Suspension Analysis: Roll Stiffness & Ride Frequency
#31
Rennlist Member
Years ago I made a spreadsheet to try to calculate some of this. It's been a while since I played with it... But looking at it now, here are some of the values:
This is my orange car - in SP2 class. I have 30mm torsion bars in it, but you can see what the equivalent spring rate would be for the coil-overs.
*Note* these are all calculations based on the best measurements/guesses I could make at the time... It doesn't mean that they're fact.
This is my orange car - in SP2 class. I have 30mm torsion bars in it, but you can see what the equivalent spring rate would be for the coil-overs.
*Note* these are all calculations based on the best measurements/guesses I could make at the time... It doesn't mean that they're fact.
#32
Arominus, from your setup I calculate
Roll Stiffness's of:
Front = 1507 lb-ft/deg
Rear = 1494 lb-ft/deg (stiffest bar position)
Total = 3001 lb-ft/deg
Roll Couple Distribution = 50.2 % front
Based on the 225/255 F/R tire width the "Grip" Distribution = 45.9 % front, I am hypothesizing that this 5% difference is a big contributor to your understeer, as you probably suspected by suggesting a move to 27 mm rear torsion bars.
Roll Stiffness if you increase rear Torsion bars to 27 mm:
Front = 1507 lb-ft/deg
Rear = 1691 lb-ft/deg (stiffest bar position)
Total = 3198 lb-ft/deg (+6.5 %)
Roll Couple Distribution = 47.1 % front
This would be a good step to reduce some of the understeer without likely creating and based on Speedy's car it sounds like 200 #/27 mm is a good combo.
I think another option for you might be to swap out the 30 mm front bar for one of the 26.8 mm bars.
Roll Stiffness if you swap to 26.8 mm front anti-roll bar:
Front = 1001 lb-ft/deg
Rear = 1199 lb-ft/deg (middle bar position)
Total = 2200 lb-ft/deg (-26.7 %)
Roll Couple Distribution = 45.5 % front
If my assumptions and measurements are right this option would provide a close to neutral car with the ability to move the rear bar either stiffer or softer to create more oversteer or understeer.
A lot of this depends on some of the variables in how the suspension works, which with more input from others on how their setups respond we could have a better of understanding of the suitable ranges of Roll Couple Distribution vs Approximate "Grip" Distribution to predict handling balance... at least that is what I am working towards.
Roll Stiffness's of:
Front = 1507 lb-ft/deg
Rear = 1494 lb-ft/deg (stiffest bar position)
Total = 3001 lb-ft/deg
Roll Couple Distribution = 50.2 % front
Based on the 225/255 F/R tire width the "Grip" Distribution = 45.9 % front, I am hypothesizing that this 5% difference is a big contributor to your understeer, as you probably suspected by suggesting a move to 27 mm rear torsion bars.
Roll Stiffness if you increase rear Torsion bars to 27 mm:
Front = 1507 lb-ft/deg
Rear = 1691 lb-ft/deg (stiffest bar position)
Total = 3198 lb-ft/deg (+6.5 %)
Roll Couple Distribution = 47.1 % front
This would be a good step to reduce some of the understeer without likely creating and based on Speedy's car it sounds like 200 #/27 mm is a good combo.
I think another option for you might be to swap out the 30 mm front bar for one of the 26.8 mm bars.
Roll Stiffness if you swap to 26.8 mm front anti-roll bar:
Front = 1001 lb-ft/deg
Rear = 1199 lb-ft/deg (middle bar position)
Total = 2200 lb-ft/deg (-26.7 %)
Roll Couple Distribution = 45.5 % front
If my assumptions and measurements are right this option would provide a close to neutral car with the ability to move the rear bar either stiffer or softer to create more oversteer or understeer.
A lot of this depends on some of the variables in how the suspension works, which with more input from others on how their setups respond we could have a better of understanding of the suitable ranges of Roll Couple Distribution vs Approximate "Grip" Distribution to predict handling balance... at least that is what I am working towards.
