KDSS Anti-roll Suspension
#1
Posted 11 November 2011 - 01:40
http://www.autozine....suspension3.htm
I have been reading a description of the fundamentals of the KDSS suspension system (bottom section of linked page) fitted in some Lexus off-roaders, and have a question relating to whether I have interpretted the system correctly or not. This in particular..
"What if the car is not cornering, but hitting a bump on the road ? Oil from the compressed wheel will flow towards other wheels, thus the suspension absorbs the bump comfortably".
I am struggling to see how this statement describes a useful characteristic. As far as I am aware, an anti-roll bar that transmits single wheel bumps to the other wheel on it's axle is a bad thing, and so a wheel that transmits bump movement to potentially all four wheels is even worse. This, as far as I can see it, is like one of the drawbacks to the live axle in single wheel bumps, in that the bump movement is transmitted to the opposing wheel (just without the extra problem of opposite camber gain?).
Presumably the reason the author says the bump transmission to other wheels is a good thing because less shock is dealt with by just the one corner? Even if this is true, I can't see how this is a benefit when it also effectively alters the camber of one-to-three extra wheels in the process.
Could anyone shed some light on this?
Also just to clarify, I am using Milliken's ride and roll rates formulae for a FSAE car (Chapter 16). The effects of mechanical roll bars on the roll rate are shown, but the consequential effect on the ride rate that a roll bar has is not mentioned or considered. I have read from a seemingly informal source that for the same vertical wheel displacement in roll (outer wheel) and in bump, the roll bar stiffness effectiveness in ride has half its stiffness effectiveness in roll. Can anyone confirm this?
Many thanks
Advertisement
#2
Posted 11 November 2011 - 03:00
Also just to clarify, I am using Milliken's ride and roll rates formulae for a FSAE car (Chapter 16). The effects of mechanical roll bars on the roll rate are shown, but the consequential effect on the ride rate that a roll bar has is not mentioned or considered. I have read from a seemingly informal source that for the same vertical wheel displacement in roll (outer wheel) and in bump, the roll bar stiffness effectiveness in ride has half its stiffness effectiveness in roll. Can anyone confirm this?
last bit first, using wheel rates. in ride the wheel rate is say 25 N/mm, with virtually no contribution from the sta bar. In roll it is say 55 N/mm. In a single wheel bump event it will be 40 N/mm. the reason is that the stab bar sees only half as much twist,.
#3
Posted 11 November 2011 - 04:13
#4
Posted 11 November 2011 - 05:32
Imagine the ARB or hydraulic connection to be rigid. For the two-wheel connection (ARB) the bump applied to one wheel will produce an equal bump on the other. For the four-wheel connection the bump applied to a single wheel produces an average deflection (bump in one and rebound in the other two) of only one third of the bump input.
#5
Posted 11 November 2011 - 09:16
#6
Posted 11 November 2011 - 13:03
I don't wish to take issue with either of the responses to your post (I particularly liked gg's "Imagine the hydraulic connection to be rigid"). However I think it would be wise to note (with apologies) that the explanations contained in your reference are not necessarily accurate, or even helpful. The section on fully active suspension, for example, contains details that are only partially based on fact & explanations that lack understanding to the point of being misleading.
Exactly. Autozine strikes again. It would be easier to start over somewhere else than to sort out that lot.
#7
Posted 14 November 2011 - 16:42
My question still stands though, is transmitting the bump to the other wheels good or bad? I can see that the shock is less focused on one wheel, but it will alter the camber of all the wheels, instead of just one.
Also, DaveW I'm not worried about eveything else Autozine says, just about the KDSS bit. Can you explain KDSS in a better way? Or link me to a source that does? I'm afraid I cannot find much more info on it.
#8
Posted 14 November 2011 - 18:33
My question still stands though, is transmitting the bump to the other wheels good or bad? I can see that the shock is less focused on one wheel, but it will alter the camber of all the wheels, instead of just one.
