6. Summary of results

The modeling carried out on \(1/8\) of the structure gives results where many values exceed the tolerance set initially (\(\text{2\%}\)). The maximum difference obtained is \(\text{89\%}\), it is located at the lowest reference value. The analysis of the isotherms shows that they are not perpendicular to line \(\mathrm{DC}\), the symmetry condition is not respected.

To find an explanation for these important differences, several complementary models have been carried out (see AnnexB). The conclusions are as follows:

• the change in the \((°C\to °F)\) unit system makes it possible to reduce the maximum difference to a value of \(\text{33\%}\),

• modeling with quadratic elements (and in \(°F\)) improves the results, the maximum difference is \(\text{6.8\%}\),

• modeling a \(1/4\) of the structure with only QUAD4 (and in \(°F\)) improves the results, the maximum difference is \(\text{–5.27\%}\),

modeling :math:`A`, performed with the NISA software, gives results identical to those of*Code_Aster.

In addition, it was not possible to obtain the original reference (Kreith’s book) cited in the ANSYS audit manual. The method for obtaining the reference solution and its uncertainty are therefore not known.

The results are considered to be acceptable taking into account the points mentioned above. However, it will be necessary to look for additional elements on the reference solution.