|
[Sponsors] |
October 12, 2022, 01:50 |
Sediment Transfer Model Instability
|
#1 |
New Member
Felix Daddo
Join Date: Aug 2022
Location: Sydney, Australia
Posts: 25
Rep Power: 4 |
Hi there,
I am trying to model the sediment transfer process from the paper "CFD Modelling of Arctic Coastal Erosion due to Breaking Waves", using the HDC process to reduce the domain size. My simulations are crashing due to very large velocities at the base of the sea bed topography and I am not sure why this is happening. I have tried many different combinations and have found no way to make this work. Any advice or solution would be greatly appreciated, I am very stressed that I can't get this working! Thanks in advance, Felix |
|
October 12, 2022, 10:04 |
|
#2 |
Senior Member
Arun Kamath
Join Date: Nov 2014
Location: Trondheim, Norway
Posts: 265
Rep Power: 13 |
Hi Felix,
From your previous posts, you have been doing good work and I think you have good handle on things. At what stage are you facing the crash? In the FNPF module or after coupling and during the CFD run? I see a few issues with your inputs, I am not sure if you have noticed them already, and can't be certain they are the reason for the crash you experience. 1. You use P 43 to restrict the HDC region to 60-61 m but then in the coupled CFD model, your domain ranges 60.5-81.35. I haven't attempted something like this before and I am not sure this would work reliably. 2. In your CFD input, you have the grod size B 1 set to 0.04, but B 10 and other objects to 0.02. Your grid then will have 0 cells in the y-direction. The output when you start reef3d::cfd should show zero cells. you don't experience that? (given that a cfd sim seems to start for you, I guess not) Or this is this an error that has crept in into the files you uploaded? Anyway, I would suggest having another look at B 1, B 10 and the objects for their y-widths. 3. You couple from t=0 to t=300. This would not be necessary. The waves approach your area of interest at around 35-40 s. You could start the coupling from t=40 and activate I 30 in ctrl.txt for cfd, so that the cfd wave tank is initialised throughout with the results from the fnpf simulation. 4. Lastly, not related to any of your choices (may be), but do you have coupling occurring? that is do you see the coupled time "progress" in the output? or do xx and yy in the line that displays [q1] xx [q2] yy remain the same throughout?
__________________
Arun X years with REEF3D |
|
October 16, 2022, 20:31 |
|
#3 | |
New Member
Felix Daddo
Join Date: Aug 2022
Location: Sydney, Australia
Posts: 25
Rep Power: 4 |
Quote:
Thanks for your reply. I have presented my preliminary results which I am happy with but am still working on incorporating the sediment transport module. I realised that the files I attached were older and not entirely correct so they're not really too useful to determine what is going wrong. I actually have the HDC procedure working very well. I think I am doing it a different way to how you must have done it in the past and that is why I only store the state files over a very small distance but it is working quite well for me. I have attached an image of the results for the Ting and Kirby experiment 1995. I am using a Dirichlet boundary at the inlet so my understanding is that only the first layer of cells require the information so I only store the state values from that area (ie. 60m - 61m as the inlet is at 60.5m). This reduces the memory requirements significantly for me. As for running the procedure over the whole time of the simulation, this just allows me to make a video of the waves propagating through the FNPF domain and CFD domain side by side to show that the coupling process is actually working. The instability occurs in the CFD program when I use turn on the sediment transport module. From looking at the results in Paraview, I get a very high velocity at the beginning of the slope. I don't get this occurring when the slope is a solid though, so this is how I know it is an issue with the sediment transport. I have attached an image of the problem where you can see the high velocity at the start of the wedge. I have attached the files for this simulation titled "HDC_Sediment_Files". I get a similar issue when I don't use the HDC procedure and only the CFD module with sediment transfer. Basically the layer above the topo object becomes NAN values after the first iteration and the simulation crashes, in the image, this is shown by the yellow layer. All other velocity is 0, except for a thin line at the free surface. I have attached an image of this and the files for this case names "CFD_Sediment_Files". I feel like I am very close to getting everything to work but there must be some small mistake I am making in the sediment transport module that is causing the simulation to crash that I need to figure out. Fingers crossed I can get it sorted soon! Thanks again for all the help. Felix |
||
October 17, 2022, 08:47 |
|
#4 |
Senior Member
Arun Kamath
Join Date: Nov 2014
Location: Trondheim, Norway
Posts: 265
Rep Power: 13 |
You seem to be doing things mostly correctly.
I would just advice using T 61 in control.txt as you had before rather than the G 61 you have now. Before I investigate deeper, are you sure about your S 20 input, the sediment diameter? You have 0.13 m as the sediment diameter, while I am used to this values being about 0.13 mm i.e. S 20 0.00013
__________________
Arun X years with REEF3D |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Thermal non-equilibrium porous media model with conjugate heat transfer | Hexahedron | FLUENT | 9 | February 22, 2023 02:55 |
Table bounds warnings at: END OF TIME STEP | CFXer | CFX | 4 | July 16, 2020 23:44 |
Basic Nozzle-Expander Design | karmavatar | CFX | 20 | March 20, 2016 08:44 |
Question about heat transfer coefficient setting for CFX | Anna Tian | CFX | 1 | June 16, 2013 06:28 |
Water subcooled boiling | Attesz | CFX | 7 | January 5, 2013 03:32 |