Stuck on Pro-Mechanica results for chassis

[Adam C]

Hello everyone,
I am attempting FEA on a chassis design using Pro-Mechanica. I am using a former team members method of constraining and loading (from 2003, he is no longer with the university and I am unable to find contact information for him).

There are 2 fixed constraints placed at the rear lower frame rails of the chassis. 2 loads are applied towards the front of the chassis near where the springs would mount. One force is straight up, one straight down of equal magnitude. The program is predicting stiffness in torsion of about 175 times (not a typo)the stiffness I would expect. The results the former student recieved were 20% greater than a physical test performed the exact same way.

The chassis is drawn using Pro-E with points and sketched lines connecting the points. In Mechanica beam elements are created referencing points, except where there is a member having more than two points lying on it, those beams are created referencing edges.

Material properties were taken from Matweb, cross section dimensions are correct.

I hope this adequately explains my situation. If anyone has any recommendations I would greatly appreciate it.

-Adam

[Adam C]

Hello everyone,
I am attempting FEA on a chassis design using Pro-Mechanica. I am using a former team members method of constraining and loading (from 2003, he is no longer with the university and I am unable to find contact information for him).

There are 2 fixed constraints placed at the rear lower frame rails of the chassis. 2 loads are applied towards the front of the chassis near where the springs would mount. One force is straight up, one straight down of equal magnitude. The program is predicting stiffness in torsion of about 175 times (not a typo)the stiffness I would expect. The results the former student recieved were 20% greater than a physical test performed the exact same way.

The chassis is drawn using Pro-E with points and sketched lines connecting the points. In Mechanica beam elements are created referencing points, except where there is a member having more than two points lying on it, those beams are created referencing edges.

Material properties were taken from Matweb, cross section dimensions are correct.

I hope this adequately explains my situation. If anyone has any recommendations I would greatly appreciate it.

-Adam

[adrial]

We also used Pro Mechanica for our chassis FEA.

We ended up modeling our constraints based on the picture that Denny Trimble has posted on the forums a bunch of times. If you search I'm sure you'll find it.

We haven't compared to a physical test yet, but we have gotten reasonable numbers.

[Adam C]

Is this the image you are reffering to:

http://students.washington.edu/dennyt/fsae/torsion_screenshot_DT-5-10-04.jpg

Currently I am only trying to replicate a former team memebers method as stated in my original post. Once I achieve that I may attempt the more complicated method shown in the picture.

I have tried analysis of the old design (the one the former student did) an I can not achieve his results with his method and his design either. Stiffness is again way off the chart.

I have tried increasingly larger moments (up to 800 lb-ft). I have plotted my rotation in degrees vrs. moment and get a linear curve. This seems good except that the values for stiffness are HUGE! I have tried removing critical members and the deflection increases a substantial amount as expected. Stiffness is still huge. I have tried swapping positions of the loads and the constraints, stiffness changes as expected but is still huge. Deflection is greatest at the loads and zero at the constraints as expected.

The program seems to be operating correctly as to how the results change with respect to how I change the FEM. I am stumped. I have stumped a fellow graduate student (I am undergrad). Is there something I am over looking?

[adrial]

Are you sure all your units are straight?
lbf vs lbm

Pro/E defaults to lbm...which would make you predict low torsional rigidities rather than high...but still worth double checking all units.

[Adam C]

Pro-E defaults are mass: lbm; force: in*lbm/sec^2 (which is lbf) I am entering my loads as a lbf.

I have now altered constraints so that they have more than zero degrees of freedom. I should have done this before as this is how they should be. Still no good. AAARGH!

[Captain Redbeard]

I really think your problem is units. You need to go back to the standard ProE app and select Edit->Setup. Then click units and make sure you are using the set with lbf NOT inlbm/sec^2. I have had the same problem and this fixed it for me.

-Redbeard

[billywight]

I'm not sure if Pro/Mech will do this, but we do our chassis analysis in Algor and instead of using a moment as a bondary condition, we use a rotational displacement of 1 degree at the front and fixed restraint at the rear. Then we check the reaction moments to get torsional stiffness. So far our results have been within 10% of actual.

[Adam C]

Adrial and Redbeard-

You were right! Switched to lbf and am obtaining resonable results. I knew (was hoping) it was a simple fix.

-Adam

[Conor]

Adam-

You're an idiot. I hope your team loses horribly and is forever digraced.