When a thermal analysis includes some type of nonlinear effect (temperature-dependent properties, radiation, and so on), the log file will include the convergence history. Once the iterations begin, monitoring this text will help the analyst decide if the analysis needs to be stopped to change some parameters. Or if the analysis needs to be performed again, changing the convergence parameters may lead to a faster solution.
The columns in convergence history are as follows (see Figure 1):
- Nonlin Iter: this is the iteration number. The interval at which the iterations are printed are controlled on the Analysis Parameters.
- Incr. Norm(T): this is the incremental norm based on temperatures. This value is compared to the Corrective tolerance entered on the Analysis Parameters
Advanced tab. - Incr. Norm(F): this is the incremental norm based on the liquid fraction. This appears in the printout only when phase change is included in the analysis (not shown in this example). This value is compared to the Corrective tolerance entered on the Analysis Parameters Advanced tab.
- Incr. Norm(Q): this is the incremental norm based on heat fluxes. This appears in the printout only when body-to-body radiation is included in the analysis (as in this example). This value is compared to the Corrective tolerance entered on the Analysis Parameters Advanced tab.
- Rel. Norm(T): this is the relative norm based on temperatures. This value is compared to the Relative tolerance entered on the Analysis Parameters Advanced tab.
- . Norm(F): this is the relative norm based on the liquid fraction. This appears in the printout only when phase change is included in the analysis (not shown in this example). This value is compared to the Relative tolerance entered on the Analysis Parameters Advanced tab.
- Rel. Norm(Q): this is the relative norm based on heat fluxes. This appears in the printout only when body-to-body radiation is included in the analysis. This value is compared to the Relative tolerance entered on the Analysis Parameters Advanced tab.
| 01 | **** BEGIN NON-LINEAR ITERATIONS | ||||
| 02 | Nonlin Iter. | Incr. Norm(T) | Incr. Norm(Q) | Rel. Norm(T) | Rel. Norm(Q) |
| 03 | ------ | ------ | ------ | ------ | ------ |
| 04 | 1 | 1.134E+02 | 1.024E+02 | 2.630E+00 | 3.333E+00 |
| 05 | 6 | 8.471E+00 | 1.039E+02 | 1.351E-01 | 4.422E-01 |
| 06 | 11 | 2.025E+00 | 2.997E+01 | 3.033E-0 | 9.484E-02 |
| 07 | 16 | 3.682E-0 | 7.019E+00 | 5.461E-03 | 2.085E-02 |
| 08 | 21 | 6.236E-02 | 1.504E+00 | 9.233E-04 | 4.409E-03 |
| 09 | 25 | 1.488E-0 | 4.227E-01 | 2.203E-04 | 1.236E-03 |
| 10 | **** END NON-LINEAR ITERATIONS | ||||
| 11 | |||||
| 12 | **** CONVERGED SOLUTION OBTAINED | ||||
|
Figure 1: Sample Text From Log File
Line numbers were added to help with the following description. |
|||||
With the above explanation in mind, the sample log file can be interpreted as follows:
- On the first iteration (line 04), the calculated temperatures changed by an average of 113 degrees, or 263%. The heat fluxes changed by 102.4 units and 3.333%.
- By the sixth iteration (line 05), the calculated temperatures changed by an average of only 8.47 degrees.
- On the 25th iteration (line 09), the calculated temperatures changed by an average of 0.01488 degrees, and the heat fluxes changed by an average of 0.4227. Since the tolerances in this example were to converge when the corrective norm was less than 0.5, the solution has converged.
- Had the solution been diverging or oscillating (the tolerances getting smaller, then increasing), the relaxation parameter could be decreased, or a more accurate initial temperature could be applied to the model.
- When the analysis includes radiation, two constants are calculated by the processor:
- The Stefan-Boltzmann constant is based on a value of 5.669E-8 W/(m 2 -K 4 )
- The temperature increment to absolute temperature is based 273.15 (K = °C + 273.15).
- Each constant is converted to the units set in the model. The summary file can be checked to see the actual values used in the analysis.
When the analysis includes temperature-dependent heat generation, the amount of heat generated is not known until the temperatures are calculated. The problem becomes an iterative solution as described above. The calculated heat generation is given for the last three iterations for each time step. For transient heat transfer, these results are in the summary file for each time step and the log file for the final time step. For steady state heat transfer, these results are given in both files. (Use the Report environment to view the summary and log files.) For example, the following lines
**** The specified heat source generated
8.233E+00 8.234E+00 8.235E+00 (Energy/time)
in the last 3 iterations, respectively.
show that the heat generated changed only slightly during the last three nonlinear iterations, which is expected for a converged solution. More importantly, the final value of 8.235 energy/time is the sum of all the temperature-dependent heat generation in the model, where the units of energy and time are based on the Model Units as listed near the beginning of the file.