Fixing engine friction torque

[NaN]

LE torque data
blue: engine torque curve
black: current friction model
red: Heywood (1988) TFMEP model


As you see current friction model is massively overestimating friction torque (I remember running into this issue when working on my sim code). Current code opens the clutch when braking to work around this.

The tfmep model as measured for several four-stoke engines over a wide range of engine speeds at wide-open throttle, looks like this:
tfmep = 97000 + 15 * rpm + 0.005 * rpm^2

For engines with throttles, one would need to add another ~50000 Pa (0.5bar) to the first coefficient to model a closed throttle. I am not sure if it really matters though. The difference would only be noticeable at low rpm.

To implement the tfmep model we need engine displacement. I'll be going through our cars and adding it (cars without engine friction would have displacement zero)

[engine]
displacement = 1.8E-3 # 1.8l or 1800cc

Help is welcome, of course.

[NaN]

Cars data is updated. I'll do some more testing before pushing the code.

[NaN]

F1-02 my frenemy

black: torque curve
blue: tfmep model
red: tfmep model with "slightly" adjusted parameters


I'll add a way to override friction coefficients to deal with such special cases, will reuse currently unused torque-friction parameter for this I think.

[Timo 6]

I'm guessing new data is needed before the cars will load with the lastest git? Time to revisit putting the data on github? - Sourceforge's SVN hosting is still down

[NaN]

I'm guessing new data is needed before the cars will load with the lastest git? Time to revisit putting the data on github? - Sourceforge's SVN hosting is still down

Yeah, svn has been down for more than a week now.

One concern has been about the amount of data if I remember correctly.

StuntRally ( https://github.com/stuntrally/stuntrally ) has got its data on github. But it will probably be much smaller due to better mesh/texture sharing, have to checkout how large it is.

[NaN]

About 1GB seems to be the soft limit for github. https://help.github.com/articles/what-is-my-disk-quota/

It might work if we split up data, have a core data repo, and extra tracks, cars repos maybe.

[CrystalH]

Cool. Since a while I found this engine friction to be incredibly strong.
I suspect now all cars are somewhat faster with the new model right?

Anyways on the offtopic part of this topic:
Stunt Rally has data with sources in 1st repo and it's like 500MB or more and
tracks in 2nd which is about 1,5GB - total with change history since start.
Quite long and it's binary so each file is just saved again.
If you get tracks repo without history it is about 500MB.
Those maybe be a bit old values, could be bit more now.
Ofc this gets compressed on repo, not a lot though (heightmap with 32bit floats at 4MB is then 3MB with 7zip)

But you have lots of png which won't get compressed at all right.
Anyways I'm for trying it, svn is ancient now. If not github then maybe other places like gitlab, bitbucket etc,
Here is a list on Wiki: https://en.wikipedia.org/wiki/Comparison...facilities