lynxbilder
|
| posted on 18/4/07 at 03:17 AM |
|
|
Chassis torsion question
I am in the design stages of a mid-engine 2 seat road car and have questions about chassis torsional stiffness. It seems to me that if suspension
pickup points are mounted to "non-deflecting" bulkheads in front and rear and the bulkheads twist relative to each other, suspension
geometry would remain as designed. The front might be rolling in a corner at 2 degrees and the back at 3 but if both of those conditions resulted in
acceptable camber change, a stable roll center, etc. I don't see why that would make the car handle poorly. I can see why that would make
spring/ARB changes not too effective but I don't see the concern with suspension geometry.
But people who know say you need a rigid chassis to keep the geometry intact so who am I to question!!
|
|
|
|
|
turbodisplay
|
| posted on 18/4/07 at 06:14 AM |
|
|
Any flex introduces unknows, if you can account for this no problem. I believe it is the spring/ shock mount is the most important as this wil take
the most force, the inner suspention mounts take less force.
Darren
|
NOTE:This user is registered as a LocostBuilders trader and may offer commercial services to other users
|
britishtrident
|
| posted on 18/4/07 at 06:33 AM |
|
|
If chassis flex was so bad it affected suspension geometry it would cause so many other concerns including fatigue failure of welds that the affect on
suspension geometry would be of little concern.
The BIG issue is the effect of torsional stiffness on how much control you have over roll couple distribution.
In short if you want the car to be predictable and easy to set up make it as stiff as you can.
[Edited on 18/4/07 by britishtrident]
|
|
|
v8kid
|
| posted on 18/4/07 at 06:43 AM |
|
|
I'll go with that. To ballance the car you need to control how much each end of the car contributes to resisting roll, for oversteer stiffen the
front for understeer stiffen the rear. If each end of the car is twisting with respect to each other it makes the changes unpredictable - in some
cases reversing the logic.
In theory it could handle acceptably but if you would ever be able to find the best setup is doubtfull.
Best just to make it stiff - its no big deal now with FEA and all the info on the web.
Of course there is another option with very soft springs torsional stiffness is less important but long travel suspension is notorious for its poor
geometry - best just make it stiff
[Edited on 18/4/07 by v8kid]
|
|
|
birt
|
| posted on 18/4/07 at 08:50 AM |
|
|
quote:
Originally posted by v8kid
I'll go with that. To ballance the car you need to control how much each end of the car contributes to resisting roll, for oversteer stiffen the
front for understeer stiffen the rear.
[Edited on 18/4/07 by v8kid]
You are right about the effect of torsional stiffness on front/rear roll stiffness distribution and yes, it is an important consideration. However
you have got mixed up with what does what.
If the front end has more roll stiffness than the rear then, as lateral aceleration increases, the front will react more load and hence offer less
overall grip.
This is because a tyre will exhibit a non linear increase in lateral force over a certain level of normal load. Since the rate drops off the combined
effect of both tyres is less than if they were both operating in the linear regime.
Therefore if you stiffen the front (or soften the rear) you will get more UNDERSTEER and if you stiffen the rear (or soften the front) you will get
more OVERSTEER.
|
|
|
Bob C
|
| posted on 18/4/07 at 09:24 AM |
|
|
A chassis twisting in torsion is acting as another undamped spring inthe suspension. It's the lack of damping (control) that's the
problem.
Bob
|
|
|
v8kid
|
| posted on 18/4/07 at 11:21 AM |
|
|
Birt I agree with you - what I said can be read two ways - sloppy of me. Although the chassis is an undamped spring if it is stiff enough in relation
to the suspension the relative movement is insignificant. The chassis as a spring generally would only be significant when either the chassis was very
flexible of the suspension was very stiff. F1 cars have very stiff suspension to cope with aero loads but thats well outside my experience.
I suppose if someone had a flexible chassis, wings, spoiler and diffuser the chassis flex would be a problem. better just make it stiff.
|
|
|
lynxbilder
|
| posted on 18/4/07 at 01:37 PM |
|
|
OK all of this makes sense. If the car is oversteering for example and you stiffen the front end, if the chassis is not stiff enough it will twist
away and the front end will have to react to spring + chassis rates, not to mention damping problems.
So chassis stiffness should be related to weight and downforce generated.
|
|
|
Doug68
|
| posted on 18/4/07 at 01:58 PM |
|
|
The reality is its virtually impossible to make the frame too stiff.
Essentially it needs to be as stiff as you can design for the weight you're prepared to carry and the materials you are going to use.
