45 replies to this topic

[Lee200]

Hi John,

No, there is no central place where all this information is located.  The old RSC forums had some in their forums, but all those threads are now lost.  And much of our current knowledge was never published as, frankly, most people really don't care about the dirty details of how GPL works.  

The Hines offered some good information when GPL first started as they were beta testers for the program and were in contact with the developers.  But ten years of investigation has revealed loads of good stuff about how GPL actually works...information the Hines probably didn't know.

You are entirely correct that real world cars may have the CofG closer to 50/50; especially when the driver and fuel weights are added.  GPL does add in a nominal driver weight, but AFAIK, that weight is added at the chassis CoG so it has no net effect on overall CoG.  The fuel; however, is added at a point .3 meters forward and slightly higher than the chassis CofG so it indeed does shift the overall CofG forward a bit.  I doubt though that it ever gets to 50/50.

Your tests using tire temps probably can't tell you much about the CofG location at rest as you have to be moving to heat up the tires and then all sorts of variables get thrown in that affect tire temp.  The best way to measure CoG location on real world cars is to simply measure the weight on each tire by putting them on four separate scales.  The difference between the front and rear is proportional to the CofG.  Unfortunately, we can't do the same for GPL cars, but I assure you that GPL uses 60% plus or minus a few for all chassis.

As you've found, GPL is really very tolerant of setup variations...it's entirely possible to be very quick even using the default setups.  But that is one of the joys of GPL; you can tinker with the setups forever.  

In general, any information you read about how real world cars work also applies to GPL, but as in any simulation, GPL doesn't completely model the real world

If you think this is all weird, just wait until you hear about GPL's antiroll bars.  

Lee

Edited by Lee200, Aug 01 2010 - 06:52 AM.

[Frenchy]

Lee200, on Jul 30 2010 - 09:18 AM, said:

I might add a few points about setup wheel/spring rate, chassis weight, and ride height.

On startup, GPL determines the amount of suspension compression by computing the sprung weight on each wheel from the chassis weight, driver weight, fuel weight, and weight distribution.  It then divides the sprung weight on each wheel by the wheel rate to determine the static suspension compression.

As mentioned before, the normal suspension travel is 19 cm/7.5" so if your goal is to have the static suspension loaded to, say, half its range, you'd want to set the wheel rate so that the static suspension compression is 3.75".  This would give you 3.75" of suspension travel in both bump and droop.

Each car weighs differently, but an approximation is the weight given in the team info menus.  Add in the weight of fuel at .75 KGs/liter or 6.26 pounds per gallon and you have a reasonable estimate of the chassis sprung weight.  Then compute the wheel rate using these formulae:

Front Wheel Rate in Pounds = (Chassis Sprung Weight * (1 - .6 )/ (2 * 3.75)
Rear Wheel Rate in Pounds = (Chassis Sprung Weight * .6) / (2 * 3.75)

All GPL cars use an approximate Weight Distribution of 60% so .6 is used the above formulae.

You may want slightly more bump travel range than droop range so higher wheel rates would do that.

As to ride height, on startup GPL goes through a pretty complicated computation to attach the top of the suspension/spring to the point on the chassis so that the actual ride height is equal to the setup ride height based on static suspension compression, vertical size of the chassis, and tire size.

Lee

from that last paragraph there I assume you mean the static compression due to fuelled wieght? In other word's, if I add race fuel to my qual setup, when I burn off the race fuel I will have a higher ride height than I would in the qual setup? Would this also mean, if I start with a low fuel setup and fill up using BRR's pitstop patch, would I then end up with a ride height of less than 2.50 in. ?

just curious, but thanks again for posting.
Cheers
David

[Lee200]

Frenchy, on Aug 01 2010 - 06:57 AM, said:

from that last paragraph there I assume you mean the static compression due to fuelled wieght? In other word's, if I add race fuel to my qual setup, when I burn off the race fuel I will have a higher ride height than I would in the qual setup? Would this also mean, if I start with a low fuel setup and fill up using BRR's pitstop patch, would I then end up with a ride height of less than 2.50 in. ?

