How to test for Torsional Rigidity?

[Hsoj]

We're to the point to test the torsional rigidity of our chassis. We've read the testing procedure in the Milliken book and are looking for better or easier solutions. We were wondering how other teams were going about testing their chassis. What is everyone else up to?

Hsoj

[Nim-T]

I know this will sound strange, but first of all, why exactly do you want to know the torsional rigidity of your chassis? Just so you can put a figure in your stats sheet or for something more?

[roadrunner]

We put ours in a torsion test rig and twisted the hell out of it.
It was mounted without the wheels but at the wheel mounts (if that makes sense).
Basically 3 of the wheels are retained and you apply a load at the other to twist it.
We found some surpising results. Not least that although out tub was solid as hell, our engine mounts and suspension mounts were soggier than sloppy seconds!!!!!
We actually calculated that as it stood we were better off having solid suspension and not running any springs and dampers at all!!!!

Word to ya moms but i came to drop bombs!!!!!

[Joseph]

Roadrunner's method is probably the best. We did a test to compare FEA with actual results on our frame before we finish the car. The frame was not complete at the time so we only tested from the main roll hoop to the front of the frame. The predicted value was 1002, actual was 963 ft*lb/deg. When the frame is complete with all mounts it should be around 2000 ft*lb/deg when tested at the wheels, including frame twist and a-arm deflection. I was surprised with the closeness of our values. I'm not exactly sure how much of it was a good model in Pro-Mechanica or luck. I would like to think that it was our superb model http://fsae.com/groupee_common/emoticons/icon_smile.gif



http://students.ou.edu/K/Joseph.D.Kliewer-1/torsion.jpg

Joseph
University of Oklahoma
http://students.ou.edu/K/Joseph.D.Kliewer-1/rallyRace.gif

[This message was edited by Joseph on February 09, 2004 at 10:55 PM.]

[Denny Trimble]

You think your frame will be stiffer when you test a longer section of it? Isn't the equation for springs in series:
(1/Ktotal) = (1/K1) + (1/K2)... where K1 would be your front section stiffness (1000) and K2 would be the rear (let's assume 10,000). That would put Ktotal at 909 right there.

Nim-T, torsional stiffness is needed in race car chassis to allow tuning of front vs. rear roll stiffnesses. Without chassis stiffness, the chassis will flex and not allow these tuning changes to take effect.

University of Washington Formula SAE ('98, '99, '03, '04)

[Nim-T]

Denny, I realise what torsional stiffness is...if I didn't know what that is I sure as hell shouldn't be desigining a chassis, as I am now. Just wanted to make sure that these measurements are being done for good reason: I know one or two teams in the past (who shall remain nameless) that have performed time consuming stiffness measurements to simply validate their FEA work and not actually use them as parameters in their suspension tuning work!

[Joseph]

Sorry for not fully explaining myself

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR> You think your frame will be stiffer when you test a longer section of it? <HR></BLOCKQUOTE>

No it won't. But as I stated this was a test to validate the FEA results. The true torsional stiffness must be measured at the wheels and loading the frame at the rockers, a-arms, and shock mounts. In the test I did we measured stiffness all the way to the front of the frame. Since this frontal zone of the frame does not contribute to torsional stiffness it was designed to be lite not stiff. It only takes one weak section and the whole frame becomes spaghetti 1/Kt = 1/Ksmall + 1/Kbig = Crappy torsional rigidity. We also were not able to constrain the main roll hoop at the corner nodes, this reduced the stiffness from 1500 to 1000 ft*lb/deg according to FEA. This test was only meant for comparison to FEA; had the frame been constrained at the nodes and twisted at the rocker arm mounts we would have seen about 4000 ft*lb/deg add that with an equal K value for the rear section and your are right at 2000 ft*lb/deg. I have a good amount of confidence in the 2000 figure because we have already validated the model by performing an actual test.

Joseph
University of Oklahoma

[Tim Heinemann]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>
Just wanted to make sure that these measurements are being done for good reason: I know one or two teams in the past (who shall remain nameless) that have performed time consuming stiffness measurements to simply validate their FEA work...<HR></BLOCKQUOTE>

I couldn't think of a better reason for testing something IRL than validating computations and FEA especially.



Tim

[Denny Trimble]

Joseph,
I understand now, thanks. Sounds like you're on top of it.

Nim, sorry if I offended you. Care to enlighten me on how you would tune for a soft frame vs. a stiff frame? Our approach is to set our frame stiffness target based on our suspension rates, so that any frame deflection is insignificant.

University of Washington Formula SAE ('98, '99, '03, '04)

[Charlie]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Denny Trimble:
Care to enlighten me on how you would tune for a soft frame vs. a stiff frame? <HR></BLOCKQUOTE>

I'd be interested to see the same..

