Measuring chassis stiffness

[singlecoil]

One of the aims, when producing a chassis, as I understand it, is to achieve a satisfactory level of stiffness.

Can anyone suggest a good method by which the stiffness of a chassis might tested, and what would be a good figure to aim for?

I assume the basic idea is to rigidly secure one end of the chassis, then to subject the other end to a measured torsional force, and measure the deflection, presumably with a dial test indicator.

[p1doc]

i tend to get fat bloke to sit on end of car if it breaks not strong lol
i know regarding the murtaya they did loads of research to keep chassis strong but not too sure how they did it,once load is applied can you not use x-rays etc to show potential cracks as with metal but for grp not sure of best method
martin

[singlecoil]

p1doc said:

i tend to get fat bloke to sit on end of car if it breaks not strong lol
i know regarding the murtaya they did loads of research to keep chassis strong but not too sure how they did it,once load is applied can you not use x-rays etc to show potential cracks as with metal but for grp not sure of best method
martin

Engineers (I'm not one myself) draw a distinction between stiffness and strength, and I think what you are talking about there is strength. Stregth is good, of course, but I'm not going to be able to measure it unless I can actually break the chassis.

[Sam_68]

Unless you have access to a proper chassis testing rig, standard solution (which I have used before) is broadly as follows:

• Bolt the suspension mounting points at both sides of one end of the car rigidly to something completely immovable (a large masonry building is good ).
• Rest one opposite corner of the car on a suitable fulcrum
• Bolt a bloody great lever arm across the suspension points of the 'free' end of the car, to enable you to apply a large, known mass a measurable distance from the centreline of the car.
• Use a dial gauge a measurable distance from the fulcrum point, to enable you to measure the deflection when you apply said large known mass to said bloody great lever arm.

Bob's your uncle...

It's not perfect, but it'll give you an adequately accurate indication of the torsional stiffness.

[p1doc]

still reckon fat bloke idea lol but above post bit more scientific!
see what you mean re stiffness vs strength though!
martin

[singlecoil]

Sam_68 said:

It's not perfect, but it'll give you an adequately accurate indication of the torsional stiffness.

I was thinking something along those lines, so thanks.

Any thoughts as to what numbers should be aimed for?

[Sam_68]

singlecoil said:

Any thoughts as to what numbers should be aimed for?

It depends what you're designing. As a general rule, light cars with supple springs don't actually need to be all that stiff. Heavy cars, or very stiffly sprung 'proper' race cars that need to maintain a level platform for aero, need to be much stiffer.

For reference, the spaceframe on a Sylva is around 1,250 lb.ft/degree; a current Caterham (one of the better spaceframe 'Sevens') is probably about 2,200 - 2,400lb.ft/degree (I tested a de Dion chassis back in the '90's and got 1,900-2000 - they've improved the design since then, but they're very cagey about quoting figures because it does them no favours at all when the numbers are compared to those you'e expect on a modern saloon car).

The Westfield FW400 monocoque was quoted as 4000 lb.ft/degree (and weight for the complete tub was quoted as 50 kilos), though I can't honestly confirm either figure).

[Sam_68]

p1doc said:

still reckon fat bloke idea

A fat bloke makes a very effective large, known mass to balance on the end of your bloody great lever arm.

[gtmdriver]

I found this on an Australian forum

'Torsional test is set up with the chassis mounted rigid at the rear, no suspension interference, front tie rail is pivoted at the centre,no suspension interference, tortional twist is calculated from the centre of the front tie rail at a measure of 4000nm dialed for four cylinder enines and 6000nm dialed for six or V8s. the twist has to sustain that weight with a maximum of two mm deflection.'

You might need two fat blokes if you have a V8?

[Sam_68]

gtmdriver said:

You might need two fat blokes if you have a V8?

Only if you're Australian.

Fortunately there isn't any minimum torsional stiffness requirement in the UK IVA test yet.

The Australian requirements are quite stiff (pardon the pun)... an Ultima or most UK Cobra replicas would fail them dismally.

eta: there's been a discussion going on for quite some time over on PistonHead's Ultima forum here, including some pics that Mr Singlecoil might find interesting.

Don't confuse a (potentially) 1,000bhp, 1-tonne, aero-stability-sensitive Ultima with a <half tonne 'Seven', though.

Edited by Sam_68 on Tuesday 16th November 18:40

[elwe]

I seem to remember that when a GTM Libra tub was tested for Australian type approval it came out at around 14,000 nm/degree of deflection (about 10,000 lb.ft). I also seem to remember that a Lotus Elise is quoted at around 10,000 nm/degree (about 7,500 lb.ft).

[Sam_68]

Yep. GTM actually quote figures on their sales specification for the car; they claim 14,000 Nm/degree for the coupe and 12000 for the convertible.

