Which solver to use for a dynamic incompressible multiphase case _ Dead water model
 

Hi there !

As a french student, I am attempting to modelize the so-called Dead-Water Phenomenon (for more details : see https://hal.archives-ouvertes.fr/fil...ater_final.pdf).
http://imageshack.com/a/img923/9584/RmoGxm.jpg
Picture from Matthieu Mercier's experiment


Being novice with Openfoam, I've been spending some days running basic tutorials and here is my problem :

The case I'd like to set up consists of a 2D water tank, with 2 layers of water (salt one and fresh one). I managed to modelize it using interFoam.
http://imageshack.com/a/img924/2573/Uz2FH2.jpg

A hull of a boat is on the top layer (fresh water). This hull has to travel from one side to the other, being towed by a constant force. It sounds a bit like the movingCone tutorial using pimpleDyMFoam. (And with some specific conditions, it should be possible to observe my phenomenon, which consist in an interfacial wave creating a pressure drop under the hull, nearly stopping the boat.)

So which solver allows to do both at the same time ? I used to have a look at interDyMFoam solver, but I didn't find any relevant case .. :(

So what do you think about that ? Did you ever seen a case close to that one ?

Thanks for your wise advice :)

Marin D.

Any idea ? :/

Hi,

Several questions:

1. Can your water layers mix? interFoam is for two immiscible fluids.

2. What is a shape of your boat? Question here is whether you can use layer addition and removal during your boat motion or it will be only mesh deformation.

3. If you need to impose drag force on your boat, maybe it would be easier to go with available IBM implementations?

Hi,

1. The two fluides are considered as immiscible.

2. not sure to understand the layer addition; the shape of my boat can be approximated as a rectangle (it's a 2D case). I'd like to set up a moving mesh, to allow having a tight mesh following the boat in order to keep the same accuracy for the wave following the boat.

3.I don't know how the IBM works on Openfoam, i'll have a look at it, thanks !

But I wondered (and it may mean I'm quite far from really understanding how solvers work) : this Dead-water phenomenon has lead scientists to establish a new equation discribing the phenomenon :
https://imageshack.com/i/poIAIsksj
It ain't exaclty the classic equation, so will I be able to simulate the phenomenon or should I try to modify a solver in order to solve it ?

Thanks for your answer :)

Hi,

1. OK. interDyMFoam is suitable for the setup.

2. Layer addition and removal is linked with mesh motion. Imagine you move your rectangle. The mesh is deformed. And as you continue rectangle motion, in front it mesh will be denser and denser, behind it mesh will become sparser and sparser. To more-or-less keep mesh density constant layer addition and removal could be added; if cell is inflated beyond certain limit, it is split into two cell with a layer of internal faces, if cell is squeezed to much, it is merged to neighbour.

Yet, if you are OK with just mesh deformation, you can try to set up case similar to tutorials/incompressible/pimpleDyMFoam/movingCone.

3. Here you can take a look. Though, it seems they implemented it for single-phase flows.

It would be nice to know what all those greek letters mean. Especially what is epsilon and why nu is constant.

Hi, thanks for your reply !

Ok great I see what you mean. But for now I'll try to set up the case with mesh deformation, and I will have a look at layer addition and removal as a possible improvement once I control the solver :)

Thanks for the link, interesting method. Though, I cannot find for now (still searching) proper source code fitting openfoam5.0 implemented for multiphase cases.. Moreover it seems that the method is not really appropriate for great Reynolds number (in my case : about 1e4). Looking more deeply for solutions.


Sorry, my bad :/.
-> Eta(x,t) is the spread pf the deformation of the surface
-> Xi(x,t) is the spread og the deformation of the interface (fresh/salt water)
-> Alpha is the (max deformation of the surface)/(max deformation of the interface)
-> Epsilon is the (max deformation of the interface)/(thickness of the top layer
-> Lambda is the characteristic length of the wave
-> Nu is the (thickness of the top layer)²/(Lambda)²
(alright, the notation is maybe not the most appropriate x) )

In short, I guess those parameters ain't that usefull for the comprehension of the modelization of the problem.

I try to run the case while studying in parallel the physics/applied maths of the model, so that I'm not able yet to clearly make simplifying hypotheses :/

So, unless I'm mistaken, InterDyMFoam should match with my problem ? Even with the last "Forcing" term of the equation which is quite new ? :/

(Moreover, apologies for my English, I hope you can understand me)
Best regards