Downloaded from orbit.dtu.

dk on: Aug 11, 2021

A comparative study on the flow over an airfoil using transitional turbulence models

Lin, Mou; Sarlak Chivaee, Hamid

Published in:
Aip Conference Proceedings

Link to article, DOI:

10.1063/1.4951806

Publication date:
2016

Document Version
Publisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):
Lin, M., & Sarlak Chivaee, H. (2016). A comparative study on the flow over an airfoil using transitional turbulence
models. Aip Conference Proceedings, 1738, [500006 ]. https://doi.org/10.1063/1.4951806

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

 Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
 You may not further distribute the material or use it for any profit-making activity or commercial gain
 You may freely distribute the URL identifying the publication in the public portal

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
 A comparative study on the flow over an airfoil using transitional turbulence models
Mou Lin and Hamid Sarlak

Citation: AIP Conference Proceedings 1738, 030050 (2016); doi: 10.1063/1.4951806

View online: http://dx.doi.org/10.1063/1.4951806
View Table of Contents: http://scitation.aip.org/content/aip/proceeding/aipcp/1738?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Turbulence models in CFD simulation of low-Reynolds number airfoils flow
AIP Conf. Proc. 1648, 500006 (2015); 10.1063/1.4912704

Comparative study of various approaches for modeling transitional hypersonic rarefied gas flows over blunt
bodies
AIP Conf. Proc. 1501, 1500 (2012); 10.1063/1.4769716

Control of Stall Flow over Airfoil using Vortex Generators

AIP Conf. Proc. 1376, 203 (2011); 10.1063/1.3651876

Turbulence structure of open channel flows over permeable and impermeable beds: A comparative study
Phys. Fluids 21, 125109 (2009); 10.1063/1.3276292

A Comparative Study of Foreign Direct Investment Flow Using Diffusion Models

AIP Conf. Proc. 963, 1040 (2007); 10.1063/1.2835918

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions IP: 94.3.35.157 On: Sun, 26 Jun 2016 21:36:44
 A Comparative Study on the Flow over an Airfoil using
Transitional Turbulence Models
Mou Lin and Hamid Sarlak
Department of Wind Energy, Technical University of Denmark

Abstract. This work addresses the simulation of the flow over NREL S826 airfoil under a relatively low Reynolds number
(Re = 1 × 105 ) using the CFD solvers OpenFoam and ANSYS Fluent. The flow is simulated using two different transition
models, γ − Reθ and k − kL − ω model, and the results are examined against the k − ω SST model without transitional
formulations. By comparing the simulations with the available experimental data, we find that the using the transitional model
can effectively improve the flow prediction, especially the drag coefficient results, before the stall.
Keywords: Computational fluid dynamics, turbulence modelling
PACS: 47.27.em

INTRODUCTION
Despite the fact that most of the airfoils are designed for aircraft, e.g the well-known NACA airfoil series, National
Renewable Energy Laboratory (NREL) has developed a family of airfoils especially for the variable-speed and
variable-pitch, horizontal-axis wind turbine blade [1], including NREL S826 airfoil investigated in this study. The
profile of S826 airfoil is shown in Figure 1.

FIGURE 1. S826 airfoil profile

Several researches have been carried out on investigating the aerodynamic performance of NREL S826 airfoil using
both experimental and numerical approaches. Sarlark et al. [2] measured the major aerodynamics quantities of S826
airfoil at Re = 1 × 105 for a range of angles of attack (AoA) from -8 to 20 degree in the low speed wind tunnel at
Fluid Mechanics Laboratory, DTU. The measurements of the lift and drag and pressure distribution are then used to
validate the Large Eddy Simulation (LES) results performed in the in-house code EllipSys3D with the same setup. The
comparison showed a satisfying agreement between the experiment and LES, despite a lift over-prediction in the large
AoA. Cakmakcioglu et al. [3] compared the results of 2D and 3D RANS simulation using different turbulence models
and Detached Eddy Simulation (DES) in matching the experimental data of S826 airfoil’s aerodynamics quantities.
They reported that in the pre-stall region, the 2D γ − Reθ transitional model is a preferable choice in terms of giving
satisfying accuracy with less time cost.
In this report, a comparative study on the 2D simulation of the flow over NREL at Re = 1 × 105 will be performed
in ANSYS Fluent and the open-source CFD code OpenFOAM using two different transitional turbulence models:
γ − Reθ model and k − kl − ω model, and the results will be compared with the k − ω SST model as well as DTU’s
wind tunnel experiment data [2].

