1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

'Developer Section' Response Thread

Discussion in 'General Discussion' started by iheartmods, May 3, 2016.

  1. enjoyinorc6742

    enjoyinorc6742
    Expand Collapse

    Joined:
    Dec 24, 2015
    Messages:
    1,814
    guess they made a new Micro Blogs thread
    some very interesting stuff in there already
     
  2. BombBoy4

    BombBoy4
    Expand Collapse

    Joined:
    May 16, 2015
    Messages:
    1,686
  3. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    But why those numbers are so different which should be same? There is also one named just Power in engine debug, quess that is flywheel power, looks like that at least? There are always questions :D

    Much better than my attempts as guessing and arrows, which I did offer to him, but obviously his update of graphs is a BIT more clear than this :D
     
  4. Diamondback

    Diamondback
    Expand Collapse
    Moderator
    Staff Member

    Joined:
    Apr 8, 2014
    Messages:
    881
    Because one is an estimate (torque curve app) and the others display live readings.
     
  5. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    Great, now we know a lot more again, thx! :D
     
  6. garyjpaterson

    garyjpaterson
    Expand Collapse

    Joined:
    Aug 4, 2013
    Messages:
    273
    Power at the wheels seems surprisingly close to power at the crank, only around 1.5% loss. I'd imagine that loss would have been somewhere between 10 and 20%?
     
  7. skodakenner

    skodakenner
    Expand Collapse

    Joined:
    May 24, 2015
    Messages:
    1,228
    I was wondering that as well espacially when i was making a v8 for the etk it had 550hp on the flywheel but ingame it was only around 430
     
  8. Diamondback

    Diamondback
    Expand Collapse
    Moderator
    Staff Member

    Joined:
    Apr 8, 2014
    Messages:
    881
    Yep, that's the reason I implemented a proper wheel power measurement, I always assumed that our powertrain losses are too small (most cars are actually too fast for their advertised power ratings, compared with similar real world cars).
    Now we can at least try to optimize it a bit, even though actual data on powertrain losses is very very rare.
     
  9. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    Have you seen this fairly recent paper?
    http://spiff.rit.edu/classes/phys216/workshops/w10c/car_engine/efficiency.pdf

    You could have a recreational day out of office testing everyone's cars at dyno for a 'data collection purposes', might not be very expensive either :D

    This one had interesting data points, they measured even mechanical fan losses, I was surprised that they found up to 40% losses at largest and without AWD:
    http://www.hotrod.com/articles/ccrp-0311-drivetrain-power-loss/
     
    #1249 fufsgfen, Feb 14, 2018
    Last edited: Feb 14, 2018
  10. Brother_Dave

    Brother_Dave
    Expand Collapse

    Joined:
    Aug 16, 2012
    Messages:
    342
    I can imagine its tough to get real numbers on losses. In my world of tuning a general rule is overall 10-15% loss on 2wd drivetrain and 20-25% for 4wd. Might be less for modern super- and hypercars.
     
  11. on3cherryshake

    on3cherryshake
    Expand Collapse

    Joined:
    May 12, 2015
    Messages:
    359
    Okay so I have a question for the devs:

    As a person who is constantly messing around with engines and transmissions and releasing them sometimes, does the TorqueCurve app show flywheel power? Like the engine measured at the dyno? Or at the wheels? I'm hoping it's wheel, because if it isn't, I'm seriously confused, haha! It seems every time I make an engine and use the TorqueCurve app as the flywheel torque, everything seems overpowered. If I use the TorqueCurve app as wheel torque, and I put a torque curve in the .jbeam directly from the engines I've made, and don't change it, it seems way more realistic.
     
  12. Brother_Dave

    Brother_Dave
    Expand Collapse

    Joined:
    Aug 16, 2012
    Messages:
    342
    I just realised it shouldnt be that hard to get numbers. Theres plenty of engine dynos around for say chevy 350 engine, also should be plenty of dyno runs around for cars with the same engine. Should give a hint of how much loss there is.
     
  13. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    Few posts Diamondback writes that torque curve app is estimation of power, where as others are measured.
    https://www.beamng.com/threads/developer-section-response-thread.24826/page-63#post-799739


    You don't put wheel torque to jbeam, you put torque values that don't have engine's internal losses (from bearings etc) to jbeam, then engine's friction is removed and you get flywheel power.

    Tricky part is when you have only wheel power from dyno available. In that case I believe best way is to increase values by how much you know powertrain losses to be. Then use engine spreadsheet or your own tool to get engine power without engine's internal losses and that is what you input into jbeam.

    Drivetrain losses are then friction parameters in differential and gearbox, you adjust those to get apps to show wheel power and torque of what you need them to be. Do notice that gear ratios have their effect on that, you need to know ratios and perhaps even wheel size of your source data and have to match those in BeamNG.

