Weight Bias - Redux

[Shiny Side Up!]

My brain wants think that the stiffness should align itself with the weight bias of the car. 

I think that you'll find that the optimal roll stiffness F vs R has very little to do with their weight distribution and has more to do with the overall suspension kinematics and the characteristics of the tires.

[Blooze]

Hmmm, 'splain that...

$

[Silvo 1981]

I've got all the data I could. I collected Aero as well because I thought it could have some bearing, it seemed fairly unique to each car and I didn't want to have to go back into every car to get the readings again in case it was relevant. I didn't collect tires, alignment, brake or diff settings because a lot of it seemed to be repeating or unnecessary. Some of the non-race cars with high PI made it on the list as they had both adjustable suspension and roll bars.

I've also added total figures as well as the bias ratio between front and rear to see if there are patterns that can be determined. I think I can begin to see some patterns emerging but I thought I would let everyone else see where I am with it at this stage.

[Blooze]

What an excellent job!!  Thank you.

I am off to digest...

$

[Silvo 1981]

The things I have noticed are:

The ARB bias is always 55%
The springs, bump and rebound bias are all extremely close to the published Wb. There are minor discrepancies but these are likely to be rounding errors on small numbers.
The rebound to bump ratio is near enough 66.66%

The next thing I noticed was that there were clear divisions in the totals for the rebound, bump and ARB. These divisions were also present in the total spring values although they were not obvious until you divided the total spring value by the weight of the car. These values for the totals are as follows:

ARB: 25.62, 22.18, 19.11, 16.38
Rebound: 20.1, 18.4, 16.6, 12.5
Bump: 10.1, 9.2, 8.3, 6.3

The values I have used for the springs are the averages for the groups set out above as there is more variation in these values than the above three items
Car Weight/Total Springs: 1.508, 1.741, 2.021, 2.946

By using the vehicles weight, bias and manually selecting one of the groups from the constant values above it is possible to accurately predict the race car setups. I have no idea yet as to whether or not these setups will work well or not on the production cars. I also do not know why these cars are grouped as they are but it is interesting that they can be predicted to a fairly accurate level so easily. The groups do seem to be categorized through some loose order of PI class but it is very loose and not definitive by any means.

I have attached a spreadsheet I have been working on to show how these figures can be predicted and I have also shown some testing I have done on the 2nd sheet to show the level of accuracy being achieved.

[Blooze]

Very interesting.

I have been thinking about how they would do this and a jump table has been looming large.

I looked through the car traits ( X - List in the My Cars screen)  Here's a list of what I think the stored information is.

Weight	Initial weight of the car
Wb	Published weight bias
Sb	Actual weight bias to use in calcs
Car Type	Producion, Pre-tuned, and Race
Body Style	There are a number of these: Coupe, Super Car, GT1, etc..
Engine Location	F, M, and R
Drive	F, R, and A

Now, I'm wonderin' if there is some way to use a combination of these (in an AI Matrix, perhaps) to get to that Ref constant of yours, or maybe in some fashion, they are your Ref Constant.

$

[feuerdog]

Some of my damper testing revealed an interesting similarity to your damper ratio of 66%.

In the heavy chassis(5900lb chassis) benchmark references I was using, there was a clear advantage in accel and decel performance when having rebound set with a BUMP * 1.33 calc. resulting in a bump/rebound ratio of 2:3 or 66%.

However, the "feel" of the cars was very twitchy. A 1.5 multiplier is a bit too muted by itself so I settled on a 1.4 multiplier for the new calcs.

But then again, the best rebound to spring ratio seems to be variable by wheelbase/CoG too.

Very interesting indeed.


My new personal calcs in regards to ARB tuning are inversely weight balanced to my aforementioned 46% drivetrain modified virtual workload center.

My scoreboard times(ABS and Auto trans) can speak for the success of these calcs to date, as they all use untweaked setups from this personal calculator.

[Blooze]

The 66% thing is nothing new - every shock in Forza II was initialized to this value.

Bump = Rebound / 2

Interesting that Race cars still use that ratio while the production cars are using extent of the range at around 70%.  That would be the value to wrap one's head around.  This morning I herded the Abarth 500 through the Factory Spec races and I noticed more than once as the thing went all a-wash on me as I lifted the throttle whilst in the midst of some steering that "I betcha knockin' the Bump back to 50% would go a long way towards straitnin' this thing out.

