53 replies to this topic

[BritishV8]

One of my friends recently showed me paperwork from 1969, when he purchased his Lola T190 (Formula A / Formula 5000) chassis. A short list of specs was included. Item 1: torsional rigidity of 300,000 foot pounds per degree per foot. Not bad for a seventy pound aluminum monocoque, right? But was the Lola's stiffness better or worse than a McLaren M10B?

My main question is: where can I find torsional rigidity data for various race car chassis?

If a census of cars and specs doesn't already exist, can we start one here? I think it would be extremely interesting to compare cars - model to model, racing class to racing class, builder to builder, or even technologically (generation to generation). Do you guys have any torsional rigidity measurements to add to the census?




Just in case you're curious... I'll attach a couple snapshots I took of the Lola chassis:

[Tony Matthews]

One of my friends recently showed me paperwork from 1969, when he purchased his Lola T190 (Formula A / Formula 5000) chassis. A short list of specs was included. Item 1: torsional rigidity of 300,000 foot pounds per degree per foot. Not bad for a seventy pound aluminum monocoque, right? But was the Lola's stiffness better or worse than a McLaren M10B?

300,000 foot pounds sounds a bit high to me! 30,000 more like, I suspect. I await a truck-load of derision... If I'm wrong I shall punish myself with a glass of wine.

[DaveW]

Apologies, but I'm not sure about your units, or your quoted value. I suspect you need to specify what was measured & how it was measured. I think the free-free, spindle-spindle torsional stiffness of a modern carbon tubbed race vehicle is only a small fraction (perhaps 2 or 3 %) of the value you quoted.

[Tony Matthews]

And I'm beginning to wonder if I didn't mean 3,000 f/lb, I used to know the figure for the Lotus 25 but - memories... I should look it up!

[BritishV8]

You might think it wouldn't be hard to come up with a standard measurement technique. Alas, the Lola spec sheet didn't say a word about how they got their number. (Could it by a typo? Possibly.) The exact verbiage follows:

Chassis: Aluminum alloy monocoque - bonded and riveted construction. Torsional rigidity of 300,000 foot pounds per degree per foot. Weight: 70 pounds

(Lower on the form they provided specs for wheelbase, track, etc.)

[Fat Boy]

Alas, the Lola spec sheet didn't say a word about how they got their number. (Could it by a typo? Possibly.)

Someone at the factory was taking the piss out of the original customer, who was probably anal about torsional rigidity. They wrote the number down on the paperwork for a laugh. It was never measured.

[gruntguru]

I have a question for all you chassis boffins. Is it unusual to use the units quoted here? I'm referring to the "per foot" at the end. This would make sense in some ways - to quote torsional stiffness relative to the length of the chassis wheelbase. I assume the torsional stiffness would be multiplied by the wheelbase (making the units ft.lb.ft/deg) which would make the number a property of the cross section.

[Tony Matthews]

I have a question for all you chassis boffins. Is it unusual to use the units quoted here? I'm referring to the "per foot" at the end. This would make sense in some ways - to quote torsional stiffness relative to the length of the chassis wheelbase. I assume the torsional stiffness would be multiplied by the wheelbase (making the units ft.lb.ft/deg) which would make the number a property of the cross section.

I noticed that - and it would make the number much higher, as lb/ft/inch would. When these figuers were normally included in press reports of new racing cars it was just ft.lb over the length of the chassis. I have a photo somewhere of a Penske chassis on a rig, fixed at the front, rear being subjected to a twisting force. Do you think I can find it...?

[BritishV8]

Maybe it would have been better in the first post if I hadn't even mentioned the T190. I clouded the issue. FWIW, the wheelbase of the T190 was only 92 inches. The tube-frame model that preceded it (Lola T142) and the improved model that proceeded it (T192) were both 98 inches. (Was the T192 tub really six inches longer, or wasn't a longer bellhousing part of the difference?) Nearly every reference on chassis design emphasizes the importance of torsional rigidity, but I don't recall seeing the torsional rigidity of historic racing cars compared in a QUANTIFIED manner.

[DaveW]

Is it unusual to use the units quoted here? I'm referring to the "per foot" at the end. This would make sense in some ways - to quote torsional stiffness relative to the length of the chassis wheelbase.

