[GeoMechanicsApplication] Add coupling calculator components #14162
Conversation
…ing with clang-format
…upling matrix calculation in the equation of motion utils
…he coupling calculators)
…ore specific info
rfaasse
changed the title
| InputProvider(std::function<const Matrix&()> GetNpContainer, | ||
| Geo::BMatricesGetter GetBMatrices, | ||
| std::function<Vector()> GetVoigtVector, | ||
| Geo::IntegrationCoefficientsGetter GetIntegrationCoefficients, | ||
| std::function<std::vector<double>()> GetBiotCoefficients, | ||
| std::function<std::vector<double>()> GetDegreesOfSaturation, | ||
| std::function<Vector()> GetNodalVelocities, | ||
| std::function<double()> GetVelocityCoefficient) |
There was a problem hiding this comment.
I'd like to add the aliases for this provider later (but let me know if you think differently). I think we should also discuss if we'd like a new alias for every type of getter, or if we'd like more generic aliases (e.g. MatrixGetter instead of having different ones for BMatricesGetter and NpContainerGetter)
There was a problem hiding this comment.
I have been struggling with the same. For std::function<std::vector> we have Geo::IntegrationCoefficientsGetter. Using that for other coefficients ( here Biot, DegreeOfSaturation, Bishop ) looks unnatural. We could change its name to CoefficientsGetter and use it throughout the providers.
WPK4FEM
left a comment
WPK4FEM
left a comment
There was a problem hiding this comment.
Dear Richard,
Thank you for making these coupling calculators available. A good step towards having a complete set of calculators for elements.
Please have an in depth look at the remarks about the velocity coefficient. It's use does not align with my expectation, I think it does not belong in the calculator and causes an incorrect RHS the way it's used. Try to prove that I'm wrong off course.
Regards, Wijtze Pieter
| InputProvider(std::function<const Matrix&()> GetNpContainer, | ||
| Geo::BMatricesGetter GetBMatrices, | ||
| std::function<Vector()> GetVoigtVector, | ||
| Geo::IntegrationCoefficientsGetter GetIntegrationCoefficients, | ||
| std::function<std::vector<double>()> GetBiotCoefficients, | ||
| std::function<std::vector<double>()> GetDegreesOfSaturation, | ||
| std::function<Vector()> GetNodalVelocities, | ||
| std::function<double()> GetVelocityCoefficient) |
There was a problem hiding this comment.
I have been struggling with the same. For std::function<std::vector> we have Geo::IntegrationCoefficientsGetter. Using that for other coefficients ( here Biot, DegreeOfSaturation, Bishop ) looks unnatural. We could change its name to CoefficientsGetter and use it throughout the providers.
| std::function<std::vector<double>()> GetBiotCoefficients, | ||
| std::function<std::vector<double>()> GetDegreesOfSaturation, | ||
| std::function<Vector()> GetNodalVelocities, | ||
| std::function<double()> GetVelocityCoefficient) |
There was a problem hiding this comment.
The velocity coefficient is more part of the time integration scheme than of the fill in of the matrix. I wonder if we can keep it outside the calculator.
|
|
||
| typename BaseType::RHSVectorType RHSContribution() override | ||
| { | ||
| return prod(CalculateCouplingMatrix(), mInputProvider.GetNodalVelocities()); |
There was a problem hiding this comment.
I think that the velocity coefficient should not be part of the RHS computation. Here it is because it is included in CalculateCouplingMatrix.
| const auto biot_coefficients = mInputProvider.GetBiotCoefficients(); | ||
| const auto bishop_coefficients = mInputProvider.GetBishopCoefficients(); | ||
| const auto np_container = mInputProvider.GetNpContainer(); | ||
| for (std::size_t i = 0; i < mInputProvider.GetBMatrices().size(); ++i) { |
There was a problem hiding this comment.
We are looping over integration points here. There should be as many integration_coefficients as B matrices, Biot coefficients, Bishop coefficients and shape function values in the NpContainer. For me the loop control reads more naturally as:
| for (std::size_t i = 0; i < mInputProvider.GetBMatrices().size(); ++i) { | |
| for ( auto i = std::size_t{0}; i < mInputProvider.GetIntegrationCoefficients().size(); ++i) { |
| typename BaseType::LHSMatrixType coupling_contribution; | ||
| GeoTransportEquationUtilities::CalculateCouplingMatrix( |
There was a problem hiding this comment.
Is it impossible to use the version where the return value is the coupling matrix contribution?
📝 Description$\leftrightarrow$ displacement coupling. Due to overlap in the two calculations, the
This PR adds two calculators, the
UPCouplingCalculatorand thePUCouplingCalculatorto calculate left and right hand side contributions regarding waterPUCouplingCalculatorre-uses theUPCouplingCalculator.Note that no connection is made yet to an existing element (e.g. the interface element).
🆕 Changelog
UPCouplingCalculatorandPUCouplingCalculator