Another example for the purity that is a beam axle ;)
#9
Posted 14 November 2011 - 18:36
For whatever reason most of 'my' cars use grippy D blocks, so i see a contribution from the torsional rate of the D block in the ride rate. I shouldn't have put that in my reply, normal people using sensible D blocks don't have that problem.
KDSS was bought out by Tenneco, what little information was publicly available seems to have gone.
Kinetic themselves were fairly cagey about handing out technical stuff, I was shown but not given their video, for example. That being said the ideas were pretty straightforward.
I wouldn't get hung up on camber in particular, with modern tires it is not as important as it was. Is there a ride advantage in lifting both axles in response to a single wheel bump? yes, obviously, you get less pitch and roll.
#10
Posted 14 November 2011 - 21:04
The position of the wheels can be described as a combination of movements in the following modes
Heave - all wheels move up/down
Roll - left/right wheel pairs move up/down
Pitch - front/rear wheel pairs move up/down
Warp - diagonal wheel pairs move up/down
When one wheel hits a bump and moves up say 8mm, it's 2mm movement in the heave mode, 2mm in roll mode, 2mm pitch and 2mm warp (yes, it seems unlikely, but it works out that way). So the wheel rate for a one wheel bump is the average of the rates in the four modes
Some info on modes here here
#11
Posted 15 November 2011 - 06:19
Is there a ride advantage in lifting both axles in response to a single wheel bump? yes, obviously, you get less pitch and roll.
The sales pitch (or not!) for the Austin 1800 right there.
http://www.bluestrea...800/Page16.html
Edited by cheapracer, 15 November 2011 - 06:19.
#12
Posted 17 November 2011 - 05:29
To understand what happen in a one-wheel bump I think it helps to look at the suspension "modally"
The position of the wheels can be described as a combination of movements in the following modes
Heave - all wheels move up/down
Roll - left/right wheel pairs move up/down
Pitch - front/rear wheel pairs move up/down
Warp - diagonal wheel pairs move up/down
When one wheel hits a bump and moves up say 8mm, it's 2mm movement in the heave mode, 2mm in roll mode, 2mm pitch and 2mm warp (yes, it seems unlikely, but it works out that way). So the wheel rate for a one wheel bump is the average of the rates in the four modes
Some info on modes here here
I disagree.
If one wheel moves up 8mm, that is one wheel moving up 8mm.
If the left front moves up 8mm and the right front moves down 2mm, that's the left front moving up 8mm and the right front moving down 2mm.
If the left front moves up 8mm, right front moves down 2mm and right rear moves down 2mm, that's all still just individual wheel movements.
If the left front moves up 8mm, right front moves down 2mm, right rear moves down 2mm and left rear moves up 1mm. That, finally, is "1mm" of roll.
The way you've described it isn't orthogonal.
I'll have to have a look at that paper, personally I believe only an active system with hydraulic actuators can implement a truly de-coupled suspension.
Edited by pugfan, 17 November 2011 - 05:54.
#13
Posted 17 November 2011 - 07:14
LF=r+p+b+w
RF=-r+p+b-w
LR=r-p+b-w
RR=-r-p+b+w
where roll pitch bounce and warp are scaled by track wheelbase etc to give consistent units. There's some -1s floating around in there as well, depending on your coordinate system.
A single wheel bump will have some r component, if you do it this way.
#14
Posted 17 November 2011 - 16:32
#15
Posted 17 November 2011 - 17:36
I see it this way: The situation that the LF moves up 8mm and the other wheels keep their vertical positions can be described as a combination of movements in the four modes.I disagree.
If one wheel moves up 8mm, that is one wheel moving up 8mm.
If the left front moves up 8mm and the right front moves down 2mm, that's the left front moving up 8mm and the right front moving down 2mm.
If the left front moves up 8mm, right front moves down 2mm and right rear moves down 2mm, that's all still just individual wheel movements.
If the left front moves up 8mm, right front moves down 2mm, right rear moves down 2mm and left rear moves up 1mm. That, finally, is "1mm" of roll.