You have a point GG. I note also that the chassis weight was quoted as 70 lb suggesting, perhaps, that the stiffness was for the tub in isolation. If that was the case, then it would suggest that the engine bulkhead might have been constrained to be planar when making the measurement. If "per foot" applies, as suggested by GG, would the multiplier have been the length of the tub, or the vehicle wheelbase?

An alternative thought is that the units might have been ft.lb/radian.

Edited by DaveW, 16 November 2009 - 22:40.

[gruntguru]

An alternative thought is that the units might have been ft.lb/radian.

That would certainly explain the big numbers. Don't think I'd like to see a chassis twisted a whole radian.

[Greg Locock]

That's a rather disturbing unit, since it implies that making the car twice as long would double its stiffness. Springs in series actually add stiffness according to the inverse rule 1/kboth=1/k1+1/k2. Alternatively if you use compliancs then they just add.

[gruntguru]

That's a rather disturbing unit, since it implies that making the car twice as long would double its stiffness.

Ahhh - but if the section was the same but twice as long, the stiffness in N.m/deg would be halved but stay the same when expressed in N.m.m/deg since the latter is a property of the cross section, not the whole chassis.

I wonder how buses compare to cars on torsional stiffness.

[Tony Matthews]

I wonder how buses compare to cars on torsional stiffness.

Single- or double-deckers?

[Joe Bosworth]

BV8 commented:

¨If a census of cars and specs doesn't already exist, can we start one here? I think it would be extremely interesting to compare cars - model to model, racing class to racing class, builder to builder, or even technologically (generation to generation). Do you guys have any torsional rigidity measurements to add to the census?¨

After only 14 posts we now see why such a list doesn´t exist. It is like dyno numbers, if you haven´t seen the test personally don´t believe the data!!

Regards

[Greg Locock]

Ahhh - but if the section was the same but twice as long, the stiffness in N.m/deg would be halved but stay the same when expressed in N.m.m/deg since the latter is a property of the cross section, not the whole chassis.

I wonder how buses compare to cars on torsional stiffness.

The sensible unit would be degrees of twist per Nm of torque per metre of wheelbase.

Buses and CVs generally are very floppy compared with cars.

[gruntguru]

The sensible unit would be degrees of twist per Nm of torque per metre of wheelbase.

Torsional floppiness, floppidity, compliance?

Do two "per"s make an "of" - I assume your unit is deg/N.m.m which is the inverse of the unit under discussion.

[cheapracer]

lots of torsional twist figures quoted by people all over the place - many, especially 3D ones simply go from one end of the chassis to the other but in my mind the only place that matters is where the suspension load paths are to cause the twist in the first instance.

[BritishV8]

but in my mind the only place that matters is where the suspension load paths are to cause the twist in the first instance.

Makes sense...


Mark Donohue touched on the subject in "The Unfair Advantage" - specifically in the chapter about his AMC-powered Lola T330

I was hoping that the basic problem was too much flex in the chassis. When the next car arrived we set it up on our surface plate, replaced the springs with solid bars, and tested its torsional rigidity. Woody spent a week at it, anchoring one end, twisting the other end with hydraulic jacks, and measuring the deflection with dial gauges. It turned out to have a stiffness of about 4000 foot-pounds per degree, which ought to be perfectly adequate for a race car. We could see a little bending in the middle, though, so we added a couple of extra braces between the roll bar and cylinder heads.

I'm guessing they must have used some sort of load cell at the jack(s) to get the force measurement. The standard wheelbase for a T330 was 102" (8.50 feet) according to this reference: http://forums.autosp...mode=linearplus

Possibly Donohue's T330 may have been a little longer or shorter because of the length of his AMC engine (vs. the more commonly-used Chevy engine). Bellhousing lengths also vary, right? But if we assume 102", then 4000 foot-pounds per degree multiplies out to 34,000 foot-pounds per degree per foot. I'd bet that the "300,000 foot-pounds per degree per foot" from the first post above erroneously included an extra zero... Obviously I'm assuming a lot, but would it be unreasonable to infer that - by design - a T330 (circa 1973) was about ten percent stiffer than a T190 (circa 1969)?

