Termal loads

[ThomasH]

Hi all

I have a question concerning termal loads. I want to use a gradient load om a model consisting of plates. I know that Nastran can do it but how can it be done with FEMAP as preprocessor. Termal load does as far as I can see only mean a temerature expansion over the entire section, not a gradient.

Does anybody have any experience, good or bad?

Regards

Thomas

 

[pjhype]

I think you can perform your thermal analysis, save the output and make the output set current, then go to Model, Load, From Output, and FEMAP will automatically define your nodal temperature loads. You should then use these nodal temperature loads in your structural analysis to perform a thermal-structural analysis. The thermal analysis can be either steady-state or transient. All you need to do is make the temperature output set that you are interested in for structural analysis the current set.

Generally, when you use nodal temperature loads you can define a temperature gradient. If you use thermal body loads, the entire "delta-T" gets applied to the entire model.

pj

 

[ThomasH]

Hi again

I think I have been a bit unclear. What I want to do is to apply a termal gradient on a plate element, one temperature on the upper and a different on the lower surface.

If I follow your idea, pj, the resulting load is a TEMP load in the Nastran bulk data. The TEMP load is based, as far as I know, on gridpoints. No element "location" specified.

What I need is a TEMPP1 load, elemental temperature. In Nastran bunk data I can than then give T1 one value and T2 another or the same value. This results in gradient if T1 not equal to T2 and "simple" expancion if T1 egual to T2.

My problem is that FEMAP writes T1 and leaves T2 blank, defaulting it I guess to T1. If I manually input a different value for T2 I get "preprocessor validation error". Basically, the N4W-bundle does not allow me to "hack" the bulk-data-deck.

I have been told that anything possible in Nastran is also possible in FEMAP, but sometimes it's a bit complicated.

Does anybody know a "complicated" mothod to get a temperature gradient?

Thomas