[DEMApplication] New class for global damping #13019
Conversation
| ${CMAKE_CURRENT_SOURCE_DIR}/custom_constitutive/DEM_global_damping_model.cpp | ||
| ${CMAKE_CURRENT_SOURCE_DIR}/custom_constitutive/DEM_global_damping_model_nonviscous.cpp | ||
| ${CMAKE_CURRENT_SOURCE_DIR}/custom_constitutive/DEM_global_damping_model_viscous.cpp | ||
| ${CMAKE_CURRENT_SOURCE_DIR}/custom_constitutive/DEM_global_damping_model_viscousforcedependent.cpp |
There was a problem hiding this comment.
I wonder how we should name those damping models. For the one you used, it is the same with eq. 18 in Zhao's work [2014]. However, he said, "To ensure quasi-static condition, a simple local non-viscous damping force [40] is added to dissipate kinetic energy".
- I understand that we are using "global damping" to identify it with the "contact viscous damping" we used for CLs.
- As Zhao said that's a non-viscous damping force, please check whether you can call it "viscous force dependent". I am confused about your description.
There was a problem hiding this comment.
True. I changed the model names to what I think is more representative and, maybe, correct.
| const double E = p_element->GetYoung(); | ||
|
|
||
| if (vel_magnitude != 0.0) { | ||
| const array_1d<double, 3> damping_force = -2.0 * mGlobalDamping * sqrt(m*r*E) * vel; |
There was a problem hiding this comment.
Thank you for putting this model (if it can be called a model) here. Some time ago, I was using this equation to apply some additional global damping forces to the particle together with the "standard" global damping force to force the system to calm down. It is function is like we are putting the particle in a viscous fluid. To be honest, I don't like this equation myself, as it is less physical. We can keep it here until someday we don't need it anymore.
There was a problem hiding this comment.
Additionally, I will clean up the old implementation after you merge this.
|
|
||
| self.global_damping_option = self.global_damping != 0.0 | ||
|
|
||
| existing_damping_models = ["NonViscous","Viscous","ViscousForceDependent"] |
There was a problem hiding this comment.
The name problem again
|
|
||
| // Public methods | ||
| virtual void SetGlobalDampingModelInProperties(Properties::Pointer pProp, bool verbose = true); | ||
| virtual void AddGlobalDampingForceAndMoment(SphericParticle* p_element, array_1d<double,3>& total_forces, array_1d<double,3>& total_moment) {} |
There was a problem hiding this comment.
Maybe you can add the following here:
KRATOS_ERROR << "This function (....) shouldn't be accessed, use derived class instead"<<std::endl;"
| GeometryFunctions::normalize(velocity); | ||
| const double force = DEM_MODULUS_3(total_forces); | ||
|
|
||
| if (central_node.IsNot(DEMFlags::FIXED_VEL_X)) { |
There was a problem hiding this comment.
The difference between this model and the "NonViscous" one is that this model applies a damping force proportional to the velocity component in each component of the total force. Thus, the direction of the total force is modified. The "NonViscous" one keeps the direction of the total force unchanged and reduces the components by the same proportion. I guess understanding this difference may help you get a better name.
|
|
||
| namespace Kratos | ||
| { | ||
| DEMGlobalDampingModel::Pointer DEMGlobalDampingNonViscousCteForceDir::Clone() const { |
There was a problem hiding this comment.
NICE! Those new names are much clearer now. Only one point, for "DEMGlobalDampingNonViscousCteForceDir", I am confused about "Cte". For "Constant", a commonly used short name is "Const". Or maybe "Cst" if you prefer to use three letters, although it's less common. But "Cte" may be a bit hard to understand.
There was a problem hiding this comment.
True, 'cte' is appropriate in Spanish
|
Let's see whether another team member has some more comments @KratosMultiphysics/dem |
Implementation of a new class for global damping (also with a few formatting organization).
Observations:
In this subclass, I use the GLOBAL_DAMPING parameter, instead of GLOBAL_VISCOUS_DAMPING, for the computations. Therefore, @ChengshunShang1996 please check it, and if it is okay, you could remove GLOBAL_VISCOUS_DAMPING and your implementation from here. Or let me know if you'd like me to do it.