Roll stiffness distribution and effects of changing spring rates.

[Komodo86]

Hi all, first post on the forum so go easy on me!

My question relates to a discussion on an owners club forum where the members were discussing the effects of spring rate changes on the cars handling balance. The car in question was a FWD sports car, with equal spring rates front and rear, and big front and rear anti-roll bars and around a 62% front weight distribution.

It was mooted that should one wish to increase the roll stiffness of the suspension for track use with sticky tyres, the spring rates should be increased by an equal amount front and rear so as to retain the stock cars fine handling balance. I disagreed and put forward the idea that this would infact increase understeer in this particular application, but my idea was rubbished by one member and then the thread died.

My thinking was this; given that the roll bars on the car generate a roll stiffness of around 2000lbs.in/degree at the front, and 4000lbs.in/degree in the rear, and the stock springs generate around 2500/2500 respectively, then we can see that the total roll stiffness is 4500/6500 or a 40.9% front RSD. If we double the spring rates to 5000/5000, sure we end up with an equal RSD contribution from the springs, but when considered in conjunction with the roll bars we end up with a total RSD of 7000/9000 with a 43.75% front RSD. So for the same increase in spring rate front and rear, we have actually increased the front RSD by 2.85%. In order to keep the RSD the same as stock, the rear springs would need to be increased by a greater proportion, in this case by an extra 22.3% to 6115.

While the one reply somewhat agreed with the maths, he said it was of no consideration as the imbalance could be made up by using more negative camber up front to combat the increased push, and that you should not throw away grip from one end of the car to improve the other. I see this approach as being the complete opposite, with the correct RSD balance then the neg. camber can be increased by an equal amount front and rear to generate an overall increase in grip on both axles, rather than just one.

Seeing as I had no further replies on this topic I thought I may have been barking up the wrong tree, but it could simply be that having never calculated the roll stiffness contributions, the other readers simply had no input as to whether I was talking gibberish!

I am hoping you knowledgeable folk here might set me straight on this matter...

 

[NormPeterson]

"RSD" is only part of the picture. What does the %change look like if you look at it from a lateral load transfer distribution point of view including the effects of geometric roll center heights? If you separate out the unsprung masses from the sprung mass?

Keep in mind that with greater roll stiffness and assuming that you end up at the same ride heights, you should experience less camber loss.


Norm

 

[Komodo86]

Ok well i'll run the actual values through the spreadsheet.

Original values:

Spring rate: 250/250lbin
Sway bar linear rate: 74.48/142.99lb/in

With all the other associated values, this gives me 2749/2867lb.in/deg for the springs and 2186/4198 for the sway bars, so 4935/7065 in total for a 42.55% front RSD. This works out at 45.8% front load transfer distribution.

Double the spring rates to 500/500 and we get 5498/5734 spring contribution and 7684/9931 total, 43.62% front roll stiffness, 46.9% front load transfer distribution.

So the values are less than the example in the original post, but the trend stays the same. Increasing the rear rate by that extra 22.3% brings the RSD back to 42.68% and front load transfer back to 45.6%, as predicted.

Point taken on the camber loss, but would I be on the right track with my setup verses the guy who chooses the 500/500 rates and runs a sub optimal rear camber angle to 'throw away' rear axle grip and balance the increased understeer?

 

[NormPeterson]

Does your spreadsheet include the effect of roll center height? Do you know what the RCH's are? If the heights are much different, or relatively high, spring and bar stiffnesses alone will give you a distorted picture of what you really want to know.


Norm

 

[Komodo86]

It does, front RCH has been measured and modelled at 2.5" and rear was guessed at around 5". I do need to get under the car and get some measurements from the rear suspension to get a more accurate value for the rear RCH, but I believe 5" is in the right ballpark.

The calculations are taken from Competition Car Suspension by Allan Staniforth.

 

[GregLocock]

I think the guy who will win races is the one who throws away 'fine handling balance' and instead develops a complete racing package. As a starting point either of the two approaches is equally easy to argue with, or if you prefer, equally good.

It all depends on which part of the corner the current setup fails. Increasing the roll stiffness is a bit counterintuitive to me, making the outside tires work harder than the inner ones doesn't seem like an automatic recipe for success. This assumes you have a decent geometry setup.




Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[cibachrome]

I might doubt the spring rates front and rear are equal. What is the basis for your statement (measurement, guess, lore, car magazine, SWAG, ...) ? What are the ride rates? A ride rate ratio (rear to front) of GT 1 is commonly done.

Meanwhile, a TLLTD of (wgt distribution minus a few percent) is most likely the factual number. Otherwise it would be unsaleable. The fwd inclined roll axis should be telling you this, as well as the difference in front and rear camber-by-roll ratios.

Can you name the car? I might have a K&C test on it in my tree house.

