hbf depending on what track you chose you are cleared to 120kph immediately and the track is flat for at least 2 or 3 km IIRC. Hope it can make your work just a bit less tedious, keep it up hopefully DTG keeps an eye on this topic
Yeah, I think that's the route I used, I just wish I could fit a Sand Patch train there haha. I did some testing of the coal train under mediocre conditions but the brakes seem almost worse than even the German ones, it looks almost constant, not even linearly increasing, just constant.
Wow. I guess it's a possible that SPG is the most basic of basic routes... it's the first one and maybe DTG just wanted to get something working to release the new game and focused on things such as opening the engine bay and starting the engine yourself rather than actually driving the train and they carried over the awful physics over to TSW2. As said, an accurate driving experience seems like it's way down their list of things to implement or consider
Wow, It may be interesting to create a more technical guide for people that are interested in this matters. There is a lot of information going on this forum but it is hard to find it back later on.
I'm planning on compiling a relatively short .pdf that can be uploaded/sent to DTG. My aim is to very explicitly describe how to change the physics to realistic behavior almost in a plug-and-play way for DTG, then they will have no excuse to ignore it . Then I will more thoroughly describe why the changes should be made and try to incorporate all my sources in case the methods are doubted. I'm currently investigating the realism of the application and release timings of the German freight trains. I have suspicions that they are somewhat incorrect (mostly in a visual way though) but I want to confirm with real drivers before I make any bug reports in case my knowledge is incomplete. It would be very bad if I convinced DTG to fix something that wasn't broken.
It would be nice if there was a test track where you could drive any train and see its characteristics and test things out. It could have several start points with long sections of level track, a selection of inclines and a selection of curves. No scenery. It could be part of the scenario designer and would be ideal for testing or getting to know your collection of trains.
We have something similar in TS2021 (Testtrak), so I'd love for it to be in TSW as well. Ultimately I'm still holding out hope for some kind of physics editor akin to the scenario designer where you could change mass, brake force, power etc. I know Matt said (on Uktrainsim a year or two ago) that you can't edit the locomotive files as is because they have already been "cooked" but it should be possible in principle to have a designer that adds features on top. So if you want to edit the brake force TSW will simply read that part of the file and prioritize it over the base physics. Sort of analogous to how if you unpack the asset or route files in TS2021 they will be read above the .ap archives. This way DTG wouldn't have to worry about people knowing the source code physics (not sure why they are so secretive about this) because any change made would be used in place of the default.
Can you post this idea in the suggestions part? I agree it would be nice to have something like this. Problem is that one such a route would not be sufficient. You need one for Germany, the UK, UK third rail, Swiss narrow gauge, US diesel, US LIRR .... Thinking it over, maybe it will not work as we may hope.
If you consider this as an actual resource for inclusion in TSW it has to go one of two ways. Extremely simple with universal power (like off the rails in scenario designer) and no signals or safety systems, or fully featured for testing or learning PZB, AWS, ATC etc. and signalling for every system and train there is. The latter could get out of hand quite easily. I’ll have a think about what features might be needed and do a proper suggestion in the next few days or so.
