Steering torque

[David Braga]

Hi everybody,
I am working on steering module in our FSAE team.
1)what factors affect the steering torque ?
2) what is the optimum torque range required for designing steering system with good feedback at steering wheel ?
3) How should i go about calculating this torque ...i would like to know how to calculate lateral force and vertical force causing moment ..

[PatClarke]

A word of advice David.

Put your armour on and DUCK!!!

Pat

[BillCobb]

You forgot to ask what caster setting to use...

As a matter of fact, I never did get any bites on a procedure to determine a caster setting on these cars, just some flamage from an old school mechanic who didn't know.

[woodsy96]

Aaaaaaand......bite. http://fsae.com/groupee_common/emoticons/icon_smile.gif

-Determine steering geometry for Ackermann effects
-Determine "ideal" camber for inside and outside wheel for different steered angles (for a series of corner radii)
-Make some approximate stiffness values for camber compliance from lateral load.
-Determine what KPI, caster and static camber settings to use as baseline to achieve "ideal" camber for different corner radii. More emphasis on higher corner radii (i.e. less tight corners) as turn in is more important than mid corner.
-Consider effects on braking from static camber values. Then do the process all over again.

Comments/ advances?

[BillCobb]

The initial caster setting is easily computed via simulation and the nonlinear tire data provided by the TTC. The model must include sufficient relationships to generate the net tie rod load and lateral acceleration. As Z points out, the trail part of this is important, but trail generation has a practical side.

Given tierods loads and lateral acceleration at the driver's head level, a plot of tierod load gradient (TRLG) (partial TRL by partial Ayh) vs Ay will usually show a peak (max) value long before a car reaches its max (limit) Ay. The net steering wheel torque gradient will also peak out well before max lat, so a driver will shy away from higher g's because the steering wheel torque gradient (STG) will be negative after that point. To improve this situation by making the g-level of the peak STG align with the max-lat, you can easily add some trail.

Now this can be done in several ways: lower air pressure (bad for max lat), you can add caster offset (shift the spindle off the steer axis), or change the caster angle. While caster offset provides a neat way to do the job, its expensive to miss the mark: You need new knuckles. And, you miss some of the other benefits: If you have a forward bite problem, you can cause rear dynamic load transfer by steering the front suspension. Steering forces will see a slight increase, too because of the rear suspension's roll stiffness. But, having the ability to drastically change the car at the track by moving a few cams or shifting some shims is a huge tool. Obviously the down side to adding tierod loads is the increase in understeer caused by steering system deflection(s) and from the increase in the rigid body aligning moments produced at the front wheels. (i.e. the additional aligning moments affect the whole car as a rigid body).

A major snafu in understanding all of this is the effect that mechanizarion of caster angle has on the ride/roll steer of the front suspension. Increasing the caster angle drops the outer tierod ball quite significantly. This adds an oversteering tendency (lowers the value of roll by steer gradient). So you can easily get a very mixed reaction to the car's behavior because of the muddy waters of caster change.

I've never been convinced of the merits of caster effect on camber because the steer angles are so small for a 'good' car and the camber stiffnesses of 'racing radials' are also relatively tiny. The product of the two is a gee-whizz.

Ever wonder what they mean by having a car good right off the hauler? The geometry settings are hard coded.

As for kingpin inclination. that's great for making a car drive straight because of the potential well the unsprung mass falls into. The sacrifice in steering feel because of the increased suspension sideloads (added friction) makes this a troublesome parameter. The new hardware needed to make a change at the track can be expensive or failure prone, too.

If scrub radius control during max lat or max lon is an objective, get the tire data and design to it.

As far as using different caster angles or offsets on certain race cars, they have very different tire constructions, pressures and peak properties on each side, so specific casters intended to give high coherent feel to the driver is necessary and appreciated and productive. Without all the data and an appropriate simulation, its a s.w.a.g. .

[flavorPacket]

That's all well and true, Bill (as per usual), but all this theory is going to run straight into the wheel rim or to an illegal frame design on most FSAE cars. Packaging will drive you to an ugly compromise well before you get even 3 of these parameters to your liking.

So in other words I agree 100% with one caveat:

Develop an acceptable volume in which you can package the balljoints, brakes, and tierod, and THEN start the calcs.

[BillCobb]

Well I know all too well that one auto company still tries to do the package first and then find a tire that meets the front and rear cornering compliances necessary to reach some altruistic goals (like getting a Consumer Reports recommended buy rating). But they keep failing because they need tires that can't be made, don't live/last, or cost too much to make. I submit that the tires should go on as soon as the motor is selected (IMHO of course).

