Motion Restriction in DEM

[Mohammad Wasfi]

Hi, 

This is DEME related question

Is it possible to restrict the motion in one direction and allow it into other directions? For example, is it possible to apply some force to an object (A) through an interaction with another object (B)  and only allow it (A) to move in the X direction when responding to that force but not the Y and Z directions? An example code will be much appreciated.

Thank you so much,

[Ruochun Zhang]

Hi Mohammad,

I assume you meant "restrict the motion in one direction and disallow it into other directions". You can do it. In fact, you can prescribe the motion in one direction while allowing free movements in other directions, too.

The methods you should use are the following:

SetFamilyPrescribedLinVel

SetFamilyPrescribedAngVel

AddFamilyPrescribedAcc

AddFamilyPrescribedAngAcc

A better example is DEMdemo_WheelDP.cpp. There, DEMSim.SetFamilyPrescribedAngVel(2, "0", to_string_with_precision(w_r), "0", false) means Family 2 will always have 0 angular velocity in X, w_r  angular velocity in Y, and 0 angular velocity in Z. Other simulation entities cannot change the angular velocities of the entities that are in Family 2. And, DEMSim.SetFamilyPrescribedLinVel(2, to_string_with_precision(v_ref * (1. - TR)), "0", "none", false) further specifies that Family 2 will always be moving linearly along X at v_ref * (1. - TR) velocity, not moving linearly along Y, and the linear motion is not prescribed along Z, meaning if the contact/environmental forces make the entities to move along Z, it will accept these changes, unlike in X or Y directions which are prescribed.

You should keep the last argument to be false in your case. If it is false, then angular/linear velocity directions/components that are not specified or specified with "none", will just go with the "simulation physics" (instead of any user prescription). If it is true, then this family's angular/linear velocity becomes prescribed unconditionally (the solver does not attempt to change them), and any unspecified components will stay at the current value (0 as default).

That is for the velocities. For the 2 acceleration-related methods, you can only "add" extra accelerations to a family; you cannot prescribe the acceleration that an entity experiences, because a contact is a contact, and you cannot wipe it out. (And if you do wish to manipulate contacts, custom contact models are your friends.) You can prescribe the consequence of these forces though, and that is the velocity.

The string arguments in these methods can be a number or some expressions that may involve "t", the simulation time. So you can write something like "5 / 10" or "sin(3.14 * 2 * t)" as the argument, too. More such "user-referrable variables" might be added in future versions. But if it is some super complex motion that cannot be described with a simple expression, then you are better off just fixing the family (SetFamilyFixed), then using a tracker class to manually set the positions and velocities of the entity in question, step by step.

Thank you,

Ruochun

[Mohammad Wasfi]

Hi Ruochun, 

Thank you so much for your reply. I have tried to implement these methods and I am experiencing something not expected. I apply the following methods to my cylindrical mesh:

    DEMSim.SetFamilyPrescribedLinVel(3,  "0","none", "none",  false);
    DEMSim.SetFamilyPrescribedAngVel(3,  "0", to_string_with_precision(w_r), "0",  false);

my purpose is to have my cylindrical wheel spin about its axis (y-axis)  while disallowing linear motion in the axis perpendicular to the cylinder axis (x-axis). For some reason, when I animate my simulation, my wheel is spinning about its axis (which is what I want) but it is allowed to move linearly in the X-direction while it is restricted to move in the y-direction. However, when I obtain my mesh position through the tracker, the simulation reports a change in the Y-axis but not in the x-axis. In other words, my simulation animation is not consistent with my position vector obtained from the simulation at each time step. For example, from the simulation, my tracker returns the following values for the position:

Frame: 91 of 206
mesh position: -0.1, -0.00142017, -0.111681

Frame: 92 of 206
 mesh position: -0.1, -0.00152285, -0.113109

However, in the simulation, the animation shows a movement in the x-direction. I attached some pictures to demonstrate this change. 

The weird thing is when I change the SetFamilyPrescribedLinVel to restrict the y-axis instead I still see the same result. I attached my simulation file for your reference. 

