Huge errors of nonlinear transient coupled thermal-electrical problem (linear vs quadratic elements)

[LeeAPower]

Dear Expertises

I have performed coupled thermal-electrical analyses, which involves fully coupled nonlinear transient thermal-electrical equation. For the analyses, only brick elements (default type) were considered.
When comparing the solution from linear brick elements with one from quadratic elements, there were huge difference in tempeature. In some regions (space), using quadratic elements approximated negative tempeature profile, which must be wrong...... Why there is huge difference in converged solutions from linear element and quadratic element??
From my understanding, the quadratic elements more accurately approximate the solutions in msot cases. Is there any exception??

Simply explaining my model, the case is 3D composite laminate (16 plies) with temperature-dependent thermal/electrical conductivity and nonlinear thermal radiation boundary.

Best regards,
Juhyeong Lee

 

[Mustaine3]

Read that:
Users Manual 6.5.2 Uncoupled heat transfer analysis -> Spurious oscillations due to small time increments

 

[IceBreakerSours]

"..quadratic elements more accurately approximate the solutions in msot cases."

Since I have more a structural background, I will stick to a structural view: It depends on the kinematics of interest. For instance, if there is contact involved, you'd want to stick to linear elements, and not quadratic ones.

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083