Do you guys like photos? Here are some snapshots I took of the Brian Redman T332...

Notice that, unlike the T190/192, the tub on the T332 doesn't extend rearward along either side of the engine - and the T332 tub also has a MUCH wider cross-section:


Change engine block or bellhousing length, and you'll pretty much inevitably change wheelbase:

Edited by BritishV8, 18 November 2009 - 18:24.

[DaveW]

After only 14 posts we now see why such a list doesn´t exist. It is like dyno numbers, if you haven´t seen the test personally don´t believe the data!!

Here is an example torsional stiffness estimate:

An open-wheeled, mid-engined race vehicle was installed on Multimatic's four post rig. Springs/dampers were replaced by solid links, & wheel/tyres by replaced with "rigging" or "set-up" wheels. The vehicle was stabilized by attaching a lightly pressurized pneumatic actuator to the vehicle close to each axle. It was otherwise unconstrained.

The wheel actuators were driven in "warp" using a sinusoid of +/- 2mm amplitude & a frequency of approx 1/20 Hz. (requiring 18.75 secs to complete a cycle). Various measurements were recorded with a frame rate of 100 /sec. The time histories of differential actuator displacement ((Left - Right)/2) are shown below:

http://davew.webs.com/111811264463.png

The corresponding differential contact patch load time histories were:

http://davew.webs.com/111811305156.png

More interestingly, perhaps, the two parameters were plotted one against the other, to obtain the following trajectories:

http://davew.webs.com/111811310161.png

The front axle trajectory is shown in red, the rear in green. The regression slope is recorded in the colour coded legends (813.31 N/mm front & 836.94 N/mm rear). The two estimates differ because the front and rear tracks were different. Multiplying by the appropriate track yields 1329.8 N.m/mm for the front axle & 1326.6 N.m/mm for the rear axle, which suggests a measurement/processing error of 0.24%, since the two should self-balance. Taking the average of the two estimates, the overall torsional stiffness was computed to be approx 9,300 N.m/deg. spindle-spindle.

Electronic inclinometers were attached to the vehicle at the nosebox bulkhead, and the rear wing lower element. Time histories generated by these are shown below:

http://davew.webs.com/111811240277.png

The time histories were noisy because inclinometers are also very sensitive lateral accelerometers. The vehicle did not deflect in an entirely smooth way, despite the purity of the platform motion, so the measurements were corrupted by small accelerations.

The corresponding front axle differential load plotted against (front-rear) angular deflection follows:

http://davew.webs.com/111811311514.png

The red dotted line shows the normal regression line (Slope of 10,115 N/deg.). The magnitude of the "noise" caused the regression slope to be biased. Arguably, a rather less biased estimate of slope can be obtained by "Fourier filtering" the measurements. The result is the green trajectory, with a regression slope of 11,113 N/deg., an increase of 10 percent. Further processing yielded an estimated torsional stiffness of the tub, engine & transmission (i.e. omitting the suspension) of approx 18,160 N.m/deg.

The test demonstrated that the torsional compliance was split fairly equally between the chassis and the suspension.

I think the example demonstrates a requirement for knowing test conditions, what was measured and how the results were processed.

p.s. 1 N.m = 0.738 lb.ft.

Edited by DaveW, 18 November 2009 - 20:45.

[Fat Boy]

This is probably the same method that Lola used. Honestly, do you all think this number is something real?

[DaveW]

This is probably the same method that Lola used. Honestly, do you all think this number is something real?

Er.. does your question imply that you think the estimates are not real, or are you simply substantiating Joe Bosworth's comment? If the first, I'd be grateful to know which number & why...

[Greg Locock]

If someone were to work out the mass and torsional stiffness of an aluminium tube 24 inches in diameter and 1/8 inch thick and 6 ft long then they would have good reasons to call shenanigans on the original number, even if common sense didn't do that for them. Unless Lola have managed to develop a cockpit that is stiffer than a closed tube of the same weight. Might be possible if they used diamond, otherwise not so much.

Edited by Greg Locock, 19 November 2009 - 00:00.

