Eddy Viscosity

From CIRPwiki
Revision as of 15:15, 22 July 2014 by Doyle.L.Jones (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

The term eddy viscosity arises from the fact that small-scale vortices or eddies on the order of the grid cell size are not resolved and only the large-scale flow is simulated. The eddy viscosity is intended to simulate the dissipation of energy at smaller scales than the model can simulate. In the nearshore environment, large mixing or turbulence occurs due to waves, wind, bottom shear, and strong horizontal gradients; and therefore the eddy viscosity is an important aspect which can have a large influence on the calculated flow field and resulting sediment transport. In CMS-Flow, the total eddy viscosity, νt , is equal to the sum of three parts: 1) a base value ν0, 2) the current-related eddy viscosity νc, and 3) the wave-related eddy viscosity νw and is defined as,

 v_t = v_0 + v_c + v_w (2-21)

The base value (ν0) is approximately equal to the kinematic eddy viscosity (~1×10-6 m2/s), but may be changed by user. The other two components (νc and νw) are described in the sections below.

Current-Related Eddy Viscosity Component

There are four algebraic models for the current-related eddy viscosity: 1) Falconer Equation, 2) depth-averaged parabolic, 3) subgrid, and 4) mix-ing-length. The default turbulence model is the subgrid model, but may be changed in the user.

Falconer Equation

The Falconer (1980) equation was the default method applied in earlier version of CMS (Militello et al. 2004) for the current-related eddy viscosity. The equation is given by

v_c = 0.575c_{b}Uh (2-22)

where cb is the bottom friction coefficient, U is the depth-averaged current velocity magnitude, and h is the total water depth.


Depth-averaged Parabolic Model

The second option for the current-related eddy viscosity is the depth-averaged parabolic model given by

v_c = c_{v}u_{*}h  (2-23)

where cv is approximately equal to \kappa/6=0.0667.

Subgrid Model The third option for calculating the current-related eddy viscosity, νc , is the subgrid turbulence model given by

v_c = c_{v} u_{*} h + (c_s \Delta)^2 \mid\bar{S}\mid   (2-24)

in which

cv = vertical shear coefficient [-]

ch = horizontal shear coefficient [-]

\mid\bar{S}\mid = magnitude of the deformation (strain rate) tensor  e_{ij}  = \left(\sqrt {2e_{ij}e_{ij}}    \right)

 e_y = deformation (strain rate) tensor = = \frac{1}{2} \left(\frac{\partial V_i}{\partial x_j}  + \frac{\partial V_j}{\partial x_i} \right)

The empirical coefficients cv and cs are related to the turbulence pro-duced by the bed shear and horizontal velocity gradients, and Δ is the (average) grid size. The parameter cv is approximately equal to κ/6=0.0667 (default) but may vary from 0.01-0.2. The variable cs is equal to approximately the Smagorinsky coefficient (Smagorisnky 1963) and may vary between 0.1 and 0.3 (default is 0.2).

Mixing Length Model

The mixing length model implemented in CMS for the current-related eddy viscosity includes a component due to the vertical shear and is given by (Wu 2008)
 v_c = \sqrt {(c_v u_{*c} h)^2 + (l_{h}^2)|\bar S|)^2} (2-25)

where

l_h = mixing length  = \kappa min(c_h h, y^')) [m]
y^' = distance to the nearest wall [m]
 c_h = horizontal shear coefficient [-]

The empirical coefficient ch is usually between 0.3 and 1.2. The effects of bed shear and horizontal velocity gradients, respectively, are taken into account through the first and second terms on the right-hand side of Equation (2-26). It has been found that the modified mixing length model is better than the depth-averaged parabolic eddy viscosity model that accounts for only the bed shear effect.

Wave-Related Eddy Viscosity

The wave component of the eddy viscosity is separated into two components

 v_w = c_{wf}u_w H_s + c_{br}h \left(\frac{D_{br}}{\rho}   \right)^{1/3}  (2-26)

where

 c_{wf} = wave bottom friction coefficient for eddy viscosity [-]
 u_w = peak bottom orbital velocity [m/s] based on the significant wave height Hi [m] and peak wave period Tp [s]
c_{br} wave breaking coefficient for eddy viscosity [-]
D_{br} = wave breaking dissipation [N/m/s]

The first term on the right-hand side of Equation (2-27) represents the component due to wave bottom friction and the second term represents the component due to wave breaking. The coefficient cwf is approximately equal to 0.1 and may vary from 0.05 to 0.2. The coefficient cbr is approxi-mately equal to 0.08 and may vary from 0.04 to 0.15.

