How does torque affect car performance?

  • Thread starter Tekku
  • 135 comments
  • 17,289 views
I read through this whole post hoping to find a clue on how PD comes up with these hp and torque curves. My understanding is that hp=torque*rpm/5252 and because of this hp and torque will always be equal at 5252 rpm. A quick check of a few graph reveals that PD has a different formula for hp and it seems to differ for each car. And on top of that, I pity the guys using auto shift because in several cases the shift point is just wrong making me doubt that algorithm too. I really wonder sometimes just what is PD doing?
I just had a quick look at one graph, and torque is measured in kgfm, whereas your equation is for torque measured in pound-feet, so you need to calculate the correct constant for the different units.
 
I read through this whole post hoping to find a clue on how PD comes up with these hp and torque curves. My understanding is that hp=torque*rpm/5252 and because of this hp and torque will always be equal at 5252 rpm. A quick check of a few graph reveals that PD has a different formula for hp and it seems to differ for each car. And on top of that, I pity the guys using auto shift because in several cases the shift point is just wrong making me doubt that algorithm too. I really wonder sometimes just what is PD doing?

It is formula by wich power is calculated, but... some of the manufacturers don't like to show their true torque and power or they lie about it... this is why this formula is not always correct.
However, when you take your car to dyno and measure the torque it produces (there is always loss in gearbox etc.) you can safely use that formula to get your cars power output...
 
I just had a quick look at one graph, and torque is measured in kgfm, whereas your equation is for torque measured in pound-feet, so you need to calculate the correct constant for the different units.
My graphs are measured in ft-lb.
 
It is formula by wich power is calculated, but... some of the manufacturers don't like to show their true torque and power or they lie about it... this is why this formula is not always correct.
However, when you take your car to dyno and measure the torque it produces (there is always loss in gearbox etc.) you can safely use that formula to get your cars power output...
Who would lie about a 68 fiat 500?
 
My graphs are measured in ft-lb.
Ah, okay, must be down to some option somewhere.

Okay, so if you have both lines displayed in the units your equation is based on, then the other issue is are you expecting the graphs to cross where the values are equal? As the graph has two y-axes, one for torque and one for power, they aren't shown on the same scale.
 
Ah, okay, must be down to some option somewhere.

Okay, so if you have both lines displayed in the units your equation is based on, then the other issue is are you expecting the graphs to cross where the values are equal? As the graph has two y-axes, one for torque and one for power, they aren't shown on the same scale.
Your right. I would have argued that because they use the same x axis and it's not like hp and tq max values couldn't use the same scale when using ft-lb. But now looking at the fiat, max tq is 22.4 max hp is 16 and it shows the hp above the tq. So the graph is kinda hopeless for direct comparison. Thx for clearing this up.
 
Your right. I would have argued that because they use the same x axis and it's not like hp and tq max values couldn't use the same scale when using ft-lb. But now looking at the fiat, max tq is 22.4 max hp is 16 and it shows the hp above the tq. So the graph is kinda hopeless for direct comparison. Thx for clearing this up.
There are too many different combinations of units to bother modeling each and every one of them, especially when you consider it would have no real value in terms of practical application.
 
I read through this whole post hoping to find a clue on how PD comes up with these hp and torque curves. My understanding is that hp=torque*rpm/5252 and because of this hp and torque will always be equal at 5252 rpm. A quick check of a few graph reveals that PD has a different formula for hp and it seems to differ for each car. And on top of that, I pity the guys using auto shift because in several cases the shift point is just wrong making me doubt that algorithm too. Further more, I've noticed that the transmission adjuster hardly ever gives you the correct speed and is sometimes out by a fair bit, so what's the calculation error here? I once tried to calculate tire circumferences using speed gearing etc., and came out with all kinds of wierd sizes. I wonder sometimes just what is PD doing? I've come to the conclusion that mathematics has no real place in GT Sport!

Most likely they model the torque and then derive the power curve from that. The reason being that torque*0 is defined, while power/0 isn’t. I also imagine that for x close to 0, power/x would start to glitch out due to precision limitations when working with floating point numbers on computers and you may obtain absurd torque figures or even division by zero errors.

