Jump to content


Photo
- - - - -

KDSS Anti-roll Suspension


  • Please log in to reply
70 replies to this topic

#51 CWA

CWA
  • New Member

  • 17 posts
  • Joined: November 11

Posted 06 March 2012 - 20:17

there are suspension springs , they load up


In the kinetic system, yes. But the SAE paper desmo referred to describes a car with a mode decoupled system (with no warp stiffness) which replaces corner springs entirely.

Advertisement

#52 DaveW

DaveW
  • Member

  • 393 posts
  • Joined: January 09

Posted 06 March 2012 - 23:39

Interesting read. A few interesting concepts and opinions.

An interesting & seductive read. I confess that I don't know quite how respond, except to say that I was closely involved with a variety of active vehicles from road cars to F1 all of which simulated modal suspensions & I didn't reach the same conclusions. I think that the gist of Mr. Zapletal's argument has merit, but his conclusions fail to recognize the complexity of suspension problem.

Hence, for example, a null warp dynamic stiffness will reduce vehicle response to a warp input, but a vehicle warp displacement will not to be corrected. Also, if a warp load is required to maintain lateral balance in a corner, then there is no way of achieving that without warp stiffness. In truth, a low warp dynamic stiffness will be helpful on some circuits (e.g. Monaco), but would be found to be less acceptable (slow) on more conventional circuits (e.g. Silverstone).

I could continue, but I think might be sufficient to state that F1 suspension set-ups tend to follow some of Mr. Zapletal's reasoning.

Edited by DaveW, 06 March 2012 - 23:41.


#53 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 07 March 2012 - 02:28

Interesting read, that SAE paper. Why not decouple the modes? Do we need warp/twist mode stiffness at all?

Apart from the obvious advantage of even tyre loadings on a "warped" section of road, the lack of warp stiffness results in a reduced stiffness in single wheel bump for a given roll and pitch stiffness.

#54 DaveW

DaveW
  • Member

  • 393 posts
  • Joined: January 09

Posted 07 March 2012 - 09:53

Apart from the obvious advantage of even tyre loadings on a "warped" section of road, the lack of warp stiffness results in a reduced stiffness in single wheel bump for a given roll and pitch stiffness.

Quite right, but that is not the whole story.

The initial input at the front wheels of a kerb strike is not a pure roll input, it is the combination of a roll & a heave input (I neglect warp for now). With a stiff heave spring, a soft roll spring will reduce the load change at the loaded wheel as you suggest, but it will also increase the load reduction at the unloaded wheel. If the input is big enough, then (in this condition) the stiff heave spring will cause the unloaded wheel to become airborne. It is easy to imagine (hopefully), that at some point during the latter stages of a kerb strike both front wheels can become airborne. That is exactly what happened to Kobayashi in Singapore 2011 qualifying when he took too much kerb - his next contact was the outside kerb & his final contact was the wall. A higher roll stiffness might have slowed him down over some parts of the track, but should also have helped him to complete the lap....

Incidentally, with a modally controlled active suspension with self-levelling operating on all modes, it was still necessary to run roll & warp springs to "tidy up" the suspension travel.

Edited by DaveW, 07 March 2012 - 11:04.


#55 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 07 March 2012 - 12:29

The initial input at the front wheels of a kerb strike is not a pure roll input, it is the combination of a roll & a heave input (I neglect warp for now). With a stiff heave spring, a soft roll spring will reduce the load change at the loaded wheel as you suggest, but it will also increase the load reduction at the unloaded wheel.

Am I missing something? Wouldn't low warp stiffness improve the single wheel kerb strike by increasing the load of the 2 nearest wheels? Is this the opposite of what you are saying?

#56 pugfan

pugfan
  • Member

  • 174 posts
  • Joined: August 09

Posted 07 March 2012 - 23:02

...If the input is big enough, then (in this condition) the stiff heave spring will cause the unloaded wheel to become airborne...


I'm a bit confused as well. Considering an axle in isolation for the moment, surely a stiff roll spring and soft heave spring is needed for a lift in one wheel to cause a lift in the other?

#57 jpf

jpf
  • Member

  • 615 posts
  • Joined: March 01

Posted 09 March 2012 - 19:46

I may be even more confused. How is warp not a linear superposition of pitch and roll?

#58 DaveW

DaveW
  • Member

  • 393 posts
  • Joined: January 09

Posted 09 March 2012 - 23:56

Am I missing something? Wouldn't low warp stiffness improve the single wheel kerb strike by increasing the load of the 2 nearest wheels? Is this the opposite of what you are saying?

I was talking about the initial hit (before the rear wheel encounters the kerb).... Logically, a soft roll/warp stiffness will not move the sprung mass to alleviate the hit the unloaded wheel will take when kerbing with a high heave stiffness. It is difficult to describe, but a modal suspension does always work in a obvious way - I am trying to compile a simple model in an attempt to explain & intend to post when I am happy that it works. Pugfan, please be patient.


I may be even more confused. How is warp not a linear superposition of pitch and roll?

