Large Scale Mixed Mesh Analysis

[sema79]

I have a client that wants me to model a large structure in solidworks and run analyses on the entire model with simulation. They want the solid model to be used in simulation with 3D meshing (no surfaces). Parts of the model will be very large (1000's of square feet) and very thin (<1" thick). I doin't think it will ever run as 3D meshing but this is a perfect situation to use mixed meshing with surfaces and solids. I may be able to satisfy them if I can produce a mixed mesh study that gives good results. I really need to specify a mesh size on the surfaces in feet and mesh on the solid components in inches. I would really like to avoid modeling smaller portions in problem areas until after the larger model has indicated where the problems are. This is kind of what the client is requiring. I would like to hear from someone else as to whether this is feasable before I tell him we have to run smaller studies only. Does anyone know the most efficient way to accomplish this? Any advice on the situation is greatly appreciated.

 

[EngAddict]

You shouldn't be using all solids. Quite possibly with large models, mostly shells with only a few solids (if required). However I would need to see the model. Probably steer clear of beam elements by the sounds of what they are after.

Just got to be careful with mixed meshing that you can verify your results. Run it rough first, draft quality mesh and then refine in high stress areas like you say. Manual mesh control in critical and some non-critical areas. Run high quality and check convergence when your model is working well. Check both deflection , stress and reactions by hand in appropriate areas to ensure your mixed mesh model is accurate.

There is no reason why you can't go this approach if you can prove the results to be valid.

 

[corus]

Clients have generally no idea what is onvolved in modelling and just assume you can include every little nut and bolt in a model when it can be quite easily broken down into smaller more manageable models. You could of course have huge elements in your model that will allow it to be run in 3D solida, and no doubt when they see the colour pictures they'll be very pleased, not that the results will have any use. Convince them that the way they suggest is wrong and you can produce more meaningful and more accurate results by using smaller models of parts of the whole.

Tata

 

[kellnerp]

If you model it in Cosmos/M which you should have no problem. You should have Cosmos/M if you have CosmosWorks. I would be leary of doing this in CosmosWorks because you may run into some undocumented limitation or quirky behavior.

The fact that they are requiring you to use bad practice might make me leary of the client. They may want both global answers and local high detail answers. This suggests making more than one model and if they don't understand that there may be trouble for you down the road.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."