International Conference of Numerical Analysis and Applied Mathematics 2015 (ICNAAM 2015)
AIP Conf. Proc. 1738, 030050-1–030050-4; doi: 10.1063/1.4951806
Published by AIP Publishing. 978-0-7354-1392-4/$30.00

030050-1
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions IP: 94.3.35.157 On: Sun, 26 Jun 2016 21:36:44
 SIMULATION SETUP
The scale of the 2D computational domain is shown in Figure 2(a). The half-width of the domain and the incoming
distance from the velocity inlet to the airfoil are 15 times of the chord length (15c). The downstream distance from the
airfoil to the outlet is 20c.
The mesh for the computational domain is divided into the C-type structured mesh, shown in Figure 2(b), with the
extra refinement on the leading and trailing edge of the airfoil. After a mesh independence check on the meshes from
350 to 500 nodes on the airfoil, the mesh with 400 nodes on the airfoil is selected. A total number 2.2 × 105 cells is
used on the entire domain, with 200 nodes on the downsteam direction and 400 nodes on the vertical direction. The
near wall mesh is refined asymptotically to y+ = 1 as required by the wall resolved turbulence models. The growth
ratio of the nodes distance to the wall is 1.05.

(a) Computational domain scale (b) Mesh layout

FIGURE 2. Computational domain and mesh

The boundary conditions of the turbulence quantities of the computational domain are defined based on the
turbulence models used in the cases:

Velocity inlet/outlet. In k − ω SST model, the turbulence quantities k and ω are estimated by the mean velocity
U, the turbulence intensity I and the viscosity ratio µt /µ [4]:

3 ρ k µt −1
k = (UI)2 ; ω = ( )
2 µ µ

According to the wind tunnel set-up, the inlet turbulence intensity I = 0.2% and the viscosity ratio µt /µ = 0.1 in this
external flow case. In γ − Reθ model, the boundary conditions of k and ω are identical with k − ω SST model. The
boundary conditions of the two extra quantities in γ − Reθ model, γ and Reθ , are specified as fixed value (γ = 1) and
zero-gradient respectively. In k − kL − ω model, the kt is estimated in the same way as k, and the laminar energy kL is
set as a small value such as 1 × 10−8 .

Airfoil Wall. In k − ω SST model, k is set as fixed value 0 at the wall and a wall function for ω is used to get the
asymptotic behaviour at wall:
6(µ )
ω = 10 (1)
β1 (∆d1 )2
In γ − Reθ model, the boundary condition for γ and Reθ are defined as zero-gradient [5]. And in k − kL − ω model, the
boundary condition for ω is zero-gradient [6].
Other boundary conditions shared by all models for velocity and pressure are summarized in Table 1.

TABLE 1. Boundary condition of U and p

Velocity inlet Airfoil Wall Velocity outlet
U Fixed value Fixed value Fixed value
p Zero gradient Zero gradient Fixed value

030050-2
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions IP: 94.3.35.157 On: Sun, 26 Jun 2016 21:36:44
 RESULT
Based on the set-up discussed in the previous section, the simulations are performed at Re = 1 × 105 with different
attack angles from 0 to 14 degree.
2 0.25
Experiment
Experiment
k−ω SST k−ω SST
1.8 k−k −ω
L k−k −ω
L
γ−Reθ 0.2 γ−Re
1.6 θ

1.4
0.15
1.2
Cl

Cd
1 0.1

0.8
0.05
0.6

0.4 0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
AoA (◦ ) AoA (◦ )

(a) Lift Coefficient (b) Drag Coefficient

FIGURE 3. Lift and drag coefficients at Re = 1 × 105 from experiments and different turbulence models
Figure 3 shows the results of the lift and drag coefficients of the airfoil from CFD simulation and experiment data. In
Cl plot, we can see that the two transitional models, γ − Reθ and k − kL − ω , give results that match the measurements
quite well in the pre-stall range of AoA, both with a maximum relative error smaller than 4%. For k − ω SST model
without transitional formulation, the results are still good at the small AoA, but start to underestimate the lift as AoA
increases before the stall. In the post-stall range of AoA, γ − Reθ model gives a correct stalling AoA at 8◦ while the
result from other models show delay the stall from 4◦ to 8◦ , causing the deviations from the measurements in lift and
drag coefficients in this range.
In Cd plot, we can see that both transitional models improve the problem of the significant drag underestimation
(> 30%) in the results of k − ω SST model in the pre-stall range, despite the fact that they still fail to agree with the
measurement after the stall. At the stalling point, drop of the drag coefficient occurs in the experiment data and only
γ − Reθ model capture this phenomenon at this AoA.
Figure 4 shows the plots of pathlines for the cases of 0 AoA from three different turbulence models. We can see that
the major difference between the results from two transitional models and k − ω SST model results is that transitional
cases show the generation of the separation bubble at the upper surface near the tail and the concave part on the
bottom surface of the airfoil. These locations are the turning points of the airfoil geometry, hence are sensitive to the
flow separation. Furthermore, this difference also explains how the transitional models improve the problem of drag
underestimation in k − ω SST cases: the present of the separation bubble thicken the boundary layer thickness, thus
increase the total drag encountered by the airfoil in the flow [7].

