Documentos de Académico
Documentos de Profesional
Documentos de Cultura
k-ε MODEL
Abstract
Three versions of low Reynolds number k-ε model have been tested against the DNS data for a
transitional oscillatory boundary layer. The original model is observed to predict the transitional
characteristics better than the other versions. However, the later versions perform better by virtue of
the magnitude of velocity, turbulent kinetic energy and bottom shear stress.
Keywords: Oscillatory boundary layer, k-ε model, bottom shear stress, turbulence.
1. Introduction
The coastal bottom boundary layers undergo transition from laminar to turbulent or vice versa in
response to changing field conditions. Due to the importance of transitional characteristics of
bottom boundary layer in relation to the sediment movement, a lot of research has been done in the
past. Although many experimental and analytical studies have been carried out, the idea of using
turbulence models to tackle this phenomenon is relatively new. With the availability of excellent
computing facilities at affordable costs, this option is gaining more popularity among the hydraulic
and coastal engineers. The benefit of using a good turbulence model is that it produces detailed
boundary layer properties to facilitate precise estimation of sediment movement under a variety of
hydraulic conditions. For a turbulence model to be good, an essential requirement is computational
economy with reasonable accuracy. With the large number of turbulence models available, it is
very difficult for practicing engineers to choose the most suitable one. The present study deals with
the application of a low Reynolds number k-ε model to a transitional oscillatory boundary layer.
The term low Reynolds number implies that this model is applicable over the whole cross-stream
dimension including the low Reynolds number region (viscous sublayer).
1
newer versions with reference to the available Direct Numerical Simulation (DNS) data for
oscillatory boundary layers. Another interesting result of this study concerns the nature of damping
functions used in Low Reynolds number k-ε models. The damping function used in many of the
newer versions involves the wall coordinates y+ (=yuf/ν , y = cross-stream distance, uf = shear
velocity and ν = kinematic viscosity). In case of oscillatory flow the bottom shear stress goes to
zero twice in a wave cycle and at that time the damping function becomes zero in the whole of the
cross-stream dimension, which is physically incorrect.
In the present study, two of the available versions of k-ε model that meet the requirements of
damping function outlined by Sana and Tanaka(2000) have been tested against the DNS data by
Spalart and Baldwin(1989) for sinusoidal oscillatory boundary layer on a smooth boundary. Here
the k-ε models proposed by Yang and Shih(1993)(YS Model) and Abe et al.(1994)(AKN Model)
have been tested.
∂ε~ ∂ ν t ∂ε~ ε~ ∂u ε~ 2
2
=
ν+ + C1 f 1ν t − C 2 f 2 +E (4)
∂t ∂y σ ε ∂y k ∂y k
Here, Cµ , C1, C2, σk and σε are model constants and fµ , f1, f2, D and E are model functions. The
value of Cµ is 0.09 and those of other model parameters, for the three versions considered here, are
presented in Table 1. In this table, Rt = k 2 /(ε~ν ) , Rk = k y / ν and y * = (ν ε)1 / 4 y / ν
(Kolmogorov’s length scale).
A Crank-Nicolson type implicit finite difference scheme was used here. The grid spacing near the
wall was varied exponentially in order to achieve better accuracy. In space 100 and in time 6000
steps per wave cycle were used. The convergence was based primarily on velocity, k and ε and
2
then on maximum bottom shear stress. The convergence limit was set to 1 ×10 −6 in the present
study.
The magnitude of the maximum wall shear stress predicted by the YS and AKN models is closer to
the DNS value than than that by the JL model. It can be observed that this sudden increase in
shear stress occurs at the inflection point in the temporal variation of acceleration
(d(U/Uo)/d(ω t)).
4. Conclusions
The newer versions of k-ε model perform better than the JL model in predicting near wall
velocities, turbulent kinetic energy and bottom shear stress are concerned. During deceleration
(ω t=150o) none of the tested models could predict the velocity as well as turbulent kinetic energy
precisely. Further investigation on the predictive ability of these models has to be carried out in
order to propose the possible improvement for better prediction during deceleration.
References
Abe, K., Kondoh, T. and Nagano, Y., 1994, A new turbulence model for predicting fluid flow and
heat transfer in separating and reattaching flows-I. Flow field calculations, Int. J. Heat Mass Trans.,
Vol. 37, No.1, pp.139-151.
Jones, W. P. and Launder, B. E. 1972, The prediction of laminarization with a two-equation model
of turbulence. Int. J. Heat Mass Transfer Vol.15, pp.301-314.
3
Patel, V. C., Rodi, W. and Scheuerer, G. 1985, Turbulence models for near-wall and low Reynolds
number flows: A review. AIAA J. Vol. 23, No.9, pp.1308-1319.
Sana, A. and Tanaka, H. 2000, Review of k-ε model to analyze oscillatory boundary layers,
Journal of Hyd. Engg., Vol. 126, No.9, pp.701-710.
Spalart, P.R. and Baldwin, B. S. 1989, Direct simulation of a turbulent oscillating boundary layer,
In Turbulent Shear Flows 6, pp.417-440.
Tanaka, H. and Sana, A. 1994, Numerical study on transition to turbulence in a wave boundary
layer. In Sediment Transport Mechanism in Coastal Environments and Rivers, pp.14-25.
Yang, Z. and Shih, T. H., 1993, New time scale based k-ε model for near-wall turbulence, AIAA
J., Vol. 31, No.7, pp.1191-1198.
4
Tables and Figures for the paper
ANALYSIS OF A TRANSITIONAL OSCILLATORY BOUNDARY LAYER BY k-ε MODEL
by
AHMAD SANA AND SHUY ENG BAN
Table 1. Model parameters for the low Reynolds number k-ε model
Model Parameter
C1 C2 σ k σ ε fµ
JL 1.55 2.0 1.0 1.3 exp{ −2.5 /(1 + R t / 50 )}
YS 1.44 1.92 1.0 1.3 3
{1 − exp( −0.00015 Rk − 5 ×10 −7 Rk −10 −10 Rk )}1 / 2
5
(1 +1 / Rt )
AKN 1.50 1.90 1.4 1.4 [1 + (5 / Rt
3/ 4
) exp{− ( Rt / 200) 2 }]{1 − exp(− y * / 14)}2
Model Parameter
f1 f2
JL 1.0 2
1 − 0.3 exp( −Rt )
YS Rt /(1 + Rt ) Rt /(1 + Rt )
AKN 1.0 [1 − 0.3 exp{ −( Rt / 6.5) 2 }]{1 − exp( − y * / 3.1)} 2
Model Parameter
D E ~
ε 0
JL 2ν(∂ k / ∂y ) 2 2ν νt (∂ 2 u / ∂y 2 ) 2 0.0
YS 0.0 ν νt (∂ 2 u / ∂y 2 ) 2 ν(∂2 k / ∂y 2 )
AKN 0.0 0.0 ν(∂2 k / ∂y 2 )
5
Figure 1. Comparison of velocity profile between JL, YS, AKN model and DNS data
6
Figure 2. Turbulent kinetic energy profiles
7
Figure 3. Wall shear stress, free-stream velocity and acceleration