Thank you so much, this is really helpful!
On Sunday, September 4, 2022 at 4:15:37 AM UTC-7 Ruochun Zhang wrote:
> Hi Mohammad,
>
> 1. The prerequisite is to know that C::GPU works in center-of-mass frames,
> like Chrono in general. The accompanying vector and matrix that you specify
> during *AddMesh *are intended to be used as some sort of pre-processing
> for your mesh file. To cut to the chase, if your mesh file is created such
> that the mesh is already in its center-of-mass frame, then for all
> purposes, you should always call *gpu_sys.AddMesh("../Screw_CM.obj",
> ChVector<float>(0,0,0), ChMatrix33<float>(1),
> screw_mass);*
>
> If your mesh is not already in its center-of-mass frame, or you want to
> scale it before loading it, then you can supply those arguments.
> *ChMatrix33<float>(1)* would create an identity matrix, and this matrix
> will be applied to every point in your mesh (as a side note, the matrix is
> applied before the vector is also applied to every point in your mesh), as
> a part of the pre-processing. Obviously, the identity matrix does nothing
> to your mesh. If it was a diagonal matrix, it would "scale" your mesh; if
> it was a unitary matrix, it would rotate your mesh; if it was a general
> matrix the it would "rotate and scale" your mesh, hence its name.
>
> Because of what I said, I do not recommend using this *ChVector *argument
> to set the initial CoM (I just want to be more specific; you can call it
> initial position, too) of the mesh in simulation, it is against its design
> purpose. I said not recommend, but it can functionally work as a initial
> position specifier: think about it and you'll figure out why. Also, while
> other things may work fine, you will get strange torque readings from the
> mesh if you do that: again you can figure out why.
>
> In principal, use *ApplyMeshMotion *to control mesh position/velocity in
> simulation, including the initial position.
>
> 2. As you can see, this is managing a ChBody and has nothing to do with
> the mesh in your gpu_sys. *SetPos *just sets the position of that ChBody.
> Because it has nothing to do with *gpu_sys*, if you need it to be at a
> location, you need to set it separately. In this co-simulation example, the
> mesh you loaded is the proxy in DEM simulation for this ChBody. The *gpu_sys
> *will collect the contact force on the mesh, feed this info to the
> ChBody, and the ChBody will use this force info to update this body's
> location, and then the new location is used to update the mesh location in
> *gpu_sys*. ChBody can manage more complex rigid body motions such as
> those involving joints, so it is used to help manage external objects in
> the DEM simulation.
>
> 3. If you now understand the above, then you can see you are really just
> setting the mesh position (in *gpu_sys*) for the very first time here.
> *ApplyMeshMotion
> *is called at every time step, so every quantity here is not just some
> initial value, but the values throughout the simulation. Some of the values
> are changing every step (because *t *changes), such as *mesh_rot*; others
> such as velocities happen to be constants in this simulation. *mesh_rot* is
> a quaternion which represents the rotation status of the mesh (or
> specifically, the mesh's local coordinate system). If that sounds
> unfamiliar then it is probably helpful to read a bit more about quaternions.
>
> Hope it answers your questions!
> Ruochun
> On Saturday, September 3, 2022 at 5:23:44 PM UTC-5 [email protected]
> wrote:
>
>> Hi Ruochun,
>>
>> Thank you so much for your help. This actually helped a lot. I have some
>> questions that I hope you could answer. I am trying to control the mesh
>> motion using *ApplyMeshMotion *as you mentioned. However, I feel like I
>> do not fully understand everything related to that so I wanted to ask some
>> questions and hopefully, this will help others as well. I have looked at
>> the documentation and they are really brief so I hope you could answer some
>> of these questions here. I will use examples from my code(attached here). I
>> might be using some stuff incorrectly so I think it would be a good way to
>> see if what I am doing is correct or not.
>>
>> 1- gpu_sys.AddMesh("../Screw_CM.obj", *ChVector<float>(0,0,-90)*,
>> *ChMatrix33<float>(1)*, screw_mass);
>>
>> - *ChVector<float>(0,0,-90): *does this represent the
>> initial position where the mesh is initiated, and does it affect other
>> parts of the simulation?
>> - *ChMatrix33<float>(1): *what does this represent? I
>> looked through the documentation and it says that this is the rotscale.
>> Could you give a little more explanation as to
>> what that is?
