2. Validation

2.1. A to E models

For each model, we test the value of TOUS the Gauss points of the integration diagrams.

The values of CH1 are tested in “NON_REGRESSION”

The values of CH2 are tested in “ANALYTIQUE”

2.2. F modeling

This modeling projects a field from a FEM tetrahedron cell to an X- FEM tetrahedron cell. Field CH1est is therefore stored on a single Gauss point, while field CH2est stored on a family of 90 Gauss points. In addition, the interface in question corresponds to the division of the tetrahedron element into 4 sub-tetrahedra, which implies that only 60 Gauss points out of the 90 available actually store a value.

Let \(C\) be the value stored by the single Gauss point in field CH1. The first 60 Gauss points in field CH2stockent the value \(C\), while the last 30 Gauss points store the value 0, so the sum of the values of CH2est equal to \(60C\) and the largest stored value is \(C\).

In order not to make the command file heavier by using 90 analytical value tests:

• The only value of CH1est tested in “NON_REGRESSION”,

• The sum of the values of CH2 and the largest value of CH2 are tested in “ANALYTIQUE”.