Share your analysis knowledge...

[Vesh]

Hi I am wanting to know peoples opinions and suggestions on different analysis techniques on the following structure...

View attachment 2789

This is for my thesis and I completed the practical "real life" experiment today at uni and measured displacement at certain nodes and strain in 9 of the members.

Now my aim is to replicate my prac experiment as best as possible on Pro/E and simulate it as accurately as possible in Mechanica, to 'hopefully' get the same results as practice.

Please provide ANY information you think may help in simulating this as accurately possible.

- I will not be simulating the assembly, just the chassis.prt file
- The two rear constraints are a single bolt each side (pin joint)
- The front support is just a flange with the member resting on it

I highly appreciate the feedback.

Cheers.


Edited by: Vesh

 

[CaptnPea]

Element type: Beams.... welded together 0 DOF connections between.


If you are going to limit yourself to only using one part file vs an assembly this is the only way I see doing it. In my experience with working with tubular chassis (a long time ago in univ solor car) the results from FEA were generally a little conservative (showed more deflection than observed). This may be compensated by breaking your beams up and adding a thicker section (work the % of the I value)at the ends to simulate the strength gained by adding the weld material, as in the real situation the beam is supported a little further than at a point is in the analysis. BUT, be way careful, we broke a chassis that the analysis indicated we shouldn't have. We blew it by not being cognicent of the HEZ at the welds and our tubes (6061T6 Al alloy) lost a lot of strength at the HEZ. We tested welding samples in an tensile testing machine to adjust our assumptions then went back and modified the design, that second chassis ran 3-4 more races and only gained a little more weight than the original.


Constraints: (assumes blue-x red-y green-z)


Rear fix: TxTz Front Lft: Tz , but what about the Ty DOF? Not knowing your physical set, like how tight the bolts are, but, I would think that adding Ty to one of the rear nodes would get you in the ballpark.


As with all analysis, there are limits to what accuracy you can achieve, you should learn a lot, and I bet it will take many shots at it to find accurate correlation, you have a huge advantage in that you have allready created your physical model to varify against, use that tool to learn more about FEA.

 

[Vesh]

Thanks alot for the reply mate. I completely understand what your saying in regards to the structures nodes being the weak link in the model. I am getting results in FEA higher then practice (more deflection and strain).

How do I go about modelling the chassis as youve suggested, ie beam elements simulating welds?

For example if a member is 25.4mm OD & 1.2mm wall, are you saying apply a ~30mm OD and ~5mm wall for each end around say 1/2" in length?

ie, a beam element representation of the solid model below?

View attachment 2806

This makes me wonder whether modelling all the nodes using solid elements is a better idea? (Still keeping the majority of the length of the memebrs as beam elements)

Re constraints, those you provided seem right, although ill play with them later after I try the above modelling methods...

Cheers...


Edited by: Vesh

 

[CaptnPea]

I don't think solids are the answer... you'll be there a long time. You may be able to use shell elements though. This would allow you to start intersecting the tubes like they are in reality (I assumed fishmouthed together). I haven't done that but, it shouldn't be too hard to get the CL geometry, may be a bit tedious. I would start by playing games with just intersection and experiementing to get a method developed, then start doing the rest, stopping everyonce in awhile, and running a load case,to make sure things are connecting the way you like.


"For example if a member is 25.4mm OD & 1.2mm wall, are you saying apply a ~30mm OD and ~5mm wall for each end around say 1/2" in length?"


This sounds about what I was thinking, but, I would keep track of the I value and see how much that is changing, S=My/I, but this is a pretty quick and dirty fix. Asa guess I would go quite abit stronger 2x or more, and have the length at the end of the beam that I bolsters the"welded node"be about the same amount of tubing that is actully welded away from the therorectical intersection point at a node.

 

[Guest]

oh,it's so comlex.