As for power vs torque, hp is a unit while torque is a physical quantity. If you are measuring torque in lbft, power in hp and engine speed in rpm then the 5252 factor is correct. If you measure torque in Nm, power in W and engine speed in radians/s then the factor is 1 and you get the fundamental relation between the quantities: power = torque*angular velocity.

In the physics engine they most likely only calculate the torque output, because you don’t really have any use of the power figure in the actual simulation. Power is more for analytical purposes, like estimating how fast a car might accelerate, or what top speed it might be able to reach.
 
The hp torque curves in game tell you a lot about how to best drive the car and the cars power delivery without even driving it. They give a good idea about shift points.
I just wish they provided an exact scale on the rpm.
I have no idea why simple things like this cause so much confusion. Mention torque on the internet add hp and gears and all is lost.
 
In the physics engine they most likely only calculate the torque output, because you don’t really have any use of the power figure in the actual simulation. Power is more for analytical purposes, like estimating how fast a car might accelerate, or what top speed it might be able to reach.
I don't know what people do in the motoring industry, but in cycling, everybody uses an equation based on power, rather than torque. If people used torque instead, they would need a more complex equation, the end result of which would just be to essentially convert it to power anyway. This is the standard equation used for cycling, which is almost identical to what would be used for a car, the only alteration that would be required that I can think of off the top of my head is that the Crr term would need to be multiplied by (mass + aero downforce) rather than just mass.

MathModelofCycling.png
 
My guess is the graphs are always drawn in some standard metric units (or perhaps the simulation units), and the axes scales are just labeled differently. Hence not crossing at 5252. I remember Ducati released a power / torque curve for its XDiavel, annotated in mixed units, but the axes were manipulated so the curves crossed at 5252; I can only imagine it must have been very important, for some reason...


The game does indeed do all of its calculations beginning with torque, because it's a dynamic simulation and it's easier to directly translate it into an impulse / force for the next physics tick. If using power, you'd usually have to involve a speed somewhere.
 
I don't know what people do in the motoring industry, but in cycling, everybody uses an equation based on power, rather than torque. If people used torque instead, they would need a more complex equation, the end result of which would just be to essentially convert it to power anyway. This is the standard equation used for cycling, which is almost identical to what would be used for a car, the only alteration that would be required that I can think of off the top of my head is that the Crr term would need to be multiplied by (mass + aero downforce) rather than just mass.

View attachment 868629
Gosh. Suddenly, I feel the need to study physics all over again just to understand these formulas. But even then, maybe it's simply isn't enough. :dunce:
 
The game does indeed do all of its calculations beginning with torque, because it's a dynamic simulation and it's easier to directly translate it into an impulse / force for the next physics tick. If using power, you'd usually have to involve a speed somewhere.
Even if you work in terms of force rather than power, you still need to know speed, as some of the forces depend on speed, e.g. the aero drag force.

(As power = force x velocity, you can convert the equation I posted above to force by dividing each term by Vg, which still leaves Va^2 in the aero drag term)
 
Even if you work in terms of force rather than power, you still need to know speed, as some of the forces depend on speed, e.g. the aero drag force.

(As power = force x velocity, you can convert the equation I posted above to force by dividing each term by Vg, which still leaves Va^2 in the aero drag term)
The difference being that the aerodynamic speed does not in itself constitute a momentum contribution and so does not upset the energy conservation / stability behaviour of the time stepping simulation.

Of more concern is the k.e. term and its dependence on a final speed we haven't worked out yet... Plus it's now divided by Vg, arbitrarily.
And the bearing resistance term will be messed up, generating a force at rest; and also still contains Vg.


And that's just the drag equation, lumped to act... somewhere? Probably the C.o.G. But the rolling resistance acts at the wheels and levers around the C.o.G., causing a rotational force on the bike (not described). Likewise, the aero force acts on the C.o.P. and levers around the C.o.G. similarly, probably oppositely. As I said, it is not a true dynamic representation.

Now imagine taking that approach and deriving steering behaviour, tyre model, suspension, drivetrain dynamics etc. Hence multi-body, distributed models - each element having its own velocity, but more general (simple) governing equations and constraints.


And now I suspect we are fully off-topic :)
 
Back