Hi all,
I'm adding engine inertia to my car sim.
First step was to detach the rpm of the engine to the wheel axle
rotations; so I got spinning wheels with 0 rpm. :)
Next step was to add a spinning mass with inertia that is now the
drivetrain. I've added it in a few parts:
- engine rotationSpeed and inertia
- transmission inertia
- driveshaft inertia
The wheel inertia was already there.
Ofcourse, things are not working out! ;-)
First, are the final drive gears always in the differential? I
previously thought they were located in the gearbox itself.
2nd, I've read about open, locked and limited-slip differentials in
RCVD. It seems I've been using an open model before, since I let both
wheels rotate independently. But I need two variables when deciding
engine rpm change.
One is inertia; this is the effective inertia, which we talked about
not too long ago. Something like:
Iengine*Ntf^2+Itransmission*Ntf^2+Idriveshaft+Iwheels.
Although I have an analog clutch between the engine and the
transmission, I think I get the picture (when the clutch is depressed,
I multiply the clutchFactor (0..1) with either the first term
(Iengine*Ntf^2) or the other terms, depending on which side you're
looking (engine acceleration, or wheel acceleration).
Second is torque (together, I can do
engine_acc=torque/effective_inertia). The torque comes from the engine
torque (from the rpm->torque graph), some friction, and also feedback
from the wheels (that try to oppose the engine's rpm). However, I'm
not clear about the workings of the differential, esp. in the case of
limited slip.
If you have a locked axle, I'd say the feedback torque generated by
the wheel (rolling resistance torque + road reaction torque (FxR) gets
fed back into the total torque for the engine's acceleration.
However, since I'm used to the open differential, it seems, as the
wheels may rotate freely, that the wheels will never generate feedback
torque.
Let me see, in an open differential, the torque is split 50/50
(normally) between both wheels, but the axle rotation speed doesn't
matter to the differential. In a locked differential, the rpm of the
wheels is kept the same, so in that case more torque is given to the
slower wheel (probably proportional to the delta_rpm).
An idea is to have a percentage of 'lock'. Then I interpolate between
open and locked situation, giving a limited-slip type simulation.
Well, that's my view at this time.
But how would I get the torque back from the wheels to the engine?
Obviously, if a wheel is held firmly so that it won't rotate, the
engine's rpm will be (near) 0. But that means the torque must be
transported back through the differential, through the transmission to
the engine itself.
I know the transmission will have some sort of efficiency, since
automatics are 'bad' in engine braking. That's ok. I could say '50% of
the wheel torque makes it back into the engine'.
But as for the differential I don't know. It seems, since the wheels
may spin freely in the open case, that no torque will be passed back.
This doesn't seem right somehow.
Does anybody have a view on this?
Thanks for any help to get my car driving again ;-)
(currently I can get the engine to rotate steadily at 4000rpm for
example without the car having any speed!)
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/
Hi Ruud,
the differential is definitely maybe the hardest part to comprehend. :)
You correctly identified the locked axle and the fully open differential
as the two extreme cases. The main difference between the two, though,
is that any (semi-)open differential introduces an additional degree of
freedom in the movement with respect to the locked axle.
The RPM of the engine and the output RPM on each of the wheels are
always in a direct relationship with a locked axle, determined by the
gearing with a locked axle.
In an open differential, it is only the average of the RPMs on the
driven wheels that is proportional to the engine RPM in the same
fashion, but the difference in RPM on both wheels can be arbitrary.
In my sim, I therefore do not deal directly with the RPMs of the engine,
the left and the right wheel independently, but put the focus of my
calculations on the differential. Therefore I introduce the 'symmetric'
RPM (which matches the engine RPM as seen through the gearing and also
equals the average of the wheel RPMs), and an 'antisymmetric' RPM (half
the difference between the wheel RPMs). For both of these it is possible
to calculate the 'symmetric' torques (the properly scaled torque on the
engine plus the torques on both wheels) and inertias (in a similar
fashion), while for the 'antisymmetric' case I took half the difference
between the torques on the left and right wheel to be the
'antisymmetric' torque (with modifications for semi-locked axles, as
will be noted below), and for the 'antisymmetric' inertia the average of
the inertias of the left and right wheel assemblies. Taking that into
account, both the symmetric and antisymmetric RPM can be updated as per
any rotating assembly.
The actual RPMS on the wheels can then be had by adding or subtracting
the antisymmetric RPM to / from the symmetric one, depending on which
wheel you are updating, while, as noted before, the engine RPM matches
the symmetric one.
When the differential is not fully open, the antisymmetric torque also
acquires an additional component that is a function of the antisymmetric
RPM (directly proportional to it in a simplified viscous differential
which my sim currently uses) and the input engine torque.
