Chassis Analysis constraints...

[mnowicki]

So I have a model ready to be analyzed in pro-mechanica. I have given each member specific beam properties. Now I'm ready to constrain and add forces. Can someone help me out on where to apply these forces and what members to constrain in order to get an analysis that is usefull. Thankyou. Any help would be greatly appreciated.

Matt Nowicki
mnowicki@seas.upenn.edu
University of Pennsylvania

[mnowicki]

So I have a model ready to be analyzed in pro-mechanica. I have given each member specific beam properties. Now I'm ready to constrain and add forces. Can someone help me out on where to apply these forces and what members to constrain in order to get an analysis that is usefull. Thankyou. Any help would be greatly appreciated.

Matt Nowicki
mnowicki@seas.upenn.edu
University of Pennsylvania

[Garbo]

The loads and constrints should be applied through the hubs with a full suspension model in order to simulate the loading the car will see on track.

Care should also be taken to ensure that the model is not over-constrained. On track, the wheelbase and track width are not constrained and constraining them in your test gives an inaccurate estimation of stiffness. We saw a 50% decrease in estimated stiffness just by freeing up over-constraints on the hubs (this was consistent with physical testing).

The more difficult question is determining how stiff is stiff enough...

Cheers
Garbo

[awhittle]

I like to design my chassis with the suspension on the chassis. Early in the design I just have the suspension modeled by slender thick walled tubes so the they have almost no bending strength but they are stiff in compression. I shoot for 3000 ft-lbs /degree of chassis twist for a car that weighs 1000 lbs with driver and fuel. Your cars can get by in the 1800 ft-lbs/deg range given 450 lb car and 150 lb driver. Note that the motor on this chassis goes in thru the seat as there are no tubes required behind the drivers back. The red line represents the main load path for torsion and beam loading thru the chassis.

It would be cool to make the rear most bulkhead a bolt on aluminum bulkhead mounting the gearcase springs rockers dif and all that other stuff back there making drivetrain removal a breeze. The entire real assembly could be attached with three bolts and the motor could remain unstressed. Lots of room for headers, oil tanks, battery....

Constraints: node 22 XYZ, node 4 YZ and node 49 Z

Hope this helps

Andy
http://www.ncs-stl.com/images/SampleFeaChassis.jpg
http://www.ncs-stl.com/images/SampleFeaChassis2.jpg

[murpia]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by awhittle:
http://www.ncs-stl.com/images/SampleFeaChassis2.jpg </div></BLOCKQUOTE>
Interesting stuff Andy!
I note the triangulation of the chassis floor is assymetric, single diagonals only. Any effect on torsional stiffness clockwise vs. anti-clockwise?
Regards, Ian

[awhittle]

There is no differance in deflection as the chassis is torqued in either direction. I did that chassis a few years ago, as a sample of how I would design a FSAE chassis. Most of the load paths I see in most schools setups make no logical sence to me and the result is tubes running in all sorts of directions to bandaid some other problem. I start with the chassis (aero and strength) and work from there with packaging. For the mostpart that chassis could se shrink wraped with film similar to boat storage film or an old homebuilt airplane and the bodywork wouls be essentualy done. Not counting the downforce stuff. Chassis stiffness is everything. I happen to be Megasquirting a 71 Hemi Cuda convertable right now for a customer. I am amazed how bad the car handles in bumps. When the car is on jackstands, I can't even open the door. Cars have come a long way over the years.

AW

[LZH]

Hi awhittle,
I notice that the way you constrain the model is different from ours.Normally, we do the TR test by constraining the rear hubs in all 6 directions (x, y, z, Rx, Ry, Rz). However, I notice you only constrain at most 3 directions at one side of the hub & 2 on the other side; letting the hub to be to move in x direction. Could you kindly explain the reason? & also the difference between your method & our method?...Thx... http://fsae.com/groupee_common/emoticons/icon_smile.gif

[billywight]

Just constrain the chassis and apply loads that best describe the actual loadng - take a few minutes to think about it, it;s really not too hard. Of course you'll have to make comprimises, but that's why it's a model. Just be sure that however you end up loading it that you can do a physical test with the same conditions. This way you can compare your FEA results to the physical test.

Also remember that you can apply beam end releases to make things more accurate at the suspension points and rigid members where appropriate to make the loading easier...

[markocosic]

Hi Andy,

If you don't mind saying, what software are you running the analysis in? Are we assuming ideal beams/pin-joints or is there offset/pre-bend in the beams, stiffness in the joints or some other assumptions? In your experience, how closely does the analysis comapre with real-world testing? (with the particular assumptions you've made)

I've had within +/- 2.5% with pin-joint/ideal beam assumptions and relatively simple/slender structures but have never built/tested anything as complex or 'short and stout' as a race vehicle spaceframe.

