Is this skinner bug?

[In-hak Min]

I am using skinner class.
I have created 3 elements, that mixed 2 hex and 1 prism. (Please refer the attachment.)

‌

Why the extracted skin quad count is 13? 11 is right.
Skinner is not supported mixed type?

Here is the reproduce code.

---

#include "moab/Core.hpp"
#include "moab/Skinner.hpp"

moab::Core *mb;

int main()
{
    mb = new moab::Core();

const unsigned NUMVTX = 14;  // The number of vertexes

// This double array stores the x, y, z coordinate of each vertex.
    const double vertex_coords[3 * NUMVTX] = {
      6776.294, 1105.722, 522.2,
      6588.656, 1105.722, 471.4,
      6776.303, 1056.417, 471.4,
      6776.294, 1105.722, 471.4,
      6588.656, 1056.417, 565.7191,
      6588.656, 1105.722, 565.7429,
      6588.617, 1056.417, 471.4,
      6682.527, 1056.417, 471.4,
      6588.611, 1056.417, 522.2,
      6682.476, 1105.722, 522.2,
      6682.476, 1105.722, 471.4,
      6682.424, 1056.417, 522.2,
      6588.656, 1105.722, 522.2,
      6776.303, 1056.417, 522.2
    };

moab::EntityHandle conn_hex[2][8] = {
        { 8, 11, 2, 7, 12, 10, 13, 9 },
  { 11, 8, 3, 4, 10, 12, 14, 1 }
};

moab::EntityHandle conn_prism[1][6] = {
        { 12, 5, 6, 10, 9, 13 }
};

moab::ErrorCode rval;
    for(int i=0; i<NUMVTX; i++) {
        float x = vertex_coords[i*3];
        float y = vertex_coords[i*3+1];
        float z = vertex_coords[i*3+2];
        std::vector<double> coord(3);
        coord[0] = x;
        coord[1] = y;
        coord[2] = z;
        moab::EntityHandle h;
        rval = mb->create_vertex( coord.data(), h );
    }

moab::Range range;

for(int i=0; i<2; i++) {
        moab::EntityHandle element;
        rval = mb->create_element( moab::MBHEX, conn_hex[i], 8, element );
        range.insert( element );
    }
    for(int i=0; i<1; i++) {
        moab::EntityHandle element;
        rval = mb->create_element( moab::MBPRISM, conn_prism[i], 6, element );
        range.insert( element );
    }

mb->add_entities(0, range);

int count;
    mb->get_number_entities_by_type(0, moab::MBHEX, count);
    printf("Total hex count is %d\n", count);

mb->get_number_entities_by_type(0, moab::MBPRISM, count);
    printf("Total prism count is %d\n", count);

moab::ErrorCode ret;

moab::Range faces;
moab::Skinner skinner(mb);
ret = skinner.find_skin(0, range, false, faces);

moab::Range tri_range;
unsigned int tri_count;
moab::Range quad_range;
    unsigned int quad_count;

for( moab::Range::iterator it = faces.begin(); it != faces.end(); ++it ) {
moab::EntityType type = mb->type_from_handle(*it);
if(type == moab::MBTRI) tri_range.insert(*it);
if(type == moab::MBQUAD) quad_range.insert(*it);
}

printf("Total skin triangle count: %d\n", tri_range.size());
printf("Total skin quad count: %d\n", quad_range.size());

delete mb;
}

[Vijay M]

Thanks for the test program. Looking at the code, nothing looks odd. But I’ll see if I can replicate the issue and get back to you.

[In-hak Min]

@Vijay Mahadevan Thank you. :)

[Iulian Grindeanu]

that is not a prism, according to our conventions:

‌

[In-hak Min]

actually i want to use penta. but i can not found penta type in moab. is same shape penta and prism? i think prism is seems penta.

[Iulian Grindeanu]

35 moab::EntityHandle conn_prism[1][6] = {
36 { 12, 9,5, 10, 13, 6 }
37 };

with the correct one you get expected result
./test4
Total hex count is 2
Total prism count is 1
Total skin triangle count: 2
Total skin quad count: 11

[Iulian Grindeanu]

‌

[In-hak Min]

is this wrong vertex connection?

i will check after some hours.

and get back result.

thank you.

[Iulian Grindeanu]

yes, prism is triangular prism; we do not use penta as a name; pentahedron ? has indeed 5 faces

[In-hak Min]

We called penta for “pentahedral” in LsDyna.
And below image is supported vertex connection in inputdeck.

‌

So I have translate connection as below when build moab::MBPRISM.

1, 2, 3, 4, 5, 6 → 1, 2, 5, 4, 3, 6

Is this right?

Thank you.

[Vijay M]

@Iulian Grindeanu Did this get resolved in the mailing list ?

[Iulian Grindeanu]

yes, that would be correct, our convention is different compared to LSDYNA, but it is similar to any other convention, vtk, cubit, nastran, ansys

[Iulian Grindeanu]

LsDyna has a different convention for prism