Dude, thank you! i don't have the 26.8mm front bar anymore but i have a 25.5 hanging around, how would that look? i'm tempted to throw it on.
#33
Advanced
Thread Starter
Join Date: Jan 2014
Location: Greater Atlanta, GA
Posts: 69
Likes: 0
Received 0 Likes
on
0 Posts
I realized after writing the previous post that I had made an error in my spreadsheet for the 30mm front roll bar, giving it the wrong motion ratio. This means when I calculated the numbers for Dimi's and Arominus's car
Arominus your corrected Roll Stiffness's are:
Front = 1278 lb-ft/deg
Rear = 1494 lb-ft/deg (stiffest bar position)
Total = 2772 lb-ft/deg
Roll Couple Distribution = 46.1 % front
This roll couple distribution is very close to the "Grip" distribution of 45.9 % front. Though the reported issues of some understeer and lift-oversteer might be related to other setup variables which have influence over the transient behaviors. That being said the 27 mm torsion bars could still be a fair change, creating a 47.8 % front roll couple distribution with the rear bar on the middle position.
Arominus your corrected Roll Stiffness's are:
Front = 1278 lb-ft/deg
Rear = 1494 lb-ft/deg (stiffest bar position)
Total = 2772 lb-ft/deg
Roll Couple Distribution = 46.1 % front
This roll couple distribution is very close to the "Grip" distribution of 45.9 % front. Though the reported issues of some understeer and lift-oversteer might be related to other setup variables which have influence over the transient behaviors. That being said the 27 mm torsion bars could still be a fair change, creating a 47.8 % front roll couple distribution with the rear bar on the middle position.
#34
Advanced
Thread Starter
Join Date: Jan 2014
Location: Greater Atlanta, GA
Posts: 69
Likes: 0
Received 0 Likes
on
0 Posts
Years ago I made a spreadsheet to try to calculate some of this. It's been a while since I played with it... But looking at it now, here are some of the values:
This is my orange car - in SP2 class. I have 30mm torsion bars in it, but you can see what the equivalent spring rate would be for the coil-overs.
*Note* these are all calculations based on the best measurements/guesses I could make at the time... It doesn't mean that they're fact.
This is my orange car - in SP2 class. I have 30mm torsion bars in it, but you can see what the equivalent spring rate would be for the coil-overs.
*Note* these are all calculations based on the best measurements/guesses I could make at the time... It doesn't mean that they're fact.
#35
Advanced
Thread Starter
Join Date: Jan 2014
Location: Greater Atlanta, GA
Posts: 69
Likes: 0
Received 0 Likes
on
0 Posts
I have done skidpad testing in a formula car to validate that the roll center heights do not affect "quasi-steady-state" handling balance. In "quasi-steady-state" all the weight has transferred at this point and so it is just the level lateral force the tires are generating to keep the car running in a circle on the skidpad. The final weight transfer distribution is proportional to the roll couple distribution. So if one axle has all the roll stiffness it will have all the weight transfer (not likely to happen but good visual); if this hypothetical car had the same tires at each corner and had equal static weight on each tire then in a corner the axle without any roll stiffness will have the same weight on each tire while the axle with all the roll stiffness will have one heavy laden tire and one unladen tire. This is important because tires have what is called tire load sensitivity which means that if we double the load on the tire we don't get double the lateral grip out of it, we get less. Meaning the axle with all the roll stiffness loses lateral force capability the more the weight gets transferred; and the less grip that axle creates. This is why I believe that you want to aim to have the roll couple distribution within a few percent of the "grip" distribution if you want neutral at limit handling balance in mid-corner or "quasi-steady-state".