The way you've described it isn't orthogonal.
I'll have to have a look at that paper, personally I believe only an active system with hydraulic actuators can implement a truly de-coupled suspension.
Wheel positions if you look at the different modes in sequence:
2 mm in heave mode:
All wheels at +2 mm
...and 2mm in roll mode:
LF and LR at +4 mm
RF and RR at 0 mm
...and 2 mm in pitch mode:
LF at +6 mm
LR at +2 mm
RF at +2 mm
RR at -2 mm
...and 2mm in warp mode:
LF at +8 mm
LR at 0 mm
RF at 0 mm
RR at 0 mm
#16
Posted 18 November 2011 - 00:42
I see it this way: The situation that the LF moves up 8mm and the other wheels keep their vertical positions can be described as a combination of movements in the four modes.
Wheel positions if you look at the different modes in sequence:
2 mm in heave mode:
All wheels at +2 mm
...and 2mm in roll mode:
LF and LR at +4 mm
RF and RR at 0 mm
...and 2 mm in pitch mode:
LF at +6 mm
LR at +2 mm
RF at +2 mm
RR at -2 mm
...and 2mm in warp mode:
LF at +8 mm
LR at 0 mm
RF at 0 mm
RR at 0 mm
What Greg posted certainly smells like an orthogonal matrix. This particular thought experiment of one wheel 8mm in jounce troubles me though. Surely with the car level to the road, there is no roll, pitch or heave component? I guess the problem is the car doesn't know it's level to the road.
The other problem I see is if roll, pitch and heave (as defined by the matrix) all have different stiffnesses (the aim of a de-coupled suspension) a strictly one wheel bump should ideally produce no body deflection but with different stiffnesses all sorts of weird responses could result.
Edited by pugfan, 18 November 2011 - 00:43.
#17
Posted 18 November 2011 - 01:51
Intuitively it does have a roll component- the sta bar comes into play for a start!What Greg posted certainly smells like an orthogonal matrix. This particular thought experiment of one wheel 8mm in jounce troubles me though. Surely with the car level to the road, there is no roll, pitch or heave component? I guess the problem is the car doesn't know it's level to the road.
The other problem I see is if roll, pitch and heave (as defined by the matrix) all have different stiffnesses (the aim of a de-coupled suspension) a strictly one wheel bump should ideally produce no body deflection but with different stiffnesses all sorts of weird responses could result.
#18
Posted 18 November 2011 - 02:48
If one wheel is in jounce the road is not level.What Greg posted certainly smells like an orthogonal matrix. This particular thought experiment of one wheel 8mm in jounce troubles me though. Surely with the car level to the road, there is no roll, pitch or heave component? I guess the problem is the car doesn't know it's level to the road.
Another example. Car is cornering hard and rolling 5 degrees. The corner is banked 5 degrees so the car is actually level, but the suspension is in a "5 degrees of roll" mode.
#19
Posted 18 November 2011 - 03:37
Intuitively it does have a roll component- the sta bar comes into play for a start!
True, but an anti-roll bar senses axle roll for want of a better term rather than body roll.
Something like the following seems more 'correct' to me:
| LF | = | 1 -1 1||P| + |OLF|
| RF | = | 1 1 1||R| + |ORF|
| LR | = |-1 -1 1||H| + |OLR|
| RR | = |-1 1 1| + |ORR|
O stands for one wheel.
Obviously this is under-determined though.
Edit: I've tried to use a monospace font for the matrix formatting but obviously the bulletin board has seen me coming.
Edited by pugfan, 18 November 2011 - 03:40.
Advertisement
#20
Posted 18 November 2011 - 04:25
If one wheel is in jounce the road is not level.
Another example. Car is cornering hard and rolling 5 degrees. The corner is banked 5 degrees so the car is actually level, but the suspension is in a "5 degrees of roll" mode.
In the banked example, ideally the car would level with the road rather than the earth's surface though.
#21
Posted 18 November 2011 - 04:32
#22
Posted 18 November 2011 - 04:45
Why?
