Contact Patch Load Optimisation

[Henry 1.9GTi 36]

I have had the privilage of getting my 205 on the Brookes 4-Post rig for a test session as part of a university project. The purpose of this project was to build a 7 degree of freedom (un-sprung mass x4 Translate up and down, body; roll, pitch, heave) computer model, input it with rig data and compare and then set about optimising the damping for best contact patch load RMS (root mean sqaure). The jist of CPL:

 

From figure 1 (a graph showing how increasing normal load on a tyre results in diminishing returns in terms of lateral force production) it is possible to understand that if the CPL variation is greater, then the mean value for lateral force will subsequently be less. It is therefore desirable to minimise this variation by correct selection of spring and damper rates. A softer spring will produce less CPL variation. However, a softer spring could lead to one or more of the following:

 

• Increased body roll

o Tyre geometry primarily concerned with camber control and bump steer adversely affected; reduction in mechanical grip.

o Poor driver feel

• The need for more suspension travel

o Raised body and hence higher centre of gravity; increased weight transfer leading to reduction in mechanical grip

• Grounding

o The body bottoming out or hitting the bump stops leading to undesirable handling characteristics.

• Significant ride height and attitude changes

o Loss of aerodynamic down force; large reduction in grip

 

It is desirable to run the race car as soft as possible to optimise CPL variation without losing grip due to above mentioned points.

 

Some comparison (Right Rear): Exp4Rig being data from the 4-Post rig; not a bad match to the model. The 2nd increase due to the hub mode is overestimated, probably due to the dampers being modelled as linear.

 

 

edit: axes seem to have been cut off. x: Freq 0-40Hz y: Contact Patch Load Amplitude Ratio

 

Baseline CPL RMS Optimised CPL RMS Optimised Damping Ratio (ζ) Optimised Damping Rate (Ns/mm) Baseline Damping Rate (Ns/mm)

FL 361.18 356.75 0.4 2.5797 2.0598

FR 379.31 375.00 0.4 2.7095 2.0598

RL 299.82 293.49 0.38 1.8383 2.4513

RR 301.47 297.05 0.36 1.9169 2.4513

 

Results from the damper optimisation above.

 

And here we see the effect of optimising the right rear damper on the CPL

 

 

edit: x: time 0-60s y: Contact Patch Load (N) 0-1000

 

Please take with a pinch of salt as the baseline setup is derived by the 4-Post rig software and may be inaccurate. But it seems the rear of the car is well controlled, the lower damping predicted by the simulation only gave a slight decrease in RMS value over the whole frequency range. This higher damping will speed up weight transfer and should help initial turn in. The front is based upon a working suspension which my car certainly did not have; will update soon!

 

Cheers,

Henry.

Edited April 5, 2011 by Henry 1.9GTi

[rallysteve 22]

Ah, the wonders of Matlab

 

Steve

[Cameron 16]

This should be interesting.

[GeorgeXS 32]

You're at Oxford Brookes Henry?

I'm there! First year Auto student, so i don't get to play with the nice bits of kit yet

Looks interesting, ill be following this.

[Henry 1.9GTi 36]

Table above without bullitin board editing

 

 

Right, so after doing some damper optimiseation on working suspension it was time to take a look at what was going on at the front. When I got the car home I reassembled the suspension without the spring. The suspension was free to move through its full range of travel without much effort other than the resistance you would expect from the damper. Certainly not 1000N (~100kg) of stiction:

 

 

The front contact patch load variation and RMS value was truely awful which is another reason I think the car is tending to understeer and doesnt respsond well to tyre pressure changes.

Heres what the front contact patch load looked like compared to the simulation which does not model stiction and friction.

 

 

As you can see the frequency of the peak is miles out. This first peak is associated with the body mode; the body going through resonance, which simplified is determined by the mass at that corner and the spring rate. With the 250lb spring I have the natural frequency should be about 2.1Hz. As you can see the peak is around 4.75Hz. Way out. Also worth noting there is no second peak usually associated with the hub mode. By estimating the damped natural frequency of a corner with extremely high, unrealistic, damping i.e. the hub and body locked together the natural frequency is ~5.5Hz based upon a tyre spring rate churned out by the rig. So it looks like the car is simply bouncing on the tyre and you could see this when the car was on the rig. Due to a tyre having very low damping the peak is higher and the area under the graph makes the RMS value very high.

