User Forum of Software BASEMENT

sterrier

User

Registered: 2020-01-09

Posts: 4

Velocity instabilities

Dear developers,

I'm trying to model the flow of a river into a reservoir. However, I obtain an unrealistically large vortex and locally high velocities (C in figure below). All non-compulsory parameters are set to their default values and the model is only water (no suspension or bed load). I've attempted to change parameters (minimum_time_step, maximum_time_step, CFL) but the phenomenon still occurs. Looking at the results, for a relatively long time the velocities are normal (A in figure below), then the velocities suddenly diverge in two adjacent elements (B in figure below, a minute later than A). Shorty afterwards a larger vortex (C in figure below) with velocities >2 m/s has formed and never disappears.

Do you have an idea of what is causing this problem, and how it can be solved?

Figure information: flow from top to bottom; identical color scale for the three figures (blue=0.01, red>=1 m/s); A & B are element results, C is node results.

Best regards,

Stéphane

Matthias Bürgler

Developer

From: Zürich

Registered: 2019-04-04

Posts: 171

Re: Velocity instabilities

Hi Stéphane
I am not sure what causes the problem in this case but here are two ideas:

1. High velocity gradients
In the region where the divergence of the velocity occurs the grid resolution seems to be quite coarse, i.e. 1 edge length for whole width of the main flow path (green). This results in high velocity gradients between the cells in the main flow path (green) and the cells on each side respectively (blue). From the shear flow between the green cells and the blue cells I would expect the formation of a vortex in this area, but since the mesh is so coarse a vortex cannot really form. Refining your mesh, especially in this narrow area could reduce the velocity gradient and hopefully the instabilities.

2. Large bed slopes
In BASEMENT v2.x the bed elevation of elements can be sloped, which can cause problems with wetting/drying of cells. Judging from the contour lines, there is a relatively high bed slope discretized by relatively few cells. Again, a finer mesh resolution of regions of high bed gradients could reduce problems.

Which CFL numbers did you try and what are the water depths?

Best regards
Matthias

sterrier

User

Registered: 2020-01-09

Posts: 4

Re: Velocity instabilities

Hi Matthias,

Thank you for your reply. After a break on the project, we are resuming our work on it.

The grid is indeed coarse to limit the calculation time (months in real time). And you are right, the banks are rather steep (close to a slope of 1 in some places)

The CFL was 1.0. A CFL = 0.8 delays the occurrence of the vortices by a few hours only. The water depth is around 50 m in the deepest parts (old riverbed before the reservoir impounding).

We attempted a refined mesh (see figure below) with the following results:

1. The vortices still occur. The maximum velocity is reduced, but there seem to be more vortices.

2. The software exited (disappeared without an error message box) in the middle of the simulation. It's not the first time we observe that. Any idea of what could be the cause?

Best regards,

Stéphane

Matthias Bürgler

Developer

From: Zürich

Registered: 2019-04-04

Posts: 171

Re: Velocity instabilities

Hi Stéphane

Do you get an error message in the .err file?

Best
Matthias

sterrier

User

Registered: 2020-01-09

Posts: 4

Re: Velocity instabilities

One simulation has an empty .err (only the headers of the simulation start).
One other simulation has the following message:

ERROR -> CGNS ERROR: could not find 'TimeIterValues' node error: goto path not found