A couple of reasons off the top of my head:
- reduce lateral weight transfer
- maintain suspension travel for road irregularities.
#23
Posted 18 November 2011 - 06:51
#24
Posted 18 November 2011 - 10:17
Nice in theory...
Did you see the mechanical implementation in that paper that Johan Lekas linked to? There's no way that would ever make production.
Personally I think mode de-coupled suspensions are always likely to remain a theory because active or passive hydraulic and mechanical implementations all have drawbacks for anything other than a prototype/technical demonstrator.
#25
Posted 22 November 2011 - 12:16
The way I picture it, the 2D plane created beneath the car by all four contact patches when one wheel is in bump can be returned to the same orientation as that of the car by moving it around in the four modes.
All these matrices posted are confusing I have to admit, even in the SAE article I have trouble following how the end product is calculated (Ri,i = (Kh+Kp+Kr+Kx)/4). This is down to my low level of maths though, and is something I'm working on.
"Personally I think mode de-coupled suspensions are always likely to remain a theory because active or passive hydraulic and mechanical implementations all have drawbacks for anything other than a prototype/technical demonstrator."
Pugfan, what drawbacks are there to hydraulic implementation of this passive system? The drawbacks of a mechanical system are clear enough, I can't see any main drawbacks with the hydraulic stuff that would make this less suitable to the 'normal' way of doing things.
#26
Posted 22 November 2011 - 21:18
#27
Posted 23 November 2011 - 08:10
#28
Posted 23 November 2011 - 20:48
#29
Posted 24 November 2011 - 01:36
Pugfan, what drawbacks are there to hydraulic implementation of this passive system? The drawbacks of a mechanical system are clear enough, I can't see any main drawbacks with the hydraulic stuff that would make this less suitable to the 'normal' way of doing things.
Only in that I don't think they meet my definition of de-coupled but I've been talking in a purely theoretical sense about completely de-coupling all modes.
The matrices simply map the wheel displacements into a different set of parameters. I think but have zero theretical calculations or experimental evidence or even much experience to back up my opinion that Roll, Heave, Pitch and one wheel displacement for each wheel are the parameters of interest rather than Roll, Heave, Pitch and Warp.
Many real world examples in race and production vehicles obviously show that pragmatically implemented de-coupled systems can certainly have advantages over 'conventional' suspensions.
#30
Posted 05 December 2011 - 20:09
Erik Zapletal, Balanced Suspension - http://www.scribd.co...nced-suspension
As far as I can tell it was published before Creuat started implementing mode decoupling, and before they published their paper above. What I picked up on was Zapletal's description of one-wheel bump as basically warp, and not the average of Pitch, Roll, Heave and Warp as you guys and Creuat have since mentioned. Is this something you were getting at pugfan?
Opinions? Particularly on the part that says how side-pair-models meet the requirements of racing suspension the most, and how Zapletal is surprised the racing community hasn't picked up more on this 50 year old technology..
#31
Posted 07 December 2011 - 21:55
Think you better stop using ladder chassis....and 2mm in roll mode:
LF and LR at +4 mm
RF and RR at 0 mm
...and 2 mm in pitch mode:
LF at +6 mm
LR at +2 mm
RF at +2 mm
RR at -2 mm
#32
Posted 07 December 2011 - 22:58
Think you better stop using ladder chassis.
..what?
#33
Posted 10 December 2011 - 07:58
Putting aside practical fitment issues, what am I missing?
#34
Posted 10 December 2011 - 15:25
Is there any particular reason why 2 anti roll bars that each go from the RHF wheel to the LHR wheel and obviously from the LHF wheel to the RHR wheel wouldn't work?
Putting aside practical fitment issues, what am I missing?
Z-bars?
Hmm, that would stiffen roll and pitch, and let warp and bounce be very soft wouldn't it.. I guess because the roll and pitch stiffnesses would end up being the same? Complicating the set up to decouple these modes allows more control over the individual rates? perhaps
Edited by CWA, 10 December 2011 - 16:12.