One thing in favour of those soggy old bendy motors (as versus almost all modern car it seems!) is that they don't get all confused when they're not driving on a billiard table... (handling balance of new (&German) cars appears to go to pot on anything other than perfect/planar surfaces where older (&british/french) motors tend not to give a damn)

[awhittle]

I use Grape FEA that is free will run in unregistered mode after ignoring a few screens. http://www.grapesoftware.mb.ca/features.htm It can't do pinned connections but long slender tubes is just fine for our purposes. It is very easy to remove on tube and see your deflection go huge or what tube needs to go to a node and not mid span of some other tube.

Regarding constraints, any time I am checking for torsion, I only constrain the chassis in as few direction that keeps the chassis stable for the program. By default, one and only one point has to be restrained in all three directions but never held in rotation. A secound point has to be held vertical and longitunal. The third has to hold the back of the chassis so that the chassis is stable as far as the sofeware is concerned. EZ http://fsae.com/groupee_common/emoticons/icon_smile.gif

I was playing last night and cut out 4 more tubes but had to add one and only lost a few percent of the strength. But it made the chassis a little more dificult to build. Most likely a wash. I would most likely go with easy to build.

Regarding actual vs FEA results. I have been close. I know that every chassis I have built with 3000 ft-lbs/deg responded just as the books claimed regarding sway bars, shocks.... None the weak chassis's worked at all.

Note: All of the tubes in thia chassis are modeled as 1" 060" wall for ease of fab. You guys could better ballance the loads given time and effort. This car was designed as a real autocross car for years of use and not a one event car.

Hope this helps

AW
http://www.ncs-stl.com/images/SampleFeaChassis3.jpg

[BeaverGuy]

With this chassis, are you asuming a direct acting shock absorber, because I would think that the altered load path from a rocker and shock would greatly change how you designed the chassis. I also noticed that you changed your constraints in the second model, any reason for this?

[awhittle]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">With this chassis, are you asuming a direct acting shock absorber, because I would think that the altered load path from a rocker and shock would greatly change how you designed the chassis. </div></BLOCKQUOTE>
Technicly it would but generally the load will take the same general path from the lower outboard upright lower ball joint up into the place the rocker mounts. If the basic design of the chassis is correct, the details of the rocker and how it spreads the load into the bulkhead is workable.
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">I also noticed that you changed your constraints in the second model, any reason </div></BLOCKQUOTE>
Good eyes http://fsae.com/groupee_common/emoticons/icon_smile.gif It makes no differance in the results. I was just prooving it to myself. You may have noticed that the colors in one screen shots was axial stress the other was axial load.

AW

[LZH]

IC. Just being curios...Is the constrain you are employing now in the FEA model able to be used in the real torsional test? How you would constrain your chassis during actual torsional test? http://fsae.com/groupee_common/emoticons/icon_smile.gif

[awhittle]

The simple answer... four links to replace the shocks, three jackstands, a chain with turn buckle and a concrete insert in the floor, a little cable, a few dial indicators, and some weight in buckets,

This is all accodemic as once you have a stiff chassis in FEA it will be stiff in the real world. Just dont make a dumb error on the rockers and shock mounts. The sway bars have to be at hard parts. Once there the real speed is in smooth power, aero, alignment, shocks and a seat that supports you well from the ribs and not your sholders. 1800+ ft-lbs / degree and the chassis is done. Less and you have a fundimental error. If every tube that you take out drops the strength by 15% then you are getting there. Look for the green ones http://fsae.com/groupee_common/emoticons/icon_smile.gif Look for a clean load path.

Remember to pick your enimies well... some day you will be just like them.

Hope this helps

AW

[TAMUCC]

How exactly are the upper and lower A-Arms constrained to the frame?
Aren't all of the suspension forces translated through the push and pull rods, since the suspension is effectively floating on the springs?

[awhittle]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">the suspension forces translated through the push </div></BLOCKQUOTE>

The way I do it is make all of the suspension pieces modeled in long slender tubes so that they realy only transmit loads in compressive or tension. Being slender, they will not create significant torsion or shear loads.

Hope this helps

AW

[Marshall Grice]

or you can change your beam end constraints to 0.

[mnowicki]

Thanks a lot for the information. I have all that done but now i'm stuck on how to calculate the angle of deflection? What parts of the chassis am I measuring the change in? Can someone be kind enough to explain briefly? Thanks a lot.

Matt Nowicki
University of Pennsylvania
mnowicki@seas.upenn.edu

[billywight]

Try giving it a displacement boundary condition of 1 degree, run the analysis and check the reaction moments... (restraints in the rear and displacement at the front)

[TAMUCC]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by awhittle:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">the suspension forces translated through the push </div></BLOCKQUOTE>

The way I do it is make all of the suspension pieces modeled in long slender tubes so that they realy only transmit loads in compressive or tension. Being slender, they will not create significant torsion or shear loads.

Hope this helps

AW </div></BLOCKQUOTE>
That was not what i was talking about, but it's alright.