[Fat Boy]

Er.. does your question imply that you think the estimates are not real, or are you simply substantiating Joe Bosworth's comment? If the first, I'd be grateful to know which number & why...

It was my attempt at humor. The car was built in 1969. I'm fairly certain they didn't have access to anything even remotely close to the tools you have at your disposal, so they couldn't possibly have use the same method of measurement. I previously had written that I thought the number was someone at the factory taking the piss out of the original owner.

Something like this:

Customer: I'm very interested in having a racecar with a high torsional stiffness

Salesman: Um, ya mate, ours is the best at that. Nigel will give you that number when your chassis is done. Now about that deposit.....

<<< Later >>>

Salesman: Nige, the bloke that bought this rolling trash-heap wants to know about torsional something or other. Can you put something on the paperwork that says we know what it is?

Nigel the Mechanic: Ya, soon as I nip up this Jubilee clip with the spanner.

<<<< Even Later >>>>

Nigel the Mechanic: Torsional stiffness was measured at 3,000....scratch that....30,000......scratch that.....OH BOLLOCKS! 300,000 ft.lb/deg...../foot. That bloody well ought to be good enough for the tosser that bought this thing.

<<<<< Later Yet >>>>>

Customer: This parrot is no more. It is an ex-parrot.

[BritishV8]

DaveW reported:

An open-wheeled, mid-engined race vehicle was installed on Multimatic's four post rig...

Thank you for trying to move the conversation forward! The test approach described will take me some time to fully digest, but the result seems obvious:

18,160 N.m/deg = ~13400 foot-pounds per degree = ~335 percent of the stiffness Mark Donohue believed was credible for his T330

Now... just what sort of "open-wheeled, mid-engined race vehicle" was tested? That description fits everything from Formula Student to Formula One. Is this a recent chassis design?


------


Joe Bosworth's posts are always pragmatic and helpful. I agree that individual statistics are useless... but if someone collected and plotted enough data points, it wouldn't be hard to discard outliers and to spot trends. Just like horsepower numbers! How big an edge might McLaren have had when they went to carbon, and wasn't it a relatively small improvement compared to the change from tube chassis to aluminum monocoque? Or, how bad might it have hurt stiffness when Formula Ford went to enlarged cockpit apertures, and will the new side-impact absorbing side pods make up the difference? I don't care to know numbers. I just want to develop some intuition.

[McGuire]

I agree that individual statistics are useless... but if someone collected and plotted enough data points, it wouldn't be hard to discard outliers and to spot trends.

That assumes that the bulk of the data is not only reliable but comparable.

[cheapracer]

18,160 N.m/deg = ~13400 foot-pounds per degree =

I have issues with these figures - not that they aren't some sort of initial indication.

What happens for the second degree on? It may take 13,400 for the first degree but 14000 may turn it into spagetti.

They don't tell you what will happen in localised areas such as shock/spring mounts.

etc. etc.

A longer engine in the F5000 shouldn't make a significant difference as we hope the gearbox/engine/suspension unit doesn't flex at all and takes the load path right up to the rear bulkhead regardless of being AMC or Chev.

[DaveW]

What happens for the second degree on? It may take 13,400 for the first degree but 14000 may turn it into spagetti.

Ouch... The "per degree" is a unit. It makes no statement about how far a chassis can be twisted, any more than driving at 200 kph implies that you will travel 200 km IN 1 hour. In fact, the last plot shows that the chassis, etc. was twisted a maximum of 0.15 degrees during the test, & was reasonably linear over that range.

They don't tell you what will happen in localised areas such as shock/spring mounts.

Quite true, although in this case the difference between the two stiffness estimates does suggest something about likely local compliances, often described collectively as "installation stiffness", although the latter does (or should) include internal damper stiffness. I am sometimes asked how stiff a chassis "needs" to be. The results for the example suggested that the structure was a reasonably efficient compromise, because increasing chassis stiffness would quickly become pointless without also increasing suspension stiffness.

A longer engine in the F5000 shouldn't make a significant difference as we hope the gearbox/engine/suspension unit doesn't flex at all and takes the load path right up to the rear bulkhead regardless of being AMC or Chev.