Also, with the R cars following the Spring Bias for their Dampers, it would most likely behoove us to stick with that should we decide to alter the Spring Bias.  Last night I got tired of having to be so tender on turn-in with the FXX else it snap out away with me, I went ahead and rebalanced the car to 46% instead of the 41%.  I reset the Damper's Bias to 46% as well.  That pretty well solved the problem.

$

[Blooze]

Ya know, I had me a thought.

The purpose built racers are set up suspension-wise not unlike they were last year.  But the production cars have all these strange numbers - much stronger than they were last year, stronger than this year's race cars.

We have all read the interviews were DG and company all claim that absolute neophytes will have playing this game.  And, drifting is a pretty visible part of the game.  It shows up on all the trailers, is even the focus of a few.  And now one of the up-front guys made his bones as a drifter.  All of the production cars, the ones the drifter cats are looking for have...

• Much stiffer springs - double or more from what they were last year.
• Much stiffer ARB - same thing - lots more than last year.
• And the Dampers as well, coupled with Bump damping that is even 50% on top of what was added.
• All the cars are more centered, its disquised with the phony Wb, but they all are in the neighborhood of 50%

I don't mess with it much, not at all actually, but if I were going to set up a car to drift, all those things would be high on the list.

The race cars - well, they need to be racey.  And we see a much more recognizable set of parameters on them.

Hmmm

  $

[Shiny Side Up!]

Hmmm, 'splain that... $

I'll try to give it a shot... not sure how coherent my train of thought will be!

Both ends of the car typically have different numbers for the variables contributing to roll stiffness.  For example, contributions to the roll stiffness include spring rate, tire stiffness, motion ratio, roll bar geometry and track width...  for any given car, these could all potentially be different front to rear (and in oval racing, different on all 4 corners!).  On top of this, in the real world, track and tire conditions are constantly changing and the ideal roll stiffness at each end, and the ideal ratio between them, becomes a moving target even if you are able to calculate an ideal starting point.

If we make the assumption that FM3 simplifies this and uses the same track width F/R, the same SLA geometry for motion ratios and roll bars, and the same tire stiffness, you still end up with a not-so-straightforward problem in that you are trying to find the optimal roll at each end to both
1) control roll at each end to maximize the tire contact patch with an eye to maximizing grip at each end, and
2) balance the car for oversteer/understeer (which likely will require you to compromise on what you would do for #1!)

The first issue is a little more straight forward in that you can try to boil it down to weight distribution and geometry - but it still gets bogged down with tire operating conditions when you are on-track (temp, pressure, slip angle, and camber thrust among others).

The second issue is much more subjective since it depends so much on the driver (but then, the tire operating conditions in #1 do, too).

When you factor all this fun into account, most RWD cars in the real world end up with the F:R roll resistance ratio from 45:55 to 70:30.  Although it is also true that most cars have a F:R weight distribution in the same relative range, it is extremely uncommon for the best roll ratio to correspond to the weight distribution.  A general rule of thumb often employed for a RWD car is that a good starting point is around 60% F:R irrespective of weight distribution.  Ratios in this range typically have fast transient response and good capability to get the power down.  Obviously, there are a lot of other factors that go into it and some cars vary wildly from this, but it is a starting point.

[feuerdog]

Finding balance based only on weight is an incomplete idea though.

The main reason we started biasing springs, based on weight, was to enable a more aggressive application of throttle, but even this is wrong.

If we bias anything solely to accelerative forces, there is a good chance that decelerative forces will become affected.

The problem is solved by looking at the forces at work of the chassis, accelerative grip(accel), decelerative grip(braking), yaw change(directional change), lateral grip(steady state cornering).

The accel and decel biases are easy to understand. The steady state cornering forces have alot to do with controlling the mass/weight of the car as much as possible to keep the tires in a state of grip. It's here where a good portion of the roll resistance comes into play.

Braking forces are predictable for the most part because all cars have 4 wheel braking.
Acceleration forces are more varied due to the three types of drivetrain and thier inherent imbalances.
Drivetrain type can even affect deceleration foces via engine braking.

The yaw change, or initiation of actually turning, is a unique force in itself because it is a highly variable and transient force. Depending on the other forces at work the turn initiation can cause understeer or oversteer. It's not a major force to adjust for but it is something to be considered as far as work being done.

And none of this even mentions aerodynamics, unsprung mass, or tire size(grip) variables.

Given all of the forces at work, and how they are all interdependant, how can we possibly attribute ARB settings to one single aspect of the chassis?

In my opinion we have to tune the suspension as a single entity, with every component having its own area of dominance but with the proper allowances for its effect on the other components. We have to look at the relationships of the parts as compound problems, not simple relationships.