If heave is H*[1,1,1,1], pitch is P*[1,1,-1,-1], roll is R*[1,-1,1,-1] & warp is W*[1,-1,-1,1], then I think you are looking for values for P & R that will equilibrate [W, -W, -W, W] with [P+R, P-R, -P+R, -P-R]. Try it & see....

Edited by DaveW, 10 March 2012 - 00:00.


#59 jpf

jpf
  • Member

  • 615 posts
  • Joined: March 01

Posted 10 March 2012 - 06:14

Ah, of course. Thanks.

Advertisement

#60 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 10 March 2012 - 06:20

I may be even more confused. How is warp not a linear superposition of pitch and roll?

Let me re-phrase Dave's answer in layman's terms (the kind I need in order to understand these things). Any combination of pitch and roll will still end up with the four contact patches "in plane" ie a flat road. To warp a suspension requires a warped road or a wheel lift.

#61 DaveW

DaveW
  • Member

  • 393 posts
  • Joined: January 09

Posted 20 March 2012 - 10:51

Am I missing something? Wouldn't low warp stiffness improve the single wheel kerb strike by increasing the load of the 2 nearest wheels? Is this the opposite of what you are saying?

Here is my long-delayed reply.

I modeled the front axle of a vehicle, contacting a notional kerb on its left side. The kerb profile is shown here.

The modal (heave & roll, respectively) responses for a typical F3 setup are shown here, and the contact patch loads (left & right, respectively) are here.

The equivalent modal responses for a vehicle with F1 vertical stiffness levels but no roll stiffness are shown here, and the contact patch loads are here.

To explain, since I modeled the front axle, only the initial responses are relevant, and since the input was vertical, roll stiffness is the same as warp stiffness. For the "F3" case, the strong bar caused a rapid roll response, which backed off the heave response at the unloaded side, reducing the contact patch load variation of that side. Thus, whilst the loaded wheel is launched by the kerb, the unloaded wheel remains in good contact with the road. This contrasts with the the zero roll stiffness case, where the the high vertical displacement is initially "shared" by the unloaded wheel, causing both wheels to lose contact with the road. Hence my comment about Kobayashi's incident & why I don't think that zero warp stiffness is always a good strategy.

Very simplified, but I hope helpful.



#62 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 20 March 2012 - 12:38

Very interesting Dave and thanks for the considerable effort you must have put in. I don't understand your outputs 100%, for example Graphic #1 - is this two bumps/kerbs?

Are you sure your zero front roll stiffness is equivalent to zero warp stiffness (ZWS)? The lack of any roll response in Graphic 3 doesn't fit with my (intuitive only) view of how a ZWS system with normal levels of roll, pitch and heave stiffness would respond. A single wheel kerb strike should produce pitch and roll responses via the lateral and longitudinal "Z bar" or equivalent couplings.

#63 DaveW

DaveW
  • Member

  • 393 posts
  • Joined: January 09

Posted 20 March 2012 - 20:36

Very interesting Dave and thanks for the considerable effort you must have put in. I don't understand your outputs 100%, for example Graphic #1 - is this two bumps/kerbs?

Thanks, but it was not too onerous. I used the opportunity to become more familiar with SciLab. I admitted defeat for a while, becoming convinced that my distant knowledge of Matlab was not helping. Finally I tried SciLab version 5.4 (pre-release) & it came together fairly quickly. My kerb input was obtained by adding together different generator elements, & two peaks was more interesting than one (I thought)...

Are you sure your zero front roll stiffness is equivalent to zero warp stiffness (ZWS)? The lack of any roll response in Graphic 3 doesn't fit with my (intuitive only) view of how a ZWS system with normal levels of roll, pitch and heave stiffness would respond. A single wheel kerb strike should produce pitch and roll responses via the lateral and longitudinal "Z bar" or equivalent couplings.

Perhaps I was being too simplistic, although I have tested an F1 vehicle which had a front roll stiffness (& I too mean roll) rather less than 10 percent of the heave stiffness, courtesy of a stiff "heave" spring & rather notional "corner" springs & bar. Interestingly, the heave stiffness was also around 3 times the tyre stiffness. Such a set-up is extreme and, in the limit, it does require the sprung mass to be moved around to maintain continuous tyre contact.



#64 RDV

RDV
  • Member

  • 6,731 posts
  • Joined: March 02

Posted 24 March 2012 - 19:17

GL-Having seen one of these Kinetic systems up close and personal I am not going to write them off.

..one has run the Kinetic system in track racing, and can say very effective, with immense possibilities. Our system had installation problems (never really worked out the bugs in the hydraulics and mechanical bits, our problem, not Kinetics...), more due to a competitive championship...when we got the edge with a conventional car, dropped the Kinetic H2O system as lacked development. Have discussed it with the people who ran it in rally cars, and they were pretty happy with it. Suspect our small suspension movement and mega aero car did not get as much gain from it as a rally car on gravel and flying...