Apologies, but that is a big hope. Most LMP vehicles, for example, include a truss connecting the back of the engine to the engine bulkhead. That is not an oversight, I think. I have seen one vehicle where, for various reasons, 40% of input energy was actually dissipated by the engine during a rig test. It wasn't wildly successful, partly because the engine was unable to survive a race weekend.

[gruntguru]

What happens for the second degree on? It may take 13,400 for the first degree but 14000 may turn it into spagetti.

That is unlikely as it would typically require some kind of failure or at the very least a severe design fault. The opposite case of a rising torsional rate seems more likely (although still quite remote) to me.

[Fat Boy]

I have issues with these figures - not that they aren't some sort of initial indication.

What happens for the second degree on? It may take 13,400 for the first degree but 14000 may turn it into spagetti.

Look at DaveW's data. It's pretty linear. I think you'll find this with any chassis worth using. It's pretty damn tough to make a spring with a falling rate. Rising rate is easy.

On older aluminum monocoque's like we have here, you'll often find some offset at very low deflections. The reason is that the initial small amount of deflection is the panels moving slightly and 'engaging' the rivets. It makes for a weird handling car. The epoxy that is used on these chassis' is supposed to reduce this effect, but it's still there, especially on the stuff that has a lot of miles on it. At some point, you either build a new chassis or relegate the car to show duties only.

[Fat Boy]

I know this sounds amazing, but when searching the web, I actually found footage of the testing of this particular chassis. Have a look.

Lola Chassis Stiffness test - 1969

[DaveW]

I know this sounds amazing, but when searching the web, I actually found footage of the testing of this particular chassis. Have a look.

Lola Chassis Stiffness test - 1969

& I bought it... brilliant.

[gruntguru]

I know this sounds amazing, but when searching the web, I actually found footage of the testing of this particular chassis. Have a look.

Lola Chassis Stiffness test - 1969

[cheapracer]

Look at DaveW's data. It's pretty linear. I think you'll find this with any chassis worth using. It's pretty damn tough to make a spring with a falling rate. Rising rate is easy.

.

I certainly take your point, a good chassis is designed (hoped?) to be like that but they also are certainly not made from spring steel.

I guess I should shift my point to the area of just how much rising rate, a small tube chassis will probably be significantly less than a large tubed one.

[RDV]

Don't laugh! Tightening the jubilee clip was crucial to the final values, in slug-furloughs per Furman

[gruntguru]

furloughs

You mean furlongs? The slug-furlough/Furman is a very lazy unit not applicable to torsional stiffness of racecar chassis.

[RDV]

...my bad, furloughs actually associated with moment of inertia...;)

Edited by RDV, 20 November 2009 - 16:09.

[Fat Boy]

Don't laugh! Tightening the jubilee clip was crucial to the final values, in slug-furloughs per Furman

It is for a racecar chassis. I'd prefer to use slug-fuloughs per Firman.

[Fat Boy]

I certainly take your point, a good chassis is designed (hoped?) to be like that but they also are certainly not made from spring steel.

I guess I should shift my point to the area of just how much rising rate, a small tube chassis will probably be significantly less than a large tubed one.

I think you'll find on a tube chassis that the spring rate is quite linear. Remember, 'spring steel' and hot-rolled have essentially the same modulus. Hot-rolled will yield sooner, but up to that point they'll act the same.

Where you find rising rates is in a monocoque that has been run long enough to loosen the rivets a bit. As the chassis deforms it engages the panels more and what started out as a noodle chassis suddenly isn't too bad.

[DaveW]

Where you find rising rates is in a monocoque that has been run long enough to loosen the rivets a bit. As the chassis deforms it engages the panels more and what started out as a noodle chassis suddenly isn't too bad.

An example of a rising rate characteristic (caused by a less than perfect installation) c/w freeplay can be found here:

http://davew.webs.com/112016435307.png

Actually, of an installed sway bar (arb) with blade adjuster, rather than a pop-riveted tub. Characteristics are likely to be similar, however.

Edited by DaveW, 20 November 2009 - 17:03.

[gruntguru]

http://davew.webs.com/112016435307.png

Actually, of an installed sway bar (arb) with blade adjuster, rather than a pop-riveted tub. Characteristics are likely to be similar, however.