 

[Komodo86]

The car is a Honda Interga R <2000. Factory spring rates are given from nurmerous sources as being 250/250 (240/240 according to some sources).

"What are the ride rates? A ride rate ratio (rear to front) of GT 1 is commonly done."

Motion ratios are measured at .7 F/R so the ride rates are 122/123lbin given an unsprung mass measured at 61/68lbs. I do not follow the second part of that statement?

"Meanwhile, a TLLTD of (wgt distribution minus a few percent) is most likely the factual number. Otherwise it would be unsaleable. The fwd inclined roll axis should be telling you this, as well as the difference in front and rear camber-by-roll ratios."

I am not entirely sure I follow you here? Are you saying that the front LLTD should be 62%, matching the weight distribution? The stock car is known to be very tail happy for FWD, it is extremely resistant to understeer.

 

[Komodo86]

"I think the guy who will win races is the one who throws away 'fine handling balance' and instead develops a complete racing package. As a starting point either of the two approaches is equally easy to argue with, or if you prefer, equally good."

I did suspect there would be no 'right' answer, but some opinion on the theory was needed to see if I was going the wrong way. Here, competition sprint/hillcimb cars are set up with the front heavy springs and then the geometry is set to match, the way I saw it, this was thowing away potential grip. In the US, autox cars are set up rear biased which makes more sense the way I am seeing it, although those guys go too far and end up cocking a wheel way up in the air.

"It all depends on which part of the corner the current setup fails. Increasing the roll stiffness is a bit counterintuitive to me, making the outside tires work harder than the inner ones doesn't seem like an automatic recipe for success. This assumes you have a decent geometry setup."

Nothing fails as such, but the car was designed around mid 90s tyre technology, throw on some >'10 spec R compounds and you are generating more lateral acceleration than the car was originally designed for, so the increase in roll stiffness is the usual method to bring the roll angle back towards stock levels to maintain the factory designed geometry.

 

[NormPeterson]

Those bar rates - are those numbers taken at the bar ends or at the wheels (through the bars' own motion ratios)?

The TLLTD for a production car should fall within a few percent of the weight distribution numbers. Expect the front TLLTD% to run a few % higher than the front weight% for a RWD car, perhaps a few % less for FWD.


What's commonly done in autocrossing FWD cars is done to provide the best overall result in a rather narrowly defined set of circumstances involving low speeds, near-continuous rapid-fire maneuvering, and the lack of much that qualifies as a "straight" where WOT acceleration is possible without getting wheelspin. A successful autoX setup would not be appropriate for higher speed activities, and in many cases isn't all that acceptable for street driving either (too oversteerish). But within its activity, an autocross setup may not be as nutty as it looks - once a car becomes a tricycle, where do you think any additional LLT goes, and what might that mean in terms of tire loading and slip angles?


Norm

 

[Komodo86]

The bar rates are as measured at the wheel through their own motion ratios of .53/.55.

For comparison, the base spec car does have a TLLTD of around 57%, which seems to tie in with what you say, but then it is a typical understeering FWD production car. With the substantial increase in rear roll resistance coupled with the LSD in the Type R, Honda managed to produce something that doesn't feel much like a FWD at all, it does certainly appear to be a special case among common FWD setups.

As for lifting a wheel, once that axle has contributed all of its load transfer, any more must come from the remaining axle. Thus lifting a rear wheel = increased front load transfer = increased understeer. In this case, for 1G lat. acc. the inside rear appears to be planted on the ground for all three setups 250/250 500/500 and 500/615, so I don't forsee any problems there. According to my workings the inside rear does not become fully unloaded until you reach 40% FLLTD, but we are well into the realms of akwardly set up autox setups then (i.e. no front bar, 500/700 springs, quite a common setup amongst Honda autox drivers.)

I think I may have just found the flaw in my idea though, I recall writing in the sums necessary to account for wheel lift (not given in the book), but they are not taken into account on the number I am looking at, they run on a seperate table where I was attempting to model tyre behaviour. I guess given a target increase in peak G, I could well run into wheel lift problems with the stiffer setup.

I'll check it out and report back!

 

[Komodo86]

Just checked out the wheel lift.

Stock car lifts at 1.2G
500/500 lifts at 1.25G
500/615 lifts at 1.2G

So it still hints at 500/500 increasing understeer and 500/615 retaining the stock balance? I guess the question would rely on whether the tyres are capable of generating more than 1.2G!

 

[GregLocock]

Roll stiffness distribution tends to be rearward biased on good FWD cars, sometimes to a surprising degree. The reason for this is that good grip under braking and acceleration for the front wheels is far more important than getting the axle weight related understeer right. Of course you then need to reclaim the understeer by other methods.

No names no pack drill, but a very good handling FWD has a 60/40 weight bias and a 43/57 LLTD.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Komodo86]

On this subject of LLTD, I have come to realise that this number could well be skewed by the fact that the car in question uses linear front spring rates but progressive rears. While the advertised rates are 250lbin for both axles, clearly the load transfer on the rear axle is not constant.

