Surface temperatures of shell elements?-NX NASTRAN THERMAL

[tlemonds]

Hey guys,

I am using the NX NASTRAN non-linear transient thermal solver to analyze some metallic and composite near-firezone structures. Post-processing in FEMAP. I have everything modeled as plate elements and I'd like to keep it that way.

Does anyone know how to (if possible) display/output bottom and top surface temperatures?

I've read the NX NASTRAN Thermal Analysis Guide manual front and back several times, looked through all the examples...cannot find any explicit or implicit mention of this capability. I do know that temperatures are calculated at grid points, and the element temperatures I am seeing displayed on FEMAP are really just averages based on those grid point temperatures. But I don't understand the nitty gritty of how those grid point temperatures and their corresponding elemental average temperatures relate to the two faces of a plate element. I can see when I change the thickness on my PSHELL card that--logically--the temperatures change as well. So obviously the solver consider the thickness in conjunction with conductivity to generate temperatures.

My best guess is that they are to be considered as "mid-plane" temperatures. If this is the case, I suppose I could just hand-calc the top and bottom surface temperatures based on thickness and conductivity. Could anyone confirm or deny this theory?

I am attempting to correlate to test data taken with thermocouples mechanically fastened to composite/metallic components. This is a radiation dominated problem so relating surface temperature to the physical/test surface which the thermocouple was fastened to is pretty important.

Thanks,
Tyler

 

[tlemonds]

Upon further investigation, I've found that my log file is reporting the exact same number of degrees of freedom as I have grid points. That would seem to suggest that I only have one degree of freedom per grid point. If that is the case, I would think only in-plane conduction is taking place. Perhaps when I was tweaking with thickness earlier and saw change in effects, all I was accomplishing was increasing the mass calculated and used in conjunction with specific heat.

So I'm beginning to think my model does not perform through-thickness conduction. There has to be a way for NX NASTRAN to consider out-of-plane conduction without modelling my article as solid elements...

 

[tlemonds]

Upon inquiry to Siemens, I have found out the only way to do what I intend is to model everything as CHEXA's instead of CQUAD4's.