#35
Posted 10 December 2011 - 16:04
Hmm, that would stiffen roll and pitch, and let warp and bounce be very soft wouldn't it.. I guess because the roll and pitch stiffnesses would end up being the same? Complicating the set up to decouple these modes allows more control over the individual rates? perhaps
I'm talking about them being the only rollbars with 'standard' side to side anti rollbars removed.
I see it as stiffening warp lessening pitch at turn in under brakes, lessening pitch overall and distributing energy of one wheel to all 3 others instead of just it's opposite side brother (when using standard roll bars).
#36
Posted 21 December 2011 - 23:43
I'm talking about them being the only rollbars with 'standard' side to side anti rollbars removed.
I see it as stiffening warp lessening pitch at turn in under brakes, lessening pitch overall and distributing energy of one wheel to all 3 others instead of just it's opposite side brother (when using standard roll bars).
If the bars connect diagonal wheels together, they wouldn't stiffen the warp mode, just like normal anti-roll bars across each axle don't stiffen pitch movements.
With this diagonal connection you have described (what I understand to be Z-bar's, or X-bars), they will resist pure pitching as one front wheel wants to go up and the rear wants to go down. Similarly they will resist pure roll, as say a left wheel wants to go up and the right wants to go down. The diagonal wheels want to move in opposite directions in both cases so the bar acts. When the warp mode is activated, both wheels move in the same direction so the bar does not act. I'm picturing this as a mechanical connection so this logic makes sense to me, correct me if I'm wrong.
What you said about warp mode being stimulated when on the brakes and turning in is interesting though, I did consider this a few days ago. Isn't that movement just a mix of roll and pitch modes, which are different to warp? In the situation you described, all four wheels still make a flat 2D plane, which is flat ground, but the chassis plane is at different angles to it around two axes. When the axles are warped, the four wheels can't make a 2D plane. Which is how warp mode is only excited by uneven ground, surely?
#37
Posted 22 December 2011 - 07:26
If the bars connect diagonal wheels together, they wouldn't stiffen the warp mode, just like normal anti-roll bars across each axle don't stiffen pitch movements.
Not 'Z' bars, I am considering the bars just as they sit laterally - ie; the RH front wheel goes up so the LH rear wheel goes up (simplistic speak).
Obviously we are not considering real world fitment issues, just the theory.
#38
Posted 22 January 2012 - 13:40
Does anyone know how the handling balance is altered on cars that implement this system? As it's used on the McLaren MP4-12C, normally you'd assume the roll bars would be used to tune the balance. But by looking at this system, as only one roll spring is actuated, I don't see how it can be altered to adjust the balance between from and rear axles. Unless in the case of the McLaren, front and rear corner springs are what provide this balance? Or the roll axis created by suspension geometries? Anyone have any ideas?
Cheers
#39
Posted 22 January 2012 - 22:21
Advertisement
#40
Posted 23 January 2012 - 02:08
That's right, the road springs provide some of the roll stiffness for each axle, so you can alter the front to rear roll stiffness balance just by altering the stiffness of one a/r bar. The posh acronym is LLTD -lateral load transfer distribution. It is one of the most important easily adjustable parts of your understeer budget. The magic number associated with it is that it should be roughly 0-5% more front biased than your weight distribution, for a RWD car (more being better for traction, less being better for ride). That can lock you into a bad 'ride' tune, hence the need for active roll systems or active dampers.
Thanks for the advice on values Greg, I had heard the term LLTD before.
But as the system on the McLaren is used instead of anti-roll bars and is interlinked front and rear, I can't see exactly how you can effectively change the 'front' or 'rear' roll bar stiffnesses. There are two roll accumulators which provide springing, but as far as I can tell from the design of the hydraulic circuitry in the diagram, these won't alter the handling balance. It's for this reason I am assuming that the corner springs on the McLaren would be used to adjust the handling balance, does this sound correct?
Edited by CWA, 23 January 2012 - 02:09.
