Chassis torsional stiffness 1965 and today 3

[Tmoose]

http://www.sportscars.tv/Newfiles/Chaparral2.html

About 1/3 way down the page.
"The Chaparral Frame

The basic frame of the Chaparral 2 is a most interesting and unusual piece of engineering. It was conceived jointly by Hall and Andy Green of PlasTrend, Fort Worth. The design requirements laid down by Hall called for a maximum frame weight of 150 Ib, combined with an axle-to-axle torsional rigidity of at least 3000 lb-ft deg."

3000 lb-ft/degree ~ 4040 Nm/degree

Are all modern cars so much stiffer than an early Chaparral?

http://youwheel.com/home/2016/06/20/car-body-torsional-rigidity-a-comprehensive-list/

Please say it isn't so !!

 

[3DDave]

Reminds me of this:

http://www.dailymotion.com/video/x2...-a-cargo-ship-twist-and-contort-in-storm_news

I wonder what the torsional stiffness is for a 737.

 

[pontiacjack]

Every torsional stiffness number in this thread amazes me. When I re-did my '59 GMC pickup (from the ground up), I knew the frame design wasn't very rigid (just C-channels) and tried to improve it. To measure any improvement (of the bare chassis) I put the extreme rear of the frame on jack stands and a single stand under the front crossmember. I shimmed the rear to make the very fronts of the rails level. Then I measured the torque to twist it one degree. I had no idea of the absolute value expected. I just wanted to assess the X-bracing I added, by comparing before/after torques. Before was 95 lb.ft. and after was 135 lb.ft.- I was pleased to see this 40% improvement(!).

I believe the torsional stiffness of my truck frame is probably very nonlinear, rising steeply past the one degree measurement I took (but certainly orders of magnitude below figures in this thread). What amount of twist is typically used to rate chassis torsional stiffness?

 

[SwinnyGG]

The reason for your low values is the conditions of the test- you can't put very much torsional energy into the chassis unless it is actually constrained. In other words, the chassis needs to be rigidly fixed to the test equipment.

If the frame is just sitting on jack stands, it will take very little torque to lift one corner. I suspect the difference you saw in your test was due more to inconsistent setup (not a slight against you- test of this kind would need fractions-of-a-millimeter precision to be repeatable, which is not possible with a tape measure) and the increased weight of the frame after modification.

You're correct in that a conventional ladder frame from a truck is not going to be very rigid, compared to a unibody car.

A modern box-section truck frame, depending on the specifics of construction, could be expected to deliver 3,000-3,500 ft. lb. per degree.

I would expect an older c-channel frame, as long as the connections are in good condition, to deliver something like one half to one third of that number, in the realm of 1,000 ft-lb per degree.

 

[NoahLKatz]

> If the frame is just sitting on jack stands, it will take very little torque to lift one corner.

Not sure about that.

If the jackstands are 3' apart and each support 1000 lb (assume they're at the front where most of the weight is), it would take 3000 lb-ft.

 

[GregLocock]

Whatever the specific issues with that setup, it is quite difficult to get an accurate number, and for a slightly stiffer chassis (Ford Tbird 55) I ended up modelling the test rig as well, in order to get good correlation with my FEA. If anybody wants to do this test in real life there is a good FSAE paper on it.

Cheers

Greg Locock


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

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

 

[pontiacjack]

I see from my notes that I misspoke above- my test only involved half a degree of twist; holding a torque wrench steady at over 100 lb.ft. was not easy for me. I simply multiplied my readings times two to get the 95/135 numbers.

-----------------------------------------------------------------------------------------------------------------------------------------

The frame was not lifting from the rear supports; I had thought it might, until I began testing.

Yes, my torsional testing certainly did deal with "... fractions of a millimeter..."; As an engineer and machinist, I routinely calculate/measure/machine to fractions of .001" (.001" is roughly 1/40th of millimeter). In fact, I milled a scrap piece to .209" thick to space up one end of my 24" level (tangent of 0.5 degrees times 24" = .2094").

"... quite difficult to get an accurate number..." was said, along with "... inconsistent setup...". I don't agree. My measurements were very accurate and repeatable.

"... an old C-channel frame to be in the realm of 1,000 ft.lb. per degree..." is not what I would expect. My frame had no damage and only a little surface rust.

----------------------------------------------------------------------------------------------------------------------------------------

Now, having said all that... I must remind you that my test was limited to a one-half-degree twist. I imagine that this limit is responsible for my low torque values. Without specifying the displacement, torsional stiffness is not very meaningful since it definitely won't be constant throughout the twist range.

 

[GregLocock]

The Tbird was of the order of 900-1500 lb ft/deg, from memory. Crucially it had an X frame under the floor. I guess it had been designed by ex-aircraft engineers, as there was no obvious point of softness - in order to stiffen it every part needed to be boxed. I did add one more X, where the frame kicks up over the axle. The efficiency of the front and rear most cross members was a bit suss, but the design was all channels, and channel to channel joints are pretty rubbish.

