Some advice on using Delft3D to analyse a deep fjord  DFlow Flexible Mesh  Delft3D
intro story DFlow FM
DFlow Flexible MeshDFlow Flexible Mesh (DFlow FM) is the new software engine for hydrodynamical simulations on unstructured grids in 1D2D3D. Together with the familiar curvilinear meshes from Delft3D 4, the unstructured grid can consist of triangles, pentagons (etc.) and 1D channel networks, all in one single mesh. It combines proven technology from the hydrodynamic engines of Delft3D 4 and SOBEK 2 and adds flexible administration, resulting in:
An overview of the current developments can be found here. The DFlow FM  team would be delighted if you would participate in discussions on the generation of meshes, the specification of boundary conditions, the running of computations, and all kinds of other relevant topics. Feel free to share your smart questions and/or brilliant solutions!
=======================================================  Sub groups

Message Boards
Some advice on using Delft3D to analyse a deep fjord
AE
Alistair Everett, modified 7 Years ago.
Some advice on using Delft3D to analyse a deep fjord
Youngling Posts: 1 Join Date: 2/22/13 Recent Posts 00
Hi all,
I intend to model a deep coastal fjord (up to ~1000m deep), the fjord is approximately 100km long and varies from 510km wide. The waters in the fjord are strongly stratified with warm salty water (~34C, 34psu) at depth and cool fresher water towards the surface (~0C, 1020psu). Fresh water is released at depth in some areas of the fjord and therefore forms a vertically buoyant flow. I am particularly interested in modelling this flow with reasonable accuracy as it may play an important role in mixing and exchange within the fjord. It is also important that I can analyse flows into and out of the fjord at different depths and how they are affected by the volume of freshwater released, tides and other forcings.
I'm currently looking into a number of potential models which I could use for this and was just after some advice from others who have experience of using D3D on similar problems before I commit to one model. So, having looked through the D3DFlow manual and doing a bit of experimentation, I have a number of questions which perhaps someone might be able to help me with:
Firstly, is D3D a suitable model for this kind of problem? Has anyone had experience of modelling a similar fjord like this with D3D (or another model)? Were your results reasonable and did you validate with any field data? Do you have any advice/warnings or problems I might encounter using D3D (outside the normal of course!)?
I understand that when using a sigmacoordinate system in D3D the assumption is made that the depth is much smaller than the horizontal length scale, I think I would be right in saying that this approximation is not valid for the dimensions given above, and therefore I would have to use the nonhydrostatic model in a Zcoordinate system? Does anyone have experience of using the nonhydrostatic pressure model with rapid convection, is it reasonably reliable? Are there any other areas of D3DFlow which still rely on the shallow water approximation, even when using the Zcoordinate model, which might introduce error to this problem?
I have attempted (quickly) to introduce a horizontal fresh water input at depth within a fictitious domain filled with salty water, I was hoping this would have an initial horizontal velocity, but would then rise up to the surface some distance away, but this doesn't seem to behave as I would've expected, instead it generates surface waves directly over the point of input. I haven't been using D3D for long so I may simply have made a mistake, does anyone have experience of this? Is it possible to create horizontal inputs at depth along a boundary in D3D? For now I'd be happy to know it is possible, in which case I will spend more time investigating it!
Thanks in advance for any help anyone might be able to provide,
Alistair
I intend to model a deep coastal fjord (up to ~1000m deep), the fjord is approximately 100km long and varies from 510km wide. The waters in the fjord are strongly stratified with warm salty water (~34C, 34psu) at depth and cool fresher water towards the surface (~0C, 1020psu). Fresh water is released at depth in some areas of the fjord and therefore forms a vertically buoyant flow. I am particularly interested in modelling this flow with reasonable accuracy as it may play an important role in mixing and exchange within the fjord. It is also important that I can analyse flows into and out of the fjord at different depths and how they are affected by the volume of freshwater released, tides and other forcings.
I'm currently looking into a number of potential models which I could use for this and was just after some advice from others who have experience of using D3D on similar problems before I commit to one model. So, having looked through the D3DFlow manual and doing a bit of experimentation, I have a number of questions which perhaps someone might be able to help me with:
Firstly, is D3D a suitable model for this kind of problem? Has anyone had experience of modelling a similar fjord like this with D3D (or another model)? Were your results reasonable and did you validate with any field data? Do you have any advice/warnings or problems I might encounter using D3D (outside the normal of course!)?
I understand that when using a sigmacoordinate system in D3D the assumption is made that the depth is much smaller than the horizontal length scale, I think I would be right in saying that this approximation is not valid for the dimensions given above, and therefore I would have to use the nonhydrostatic model in a Zcoordinate system? Does anyone have experience of using the nonhydrostatic pressure model with rapid convection, is it reasonably reliable? Are there any other areas of D3DFlow which still rely on the shallow water approximation, even when using the Zcoordinate model, which might introduce error to this problem?
