And what happens when gearbox IS in neutral?

In neutral, lay shaft and drive shaft are disconnected and rotate independently. There is no torque transfer.

And how do I calculate each of their rotations seperately?

And how do I calculate each of their rotations seperately?

Depends on how you model your drive train. Usually, on one side gearbox is connected to clutch so lay shaft velocity you figureout from there and on another side drive shaft is connected to differential and wheels, which influences velocity of driveshaft.

OK! So far all my attempts to implement this failed.

By the way - the ground reaction torque, is it longitudal friction force multiplied by radius, longitudal friction force dividen by radius, longitudal friction force multiplided by inertia or longitudal friction force dividen by inertia of wheel?

Multiple friction force by the radius of the wheel to get torque. When you divide torque by moment of inertia you get angular acceleration.

Yeah... That's what I am doing currently. The fact is that it's failing.

My current system looks like this:

[attachment=31799:Problem.png]

• Is it correct system?
• What may cause it not to work?
• What happens at differential, can wheel av differ from gearbox av?

i'd start with the simplest system and build up from there:

engine rpm = throttle position (0-100%) * constant

engine torque = engine rpm * constant (not a realistic torque curve, but its a start).

assume a 1:1 final drive ratio with no parasitic drivetrain drag. and clutch fully engaged with no slippage or torque converter above stall speed.

axle torque = engine torque.

axle torque * tire radius = acceleration force.

acceleration = force / vehicle mass.

that will get you moving.

then you can add in things like gear ratios and differentials (which are just torque multipliers), clutches, wheel slippage, more accurate torque curves for the engine, parasitic driveline drag, etc, etc.

also keep in mind how in-depth one needs to go when modeling things. i mean you could model the parasitic drag based on oil viscosity if you wanted, but there would be little point.

already much of what you're doing seems like overkill.

having a decent physics book and engineering mechanics book at hand would definitely help. personally, i'd recommend Tippler and Hibbler, respectively (IE the "gang of two" when it comes to physics).

its seem you're trying to do this with little knowledge of either physics or cars. that's a tough prospect.

i'd start with the simplest system and build up from there:

engine rpm = throttle position (0-100%) * constant

engine torque = engine rpm * constant (not a realistic torque curve, but its a start).

assume a 1:1 final drive ratio with no parasitic drivetrain drag. and clutch fully engaged with no slippage or torque converter above stall speed.

axle torque = engine torque.

axle torque * tire radius = acceleration force.

acceleration = force / vehicle mass.

that will get you moving.

then you can add in things like gear ratios and differentials (which are just torque multipliers), clutches, wheel slippage, more accurate torque curves for the engine, parasitic driveline drag, etc, etc.

also keep in mind how in-depth one needs to go when modeling things. i mean you could model the parasitic drag based on oil viscosity if you wanted, but there would be little point.

already much of what you're doing seems like overkill.

having a decent physics book and engineering mechanics book at hand would definitely help. personally, i'd recommend Tippler and Hibbler, respectively (IE the "gang of two" when it comes to physics).

its seem you're trying to do this with little knowledge of either physics or cars. that's a tough prospect.

First step is way too wrong. My current system(which I called "Arcady") uses way better method(uses wheel AV).

Rest of it is exactly what I do in my "arcady" vehicle physics system version. I've done it all already, all worked(but the realism ISN'T there at all.

Now I am trying to get realistic version which assumes inertias aswell as fact that 1500 Nm torque doesn't mean 5x more longitudal force than 300 Nm torque aswell as fact that 500 kg car won't accelerate 2x faster than 1000 kg car. It's important do do ALL the motion through friction. But to get it drive through friction, I need to get the wheels turn correctly. And for that I need to transfer engine torque and angular velocity correctly. So far it's been almost working, but not yet at full rate.

What I am gone try to do now is what first car manufacturers did: add all in order of priority.

First I will add engine, driveshaft and wheels. Set up all the equations until it works as needed. Propably something like this:

torque = interp(xcurve, ycurve, rpm) * throttle # Throttle is being controller by limitters, idle controls etc. I already know how to make them;)
torque -= frictional_torque
engine.av += torque / (engine_inertia + wheels_inertia + driveshaft_inertia) * dt
for w in wheels:
    w.angular_velocity = engine.av
    calculate_friction_forces_using_pacejka()
    w.angular_velocity -= longitudal_friction_force * w.radius / (engine_inertia + wheels_inertia + driveshaft_inertia) * dt
engine.av = w.angular_velocity

I'll see if something like this works.

After that I will add in clutch.

After that I will start adding in transmission.

I am still not sure how the ignition(engine starting) works and how to best simulate it.

Yeah... That's what I am doing currently. The fact is that it's failing.

My current system looks like this:

Problem.png

• Is it correct system?
• What may cause it not to work?
• What happens at differential, can wheel av differ from gearbox av?

Yes, wheels angular velocity will be different from gearbox angular velocity. First of all because of the differential gear in the differential, which is installed between gear box and wheel. Second because having different angular velocity on left and right wheel when cornering is exactly the point behind the differential. In regards to differential both torque and angular velocity before and after differential have a very simple relationship. TorqueGearBox * GearRatio = TorqueOnWheels and GearBoxAngularVelocity = LeftWheelAngularVelocity+RightWheelAngularVelocity.

I'm not sure what "av" in your diagram represents. Is it a single variable or you integrate it separately for engine, clutch plate, drive shaft of gearbox and wheels? Because with a single variable you just can't track engine for example if clutch is not locked.

The way I do this part is by integrating velocity on each element that can receive forces from outside of drive train or if it can "disconnect" drive train. For example engine brings torque from combustion and wheels bring torque from friction. Clutch and gearbox can disconnect drive train. For this reason, each of the elements will integrate angular velocity and pass it further. If clutch or gearbox are locked they will just pass it further without integration.

"Ignition" is rather simple - when you enable starter, it will apply torque to the engine crankshaft, when rpm is higher than some stall rpm of the engine (where engine produces less torque than necessary to keep it spinning) starter is disabled and engine will throttle automatically to idle rpm. Or just make starter as a separate button, as long as its pressed you add torque to crankshaft.