Benefits of body roll for street performance cars? 3

[ksw100]

Active technologies can essentially eliminate body roll, I'm wondering about the possible advantages of allowing some body roll in performance street cars. What are, if any, the conditions where body roll is beneficial?

Googling around, I've read that one advantage of allowing some body roll is driver feedback at the limits.

 

[cibachrome]

Define 'performance'. Ax, Ay, top speed, good ride, ...

 

[ksw100]

I should have used a better descriptor, by performance street cars, I meant cars like the 911, Cayman, Corvette, Emira etc. for street use as opposured to their track only variants. I'd say being used for spirited weekend canyon carving but not dedicated track use.

Thanks for asking!

 

[cibachrome]

Roll is methodically used to affect the transient & steady state responses of all vehicles via coupling to wheel camber & steer alterations. This is done via the kinematics & compliances designed/allowed in the front & rear suspensions. Couple of talking points:
1). In order for that to work as intended, a certain amount of body roll (per g of lateral acceleration) is prescribed. Since body roll can be uncomfortable/confusing to a driver at max lat it is well regulated by springs, bars, 'roll centers', and sprung mass & inertia. Typical values for roll-steer (steer due to roll expressed as a percent. Camber typically just degree camber per degree roll.
2). very little roll occurs from steer. This is driven by front caster, rear suspension roll stiffness and tire spring rates. Instead, roll is produced mainly by lateral acceleration (yawrate, speed, and sideslip characteristics) which is generated by steer, wind, or road irregularities).
3). Yawrate & sideslip are direct functions of speed, so their frequency responses have speed dependency (amplitude, damping, frequency @ peak). But roll is generally not speed dependent, so that's where the fun begins. If a carelessly designed vehicle has frequencies that overlap with roll, then unwanted excitation can/will occur. typically at speeds above 140 kph for cars. Much lower speed for trucks. There are a few of these around. Their reputations preceed them, but the blame is not apparent.
4). these subtle roll steer & camber effects alter the total vehicle under/oversteer, hence gains & response times. So, if your 'sporty' feeling car has low roll per g, then the couplings can be quite large in order to achieve the desired responses (usually linear & mid-range transients). You probably would be surprised by how some of the cars you listed have quite large roll steer coefficients. This results as low front weight distributions plus large, grippy, & high load capacity tires are chosen. Even these tires stop listening to steer/slip angle changes near max lat, so roll influences on tires (except for load transfer) are diminished. This means some cars turn into shit storms at max lat even though they are just so lovely driving home from the office.
5) Did I mention that the roll transient response is usually quite sluggish (as in 2 - 3 times the yawrate response) so it can 'feel' kinda funny getting into a steady state turn.
6) Your 'sporty' cars usually have very low understeer, and even with high stiffness tires, the Ay transient response time would be quite long (above 0.35 seconds) without the added roll steer, especially rear roll steer.
7) Since vehicles usually have payload variations (passengers, fuel, cargo), suspensions (usually the rear) have load/position dependent roll steer that greatly improves the steady state & transient responses. So, your sled might go from 2% roll steer at 1 passenger, empty fuel tank, to 15% roll steer when fully loaded. That is if an on-board automatic load leveling system isn't employed.

There's a few more considerations to mention, but this ought to get you thinking....

 

[XR250]

Well an ND Miata has plenty of roll and "Miata Is Always The Answer"

 

[cibachrome]

Miata:
Front: 5%
Rear : 0.2%
Avgerage of 2 cars.

Same as BMW 530, Toyota Camry, Cadillac STS, Kia Spectra, Acura MDX, Chevy TrailBlazer, Chrysler 300... So,what ???

 

[XR250]

Well, just sayin' an ND Miata is regarded as one of the most satisfying street cars to drive and it has plenty of roll.
The OP was asking about advantages of body roll.

 

[BrianPetersen]

Look up the Citroen Xantia Activa.

Citroen doesn't use that system any longer. In a way, that's sad. In a way, it's cancellation is understandable given the system's complexity and cost.

 

[ksw100]

What is the parameter being measured here? And how is it calculated?
"Miata:
Front: 5%
Rear : 0.2%"

 

[cibachrome]

FRONT Steer due to body roll. 0.05 deg steer per degree roll (usually expressed as a percent from ancient history).
REAR Steer due to body roll. 0.002 deg steer per degree body roll.

