Transient thermal coupled with Structural Modal analysis

[harshn]

Hi,

I am performing a coupled analysis, in which I am trying to analyze effect of temperature on the modal frequencies of the structure.

For this the logic I am thinking of is to do a transient thermal analysis with controlled time step and then move to modal analysis after every time step do the modal analysis, save the modal frequency and return to next load step in thermal analysis.

I am not very comfortable with transient analysis, can someone advise me how can I implement this strategy.

Thanks and Regards
Harsh

 

[Stringmaker]

This sounds like a lot of unnecessary work and run time. What do you want to get of this analysis exactly? Would running thermal/modal analyses at a set number of points and comparing the results either graphically or another manner not serve your purpose here?

 

[cbrn]

Hi,
Harshn, just like Stringmaker says, I'd forget about this very complicated way (moreover, what you say is not possible: you can not interrupt a transient analysis to embed another kind of analysis between steps).
At best, you would have to write an APDL which would execute repeatedly the thermal (remember to activate prestress!!!) / modal analysis with varied boundary conditions. This would be useful only if you wanted to investigate more than, I'd say, 10 different conditions.
For normal engineering purposes, and unless you work for a gas/steam-turbines manufacturer, I think that 5 thermal conditions are sufficient to characterize the behaviour of a structure (the variation law of the stiffnes, i.e. of the frequencies, should not be of higher order than 2).

Regards

 

[harshn]

Hi Stringmaker and cbrn,

I agree with the fact that is a computationally expensive and it looks like an unnecessary computation. That is why I want to do this and once I have the temporal data of temperature and modal characteristic, I will like to fit a stochastic model to this and hope that it fits well and no one has to look at this computation again for implementing other strategies which makes extensive use of modal characteristics.

Regards

 

[cbrn]

Hi,
I understand. This task should be very easy with APDL. Only computationally expensive, but you can:
- save disk space by making ANSYS save only the results you need
- avoid to expand the modes, since the only thing you need is the frequencies
- make the model as fast as you can, for ex. by using SOLID92 instead of tetrahedron-degenerated SOLID187 (if you mesh with WB, for example), or by using mapped / swept mesh everywhere you can
- use relatively coarse mesh, since the eigenmodes extraction is accurate even with few elements
- prepare the APDL, verify it with a very simple simulacrum model, then let it run overnight if needed.

Regards