Yes, GPL includes the fuel weight when determining static compression.

It uses the actual fuel weight from the setup to compute the static compression and suspension attachment point.  The suspension attachment point is constant from then on.  So as you burn off fuel your ride height will increase as there is less load on the suspension and they will compress less.

I know nothing about brr's pitstop patch, but assuming it only adds fuel and doesn't adjust the suspension attachment point, then the heavier weight will compress the springs more resulting in a lower ride height than you had before refueling.  Whether it is more or less than the original ride height depends on how much fuel you take in relation to the setup fuel setting.

If I understand your last question, the answer is probably yes.  However, any advantage you might gain from a lower ride height is probably more than offset by the increase in fuel weight.  Interesting question though.  

Lee

[Frenchy]

Lee200, on Aug 01 2010 - 07:21 AM, said:

Frenchy, on Aug 01 2010 - 06:57 AM, said:

from that last paragraph there I assume you mean the static compression due to fuelled wieght? In other word's, if I add race fuel to my qual setup, when I burn off the race fuel I will have a higher ride height than I would in the qual setup? Would this also mean, if I start with a low fuel setup and fill up using BRR's pitstop patch, would I then end up with a ride height of less than 2.50 in. ?

Yes, GPL includes the fuel weight when determining static compression.

It uses the actual fuel weight from the setup to compute the static compression and suspension attachment point.  The suspension attachment point is constant from then on.  So as you burn off fuel your ride height will increase as there is less load on the suspension and they will compress less.

I know nothing about brr's pitstop patch, but assuming it only adds fuel and doesn't adjust the suspension attachment point, then the heavier weight will compress the springs more resulting in a lower ride height than you had before refueling.  Whether it is more or less than the original ride height depends on how much fuel you take in relation to the setup fuel setting.

If I understand your last question, the answer is probably yes.  However, any advantage you might gain from a lower ride height is probably more than offset by the increase in fuel weight.  Interesting question though.  

Lee

Thanks for your quick reply.

Yeah, any advantage gained would be more than offset by my consummate lack of skill . Just interested to understand how it all works.

Cheers
David

[Lee200]

Frenchy, on Aug 01 2010 - 07:28 AM, said:

Yeah, any advantage gained would be more than offset by my consummate lack of skill . Just interested to understand how it all works.

It's an interesting question though David and I thought about it a bit more over a cup of coffee.

Remember way back in '99, Papy introduced its v1.1 patch to GPL which raised the minimum ride height from 1.0" to 2.5".  They didn't say whether this only affected the setup minimum or also affected the physics...I suspect it effectively put a limit on the physics too.  And if it affected the physics, just where in the code did it do so?

So it may be that regardless of the fuel load, the actual static ride height can never go below 2.5" now.  Of course, under heavy load, the chassis may bottom as we know.

Ride height is also just one of the areas where all this fuel load and CofG stuff comes into play.  One of the first things GPL does on start is to compute the roll, pitch, and yaw moments of inertia which are used in conjunction with the instantaneous torques on the chassis to determine the amount of spin in each of those axes.  The moments of inertia are computed only once using half of the maximum possible fuel load and don't change after that.  So regardless of the current fuel load, you will get the same spin characteristics.

Lee

Edited by Lee200, Aug 01 2010 - 07:49 AM.

[ginetto]

Thanks for the info
Another question; how exactly the bump and rebound ("clicks" of the suspensions right?) get involved into this?
Should these 2 values be set according a formula with the variation of the wheel-rate or just following the weight of the car?

[Lee200]

Ginetto, on Aug 02 2010 - 04:40 AM, said:

Thanks for the info
Another question; how exactly the bump and rebound ("clicks" of the suspensions right?) get involved into this?
Should these 2 values be set according a formula with the variation of the wheel-rate or just following the weight of the car?