Thank you so much, 

[Ruochun Zhang]

Are you sure the mesh you loaded from the file is centered (i.e. its MOI is at (0,0,0) in the mesh file)? Can you show a rendering of the screw.obj file you used in some CAD tool for us to understand?

Thank you,

Ruochun

[Mohammad Wasfi]

Hi, 

This is a picture of my mass properties page in Solidworks. I also included a picture of the module.

Thank you, 

Mohammad

[Ruochun Zhang]

Hi Mohammad,

Then it's hard to say. I can only say the tracker reports what the solver sees; indeed, in the solver, X of the mesh is not modified, while Y and Z are. I don't know why it does not translate to the rendering. What I thought was since the X of the CoM truly did not change, then the X-translation you see has to be the contribution of the rotation, and that means the CoM is in fact somewhere away from the center axis of the mesh. But I cannot be sure with the given information.

Can you render a movie and share with us? If you are going to generate the movie: since you are already outputting vtk files, can you please render the mesh as, well, a mesh, instead of some points? You can load the vtk time series directly into Paraview. And if you can share the mesh file, I can probably run it myself to understand what happened.

Ruochun

--
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 projectchron...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/91f6ec2b-f5b8-4014-8176-0cbd87010984n%40googlegroups.com.

--

Ruochun Zhang
Email: ruoc...@gmail.com
Email: ruochu...@alumni.rice.edu
Tel: 832-353-5111

[Radu Serban]

Mohammad and Ruochun,

 

I’m late to this party but something is concerning here.  If you want to constrain the motion of a mechanism part, you should use precisely that: constraints!  In other words, you should model the mechanical multibody system you are co-simulating with the granular material so that it respects whatever kinematics you desire.  The granular code does *not* do multibody dynamics.  As such, imposing the state (velocity in this case) of any part of the multibody system is not proper force-displacement co-simulation (which is what DEME *should* be doing) but just a hack.  Multibody dynamics is not as simple as that. I will not go into details here, but what you are likely seeing here is due to constraint drift.

 

Since it is already possible to do a proper co-simulation of a Chrono multibody system with DEME granular material, why not just use a cylindrical joint (which I assume is the kinematics you are looking for) in your Chrono model? 

 

Sure, for a very simple case like what you have here, you may be able to get away with prescribing the full state of your “screw” (that means also prescribing position-level state which I gather you are not doing).  But then you do not solve a dynamics problem for your mechanism. That also means you wouldn’t be able to get out of the simulation quantities that (I would think) are important, such as reaction forces on that screw.

 

--Radu

To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/CAHvQpOtftnbiQx36NMvifUF0RA5uyFoWX%3DBYNWy9N6TH%2BCUPDQ%40mail.gmail.com.

[Mohammad Wasfi]

Hi Radu, 

Thank you so much for your reply. I believe that we have figured out the source of the issue we faced. It seemed that my mesh's COM point was located at a point (x,y,z) and not at (0,0,0) as I assumed when I wrote my code. I have used the method InformCentroidPrincipal to tell the solver where exactly my COM point was. After, it seemed that everything is working as expected. 

I would for sure start looking into co-simulation between Chrono and DEME as I start simulating larger mechanical objects. I appreciate your help!

Thank you so much, 

[Mohammad Wasfi]

Hi Ruochun, 

I have one more question regarding this topic. Is it possible to apply constant normal/shear force to a mesh? In particular, I am interested in applying a normal force, let's say a 100N, to the screw. Would that be possible in DEME?  Would that be equivalent to applying an acceleration in the normal direction with a value equal to (100/mass of mesh)?

Thank you so much in advance, 

[Ruochun Zhang]

Hi Mohammad,

Please read one of my previous replies in this thread. I talked about it, and you can add extra accelerations to simulation objects through prescription methods. That is equivalent to applying forces. Off-topic comment: Having graduated with a degree in physics, I always believe that the idea of forces is such a crude thing in describing interactions compared to fields, and that is why I don't connect the idea of force to these user methods in DEME :)

Thank you,

Ruochun