Some people subscribe to 'controlled flex' as a theory but basically thats complete rubbish.
|
|
|
ScotJebus
|
| posted on 18/4/07 at 08:59 PM |
|
|
aye, for a car you want it to be a stiff as you possiably can make it, the confrolted flex donnt come in untill you thinking motorbikes and they got a
different set of rules for a lot of this stuff
|
|
|
JoelP
|
| posted on 18/4/07 at 09:19 PM |
|
|
your chassis is a collection of points being forced in different directions. With mine i reduced the number of points (ie combine the suspension
pickup points with an engine or diff mount, the seat mount beside the tunnel bracing) and then link the points as directly as possible. If you cant
decide whether a link is needed (i had it today when pondering a line between engine mount and roll cage) just do it in smaller tubing, for a lower
weight penalty.
|
|
|
lynxbilder
|
| posted on 18/4/07 at 10:31 PM |
|
|
Mine is a self designed fiberglass tub using homebuilt airplane techniques. Fiberglass is soft compared to carbon and any type of metal so it has to
use fat sidepods, etc. and it is still tough to get past about 3000 ft lb per degree. Suspension points will be on bulkheads with forces in shear so
nothing can move there really at all unless it tears.
|
|
|
Doug68
|
| posted on 19/4/07 at 08:30 AM |
|
|
Now you've gone and done it!
I think a bunch of the regulars will be along soon to tell you thats a mad idea.
Of course it should work providing the design is good and its made correctly just like any other material / technique.
'Hard' & 'Soft' I'm not sure what thats supposed to refer to. Whether your chassis ends up stiff or not depends
upon the Youngs modulus of elasticity of the material used combined with the area or more properly the effective second moment of the area of the
chassis.
Another point might be that aircraft are deliberately designed to flex as a method of absorbing loads (just what you don't want to happen in a
car) and somewhat perversely the nature and magnitude of loads on aircraft appear to be more readily estimated that those that road going vehicle will
be submitted too.
You might find this article interesting:
http://www.mscsoftware.com/support/library/conf/wuc96/07b_brow.pdf
|
|
|
whittlebeast
|
| posted on 19/4/07 at 11:02 AM |
|
|
Almost every post above is great info. Bottom line is it has to be very strong in torsion. It has everything to do with making the springs,
shocks(dampers), and bars do what they are designed to do in a corner. I shoot for the weight of the car in lbs *3 so a 1000 lb car needs 3000
ft-lbs/deg stiffness. a book chassis is about 1/3 what it needs to be. My new chassis is about 9000.
AW
[Edited on 19/4/07 by whittlebeast]
|
|
|
v8kid
|
| posted on 19/4/07 at 11:46 AM |
|
|
I know I'm digressing a bit but how do we measure the stiffness? I've crunched the numbers through my FEA programme with a satisfactory
outcome but reality is often different from computer programmes.
The difficulty is in getting repeatable results - how do we translate the constraints used in the computer model into reality so we can measure the
torsional deflection. I have a sneaky feeling that the tie down arrangements will add to the stiffness unless done correctly.
|
|
|
lynxbilder
|
| posted on 19/4/07 at 03:29 PM |
|
|
Doug68: By "soft" I was referring to modulus of elasticity, and the low number means large cross sections in the monocoque as you noted.
I am using scale modeling to estimate its rigidity and I am in the neighborhood of 3000 to 3500 ft lb per degree but I need a factor of safety because
my modeling is probably inaccurate. Probably looking at a 200 lb chassis for a mid-engined 2540mm wheelbase.
And thank you for the article link!
I chose fiberglass because I like to work with it and it works well for the home shop. I like the idea of sculpting the contours of the lower
bodywork and interior right into the chassis.
V8kid: Take a look at "Chassis Engineering" by Herb Adams. He takes a very non-mathematical look at chassis design (which is of course,
risky!) but he has a section on torsion testing the final chassis. Very "seat of the pants" as we say here in the States. (Wonder what
that means?)
[Edited on 19/4/07 by lynxbilder]
[Edited on 19/4/07 by lynxbilder]
|
|
|
Aloupol
|
| posted on 19/4/07 at 05:12 PM |
|
|
Another concern about poor stiffness frame: suspension movements are damped, when chassis distorsions aren't.
If you could put a damper on each chassis tubes and nodes to lower the movements then the "soft" chassis will perhaps handle well.
And if you remove the dampers (let only the springs) from your perfect engineered suspension, the car will handle as a barrow.
|
|
|
chriscook
|
| posted on 19/4/07 at 07:11 PM |
|
|
It's just as important to get the constraints right in the FE model. There should be no more than 6 constraints otherwise it is over constrained
and will have the effect of falsely stiffening the chassis
Chris
quote:
Originally posted by v8kid
I know I'm digressing a bit but how do we measure the stiffness? I've crunched the numbers through my FEA programme with a satisfactory
outcome but reality is often different from computer programmes.
