Empirical determination fo the chassis torsional stifness.

[MRDAGERUS]

I'd like to find out what is the torsional stiffness of the chassis/body assembly of the car I'm working on.
The engine and transmission is out for repairs.
Since the body sits on axle stands already, no engine in it, I'm thinking about proceeding in the following manner:
1. Level it on 4 stands and measure the height
2. Keep it on axle-stands on tree corners, load the corners with ballast to keep it from jumping of ..and
3. Remove one stand and put known ballast on one corner that is not supported.
4. Now, measure with a 'straight edge' how much the level on this corner drops down for a given ballast weight
Since I'm looking for Nm/deg result, I need to measure the "arm", which would be(?) the distance from the unsupported corner to the stand diagonally opposite to my measuring point .
Do I have it right?

 

[MRDAGERUS]

Correction :

I skipped one (too obvious) step (3.a)
1. Level it on 4 stands and measure the height
2. Keep it on axle-stands on tree corners, load the corners with ballast to keep it from jumping of ..and
3. Remove one stand and measure the height of the "drooping" corner
3.a. put known ballast on one corner that is not supported.
4. Now, measure with a 'straight edge' how much the level on this corner drops down for a given ballast weight

 

[MRDAGERUS]

OK, I guess, I answered my question:

First, place car/chassis on axle stands at the points corresponding to the suspension attachment points; wheels off the ground.
1. Level it on 4 stands.
2. Keep it on axle-stands; load the rr corners with ballast to keep it from jumping off.
If you’re testing fully dressed car the weight of the engine and transaxle will keep it pinned down.
3. Remove one (front driver sd) stand and a wheel and measure the distance from the fixed chassis/suspension point to the ground.
If you’re to use a dial indicator, set it up and zero it.
4. Put known ballast on one corner that is not supported**, [F].
5. Now, measure how much the level on this corner drops down for a given ballast weight [D]
6. Measure the distance to the opposite side stand (closest to the measuring point), [L].

The torsional rigidity can be calculated by finding the torque applied to the frame and dividing by the
angular deflection.

K^2= FxL/tan(D/L)

This method of frame testing is relatively straightforward and the advantage is the frame stiffness can be determined without including the suspension components. The primary disadvantage is the artificially created load paths do not load the frame in the same manner as on the track. Also, the choice of what rear nodes to fix and what front nodes to apply the load can affect the results significantly. For this reason a whole car chassis torsion test is the preferred method for capturing the true vehicle stiffness.

Chassis testing, however, has very high non-linearity in the early stages. For small forces gaps in the suspension and compression of various bearing elements occurs. As these gaps are closed and bearing friction is overcome the slope of the load deflection curve becomes linear.
** For this reason, it is necessary to map the force displacement characteristic of the structure, rather than finding one stiffness value. To get better data small steps of load should be applied, and the corresponding displacement measured. It is also interesting to note that the force deflection curve has some hysteresis. To accurately gauge this characteristic, it is helpful to add or remove the load in finite steps and record the deflection. This will build a load deflection “path” that rises and then falls again. At high loads the deflection is linear. This represents the deformation of the elastic frame and suspension members after gaps are closed. Therefore one should start at loads higher than 40-50 lbs. Then, load it in 10 lbs increments.

 

[cibachrome]

In theory you are doing it OK. In practice, there are some issues:

1) Without the motor, the stiffness won't be the same as when the motor is installed because the engine and transmission contribute to the body/frame stiffness.

2) The nonlinearity is actually axial, not necessarily rotational. This is because the motor, the doors, the roof, the windscreens, the rear wheel kick-up and even the bumpers cause the displacements to vary along the wheelbase. Its the rear of the car that usually is the softest.

Try taking dial indictor readings every 6 inches or so along the wheelbase. Also include points ahead of the front axle and behind the rear axle.

This may help you decide where braces might be added to 'improve' your torsional rigidity.

 

[GregLocock]

Although what you are measuring is useful, it is a combination of vertical bending and torsion. If you want to measure torsion alone then you need to twist the body (ie push up on the other corner), not just stand on one corner.

