CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Pressure distribution over a bluff body RANS OR LES? (https://www.cfd-online.com/Forums/main/226381-pressure-distribution-over-bluff-body-rans-les.html)

sonsonst April 25, 2020 13:48

Pressure distribution over a bluff body RANS OR LES?
 
Hello,

I am dealing with a geometry which is similar to the classical turbulent flow over an inclined plate (high AoA 135º). Let's, assume Re=1e6 and that my primary interest is knowing the pressure distribution on the forebody (upstream part).

I know that RANS has its limitation when it comes to large flow separation and swicthing to a LES/Hybrid-RANSLES is optimal. As you already know there is a massive increase in computational cost switching to a hybrid model. However, if I am not interested in what happens downstream - could I rely on results from steady-state RANS simulation? I am not actually interested in the transient solution. Even if the simulation is not accurate would I be in the ballpark? (hard to tell I imagine)

I have tried to find something similar in the forum or in the literature, though there is much said on bluff bodies and theses classical problem, I have not found an answer to my question. I am trying to model both and compare results.

Many thanks in advance

FMDenaro April 25, 2020 14:03

Quote:

Originally Posted by sonsonst (Post 767349)
Hello,

I am dealing with a geometry which is similar to the classical turbulent flow over an inclined plate (high AoA 135º). Let's, assume Re=1e6 and that my primary interest is knowing the pressure distribution on the forebody (upstream part).

I know that RANS has its limitation when it comes to large flow separation and swicthing to a LES/Hybrid-RANSLES is optimal. As you already know there is a massive increase in computational cost switching to a hybrid model. However, if I am not interested in what happens downstream - could I rely on results from steady-state RANS simulation? I am not actually interested in the transient solution. Even if the simulation is not accurate would I be in the ballpark? (hard to tell I imagine)

I have tried to find something similar in the forum or in the literature, though there is much said on bluff bodies and theses classical problem, I have not found an answer to my question. I am trying to model both and compare results.

Many thanks in advance




Very difficult to say that the pressure distribution in the upstream part is not influenced by the downstream region where separation is present.
I would not use RANS.

duri April 28, 2020 07:08

Quote:

Originally Posted by sonsonst (Post 767349)
I know that RANS has its limitation when it comes to large flow separation and swicthing to a LES/Hybrid-RANSLES is optimal. As you already know there is a massive increase in computational cost switching to a hybrid model. However, if I am not interested in what happens downstream - could I rely on results from steady-state RANS simulation? I am not actually interested in the transient solution. Even if the simulation is not accurate would I be in the ballpark? (hard to tell I imagine)


To my understanding any turbulence model RANS or LES has its own assumptions and tuning coefficients. The selection of turbulence model should be based on your application, desired results and cost. All turbulence model can fairly predict pressure distribution and forces because they are mostly affected by inviscid solution (except skin friction) viscous effect are secondary through displacement thickness. Since this is flat plat with inclined orientation I assume the separation is from corner, turbulence model may not play critical role in force estimation. LES or DES is more useful in wake investigation. Again RANS or LES has multiple options to choose and which could affect the solution.

It is always better to validate the model against some test data before using it for prediction. Probably the validation step will guide you to choose the correct the model. I would always start from cost effective solution.


All times are GMT -4. The time now is 07:12.