(a) k − ω SST (b) k − kL − ω Transition (c) γ − Reθ Transition

FIGURE 4. Pathlines around the airfoil at AoA = 0◦ using different turbulence models
Figure 5 shows the turbulence kinetic energy (TKE) contour for the cases of 0 AoA using different turbulence
models. We can see that while k − ω SST model predicts that TKE production is mostly generated in the wake after
the trailing edge of the airfoil, the results of the transitional models shows that the major TKE generation occurs in
the location where the laminar separation bubbles are generated in Figure 4, indicating that the type of the transition
processes in these cases is the separation-induced transition. In terms of magnitude, γ − Reθ model generates the
largest TKE while k − ω SST model shows the smallest TKE production.
Figure 6 shows the pressure coefficient (C p ) distribution along the airfoil surface for the 0 AoA cases (in reverse Y
axis to match the airfoil geometry). We can observe the fluctuation of the C p curve at the location where the laminar

030050-3
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions IP: 94.3.35.157 On: Sun, 26 Jun 2016 21:36:44
 (a) k − ω SST (b) k − kL − ω Transition (c) γ − Reθ Transition

FIGURE 5. Contours around the airfoil at AoA = 0◦ using different turbulence models

separation bubbles (LSB) occur in the transitional model results. The location where the LSB start to occur in γ − Reθ
model result is 0.48 x/c (lower side) and 0.67 x/c (upper side). In k − kL − ω model, the location is 0.48 x/c (lower
side) and 0.65 x/c (upper side).

0.5
Cp

0
Experiment
k−ω SST
−0.5 k−kL−ω
γ−Reθ

−1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/c

FIGURE 6. Pressure distributions on the airfoil at AoA = 0◦ using different turbulence models

CONCLUSION
According to the discussion of the NREL S826 simulation results in the previous sections, we find that using the
transitional RANS model, both k − kL − ω model and γ − Reθ transitional model, improves the result of the lift and
drag coefficient effectively before the stall, particularly the drag underestimation problem in the k − ω SST model.
The explanation to this improvement is that the transitional model’s correctly predict the critical physics feature in the
transitional process such as the separation bubble generation and reattachment.
Comparing γ − Reθ model and k − kL − ω model, we find that k − kL − ω model gives a better result of lift and drag
coefficients before the stall, but it over-predict the stalling angle and result in a significant drag overestimation in the
post-stall region while γ − Reθ model gives the correct stalling AoA and the trend of decreasing lift after the stall.

REFERENCES
1. D. M. Somers, National Renewable Energy Laboratory, Subcontractor Report (2005).
2. H. Sarlak, R. Mikkelsen, S. Sarmast, et al., ŞAerodynamic behaviour of NREL S826 airfoil at Re = 100,000,Ť in Journal of
Physics: Conference Series, IOP Publishing, 2014, vol. 524, p. 012027.
3. S. Cakmakcioglu, I. Sert, O. Tugluk, and N. Sezer-Uzol, Ş2-D and 3-D CFD Investigation of NREL S826 Airfoil at Low
Reynolds Numbers,Ť in Journal of Physics: Conference Series, IOP Publishing, 2014, vol. 524, p. 012028.
4. F. R. Menter, Two-equation eddy-viscosity turbulence models for engineering applications (1994).
5. R. B. Langtry, and F. R. Menter, AIAA Journal 47, 2894-2906 (2009), ISSN 0001-1452.
6. D. K. Walters, and D. Cokljat, Journal of Fluids Engineering 130, 121401 (2008), ISSN 00982202.
7. H. Schlichting, K. Gersten, and K. Gersten, Boundary-layer Theory, Springer Science & Business Media, 2000.

030050-4
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions IP: 94.3.35.157 On: Sun, 26 Jun 2016 21:36:44