[philiplu]

Hello, Phicau,
thank you so much for the reply.
It can be verified: under normal condition, the OLA-BC works very well

I still wonder, under breaking condition, if an efficient way to estimate the reflection?
It's not specific to OLA, but in the future, if you or anyone can share me piece of hint, it would be great. If you don't get free time, it's Ok

thank you again for the last reply, wish you and all: nice weekend

[GrivalszkiP]

Originally Posted by Phicau

Hi Peter,
The breakwater tutorial includes all the files that you need to set and process the free surface elevation gauges. The results can be generated at runtime.

Best,
Pablo

Thank you for your reply!

I found the files, and it works, but I think something is not OK. I tried it on my simulation, where I have set the water level to 0.45 m (setFieldsDict). (It is on cell edge, so "under" the free surface the cell value is 1, "above" it is 0.) After the simulation, the generated free surface time series shows that the free surface at the beginning is at 0.4725m. I found out that the difference (0.0225m) is 1/4 part of my cell size (0.09m). After this, I trided out the breakwater tutorial, and in your simulation, the situation is the same - it has a cell size/4 error (You set 0.8 m water level, and the time series shows 0.805m, 5mm error, at 2 cm cell size, 1/4 as well)

How can I get rid of this error? I'm not experienced in programming, and python.

Thank you in advance!
Péter

[Phicau]

Hi Péter,
to solve this you can change the interpolationScheme to cell in the controlDict section where the gauges are defined.

Best,
Pablo

[philiplu]

Hello, Phicau,
In OLA, if there's some criteria:
for irregular waves, how long it'll take till the field gets fully active, i.e. the shortest time needed

thank you in advance
Philip

[Phicau]

Hi Philip,
energy travels at the group celerity, you can calculate it from there.

Best,
Pablo

[philiplu]

Hello, Phicau,
Thank you, so you mean, when arriving, the OLA-variables e.g "u", "p", "alpha" get immediately active as fully developed

thank you again for the reply
actually i still look for sth of reflection, as e.g. a 3Ps method isn't valid anymore under breaking due to nonlinearity. But as it's not OLA-related, so I keep quiet and instead, just wish you a nice day

[Phicau]

Hi All,
I have just pushed a new update in olaFlow's GitHub to make it compatible with OpenFOAM v9. All the info on how to update: https://olaflow.github.io/source-code/
Enjoy!

Hi Philip,
You may have more luck if you ask the general coastal engineering questions in CoastalList instead.

Best,
Pablo

[philiplu]

hello, Phicau,
thank you for the both

philip

[abas.rahmani86]

Dear colleagues,

Is it possible to model two miscible fluids in Oaflow which have different densities? I want to model the interaction of this with free surface variations.

[OceanMan]

Hi Pablo,
I have a problem when using olaFlow. I solve 2 cases (figure attach). Case 1 was solved, case 2 have an error when I used to fine mesh.
I don’t know how to solve this problem. I hope you can help me.
Thank you very much!

[Phicau]

Hi Abas,
Currently there is no public version of olaMixingFoam, although it is easy to create. What is not so easy is generalising the wave generation BCs to work with the two mixable fluids (water + X). However, if fluid X never reaches the wave generation boundary, the olaFlow BCs will work perfectly out of the box.

Hi OceanMan,
I cannot possibly guess what is wrong. Please see How to give enough info to get help

Best,
Pablo

[GrivalszkiP]

Dear Pablo,

I want to use olaFlow for my further research. For this, I want to make a grid sensitivity analysis with a benchmark case - it is an abutment in a flume, as an obstacle for solitary wave (Lara et al. 2012, I think you know this paper).

I set different constant grid sizes (10-7.5-5-2.5 cm). For 10, 7.5 and 5, everything went fine. At the finest grid (2.5 cm), the solution crushed. I tried different wall boundary conditions then different turbulence models (kOmegaSSTStable, then kOmegaSSTBuoyancy). I have a feeling that here is the error: the kOmegaSSTStable could not even find the wave height and time. The kOmegaSSTBuoyancy version have found the first wave. but then crushes (attached graph). I made some Paraview plots where I can find that the k values are totally unrealistic, and hard to explain (Attached figure, the wave is near the abutment, it is a contour plot for k=0.1). I also recognised, that in this finest mesh, velocities near the wall are unrealistic (figure).