-Charlie Ping

I just need enough to tide me over until I need more.

[Nim-T]

I have now realised how much I spend on FS work - when I questioned the measurements I was in my own little monocoque world...for a spaceframe chassis I totally agree with the testing and validation work.

I've just finished writing a project progress report and I had to change my original plan. I was originally going to test the TS to conclude the project, but it came down to that annoying nag - time (not enough of it).

I remember going through some SAE papers a while ago and coming across some TS testing work done on a Winston Cup car chassis...I'm sure someone has read this. Is the paper worth chasing up for some bedtime reading?

Nim

[Travis Garrison]

(repeated editing here as I was apperently sleep typing)

Nim-T, I don't mean to pick on you, but in one of your other posts you mentioned that you were very confident in your FEA results regarding your monocoque...but here you seem to be downplaying the usefulness of validating FEA results with real world tests here...I assume you did at least a torsional test to compare with your FEA results somewhere along the line before using FEA results as a component of your safety equivalency report?

I just mention it because my quite limited FEA experience, and everything I have read on the subject, seems to indicate that FEA without some sort of physical validation is likely very questionable...esp. with composites. To me doing torsional testing simply to validate the FEA would seem to be invaluable, especially to the next team who might be looking for areas to improve upon.

By the way, our previous tube car Viking 28 ( http://dot.etec.wwu.edu/fsae/v28/image018.jpg )weighed in at 2300 Ft lb / degree and the chassis in theory weighs ~45 lbs with roll bars (not exactly easy things to remove, besides, roll bars should at least pretend to be an integral part of the chassis)...of course the car predates anyone currently in the program so exact chassis weight is unknown...total vehicle weight is 440 lbs wet. However the car isn't a monocoque...more like a twin torsion tube chassis, using 6" CF tubes on either side of the driver. Funny looking, but effective, and easy to build. The torsional testing was down with large steel bars, attached to the shock mounts. The reason for this test method has more to do with access to the chassis' hard points than anything else...Additionaly the major draw back to this car is that while the shock mounts are very sound, the a-arm mounts are likely significantly less stiff, essentially the perfect example of why one should test TS at the wheels, although if just validating FEA, testing the frame by itself seems to make more sense to me. Basicaly what I'm trying to say is that I've given you useless numbers Hosj http://fsae.com/groupee_common/emoticons/icon_smile.gif

Our new car moves everything closer to the tubes, which should result in a much stiffer package. This time around we will hopefully have the time to do a proper tests of torsional stiffness...both of the fram and of the whole suspension/frame package.

Travis Garrison
WWU FSAE

[This message was edited by Travis on February 09, 2004 at 11:53 PM.]

[This message was edited by Travis on February 09, 2004 at 11:57 PM.]

[Tim Heinemann]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Nim-T:
I have now realised how much I spend on FS work - when I questioned the measurements I was in my own little monocoque world...for a spaceframe chassis I totally agree with the testing and validation work.
<HR></BLOCKQUOTE>


Okay, let me modify my last assertation: I couldn't think of a better reason for testing something IRL than validating computations and every kind of FEA especially and most hyper-especially every kind of FEA done on a complex anisotropic structure like a fibre reinforced plastics monocoque!

Do you honestly trust your FEA without any kind of validation and testing? Is that really what you want to express by your posts or am I getting something terribly wrong here?



Tim

[Frank]

we remove the wheels
install solid shocks
attach lasers to the bulkheads
and apply a moment to the hubs

800 ft.lb / deg

the twisting comes from the drivers bay (you can see it visually)

[Ashley Denmead]

Frank, i'm interested to know what your ridgidity value is considering you run a car with no diff? and also in tuning did you ever have any trouble losing tracion on the inside rear?

cheers

ash

P.S congratulations on your result in F-saea last year http://fsae.com/groupee_common/emoticons/icon_wink.gif

The edge is not the limit it is mearly the beginning!

Deakin University Phantom Engineering Team Captain

[jack]

dont you want to loose traction on the inside rear with a spool car? thats how i thought it was done...in anycase there are two ways to do it, suspension geometry (jacking), or torsional rigidity (or lack of). it seems like the first option would be more predictable.

jack @ WWU
http://www.etec.wwu.edu/

[Angry Joe]

Joseph, I mean no offense but I don't think I agree with your testing method. In order to get accurate numbers the chassis should be loaded through the suspension members, with the shocks replaced by rigid links. That is the reason for torsional rigidity is to eliminate deflection due to suspension loads, after all...



