Scribd
Upload
85 views

Axisymmetric Diffuser

The document analyzes flow separation and reattachment in an axisymmetric diffuser using RANS turbulence models. It compares experimental data to simulations with the Standard K-Epsilon, Realizable K-Epsilon, and SST K-Omega models. Only the SST K-Omega model accurately predicts the region of flow separation due to adverse pressure gradients. While all models perform well without separation, only SST K-Omega captures the reversed flow and separated boundary layer profile downstream. The conclusion is that SST K-Omega best predicts flows with adverse pressure gradients near walls.

Uploaded by

BHAVESH JOSHI
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
85 views

Axisymmetric Diffuser

The document analyzes flow separation and reattachment in an axisymmetric diffuser using RANS turbulence models. It compares experimental data to simulations with the Standard K-Epsilon, Realizable K-Epsilon, and SST K-Omega models. Only the SST K-Omega model accurately predicts the region of flow separation due to adverse pressure gradients. While all models perform well without separation, only SST K-Omega captures the reversed flow and separated boundary layer profile downstream. The conclusion is that SST K-Omega best predicts flows with adverse pressure gradients near walls.

Uploaded by

BHAVESH JOSHI
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

Flow in an Axisymmetric

Diffuser with RANS Models


- by Rajat Walia
Here is everything about it.

The expansion of the diffuser wall will give rise to adverse
pressure gradient.

Flow is turbulent with the Reynolds number of 2.8 * e5.

Slip wall is given to the top surface, experimentally


measured location of a streamline is used as external
boundary.

This test case very sensitive to the capability of the various


turbulence models to accurately predict separation with
reattachment.

Various velocity profile along the stream wise direction, skin


friction coefficient & pressure coefficients on the wall are
studied and compared with experimental results.

2 Equation RANS models, Standard K-Epsilon, Realizable


K-Epsilon & SST K-Omega are used to study the flow
separation & reattachment due to adverse Pressure
gradient.
Meshing
Hexahedral cells

Mesh 1 Mesh 2
Y+ 1 0.32
Number of Cells 7200 8400
First cell Height 2.5 * e(-5) 6.0 * e(-6)
Compute Wall y+ on Bottom Wall

Bottom Wall
Results & Discussion..
Coefficient of Friction plot along the stream wise direction on bottom wall.

Only K-OMEGA SST model was able to predict negative wall shear stress in the region 0 < XD < 2.

K-Epsilon standard (ske) & K-Epsilon Realizable (Rke) fails to predict separation zone.
Results & Discussion..
Coefficient of Pressure plot along the stream wise direction on bottom wall.

All models fails to predict flattening of the pressure distribution in the region 0 < XD < 3 &
overestimate the pressure.

However K-OMEGA SST gave better agreement and closer results with the experimental values.
Results & Discussion..
Velocity contour

Standard K-Epsilon

Realizable K-Epsilon

SST K-Omega

Negative Value of Velocity

SST K-OMEGA clearly shows separation zone indicated by region of negative velocity.

K-Epsilon variant fails to predict separation zone.


Results & Discussion..
Velocity vectors
Standard K-Epsilon

Realizable K-Epsilon

SST K-Omega

Back Flow predicted by K-OMEGA SST

SST K-OMEGA clearly predicts back flow near the wall due to adverse pressure gradient.

K-Epsilon variant flow remains attached.


Results & Discussion..
Velocity plot at various stream wise locations.

Plot Location

Flow remain attached in the region where cross-section is constant, All 3 model gives
good agreement to the experimental results.
Results & Discussion..
Velocity plot at various stream wise locations.

Plot Location

Further downstream, due to adverse pressure gradient, Only K-OMRGA SST model able
to predicts separated boundary layer profile properly.
Results & Discussion..
Velocity plot at various stream wise locations.

Plot Location

Further downstream, SST K-OMEGA gave very close agreement to the experimental data.

K-EPSILON variants shows attached flow close to the wall.


Conclusion..
K-Epsilon variants & K-Omega SST works well and predicts the flow field properly when there
is no adverse pressure gradeint.

K-Epsilon standard model overestimates the turbulent stress in adverse pressure gradient
making the flow remain attached to the wall.

K-Epsilon Realizable modifies the epsilon equation & introduces effect of mean flow distortion
on turbulent dissipation does not help in improving the flow prediction in adverse pressure
gradient.

Honors the Bradshaw’s observation that the shear stress in a boundary layer is proportional to
the turbulent kinetic energy k, To ensure this condition, Menter’s k-ω SST model limits the
turbulent by clipping turbulent viscosity.

If the interest is to predict flow behaviour near the wall, one should use K-Omega SST model
by placing first cell in viscous sub layer and with enough number of prism layers with in
boundary layer.

If what is happening near the wall is the mere interest, May be high Reynolds number flow
and One want to calculate pressure drag, lift or to capture shock wave then K-epsilon
realizable works well. Using the wall function & placing the first cell centroid in Log-layer is a
better approach.

You might also like