Hi Ginetto,

My understanding is that the damper force is simply added to the wheel/spring rate force, bump rubber force, and ARB force.

Instead of adding force per inch of suspension compression, dampers add force per speed of the suspension compression.  So the faster the suspension moves, the more force is added.  As speed is distance over time, GPL just takes the difference in suspension compression between each cycle and adds force based on the difference.

When hitting a dip, heavier cars should produce more downward force resulting in faster suspension compression in bump; however, they MAY in fact resist upward force at a crest resulting in slower suspension decompression in droop.  So you'd think the bump force should be more than the rebound force.  When in roll; however, the two opposite suspensions are moving at the same speed although in opposite directions so the bump and rebound force should be about the same.

In GPL, the bump resistance is less than the rebound resistance.  Here's a chart of GPL's damper force courtesy of Gene.  The bottom axis isn't labeled, but it the units are in distance per time.

To answer your question, seems to me that the heavier the car, the higher the bump and rebound settings should be.  But since the wheel/spring force gets added in, a high wheel/spring setting may allow lower bump and rebound settings.  

Lee

Attached Files

•  dampers.jpg   33.95K   220 downloads

Edited by Lee200, Aug 02 2010 - 06:44 AM.

[John Woods]

This kind of stuff makes me wish I was smart. All this time, and it never occurred to me bump/rebd was asymetric.
First, it never occurred to me lower shock rates could complement stiffer springs...I thought exactly the opposite; that stiffer shocks were needed to control stiffer springs. (If I read things correctly).
From the graph, it appears a setting of 4 for bump and 1 for rebd would equal a 50%-50% bump/rebd setting? But 50/50 may not be a good idea because?
An ALMS, (or maybe Rolex Series), Chief Mechanic once said in an interview, "Well, we worked on the car until the last minute and got no practice time, so we sat the shocks up 50/50 and we'll have to make changes as we see how things go in the race." So I've been starting out with settings like 3/3 and tweaking...thinking I knew what I was doing.

Edited by John Woods, Aug 02 2010 - 09:35 AM.

[Lee200]

John Woods, on Aug 02 2010 - 09:24 AM, said:

This kind of stuff makes me wish I was smart. All this time, and it never occurred to me bump/rebd was asymetric.
First, it never occurred to me lower shock rates could complement stiffer springs...I thought exactly the opposite; that stiffer shocks were needed to control stiffer springs. (If I read things correctly).
From the graph, it appears a setting of 4 for bump and 1 for rebd would equal a 50%-50% bump/rebd setting? But 50/50 may not be a good idea because?
An ALMS, (or maybe Rolex Series), Chief Mechanic once said in an interview, "Well, we worked on the car until the last minute and got no practice time, so we sat the shocks up 50/50 and we'll have to make changes as we see how things go in the race." So I've been starting out with settings like 3/3 and tweaking...thinking I knew what I was doing.

Me too...wish I were smart.  

Don't get me wrong; I think in the real world and probably in GPL, you should use higher damper settings and wheel/spring rates for heavier cars.  But it all comes back to how you want to make the car feel.

Yes, you are reading the graph correctly.  4 to 1 results in the same force in bump and rebound.

Well no one took my bait so I'll throw in what we know about roll bars anyway.  

The sprung weight transfers from the inside tires to the outside tires in a turn and as I said before, we have no control over this as for any given lateral G, the load transfer will always be the same.  Load transfer is resisted by the springs and bump rubbers and roll bars (ARBs).  The sum of the spring, bump rubber, and ARB forces is the total roll resistance force.

In the real word, what is totally magic is that the proportion of total load that is transferred to the front or rear outside tires is directly proportional to the roll resistance at each end.  For example, if the front has 75% of the total roll resistance and the rear has the remaining 25%, 75% of the total load transfer will go to the front outside tire and only 25% to the rear.  This is great stuff for fine tuning the handling of a car as we can use the ARBs to proportion how much load we want to go to the front and rear tires thus making the car understeer or oversteer.  The more you load a tire, the less grip it has and vice versa.

