[eduard.puig.montella]

Hi Legacy,

Have you started your case from scratch? It is difficult to answer your question without more details on your case. I suggest you start with the tutorial available in sedFoam called 2DPipelineScour. It has a similar geometry, and you should be able to adjust it to match the boundary conditions of your case.

I have adjusted the 2DPipelineScour tutorial (tar.gz file attached) to reproduce an oscillatory flow similar to the one in the article you are referring to. The case runs smoothly, and you can observe the flow reversing in the following picture:



Maybe setting zeroGradient velocity at the outlet will help get rid of the instabilities at the outlet. You should also adjust the geometry to fit your case and make sure the boundary conditions are the same as in the article. For instance, I haven't modified the omega and k inlet conditions to be sinusoidal as in the article.

I hope this setup helps you model the scour as in the article.

2DPipelineScourOscil.tar.gz

[Legacy]

Dear Eduard Puig Montella,

Thanks for quick response. I tried your case and adjusted my the simulation setup similar to the article. I tried different ramp up times but after 9-10 seconds simulation starts to diverge. Here is test case for the article. Could you please have a look at it?

Thanks.

[eduard.puig.montella]

Hi,
I tested your case with a lower Courant number (set to 0.1) and the simulations ran without problems. See the following video:

What version of SedFoam are you using? I am using the latest version, SedFoam2312. If the problem persists, can you tell us where the instabilities are appearing? Is it at the boundary conditions, near the cylinder, at the interface, etc.?

[Legacy]

Thank you for spending time on this. The instabilities appear near the cylinder. My run crashes around t=9.5 seconds. Suspended sediments gets "stuck" to cylinder and it causes sharp gradients in the U.a field.

I am using version 2212. I think it has something to do with my installation. I will give it a try with the docker image of sedFoam.

Thanks again!

[Legacy]

Hi,

I compiled the sedFoam (latest version) with OpenFOAM v2312. I still have the same problem. Attached, please see the instabilities around the cylinder. The suspended sediment "attaches" to the cylinder and leads to instabilities.

You can see on the top left of the cylinder where U.a (vertical component) = -2 m/s.

[eduard.puig.montella]

I am really surprised you are encountering such instabilities because I do not observe them, at least not for the first 15 seconds. Please run the following case without any modifications.

testCase_eduard.tar.gz

[daiml95]

Hi,
I'm learning the tutorial cases and I meet a problem. In the 1D precursor case of 3DScour, how can I control the fully developed velocity using gradPMEAN? What is the corresponding relationship between them? If I want a flow velocity of 0.326 m/s, how should I determine the value of gradPMEAN?
I'm looking forward to your reply.

[eduard.puig.montella]

hey daiml95,
the easiest and most practical way is to use meanVelocityForce in fvOptions.

Just add this into your constant/fvOptions:

meanVelocityForce
{
    type            meanVelocityForce;
    active          true;
    selectionMode   all;
    fields          (U.b);
    Ubar            (0.326 0 0);  // Target mean velocity in x-direction

alternatively, if you really need to use gradPMEAN , then, unless you have a Poiseuille flow i think the better approach is to play with different gradPMEAN values in a 1D case until you get 0.326 m/s

[cgotelli]

Hello,

I would like to use SedFOAM to model sediment transport in a curved open channel and include turbines represented using the turbinesFoam package.

Since turbinesFoam applies actuator-line forces through fvOptions, I would like to confirm whether SedFOAM is fully compatible with fvOptions.
• Does SedFOAM support fvOptions in its current implementation?
• If yes, are there any recommendations or caveats when applying actuator-line forces, especially regarding which velocity field or phase should be targeted?
• If not fully compatible, are there known workarounds or patches to enable fvOptions in SedFOAM?

Any guidance or references would be greatly appreciated.

Thank you!

[eduard.puig.montella]

Hi Clemente,
Short answer: yes, sedFoam is fully compatible with OpenFOAM’s fvOptions in its current implementation. That means you can introduce source terms like actuator-line forces. In fact, sedFoam applies fvOptions.correct() to the momentum equations of both phases.
I have never worked with turbinesFoam or actuator-line forces, but sometimes I use fvOptions to control the mean water velocity, for example in constant/fvOptions with a block such as:

meanVelocity1
{
    type            meanVelocityForce;
    selectionMode   all;
    fields          (U.b);
    Ubar            (1 0 0);
}

I guess you should apply the actuator-line force to the fluid phase velocity field, not the sediment phase. Then the turbine’s blades will exert forces on the water, and sedFoam will transfer momentum to the sediment through the drag coupling. In sedFoam, Ub denotes the fluid (water) phase velocity and Ua denotes the sediment phase velocity.
In general, I don’t think a workaround is required if you’re using a recent sedFoam version that already supports fvOptions. Just make sure to compile sedFoam and turbinesFoam with the same OpenFOAM version, load the turbinesFoam library so that the custom actuator-line fvOption is recognized, and you should be good to go.
I hope this helps, and good luck with your simulation!
Eduard

[cgotelli]

Hi Eduard,

Thanks a lot for the clarification regarding fvOptions in sedFoam; that was very helpful

I would also like to ask for your personal opinion about the new solver, sedExnerFoam. In the recent preprint the authors mention:

“Additionally, sedExnerFoam could be employed alongside two-phase flow models, such as sedFoam (Chauchat et al., 2017), to derive more accurate and robust closures for sediment transport fluxes through an upscaling process.”

My main questions are:

• Is sedExnerFoam worth adopting now? In your view, is it sufficiently mature/validated beyond the preprint? Are there benchmark cases or comparisons (lab data or two-phase references) you’d point to? Any caveats you’ve seen in practice?
  
• If you do recommend sedExnerFoam, what new insights would justify using sedFoam in a complementary way? Concretely, in your opinion, in which situations does a two-phase run add valuable physics that can meaningfully inform/upscale closures (bedload laws, erosion–deposition, etc.) for sedExnerFoam?
  

In short, is it worthwhile to combine both solvers, or is it better to focus only on a single model?

Thank you again for your time!

Best,
Clemente

[zjy]

Originally Posted by eduard.puig.montella

I am really surprised you are encountering such instabilities because I do not observe them, at least not for the first 15 seconds. Please run the following case without any modifications.

Attachment 100881

I followed your suggestion and ran the testCase_eduard.tar.gz example without making any modifications, but the issue still persists. The error occurs around 10 seconds into the simulation, and I observe instability, possibly due to some abnormal computation.

I am using OpenFOAM version 2412 and the latest version of sedfoam. I have attached the relevant log file for your reference.

Please review the log file and provide any further suggestions or possible causes to help resolve this issue. I appreciate your help!

Best Regards!38498af8-f7af-4d6f-82f2-aa8512b7f28c.jpg

[eduard.puig.montella]

Thanks for reporting this issue. The instability comes from the shear-induced pressure term. We will check it in detail. If your case can neglect this effect, you can disable it in constant/granularRheologyProperties by setting PPressureModel none;. This should avoid the divergence, though results will change, especially in strongly sheared, collisional regimes. We will update the case as soon as possible.