how good is Proe mechanica for FEA

[dopa]

hi
I want to know how good is proe mechanica for FEA ....I know that there are some good softwares like ansys and catia which are generally used... will the reults be different if i use proe mechanica structural analysis.

[dopa]

hi
I want to know how good is proe mechanica for FEA ....I know that there are some good softwares like ansys and catia which are generally used... will the reults be different if i use proe mechanica structural analysis.

[Conor]

Dopa,

If you are good at modeling in ProE, then Mechanica is a great FEA package. Compared to Ansys or Algor, I think it's much much more user friendly. We also use the Mechanica package where I work and our engineers have been able to fully replicate FEA results with physical testing - it's good stuff.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by dopa:
hi
I want to know how good is proe mechanica for FEA ....I know that there are some good softwares like ansys and catia which are generally used... will the reults be different if i use proe mechanica structural analysis. </div></BLOCKQUOTE>

[dopa]

thanks Conor http://fsae.com/groupee_common/emoticons/icon_smile.gif

[roostmeyer]

CosmosWorks is more user friendly, but if your using Pro-E, like Conor said its the way to go.

I will say that I've had trouble in the past with mechanica and the default unit system, but its nothing you shouldn't be able to figure out.

[Pennyman]

Mechanica and Cosmos are both user friendly, but I'm not sure that they incorporate more complicated "dynamic" loading conditions. For the static loads they're great, but you'll need a more advanced validation program if you choose to test your models in dynamic situations.

I know this has probably been covered many times over too, so simply searching "FEA" or "dynamic loading" will most likely help you out too.

[Gareth]

First, you should read a book on FEA. Garbage in is always garbage out. There is a lot to learn in this field...

Mechanica uses a variable order 'p' element (the p is for polynomial) unlike most other solvers that use 'h' elements. Pro/M has some logic to increase the element order until it's satisfied that it's reached convergence. The problem is that near artificial stress concentrations the stress field is non-physical and the solver spends several iterations trying to converge on a solution that will never be correct. The upside is that it's not as sensitive to mesh quality so you can get away with the meshing tools in Pro/M.

As for all the other solvers that use h elements the only way to ensure grid convergence is by iteratively meshing the parts until you have adequate consistency. Of course this isn't practical for every part so it takes time/experience/reading to understand where a fine mesh is needed. The benefit is faster solution times once the part is meshed. If you going to tet-mesh you should use 2nd order tet10 elements to get decent results. The solution time is longer...but the results will be meaningful. Tet4s are too stiff.

If you're looking for solver recommendations then I'd suggest ABAQUS for linear/non-linear statics. Ansys is good too, but it's harder to learn, unless you're using Workbench. Be careful with WB as it's a little like CosmosWorks and tries to hide the mesh. You're doing finite ELEMENT analysis...if you can't see the elements, how do you know you're getting good results? Sorry for the gripe...I've seen some unbelievably wrong results come out of it because the user wasn't aware of what WB was doing. Again, the solver is very good. LS-Dyna is the most commonly used crash/impact FE solver. If you're doing optimization then OptiStruct is the best tool for the job.

When I looked at Cosmos a few years ago I was impressed with its ease of use, but not its actual setup. I found it difficult to create the correct constraints and its meshing tools are nearly non-existent. Perhaps things have changed, but it definitely seemed like a shoot-first-ask-questions-later kind of tool. For this reason, I would be very surprised if you got the same answer in Cosmos as any other solver. For even the most simple problems it takes a surprisingly fine mesh to get 'exact' answers (think benchmark beam-in-bending problems) but this is mostly because FE is an approximation.

I weren't asking about dynamics, but for those interested Pro/Mechanism uses the ADAMS solver behind the scenes. I've used in the past and it's okay. It's not really setup to do much more than simple mechanisms though. I looked at damper/anti-roll kinematics and in hindsight I would have just used ADAMS. I was in the same boat as you...I knew Pro/E and I didn't know ADAMS. The problem with those decision is that inevitably you're constrained by the software. It's usually worth choosing the right tool for the job, even if you have to spend time learning it. I was using Wildfire.

Hope this helps...

[Conor]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Gareth:
First, you should read a book on FEA. Garbage in is always garbage out. There is a lot to learn in this field...

Mechanica uses a variable order 'p' element (the p is for polynomial) unlike most other solvers that use 'h' elements. Pro/M has some logic to increase the element order until it's satisfied that it's reached convergence. The problem is that near artificial stress concentrations the stress field is non-physical and the solver spends several iterations trying to converge on a solution that will never be correct. The upside is that it's not as sensitive to mesh quality so you can get away with the meshing tools in Pro/M.

As for all the other solvers that use h elements the only way to ensure grid convergence is by iteratively meshing the parts until you have adequate consistency. Of course this isn't practical for every part so it takes time/experience/reading to understand where a fine mesh is needed. The benefit is faster solution times once the part is meshed. If you going to tet-mesh you should use 2nd order tet10 elements to get decent results. The solution time is longer...but the results will be meaningful. Tet4s are too stiff.

If you're looking for solver recommendations then I'd suggest ABAQUS for linear/non-linear statics. Ansys is good too, but it's harder to learn, unless you're using Workbench. Be careful with WB as it's a little like CosmosWorks and tries to hide the mesh. You're doing finite ELEMENT analysis...if you can't see the elements, how do you know you're getting good results? Sorry for the gripe...I've seen some unbelievably wrong results come out of it because the user wasn't aware of what WB was doing. Again, the solver is very good. LS-Dyna is the most commonly used crash/impact FE solver. If you're doing optimization then OptiStruct is the best tool for the job.

When I looked at Cosmos a few years ago I was impressed with its ease of use, but not its actual setup. I found it difficult to create the correct constraints and its meshing tools are nearly non-existent. Perhaps things have changed, but it definitely seemed like a shoot-first-ask-questions-later kind of tool. For this reason, I would be very surprised if you got the same answer in Cosmos as any other solver. For even the most simple problems it takes a surprisingly fine mesh to get 'exact' answers (think benchmark beam-in-bending problems) but this is mostly because FE is an approximation.

I weren't asking about dynamics, but for those interested Pro/Mechanism uses the ADAMS solver behind the scenes. I've used in the past and it's okay. It's not really setup to do much more than simple mechanisms though. I looked at damper/anti-roll kinematics and in hindsight I would have just used ADAMS. I was in the same boat as you...I knew Pro/E and I didn't know ADAMS. The problem with those decision is that inevitably you're constrained by the software. It's usually worth choosing the right tool for the job, even if you have to spend time learning it. I was using Wildfire.

Hope this helps... </div></BLOCKQUOTE>

I've been doing some recent graduate research utilizing Optistruct and it is one awesome program. I had never heard about it during my years of FSAE, but would have taken full advantage of the topology optimization abilities of the software. My only complaint is that it could be more user friendly like Mechanica, but I think Altair is working towards that. If anybody has a chance to get their hands on this program, it's VERY powerful.