# A roughness-dependent model

(Difference between revisions)
Jump to: navigation, search
 Revision as of 15:20, 19 June 2007 (view source)R.absi (Talk | contribs) (→Algebraic eddy viscosity model [Absi (2006)])← Older edit Latest revision as of 12:47, 22 June 2007 (view source)R.absi (Talk | contribs) (→The mean velocity profile) (13 intermediate revisions not shown) Line 1: Line 1: - ==Two-equation eddy viscosity model== + ==Two-equation $k$-$\epsilon$ eddy viscosity model==
$[itex] - \nu _t = C_{\mu} {{k^2 } \over \varepsilon } + \nu _t = C_{\mu} {{k^2 } \over \epsilon }$(1)
[/itex]
(1)
where: where: $C_{\mu} = 0.09$ $C_{\mu} = 0.09$ + + [http://www.cfd-online.com/Wiki/Standard_k-epsilon_model $k$-$\epsilon$ model] ==One-equation eddy viscosity model== ==One-equation eddy viscosity model== Line 12: Line 14: \nu _t  = k^{{1 \over 2}}  l \nu _t  = k^{{1 \over 2}}  l [/itex]
(2)
[/itex]
(2)
+ + [http://www.cfd-online.com/Wiki/Prandtl%27s_one-equation_model One-equation model] ==Algebraic eddy viscosity model== ==Algebraic eddy viscosity model== Line 20: Line 24: $l_m$ is the mixing length. $l_m$ is the mixing length. - ===Algebraic model for the turbulent kinetic Energy=== + ===Algebraic model for the turbulent kinetic energy===
$[itex] k^{{1 \over 2}}(y) = {1 \over {C_{\mu}}^{{1 \over 4}}} u_\tau e^{\frac{-y}{A}} k^{{1 \over 2}}(y) = {1 \over {C_{\mu}}^{{1 \over 4}}} u_\tau e^{\frac{-y}{A}}$(4)
[/itex]
(4)
- $u_\tau$ is the shear velocity and $A$ a model parameter. + $u_\tau$ is the shear velocity and $A$ a model parameter. + + For steady open channel flows in local equilibrium, where the energy production is balanced by the dissipation, from the modeled $k$-equation [[#References|[Nezu and Nakagawa (1993)]]] obtained a similar semi-theoretical equation. + + ===Algebraic model for the mixing length=== + + For local equilibrium, an extension of von Kármán’s similarity hypothesis allows to write, with equation (4) [[#References|[Absi (2006)]]]: - ===Algebraic model for the mixing length, based on (4) [[#References|[Absi (2006)]]]===
$[itex] l_m(y) = \kappa \left( A - \left(A - y_0\right) e^{\frac{-(y-y_0)}{A}} \right) l_m(y) = \kappa \left( A - \left(A - y_0\right) e^{\frac{-(y-y_0)}{A}} \right)$(5)
[/itex]
(5)
- $\kappa = 0.4$, $y_0$ is the hydrodynamic roughness + $\kappa = 0.4$, $y_0$ is the hydrodynamic roughness. + For a smooth wall ($y_0 = 0$): +
+ $+ l_m(y) = \kappa A \left( 1 - e^{\frac{-y}{A}} \right) +$(6)
===the algebraic eddy viscosity model is therefore=== ===the algebraic eddy viscosity model is therefore=== Line 39: Line 53: \nu _t(y)  = \kappa \left( A - \left(A - y_0\right) e^{\frac{-(y-y_0)}{A}} \right) \nu _t(y)  = \kappa \left( A - \left(A - y_0\right) e^{\frac{-(y-y_0)}{A}} \right) u_\tau  e^{\frac{-y}{A}} u_\tau  e^{\frac{-y}{A}} - [/itex]
(6)
+ [/itex]
(7)
+ + + ==The mean velocity profile== + + For local equilibrium, we are able to find the mean velocity profile $u$ from the turbulent kinetic energy $k$ (equation 4) and the mixing length $l_m$ (equation 5), by: +
+ $+ {{d u} \over {d y}} = C_{\mu}^{1 \over 4} {{k^{1 \over 2}} \over {l_m}} +$(8)
+ + Figure (1) shows that the velocity profile obtained from equations (8), (4) and (5) (solid line) is more accurate than the logarithmic velocity profile (dash-dotted line). + + [[Image:fig7a.jpg]] + [[Image:fig7b.jpg]] + + '''Figure 1''', Vertical distribution of mean flow velocity. + $A = {{h} \over {c_1}}$; $c_1 = 1$; + Dash-dotted line: logarithmic profile; solid line: obtained from equation (8); symbols: experimental data ([[#References|[Sukhodolov ''et al.'' (1998)]]]). a) profile 2: $y_0 = 0.062 cm$; $h = 145 cm$; $u_\tau = 3.82 cm/s$. b) profile 4: $y_0 = 0.113 cm$; $h = 164.5 cm$; $u_\tau = 3.97 cm/s$ ; (values of $y_0 , h, u_\tau$  are from  [[#References|[Sukhodolov ''et al.'' (1998)]]]); Figure from [[#References|[Absi (2006)]]]. == References == == References == - * {{reference-paper|author=Absi, R. |year=2006|title=A roughness and time dependent mixing length equation|rest=Journal of Hydraulic, Coastal and Environmental Engineering, Japan Society of Civil Engineers, (Doboku Gakkai Ronbunshuu B), Vol. 62, No. 4, pp.437-446}} + * {{reference-paper|author=Absi, R. |year=2006|title=[http://www.jstage.jst.go.jp/article/jscejb/62/4/62_437/_article A roughness and time dependent mixing length equation]|rest=''Journal of Hydraulic, Coastal and Environmental Engineering, (Doboku Gakkai Ronbunshuu B),'' Japan Society of Civil Engineers, Vol. '''62''', No. 4, pp.437-446}} + + * {{reference-paper|author=Nezu, I. and Nakagawa, H. |year=1993|title=Turbulence in open-channel flows|rest=A.A. Balkema, Ed. Rotterdam, The Netherlands}} + + * {{reference-paper|author=Sukhodolov, A., Thiele, M. and Bungartz, H. |year=1998|title=Turbulence structure in a river reach with sand bed|rest=''Water Resour. Res.'', Vol. '''34''', pp. 1317-1334}} + [[Category:Turbulence models]] [[Category:Turbulence models]] {{stub}} {{stub}}