Cheers

Greg Locock


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

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

 

[Tmoose]

Hi Greg,

Can you share any details of why you were evaluating an out-of-production ( I infer/assume from your comments ) T-bird?

The best I come up with is a university prof or boss wanted to beef up HIS bird .

regards,

Dan T

 

[NoahLKatz]

> As an engineer and machinist, I routinely calculate/measure/machine to fractions of .001"

Yes, but with items much stiffer and much smaller than a truck chassis.

 

[GregLocock]

I was working with a client who wanted to restore his Tbird, while bringing the suspension and handling up to date. Fitting a rack and pinion was harder than the IRS!

Cheers

Greg Locock


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

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

 

[pontiacjack]

In this thread I count at least 14 examples of chassis torsional rigidity values. But none of them (other than my two) specify the total angular deflection used for measurement. Why not? Can't some of you provide some numbers? Or perhaps provide a reference to an industry-standard?

 

[NoahLKatz]

If I remember correctly, deflections that are no more than a few % of the length over which they're measured are considered small and within the linear range, which seems to fit the situation here.

 

[SwinnyGG]

Noah is correct.

Stiffness values are quoted as lb-ft or Nm/deg because over the realistic range of chassis displacements that don't involve accidents, the behavior is linear, or at least near enough to linear that it isn't worth listing an equation instead of an easy to understand scalar.

 

[pontiacjack]

Noah- "... deflections that are no more than a few % of the length over which they're measured..." does not relate to the angular deflection of one end of a chassis. Perhaps you meant "deflections that are no more than a few % of the value which causes deformation"?

jgKRI- How did you conclude that torsional resistance of a [typical] chassis is "near linear"?

 

[NormPeterson]

Given that chassis/body roll is something like 5°/g or less and that the suspension roll resistances aren't intentionally made outlandishly large or small for some specific reason, the amount of chassis twist is only going to be a smallish fraction of that. All the chassis is doing is "stealing" some of the roll moment from where it wants to go to where the relative stiffnesses are forcing it to go.

Nonlinearities in the chassis torsional stiffness vs amount of twist imply that nonlinear structural elements are present somewhere in the load path(s), such as the bushings in bodywork resiliently mounted over a separate frame.


Norm

 

[NoahLKatz]

> How did you conclude that torsional resistance of a [typical] chassis is "near linear"?

Why put quotes on something I didn't exactly say?

I said why, but to elaborate, metals have a very linear stress-strain curve until yield is approached.

Do you think twisting your frame a few degrees is anywhere near yield?

Just remembered something though - older vehicles may have bolted and/or riveted joints, which if not sound could behave nonlinearly even for small deflections.

Norm, suspension roll resistances are not in the load path (or shouldn't be) when measuring chassis stiffness.

 

[GregLocock]

If you've done the experiment you would know if they were linear. In my experience they are fairy linear, with a bit of early softness. Here's a graph

http://forums.pelicanparts.com/pors...335-torsional-stiffness-1988-911-coupe-2.html

In a proper lab we'd run several repeats.

Cheers

Greg Locock


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

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

 

[SwinnyGG]

jgKRI- How did you conclude that torsional resistance of a [typical] chassis is "near linear"?

Through my own experience, limited as it is, and because when people with a lot more experience tell me so I tend to believe them.

If you approach it from a first principles standpoint, it's easy to deduce that the behavior should be linear or close to it.

Once enough load is applied to the system to take up any irregularities between the chassis, fixture, and load application scheme, the chassis is basically a big spring that isn't flexed anywhere near yield. The amount of local strain present anywhere in a modern chassis that is twisted a degree or two is very very small.

 

[GregLocock]

One way to think about it is the 4WDers RTI test, basically drive one wheel up a ramp until the car teeters on 2 wheels. You don't expect the chassis to bend permanently after that, yet the torque involved is quite high (the chassis tosrion in that test is quite measurable and significant), so we're still in the elastic and hence linear range.

Cheers

Greg Locock


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

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

 

[NormPeterson]

Norm, suspension roll resistances are not in the load path (or shouldn't be) when measuring chassis stiffness.

I realize that, but chassis torsional stiffness isn't a property that exists in its own vacuum. Mostly I was trying to justify the notion that using a degree or less of chassis twist for torque measurement wasn't out of the ballpark when the amount of cornering roll envisioned is maybe 5° or less. As an occasional track day participant, perhaps I see a greater value in being able to adjust my car's handling behavior via roll stiffness tuning and damper adjustments than the average driver who rarely exceeds 0.3g in any direction by direct intention. IOW, a downstream consideration from the chassis stiffness itself.

Might be a different story if you were into Jeeps and rockcrawling.


Norm