ABAQUS ERROR RADIATION PROBLEM

[emrecan88]

I am runnin simulations for heat transfer analysis counting conduction, when I add the effect of surface radiation to environment with *radiate or *sradiate , I am getting the error message:

***WARNING: Temperature at 1 nodes has fallen below absolute zero. Increment
will be attempted again with a new time increment. The nodes have
been identified in node set WarnNodeTempBelowZeroStep2Inc45.

CONVERGENCE CHECKS FOR EQUILIBRIUM ITERATION 1


AVERAGE HEAT FLUX 1.78 TIME AVG. HEAT FLUX 2.75
LARGEST RESIDUAL HEAT FLUX 7.470E-07 AT NODE 60122 DOF 11
LARGEST INCREMENT OF TEMP. 3.352E-02 AT NODE 60122 DOF 11
LARGEST CORRECTION TO TEMP. -6.868E-05 AT NODE 60122 DOF 11
THE HEAT FLUX EQUILIBRIUM EQUATIONS HAVE CONVERGED

MAXIMUM NON-PRESCRIBED INCREMENT IN TEMPERATURE IS 3.352E-02
TIME INCREMENT MAY NOW INCREASE TO 6.371E-05

***WARNING: TEMPERATURE AT NODE 3 (-1.50112E-03) HAS FALLEN BELOW ABSOLUTE
ZERO (0.0000). INCREMENT WILL BE ATTEMPTED AGAIN WITH A TIME
INCREMENT OF 1.59273E-05


***ERROR: TOO MANY ATTEMPTS MADE FOR THIS INCREMENT



Did anyone experience that problem, please help me?

My initiol condition is 373 K at each node, so negative temperature values at nodes are nonsense.



Thanks



Emrecan Soylemez

 

[corus]

It would be very unusual for convergence not to be obtained. I'd check your model constants, such as stefan boltzmann's constant to make sure you have the right units. I'd also check the node location for the warning message and the boundary conditions there about.

corus

 

[emrecan88]

All units are already checked. And nodes are well located. Any other opinion?

Thanks

 

[corus]

Do you have temperature dependent properties that may be incorrect?

corus

 

[emrecan88]

The properties are temperature dependent but they were checked.

 

[corus]

If the change in temperature is too severe then you can get divergence in some methods, though I'm not sure what method Abaqus uses to solve for radiation. You can improve convergence by setting the initial temperature of the body to the same order you expect the converged temperature to be at. Otherwise try a transient analysis so that the temperature change is relatively small for each step.

corus