However from a practical perspective your method is simple and useful. Here's what ACBC wrote:

"35. Measured over the wheelbase, one axle restrained and lifting one corner of the opposite axle, torsional stiffness of the complete structure must exceed 2500 foot lbs [3456 Nm] per degree. The variation in stiffness per foot of wheelbase must not exceed +/- 10% or 15% between two adjacent sections otherwise fatigue failures are likely."

We can argue about numbers (that value is low for a road car, it was very high for a 1960s F1 car).

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[dynatune]

Executing correctly vehicle torsion measurements is really difficult for all the above mentioned reasons. Beyond that the quality of your "fixture" define in a significant matter the quality of the results. If you are trying to verify weather your stiffness is above the mentioned 3456Nm/° you might as well leave the effort. Nowadays anything below 25000Nm/° is considered uncompetitive, in production cars and certainly in race cars. It goes without saying that measuring such "stiff" systems do require very carefull procedures and very stiff fixtures in order to not measure "garbage". Most people relay nowadays on a proper finite element model of the car to determine those numbers, it is easier and more robust. If you really want to know, go that road. It will save you time and pain.

 

[GregLocock]

Another option is to put a cats cradle of string inside the car between all the obvious points and drive round and see which go slack. The Mazda technique.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[MRDAGERUS]

Thank you all for your valuable input.

I'm just trying to find a "quick and dirty" way of verifying Lotus Esprit torsional stiffness quoted by unknown source, see table below. Comparing to other Marques, it seems quite low. Unfortunately the back bone chassis type is not conductive to any stiffness record numbers.

Chassis Stifness:
Alfa 159 - 31.400Nm/degree
Aston Martin DB9 Coupe 27,000 Nm/deg
Aston Martin DB9 Convertible 15,500 Nm/deg
Aston Martin Vanquish 28,500 Nm/deg
Audi TT Coupe 19,000 Nm/deg
Bugatti EB110 - 19,000 Nm/degree
BMW E36 Touring 10,900 Nm/deg
BMW E36 Z3 5,600 Nm/deg
BMW E46 Sedan (w/o folding seats) 18,000 Nm/deg
BMW E46 Sedan (w/folding seats) 13,000 Nm/deg
BMW E46 Wagon (w/folding seats) 14,000 Nm/deg
BMW E46 Coupe (w/folding seats) 12,500 Nm/deg
BMW E46 Convertible 10,500 Nm/deg
BMW X5 (2004) - 23,100 Nm/degree
BMW E90: 22,500 Nm/deg
BMW Z4 Coupe, 32,000Nm/degree
BMW Z4 Roadster: 14,500 Nm/deg
Bugatti Veyron - 60,000 Nm/degree
Chrysler Crossfire 20,140 Nm/deg
Chrysler Durango 6,800 Nm/deg
Chevrolet Corvette C5 9,100 Nm/deg
Dodge Viper Coupe 7,600 Nm/deg
Ferrari 360 Spider 8,500 Nm/deg
Ford GT: 27,100 Nm/deg
Ford GT40 MkI 17,000 Nm/deg
Ford Mustang 2003 16,000 Nm/deg
Ford Mustang 2005 21,000 Nm/deg
Ford Mustang Convertible (2003) 4,800 Nm/deg
Ford Mustang Convertible (2005) 9,500 Nm/deg
Jaguar X-Type Sedan 22,000 Nm/deg
Jaguar X-Type Estate 16,319 Nm/deg
Koenigsegg - 28.100 Nm/degree
Lambo Murcielago 20,000 Nm/deg
Lotus Elan 7,900 Nm/deg
Lotus Elan GRP body 8,900 Nm/deg
Lotus Elise 10,000 Nm/deg
Lotus Elise 111s 11,000 Nm/deg
Lotus Esprit SE Turbo 5,850 Nm/deg(?)
Maserati QP - 18.000 nm/degree
McLaren F1 13,500 Nm/deg
Mercedes SL - With top down 17,000 Nm/deg, with top up 21,000 Nm/deg
Mini (2003) 24,500 Nm/deg
Pagani Zonda C12 S 26,300 Nm/deg
Pagani Zonda F - 27,000 Nm/degree
Porsche 911 Turbo (2000) 13,500 Nm/deg
Porsche 959 12,900 Nm/deg
Porsche Carrera GT - 26,000Nm/degree
Rolls-Royce Phantom - 40,500 Nm/degree
Volvo S60 20,000 Nm/deg
Audi A2: 11,900 Nm/deg
Audi A8: 25,000 Nm/deg
Audi TT: 10,000 Nm/deg (22Hz)
Golf V GTI: 25,000 Nm/deg
Chevrolet Cobalt: 28 Hz
Ferrari 360: 1,474 kgm/degree (bending: 1,032 kg/mm)
Ferrari 355: 1,024 kgm/degree (bending: 727 kg/mm)
Ferrari 430: supposedly 20% higher than 360
Renault Sport Spider: 10,000 Nm/degree
Volvo S80: 18,600 Nm/deg
Koenigsegg CC-8: 28,100 Nm/deg
Porsche 911 Turbo 996: 27,000 Nm/deg
Porsche 911 Turbo 996 Convertible: 11,600 Nm/deg
Porsche 911 Carrera Type 997: 33,000 Nm/deg
Lotus Elise S2 Exige (2004): 10,500 Nm/deg
Volkswagen Fox: 17,941 Nm/deg
VW Phaeton - 37,000 Nm/degree
VW Passat (2006) - 32,400 Nm/degree
Ferrari F50: 34,600 Nm/deg
Lambo Gallardo: 23000 Nm/deg
Mazda Rx-8: 30,000 Nm/deg
Mazda Rx-7: ~15,000 Nm/deg
Mazda RX8 - 30,000 Nm/degree
Saab 9-3 Sportcombi - 21,000 Nm/degree
Opel Astra - 12,000 Nm/degree
Land rover Freelander 2 - 28,000 Nm/degree
Lamborghini Countach 2,600 Nm/deg
Ford Focus 3d 19.600 Nm/deg
Ford Focus 5d 17.900 Nm/deg