Wave Radiation Stresses

The wave radiation stresses, sij , are calculated using linear wave theory as (Longuet-Higgins and Stewart 1961; Dean and Dalrymple 1984)

S_{ij} = \iint E_w (f,\Theta) \left[n_g w_i w_j + \delta_{ij} \left(n_g - \frac{1}{2}   \right) \right]df d\theta (2-27)

where

f = the wave frequency [1/s]
\theta = the wave direction [rad]
wi = wave unit vector = (cos \theta, sin \theta) [-]
 \delta_{ij} =  \left\{
\begin{align}
&1 \quad \text{for i} = j \\
&0 \quad \text{for i} \neq j
\end{align} 
\right.


and

n_g = \frac{c_g}{c}=\frac{1}{2}\left( 1+\frac{2kh}{sinh 2kh}  \right) (2-28)

in which cg is the wave group velocity, c is the wave celerity, and k is the wave number.

Wave Flux Velocity

In the presence of waves, the oscillatory wave motion produces a net time-averaged mass (volume) transport referred to as Stokes drift. In the surfzone, the surface roller also provides a contribution to the mean wave mass flux. The mean wave mass flux velocity, or simply the mass flux ve-locity for short, is defined as the mean wave volume flux divided by the local water depth and is approximated here as (Phillips 1977; Ruessink et al. 2001; Svendsen 2006)

U_{wi}= \frac{    (E_w + 2 E_{s\gamma}) w_i}     {\rho hc} (2-29)

where

 E_w = wave energy 1/16 \rho gH_s^2 [N/m}

Hs = significant wave height [m]

E_{s\gamma}= surface roller energy density [N/m]

wi = wave unit vector  = (cos \theta, sin \theta) [-]

c = wave speed [m/s]

\theta = mean wave direction, [rad]

The first component is due to the wave energy ( is the Stokes velocity), while the second component is due to the surface roller ( is only pre-sent in the surfzone).

Wind Surface Stress

The wind surface stress is calculated as

\tau_{si} = \rho_a c_D WW_i  (2-30)

where

 \rho_a = air density at sea level [~1.2 kg/m3]
 c_D = wind drag coefficient [-]
 W_i = 10-m wind speed [m/s]
W = 10-m wind velocity magnitude  = \sqrt{W_i W_i} [m/s]

The wind speed is calculated using either an Eulerian or Lagrangian reference frame as

W_i = W_i^E - \gamma_w U_i   (2-31)

where

W_i^E = 10-m atmospheric wind speed relative to the solid earth (Eulerian wind speed) [m/s]
\gamma_w  = equal to 0 for the Eulerian reference frame or 1 for the Lagrangian reference frame
U_i = current velocity [m/s]

Using the Lagrangian reference frame or relative wind speed is more accurate and realistic for field applications (Bye 1985; Pacanowski 1987; Dawe and Thompson 2006), but the option to use the Eulerian wind speed is provided for idealized cases. The drag coefficient is calculated using the formula of Hsu (1988) and modified for high wind speeds based on field data by Powell et al. (2003)

c_D = \left\{
\begin{align}
&\left(\frac{\kappa}{14.56 - 2 ln W}\right)^2 for W \leq 30 m/s \\
&10^{-3} max(3.86 - 0.04 W, 1.5) for W > 30 m/s
\end{align}

\right.
(2-32) Powell et al. (2003) speculate that the reason for the decrease in drag coefficient with higher wind speeds is due to increasing foam coverage leading to the formation of a “slip” surface at the air-sea interface.

fig_2.2.png Figure 2.2 Modieifed Hsu (1988) wind drag coefficient

Wind measurements taken at heights other than 10 m are converted to 10-m wind speeds using the 1/7 rule (SPM 1984; CEM 2002)

 W_i = W_i^z\left(\frac{10}{z}  \right)^{\frac{1}{7}} (2-33)

where z is the elevation above the sea surface of the wind measurement and W_i^z is the wind velocity at height z .

Boundary Conditions