 

By comparing body and hub accelerations:

 

 

It is possible to see that the front hub and body are experiencing the same acc due to the stiction and bouncing on the tyre. The rear hub and body accs are as expected and the rear hub (un-sprung) mass is well controlled. Thanks GrpN billies!

 

So it is reasonable to assume that the front stiction is induced by bending load as a result of changing the spring axis. On standard suspension the spring axis is offset from the damper. Coilovers change this, and therefore a bendind load is induced:

 

 

I have read a few papers on spring design and side loads but unfortunately can't find anything on how race cars deal with this or whether their geometry takes this into account. Or whether dampers are designed for high bending loads.

 

Running longer wishbones compounds the problem, but adding negative camber at the top mount will help.

 

My next goal is to get GrpN billies on the front with custom spring and eccentric top mounts and retest the car. And then race the little bugger!

 

Cheers,

Henry.

 

George:

 

yes im at brookes, but only for another month!! good luck with the degree.

[Batfink 201]

If you look at the size of the struts on a race coilover they are huge in comparison to conventional suspension. They can take far greater bending loads..

[Cameron 16]

The bending loads experienced in the struts when the car is stationary are nothing compared to those under braking / cornering, that's the whole reason why the standard springs have their axis at an angle to that of the strut; the force caused by the compressed spring then (partially) counters the bending moments caused by braking / cornering. OE dampers are certainly designed for high bending loads, and aftermarket coilovers should be even more so since they don't get any help from the spring as the axes are concentric.

 

I'd be surprised if the stiction was caused by the dampers to be honest, unless the bushes are totally shot. If they're any reasonable make then the bushes should easily cope with all the bending moments.

Edited April 18, 2011 by Cameron

[Henry 1.9GTi 36]

the thing is during a heave run on the rig there is no cornering and braking load to induce bending. It is primarily down to the forces involved in compressing the spring. Every single moving part in the front suspension is free, i have checked arb, arb drop links top mount bearings CV joints, wishbones, dampers checked out ok on the dyno. So the only assumption to make is its induced by the change in spring axis. I would love to get another 205 on the rig with the same suspension and see whats what, but that is unlikely unless george does his project on it in a few years! hint hint

 

will continue with the plan of the GrPN billies and see how it goes. Worst comes to worst I solve a problem that was cuased by something else but still have good suspension at the end of it.

[Cameron 16]

How were you measuring the force when compressing the strut? Were you just cycling one wheel up and down throughout a small amount of the suspension travel, and were you doing this with the springs in place?

 

I think the hysteresis you see there is a sum of the stiction in all of the suspension parts, rather than just in the damper. There will be stiction in the BBJ, in the track rod BJ's, hysteresis and / or stiction of the wishbone bushes, stiction in the damper; and if you had the springs / ARB fitted - losses from friction between the spring and spring pan, stiction in the ARB bushes; all of this will add up. Maybe it would be a good idea to get a force gauge out and measure the force required to move all of these parts separately rather than wiggling them by hand and assuming they're ok? I wouldn't be surprised if you saw 10-20N in each of these parts, maybe more! It's certainly worth looking into to get a greater depth of understanding here.

 

I remember our FS car had a huge amount of stiction when we first put it together, and it was enough to make the ride quite harsh. This was despite the wishbones being pretty easy to move when you disconnected everything and tried them separately. We took everything apart and smoked the spherical bearings in and it was buttery smooth afterwards, and felt miles better to drive.

Edited April 19, 2011 by Cameron

[Henry 1.9GTi 36]

the force is measured by the rig by £20k worth of load cell, which samples at about 20KHz and its displacement is accurate to 2 microns or something!

 

We did disconnect the ARB and steering arms as they are quick and easy to do, but it made no noticeable difference to the stiction. I agree there will definately be some stiction and friction in all the moving parts, but 1000N is a problem, not a cumulative effect of the parts. To compare, the rear had about 150N of stiction was was deemed normal and acceptable. It also had about 50N friction compared to the fronts 175N when it actually moved.

 

The stiction and friction was meaured by running the car in warp, otherwise the front just didnt move.

 

I would have expected that that amount of stiction if caused by a suspect joint/bushing would be pretty obvious and not able to move the part by hand. Everything on the car has been replaced with new parts in the past that havn't covered that much mileage. Anyway this is the current situation, pretty interesting to see regardless of whats causing it. Will report back when the changes have been made.

 

Cheers,

Henry.