Why the hysteresis - sliding joint with friction somewhere?

[DaveW]

Why the hysteresis - sliding joint with friction somewhere?

Apologies, I was lazy & didn't mention that "corner" springs & dampers were also present (I did say the bar was "installed", however). Overall suspension friction was also present (low in this case).

Tests of this type are simple to execute on a multi-post rig, but can be very useful both to race engineers & designers. Here is another example, this time showing a family of trajectories as the blade angle is increased from 0 to 90 degrees. The stiffness doesn't exactly vary linearly with blade angle, see here. The characteristic was caused by non-normal & short drop links that forced the bar to bend as well as twist. Incidentally, suspension friction caused the change in load with no change in displacement at the trajectory extremes. The family of trajectories also emphasises on-centre freeplay, caused mainly by the blade bearing (I think).

Edited by DaveW, 21 November 2009 - 07:49.

[Greg Locock]

Ball joint friction, hysteresis in rubber components, shock absorber friction are the main contributors to the loop height. Fairly important parameter for secondary ride.

[gruntguru]

The stiffness doesn't exactly vary linearly with blade angle, see here. The characteristic was caused by non-normal & short drop links that forced the bar to bend as well as twist.

"Non-linear" is putting it very mildly. I can see the driver getting very confused if you gave him 0-90 degrees to play with. I assume you only gave him the 45-90 degrees range?

Don't suppose you have a photo of the geometry?

[DaveW]

"Non-linear" is putting it very mildly. I can see the driver getting very confused if you gave him 0-90 degrees to play with. I assume you only gave him the 45-90 degrees range?

Don't suppose you have a photo of the geometry?

Could certainly create communication problems between driver & engineer... (This should help oversteer.. No it didn't, made it worse. Etc.)

No photo's I'm afraid, but start with Tony Matthews beautiful "sketch" shown here. Lose the 3rd spring & replace the "Lola shuttle" with a T-bar installed vertically. The bar is retained at the lower end by horizontal arms comprising a single piece rotatable double blade. The drop links are nominally parallel (when the assembly works predictably), but alternative attachment points angle the drop links, rather like those shown in Tony's drawing. When the drop links are not parallel the torsion bar is both twisted & bent. Then the assembly has counter-intuitive characteristics such as the one I posted.

p.s. The blade is not cockpit adjustable - yet another FIA "cost saving" regulation!

Edited by DaveW, 22 November 2009 - 00:30.

[mariner]

I can see the huge advantage of using a four post rig to do torsional testing, I would presume that you can also derive the natural frequency of the twisting as well because it is dynamic. Hopefully that would be so high it would never effect things like , say, turn-in, but it must be nice to know that rather than assume it from a static test.

I wonder if it is possible to instrument a dynamic four post testwhere WHERE the twist happens in the chassis. The old fashoined methods often had dial gauges along the frame stations within the wheelbase so it was "easy" to identify exactly which part of the chassis was weakest in torsion and fix that. I have long been fascianted by exactly where chassis flex particularly when the engine is a stressed menber. Lacking the skills and tools to do this I can , sadly, only try to visualise it but on a on F1 car the cross section of the V-8 is really small so I wonder if that flexes the most?

I dont mean the whole engine necessarily but the detail bits like brackets and mounting lugs. Similarly modern push rod suspensions have really high leverage moments in them and extreme loads so even a very small amount of flexing will affect the true rigidity in terms of predictablity. The same may be true of any engine/chassis interface. When Lola built a ( unsuccessful ) carbon F1 car a while back they had to fit CF compression posts inside the fuel cell because the width of the Ferrari engine did not match the width of the fuel cell rear bulkhead so all the torsional loads would otherwise have gone into the middle of a flat panel which would then "pant", hence the posts going into a stong corner section at the front of the fuel cell box.

I do not know if similar thing are done in modern F1 cars but I can see that such detail could make or break (sorry) the overall torsional stiuffnes hence the benefit of measuring the twist at points along the wheelbase. Can the four poster technique use some sort of laser measuring at intermediate points?

[DaveW]

I wonder if it is possible to instrument a dynamic four post testwhere WHERE the twist happens in the chassis.