As far as ARBs are concerned, i've been focusing on tuning them with consideration to a balance of several elements:
1. Primarily they reduce body roll - in this matter i'm using the accumulated value of the two bars. The higher the number the stiffer the overall roll resistance.
2. Secondary biases - Drivetrain, braking, turning, and weight distribution. We already know how successful the 60% front dist. FF chassis can be with the 0f/40r ARB setting, and rearward biased spring rates. All this is a matching the drivetrain, braking, weight, and turning imbalances and counteracting them with imbalances of our own.

By understanding and combining the effects of each of these counterbalances, we can set them to work in cooperation with eachother, which acheives:
1. Minimal adjustments and therefore a more balanced chassis.
2. Fewer side effects of extreme counterbalance settings.

I think the important thing to try and understand is how the settings need to be considered as a single entity, and then adjusted as a whole in order to compensate for inherent chassis imbalances in as minimlistic a way as possible, which means in cooperation with eachother.

My brain hurts now, so i'm going to go lie down for a bit.
 

[Blooze]

Guys, that is all good stuff... but I fear we might have run off into the land of amock here.

In hopes of bringing this thing back in line, let me make myself a little more clear about the question I would like answered/discussed etc...

Now, the purpose of this thread, hell, this board even, is to fellowship around tuning the cars in the game Forza.  It is our luck in many ways that Real World and Forza can mesh and intermingle in this arena and that is in particular what draws many of us to the game.  But, to get all a tangle in measurements that we have no access to - to see, let alone alter - while perhaps interesting, does little to enlighten the issue at hand.

I contend that the Published Weight Bias has little or nothing to do with the physics engine - it is no more than an interesting tidbit on the track, a waterfall, and like Kaido's Waterfall, paying too much attention to it will put you into the fence if you're not careful.  Furthermore, I contend that the Spring Bias, is important to the game, along with a calculated value, SWR...

(Front Spring + Rear Spring)/Curb Weight

(For intance - weight changes are applied to the springs according to the SWR, not the Sb, and most certainly not the Wb)

The argument for my position is manifest, and I won't repeat all of it here.  I understand this.  I can grasp why they are doing things like this.  There are 428 cars in the database - coming up with a unique set of parameters for each car on a one by one basis would be prohibitive.  So, they are calculating many of the values, deriving them from a few data items that are stored in the car record.

You can see what a bunch of the data record fields are by checking the Car List from the Car Detail screen.  Part of the program's process is to load a runtime structure, or object, with these items of data, and to populate other values in the structure with calculations so they don't have to be recalculated each time they are addressed.  Now, all of the car's data items are not displayed, and tow of these hidden fields are the initial SWR and Spring Bias.  They are saved in a data structure stored on your xBox once you buy a car, or build a Tune for one in Free Run or something.  

Now, you can get the...

• Total Spring Weight by multiplying the SWR x the Curb Weight.  Front and Rear are divied up according to the Spring Bias.
• Total ARB by using the SWR as a varible in a trend calculation, the constants in which are located in a jump table associated with the data items, Car Type, Body Style, Engine Location, and Drive.  There are essentially two, one for production and another for race an pre-tuned.  Tonka figued this out last year.  I jury rigged a new one for the race cars using Phil's data and Tonka's old equations.  It's pretty good too - within a few thousands.  I could have got it closer but I was using a binary hunt and peck process and it was getting late.  Suffice it to say, if we can do it - so can they.  Mark this point... they take that total ARB and divvy it up by 55% to the front.  For every car!
• Total Rebound Damping by using the SWR and another trend calc.  There appears to by two, one for production types, and the other for race and pre-tuned types.  The Total Rebound is divvied up by the Spring Bias.  Bump is derived from the Rebound values using a jump table - basically 70% for production cars and 50% for everything else.

That's really not too bad when you think about it.  For a few fields and a bit of code you can describe the handling characteristics of the world's four wheeled objects.

Now let's consider the Forza Mission Statement.  The stuff that gets talked up in the conference rooms at the back of the building.  They want realism.  They have built their reputation on it.  They claim it in the media.  They want to provide a realistic driving experience.  Then the issue comes to balancing what it takes to be realistic and fiscally responsible at the same time.  Hence, the formulaic mechanism.

So... I don't know whether you buy all this stuff I just said or not.  But if you consider it, does a straight 55% biased to the front arrangement of the ARB for every car in the world make sense?  Last year the ARB were distributed by the inverse of the Sb.  Were they that wrong?  Is there some automotive geometry rule that supports the 55% deal?