With all of that in mind, you are correct that the roll center hieghts can have a profound effect on the transient handling characteristics. This is because the roll center heights greatest effect is on the weight transfer RATE. Since we know weight transfer is proportional to lateral grip level, the end of the car that transfers weight faster will be perceived as having more grip during the transient situation. So raising or lowering the roll centers at either end can help with transient handling balance issues. My preference from doing the skidpad testing in the formula car is to keep the roll centers low to keep the chassis reactions more readable by the driver (human reaction time limits) but high enough to help keep the body roll in the desired range without needing excessive springs or anti-roll bars. On the formula car with a 12" CG height, my preference on roll center height was about 1-2".
This may disagree with what some books say as it is based on my observations studying vehicle dynamics on a formula car. YMMV
#36
Rennlist Member
I'm using the Tarret adjustable bars - that's why the drop link angle comes into play.
#37
Rennlist Member
#38
Advanced
Thread Starter
Join Date: Jan 2014
Location: Greater Atlanta, GA
Posts: 69
Likes: 0
Received 0 Likes
on
0 Posts
I really don't know if I'm doing that calculation correctly... But I'm taking half the length of the sway bar (because, in a static view, the sway bar seems to push one side up and one side equally down) and the modulus of rigidity to calculate the torsional twist in lbs/degree (to try and get a spring rate equivalent to match a conventional coil spring). And then the wheel rate is based on the moment of the lever arm and the angle of the drop link and the motion ratio (where the sway bar connection is on the control arm).
I'm using the Tarret adjustable bars - that's why the drop link angle comes into play.
I'm using the Tarret adjustable bars - that's why the drop link angle comes into play.
#39
Three Wheelin'
Mike, you didn't miss it. I have not used roll center heights in any of the calculations. Roll center heights do not affect the roll stiffness, they do affect how much the body will roll but not the roll stiffness nor the weight transfer.
I have done skidpad testing in a formula car to validate that the roll center heights do not affect "quasi-steady-state" handling balance. In "quasi-steady-state" all the weight has transferred at this point and so it is just the level lateral force the tires are generating to keep the car running in a circle on the skidpad. The final weight transfer distribution is proportional to the roll couple distribution. So if one axle has all the roll stiffness it will have all the weight transfer (not likely to happen but good visual); if this hypothetical car had the same tires at each corner and had equal static weight on each tire then in a corner the axle without any roll stiffness will have the same weight on each tire while the axle with all the roll stiffness will have one heavy laden tire and one unladen tire. This is important because tires have what is called tire load sensitivity which means that if we double the load on the tire we don't get double the lateral grip out of it, we get less. Meaning the axle with all the roll stiffness loses lateral force capability the more the weight gets transferred; and the less grip that axle creates. This is why I believe that you want to aim to have the roll couple distribution within a few percent of the "grip" distribution if you want neutral at limit handling balance in mid-corner or "quasi-steady-state".
With all of that in mind, you are correct that the roll center hieghts can have a profound effect on the transient handling characteristics. This is because the roll center heights greatest effect is on the weight transfer RATE. Since we know weight transfer is proportional to lateral grip level, the end of the car that transfers weight faster will be perceived as having more grip during the transient situation. So raising or lowering the roll centers at either end can help with transient handling balance issues. My preference from doing the skidpad testing in the formula car is to keep the roll centers low to keep the chassis reactions more readable by the driver (human reaction time limits) but high enough to help keep the body roll in the desired range without needing excessive springs or anti-roll bars. On the formula car with a 12" CG height, my preference on roll center height was about 1-2".