#65 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 27 March 2012 - 04:29

The "Pull to Damp" thread has moved towards a discussion of roll resistance and roll damping which prompted me to post this sketch but I thought it might be better posted here - particularly since the artist is Erik Zapletal mentioned earlier in this thread. I have copied it without permission but I'm sure Erik won't mind.
Posted Image

This suspension layout is a simple implementation of zero warp stiffness with fully decoupled roll and pitch stiffness.

If the longitudinal leaf springs were replaced with rockers, each connected to the chassis by a single coil spring and damper, the roll and pitch damping would also be decoupled and independently adjustable. The warp mode would be undamped. Dive and squat could be controlled seperately without affecting the roll moment distribution which (I think) is entirely kinematic (no elastic component) and a function of side pivot location.

Edited by gruntguru, 27 March 2012 - 04:43.


#66 Johan Lekas

Johan Lekas
  • Member

  • 45 posts
  • Joined: May 09

Posted 27 March 2012 - 08:51

Posted Image
Design with decoupled roll mode, free (undamped) warp mode
Roll resistance distribution front/rear can be changed by adjusting the leverage ratios of the rockers



#67 DaveW

DaveW
  • Member

  • 393 posts
  • Joined: January 09

Posted 27 March 2012 - 19:30

Roll resistance distribution front/rear can be changed by adjusting the leverage ratios of the rockers

Forgive me, Johan, but are you sure. It seems to me that changing roll resistance distribution will, in a turn, introduce a warp displacement of the suspension. Zero warp resistance implies that the displacement will not generate a warp load, so it follows that changing the roll resistance distribution will change the "average" roll resistance, coupled with some warp deflection in a turn, but no change in "balance".



#68 Johan Lekas

Johan Lekas
  • Member

  • 45 posts
  • Joined: May 09

Posted 28 March 2012 - 08:12

... but are you sure.

Well I was, kind of, but now... :)

It seems to me that changing roll resistance distribution will, in a turn, introduce a warp displacement of the suspension. Zero warp resistance implies that the displacement will not generate a warp load, so it follows that changing the roll resistance distribution will change the "average" roll resistance, coupled with some warp deflection in a turn, but no change in "balance".

I think I understand what you mean: If the rockers are the same geometry front and rear the axles can be moved in warp without roll movement of the chassis, and vice-versa. If the geometry of one of the rockers is changed, moving the axles in warp will also give roll movement of the chassis, and vice-versa. I can also see that the average roll resistance changes.
But that it would not change balance (roll resistance distribution front/rear)... I'll have to think about that!
Thanks for your input!


#69 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 07 January 2013 - 00:27

Well I was, kind of, but now... :)

The key here is that regardless of the ERMD setting, the suspension can be "warped" without displacing any of the springs. The fact that the chassis may simultaneously roll a little, depending on the ERMD setting is not relevent.

Johan. Not sure if I asked this question earlier, but what are your thoughts on the complete lack of warp mode damping? I ask because I recently saw University of Western Australia's latest FSAE chassis and it has zero warp stiffness, but with the dampers in the conventional location ie one per wheel.

The UWA layout is functionally similar to the sketch in Post#65 but with the "Longitudinal Leaf Springs" replaced by rocker beams and the pivots attached to an ARB rather than springs.

Edited by gruntguru, 07 January 2013 - 00:31.


#70 Johan Lekas

Johan Lekas
  • Member

  • 45 posts
  • Joined: May 09

Posted 07 January 2013 - 18:44

Johan. Not sure if I asked this question earlier, but what are your thoughts on the complete lack of warp mode damping?

I've been thinking along these lines:
The springs involved when the suspension moves in warp are the tires, and the mass moving consists of the axles and roll suspension components (the chassis does not move). So what could theoretically happen is that there would be some oscillations in this spring/mass system. The frequency would have to be relatively high since the tires are stiff and the mass not that big. The oscillation movement would also have to pass the roll spring/damper unit, which probably could let small displacement movements past without damping. I still think it's unlikely that there will be any oscillations since there is friction in links etc and there is some damping in the tires.

Do you have any pictures of or links to the UWA system? Would be interesting to check it out!


#71 gruntguru

gruntguru
  • Member

  • 5,194 posts
  • Joined: January 09

Posted 07 January 2013 - 23:17

Do you have any pictures of or links to the UWA system? Would be interesting to check it out!

UWA 2012 photos courtesy of Rex Chan - University of Melbourne.

Car is not complete (dampers attached to axles but not chassis etc). Note:
- "E" shaped plate acting as spring & lat/long location link at centre of each (beam) axle.
- Underbody aero (undertray) acts as "twin rocker beams" connecting front and rear axles. Also restrains rotation of axles about Y and Z axes.
- Roll mode controlled by single ARB connected to "rocker beams" at a point (which is longitudinally adjustable to set F/R roll stiffness)
- Final pic shows car with approx 150mm of shim under LF tyre (3 remaining wheels on ground, dampers not connected) to demonstrate soft twist and travel in single wheel bump. These cars are only required to have +- 25mm of travel.