And the damping.  Last year Bump was 50% of Rebound across the board.  This year it is 70% for production cars and 50% for cars that have been fine tuned - the racers.  What would be the impedous for putting the production cars near the high end of the industry norm?

And, while we are at it.  Production cars have an SWR that damn near doubles their suspension setup values.  Is there an industry standard that indicates they were too soft last year?

My thought is that if we can get a good understanding of these changes to the system will help us get a grip on the things we need to change with our tuning to improve the performance of our cars.  I mean, I've been known to subscribe to "Shoot and Holler Sh*t" methodologies as much as the next guy.  But I do try to avoid it when I can...

$

[Shiny Side Up!]

Guys, that is all good stuff... but I fear we might have run off into the land of amock here...

...But if you consider it, does a straight 55% biased to the front arrangement of the ARB for every car in the world make sense?  Last year the ARB were distributed by the inverse of the Sb.  Were they that wrong?  Is there some automotive geometry rule that supports the 55% deal?

And the damping.  Last year Bump was 50% of Rebound across the board.  This year it is 70% for production cars and 50% for cars that have been fine tuned - the racers.  What would be the impedous for putting the production cars near the high end of the industry norm?

And, while we are at it.  Production cars have an SWR that damn near doubles their suspension setup values.  Is there an industry standard that indicates they were too soft last year?

$

For the question about a universal 55% ARB Front Bias, I don't think it makes sense in the real world as it simply isn't true... however, it is in the range of a generally acceptable %, and from that point of view it is not out-to-lunch.    I have seen some real world examples of cars that met last year's assumption of inverse relationship for ARBs wrt the Sb, and in fact it is not uncommon to be in that ballpark - but the examples that spring to mind are light, high power/weight ratio, purebred machines - from that view it would suprise me if it is a common production car setup given the completely different performance goals!

I will have to look some more into the second and third questions... I must confess that I really haven't paid much attention to the SWR debate and discussions and need to try to square it with my experience to talk intelligently to it (as soon as I started typing a response to question 2 including "motion ratio" I realized I was veering off track again...).

[Blooze]

Thanks!  I guess some research and 'spirimentating is in order.  I need to get one of those Suspension books.  Google is my friend and all that.

The inverse thing always made sense to me for nose heavy cars, specially if they were driving off the rear wheels.  And the fixes made sense and worked.  If it's pushing, soften the front - if it's loose, soften the back.  But then, not so much for RR cars.  It seemed like you would want to keep the back end flat to keep the driving wheels on the ground.

Any hoo... I have me an Acura that I've been messing with - I think I will invert the ARB and set the bias to the Sb later on tonight and see if that works as good in this version as it did in the last one.

As far as the SWR thing goes - I doubt you'll find much about it anywhere but here.  I certainly doubt that Turn 10 would own up to it.  I came up with it mostly like how that dude found Neptune... he never really saw it, but everything around it acted like it was there.

$

[feuerdog]

In the idea of formulaic default tuning the 55% front ARB makes perfect sense. It's safe.
It promotes a slight understeer trend for most cars, or at the very least compensates slightly for a car that is inherently overtseery.
It's a safe default ratio of 11:9

My latest calcs, based on real world relationships have been quite successful and balanced(in combination with other chassis balance tuning) with ratios between 1:3 thru 5:3

The increase in spring strength I think has more to do with simply emulating real world values, nothing more.

The damping ratios make more sense now too, again, more in line with real world relationships.

As i've mentioned elsewhere as per some of my chassis dynamics books, typical damping ratios look something like this:
1:5 luxury car - 80%
1:4 typical modern car -75%
1:3 sports sedan - 66%
1:2 sports car - 50%
1:1 racing car - 0%


For reference i've been testing:
1:1.33 (FM3 baseline? Too twitchy with current FM3 calcs) 25%ish
1:1.45 (soon to be calc setting) 31%ish
1:1.5 (testing with higher spring rates. Too muted with current FM3 calcs) 33%ish

The relationship of lower damping rates for racing cars makes realistic sense because of the amplified aerodynamic properties. Namely, the increased downforces on the chassis amplify the effort the rebound damping is trying to perform. Therefore the need to regain chassis response while under the effects of aerodynamic downforce is to reduce rebound damping.
But thats only half the story.
As the spring rates increase for faster cars due to both higher velocities and the need for aerodynamic compensation, the resistance needed in bump damping is reduced. Some kind of compliance needs to maintained.
The combination of the lower bump in relation to increased spring value and the lower rebound in conjunction with the effect of downforce bring the ratio closer to 1:1.

This is all very simplified for FM but they do represent realistic traits.