[Cameron 16]

I think you misunderstood my question, I mean how does the rig test for stiction? I guess from your chart that it does it by cycling one wheel up and down through a small distance, but what frequency does it do this at? I imagine the frequency is quite low..

 

Comparing the rear to the front isn't really representative though, if you think about what's moving at the rear it's just a shaft rotating in a couple of bearings, a rubber seal, and two bushes on the shock. It doesn't really compare to the amount of contributing factors at the front where you have 2 wishbone bushes, 2 ball joints, ARB bush and top mount (and driveshaft if you want to get into fine details), so I'm not surprised you see far less stiction at the back.

 

Don't think I'm having a go or knocking your approach, I have zero experience with rig testing! I'm just pointing out that if the components of the front suspension have say 50N stiction to overcome, which is quite reasonable, then there's at least 300N of your 1000N.

[Henry 1.9GTi 36]

the sitction measure was run by cycling the car in warp at a low frequency. i.e left front right rear go up and right front left rear go down. This was after the results from the heave run, the guy running the rig who has done a phd on it and written all the matlab code ran it through some warp runs at large amplitude with constant frequency so we could see what was going on. So I trust him when he says this is not normal and also had a discussion with my lecturer who is a vehicle dynamics guy and he was quite bemused by why I had changed the spring axis lol.

 

Ok so say 300N may be reasonable. still 700 left either way, for want of better words, it's buggered. Gonna order the dampers and get the custom springs made and then draw up a top spring collar to account for the angle and try and fit it all together! It nothing else it's pretty fun

[Rippthrough 98]

I'd stick a couple of roller needle thrust bearings under the springs whilst you're at it, that usually makes a decent improvement when there's coilover units up front.

[Cameron 16]

Yeah I wasn't saying your rig operator doesn't know what he's doing!

 

We used thrust bearings on our FS car's dampers, they actually made quite a bit of difference when we tested the dampers and that was with pretty soft springs.

[welshpug 1,666]

is that to allow the springs to turn on the pans whilst compressing?

[Batfink 201]

Would you do that as well as having a bearing at the top?

[Rippthrough 98]

is that to allow the springs to turn on the pans whilst compressing?

 

Yep.

Usually worth it even with a spherical bearing at the top tbh.

[Henry 1.9GTi 36]

after a bit of digging, old BTCC cars used linear bearing dampers to accomodate shaft motion and remove static and dynamic friction. Penske supply the btcc which now use push rod actuated horizontally mounted dampers because getting a McPhearson strut to work well was too expensive apparently, and as always regs are moving to reduce cost.

 

E-mailed penske about dampers for the 205 and yes they are available with linear bearing technology..... £3000 each, and that was the cheapest 1 way adjustable. Oh how I would love a set. So back to the GrpN billies for spring axis mods untill I suddenly become rich.

Edited May 9, 2011 by Henry 1.9GTi

[Rippthrough 98]

Yeah, they banned those in the wrc because of the costs.

[EdCherry 34 1 Cars]

Still got bending loads in dampers mounted on a rocker/crank though.

 

If you really serious about using a penske design give quantum racing shocks a call. Built identical to most penske's according to both QRS and a friend who runs them on several cars (although havn't had one apart yet), relatively good pricing to against what penske quoted me for a set of dampers to my design for a 205.

[RossD 44]

Fascinating topic - I'll be interested to see your results witht he new front dampers (Vested interest - I run them!)

[Cameron 16]

Still got bending loads in dampers mounted on a rocker/crank though.

 

Explain how?

Surely if you have spherical bearings at both ends the loading is 100% axial.

 

Funny you should mention rockered dampers though, this it the approach I'm going with for a double wishbone setup.

Edited May 13, 2011 by Cameron

[Rippthrough 98]

Not entirely, there'll always be some load transmitted through the spherical bearings, they're not frictionless.

 

That's why the Atom started riding so much better when they switched the pivot and shock bearings from phosphor bronze plain units to roller needles.

[Cameron 16]

But it's negligible, surely?

 

I know you get stiction from sphericals (particularly new ones), our entire FS suspension was on them. The combination of 12 brand new spherical bearings on each corner did make the ride harsher (was incredibly smooth after we smoked them in), but actually putting significant flex loads in the shocks from it? I can't really see it happening..

[Rippthrough 98]

^^Pretty much, it's noticable when we stick fresh bearings on a car but then they are running 8 dampers and the bearings are 16mm bore, so that's a fair bit of stiction 'fore they bed in!