This may disagree with what some books say as it is based on my observations studying vehicle dynamics on a formula car. YMMV
I have done skidpad testing in a formula car to validate that the roll center heights do not affect "quasi-steady-state" handling balance. In "quasi-steady-state" all the weight has transferred at this point and so it is just the level lateral force the tires are generating to keep the car running in a circle on the skidpad. The final weight transfer distribution is proportional to the roll couple distribution. So if one axle has all the roll stiffness it will have all the weight transfer (not likely to happen but good visual); if this hypothetical car had the same tires at each corner and had equal static weight on each tire then in a corner the axle without any roll stiffness will have the same weight on each tire while the axle with all the roll stiffness will have one heavy laden tire and one unladen tire. This is important because tires have what is called tire load sensitivity which means that if we double the load on the tire we don't get double the lateral grip out of it, we get less. Meaning the axle with all the roll stiffness loses lateral force capability the more the weight gets transferred; and the less grip that axle creates. This is why I believe that you want to aim to have the roll couple distribution within a few percent of the "grip" distribution if you want neutral at limit handling balance in mid-corner or "quasi-steady-state".
With all of that in mind, you are correct that the roll center hieghts can have a profound effect on the transient handling characteristics. This is because the roll center heights greatest effect is on the weight transfer RATE. Since we know weight transfer is proportional to lateral grip level, the end of the car that transfers weight faster will be perceived as having more grip during the transient situation. So raising or lowering the roll centers at either end can help with transient handling balance issues. My preference from doing the skidpad testing in the formula car is to keep the roll centers low to keep the chassis reactions more readable by the driver (human reaction time limits) but high enough to help keep the body roll in the desired range without needing excessive springs or anti-roll bars. On the formula car with a 12" CG height, my preference on roll center height was about 1-2".
This may disagree with what some books say as it is based on my observations studying vehicle dynamics on a formula car. YMMV
I understand all of what you have said above, and good to see you have backed it up with real-world experience! Great work.
I'm still going to challenge you on the roll centre assumptions. Roll centre force transfer is instantaneous, however that doesn't make it transient (unlike shocks, which are transient).
And if roll centre height affects final roll angle then surely it has affected roll stiffness? Roll stiffness being measured in degrees per g.
I wrote a spreadsheet about 20 years ago which included RC height, based on the formulae from "the Milliken".
Quite enlightening, I'd share a copy of the spreadsheet with you if I still had it. It gave vertical tyre loads at each wheel, under any combination of forward and lateral acceleration (yep also incorporated antidive/antisquat)
Happy to continue this discussion offline if you like?
Cheers,
Mike
#40
Rennlist Member
Van, I think your measurements look good. What confuses me is how a bar that stiff would only contribute 54 lb/in to the wheel rate. When I calculate its stiffness, assuming pure roll and using your lever arm and my bar length (I measured 41.25" from end to end where it matches up to the control arm) I get that it provides 473 lb-ft/deg to chassis roll stiffness. From your provided wheel rate from the spring (331 lb/in) I calculate a roll stiffness of 815 lb-ft/deg. So I would expect that you have a wheel rate of 192 lb/in contribution from the Tarret roll bar.
Also, that bar I'm using is smaller than the stock Tarret bar, so don't be fooled by that.
I also don't know if it's right to calculate the torsional force from the center of the bar... If that calculation should be over the whole length of the bar, then the spring rate value will go down by half...
#41
Three Wheelin'
I was thinking about this last night, and I think I have forgotten a major piece of the equation. My values are really in lbs/degree rotation... but somehow the "inch" part got in there, too... I think we have to figure out how many degrees of rotation it would take for the drop link to travel 1" - it might be a few degrees, which would make that value higher.
Also, that bar I'm using is smaller than the stock Tarret bar, so don't be fooled by that.
I also don't know if it's right to calculate the torsional force from the center of the bar... If that calculation should be over the whole length of the bar, then the spring rate value will go down by half...
Also, that bar I'm using is smaller than the stock Tarret bar, so don't be fooled by that.
I also don't know if it's right to calculate the torsional force from the center of the bar... If that calculation should be over the whole length of the bar, then the spring rate value will go down by half...
Cheers,
Mike
#42
Rennlist Member
Dimi what are you using on the rear? I'm thinking of trying the 30 series Konis with stock TBs and would appreciate your opinion?