    This I think is quite accurate explanation, but torque curve app is estimation of flywheel power:
    https://www.beamng.com/threads/developer-section-response-thread.24826/page-63#post-799729

    You can use this tool to get real world flywheel torque converted to values that go to jbeam:
    https://www.beamng.com/threads/torque-vs-horsepower.52365/page-2#post-799279
    Make sure dynamic friction and friction is same as in your jbeam. It has protection but no password so you can tamper with rpm values too if you want when you unprotect the sheet, but don't change constants.

    Or you can use this original version by 440cid:
    https://www.beamng.com/threads/flywheel-torque-power-spreadsheet-calculator.45889/

    Ahem, https://www.beamng.com/threads/developer-section-response-thread.24826/page-63#post-799895
     
  14. Brother_Dave

    Brother_Dave
    Expand Collapse

    Joined:
    Aug 16, 2012
    Messages:
    342
    I knew i shouldve opened that pdf. Quite extensive my good man.
     
  15. on3cherryshake

    on3cherryshake
    Expand Collapse

    Joined:
    May 12, 2015
    Messages:
    359
    Well here's my logic: a 4.6L Modular in the Mercury Grand Marquis makes about 210 hp and 270 lbs-ft. If you give the Grand Marshal an engine (which I made) and put in the .jbeam the torque curve of said Modular, it goes into the game and loses horsepower. I'm going off of the basis that it's the drive-train losses, because if I add that power back everything seems ridiculous.
     
  16. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    Reason why it loses power is because of engine has these lines:
    "friction":40
    "dynamicFriction":0.2

    If you open jbeam that has transmission it will have line friction that is transmission losses, differential has also line friction, which is losses at differential.

    So Friction = losses, but you need to know what those losses are, for engine it is piston rings, bearings, cams, belts etc, engine stuff and only engine stuff. Reason for to keep these separate is that when you put some other transmission or diff in, then you get realistic figures again, autobox loses more energy than manual etc. You could put all friction to engine and none to powertrain, but that would be silly.

    That is why 440cid made spreadsheet that converts real world torque data to what is needed to input into jbeam so that you get your desired flywheel power easily.

    Also look at this picture Diamondback posted to devblog, it clearly shows how flywheel power and torque are what torque curve app estimates 467Nm vs 469Nm:


    Only wheel power and torque is shown by engine debug app, coming update has those labeled much more clear manner.

    Diamondback will be angry at me as I messed up with his picture, but maybe this makes it bit more clear:
    upload_2018-2-14_18-22-15.png
     
  17. on3cherryshake

    on3cherryshake
    Expand Collapse

    Joined:
    May 12, 2015
    Messages:
    359
    Okay, that seems to make sense. It's still a lot and I'm going to keep doing what I'm doing though.

    The engine in Automation makes 235 hp and 270 lbs-ft, I think. It makes 215 and 255 in the app. What you're saying is that the torque parameters I put into the .jbeam is what the engine is rated at, and the app is an estimate after the internal frictions and such are accounted for? That's okay with me :D
     
  18. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    Actually no.

    Values you put in jbeam is something you can't find anywhere.

    That is why easiest way to get flywheel figures you can find, converted to jbeam values is to use spreadsheet linked.

    In jbeam you need to input flywheel power that has been increased by amount of engine's friction, such data does not exists, but can be easily calculated (just use 440cid's spreadsheet it makes life so much easier as you don't need to tune anything, just input values and it tells you what to write to jbeam).

    You will put those automation values to spreadsheet and that will tell you what you need to put into jbeam to get ingame flywheel torque correct.

    Then if you know wheel power, you adjust transmission and differential friction until you get correct amount of wheel power shown by engine debug app, but even official vehicles don't have this tuned in perfectly yet.

    Of course you can use whatever method you like, but method I explain will get you correct power and very easily, no need to tune torque curve at all, as long as it is NA engine.

    I get correct amount of power by simply writing torque from my data to spreadsheet, copying values from export page I made to jbeam, saving the file and loading to beamNG, done, no hazzle, no wasted time tuning over and over again.

    Then if I have wheel power known (I can of course calculate that easily) All I need to do is to tune friction of transmission and differential, that is adjusting two values and I have that part of engine done in a flash.

    Don't really see point of doing it more difficult or inaccurate way.


    With turbo engines two pages is not enough to explain all that is needed to do, those are a huge amount of work to get right, so obviously above applies only to NA engines.
     
  19. on3cherryshake

    on3cherryshake
    Expand Collapse

    Joined:
    May 12, 2015
    Messages:
    359
    So what do I do then? I take the automation numbers, and use 440cid's spreadsheet to get what I actually need to put into the jbeam, and then adjust transmission and diff friction to create accurate power loss?

    Could you link said spreadsheet please?
     
  20. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    2,095
    Exactly like that and I promise it will be amazing how easy it is with that workflow :D

    Links in here, there is also my version of the spreadsheet that is my attempt of making less confusing to beginners, use which you like, result should be the same:
    https://www.beamng.com/threads/developer-section-response-thread.24826/page-63#post-799957