I don't have a clue what is the problem. I use the suggested preset as in the tutorials (only modification is roughness height on walls at nut dictionary). All the rougher grids are fine, the error occurs only in the finest case. Every setup is the same, except blockMeshDict.

Do you have any advice to solve this problems?

Thank you in advance!

Peter

[yannaixiao]

Dear all,
I have established a three-dimensional model to simulate the impact of waves on structures. From the results, I found that the wave has obvious attenuation phenomenon and can not reach the preset wave height in the last few cycles.The following figure shows the wave height curve 50 meters away from the wave making boundary.
The wave making model adopts the regular wave stokes Ⅲ, with a wave height of 10 meters, a period of 12 seconds.
As for the boundary conditions, the left side of the model is the wave making boundary, the right sides are the wave absorption boundary, and the bottom, front, rear and structure are the wall boundary.
I don't know what caused the wave attenuation.

[abas.rahmani86]

Originally Posted by Phicau

Hi Abas,
Currently there is no public version of olaMixingFoam, although it is easy to create. What is not so easy is generalising the wave generation BCs to work with the two mixable fluids (water + X). However, if fluid X never reaches the wave generation boundary, the olaFlow BCs will work perfectly out of the box.

Hi OceanMan,
I cannot possibly guess what is wrong. Please see How to give enough info to get help

Best,
Pablo

Thank you a lot, Pablo.
I want to use olaFlow to model internal waves. Could you please guide me more about creating olaMixingFlow?

wish the best

[abas.rahmani86]

Originally Posted by yannaixiao

Dear all,
I have established a three-dimensional model to simulate the impact of waves on structures. From the results, I found that the wave has obvious attenuation phenomenon and can not reach the preset wave height in the last few cycles.The following figure shows the wave height curve 50 meters away from the wave making boundary.
The wave making model adopts the regular wave stokes Ⅲ, with a wave height of 10 meters, a period of 12 seconds.
As for the boundary conditions, the left side of the model is the wave making boundary, the right sides are the wave absorption boundary, and the bottom, front, rear and structure are the wall boundary.
I don't know what caused the wave attenuation.

Dear 闫乃笑,
I guess it's due to the over-production of turbulence beneath surface waves in RANS models. Please see the Larsen & Fuhrman (2018) and update your olaFlow from here:
https://olaflow.github.io/blog/turbu...ons-revisited/

Larsen, B.E. & Fuhrman, D.R. (2018)
On the over-production of turbulence beneath surface waves in RANS models.
Journal of Fluid Mechanics, 853, 419–460.
doi:10.1017/jfm.2018.577

[Phicau]

Hi Péter,
I would suggest that you play with the turbulence seeding values and follow Larsen & Fuhrman (2018) recommendations. Also, I am not sure that using roughness is helping, since the surfaces were quite smooth to start with.

Hi 闫乃笑,
this issue has been discussed previously in the olaFlow/olaFoam thread, please use the search tool in the forum to find a full answer. There are multiple causes that may be playing a role such as the numerical schemes, or as Abas mentions, over-production of turbulence.

Hi Abas,
In that case I think the best way is for you to use interMixingFoam and link the olaFlow boundary conditions dynamically https://openfoamwiki.net/index.php/C...ary_Conditions . If that does not work you may need to go for the complex route that I discussed before.

Best,
Pablo

[Stan Zhou]

Dear Pablo,
After dealing with the trifles at hand, I finally find time to continue my study using olaFlow.
Recently I’ve been reading the previous posts in IHFOAM, OlaFoam, and OlaFlow to get some hints for solving my problems. I’ve summarized some colleagues’ experience and I’d like to share my superficial thoughts with everyone facing the same challenges. If my understanding is wrong, please point it out

For the problem that I had:
1. The wave height decreases after some time (about several periods), even near the wave generation BC (The wave absorption at the wave generation BC has been activated.). Meanwhile, when the overtopped water reaches the wave absorption BC, some water flows back to the structure.