## Two-equation $k$-$\epsilon$ eddy viscosity model

 $\nu _t = C_{\mu} {{k^2 } \over \epsilon }$ (1)

where: $C_{\mu} = 0.09$

## One-equation eddy viscosity model

 $\nu _t = k^{{1 \over 2}} l$ (2)

## Algebraic eddy viscosity model

 $\nu _t(y) = {C_{\mu}}^{{1 \over 4}} l_m(y) k^{{1 \over 2}}(y)$ (3)

$l_m$ is the mixing length.

### Algebraic model for the turbulent kinetic energy

 $k^{{1 \over 2}}(y) = {1 \over {C_{\mu}}^{{1 \over 4}}} u_\tau e^{\frac{-y}{A}}$ (4)

$u_\tau$ is the shear velocity and $A$ a model parameter.

For steady open channel flows in local equilibrium, where the energy production is balanced by the dissipation, from the modeled $k$-equation [Nezu and Nakagawa (1993)] obtained a similar semi-theoretical equation.

### Algebraic model for the mixing length

For local equilibrium, an extension of von Kármán’s similarity hypothesis allows to write, with equation (4) [Absi (2006)]:

 $l_m(y) = \kappa \left( A - \left(A - y_0\right) e^{\frac{-(y-y_0)}{A}} \right)$ (5)

$\kappa = 0.4$, $y_0$ is the hydrodynamic roughness. For a smooth wall ($y_0 = 0$):

 $l_m(y) = \kappa A \left( 1 - e^{\frac{-y}{A}} \right)$ (6)

### the algebraic eddy viscosity model is therefore

 $\nu _t(y) = \kappa \left( A - \left(A - y_0\right) e^{\frac{-(y-y_0)}{A}} \right) u_\tau e^{\frac{-y}{A}}$ (7)

## The mean velocity profile

For local equilibrium, we are able to find the mean velocity profile $u$ from the turbulent kinetic energy $k$ (equation 4) and the mixing length $l_m$ (equation 5), by:

 ${{d u} \over {d y}} = C_{\mu}^{1 \over 4} {{k^{1 \over 2}} \over {l_m}}$ (8)

Figure (1) shows that the velocity profile obtained from equations (8), (4) and (5) (solid line) is more accurate than the logarithmic velocity profile (dash-dotted line).

Figure 1, Vertical distribution of mean flow velocity. $A = {{h} \over {c_1}}$; $c_1 = 1$; Dash-dotted line: logarithmic profile; solid line: obtained from equation (8); symbols: experimental data ([Sukhodolov et al. (1998)]). a) profile 2: $y_0 = 0.062 cm$; $h = 145 cm$; $u_\tau = 3.82 cm/s$. b) profile 4: $y_0 = 0.113 cm$; $h = 164.5 cm$; $u_\tau = 3.97 cm/s$ ; (values of $y_0 , h, u_\tau$ are from [Sukhodolov et al. (1998)]); Figure from [Absi (2006)].

## References

• Nezu, I. and Nakagawa, H. (1993), "Turbulence in open-channel flows", A.A. Balkema, Ed. Rotterdam, The Netherlands.
• Sukhodolov, A., Thiele, M. and Bungartz, H. (1998), "Turbulence structure in a river reach with sand bed", Water Resour. Res., Vol. 34, pp. 1317-1334.