Modelling Rivers

River

Rivers are natural water courses. They have vital contributions to the social and economical welfare of societies living along their course. They provide continuous, usually dependable, flow and supply of water. Accordingly, centralised civilisation began emerging independently along number of river valleys; e.g. the Nile, Tigris and Euphrates, the Indus and the Yellow Rivers. Since then, mankind has been in continuous struggle to understand and live in harmony with this gift/force of nature. Modelling rivers is the modern way to understand their behaviour and is an essential tool to tackle several river problems.

Delft3D is the state-of-the-art modelling framework for hydrodynamics, water quality, ecology, waves and morphology.
 
Here, in this space, we share knowledge & experiences, and discuss issues related to modelling of rivers.
 
 

** PLEASE TAG YOUR POST! **

 

 

 

Sub groups
D-Flow Flexible Mesh
DELWAQ
Cohesive sediments & muddy systems


 

Message Boards

qh-relation fails

TG
Tobias Gierra, modified 9 Years ago.

qh-relation fails

Youngling Posts: 5 Join Date: 3/23/11 Recent Posts
Hello.

We're trying to simulate a simple steady state discharge scenario in a rectangular channel. At the upstream boundary we defined a total discharge (constant or increasing) and at the downstream boundary is a qh-relation according to the mannings-formular. Every time we try to run the simulation with different cell sizes (1 to 0.05 m) or time steps (1 to 0.001 s) over a period of more than a day the solver reaches the steady state but after a while the qh-relation fails and we get waves and increasing water levels.
Does anyone know the reasons for this behavior?
We got the same results with the latest sources or a 5 years old binary distribution of Delft3D. If it would be helpful, I will attach the simulations files.

Sincerely yours,
Tobias.
Bert Jagers, modified 9 Years ago.

RE: qh-relation fails (Answer)

Jedi Knight Posts: 201 Join Date: 12/22/10 Recent Posts
Hi Tobias,

Generally if boundary conditions fail, it's because initial oscillations cannot escape. The alpha factor can help to remove such fluctuations at water level boundaries. However, I understand from your post that the simulation reaches a steady state, but afterwards starts to oscillate (again). I haven't seen this before, but you indicate that it's consistent across a wide range of time steps and grid resolutions ... therefore, it would indeed help if you can provide input files of one of these simulations.

Bert
TG
Tobias Gierra, modified 9 Years ago.

RE: qh-relation fails

Youngling Posts: 5 Join Date: 3/23/11 Recent Posts
Hello Bert,

attached you will find an example which reaches a staedy state first and fails afterwards. The time simulations failing varies depending on time step, cell size and boundary conditions. Hopefully, there is only an option missing in MDF.

Kind regards,
Tobias.
Mohamed Yossef, modified 9 Years ago.

RE: qh-relation fails (Answer)

Youngling Posts: 23 Join Date: 2/23/11 Recent Posts
Dear Tobias

I have checked your model and it seems that the time step you used is too large (you can check it from: quick plot --> set file dependent options --> Check Time Step). I managed to successfully run the simulation with a smaller time step (Dt = 0.0015) and a "relaxation factor for Q-H forcing" set to 0.5.

Mohamed Yossef
TG
Tobias Gierra, modified 9 Years ago.

RE: qh-relation fails

Youngling Posts: 5 Join Date: 3/23/11 Recent Posts
Hi Mohamed,

thanks for your testing. I'm currently testing a long time simulation with your configuration since we also tested the relaxation factor and very small time steps. In this cases we got the failure not until 30 hours of simulation time.

I chose the 'big' time step for better reproducing of the failure. So why does the qh-relation fail after a steady state was reached? In fact we want to run a long term sediment simulation and so we would prefer smaller time steps.

Tobias.
Mohamed Yossef, modified 9 Years ago.

RE: qh-relation fails (Answer)

Youngling Posts: 23 Join Date: 2/23/11 Recent Posts
Hi Tobias

You can run the morphological simulation with a relatively large morphological acceleration factor. The Morphological time step should satisfy Courant criteria based on celerity of bed perturbations, which often gives an opportunity for 'morfac' of several hundreds. If you use the acceleration factor of 100, you need to run an hour of flow for every 100 hours of morphology.

Mohamed