So much for the real world...how about GPL?

In GPL, the ARBs resist roll by adding roll resistance to the outside wheel just like the real world.  It does this by adding force to the outside tire based on the difference in suspension compressions on opposite wheels and the ARB setting, but also removes the same amount of force from the inside tire.  This is very strange and I'm not at all sure that real world ARBs do this.  Looking at a ARB which is attached to opposite side suspensions, you might expect that as the outside suspension is loaded with force under roll, the inside suspension would unload by the same amount of force, but is that really true?

I've read several books and articles on this and have found no definitive answer.  Milliken's "Race Car Vehicle Dynamics" has only one line in 800 pages that says, "Twisting of the bar adds load to one wheel and removes it equally from the other.", and Staniforth's "Competition Car Suspension" says the roll bar transfers load from the inside to the outside tire.  In both books when a numerical example of load transfer is given, nothing is shown about load being added back to the inside tire.

Regardless of whether GPL's is correct or not, that's how it works.  What is important though is that GPL's method does not provide load transfer from one axle to the other...you can't apportion more load to the front or rear based on the relative proportions of roll resistance on each axle.  All you can do with the ARBs is to even out the load between the two wheels on the same axle which SHOULD even out their grip and temperatures.

Remember that total roll resistance includes the springs and bump rubbers (and dampers too when the suspension is moving) so they have an effect on load transfer just like the ARBs.  What is different about the ARBs is that they only work when the chassis rolls; they have no effect under straight line dive or squat.

EDIT:

As the ARBs only work in roll, clever suspension geometry can also have a similar effect during pitch.  Called antidive and antisquat, these effects come into play whenever the chassis pitches forward under deceleration or backward under acceleration.  GPL properly models these effects by adding force in the opposite direction so that the pitch change is lessened during longitudinal load transfer.

Lee

Edited by Lee200, Aug 03 2010 - 06:41 AM.

[John Woods]

Amazing. Information nowhere else available? Incredible!
IMHO

What more can there be?

[FloP]

Thank you very much for this wealth of information, Lee! I'm reading every post of yours with great interest!

[John Woods]

Two questions I'd like to know answers to:

How to optimize correlation of shocks and diff to achieve maximum launch, sustained acceleration, and top speed.

How to correlate that to original 11 track specific grip coefficients.

Just wondering. Figured I'd skip the bait and go the whole way.

But I still want more bait.

[Lee200]

Well since we have a few crazies like me who are interested in how GPL works, here's a bit on the how the computer simulation actually operates.  It's not essential to know and probably has no practical application to making your car run any faster, but does give you some more insight on this wonderful program.

Inside the guts of GPL are constructs call "rigid bodies" which are used to simulate parts of the car.  I think of a rigid body as being a homogeneous sphere that has no size and cannot be compressed or twisted, but it can move.  In a "Six Degree of Freedom" model, a rigid body can move in six dimensions:

Linear Motion:

1)  Forward or backward along the longitudinal axis
2)  Left or right along the lateral axis
3)  Up or down along the vertical axis

Rotational Motion:

4)  Pitch about the lateral axis
5)  Roll about the longitudinal axis
6)  Yaw about the vertical axis

GPL uses twelve...yes, twelve different rigid bodies.  They are:

1)  Chassis
2)  Fuel
3)  Front Left Wheel Hub
4)  Front Right Wheel Hub
5)  Front Left Wheel
6)  Front Right Wheel
7)  Rear Left Wheel
8)  Rear Right Wheel
9)  Engine
10)  Clutch
11)  Outshaft--differential
12)  Head--player's head

Not all of these rigid bodies are allowed six degrees of movement.  The chassis and head do, but the wheels can only rotate about one axis and move up and down in another while the fuel has no freedom of movement at all.

The front wheels have hubs while the rear do not.  This allows the front wheels to be steered.

Note also that the head, which simulates the driver's head, is given its own rigid body, but apparently Papy didn't have time to finish the code that would make it move independently  to simulate bumps and loads on the driver.  