Lehigh Formula SAE Alumni
Team Captain 2002-2003

www.lehigh.edu/~insae/formula (http://www.lehigh.edu/~insae/formula)

[ben]

I get the impression over constraining the chassis is the biggest issue. You aren't replicating the loads accurately because you haven't considered local compliance due to cornering forces.

Surely a suitably constrained test that can be replicated in the FE model to validate the latter is the main aim of all this?

Ben

University of Birmingham
www.ubracing.co.uk (http://www.ubracing.co.uk)

[Angry Joe]

Perhaps, if I knew how to create pivoting linkages in FEA...



Lehigh Formula SAE Alumni
Team Captain 2002-2003

www.lehigh.edu/~insae/formula (http://www.lehigh.edu/~insae/formula)

[ben]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Angry Joe:
Perhaps, if I knew how to create pivoting linkages in FEA...



Lehigh Formula SAE Alumni
Team Captain 2002-2003

http://www.lehigh.edu/~insae/formula<HR></BLOCKQUOTE>

You use pin flags in NASTRAN I think. You can set the DOF on a node to allow free rotation in a particular direction.

Ben

University of Birmingham
www.ubracing.co.uk (http://www.ubracing.co.uk)

[Joseph]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR> Angry Joe: In order to get accurate numbers the chassis should be loaded through the suspension members, with the shocks replaced by rigid links. <HR></BLOCKQUOTE>

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR> Joseph: The true torsional stiffness must be measured at the wheels and loading the frame at the rockers, a-arms, and shock mounts. <HR></BLOCKQUOTE>

I intend on including the defection of the A-arms and the push rods. I will assume the rockers and the shocks are ridged. I want to find the torsional spring rate of the entire system minus the shock displacement.

Since I cannot create a pivoting link there are two ways to do the preceding analysis.

1. Use a free body diagram to determine the reaction forces at all mounting points. Then add measures to the model to find deflection at each point in the proper direction. Then sum the deflections to find the resulting deflection at the wheels. This could take some time but it could be done.

2. Define the model with all ridged joints. Then modify the beam elements. At the points where you want a pin joint to be simulated define a cross sectional area but set the I value to 0 that way the beam will only transmit compression and tension forces. With I = 0 no bending forces can be transmitted. This would be the easiest way but I don't know if it will work yet.

Joseph
University of Oklahoma
http://students.ou.edu/K/Joseph.D.Kliewer-1/rallyRace.gif

[Angry Joe]

I agree recreating this in FEA can be complicated. Physical testing is a bit simpler, though you need to be sure there is no slop in the suspension mounting



Lehigh Formula SAE Alumni
Team Captain 2002-2003

www.lehigh.edu/~insae/formula (http://www.lehigh.edu/~insae/formula)

[RagingGrandpa]

I haven't really heard how you guys are testing in FEA though. Someone sent me this pic a while back: (Hypermesh FEA with NASTRAN analysis)
http://www-personal.engin.umich.edu/~lovelljc/03ChassisFEA.jpg
It was explained that the rear shock mounts were constrained to zero movement in any direction (DOF 1-6) and the front mounts were forced to each move a small amount, so a forced displacement of say Z = +0.1" on the left tab and another of Z = -0.1" on the right tab. After analysis, the force vectors shown in the picture are generated and can be measured and used (just the forces at the front of the frame, mind you) along with the known displacements and the distance between the displacements to calculate TS. Does this sound reasonable for frame-only testing?

"...with powershifts and tiresmoke for all"

[Angry Joe]

tough to say without knowing the details. What directions were the mounts moved, and did they include the front shock mounts?



Lehigh Formula SAE Alumni
Team Captain 2002-2003

www.lehigh.edu/~insae/formula (http://www.lehigh.edu/~insae/formula)

[RagingGrandpa]

the only mounts moved were the front shock mounts, as shown by the purple resulting force vectors in the pic

"...with powershifts and tiresmoke for all"

[Matt Gignac]

When modeling the chassis in an FEA program to check for torsional stiffness, what seems to be the concensus about the suspension arms? Do you model them with the actual physical properties that they should be, or do you model them as being completely rigid. Or maybe just the shocks and bellcrank rigid? Between having them rigid and normal, there's a swing of 700 lbft/degree for us.

The way I see it, having the suspension rigid will give a good measure of the frame's stiffness itself, but with the a-arms not rigid, it would give us a better idea of how the car would behave.

[roadrunner]

I am of the opinion that you test the chassis in FEA the same as you would in real life. You want chassis stiffness?, then you test without any suspension and just lock and load at the suspension mounting points, in both reality and VR.

If you want to know the effectivness of your suspension then test the suspension on its own, i.e. for stiffness in your wishbones/pushrods etc, using basic stress/strain methods. Calculate the expected forces using the calcs from when the suspension geometry was designed. Then use dynamics packages like ADAMS where you can set the specific stiffnesses for components which you already know from the above tests and actually model the system.

