Inquiry on CFL Number Sensitivity and Robust Parameter Settings in IBLevelSet Simulations

[LIMING CHAO]

Dear all,

I am currently using IBLevelSet to evaluate the propulsion performance of oscillating foils with different amplitudes under incoming flow, all at the same oscillation frequency. The foil motion follows a fixed sinusoidal pattern.

During my tests, I observed an unexpected behavior: when using the same grid resolution and time step, foils with larger amplitudes sometimes result in a smaller CFL number, while those with smaller amplitudes yield a larger CFL number, which in some cases causes the simulation to diverge.

I suspect this issue may be related to the parameters LS_TAG_VALUE or LS_TAG_ABS_THRESH, as they may influence how the level set field evolves during motion. I would like to ask whether there are robust values or recommended strategies for these parameters that can ensure stable simulations across different frequencies and amplitudes.

Best,

Li-Ming

[Boyce Griffith]

Sorry for the delay on this.

Can you post a screenshot of the dynamics just before the simulation diverges? Do you have “strong” vortices hitting outflow boundaries, which can result in “back flow divergence”?

--
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/808d6d6a-a6e0-4714-a75f-0bd934f219a4n%40googlegroups.com.

[LIMING CHAO]

Dear Prof. Boyce,

I did not save the visualization files, so unfortunately I cannot provide you with screenshots. However, I agree with your point. Do you think it would be possible to add a forced boundary outlet in order to avoid backflow?

Best

Li-Ming  

To view this discussion visit https://groups.google.com/d/msgid/ibamr-users/1DB1BF95-2F23-43F3-A77E-0B67BF65D9EB%40gmail.com.

[Boyce Griffith]

On Sep 7, 2025, at 8:07 PM, LIMING CHAO <chaoli...@gmail.com> wrote:

Dear Prof. Boyce,

I did not save the visualization files, so unfortunately I cannot provide you with screenshots.

Very understandable.

However, I agree with your point. Do you think it would be possible to add a forced boundary outlet in order to avoid backflow?

Yes.

Briefly, if you run into any weird flow patterns near physical boundaries where you are setting normal traction boundary conditions, then it probably has something to do with “back flow divergence”. The basic problem seems to be that stagnation points along these types of boundaries are unstable — so, in fact, back flow is OK as long as there is back flow along the entire boundary.

If you have a flow that is coming in on the left and leaving on the right, you can help to stabilize the outflow boundary by adding a force that is of the form

f(x,t) = - eta(x) * min(u(x,t), 0)

You want the force to be operating only close to the outflow boundary — I would make eta a smooth field that is concentrated at the boundary and takes a maximum value of around 0.5 * rho / dt.

This will act as a damping term in locations where you have “spurious inflow” from the outflow boundary.

Best

Li-Ming  

On Sat, Sep 6, 2025 at 10:50 AM Boyce Griffith <boy...@gmail.com> wrote:

Sorry for the delay on this.

Can you post a screenshot of the dynamics just before the simulation diverges? Do you have “strong” vortices hitting outflow boundaries, which can result in “back flow divergence”?

On Jul 7, 2025, at 8:04 PM, LIMING CHAO <chaoli...@gmail.com> wrote:

Dear all,

I am currently using IBLevelSet to evaluate the propulsion performance of oscillating foils with different amplitudes under incoming flow, all at the same oscillation frequency. The foil motion follows a fixed sinusoidal pattern.

During my tests, I observed an unexpected behavior: when using the same grid resolution and time step, foils with larger amplitudes sometimes result in a smaller CFL number, while those with smaller amplitudes yield a larger CFL number, which in some cases causes the simulation to diverge.

I suspect this issue may be related to the parameters LS_TAG_VALUE or LS_TAG_ABS_THRESH, as they may influence how the level set field evolves during motion. I would like to ask whether there are robust values or recommended strategies for these parameters that can ensure stable simulations across different frequencies and amplitudes.

Best,

Li-Ming

--
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/808d6d6a-a6e0-4714-a75f-0bd934f219a4n%40googlegroups.com.

--
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/1DB1BF95-2F23-43F3-A77E-0B67BF65D9EB%40gmail.com.

--
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/CACn83E1TkcRPvn%3D30E%2B68DO_OJ-vbXixzMi6DT9hAz-Dz9rBRA%40mail.gmail.com.