[ThatHimalayanGuy]

Hi all,
I am trying to model a free surface flow problem for a small hydropower plant. The doamin of the problem looks like following when clipped in half (domain is symmetrical):
domain.jpg


The initial water surface levels are 2.5 m upstream side (F 72 -13.4794 8.25 0 5 2.5) and -5 m downstream side (F 72 8.25 16.5206 0 5 -5):
initial_WSL.PNG


The inflow rate is fixed at 9 m3/s (W 10 9), upstream water level is fixed at 2.5 m (F 61 2.5), the downstream level is guided by the flow rate and expected to reach steady condition.

I have tried multiple combination of discretization methods, numerical solvers, turning on and off initialization, turbulence model, adaptive time stepping on/off and grid spacing (15 cm ~ 850k cells, 10 cm ~ 2.8 million cells). However, I am unable to reach a steady state and the solution blows up after few thousand iterations. I am also unable to get the outflow value equal to inflow.


I have attached the control files for DIveMESH and REEF3D. I would be very grateful if somebody could help me figure out how to fix the problem.


I have attached plots of evolution of flow rate for a case without turbullence modelling and fixed time stepping. But the evolution for any other case is similar to this. FlowRate_evolution.PNG


Thank you in advance.

-Rabs

[valgrinda]

I also need the STL in order to have a look. You could send that via private message.

[valgrinda]

I think do to the turbulent nature of the flow, there are some fluctuations in the outflow, especially because there are som strong eddies directly at the end of the domain. Overall, the outflow then oscillates around the inflow value, so ok. Have a look at the attached results. Some more thoughts:

- do you need free surface? I think not.
- as always, a finer grid will improve things
- you need a turbulence model; with the current mesh RANS, with finer meshes you could use LES
- the CFL number was was too small
- use direct forcing, it works better and will soon be the default for the immersed boundary
- it looks like you run an older version of REEF3D, e.g. your F 40 options no longer exists
- best with I 10 1, the potential flow initialization will speed up the simulation

[ThatHimalayanGuy]

Hi Prof. Bihs,
Thank you very much for putting in time to look into the issue. That was very kind of you


I updated to the latest patch of REEF3D and now the simulation runs without explosion. Although G 3 1 function did not work for me.


The outflow rate oscillating around inflow does make sense considering the turbulent nature at outflow.



I have few more questions:
1. Is there option to recontinue a simulation from end time of an existing simulation (hot strat)? Last time I asked the question in forum, I was told the function is currently broken.


2. If we can hotstart now, is it possible to hotstart with finer meshing?



2. Is there any trick around how to filter out water that is stationary within the solid object stl? Think would help in postprocessing.

[valgrinda]

Hot start is working, but only for the exact same mesh.