Reproducing 10-sphere Drafting–Kissing–Tumbling in ConstraintIB/falling_sphere

33 views
Skip to first unread message

JATIN RAJ

unread,
Jun 25, 2025, 12:11:32 PMJun 25
to IBAMR Users

Hi everyone,

I’m using the ConstraintIB/falling_sphere example to simulate many spheres settling in a uniform grid (128×512×128, no AMR). When I increase the number of spheres, they seem to  overlap after coming close enough. I then implemented the inter-particle repulsive force exactly as in the Drafting–Kissing–Tumbling example shared here:
https://groups.google.com/g/ibamr-users/c/6Zs0y7zS6nc/m/P5OPgSacAgAJ

That fixed the overlap, but when I try to reproduce the 10-sphere results (Fig. 8 (a) & (b)) from
Sharma & Patankar, J. Comput. Phys. 205 (2005) 439–457 (https://doi.org/10.1016/j.jcp.2004.11.012),
my trajectories still don’t match.

Questions:

1. Is this example well suited for such kinds of applications and
2 . could there be any flaw in the implementation
3. Are my chosen parameters (penalty stiffness εₚ, interaction range ξ, time step Δt) in a reasonable range

Any pointers or suggestions would be greatly appreciated. Thanks!

Please find my input files and corresponding results in the drive folder below for reference.




Thank you,
Jatin 




Sharma & Patankar's result :
Screenshot from 2025-06-25 00-15-42.png

my result:

result.gif
a more detailed comparison including the 2 sphere DKT can be found in the following docs file :



Boyce Griffith

unread,
Sep 6, 2025, 4:06:41 PM (2 days ago) Sep 6
to IBAMR Users

On Jun 24, 2025, at 12:58 PM, JATIN RAJ <jati...@tifrh.res.in> wrote:

Hi everyone,

I’m using the ConstraintIB/falling_sphere example to simulate many spheres settling in a uniform grid (128×512×128, no AMR). When I increase the number of spheres, they seem to  overlap after coming close enough. I then implemented the inter-particle repulsive force exactly as in the Drafting–Kissing–Tumbling example shared here:
https://groups.google.com/g/ibamr-users/c/6Zs0y7zS6nc/m/P5OPgSacAgAJ

ConstraintIB allows for some slip between the fluid and the structure, and so in some situations you will need a contact model to avoid overlap between independent structures that are talking to a common background fluid.

That fixed the overlap, but when I try to reproduce the 10-sphere results (Fig. 8 (a) & (b)) from
Sharma & Patankar, J. Comput. Phys. 205 (2005) 439–457 (https://doi.org/10.1016/j.jcp.2004.11.012),
my trajectories still don’t match.

I think it is hard to know whether you should expect to get exactly the same results for the multi-sphere case. The dynamics are pretty complicated, and I would expect that small perturbations could yield very different dynamics if any of the details are different.

I would focus on comparisons with one- and two-sphere cases with highly-resolved benchmarks. (I don’t have specific suggestions.) If you look at the Sharma paper, they do make some comments on limitations of spatial and temporal resolution. Whatever you are comparing to, you will want to make sure that you are exactly matching boundary conditions and dimensions of the computational domain — those will definitely have an impact.

I don’t have experience with the contact model that is in this example, but I would expect that you would want the penalty forces to be active only when points are somewhere between one and two grid cells apart.

Ultimately, what I think you want is to do a grid-convergence study to see what the resolved dynamics look like and how sensitive they are to the numerical parameters used by the contact model.



— Boyce

Questions:

1. Is this example well suited for such kinds of applications and
2 . could there be any flaw in the implementation
3. Are my chosen parameters (penalty stiffness εₚ, interaction range ξ, time step Δt) in a reasonable range

Any pointers or suggestions would be greatly appreciated. Thanks!

Please find my input files and corresponding results in the drive folder below for reference.




Thank you,
Jatin 




Sharma & Patankar's result :
<Screenshot from 2025-06-25 00-15-42.png>

my result:

<result.gif>
a more detailed comparison including the 2 sphere DKT can be found in the following docs file :




--
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 visit https://groups.google.com/d/msgid/ibamr-users/959bc8d7-9c5c-4c42-8415-32e389f5a358n%40googlegroups.com.
<result.gif><Screenshot from 2025-06-25 00-15-42.png>

Reply all
Reply to author
Forward
0 new messages