Element = "p:2" VS Salome Second Order Mesh

[CrocoDuck]

Hi Guys,

I have seen that many solvers offer the possibility of specifying the type and order of the basis function through the Element keyword. As far as I understand, specifying Element = "p:2" will increase the element order from linear to quadratic for all solvers that support p-elements.

As far as I understand, the order of the element and the number of nodes in it goes hand to hand. I guess than that Elmer will calculate the coordinates of the mid-nodes when using Element = "p:2"?

Also, in the past I normally relied on Salome to produce second order meshes, as in the example below.



My assumption was that when a mesh is created with Salome with the Second order box ticked, and then exported as a UNV to be used in Elmer (or with salomeToElmer) then Elmer will treat the elements as quadratic and apply quadratic base functions, correctly recognizing the vertex nodes and the mid-nodes in the mesh. Is this the correct assumption?

[raback]

Hi

There are two ways to introduce higher order elements in Elmer.

a) You can use nodal higher order elements that should be defined already in the mesh generator. This is the approach you are describing. Sometimes there may be hick-ups mainly because higher order elements have more issues with ordering and the chosen workflow might not support all the elements.

b) You can use the hierarchical p-element base, this is done with "Element = p:2", for example. Not all modules support this as it was not there in the early years. The hierarchical bases uses some smart multiplication and can go up to p:10. There number of dofs and also integration rules grow rapidly but often p:2 or p:3 are worth the effort. It should be noted that the p-elements are not seen in visualization since the mesh used for visualization is just linear.

-Peter

[CrocoDuck]

Hi Peter,

Thank you, that clarifies.

Cheers!