Opinions and solutions:
These problems may be related to several factors. Firstly, RANS models over-estimate the turbulence in wave propagation. Secondly, the previous AWA is suitable for shallow water waves and it may cause wave attenuation to some extent and lead to some backflow. Thirdly, the Courant number set in controlDict may not be small enough.
It could be solved by adopting the laminar model (This may not be a good choice since turbulence occurs in wave-structure interaction more or less), reducing Courant number, using a smaller nut, adopting ER-AWA developed by Pablo and so on. By the way, I noticed that Li Yuzhu adopted Reynolds stress turbulence models (RSMs) to simulate waves. It was shown that RSMs sufficiently reduced the turbulence over-production and the result seemed to be pretty good (Especially for breaking waves). Will this method be further implemented into olaFlow?

2. Same case settings, different results.
The simulation I’ve been doing is a 2D breakwater case. I want to obtain the characteristics of wave propagation on the reef, pressure on the revetment breakwater, and the overtopping volume over the caisson. The fluid domain is about 950m in x direction and 120m in z direction (3.5 milion cells). The preset wave height is 7m, period is 10s and the water depth is about 100m ( The wave length is 156m, Stokes II wave). When the wave propogates onto the reef, the water depth suddenly reduces to several meters. I adopted the widely used K-Epsilon model as turbulence model (Now I reckon that it’s not a good choice since it is suitable for conditions with low Reynolds number but in this case, there suppose to be severe wave overturn and wave breaking due to the extreme wave conditions and the change of terrain). In order to get enough resolution for capturing waves, the smallest cell size in z direction is 0.05m (much smaller than H/20). To get accurate pressure on the breakwater, the cell sizes near the breakwater in x and z direction are both 0.05m (much smaller than L/200). However, I’m not sure whether these cell sizes are good enough since once the wave propogates on the reef, the water depth drops suddenly. As a result, the wave height, period and wave length all changes, so I don’t know whether the wave characteristics could be captured well even though the cell sizes here are quite small. In addition, since the coefficients of porous medium also play a significant role in the simulation, we adopted the coefficients obtained from permeability tests. The blank part beneath the porous medium is impervious rock.

Since the wave breaks and overturns very badly, I thought I should run it again to double check the result before verify the grid sensitivity. However, even for the same case, the results are different when it is run for a second time. As Pablo suggested, I did a few small cases before I set out to this complex case. However, the small cases’ results seemed to be good but things changed when it came to my large case. Therefore, my tutor and I start to wonder if it was the inherent disadvantage of turbulence model that made the wave extremely nonlinear. Due to this severe nonlinearity, the characteristics of wave propagation on the reef and the wave breaking after hitting the caisson are quite different even with the same initial settings. Do you have any comments on this? By the way, would you mind having a look at my case settings (Due to some reasons, the case settings are not suitable for posting here, hope you could understand. If you don’t mind, could I send it to your e-mail?)? I’ll be very appreciated!

[Phicau]

Hi Stan,
Thanks for your detailed description, here some comments:

- The wave height decreases certainly because over-production of turbulence. If you want to use k-epsilon or RNG you should use the stabilised version ( https://github.com/BjarkeEltardLarsen/RANS_stableOF50 ) to solve your issues.
- I see no advantage of k-epsilon over the SST model, so that it the one I use and I recommend using. Either the buoyancy correction ( https://github.com/BrechtDevolder/bu...rbulenceModels ) or the stabilised version (link above) will most likely work. If you want an up-to-date version, I have just updated those 2 to work with the latest OF and OF+ versions ( https://github.com/phicau/olaFlow_su...enceMultiphase )
- If RSM is available in OpenFOAM it is already compatible with olaFlow, but I have no experience using those type of models.
- AWA is not meant to work inside porous media. I would split the right boundary in 2. Set a wall where porous media are and an atmospheric BC on top of it to have overtopping flow away.

Best,
Pablo

[erickmartinez]

Hi,

Does olaIsoFlow help refine the interface between phases automatically or must we include any refinement such as dynamicFvRefineMesh? Additionally, On the website the latest olaIsoFlow I found was for 17xx, is it compatible with 20xx versions? Thanks!

[GrivalszkiP]

Hi Pablo,

I have a case with free surface gauges (as can be seen in the breakwater tutorial). It turned out that I placed the gauges to wrong place. I corrected the position of the gauges in the controlDict file, and tried to run the following:

postProcess -func gaugesVOF

Nothing happened. However, run again the whole simulation(s) would be really time-consuming. It is a parallel simulation if it matters, and reconstructPar is not an option.

What is the correct line for re-calculate the free surface time series?

Thank you in advance:

Peter