The online racing simulator
Power curve engine
(27 posts, started )
Power curve engine
Hi all
I am developing software to automatically calculate the best possible gear ratio, depending on track properties, car setup and drivers driving stile.
Well, I have already built that software and I am using it on racing go-karts, and some cars, but if I want to improve it even more, I need to do a lot more testing on track… or just use LFS and its telemetry output. I can read LFS binary files… but I could not find information of PoweCurve of cars engines (kW/RPM, HP/RPM, Nm/RPM…) directly in file.
So where can I get engines porwecurve characteristics? And how are they connected with throttle and turbine pressure at turbo engines.
Thanks for help
Quote from GrEgce :So where can I get engines porwecurve characteristics?

As far as I know, nowhere. They can only be estimated, as can be seen in Bob's VHPA.
In the past, I have extracted estimated curves from RAF outputs. More info on that can be found here:

It should be noted that the torque curves apparently changed (mostly at low revs) for patch Y, but I have not got around to re-extracting the data. I was actually going to ask Scawen if he could add a small data table to the .bin export. This would also be massively convenient as it export uses the current setup, so intake restriction could be closely investigated.
:S J if that is truth!?!? That Bobs VHPA uses only estimated power curve, than any useful program to calculate gear ratios can be just our wet dream!!!!
“I build my own dynos for engines and I know that just a small power curve modification, in setup range, can make huge change in gear ratios… so therefore we prepare engines setup separately for every different track, driver, and grip!!”
I hope that programmers of LFS will provide us engines power curve properties, and save me a lot of programming to build my dyno for LFS engines too. Because it would be nonsense to calculate back must constant values. :S
And again… I don’t want to be rude, I am using LFS because it is the best simulator and because it has very good support!!!!
I build my own engines and I know that small power curve modifications don't need any gear ratio changes at all, unless it's a fundamental change to the curve (i.e. not small) - e.g. where the power is in the rev range. Just filling a hole in a curve with a modified airbox (for example) doesn't generally make a difference.
Hi BOB you speed me up

tnx for link... i read a little further...
Depending on discussion here: and your last answer. i can gather that you still didn’t get polynomial constants for power curve :S
No, I've not looked at creating polynomials to exactly match the torque curves. I would have thought that they are polynomial in nature, rather than just a single polynomial curve. The engine modelling in LFS looks more detailed than a simple curve generator.
So you are using filtered power curves that you paste them on link above in your VHPA?
Till now I was using your curves digitalized from VHPA graph
I know that engine modeling in LFS is far more than just a curve… it should be a whole system, but still there is some logic of how it works. And it is still far simpler than real engine (no or small temperature relations, no wear, no fuel regulation, no air properties…)
Tnx Bob for your help!!
The curves in VHPA, for the moment, match the peak power and reak torque values, but otherwise are largely from my creativity. Once I've extracted the curves again, I will of course update the vehicle presets for VHPA to be as close as my curve system allows.
Dammed i am dumb!!!

i just dont get it how did you manage get CdA form CAR_info.bin file (or at all)

// Now 4 aero blocks - rear wing, front wing,
undertray, body : 20 bytes each)

128 Position : x, y, z
Lift : multiply by speed squared to get lift value
Drag : multiply by speed squared to get drag value
for XFG
if i read this at Offset 188+4 (192) i get 0,4. And as it writes above... i should multiply 0,4 by speed squared...
But if i look at VHPA... it writes 0,66 for CdA

what am i dooing wrong??

dragforce=CdA*v^2 (1)
dragforce=Cd*A*v^2 (2)
dragforce=C*A*r/2*v^2 (3)

As i understand that i should use only the equation (1)
The value LFS outputs is 0.5 * Cd * Af * p

Given that 0.5 is a constant and p (rho) is known (1.205 in LFS and VHPA), I can get CdA back out, which is what VHPA takes as an input (as this figure is what you'll find when you research cars, not 0.5CdAp).
so every thing is as it says

tnx mate, once again

i hope that, i ll be able to introduce alpha version soon!!!
So I presume you are using a fixed first and top gear, and then calculating the intermediate gear ratios that give the largest combined area under the used curve? Thus, the fastest acceleration given your specified first and last gears. If so, sounds like it would be handy. Assuming you get this to work, would you be willing to share what you achieved so I could integrate this into VHPA? Would save me having to recreate it from scratch.

Even more complicated!
The program reads RAF of your (best) lap and determines locations and distances of streight accelerations and it builds database with straight lengths, its starting speed and its ground level properties. that means that program is reading driving stile properties of the driver and properties of the track.
Then this data base is used to find the fastest gearbox setup with "millions" of different tests.

This is my program in short. I am pleased that you invited me to particiapte in a bit of VHPA project, and i will give the program to LFS public... the only problem is, that i am using LabVIEW to write a program, as i am an Mechanical Engineer not a Programmer by profession so as for now there is no C, c++ or other konwn translater from LabVIEW
One more question

how come that perpendicular wheel loads summed together are not the same sa car weight on the flat area?

this gives no sense... this means that system is not stable and the car is falling with acceleration (on flat area!!)?

and why do left wheels have different sign than right? (in RAF)
Quote from GrEgce :how come that perpendicular wheel loads summed together are not the same sa car weight on the flat area?

did you consider the driver and fuel load?
I read total car weight from RAF ( kg) as i know this shlould be right for FXG at WR lap.

and if i sum all wheel loads i get 860kg. So there is 150kg missing...

there is a chance that this perpendicular loads are suspenssion loads, then i have to add 20kg for every wheel and unknown value for suspansion weight. But at the end there are still just 940kg...

well this is not so extremly important as i use this to calculate rolling resistance. but it is a little strange.
Then add 70kg driver weight and 2kg fuel weight and you have it, no?
the driver and fuel are unsprung mass?
#20 - w126
Quote from GrEgce :there is a chance that this perpendicular loads are suspenssion loads, then i have to add 20kg for every wheel and unknown value for suspansion weight. But at the end there are still just 940kg...

They are suspension loads. Could you attach this RAF file here?
Unsprung masses are known. They are in the .bin flies and can also be seen in VHPA.
OK here it is... one WR file

data in here are even worse that i described above.

Please see the RAF at the 43rd s where the altitude is constant.

:S sorry i missed unspring mass in .bin file
Attached files
WR.raf - 1.7 MB - 137 views
#23 - w126
I have just checked the RAF file. At 43.00 s the sum of vertical loads is 9544.29 N. At 0.00 s it is 9571.98 N. These are suspension loads taken directly from RAF file. The whole car mass is 1010.90 kg and the unsprung mass is around 50 kg, so the loads seem to be correct.

The image below shows the sum of tyre normal loads calculated with my RAFTyreExtract (I had to fix it because it turned out to be incompatible with FWD car data). These tyre normal loads are approximately the sum of suspension vertical loads, unsprung mass weight and the forces equivalent to wheel vertical accelarations. Again, it looks reasonable, as it oscillates around 10 kN.
Attached images
i wonder where that oscillation comes from... looking at the forces view in game and considering the smoothness of the tracks i always get this eerie feel like the car is constantly falling through the road and bouncing back up again
If that graph is vertical tire load, I'd guess the oscillations are primarily a tire spring effect. If you lower the air pressure a bunch the frequency should get reduced if that is the case.

Power curve engine
(27 posts, started )