FEA of suspension and chassis together is a bitch and not something to be attempted by those frightened of such packages. ADAMS package is scary enough!!!!! http://fsae.com/groupee_common/emoticons/icon_wink.gif

The best test is race track tesing, the only way you will truly know if you right.

Word to ya moms but i came to drop bombs!!!!!

[RagingGrandpa]

lol so when a judge comes up to you and asks, "How stiff is your chassis?" you're going to say "Enough"? I would tend to agree with that mentality when discussing the car with a casual observer, but that won't cut it with professionals.

"...with powershifts and tiresmoke for all"

[Lukin]

In regard to the stiffness values for the chassis and the amount of work that goes into it, should the aim of designing a chassis with an extremely high stiffness take priority over other chassis factors?

Unless the chassis is a complete joke, it will take a relatively large torsional load to cause a relatively small degree of deflection. This same load will more than likely cause a much larger change in the suspension geometry, and both the chassis and suspension will deflect in a largely predictable fashion.

Not that I think it's a matter of welding a few bits of tube together to make the chassis, but wouldn't other factors be deemed just as important.

Since Formula SAE are largely underpowered for the tyres, the chances of losing traction in dry conditions is relatively small. Therefore I think that weight transfer isn't hugely important, in the sense that it would take a huge amount of weight transfer under braking/acceleration to cause wheelspin.

Almost all of the cars running (ie endurance) is in twisty conditions with very few fast corners where stability is a requirement. Since most of the corners are sharp where good turn in is needed, would it be better to have the wheelbase sit on the minimum with a relatively large front:rear track ratio? I think this would make the car 'twitchy' but as there isn't surplus power a decent suspension design should give tremendous 'chuckability' over a car with a longer wheelbase.

Also, reducing the polar moment of inerta will help this, as is the case with the Mclaren F1 road car. This will be harder and will involve a fair bit of smarts in packaging and trying to make sure moving everything closer to the car centre doesnt raise the height of the COG.

Sorry if these factors have come up before, I couldnt find much on it in these forums. If someone disagrees whole heartedly with anything, please let us know.

[Denny Trimble]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Lukin:
...
Since Formula SAE are largely underpowered for the tyres, the chances of losing traction in dry conditions is relatively small. Therefore I think that weight transfer isn't hugely important, in the sense that it would take a huge amount of weight transfer under braking/acceleration to cause wheelspin.
...
<HR></BLOCKQUOTE>

Lukin,
Sounds like you need a better engine man http://fsae.com/groupee_common/emoticons/icon_smile.gif We're FAR from engine-limited in FSAE events, even with a 45:55 weight distribution. We also see a good amount of weight transfer from accel and braking, not as much as an aero car, but over 1.5g's.

University of Washington Formula SAE ('98, '99, '03, '04)

[Charlie]

Whoa, I'm with Denny, these cars can easily overwhelm the tires out of a slow corner if your engine dept is in order!

-Charlie Ping

I just need enough to tide me over until I need more.

[Lukin]

My Bad! I spoke to the guys from last year today and they had the same problem with lack of traction in straight line and also on corner exit, it was especially bad when the tyres werent completely up to temperature.

[Frank]

i believe the only way to make a FSAE car go faster over 75m is to have a rear CG.

check your acceleration logs, is it linear?

with a co-efficient of friction of 1.4 at the rear tyres, how many G's can you do?

try putting a 30 pound weight over the rear axle, and see what happens to your 75m times

[Kevin Hall]

Frank

We've all tried ballast, whether it's some lead on the floor to lower CG and try running skidpad, or trying to improve traction in braking or acceleration with similar ballast. The added traction is eaten up by the added weight that needs to be accelerated.

Kevin Hall
University of Saskatchewan
'03-'04 Team Director

[Popeye]

Getting back to torsional stiffness test of the vehicles.

Do people agree with the train of thought that there is a torsional stiffness of the frame and a torsional stiffness of the entire vehicle/suspension as a whole? It would be my impression that there would be four more significant outcomes of physical testing.
1-validate FEM by simply replicating constraints and loading conditions in reality with those used in a FEM package.
2-determine the torsional stiffness of the frame alone between the bell cranks
3-determine the torsional stiffness of the frame and all suspension components to determine expected "slop" of suspension components (using solid pieces in place of shocks, loaded through uprights with all pushrods, bellcranks and a-arms in place)
4-to improve vehicle simulations in adams with respect ot roll center position and overall vehicle response.

Lastly, why would one want to constrain their vehicle at the main roll hoop as it is not going to replicate accurately the load paths through the chassis as would be seen during vehicle use.

Regards,

Pops