Well, that's how I do it. Perhaps someone else might have another
approach to share.
-Gregor
*saved*
Ash - watching and learning intently :)
"Gregor Veble" <[email hidden]> wrote in message
news:[email hidden]...
Quoted text (47 lines)
| Hi Ruud, | | the differential is definitely maybe the hardest part to comprehend. :) | | You correctly identified the locked axle and the fully open differential | as the two extreme cases. The main difference between the two, though, | is that any (semi-)open differential introduces an additional degree of | freedom in the movement with respect to the locked axle. | | The RPM of the engine and the output RPM on each of the wheels are | always in a direct relationship with a locked axle, determined by the | gearing with a locked axle. | | In an open differential, it is only the average of the RPMs on the | driven wheels that is proportional to the engine RPM in the same | fashion, but the difference in RPM on both wheels can be arbitrary. | | In my sim, I therefore do not deal directly with the RPMs of the engine, | the left and the right wheel independently, but put the focus of my | calculations on the differential. Therefore I introduce the 'symmetric' | RPM (which matches the engine RPM as seen through the gearing and also | equals the average of the wheel RPMs), and an 'antisymmetric' RPM (half | the difference between the wheel RPMs). For both of these it is possible | to calculate the 'symmetric' torques (the properly scaled torque on the | engine plus the torques on both wheels) and inertias (in a similar | fashion), while for the 'antisymmetric' case I took half the difference | between the torques on the left and right wheel to be the | 'antisymmetric' torque (with modifications for semi-locked axles, as | will be noted below), and for the 'antisymmetric' inertia the average of | the inertias of the left and right wheel assemblies. Taking that into | account, both the symmetric and antisymmetric RPM can be updated as per | any rotating assembly. | | The actual RPMS on the wheels can then be had by adding or subtracting | the antisymmetric RPM to / from the symmetric one, depending on which | wheel you are updating, while, as noted before, the engine RPM matches | the symmetric one. | | When the differential is not fully open, the antisymmetric torque also | acquires an additional component that is a function of the antisymmetric | RPM (directly proportional to it in a simplified viscous differential | which my sim currently uses) and the input engine torque. | | Well, that's how I do it. Perhaps someone else might have another | approach to share. | | -Gregor
Ruud van Gaal wrote:
> In a locked differential, the rpm of the
> wheels is kept the same, so in that case more torque is given to the
> slower wheel
If the differential is locked then as you say the wheel rotational
rate is the same for both wheels. As it is all one piece when locked,
I wouldn't use the concept of a distribution of torque given to one wheel
or the other. There is torque coming in from the differential and
being absorbed by both tires differently. The sum of all those torques
over the inertia is the rotational acceleration of the whole locked
assembly. (And properly taking into account the driveshaft and
anything else locked with the drive train, of course.)
An example of a situation where the torque fed back from each wheel
(F cross R) is different is in low speed cornering. The longitudinal slip
ratios will be noticibly different on the inside and outside wheels from
the geometry of the turning situation.
> I know the transmission will have some sort of efficiency, since
> automatics are 'bad' in engine braking. That's ok. I could say '50% of
> the wheel torque makes it back into the engine'.
The tranmission efficiencies are much higher than 50%. (Closer to 1
than 50% ;). Gillespie gives some examples in "Fundamentals of
Vehicle Dynamics". I've heard estimates of the torque lost in the
total drive train for Trans-Am cars as something like 7% to 11%.
I believe Trans-Am cars are now also allowed a differential
(rear end) temperature gauge in cockpit, so presumably they can be
concerned about heating it.
- Matt
On Fri, 16 Mar 2001 14:42:14 +0100, Gregor Veble
<[email hidden]> wrote:
>the differential is definitely maybe the hardest part to comprehend. :)
And there are a lot of types with varying behaviors! :(
Thanks for the replies, Gregor and Matt. I'm beginning to understand
the mathematical principles of these things. I've done an open
differential with no locking, and will dive into it some more later;
for now I'm again swamped with work, so I'll have to postpone things
for a week or so.
Talk to you later!
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/
Does it let one wheel spin-up freely if that wheel is off the ground? And,
if it's "perfect", then there should be no driving force from the wheel
that stays on the ground...
On Tue, 20 Mar 2001, Ruud van Gaal wrote:
Quoted text (19 lines)
> On Fri, 16 Mar 2001 14:42:14 +0100, Gregor Veble > <[email hidden]> wrote: > > >the differential is definitely maybe the hardest part to comprehend. :) > > And there are a lot of types with varying behaviors! :( > Thanks for the replies, Gregor and Matt. I'm beginning to understand > the mathematical principles of these things. I've done an open > differential with no locking, and will dive into it some more later; > for now I'm again swamped with work, so I'll have to postpone things > for a week or so. > > Talk to you later! > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Free car sim : http://www.marketgraph.nl/gallery/racer/ > >On Tue, 20 Mar 2001 19:03:10 GMT, Doug Milliken
<[email hidden]> wrote:
>
>Does it let one wheel spin-up freely if that wheel is off the ground?