In addition, a friend of mine, (also a Lotus owner), has started a tubular replica design of his own and he is shooting for much stiffer chassis. He was inspired by Jan Hoel's Lotus Esprit S3R development. Please take a look at his Facebook picture gallery

https://www.facebook.com/media/set/...1404582&set=pu.178516808826230&type=3&theater

I'm impressed!
And you?

 

[GregLocock]

The Lotus figure is correct, if you have the stiffening braces running from the intercooler's radiator duct to the chassis. If you don't have them then knock 20% off that figure. I put the braces in there, for the introduction of the intercooled turbo model. My mechanic Big John fabricated the first set. I don't know if they went in across the board in production, I think so as they had such a positive effect on handling.

Basically the issue is that the spine chassis is not using the full width of the car to develop torsional stiffness, and Chapman had decreed after the original Elite's problem with body cracking that the glass fibre tub was to be an unstressed unit.

Since he was dead or hiding in South America by the time I was working on the car, and we had a reasonable durability test on pave to check things, we were confident that the GF could be used as a torsional stiffener.

If the body is overstressed you'll probably notice gel coat cracking over the front wheels.

The effect on handling is profound, with the braces you should get a lot more understeer, it'll need new front bar and shocks, and probably springs. John Miles did the tuning, from memory.

Other cars that had surprsingly low torsional stiffnesses in the eighties were the Porsche 911 cabriolet, and some BMWs.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[MRDAGERUS]

When using all my power, one area that I can feel is body torsional stiffness, or should I say, lack of it. The interesting aspect is that it handles very well, but you can feel it twists out of the bends. I believe Lotus may have compensated for this in the springing on later cars, which is why they exhibit less harsh ride and a bit more of predictability.