Gas turbine engines run with free stream turbulence (FST) intensities as high as 20%. FST can be defined as the turbulence in the approach stream which affects turbulent boundary layers. The main objective of the present study is to test and compare the prediction capabilities of low Reynolds number k-ε models in predicting turbulent

6th OpenFOAM Workshop, The Pennsylvania State University, University Park PA, USA, June 13-16, 2011

Two Applications of OpenFOAM:

Evaluation of k-ε Models in Boundary Layer Predictions under

High Free Stream Turbulence

Simulation of the Flow in Simplified Draft Tube of a Francis

Turbine

Introduction

Results

Conclusions

Computational Approach

Introduction

Low Free Stream Turbulence (Tu=1%)

Moderately High Free Stream Turbulence (Tu=6.53%)

Very High Free Stream Turbulence (Tu=25.7%)

Tu=0%

Tu=3%

Results

Fig. 1.2. Skin friction coefficient (left) and TKE profiles (right)

Fig. 1.3. Skin friction coefficient (left), TKE profiles at x=1.32m (middle) and TKE profiles at x=1.73 m (right)

Fig. 1.4. Skin friction coefficient (left), TKE profiles at x=1.08m (middle) and TKE profiles at x=2.08 m (right)

Fig. 2.4. Axial velocity profiles

Fig. 2.6. TKE profiles

Fig. 2.2. Computational grid

Fig. 2.3. Pressure (left) and axial velocity (right) contours

in the meridian plane

Conclusions

Fig. 2.5. Circumferential velocity profiles

Fig. 2.7. Flow stream-traces

Fig. 2.1. The FLINDT project draft tube

Uz [m/s]

Hydraulic power plants often operate under off-design conditions. In these cases, water exits the runner with a strong swirl which leads to formation of an unsteady

vortex rope. The main goal of the present study is to investigate the capability of commonly used two-equation turbulence models in predicting the flow behavior under these conditions.

Models give reasonably good results for low FST conditions Predictions become poorer (even up to more than 100% for TKE) as FST increases For very high FST conditions models fail to predict the turbulent boundary layer behavior Modifications are needed for all models. This would be the future step in this study. OpenFOAM would be a very useful tool since it makes the implementation of new models easier.

The average location of the vortex rope can be captured by the k-ε model, however, deviations from the experimental data is seen near the centerline where the vortex rope exists. This shows that the correct behavior of the flow near the vortex rope can not be predicted by these models. Again, turbulence models should be modified near the centerline and OpenFOAM can be used because of the availability of the source code.

Hosein Foroutan and Savas Yavuzkurt Department of Mechanical and Nuclear Engineering, The Pennsylvania State University

P/ρ [m2/s2]

flow under high free stream turbulence conditions. Since a basic study can eliminate other effects and focuses only on effects of FST, a 2-D flat plate boundary layer with constant free stream velocity is chosen for this study. Two low Reynolds number k-ε models namely Lam and Bremhorst (LB), and, Launder and Sharma (LS) are considered. Fig. 1.1. Effect of FST on turbulent boundary layer

Simplified draft tube based on the hydraulic diameter

Parameterizing blockMeshDict using an m4 script

540,000 cells

k-ε with wall functions, average y+ = 31 – 33

Inlet boundary conditions : profile1DfixedValue

(Available thorough OpenFoamTurbo library)

SimpleFoam