Interesting question - actually, several questions...

Chassis compliance does appear to affect suspension performance, sometimes quite dramatically, especially when a vehicle has been designed for minimum weight, perhaps to provide the maximum scope for c.g. adjustment. It is general knowledge, I think, that some F1 vehicles last year carried up to 100 kg ballast, implying that a dry & unballasted vehicle will weigh not much more than 400 kg. It is very difficult to achieve that weight without sacrificing stiffness, bearing in mind that cross sectional dimensions will also have been minimized for aerodynamic reasons.

I bought a pair of servo-controlled inclinometers with the idea that they could be used to identify the distribution of chassis compliance. After playing with them for some time, I concluded that they could not be used dynamically because they are sensitive to lateral acceleration, & that can't be controlled when a vehicle is minimally constrained. Constraining the vehicle might help, perhaps, but there is always the possibility (probability) that the constraint(s) would affect compliance. However, with care, they can be used quasi-statically, provided that the transducers can be attached to representative vehicle structure (essentially, bulkheads, where deflection can be expected everywhere to be representative of the sectional deflection). Road vehicles, which are not normally well endowed with structural bulkheads, proved that it was possible to get a broad range of estimates, depending upon the precise location of the transducers (interestingly, however, it was possible, with the transducers fixed, to measure the difference in torsional stiffness doors open & doors closed with reasonable consistency).

Of course, an inclinometer is not the only transducer that could be used to measure angular deflection. Some would probably cope better with lateral acceleration. But probably all would have problems identifying representative sectional deflection unless the structure was "friendly".

On related matters:

Cantilevered rocker posts are usually a significant source of "installation" compliance, which is why many vehicles now use rocker posts installed in double shear.

Observably, the torsional compliance of a carbon composite mid-engined open wheeler occurs mainly in the cockpit area and between (perhaps including) the engine bulkhead and the drive shafts. The engine itself can be a significant contributer to that compliance. An "adequate" torsional stiffness can't be specified in isolation because it depends (as is usually the case) on stiffness ratios. A softly sprung vehicle will "work" with a relatively low torsional stiffness. Increased suspension settings will continue to "work" provided that the chassis torsional stiffness (& tyre stiffnesses) is (are) high enough. To find that a vehicle doesn't respond (much) to suspension changes is usually a warning that the settings are too high for the vehicle &/or its tyres.

The torsional natural frequency of a modern race vehicle is normally between 20 &, perhaps, 35 Hz. An open-wheeled vehicle can be approximated fairly respectably by a spring element with "dumbell" elements at each end, since the wheel assemblies are the most significant inertial elements (so unsprung mass does have a major effect on the natural frequency of the torsional mode).

Edited by DaveW, 22 November 2009 - 15:36.

[GeorgeTheCar]

A little off topic, stiffness of production vehciles

I was watching the NASCAR race and one of the glass commercials comes on. Stone chip, hit a pothole and the windshield cracks asunder.

I understand that most of that comes from gluing the windshield in to retain the passenger airbag.

I think that at a minimum you would have to hit hard enough to get coil bind before getting enough chassis twist to have an impact on the windshield so I think you would also have tire and wheel damage or is my intuitive gauge off this weekend?

[DaveW]

A little off topic, stiffness of production vehciles

I was watching the NASCAR race and one of the glass commercials comes on. Stone chip, hit a pothole and the windshield cracks asunder.

I understand that most of that comes from gluing the windshield in to retain the passenger airbag.

I think that at a minimum you would have to hit hard enough to get coil bind before getting enough chassis twist to have an impact on the windshield so I think you would also have tire and wheel damage or is my intuitive gauge off this weekend?

Greg will probably be able to provide hard numbers, but I believe the front screen contributes significantly to road vehicle torsional stiffness. I do know of two racing tin tops converted (perhaps not very well) from road vehicles that use clip on screens to avoid spurious breakages. In a "proper" tin top conversion, the roll over cage becomes the primary (& very stiff) structure of the vehicle.

[GeorgeTheCar]

Yes, the roll cage was one of Roger Penske's famous" Unfair Advanatages".

A lot o people thought it was just Mark being cautious!