In the case of FSAE, there is a reason that the tire selection choice list is what it is: These tires can be made to work nicely within the size, wheel and pressure envelopes recommended. And, the analyticly chosen alignments are usually PFC to those determined experimentally.

The advantage is time saved for other design issues and the confidence, satisfaction, and glory of doing it using professional engineering tactics instead of cut and try methods. Those folks never know when they are done. That's why you still hear them say that this year's new model "has a refined suspension". Yeah, that;s because they ran out of time, missed the objectives and used a committee to choose the suspension parameter settings instead of the cold hard facts.

You should be made aware that HMS has elevated a good friend of mine to the Simulation Director position. Look on Linkedin find out who. Then consider the team's performance...

[flavorPacket]

Originally posted by BillCobb:
These tires can be made to work nicely within the size, wheel and pressure envelopes recommended. And, the analyticly chosen alignments are usually PFC to those determined experimentally.

I wasn't talking about the tire, surely by now most FSAE teams choose tires before wheels (a luxury we rarely have the production world!). I was referring specifically to the suspension architecture required to accommodate the steering geometry that such a design approach would suggest.

Maybe on the 13" Goodyear it's possible to meet all the analytic targets, but not on the Hoosier LC0 with any of the frame rules in place from 2009 onwards. Keep in mind the LC0 was never intended for these cars, it just happens to fit, weigh nothing, and last just a hair longer than an endurance.

[woodsy96]

Given tierods loads and lateral acceleration at the driver's head level, a plot of tierod load gradient (TRLG) (partial TRL by partial Ayh) vs Ay will usually show a peak (max) value long before a car reaches its max (limit) Ay. The net steering wheel torque gradient will also peak out well before max lat, so a driver will shy away from higher g's because the steering wheel torque gradient (STG) will be negative after that point. To improve this situation by making the g-level of the peak STG align with the max-lat, you can easily add some trail.

I like this. Never really occurred to me in my 4 years of FSAE.

[Zac]

Originally posted by BillCobb:
Well I know all too well that one auto company still tries to do the package first and then find a tire that meets the front and rear cornering compliances necessary to reach some altruistic goals (like getting a Consumer Reports recommended buy rating). But they keep failing because they need tires that can't be made, don't live/last, or cost too much to make.


Being on the other side of this is pretty awful. What is rough is when you develop a tire that actually meets a tough performance target (say an EMT that doesn't ride like a brick) only to get kicked off the program because of a couple dollar price difference. The icing on the cake is when you get a phone call a few weeks later from the customer asking for tips on fixing your competitors product.

[BillCobb]

And you may have heard of my "Premium Tire Supplier" proposal, eh? Took my lumps on that one (and also took my Lump Sum and gave myself a healthy raise)!

[Zac]

I've seen similar proposals from several OEMs. They have all met the same fate.

[UPS]

A word of advice David.

Put your armour on and DUCK!!!

Pat

The trick is ... There is no trick!

[CWA]

Folks I'm very sorry to bump such an old thread (it is Nov 2014 now, last post Sept 2012) but my internet searches regarding my question below have lead me here, the only place I seem to be able to find anything that touches on what I am after.

"As for kingpin inclination. that's great for making a car drive straight because of the potential well the unsprung mass falls into. The sacrifice in steering feel because of the increased suspension sideloads (added friction) makes this a troublesome parameter. The new hardware needed to make a change at the track can be expensive or failure prone, too."

Bill, could you explain the "potential well the unsprung mass falls into" statement, I am afraid I don't quite understand this mechanism. I have been trying to understand how KPI affects on-centre steering response for a while now, but still can't quite get my head around it.

"Obviously the down side to adding tierod loads is the increase in understeer caused by steering system deflection(s) and from the increase in the rigid body aligning moments produced at the front wheels. (i.e. the additional aligning moments affect the whole car as a rigid body)."

Also, the above is unrelated, but I am still curious. Bill, when you have mentioned an increase in rigid body aligning moments, what are you saying might cause this increase? An increase in trail? If so, I can't currently picture this mechanism either, would you be able to clarify?

Thanks, Chris.
Coventry, UK

[BillCobb]

Go out into the Coventry Side and find an old antique Meccano Set. (We had Erector Sets here in the 'States).

