| Author |
Message |
pinkerton5
Joined: 30 Nov 2005
Posts: 5
|
 Posted:
Wed Nov 30, 2005 4:48 pm Post subject:
beam elements? |
|
|
I am having a definte problem within cosmos works when it comes to meshing a frame I have drawn up for analysis. Let me try and be a bit more specific; the frame in question is actually a modified race car chassis that I am studying for my undergrad research, It consists of various tubing (mostly 1.5"x0.095 wall thickness, 1.5"x0.120 wall thickness, and 1.25"x0.095 wall thickness), of which I have drawn up cross sections. Now to construct the model I first constructed a 3d sketch and then selected individual lines of the sketch for the various structural members of the above mentioned cross sections etc... Where pieces of tubing meet I have utilized the trim function to create a fishmouth in the tubing, (however this won't work for me in some cases, i.e. when 1.5x0.095 meets 1.5x0.120). Now the REAL problem, when the frame comes up in cosmosworks it fails to mesh, more specifically it surface meshes ok, but the volume mesh fails. I can't keep decreasing the element size because it is ridicuiously small as it is, it seems as though solidworks/cosmos just can't find a mesh to be happy with all of the strange bends/joints in a race car chassis. Speaking with my advisor/professor (who doesn't really know solidworks, so far I'm self taught) he suggested the possibility of treating each individual tube as a beam element so the entire model wouldn't have to mesh(accuracy doesn't have to be super high), much like I have done before with an old version of ansys to create trusses. Is this possible in solidworks? Can anyone give me any ideas on how to begin to tackle this, I have been beating me head on my desk with this one for a few weeks now!!
|
|
| Back to top |
|
 |
TOP
Guest
|
| Posted:
Wed Nov 30, 2005 9:10 pm Post subject:
Re: beam elements? |
|
|
You are using the wrong FEA tool. Try NEiWorks or NEiNastran. Even
Cosmos/M will do beam elements. CosmosWorks is not for long slender
thinwall parts. |
|
| Back to top |
|
|
pinkerton5
Joined: 30 Nov 2005
Posts: 5
|
| Posted:
Wed Nov 30, 2005 9:48 pm Post subject:
not such good news |
|
|
i wish i could say that was good news to me, but considering that this is for a research project and my deadline is fast approaching i may have to try and be really creative with this. I wish I could give nastran a shot, however having the funds of a typical student (aka being broke) i was lucky to purchase the student version of solidworks/cosmosworks....which by the way they assured me would meet my fea needs perfectly!!! With that said I hopped over to nastran's site and even if i were to find the extra $$ for the student version, I don't think it could quite cover all the models I need to study....my school is of no help on this one either...as a physics major for some reason my department has a severe attitude when it comes to engineering topics....as a matter of fact I was EXTREMELY lucky to get my research approved (luckily one professor thought it was a good idea). In other words, since I'm not studying superconductors or nuclear physics I get zero help from the school as far as resources go. Any suggestions on what direction to head from here would be GREATLY appreciated!
|
|
| Back to top |
|
|
That70sTick
Guest
|
| Posted:
Thu Dec 01, 2005 8:42 am Post subject:
Re: not such good news |
|
|
Well, as far as lessons go,you are getting a great value.
KEY POINTS:
CAD expertise does not equate FEA expertise.
Sales people are not good resources for analyzing your FEA needs. |
|
| Back to top |
|
|
Jeff N
Guest
|
| Posted:
Thu Dec 01, 2005 5:10 pm Post subject:
Re: beam elements? |
|
|
The reason the mesh will not complete is probably related to running
out of memory when trying to mesh the model at a super fine resolution
or it may be having problems with aligning element nodes. I had similar
issues with small IC device models until I learned more about what the
FEA program was actually doing.
CosmosWorks recommends using shell elements for thin features such as
your tubular frame and in fact that method would be recommended by
anyone for this particular problem. The reason being that you don't
need a volume mesh to come up with a good plot. You can shell the
entire frame and specify the different thicknesses on different shell
areas. A good FEA Modeler/Engineer is always trying to optimize the
mesh model so that it's efficient (research "FEA convergence"). You
don't want too much unecessary elements because this increases the
amount of resources you need to solve it and also increase the chance
of errors due to a more complex mathmatical problem.
I would suggest starting a study using shells only and use a rough
setting for the mesh resolution. If that works, run the study and
identify the areas in which the stresses show up. You can then go back
and use mesh control to make those particular areas a denser mesh and
get more accurate plots.
You can also send me your model and I can let you know if I have any
success with meshing it. Just let me know and I can give you my gmail
address.