Measured on an MTS K&C apparatus as part of routine competitive assessments at 2 passenger load. I have a database with well over 1000 vehicles in it, for this and the rest of the parameters which effect all aspects of cornering, including the tire properties (those being measured on MTS Flat-Trac equipment).

A few choice samples of roll steer, all at 2 pass loading:
911: front 14% rear 0.008%
Cayman 14.2% & 2%
Corvette -3% & 3%
Saab9.5 0 -5%
Scion 0 10%
Audi A4 0 5
A3 4 4
A5 0 2
Quatro 0 6
Kia Rio 18 8
F430 -1.5 2
Carrera 13.5 5.6
Cayenne 3.8 5

While you might be tempted to dismiss the Miata values as 'small & insignificant', the Bridgestone Turanzas on it have high cornering stiffness, hence substantial influence.

Furthermore, your car has a low front weight distribution and relatively high stiffness tires. So, to get a reasonable (i.e. legally required) amount of understeer in the car, Mazda has chosen to
use a pretty soggy steering gear/gear mount apparatus utilizing the front tires Mz stiffness. The amount (~1.15 deg/100 Nm on-center) is about the same as we see in trucks with RB steering gears.
Typical values for on-center steering stiffness of a car rated high for on-center feel are on the order of 0.50 to 0.60 deg steer per 100 Nm per tire.

The issues with this strategy are that tire Mz is non-linear with slip angle and peaks before tire Fy does. This results in a false sensation in the steering wheel that you've reached max Ay. Caster can be rigged to try and fix this, but adds more load to the steering gear in a viscious cycle. Also, at the real Fy limit, tire Mz is probably negative, meaning the steering wheel will have lost it's returnability sensation and precipitate an oversteering tendency. Tire Mz is very wear dependent, so steering effort/feel ranges are different with tire mileage/wear/pressure characteristics.

 

[ksw100]

A few questions:
1. Is there a desireable roll steer range? Or is this an area where the character of the car is purposefully designed? For instance: rear steer, the designer may want to add some roll steer for stability versus agility.

2. What do the positive and negative values indicate? For examplem, does a positive number mean that it makes the turning radius bigger or smaller during roll?

3. Have you found that the roll steer of the inner and outer wheels are pretty symmetric? And that is why only 2 numbers are presented (front and rear) but not all 4 corners of the car? Or are the pairs of numbers presented the average of the inner-front and inner-rear roll steer numbers?

 

[cibachrome]

1) Just mu opinion, but for your 'sports' cars, roll steer is a bandaid to fix deficiencies caused by adherence to lore about weight distribution, tire sizes, total roll stiffness, etc. It's the worst one because the car rolls by convolution: You steer to get Ay, then the Ay activates the roll. So it's late, OR you need a high roll frequency to help it stay in touch with the other degrees of freedom (yawvelocity & sideslip. But REAR roll steer also serves to help with large changes in payload. The suspension compresses, the linkages or structural members move to a higher roll understeer position and the linear range dynamics are reasonably maintained.
2) The traditional convention is + for an understeer effect, - for an oversteering effect. Since the rear is always oversteering, adding roll (under) steer lessens the drama within it's ability to still affect tire force generation. Can't do much for you though at the limit because just about the only things tires listen to are camber and Fz load. Understeer makes the turn radius bigger (Curvature lower).
But it also increases the natural frequency at the expense of damping, so response times are shorter (bandwidth higher). The goal is usually to get the Ay response times closer to driver perception abilities, usually about 0.32 to 0.34 seconds. Otherwise, a sluggish response begs for additional input as the driver is lead to believe that it's not responding enough. A Neutral Steer car has the slowest response times and is overdamped in yaw velocity, but is able to generate the highest Ay if done right. Response times don't mean that much in racing because it's not a wise idea to thrash the car around with steering. "Smooth is Fast".
3) its almost always symmetric for several reasons: same parts, mirrored parts, and the desire to give the car turning with the same reactions in both directions. Oval track cars are different. The car will use a force balance to utilize the left & right steer effects, not steer displacements. Besides, we're not talking huge amounts of steer here. A 5 degree per g roll gradient in a car with 10% front roll steer creates 0.5 degrees of steer per wheel at 1.0g. At, say 1500 N/deg per tire, that can be a pretty large force to contend with if you scale it back to say 0.25g typical racing Grandma ability or nerves. When I follow a 'sports' car around an expressway exit in a shitbox econocar, I often pull maybe 0.70g's while they struggle with 0.30gs. BTW; I have an app on my phone (Physics Toolbox) which can sort thru most of this because it has all three accelerations, speed, yawrate, roll rate, pitch rate and the angles. It sends me an email with all this data vs. time. I wrote a Matlab application to turn this data into engineering metrics. I even did a constant steer increasing speed test on my golf cart and my bass boat just to show how easily it can be done. I've shown the results on a few of the Vehicle Dynamics forums if you are interested...

