The Lagrange point repulsion problem in ConstraintIB.

[Dewu Yang]

Dear all,

Has the overlapping phenomenon of Lagrange points of the structure been solved? Studying school swimming with multiple degrees of freedom results in overlap between individual fish. The NonbondedForceEvaluator that Prof. Griffith mentioned before is used for the mutual exclusion of Lagrange points and what which parameters need to be changed? I would appreciate it if you could give me some suggestions.

Best,

Dewu

[Boyce Griffith]

> On May 16, 2022, at 11:20 PM, Dewu Yang <yangd...@gmail.com> wrote:
>
> Dear all,
> Has the overlapping phenomenon of Lagrange points of the structure been solved? Studying school swimming with multiple degrees of freedom results in overlap between individual fish. The NonbondedForceEvaluator that Prof. Griffith mentioned before is used for the mutual exclusion of Lagrange points and what which parameters need to be changed? I would appreciate it if you could give me some suggestions.

I’m not sure I understand the question / problem.

If you have overlapping structures in an IB-type model, then the equations of motion are such that the two structures should move together unless you allow for slip between the structure and the background velocity (like for a porous material model). With ConstraintIBMethod, there is some "numerical slip", but it goes to zero under grid refinement.

> Best,
> Dewu

[Dewu Yang]

Thank you for your advice, Prof.Griffith. The problem I am encountering now is that when studying multiple fish swimming, there will be "bonding" or even overlapping between individuals which leads to the termination of the program. What can be done to prevent this abnormal overlap between the fish?  I think this is the difficulty of studying multi-degree of freedom cluster swimming with ConstrainIBMethod at present.

Thanks

[Amneet Bhalla]

This is not a problem with ConstraintIBMethod per se. It is a setup/modeling problem. If the kinematics are such that the two fish are bound to collide with each other, that will happen. It is like if two cars are crossing each other they will collide. The aerodynamics of the cars does not prevent car collisions (it would be cool if that were the case).

So how do we prevent car collision? By applying brakes. Here, it means apply a repulsion or a collision  model, which is no means a trivial thing because the geometry of two fish are time varying. 

I suggest re-consider what you want to study in your model. Maybe you don’t need collisions after all.

--
You received this message because you are subscribed to the Google Groups "IBAMR Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibamr-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ibamr-users/086785e0-26eb-48ec-ae77-25b53dee6213n%40googlegroups.com.

--

--Amneet 

[Boyce Griffith]

On May 19, 2022, at 11:30 AM, Amneet Bhalla <mail2...@gmail.com> wrote:

This is not a problem with ConstraintIBMethod per se. It is a setup/modeling problem. If the kinematics are such that the two fish are bound to collide with each other, that will happen. It is like if two cars are crossing each other they will collide. The aerodynamics of the cars does not prevent car collisions (it would be cool if that were the case).

So how do we prevent car collision? By applying brakes. Here, it means apply a repulsion or a collision  model, which is no means a trivial thing because the geometry of two fish are time varying. 

I suggest re-consider what you want to study in your model. Maybe you don’t need collisions after all.

FWIW, I think that all of these methods will “naturally handle contact” with enough spatial resolution. However, for a given model and set of operating conditions, it may or may not be practical to compute at that resolution.

An alternative — although it would require some work on your part to make it work — would be to use IBMethod, IBFEMethod, or CIBMethod, which all implicitly provide a numerical contact model that probably will keep the structures from overlapping.

It also possible to implement a contact model with IIMethod (see, e.g., https://arxiv.org/abs/2204.12907).

On Wed, May 18, 2022 at 8:09 PM Dewu Yang <yangd...@gmail.com> wrote:

Thank you for your advice, Prof.Griffith. The problem I am encountering now is that when studying multiple fish swimming, there will be "bonding" or even overlapping between individuals which leads to the termination of the program. What can be done to prevent this abnormal overlap between the fish?  I think this is the difficulty of studying multi-degree of freedom cluster swimming with ConstrainIBMethod at present.

Thanks

<2022-05-19_10-53.png>

On Thursday, May 19, 2022 at 1:44:38 AM UTC+8 boy...@gmail.com wrote:

> On May 16, 2022, at 11:20 PM, Dewu Yang <yangd...@gmail.com> wrote: 
> 
> Dear all, 
> Has the overlapping phenomenon of Lagrange points of the structure been solved? Studying school swimming with multiple degrees of freedom results in overlap between individual fish. The NonbondedForceEvaluator that Prof. Griffith mentioned before is used for the mutual exclusion of Lagrange points and what which parameters need to be changed? I would appreciate it if you could give me some suggestions. 

I’m not sure I understand the question / problem. 

If you have overlapping structures in an IB-type model, then the equations of motion are such that the two structures should move together unless you allow for slip between the structure and the background velocity (like for a porous material model). With ConstraintIBMethod, there is some "numerical slip", but it goes to zero under grid refinement. 

> Best, 
> Dewu 

-- 
You received this message because you are subscribed to the Google Groups "IBAMR Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibamr-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ibamr-users/086785e0-26eb-48ec-ae77-25b53dee6213n%40googlegroups.com.

-- 

--Amneet 

-- 
You received this message because you are subscribed to the Google Groups "IBAMR Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibamr-users...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ibamr-users/CAMETWJ2VYwA4%3D2EX7O0uR3sXirASv0SzerNafwChPO1K4c27uQ%40mail.gmail.com.

[Dewu Yang]

Dear prof.Griffith and prof.Amneet,

Thank you very much for your valuable suggestions, I will try according to what you said. Yes, I totally agree with what you say. I've done some simulations with ConstraintIB, and it works really well. It's a powerful tool.

Best,

Dewu