tegwin
|
| posted on 24/12/07 at 03:19 PM |
|
|
Welding in solidworks...
Im just curious about this really...
Lets say I want to make an upright on Solidworks and then analyse it using cosmos stress analysis...
The upright is to be made out of various elements of mild steel cut and welded together....
So when I draw it on solidworks do I
A- Draw the finished upright from one solid block
B- Draw each steel component and then join them together using constraints/Mating...
Does solidworks give you the option to specify how the materials are joined? Because for stress analysis of a component I guess this is crucial...
I was also considering drawing my chassis on solidworks in component form and then doing stress analysis on the final assembly...is this possible?
(Again, this brings into question wether you can specify weld lengths etc?
Its proberably worth metnioning that I dont have solid works YET...But I was just wondering if I can achieve what I want with this program if I can
aquire it.
[Edited on 24/12/07 by tegwin]
|
|
|
|
|
flak monkey
|
| posted on 24/12/07 at 03:40 PM |
|
|
Cosmosworks (solidworks FEA package) will allow you to do either, but....
Bear in mind the more 'accurate' the model, the longer it will take to analyse, but the better your results. Most FEA models are actually
simplified versions of the actual design because of this, then the skill comes from deciding whats important in the model, and whats not, and then
interpreting the results accordingly. Even the upright could take an hour or more to analyse with a very basic model.
FEA is a dangerous thing, as the wrong constraints, or the wrong load types will give you wildly different results, all of which will be wrong. There
is very little skill involved in drawing the model, but a lot of skill in applying FEA and then analysing the results.
For the chassis you will need to combine all of the elements in the assembly before you can analyse it with Cosmos. Even then, when it comes to
analysing it you will probably crash your PC. Chassis' and other tubular structures are really 'best' modelled using beam elements
(in other words using single lines to represent the chassis members, and assigning a general property to them all. This will give you a reasonably
accurate result (not as accurate as analysing a full 3d one) but it should be within 10% if done correctly. You cant do this with solidworks though.
There are special FEA packages that are designed for beam structures.
For the upright, draw it as a solid as it will make it easier to analyse, and its less likely to throw up errors part way through the analysis. You
then need to add a safety factor, or you can run the analysis with your calculated loadings and then see what the saftey factor comes out to be after
the analysis (cosmos works will tell you). Welded joints (if done correctly) will be as strong as the parent metal, therefore drawing the part as a
solid will give you the best results. Dont forget to add fillets in each area where the weld bead would be.
If you are concerned about fatigue in welded structures (i assume this is what you are thinking) Cosmos cant help you. The best way to find out about
fatigue is build one and have it tested. There is a vibration and fatigue analysis package available for solidworks, but I have never used it so dont
know whats its like. The best course of action would be to make the upright and then have it heat treated to normalise the structure of the whole
thing. This will minimise the internal stresses in the finished thing. However you will get some distortion of the part as it is stress relieved.
A bit of a long rambling post that, but hopefully theres something useful in it.
David
Sera
http://www.motosera.com
|
|
|
tegwin
|
| posted on 24/12/07 at 05:06 PM |
|
|
Chears David....Thats really enlightening
Need to get a copy and have a play!
|
|
|
balidey
|
| posted on 24/12/07 at 06:25 PM |
|
|
As Flak said,. but a few things to add....
If you are using the comsos within SW, then this is cosmoslite, which is a cut down version of cosmos, you can only analyse solid parts, not
assemblies.
If modeling two plates together then in the ideal world you would model them with a 3mm proud fillet to represent the weld, but for FEA the best way
is to put an internal fillet in. Now this is not accurate, but you can't have sharp internal intersections as the meshing doesn't like
it.
Another point to labour on is that you really really need to know what you are doing with FEA. I have had a specialist company do an FEA project for
me (my company used Mayflower for a bit of freelance work) and they made a monumental cock up, they wasted weeks of our time because of not
understanding how we wanted to restrain an item. And these are supposed experts.
I can and do basic analysis with it, but you have to understand that it is a guide only and must never be used alone. Its a tool to reduce the number
of prototypes needed. So, if you are using it to develop an upright and you have never used it before then IMHO, don't rely on your data. By all
means do it as an excercise, i know I would, but know your limits. And i'm not trying to be patronising here.
I spoke to someone else in another design consultancy company about FEA of two parts welded together, they basically said, ooh, aah, well, best thing
you can do is add the internal fillets, so they say you have to compromise in this area, and again , these are FEA experts.
[Edited on 24/12/07 by balidey]
|
|
|
|