-Jeff
pinkerton5 wrote:
| Quote: | I am having a definte problem within cosmos works when it comes to
meshing a frame I have drawn up for analysis. Let me try and be a bit
more specific; the frame in question is actually a modified race car
chassis that I am studying for my undergrad research, It consists of
various tubing (mostly 1.5"x0.095 wall thickness, 1.5"x0.120 wall
thickness, and 1.25"x0.095 wall thickness), of which I have drawn up
cross sections. Now to construct the model I first constructed a 3d
sketch and then selected individual lines of the sketch for the
various structural members of the above mentioned cross sections
etc... Where pieces of tubing meet I have utilized the trim function
to create a fishmouth in the tubing, (however this won't work for me
in some cases, i.e. when 1.5x0.095 meets 1.5x0.120). Now the REAL
problem, when the frame comes up in cosmosworks it fails to mesh,
more specifically it surface meshes ok, but the volume mesh fails. I
can't keep decreasing the element size because it is ridicuiously
small as it is, it seems as though solidworks/cosmos just can't find
a mesh to be happy with all of the strange bends/joints in a race car
chassis. Speaking with my advisor/professor (who doesn't really know
solidworks, so far I'm self taught) he suggested the possibility of
treating each individual tube as a beam element so the entire model
wouldn't have to mesh(accuracy doesn't have to be super high), much
like I have done before with an old version of ansys to create
trusses. Is this possible in solidworks? Can anyone give me any ideas
on how to begin to tackle this, I have been beating me head on my desk
with this one for a few weeks now!! |
|
|
| Back to top |
|
|
pinkerton5
Joined: 30 Nov 2005
Posts: 5
|
| Posted:
Thu Dec 01, 2005 10:09 pm Post subject:
shell meshes |
|
|
| Thanks for the tip. You're abolutely correct about the memory. It is to the point where I simply can't create any finer of a mesh due to memory limitations and for that reason i had previously explored shell meshing. Now, having only previous FEA experience with an old student version of ANSYS in which the number of elements was extremely limited, the idea of attempting shell meshing seemed itimidating, but on the contrary it was easy enough to follow the principles and techniques used by COSMOS. However the problem with the shell mesh lies in the fact that while most of the frame meshes well, even with a relatively large global element size, some members simply do not mesh at all and "disappear". The members in question are mostly areas in which there are bends, or members tangent to bent members (i.e. the portion of the member directly before or after the bend). Attempting to correct this I applied mesh controls to the areas that were not meshing, and also tried applying contols to the members that were not meshing as well as any adjoining members....end result....nothing changes....the pieces that wouldn't mesh still won't. Could this have to do with some type of gap condition? Theoretically there shouldn't be any gap, but being new to solidworks I suppose it's possible I have done somethign else wrong. Any insight? By the way thanks for the offer to look at my frame mesh for me, I just may have to take you up on it if I can't sort this out.....I suppose I'd at least like to know if I'm going about things all wrong. |
|
| Back to top |
|
|
pinkerton5
Joined: 30 Nov 2005
Posts: 5
|
| Posted:
Thu Dec 01, 2005 10:21 pm Post subject:
actually |
|
|
| Actually I see now that I can't specify contact/gap settings unless using a solid model...Any other suggestions on how to mesh those "missing"elements? |
|
| Back to top |
|
|
That70sTick
Guest
|
| Posted:
Fri Dec 02, 2005 7:27 am Post subject:
Re: shell meshes |
|
|
If you're stuck with solid mesh and running out of memory, you might
still have options.
Run your analysis with a coarse mesh to get an idea of you hotspots.
Consider carefully both areas of high stress AND areas with high stress
gradient (large change over small distance). Then refine your mesh in
the hotspots.
Every time you refine, resulting stress values will most likely
increase. Keep refining until the resulting stress values taper off.
Many of the big-boy FEA packages can do this automatically. Very
important. It's called "convergence". |
|
| Back to top |
|
|
TOP
Guest
|
| Posted:
Fri Dec 02, 2005 9:10 am Post subject:
Re: not such good news |
|
|
Believe it or not there are some free FEA codes that will do what you
want. By the way, what is the goal of your study? |
|
| Back to top |
|
|
pinkerton5
Joined: 30 Nov 2005
Posts: 5
|
| Posted:
Fri Dec 02, 2005 2:37 pm Post subject:
re:beam elements? |
|
|
| I have attempted to refine the mesh to find some convergence in the numbers. Problem is.....thus far i can't even establish a good enough mesh in the first place. Either it is too fine of a mesh to create or some members are left out etc... The ultimate goal of the study is actually rather simple...see which frame members/section see a signifigant stress under a set of static load conditions. Just attempting to spot a trend in the load patterns of the members, maybe determine if some would be more well suited if they had thicker/thinner walls, different material type etc...All in all the results don't have to be super accurate, it would just be nice to see some reasonable stresses and deflections. |
|
| Back to top |
|
|
TOP
Guest
|
| Posted:
Mon Dec 05, 2005 8:07 am Post subject:
Re: beam elements? |
|
|
When you have singularities in your mesh, refining will only make them
worse. The tool you have really isn't appropriate for what you are
trying to do. |
|
| Back to top |
|
|
wurz
Guest
|
| Posted:
Mon Dec 05, 2005 5:10 pm Post subject:
Re: beam elements? |
|
|
TOP is right - the tool you have is not the right tool for the job.
Jointed tubes are best analysed as beams, plain and simple.
Try googling for £free fea software". Anything that analyses trusses
or frames could do the job.
Martin |
|
| Back to top |
|
|
Alan Krem
Guest
|
| Posted:
Mon Dec 05, 2005 9:10 pm Post subject:
Re: beam elements? |
|
|
I typically do FEA on solid models (engine parts) but when I do a race
car chassis I use beam elements. |
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in
