Multi material support #173
Conversation
|
@kinanbezem7 feel free to try this out or ask about the user interface here |
|
I tried using for the multi_material, inputs such as: "density" : {"value": [8000, 4000], "unit": "kg/m^3"}, and then defining the matrix and fiber densities ass: However, we use: double rho = inputs["density"]["value"]; in CabanaPD_Input.pp. Similarly, with elastic_modulus, etc. So, this needs to be generalized to work with multi-material systems. In addition, for multi-material systems, the critical time step should be the minimum of the critical time steps of any of the materials. |
src/CabanaPD_ForceModels.hpp
Outdated
| // No-op for temperature. | ||
| KOKKOS_INLINE_FUNCTION | ||
| void thermalStretch( double&, const int, const int ) const {} | ||
| void thermalStretch( const int, const int, double& ) const {} |
There was a problem hiding this comment.
Is the change "const int, const int" to enable two particle indices, instead of one, for multi_material?
There was a problem hiding this comment.
No, it's particle ids
| @@ -36,38 +38,72 @@ struct ForceModel<LPS, Elastic, NoFracture> : public BaseForceModel | |||
| double theta_coeff; | |||
| double s_coeff; | |||
|
|
|||
| double c; | |||
There was a problem hiding this comment.
Why do we have "c" in the LPS?
There was a problem hiding this comment.
What does "duplicate" mean. Just a copy/paste mistake?
There was a problem hiding this comment.
No, just that I don't want to discuss the same thing in multiple comments
| theta_coeff = 3.0 * K - 5.0 * G; | ||
| s_coeff = 15.0 * G; | ||
|
|
||
| // Equivalent PMB inputs. |
There was a problem hiding this comment.
Why do we have "c" in the LPS?
There was a problem hiding this comment.
Exactly what the comment "equivalent PMB inputs" intends - if you want to mix model types you need some kind of interoperability
| ForceModel( const double _delta, const double _K, const double _G, | ||
| const double _G0, const int _influence = 0 ) | ||
| : base_type( _delta, _K, _G, _influence ) | ||
| , G0( _G0 ) |
There was a problem hiding this comment.
Why this change for multi_material?
There was a problem hiding this comment.
Just a side effect of the changes
| { | ||
| set_param( delta, K ); | ||
| c = 18.0 * K / ( pi * delta * delta * delta * delta ); |
There was a problem hiding this comment.
Why using "pi" is OK? Shouldn't we use CabanaPD::pi?
There was a problem hiding this comment.
It's in the CabanaPD namespace already
| { | ||
| c = model.c; | ||
| K = model.K; | ||
| K = ( model1.K + model2.K ) / 2.0; |
There was a problem hiding this comment.
Why do we need this line? We only need the average of "c"
There was a problem hiding this comment.
We should not leave values unset. If we do what you suggest we should not keep K at all
There was a problem hiding this comment.
So, in the "fiber_reinforced_composite.cpp" example, we have:
double theta = stacking_sequence( nply ) * CabanaPD::pi / 180;
We also have: #include <CabanaPD_Constants.hpp>.
So, could we simply use the following?
double theta = stacking_sequence( nply ) * pi / 180;
There was a problem hiding this comment.
No, it's not in the CabanaPD namespace
| G0 = model.G0; | ||
| s0 = model.s0; | ||
| bond_break_coeff = model.bond_break_coeff; | ||
| s0 = Kokkos::sqrt( 5.0 * G0 / 9.0 / K / delta ); |
There was a problem hiding this comment.
We need to make sure K here is not the average K.
| base_type::set_param( _delta, _K ); | ||
| G0 = _G0; | ||
| s0 = Kokkos::sqrt( 5.0 * G0 / 9.0 / K / delta ); | ||
| G0 = ( model1.G0 + model2.G0 ) / 2.0; |
There was a problem hiding this comment.
What is this line for?
src/force/CabanaPD_Force_LPS.hpp
Outdated
| double m_j = mu( i, n ) * model.influence_function( xi ) * xi * | ||
| xi * vol( j ); | ||
| m( i ) += m_j; | ||
| m( i ) += mu( i, n ) * model.weightedVolume( xi, vol( j ) ); |
There was a problem hiding this comment.
Using "model.weightedVolume" here is confusing because "m" is the weighted volume.
src/force/CabanaPD_Force_LPS.hpp
Outdated
| if ( m( i ) > 0 ) | ||
| theta( i ) += 3.0 * theta_i / m( i ); | ||
| theta( i ) += mu( i, n ) * | ||
| model.dilatation( s, xi, vol( j ), m( i ) ); |
There was a problem hiding this comment.
Same as above, here theta is the "dilatation", so "model.dilatation" could be confusing.
src/force/CabanaPD_Force_LPS.hpp
Outdated
| const auto vol = particles.sliceVolume(); | ||
| auto linear_theta = particles.sliceDilatation(); | ||
| auto theta = particles.sliceDilatation(); |
There was a problem hiding this comment.
Why do we change "linear_theta" to "theta"? We still use "linear_s" below.
src/force/CabanaPD_Force_LPS.hpp
Outdated
| auto model = _model; | ||
| auto neigh_list = _neigh_list; | ||
|
|
||
| const auto vol = particles.sliceVolume(); | ||
| const auto linear_theta = particles.sliceDilatation(); | ||
| const auto theta = particles.sliceDilatation(); |
Variadic models, only support binary currently
Redefine model inheritance to avoid diamonds
streeve
force-pushed
the
multi_material
branch
from
9d7db05 to
66e534d
Compare
| { | ||
| set_param( delta, K ); | ||
| c = 18.0 * K / ( pi * delta * delta * delta * delta ); |
There was a problem hiding this comment.
It's in the CabanaPD namespace already
| { | ||
| c = model.c; | ||
| K = model.K; | ||
| K = ( model1.K + model2.K ) / 2.0; |
There was a problem hiding this comment.
We should not leave values unset. If we do what you suggest we should not keep K at all
streeve
force-pushed
the
multi_material
branch
from
02bf7fb to
f0d3789
Compare
Currently binary only (written to be easy to expand however)
ForceModelsaccepts a variadic list or tuple of models and calls each model peri,jpair as needed in the force loopTemporary changes to KW to test the capability - missing new bimaterial example
Built off of #172 for better composability
Replaces #119
Closes #109