Problem about periodic boundary

[liangliang]

I want to make some cases with delft3D ,and compare it with ROMS . There are two problems
a、 I can't find periodic boundary in delft3D ,how could i implement it ?
b、 Where I can find periodic boundary source code in ROMS ?

Thank you !

[arango]

See ROMS/Nonlinear/exchange_2d.F or ROMS/Nonlinear/exchange_3d.F.

[jcwarner]

"I can't find periodic boundary in delft3D ,how could i implement it ?"
Wow. That is non-trivial. I believe delft is solved with an implicit scheme. I would suggest that you do not try this, but of course you can do what you want. If you are really interested to do that, you should talk to the Delft team first (Bert Jagers would be a good contact). Or perhaps you maybe you could try a differnet set of test cases.
-j

[kate]

I once thought about creating periodic grids from a tool that is inherently non-periodic. I believe you can do it using an iterative technique, at least for those tools using the Ives and Zacharias algorithm. I'm assuming you want to make a periodic channel (east-west) with bumps along one wall. If you start with straight walls at the periodic ends, the grid resolution will not match at the two ends, either in i or in j.

1. make that grid anyway.
2. Extract the outline of the eastern half and of the western half.
3. Add the channel length to the western half x values.
4. Create the outline file of the new region and make a grid in it.
5. Extract the outline of the eastern half and of the western half.
6. Subtract the channel length from the eastern half x values.
7. Create the outline file of the new region and make a grid in it.
8. Goto step 2.

I did this once and it was converging to a more periodic solution. Then we realized it wasn't what we wanted because our coastline was north-south and you can't make the Coriolis parameter periodic.

[liangliang]

I once thought about creating periodic grids from a tool that is inherently non-periodic. I believe you can do it using an iterative technique, at least for those tools using the Ives and Zacharias algorithm. I'm assuming you want to make a periodic channel (east-west) with bumps along one wall. If you start with straight walls at the periodic ends, the grid resolution will not match at the two ends, either in i or in j.

1. make that grid anyway.
2. Extract the outline of the eastern half and of the western half.
3. Add the channel length to the western half x values.
4. Create the outline file of the new region and make a grid in it.
5. Extract the outline of the eastern half and of the western half.
6. Subtract the channel length from the eastern half x values.
7. Create the outline file of the new region and make a grid in it.
8. Goto step 2.

I did this once and it was converging to a more periodic solution. Then we realized it wasn't what we wanted because our coastline was north-south and you can't make the Coriolis parameter periodic.

Hi,kate
Thank you for your reply !
If I set a sink at south boundary and a source at north boundary . Connect them with each other . Can i think this is a periodic boundary ?

[kate]

You want a periodic channel in which you are specifying the transport? Sure, you can make a rectangular domain and put a line of sinks on one end and a line of sources on the other. See the thread about instability of tracers when mixing sources and sinks, though.

[liangliang]

You want a periodic channel in which you are specifying the transport? Sure, you can make a rectangular domain and put a line of sinks on one end and a line of sources on the other. See the thread about instability of tracers when mixing sources and sinks, though.

Thank you! My domain is rectangular . I will have a try !