>>
>> 2- ChVector<> ball_initial_pos(0, 0, 0);
>> std::shared_ptr<ChBody> screw_body(sys_screw.NewBody());
>> screw_body->SetMass(screw_mass);
>> screw_body->SetInertiaXX(ChVector<>(screw_inertia_x,
>> screw_inertia_y, screw_inertia_z));
>> screw_body->*SetPos(ball_initial_pos)*;
>> screw_body->SetBodyFixed(false);
>> sys_screw.AddBody(screw_body);
>>
>> - When creating a system using the imported mesh, what does
>> screw_body->*SetPos(ball_initial_pos) *do? Clearly, it is a way to set
>> the position of the mesh. However, since the
>> *ChVector<float>(0,0,-90)
>> *was used before when adding the mesh, is there a need to use *SetPos?*
>>
>> *3- * ChVector<float> *mesh_pos(0, 0, 0); *
>> float rev_per_sec = 10.f;
>> float ang_vel_Z = rev_per_sec * 2 * (float)CH_C_PI;
>> ChVector<> *mesh_lin_vel*(0);
>> ChQuaternion<> *mesh_rot *= Q_from_AngZ(t * ang_vel_Z);
>> ChVector<> *mesh_ang_vel*(0, 0, ang_vel_Z);
>> gpu_sys.ApplyMeshMotion(0, mesh_pos, mesh_rot, mesh_lin_vel,
>> mesh_ang_vel);
>>
>> - *mesh_pos(0, 0, 0):* to me, it seems that we are
>> setting the position of the mesh a third time. could you explain how this
>> affects the simulation?
>> - *mesh_lin_vel*(0): does this represent the initial
>> velocity of the mesh? or is it applying a constant velocity throughout the
>> entire simulation?
>> - *mesh_rot : *Could you explain what this exactly
>> is and how changing it affects the simulation?
>> * - **mesh_ang_vel*(0, 0, ang_vel_Z): is this the
>> initial rotational speed or it is a constant rotational speed that the mesh
>> will have throughout the entire
>> simulation?
>>
>>
>> Thank you so much in advance for taking the time to help us with
>> everything,
>>
>>
>>
>>
>> On Saturday, September 3, 2022 at 12:09:24 PM UTC-7 Ruochun Zhang wrote:
>>
>>> Hi Mohammad,
>>>
>>> I think that *demo_GPU_mixer.cpp* demo does exactly what you needed.
>>> While you are looking into this demo, I just want to mention one thing to
>>> avoid confusion: in terms of controlling the mesh motion in C::GPU, the
>>> *Transform
>>> *call before *AddMesh *should only be used to "align" the mesh with its
>>> CoM before the simulation starts, so that when the mesh is loaded it is
>>> expressed in its centroid and principal frame; it should not be used to set
>>> the initial location of the mesh. Instead, like in the demo, use
>>> *ApplyMeshMotion
>>> *to control the mesh motion throughout the simulation.
>>>
>>> Hope it helps!
>>> Ruochun
>>>
>>> On Saturday, September 3, 2022 at 10:58:19 AM UTC-5 [email protected]
>>> wrote:
>>>
>>>> Hello there,
>>>>
>>>> I am trying to perform a simulation using the gpu module where I want
>>>> to import a cylindrical screw (picture attached) into a bed on granular
>>>> media and then have the screw spin on top of the bed and then obtain the
>>>> forces/torques on the screw.
>>>>
>>>> I am not sure about what would be the best way to import the screw in a
>>>> way that it's set on the bed at the start of the simulation. Also, I am
>>>> not
>>>> sure how I can impose a type of motion that would force the screw to spin
>>>> on top of that bed since I am importing it as an .obj file. Finally, due
>>>> to
>>>> the shape of the screw, I assume that when the screw moves it will tend to
>>>> turn to the left/right. Is there a way to restrict the motion of the screw
>>>> to be just straight while spinning? I was wondering if there are any demos
>>>> regarding this or if someone has done something similar before. Any help
>>>> will be appreciated.
>>>>
>>>> Thank you so much in advance,
>>>>
>>>>
>>>>
--
You received this message because you are subscribed to the Google Groups
"ProjectChrono" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/projectchrono/9bc043d3-15cc-4c98-8044-9a3e7a95bf70n%40googlegroups.com.