There are a few ways to look at this. One is that the progressive coils are concerned mostly with ride, and that all body motions are controlled by the 'main' rate. Having looked at the coils on my car, they look very close at the top, so much so that I expect they will be bound at ride height. Would this have the same effect as them being linear under normal driving conditions?

Another view is that the rear axle experiences an increasing rate of load transfer in a corner. A simplistic way of looking at it without considering the geometry, but this is also possible.

Needless to say, this throws any estimates of what linear rear springs I would need out, as I now have no way to compare the load transfer effects between a linear and progressive spring.

 

[GregLocock]

Does it also have constant contact jounce bumpers (the polyurethane foam cylinders on the shocks) -they also have an effect, as do the internal rebound springs in the shocks.

At say 3 deg/g of chassis roll gain you'll have about 40mm of jounce travel on one side and and 40mm of rebound on the other, at the wheel.

So on the rebound side you'd almost certainly be seeing the soft progressive rate of the springs.

Basically if you are trying to do it properly and analytically you need to measure these things, otherwise you are playing with numbers, which can be useful, or can just be a waste of time, depending on how good your guesses and approximations are.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[evelrod]

I tend to agree with Norm and Greg. I don't have the technical experience as Greg, but I do have a wee bit of experience racing a FWD car. Increasing front roll stiffness is not only counter intuitive but counter productive. The front does the work, the back just follows. Lifting a rear wheel is not a problem as long as the front is planted and going where it's pointed. Autocross setup is NOT the hot setup for circuit racing. From what I've seen, 50/50 to 60/40 would work for autocross. My car is 70/30 with rear anti roll bar only. The front spring rate is set to go "infinite" under certain conditions which would, obviously, not be appropriate for any other application. The big difference seems to be between the real world skid pad performance and the computer generated performance. I have yet to see the two agree to any predictable degree...at least with my limited skill level.

Rod

 

[Komodo86]

Increasing front roll stiffness is not only counter intuitive but counter productive. The front does the work, the back just follows.

That was the aim, to increase total roll stiffness but NOT to increase the front relative to the rear. I would disagree that the rear just follows though. As I see it, the rear has a significant effect on how well the front can work.

Lifting a rear wheel is not a problem as long as the front is planted and going where it's pointed.

Depends on whether you subscribe to the school of thought that when the rear wheel lifts, no further load transfer can happen at that axle and any further load transfer must happen at the front. Some people agree with this, others do not. The ones that do see lifting a wheel go hand in hand with an increase in understeer, which seems logical.

Going on the above assumption, I think that a slight lift right at the peak lateral acceleration, or under combined braking and turning is probably inconsequential, however picking that wheel up in a steady state corner at mid lateral Gs is a probably a good indicator of a wonky balance. This might be necessary in some cars to overcome other shortfalls within the design, but on the car in question with double wishbone suspension all round and a well regarded chassis, there does not appear to be much benefit, only potential disadvantages.

Autocross setup is NOT the hot setup for circuit racing. From what I've seen, 50/50 to 60/40 would work for autocross. My car is 70/30 with rear anti roll bar only.

But we are considering a car that is close to 45/55 out of the factory, what you would consider a full autocross setup, yet has almost universally praised handling on circuit. I think they key here is the cars we are comparing, assuming that is your Mini in the photo? A 70/30 setup on a Honda would create an understeering barge of a car, it would be nigh on undriveable in any kind of competitive arena. Last year I changed my setup from a 58/42 to a 43/67 to closer match the Type R specifications and the improvement in handling on both circuit and autocross was significant!

The suspension design on the two cars must account for most of the difference in the roll stiffness requirements, double wishbones must surely require a different approach to a live axle or MacPherson struts, the latter of which I am aware requires a large amount of front roll stiffness to combat the rubbish camber curve...

 

[Komodo86]

Does it also have constant contact jounce bumpers (the polyurethane foam cylinders on the shocks) -they also have an effect, as do the internal rebound springs in the shocks.

I couldn't honestly say!

At say 3 deg/g of chassis roll gain you'll have about 40mm of jounce travel on one side and and 40mm of rebound on the other, at the wheel.So on the rebound side you'd almost certainly be seeing the soft progressive rate of the springs.

This sounds about right, having different rates on either side of the car sure makes for some unpredictability!


Basically if you are trying to do it properly and analytically you need to measure these things, otherwise you are playing with numbers, which can be useful, or can just be a waste of time, depending on how good your guesses and approximations are.

Very much a case of 'garbage in, garbage out' however it does give me a insight into how various changes will potentially affect the setup. The rate of the rear spring is given in the specs available online as 2.5-4.5kg/mm. Changing between a 2.5 and 4.5 rear spring appears to alter the LTD by about 1%.