[ZX9RIOT]

the sylva riot chassis passed the australian 4000nm test

[Wanchaiwarrior]

check out this page, some pics and blurb for Oz testing

http://www.lescanfield.info/ApplicationTobuild.htm...

[p1doc]

Sam_68 said:

p1doc said:

still reckon fat bloke idea

A fat bloke makes a very effective large, known mass to balance on the end of your bloody great lever arm.

exactly lol
interesting thread though-i will have to refind my murt figures for comparison!
martin

[stig mills]

We sent a Rocket to Australia and they used this rig to ensure it complies with the Australian torsional rigidity requirements of 4000 ft lbs per degree of deflection. It passed and is now road legal down under. He used a full roll cage to achieve the same results with a 7 type and failed with the Atom, he tells me there are no Atoms on the roads in Australia.

[Russ Bost]

OK - now I'm confused (nothing new there then!), looking at that rig, they've replaced the shox & springs with solid links & bolted the drive flanges to the rig - but surely that's not testing just chassis stiffness, but also bits of suspension that would never normally be loaded like that as they would be married to a spring & damper with a measure of compliance???
I would have expected them to bolt the bare chassis down at one end & then twist it from the opposite end.
Does anyone know the exact purpose of why the test is carried out in this way (I guess they are testing everything from the contact patch of the tyre inwards, but in which case you would think the wheels would also be included???) & is this just the way they do it in Aus, or is it the same everywhere?

[stig mills]

I see what you mean Russ, the poly bushes will definitely compress to make it look worse than it is. I was told that in Australia all components need to be tested but I assume they replaced the poly with steel for the test.
A crude method would be say a 10ft beam as a lever at one end of a chassis with a dial gauge and the other end bolted to something rather solid. The beam end would need support at the centre of the chassis (fulcrum point). Divide the load by 10 and allow for the beam bending, the rest is chassis twist expressed in ft lbs per degree of deformation assuming you have converted the dial gauge reading to the change in angle.
I always laugh when TV presenters refer to a chassis twisting when they test a soft top version of a car. How do they feel the difference between suspension the movements up and the body shell twisting?

[NeilE]

ZX9RIOT said:

the sylva riot chassis passed the australian 4000nm test

See here:
http://sylva-autokits.blogspot.com/2009/08/good-ne...

Edited by NeilE on Wednesday 17th November 15:34

[Sam_68]

stig mills said:

I see what you mean Russ, the poly bushes will definitely compress to make it look worse than it is.

This is one of the reasons that I don't get too fixated on chassis stiffness.

On the one hand, I can understand the Australian testing method of replacing the dampers with rigid links and testing everything, because it highlights any flexibility in the locating links as well as the chassis, which is a valid concern.

But in practice all cars function as if they were several springs working in series; first of all you have the tyres, then you have the road springs and bushes (effectively acting in parallel), then you have the chassis.

Now, when you actually do the sums to work out which bits deflect by how much, even a half-reasonable chassis, on a light car, deflects so little in comparison to the sum of the deflections from the tyres, springs and bushes, that it's really not worth worrying about as much as some people do. It's important that you have enough stiffness, but it's not the be-all-and-end-all.

Certainly, the GTM's torsional stiffness is overkill (though not to criticise GTM: it obviously has the important side benefit of giving what is a very compact car a very strong, crash-resistant passenger cell, and since the car isn't overly heavy, why not?).

stig mills said:

I always laugh when TV presenters refer to a chassis twisting when they test a soft top version of a car. How do they feel the difference between suspension the movements up and the body shell twisting?

To be fair, a lot of the problem with torsional stiffness isn't so much the 'you've got to have a stiff chassis to let the suspension work' nonsense that we've all been indoctrinated with; it's the fact that subjectively Human Beings are actually incredibly sensitive when it comes to detecting the very slight judders, rattles and squeaks that you get from even the tiniest relative movement between components.

I've got an early MX5 in the collection at the moment which subjectively feels like the least torsionally stiff car I've ever owned and fitting some basic bracing made a very noticeable difference to the frequency of the vibrations I wss getting back through the chassis. Objectively, I know that it's stiffenss when new was around 4,400 lb.ft/degree, which is more than an Elan and considerably more than most spaceframe 'Seven' types, including my Sylva, so even allowing for the fact that it will have relaxed quite a bit with age, it's not going to be as bad as all that - and certainly the steering and handling are still commendably sharp and precise.

The MX5 suffers, I think, because it has a lot of secondary components (doors, dashboard, hood etc.) that rattle and bang about and give the impression that the whole structure is very 'loose', whereas the Sylva has nothing that isn't well bolted, bonded or strapped in place, therefore feels incredibly taut despite the fact that in reality it really isn't.

Sorry... very long winded way of saying that once you've gone beyong the point of being functionally stiff enough, torsional stiffness is probably more important in terms of NVH than it is in terms of performance and handling.

Edited by Sam_68 on Wednesday 17th November 18:48