I have attempted (quickly) to introduce a horizontal fresh water input at depth within a fictitious domain filled with salty water, I was hoping this would have an initial horizontal velocity, but would then rise up to the surface some distance away, but this doesn't seem to behave as I would've expected, instead it generates surface waves directly over the point of input. I haven't been using D3D for long so I may simply have made a mistake, does anyone have experience of this? Is it possible to create horizontal inputs at depth along a boundary in D3D? For now I'd be happy to know it is possible, in which case I will spend more time investigating it!
Thanks in advance for any help anyone might be able to provide,
Alistair
Bert Jagers, modified 7 Years ago.
RE: Some advice on using Delft3D to analyse a deep fjord
Jedi Knight Posts: 201 Join Date: 12/22/10 Recent Posts 00
Hi Alistair,
Interesting case. I can't remember any other Delft3D fjord models, but there are some successful applications with narrow deep model areas.
Delft3DFLOW is based on the shallow water equations in both sigma and zlayer mode, so I wouldn't use that as main argument against the sigmaapproach. The main restriction that the "shallow water assumptions" cause is that any waves that you model should be significantly greater than the water depth. I believe that this isn't a problem in your case which might possibly even be simulated using a model without a free surface (depending on the details of the research questions you want to solve).
The main argument should be that the sigmalayer approach isn't recommended in combination with steep topographic gradients as it can lead to spurious mixing of the water layers at the location of the gradients. The horizontal diffusion term is split into a diffusion along sigmaplanes and an "anticreep" correction for the deviation from the horizontal. That correction doesn't work well in case of steep topographic gradients. So, I agree with you that you should use the zlayer approach. Since you're interested in the development of a buoyant, you need indeed a nonhydrostatic model. It should be noted that Delft3D used in both hydrostatic and nonhydrostatic mode, an anisotropic turbulence model reflecting shallow water: the horizontal turbulence can be included using eddy viscosity or HLES, whereas the vertical turbulence can be included via keps equations. Extending keps to 3D might be considered for this kind of application.
I don't know about the details of your model setup, but there have been successful simulations of buoyant plumes with the desired behavior that you describe. Why your model doesn't show this behavior I don't know. Maybe you can start with some simple square basins for some tests.
Hope this helps a bit,
Bert
Interesting case. I can't remember any other Delft3D fjord models, but there are some successful applications with narrow deep model areas.
Delft3DFLOW is based on the shallow water equations in both sigma and zlayer mode, so I wouldn't use that as main argument against the sigmaapproach. The main restriction that the "shallow water assumptions" cause is that any waves that you model should be significantly greater than the water depth. I believe that this isn't a problem in your case which might possibly even be simulated using a model without a free surface (depending on the details of the research questions you want to solve).
The main argument should be that the sigmalayer approach isn't recommended in combination with steep topographic gradients as it can lead to spurious mixing of the water layers at the location of the gradients. The horizontal diffusion term is split into a diffusion along sigmaplanes and an "anticreep" correction for the deviation from the horizontal. That correction doesn't work well in case of steep topographic gradients. So, I agree with you that you should use the zlayer approach. Since you're interested in the development of a buoyant, you need indeed a nonhydrostatic model. It should be noted that Delft3D used in both hydrostatic and nonhydrostatic mode, an anisotropic turbulence model reflecting shallow water: the horizontal turbulence can be included using eddy viscosity or HLES, whereas the vertical turbulence can be included via keps equations. Extending keps to 3D might be considered for this kind of application.
I don't know about the details of your model setup, but there have been successful simulations of buoyant plumes with the desired behavior that you describe. Why your model doesn't show this behavior I don't know. Maybe you can start with some simple square basins for some tests.
 Did you model the fresh water inlet as a discharge with (horizontal) momentum? Delft3D switches to an upwind approach at locations with momentum discharges; under weakly forced conditions the effect of changing the numerical scheme can be significant. You can try to specify Upwsrc = #NO# in the mdf file to overrule this behavior.
 Surface waves can also result if the time step is too big. Have you verified that the simulation mostly satisfies the time step criteria specified in Table 10.1 / Section 10.4.5 of the Delft3DFLOW manual?
 The default nonhydrostatic model performs a hydrostatic simulation step, followed by a nonhydrostatic correction step. This works well for weakly nonhydrostatic conditions, but might fail for a strongly buoyant yet. In that case, it might be better to use the full nonhydrostatic solver that is currently undocumented.
Hope this helps a bit,
Bert