I don't see why they would need to make it this complicated? The test route is not supposed to be realistic, just a way to test trains. In Test trak you can run third rail, overhead line, diesel etc. EDIT: Okay so I've got some confirmation about the goods brake timings on the RSN/MSB etc. 1. The brake pipe indication will not rise or decrease slower than what it does for passenger trains in P/R-brake, so this is correct or at least reasonably correct in TSW2. 2. However, the brake cylinder has different valves for G-brake than P/R-brake so it fills up or empties at different rates than when train is in P/R-brake. This means that the brake cylinder pressure gauge in the locomotive should take around 18-30s to go from zero to full pressure when applying full service brakes and 45-60s to go from full pressure to zero when releasing the brakes. This is incorrect in TSW2 because the brake cylinder pressure is coupled to the brake pipe pressure in the game so that the brakes look like they have released in about 5-10 seconds although the brakes are still applied which can be noticed by how long it takes for the deceleration to stop. 3. The easiest way to understand the difference between G-brake and P/R-brake is that the *rate of change* for the brake cylinder is greatly reduced for G-brake. Let r_G = (dp/dt)_G be the G-brake rate of change and r_P = (dp/dt)_P be the R/P-brake rate of change, then the time to make a full service application will be t_R = 1.5bar/r_P t_G = 1.5bar/r_G (rates assumed constant) The time to release a full service application will differ as can be seen from the different times above giving different release rates. However in TSW2 it seems like the time to release/apply brakes is the same no matter whether a small or a full application is made. For example it takes 50s to release full service brakes (realistic) but it also takes 50s to release a 1 bar application (unrealistic) when it should take (1 bar)/(full service drop) x 45-60s = 1/1.5 x 45-60s so between 30s and 40s. For even smaller applications it becomes even less realistic since the time stays the same when it should be even lower. Hope this makes sense.
Not true, the brake cylinder filling/venting times do vary depending on how much of an application is made in TSW. For example on a 10 wagon trains of PGA, with the BP and BC pressures being recorded at the 1st (red), 5th (green) and 10th wagon (Blue). 50% Application (sorry I'm using PSI). Brake cylinder filling times to approx 34psi = 14.5s Brake cylinder venting times to 5.7 psi (0.4 bar) = 25s 100% Application (Full Service) Brake cylinder filling times to 95% pressure (51psi) = 22.5s Brake cylinder venting times to 5.7psi (0.4 bar) = 35s Note: times are from the brake cylinder starting to respond to an action. Also note UK goods timings are slightly different from UIC Goods brake timings. Although I will concede the ZACNs distributor and possibly some others may have a slight issue with the timings being off.
Wow those are interesting looking graphs. How did you get them, I you don't mind me asking? I retested the brakes on RSN. Test train was the 07:00 BR185.2 service form Letmathe to Kreutzal. Contrary to my previous (quite badly designed) testing results the release time does actually vary correctly with the size of the application. It took 18 seconds +- 2s to release a 0.6 bar application and it took 48 seconds +- 2s to release a full service application. So this aspect of the braking system does in fact seem to be correct. However, the gauge indications still seem incorrect. The brake cylinder gauge in the cab fills and vents like it would for an R-brake setting (passenger timings). Do you agree/disagree that this should be the case when G-brake (goods timings) is used? EDIT: From my understanding the brake pipe filling/releasing rate is affected by the length of the train but the brake cylinder rates are affected by the P-/G-brake setting (different valve diameters).
Only the UK BR diesels actually have variable brake distributor timings at present (excluding Class 47), so the DB locos are fixed to a single rate at present which is probably either P or R, don't know which.
Was this an old graph from DTG or is SimuGraph available or are you (DominusEdwardius) a developer? Also yes, the German locomotives are fixed in terms of brake setting with passenger locomotives in R (BR143) and BR185.2,Br155 in G. However the G-brake simulation seems to be a mix of R/P and G. I talked with some Swedish drivers (same equipment as German) on another forum was told that the brake pipe will go charge/discharge depending on length of train (more wagons longer times) while independent of brake setting however brake cylinder pressure will change depending on G-brake and R/P-brake. In TSW2 the brake cylinder pressure rate is that of R/P. My guess is that DTG has correctly modelled how the actual retardation is applied/released but not the gauge in the locomotive cab. In fact, in reality, it is not uncommon to have only the locomotive in G-brake. Then the locomotive brake cylinder will apply/release slower than the wagons that are in P. Not it seems like the wagons are in G but the locomotive is in R or P. EDIT: I should say, though, that the graphs make me optimistic as they show realistic physics being modelled (air signal has a propagation speed rather than being infinite) it's just the system specific settings that are slightly wrong.
