I created two input files to solve a radiation heattransfer problem in ANSYS, one utilizing the matrix method and the other using the Radiosity method. The results are completely different, even though I have received no error messages. Why? In addition, if I run the Matrix method after running the Radiosity method, with input files beginning with /CLEAR, the result of the matrix method changes. Why? Should the answers from the two methods not be very close?
The two phenomena mentioned are related by the three factors below. This difficulty occurs if the Stefan Boltzman constant is defined in /SOLU rather than /AUX12. (1) Despite the fact that TOFF and STEF are valid commands in /SOLU (i.e. for Radiosity), entry into the /AUX12 routine initializes all the matrix method commands to the default values; therefore, the Matrix method resets the Stefan Boltzman value to the default value, a much lower number than, say, the proper SI value that might be set earlier in the input file. (2) Meanwhile, the /CLEAR command does not completely wipe out the effects of a radiation solution. In particular, it does not reinitialize the number of radiation surfaces, which is set with either the Radiosity or the Matrix method, so after the Radiosity solution, there is information left over. (3) The /AUX12 initialization will be skipped if the code senses that the number of radiation surfaces is not zero, therefore, the matrix input file will run correctly the second time through and give a different answer (this time the correct one!) than it did the first time. The proper way to proceed is to simply make sure that ALL the commands related to the matrix method are included in the /AUX12 portion of the input file. Emissivities and the Stefan Boltzman constant are the key parameters. Note that the answers for Radiosity and /AUX12 are not exactly the same, but they are very close. This is expected, as the methods are not identical. 