Then build yourself a front axle suspension model which contains rudimentary suspension architecture. Given some absurd KPI just for the sake of exaggerating, you should come to realize that the projected spindle elevation rises and falls as you steer it. Because of gravity, (thanks to the Higgs boson, either discovered or still waiting), the net effect is to raise and lower the whole front end because the tires and wheels like to stay put. Thusly, the student is asked to show that there then forewith and forevermore is, a preferred position of the steered knuckle. This positional preference in the vicinity of the gravitational potential is a 'potential well' in the vernacular of Newtonian and Relativistic physics. (As in the statistical position of an electron in a hydrogen atom). Pardon me for the analogy, I also have a minor in, you guessed it, Nuclear Physics and can prove it: I have a Nuclear Bomb Effects circular slide rule that was shown being used in the Dr. Strangelove movie.

OK, so unfortunately there are additional surface topological features in this 'well' that disturb this self centering mechanism. Things like caster, wheel offsets, overturning moments, camber alignment and a big one: non-zero tire force and moment offsets (conicity, ply steer and ply-rat). Some teams stumble around and find a best/acceptable solution. Others deal with it. Manufacturers tend to specify tire property conditions to avoid problems that these factors introduce (wander, leads and pulls, headlight aim, and dog-tracking: rear wheels don't follow in the front wheel path).

As far as the rigid body moments are concerned, it's back to vehicle dynamics basics. A rear weight biased car on the same tires and pressure front and rear is most certainly oversteering by definition because tires are softening springs. (They're not really like springs, moreso like softening dampers but that analogy makes the uneducated fall asleep). Because a vehicle in closed loop control is governed by a moment balance, the tire aligning moments resulting from slip angle generation act not only on the axle but on the whole vehicle. for example: a real vehicle at 50% weight distribution same tires set to 90 psi will have very little tire aligning moments and could measure almost 0.0 understeer by SAE procedures if the steerin' system is tight (That's a Bill Milliken phrase I can still hear him say). BUT, set the pressures back down and the understeer will jump to 1.5 deg/g. From K&C and tire data, we know that there will be additional understeer added because of the steering system compliances, etc. so it make actually come out to about 2 - 2.5 deg/g. A good simulation can break this down for you finely (not finally) in terms of the Cornerin' Compliances (That's also Bill).

So, in order to keep this vehicle stable in closed loop control, you need to get the total vehicle understeer above the Ackerman Gradient for the speed you expect to run at, otherwise its a major STB handful. So, teams add roll steer and camber, steering system deflection or split tires and or tire pressures. Some do this by design, some have it happen by accident, some are in total denial.

Now, adding understeer from tire aligning moment deflections increases tire slip angles which compound the net aligning moments and it 'snowballs' . (Back to Nuclear: there is a chain reaction). More deflection and more understeer. Worse if power assisted steering because you have a torque sensor that must deflect to read the net tierod load. Lastly, tire aligning moments peak out after about 3-4 degrees slip, so in the famous words or Terry Satchell: it's Titey-Loose.

Still cloudy? Try to run a FSA car up to 70 mph and make a moderate to severe lane change. Use your No2 Driver for this. Next best thing and save the car ? Run a constant radius test in a prking lot and measure the Tangent Speeds from the pressure change menu.

In spite of the fact that your tires say 'Racing' somewhere on them and you have chrome-moly chassis parts, The Cornering Compliances of these cars are still not much better than my golf cart's (It's a Yamaha). How do we know this ? (We can MEASURE them !) Remember, I measured the Understeer of my Bass Boat.

[BillCobb]

Here ya go Matey:

[DougMilliken]

Hi Bill,

Thanks for the memories! Tom Bundorf wrote a nice paper on cornering compliances and understeer budget, SAE #670078.

For the acronym-challenged, I believe there is one missing letter here?
tire force and moment offsets (conicity, ply steer and ply-rat).
tire force and moment offsets (conicity, ply steer and ply-rsat). [residual self aligning torque]

Next question -- why did the Chevy Sonic that I rented from Hertz give me fits when I first started driving it? After 5 minutes I adapted, but it was awful at first on a twisty freeway, hard to keep centered in a lane. My guess (only a guess, I have no data) is a combination of:
* crappy emulation of hydraulic power steering (this car has EPS)
* a soft rack mount adding even more aligning torque compliance steer
* high ball joint friction so the steering has some break-out torque
Almost like driving an ancient truck with a couple of inches (at the steering wheel rim) of slop in the steering...

-- Doug

[BillCobb]

Who knows, its a rental ! Cold weather (Morning Sickness), friction, Euro Tuning (Sonic is a Opel clone, eh?) EPS needs 48V in order to be competitive with hps imho.

Shouldn't you be driving a client's car ?)

We always called it ply-rat. And it was aligning torque, not moment, too. Self serving...