 

[ksw100]

Thank you for the explanations, it will take me time to digest.
In the example numbers you gave for the various cars, I see that the 911s and Cayman all have very high front roll steer numbers. Is that because they have a tendency to oversteer and the large front roll steer is designed in as a bandaid (which induces understeer?)

 

[GregLocock]

The disadvantages of excessive roll steer are (1) makes the car darty on rough surfaces - in Australia we expect to drive comfortably at 100 (or more) kph on gravel roads, including dropping one side off the track to allow an oncoming car through. Not all manufacturers quite managed that. (2) Some whingeing about tire wear (unproven in my opinion, may be lore left over from cross ply tires) (3) As cibachrome says, roll/yaw coupling aka threepenny bit cornering. You move the steering wheel to enter a turn, the car's ay builds up, the roll builds up, and the car steers in response so you have to correct the steering.

There are good subtle solutions to this, but I'm not telling.

As you can see from the numbers posted, whatever the optimum is, there is no agreement.

On rolly trucks where we are fighting for understeer we dial a lot of rollsteer in. I can't think of a single program I did where we gained any understeer during the program without taking 'special actions', which cause grumpiness.



Cheers

Greg Locock


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

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

 

[XR250]

That machine looks fascinating. How do the computer models typically compare to the test rig?

 

[GregLocock]

Which machine? K&C?



Cheers

Greg Locock


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

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

 

[XR250]

Yes

 

[GregLocock]

Kinematics is usually pretty good, if you've CMMed the hardpoints properly. This is pretty crucial, typical production tolerances on the height of theUCA and rack compared with the lower arm bushes, which controls your toe curve, is enough to mess things up.

Compliances not so good, we measure the bushes compliance curves in 6 directions independently, and so when installed the preloads in other axes aren't accounted for. There is also the structure stiffness, for example on a typical tall spindle SLA the spindle bends of the order of 1 deg per kN of lateral force, in camber. You can get around that side of things by using flex bodies, basically FEA, of the structure.


Cheers

Greg Locock


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

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

 

[XR250]

Interesting - thanks

 

[cibachrome]

Adding the chassis structural compliance to the mix via the high contribution Normal Modes gets you pretty darn good correlation and satisfactory functions of load. This includes hysteresis.
This was done using NASTRAN at the time to do a free-free normal modes of the dressed chassis, sorting the mode numbers/frequencies/damping by order of influence on suspension attachment points produces frequency regions of high participation. Note that these modes can be added to stick models producing algebraicly solved influences so no more F.E.M. solutions were needed. I wrote a paper back in '77, 3 years after I began working on it. Been done that way ever since. The entire scheme was developed for ride and durability modeling. Plus you could throw in real braces, springs into any of the nodes (Like cross tower braces for example). An SAE paper written on this too. (Curt Vail, et. al.) So, a 25,000 degree of freedom chassis model, might have only a few hundred modes participating in each of the ride, roll, lateral, aligning moment compliances. A huge twist axle with 50 modes was a blink of an eye turnaround job on TSO, vs. an overnight job on a mainframe.

It was eye opening to bring in race cars for K&C testing and expecting a solid brick to hold the suspension bits. In fact, the body stiffness was often the major player in just about all d.o.f.s Big, heavy, stiff suspensions were utilized to get needed stiffness because the bodies were the weaklings in the mix. Track bar brackets, steering gear mounts, engine bay, windshield, safety cages, all needed a lot more attention. Plus, you could put a 2g Inertia Relief load to the car on the K&C rig and not fear breaking stuff because the measured & simulated strains correlation looked so good.