I believe so, well, I just apply engine torque to both wheels, using
an effective inertia of wheel rotational inertia plus the drivetrain
effective inertia's (jeez, what IS the plural form of 'inertia'
anyway?). With the squared gear ratio terms as in Gillespie's book.
So they're quite independent, and I then *derive* the engine's rpm as
the average of the 2 wheels. Which isn't right ofcourse, since I want
a clutch in, so that would mean the engine is a rigid body of its own.
But at least its moving again in a fairly realistic way.
> And,
>if it's "perfect", then there should be no driving force from the wheel
>that stays on the ground...
Ah, there's where I go wrong I guess. I apply torque to both ends
equally. So even if the right rear wheel spins violently, the left
rear wheel still gets the same torque. Should I bias the torque
instead with the ratio leftRPM/rightRPM? So if the right wheel has
twice as much rotation as the left wheel, the right wheel will also
get twice as much torque (talking open differential here)?
Even though I may do it wrong now, it seems silly simulating this, as
a Torsen differential seems much better by putting power on the ground
wheel (left in my example) instead (the inverse of the open
differential). So a very little hunch says to me that open diffs
aren't used at all without locking (when my car breaks out, it is
hardly stoppable until it flips ends).
But ofcourse, I lack practical knowledge here; do you happen to know
what diffs they normally use in F1, Sports Cars, Nascar and normal
passenger cars?
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/
On Tue, 20 Mar 2001, Ruud van Gaal wrote:
> >Does it let one wheel spin-up freely if that wheel is off the ground?
>
> I believe so, well, I just apply engine torque to both wheels, using
> an effective inertia of wheel rotational inertia plus the drivetrain
I don't think this is right, but it might be easy to program... I've seen
it done this way (iirc) before, it got named "magic-diff" and was left in
that game.
To get it right, you have to apply torque to the spider gears -- which act
like little levers balancing the torque between the wheels. The engine
pushes on the center pivot of the "lever". RCVD Fig 20.2....
Real open diffs (not perfect ones) have some friction in them, so even if
one wheel is off the ground (and engine rpm is constant), the friction will
apply some small amount of drive torque to the wheel that is still on the
ground.
Same if one wheel is on glare ice, the internal friction will still
apply some torque to the other wheel so the car *might* be able to move.
Open diffs are great at tracks where the car never lifts a wheel and rarely
spins a wheel. They have little friction, so they absorb little power,
thus they are ideal for super speedways and any track where the car is
power limited.
-- Doug
Douglas Milliken <[email hidden]>
Milliken Research Associates Inc.
Quoted text (33 lines)
> effective inertia's (jeez, what IS the plural form of 'inertia' > anyway?). With the squared gear ratio terms as in Gillespie's book. > So they're quite independent, and I then *derive* the engine's rpm as > the average of the 2 wheels. Which isn't right ofcourse, since I want > a clutch in, so that would mean the engine is a rigid body of its own. > But at least its moving again in a fairly realistic way. > > > And, > >if it's "perfect", then there should be no driving force from the wheel > >that stays on the ground... > > Ah, there's where I go wrong I guess. I apply torque to both ends > equally. So even if the right rear wheel spins violently, the left > rear wheel still gets the same torque. Should I bias the torque > instead with the ratio leftRPM/rightRPM? So if the right wheel has > twice as much rotation as the left wheel, the right wheel will also > get twice as much torque (talking open differential here)? > Even though I may do it wrong now, it seems silly simulating this, as > a Torsen differential seems much better by putting power on the ground > wheel (left in my example) instead (the inverse of the open > differential). So a very little hunch says to me that open diffs > aren't used at all without locking (when my car breaks out, it is > hardly stoppable until it flips ends). > But ofcourse, I lack practical knowledge here; do you happen to know > what diffs they normally use in F1, Sports Cars, Nascar and normal > passenger cars? > > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Free car sim : http://www.marketgraph.nl/gallery/racer/ > >Doug Milliken wrote:
>
> Does it let one wheel spin-up freely if that wheel is off the ground? And,
> if it's "perfect", then there should be no driving force from the wheel
> that stays on the ground...
>
Yes, it does :). I tested the sim with a fully open differential on an
uneven terrain first, and most of the time I couldn't move the car
around at all. I was surprised at the results first, but then thought
about it some more and indeed this is what should happen.
-Gregor


