BTW: How's the snow testing in Buffalo going ? I saw Bundorf at a GM Vehicle Dynamics department reunion a few months ago.

And finally, would you believe you can extend the range of a Nissan Leaf with EPS by continually making fast lane changes ?

[DougMilliken]

This trip was for a lecture to an SAE student section--their FSAE car (client car?) didn't have room for luggage<grin>.

The big lake effect snow was on the other side of town--we are north side, central and south of Buffalo was hit (this time). A friend outside Lancaster, NY had five feet of snow the first day, with drifts well over his head. We had eight inches over several days.

Yep, Bundorf told me he enjoyed the reunion. He put together the two slide shows linked at the bottom of page, http://www.millikenresearch.com/olley.html

Is that extended range with the Leaf measured along the centerline of the road...or along the serpentine path?

[CWA]

Thanks Mr. Cobb. The self-centring mechanism you have described aligns with other descriptions I have read previously. If I understand correctly, this mechanism contributes to a purely steer-displacement-dependent self-centring steer torque. Self-centring torque can be generated with steer displacement without vehicle speed/lat acc (ignoring tyre CP rubber twist effects).

Would this steer torque generation be considered as different to the self-centring steer torque created purely by trail, which is dependent on lat acc? If so, in your experience, how should the trade-off be balanced, do you want to prioritise one mechanism over the other, does this depend on application greatly, or not so much? Do drivers classically prefer one mechanism over the other?

I had not considered the compound effect of front slip angle increase and front tyre aligning moment increase, but I can now picture it, thanks for describing this. I suppose this is considered as a different, separate mechanism to 'an increase in USG due to an increase steering system compliance due to an increase in trail'. Perhaps this is why I became confused.

Also, for some reason, this discussion has reminded me of Evan Boberg's book. It's rather off topic, but I would love to know what views either yourself Bill, or Doug have on the book, if you have read it?

[DougMilliken]

Haven't read Boberg. Is this review at all accurate?
http://www.amazon.com/review/R11TO6AFVPUVPY/ref=cm_cr_dp_title?ie=UTF8&ASIN=1414040776&channel=detail-glance&nodeID=283155&store=books
Other Amazon reviews, http://www.amazon.com/Common-Sense-Not-Required-Designing/dp/1414040776

[CWA]

Thanks for responding Doug. I can certainly understand the point of view of the chap who wrote that review, but there were more interesting parts of the book (for me, at least) than I think the review gives credit for.

Although perhaps the book does sound like total nonsense to those with industry experience; I suppose I am curious as to whether his views on industry are mirrored by any others in any way.

[BillCobb]

Quick answers, then theory.

Yes the KP steering axis moments are different from caster, etc. but the other mechanisms (caster, strut coil spring rates, tire scrub torque and direct acting anti-roll bars are in many situations all present. Each has its own deficiencies.

The recipe for fractional assignment depends on the mission: racing, speed range of operation, powertrain weight and horsepower, driver skill and strength, and turn radii expected.

For example a car who's journeys are mostly straight will demand good road feel, high gain, light effort and 'good' ride (whatever in Hell that means to people). A car who's mission is running fast in a city street maize will need to have nimble maneuvering above average cornering linearity and good returnability.

A FSAE car probably needs low speed maneuverability, +- 100 degrees of steering wheel angular range max, rim forces below wrist sprain levels, good max sideforce feedback and probably a manual steering requirement because of packaging, engine accessory, and fuel economy constraints.

So, again in FSAE_Land, the steering wheel displacement working range limit plus its turn radius constraint produces a steering wheel to g-level hardware set. To make the driver at ease, the steering work (moment times angle) as a function of lateral g-level function is desigbed. There is an optimum setpoint for the slope of the work function, so now you would be designing a tierod load gradient to piggyback onto your tire-rod to steering wheel movement function and also settle on a steering wheel radius. Again, to make the driving task pleasant. Next would be to manage tire selection and pressures able to achieve the steering gain and then compensate deficiencies in tirerod load generation by tire alone, with mechanisms to augment (amplify or attenuate) the tierod loads.

There is also the caster crutch, added to align peak steering torque gradient with the maximum tire force gradient.

Now, don't get me wrong, some evil is necessary for fun, but I'm not a big caster fan. That's because of what caster does to initial turn-in, how tire wear affects it and how pressures and wheel rimwidth change all of this.

So, a good simulation using a well defined and road test proven procedure with an optimizer wrapped around it can greatly save the day. It's a splendid thesis project and an everyday practice in the car industry.

The only other thing I would ad is that its a good thing you don't have a performance task of demonstrating the ability to parallel park these cars....