The most important rigid body is the chassis as all the other rigid bodies are connected to it.  The chassis has one other important characteristic...it has mass.

A few hundred years ago, a rather clever Englishman named Newton came up with his three laws of motion while watching apples fall on his head from a tree.  His second law states that Force = Mass * Acceleration which can be rewritten as Acceleration = Force / Mass.  So if we apply a given force to an object of given mass, the object will accelerate according to his formula.

GPL measures all the different forces acting on the chassis and combines them into a single force for each of the three linear directions of motion.  When applied to the chassis mass, an acceleration results which between two successive time points can be used to determine  the change in speed and also distance traveled for each of the three linear axes.  The mathematics of how GPL does this is not fully understood as it involves solutions of differential equations and is stuff that is way, way beyond my comprehension, but this is the gist of what happens.

So GPL keeps track of the rigid body's acceleration, speed, and position and repeatedly updates these 388 times per second!  No wonder that 1990s computers sputtered.  GPL also keeps track of the rigid body's linear momentum (which is speed times mass) for use during collisions with other objects such as the hay bales I so often visit.

GPL does exactly the same thing for rotation.  It keeps track of the rigid body's rotational acceleration, speed, and direction.  And like the linear components, it also keeps track of rotational momentum which explains why your GPL car can do all those 360s coming off a tight turn.

The rotational analogs of force and mass are torque and moment of inertia.  As we all know, torque is a force applied over a distance and is measured in pounds feet or some such.  Moment of inertia is a harder concept to grasp, but there are formulae for figuring out the moment of inertia for all sorts of differently shaped objects.  As I've mentioned before, GPL figures an overall car moment of inertia at startup and uses that throughout in its rotational computations.   The formula for rotation is Acceleration = Torque / Moment of Inertia.

To wind this up, here's a list of forces that act on the chassis.  As the forces act over a predefined distance, they also generate a torque on the chassis.  For example, the tire acts upward through the wheel to the spring which is located outboard of the chassis based on the track size.

1) Gravity
2) ARBs
3) Springs
4) Engine
5) Clutch
6) Head!

Lee

Edited by Lee200, Aug 02 2010 - 05:17 PM.

[Lee200]

John Woods, on Aug 02 2010 - 04:18 PM, said:

Two questions I'd like to know answers to:

How to optimize correlation of shocks and diff to achieve maximum launch, sustained acceleration, and top speed.

How to correlate that to original 11 track specific grip coefficients.

Just wondering. Figured I'd skip the bait and go the whole way.

But I still want more bait.

Ah Grasshopper, that is in Lesson #2.  But first you must learn to walk on the rice paper.  

Seriously John, drivetrain physics is an entirely separate subject and so far, we've only touched on suspension physics.  But if you're interested, I'll get into that later.

I don't know what you mean by 11 track specific grip coefficients though.  Are you referring to the 11 original tracks?  If so, each track does NOT have different grips as they basically all use the same track surface.  In GPL, the surface grip is set according to its composition with the best being asphalt.  So as long as you are on the track and not off in the weeds, you're getting the best grip possible.  I can tell you though that the hay bales don't offer much grip.  

Lee

[John Woods]

Okay, I can be patient.

A vague memory involves reading about track builders setting grip or traction variables. Isn't there something about that in the GPLEA Guides? (Stefan just posted them, guess I could check). Or somewhere. And it is difficult for me to believe Mosport, Kyalami, and Monza all have the same grip. I always thought the factorials of the variables offered a wide range of choice for track surface grip, and matching a setup to a particular unknown grip recipe was part of the art of being fast.

But I'll just sit quietly and try to not get too far ahead. Except, I just noted the exclaimation sign after "head." As in, head motion figures into force acting on the chassis?!!! Or is that the unfinished part? And clutch?

Well? I'm waiting. Quietly...

Edited by John Woods, Aug 02 2010 - 10:53 PM.