Hello everyone,
I am currently using Code_Saturne V8.3 to perform an LES simulation with the WALE model.
I would like to experiment with different numerical schemes for the velocity field. However, I have encountered an issue in the GUI: regardless of which discretization scheme I select (e.g., Blending SOLU/centered), the "Centering Blend" field seems restricted to a value of 1. If I try to input any other value (such as 0.9), the input box turns red and is considered invalid.
Could someone please advise if this is a known limitation in the V8.3 GUI, or if there is a specific setting (like Slope Test) that must be toggled first?
Additionally, I would like to confirm if my understanding of the physics is correct:
If I select "Blending (SOLU/centered)" and set the Centering Blend to 0.9, does this correspond to 90% Central Differencing + 10% Second Order Linear Upwind (SOLU)?
Any clarification or advice would be greatly appreciated.
Best regards,
Xian Ge
[V8.3 GUI] Unable to change Centering Blend value for Velocity (locked to 1)
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
Yvan Fournier
- Posts: 4281
- Joined: Mon Feb 20, 2012 3:25 pm
Re: [V8.3 GUI] Unable to change Centering Blend value for Velocity (locked to 1)
Hello,
The slope test does not need to be toggled, as it adjusts to the scheme.
It seems the setting is disabled whenever you are using an LES turbulence model, because in this case a sencond(order scheme is required. In practice though, we seomtimes recommend adding 2% upwind (perhaps even) to some LES computations. Enough for stability, not too much to avoid degrading the scheme order too much.
But today, this can only be done in user-defined functions, though I will discuss this withe colleagues, to see if they agree we could relax this constraint in the GUI.
Other wise, yes, if you have 0.8 blending, it means you combine the centered scheme with a weight of 0.8, and SOLU with a weight of 0.2.
The poorly documented "blending" scheme additionally uses a local "hybrid_blend" field (whose values need to be defined/computed in some cases, or are computed by the turbulence model in others).
Also note that v8.3 is retired since June 2025. Switching to v9.0 or v9.1 is recommended (as usual, v9.0 will be maintained until the v11.0 release mid 2029, and v9.1 until the v9.3 release in December 2026).
Best regards,
Yvan
The slope test does not need to be toggled, as it adjusts to the scheme.
It seems the setting is disabled whenever you are using an LES turbulence model, because in this case a sencond(order scheme is required. In practice though, we seomtimes recommend adding 2% upwind (perhaps even) to some LES computations. Enough for stability, not too much to avoid degrading the scheme order too much.
But today, this can only be done in user-defined functions, though I will discuss this withe colleagues, to see if they agree we could relax this constraint in the GUI.
Other wise, yes, if you have 0.8 blending, it means you combine the centered scheme with a weight of 0.8, and SOLU with a weight of 0.2.
The poorly documented "blending" scheme additionally uses a local "hybrid_blend" field (whose values need to be defined/computed in some cases, or are computed by the turbulence model in others).
Also note that v8.3 is retired since June 2025. Switching to v9.0 or v9.1 is recommended (as usual, v9.0 will be maintained until the v11.0 release mid 2029, and v9.1 until the v9.3 release in December 2026).
Best regards,
Yvan