The difficulty is in getting repeatable results - how do we translate the constraints used in the computer model into reality so we can measure the
torsional deflection. I have a sneaky feeling that the tie down arrangements will add to the stiffness unless done correctly.
|
|
|
lynxbilder
|
| posted on 19/4/07 at 09:58 PM |
|
|
quote:
Originally posted by Aloupol
Another concern about poor stiffness frame: suspension movements are damped, when chassis distorsions aren't.
My chassis may actually be damped somewhat, I had not thought of that. The fiberglass layup surrounds foam block formers that should have some
damping qualities. Cannot rely on that but I am hoping it will reduce noise.
|
|
|
britishtrident
|
| posted on 20/4/07 at 06:39 AM |
|
|
quote:
Originally posted by chriscook
It's just as important to get the constraints right in the FE model. There should be no more than 6 constraints otherwise it is over constrained
and will have the effect of falsely stiffening the chassis
Chris
quote:
Originally posted by v8kid
I know I'm digressing a bit but how do we measure the stiffness? I've crunched the numbers through my FEA programme with a satisfactory
outcome but reality is often different from computer programmes.
The difficulty is in getting repeatable results - how do we translate the constraints used in the computer model into reality so we can measure the
torsional deflection. I have a sneaky feeling that the tie down arrangements will add to the stiffness unless done correctly.
Type and position of constraints and loads together with choosing the correct element type is the key.
|
|
|
britishtrident
|
| posted on 20/4/07 at 06:44 AM |
|
|
quote:
Originally posted by lynxbilder
Doug68: By "soft" I was referring to modulus of elasticity, and the low number means large cross sections in the monocoque as you noted.
I am using scale modeling to estimate its rigidity and I am in the neighborhood of 3000 to 3500 ft lb per degree but I need a factor of safety because
my modeling is probably inaccurate. Probably looking at a 200 lb chassis for a mid-engined 2540mm wheelbase.
And thank you for the article link!
I chose fiberglass because I like to work with it and it works well for the home shop. I like the idea of sculpting the contours of the lower
bodywork and interior right into the chassis.
V8kid: Take a look at "Chassis Engineering" by Herb Adams. He takes a very non-mathematical look at chassis design (which is of course,
risky!) but he has a section on torsion testing the final chassis. Very "seat of the pants" as we say here in the States. (Wonder what
that means?)
[Edited on 19/4/07 by lynxbilder]
[Edited on 19/4/07 by lynxbilder]
GRP monocoques can be very stiff but need a great care to get right or the stifness can be thrown away.
best example of good practice is probably the Davrian Mk7 or its near relative the Darrian.
[I] “ What use our work, Bennet, if we cannot care for those we love? .”
― From BBC TV/Amazon's Ripper Street.
[/I]
|
|
|
Fred W B
|
| posted on 20/4/07 at 09:08 AM |
|
|
Some good discussion
here
Cheers
Fred W B
[Edited on 20/4/07 by Fred W B]
|
|
|
JonBowden
|
| posted on 20/4/07 at 09:08 AM |
|
|
I'm quite keen on the idea of monocoque cars like your design.
This site has some files relating to measuring chassis stiffness :
http://locost7.info/mirror/aussiemods.php
You probably already know, but this car a a fibreglass monocoque :
http://www.gtmcars.com/
Jon
|
|
|
lynxbilder
|
| posted on 20/4/07 at 05:37 PM |
|
|
All the comments and links have been great, I am learning a lot. If anyone is interested here are a couple of photos of my chassis tub model in 1/10
scale styrene:
I have learned a lot just twisting this thing. For example the foot box with its thin sheet over the top is very rigid. Total surprise, I thought a
bulkhead would be in order across the footbox rear. Much of the tub flexibility happens behind the firewall bulkhead so the furthest aft shear panel
(inaccurately shaped on the model) was necessary to stop the rear portions of the sidepods from twisting. Without that shear panel there definitely
would be suspension pickup point variation in twist. The top panels of the sidepods behind the firewall are narrow and flex in shear so they need to
be at least double thickness.
I am getting ready to build a more detailed model with some changes that will further help rigidity.
I have built full size fiberglass portions of the chassis for testing to get relationships between reality and the model and I am going to do a test
to determine if the odd shape of the rear sidepods will buckle the fiberglass off the foam under load.
|
|
|
JonBowden
|
| posted on 20/4/07 at 08:26 PM |
|
|
That's quite similar to a design I have in mind but using aluminum
Jon
|
|
|