#41
Posted 23 January 2012 - 03:22
#42
Posted 23 January 2012 - 05:06
#43
Posted 15 February 2012 - 10:56
#44
Posted 18 February 2012 - 10:08
I found it interesting how he is using a ladder chassis because of the system's claims that it has no warp stiffness. I understand how the torsional rigidity requirement of a chassis is usually based on the car's warp stiffness, but I don't see how the chassis is as free from the problem as he claims.
In constant cornering, if there was no 'handling balance' due to LLTD or RMD and 50/50 weight distribution, in that constant corner, the chassis have no 'warp' forces acting on it (and so would not need to be THAT torsionally rigid), right?
If this is the case, what happens in the transient, notably turn in? There is axle warp when this occurs, front axle relative to rear. But if there is no resistance to this, would there effectively be no means of adjusting the transient handling balance, as you could with say the dampers on a normal corner spring set up?
#45
Posted 04 March 2012 - 06:48
Check the advese camber angles. Cheapy that picture is a case for beam axles if ever there was one.Wouldn't hurt this branch of racing to look at these systems as the reason they don't run anti roll bars is a paranoia about side to side travel robbing ..
#46
Posted 06 March 2012 - 06:56
Interesting read. A few interesting concepts and opinions. In particular the concept of a soft "twist mode" and its application to race car suspension. Love to hear an opinion from one of our suspension experts - Greg? Dave?Just been reading the SAE paper below, has anyone read it before? If so, what do you think of it?
Erik Zapletal, Balanced Suspension - http://www.scribd.co...nced-suspension
#47
Posted 06 March 2012 - 10:47
In particular the concept of a soft "twist mode" and its application to race car suspension.
Yeah, I suppose you mean as opposed to it's application in an off-roader which has visibly more wheel travel and a frequently more excited 'crossing of the axles'?
In my mind a soft twist mode would still be of benefit. The wheel travel on a race car is designed for the biggest bumps seen on the track I presume, and the stiffness is set accordingly.. Whilst the bumps seen aren't as big, because it is stiffer it takes a much smaller bump to affect the normal tyre loads drastically enough to affect it's cornering ability. I'm sure there is more to consider though..
Interestingly my uni vehicle dynamics lecturer was not completely familiar with the concept of a 'warp' mode, even though it seems like something quite useful to consider to me. Maybe it was just called other things in his day though, and it's not such a new consideration at all.
Love to hear an opinion from one of our suspension experts - Greg? Dave?
Me too!
#48
Posted 06 March 2012 - 16:06
#49
Posted 06 March 2012 - 16:33
That's a very cool read, thanks for posting that cheapracer. Not sure how much you read of it but I have some thoughts..
I found it interesting how he is using a ladder chassis because of the system's claims that it has no warp stiffness. I understand how the torsional rigidity requirement of a chassis is usually based on the car's warp stiffness, but I don't see how the chassis is as free from the problem as he claims.
In constant cornering, if there was no 'handling balance' due to LLTD or RMD and 50/50 weight distribution, in that constant corner, the chassis have no 'warp' forces acting on it (and so would not need to be THAT torsionally rigid), right?
If this is the case, what happens in the transient, notably turn in? There is axle warp when this occurs, front axle relative to rear. But if there is no resistance to this, would there effectively be no means of adjusting the transient handling balance, as you could with say the dampers on a normal corner spring set up?
When the car turns in and the front steered wheels start to accelerate laterally before the rears do, weight transfer occurs on the front axle but not the rear and by one definition, this is axle warp. If there is not resistance to this, what happens?
Edited by CWA, 06 March 2012 - 16:33.
#50
Posted 06 March 2012 - 18:35
This is what I was wondeing before, as I commented above.. (in repsonse to cheapy's link which is different, but the issues raised with 'low warp stiffness' still apply I feel)
When the car turns in and the front steered wheels start to accelerate laterally before the rears do, weight transfer occurs on the front axle but not the rear and by one definition, this is axle warp. If there is not resistance to this, what happens?
there are suspension springs , they load up