It depends how you build it and what your interests are. If the members can be modeled as truss, beam, etc. it will be fastener than shells than solid elements. Having no idea what your goals are, it could take from seconds/minutes to orders of magnitude greater.
Generally, you would build a model that gave you a reasonable workflow and make assumptions beyond that. Some people try to capture too much detail and never ending solving the problem (i.e. they decide test is better or blow the budget). I remember one example where a person tried to account for all nonlinear behavior and was "almost there" for months. I took a stab at a simplified solution with GOOD assumptions and it matched the later performed test quite well.
It depends on how fine you want the mesh to be.
For example, your mesh might be as low as few tens of thousands, and could be up to several million elements.
So, you should balance your mesh with your computer capability.
Good luck
By the way, your problem won't be in the "meshing hours" but in solving the FE model, if your mesh consists of millions of elements, I don't think you will be able to get a solution (I might be wrong).
"By the way, your problem won't be in the "meshing hours" but in solving the FE model, if your mesh consists of millions of elements, I don't think you will be able to get a solution (I might be wrong). "
Yes you are wrong pretty much. Solvers like Nastran and so on can cope with ridiculously large number of elements. Even on my little old laptop models with half a million or more DOF could be solved quickly, in RAM, and if they were too big for that they page to disk.
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376
Starting from the complete CAD model. Assuming that the analysis goal is calculate the whole frame, you have to import the files into the pre processor, extract all midsurfaces, connect with a lot of rigid elements, model the suspension, simplify a lot of things and you be talking about 800 hours of work. The mesh quality is an issue and takes longer to make than sloppy meshes.
The model size depends but can be solved in a regular dektop computer.
"How many hours should be taking for mesh whole bus model...?"
how long is a peice of string ?
more usefully ...
how much experience do you have, with FEA ?, with stress analysis ?, with bus structures ?
how reliable an answer do you want ? are you going to test as well (so you're not completely relying on the FEA results) ? or not (so people's lives depend on getting the right answers) ??
are you building a new bus, or tweeking an old design ??
sorry, but i don't see how someone can reply "800 hours"
That's kinda an easy question to answer. 800 hours is how much we charge for doing it. Its not permitted by law or by reason to start building tools or production vehicles without a lot of testing, so, yes a prototype will be built and tested in a proving ground (700 laps in our case, which takes about 4 months to complete). The test vehicles are fully instrumented. The strain gages and accelerometers are put in the places indicated by the numerical results.
Its not perfect and failures not predicted by the numerical models take place sometimes. But mostly the models are pretty reliable. Once the model for one specific vehicle is ready and validated, any changes like a new cross member design can be done very easily.
