RANS Boundary, Region, and Initial Conditions

Non-Flow Boundaries

At wall, symmetry plane, and axis boundaries, the transported variables are defined as follows:

RANS Model Transported Variable Wall Symmetry Axis
Spalart-Allmaras models ν˜ 0 ν~/n|sym=0 ν~/n|axis=0
K-Epsilon models k k/n|wall=0 k/n|sym=0 k/n|axis=0
ε ε/n|wall=0

In addition, the wall cell value is imposed according to the appropriate method in the wall treatment. See Wall Treatment for RANS and DES—Turbulence Quantities.

ε/n|sym=0 ε/n|axis=0
(Lag) Elliptic Blending model k k/n|wall=0 k/n|sym=0 k/n|axis=0
ε ε/n|wall=0

In addition, the wall cell value is imposed according to the appropriate method in the wall treatment. See Wall Treatment for RANS and DES—Turbulence Quantities.

ε/n|sym=0 ε/n|axis=0
φ φ/n|wall=0 φ/n|sym=0 φ/n|axis=0
α 0 α/n|sym=0 α/n|axis=0
K-Omega models k k/n|wall=0 k/n|sym=0 k/n|axis=0
ω ω/n|wall=0

In addition, the wall cell value is imposed according to the appropriate method in the wall treatment. See Wall Treatment for RANS and DES—Turbulence Quantities.

ω/n|sym=0 ω/n|axis=0
Reynolds Stress model R R/n|wall=0

In addition, a method that is developed by Hadzic [340] is used to impose the value of production of each stress component in accordance with a wall function approach. See Strain Rate Modification for Reynolds Stress Transport Models.

See Reynolds Stress Model—Symmetry. See Reynolds Stress Model—Axis.
ε ε/n|wall=0

In addition, the wall cell value is imposed according to the appropriate method in the wall treatment. See Wall Treatment for RANS and DES—Turbulence Quantities.

ε/n|sym=0 ε/n|axis=0

Reynolds Stress Model—Symmetry

On a symmetry boundary, the procedure for setting the Reynolds stress tensor at the face R f is as follows. First, the tensor at the adjacent cell center R is rotated such that it is parallel to the symmetry plane to obtain a new tensor termed R' . This then enables the symmetry boundary face values of the tensor in this rotated coordinate system to be explicitly set as follows:

Figure 1. EQUATION_DISPLAY
R'f=[R110R130R220R310R33]c
(1352)

The values of R f are then computed by rotating the R'f tensor back to the original orientation of R —this is made easy by the fact that the inverse of the rotation tensor is simply its transpose, since it is an orthogonal tensor.

Reynolds Stress Model—Axis

On an axis boundary, the procedure for setting the Reynolds stress tensor at the face R f is similar to the symmetry plane. In this case, however, the face values of the tensor in the coordinate system parallel to the axis are as follows:

Figure 2. EQUATION_DISPLAY
R'f=[R11000R22000R33]c
(1353)

Flow Boundaries, Region, and Initial Conditions

When defining values for flow boundaries, region, and initial conditions, you have three choices for specifying the transported variables:

  • Specify the values of the transported variables directly.
  • Have the values of the transported variables derived from a specified turbulence intensity I and length scale L .
  • Have the values of the transported variables derived from a specified turbulence intensity I and turbulent viscosity ratio μt/μ .
RANS Model Direct Specification of transported variables Derivation of transported variables from I and L Derivation of transported variables from I and μt/μ
Spalart-Allmaras model ν˜
Figure 3. EQUATION_DISPLAY
Cμ0.2532IvL
(1354)
Figure 4. EQUATION_DISPLAY
μtμ=ρν˜fv1μ
(1355)
K-Epsilon model k
Figure 5. EQUATION_DISPLAY
32(Iv)2
(1356)
Figure 6. EQUATION_DISPLAY
32(Iv)2
(1357)
ε
Figure 7. EQUATION_DISPLAY
Cμ3/4k3/2L
(1358)
Figure 8. EQUATION_DISPLAY
ρCμk2(μt/μ) μ
(1359)
(Lag) Elliptic Blending model k As for K-Epsilon model.
ε
φ φ φ
NOTE: In porous regions, the value of φ is set to 2/3 .
K-Omega model k As for K-Epsilon model.
ω
Figure 9. EQUATION_DISPLAY
kLβ*1/4
(1360)
Figure 10. EQUATION_DISPLAY
ρk(μt/μ)μ
(1361)
Reynolds Stress model R
Figure 11. EQUATION_DISPLAY
23(Iv)2I
(1362)
Figure 12. EQUATION_DISPLAY
(Iv)2I
(1363)
ε
Figure 13. EQUATION_DISPLAY
Cμ3/4[12trR]3/2L
(1364)
Figure 14. EQUATION_DISPLAY
Cμ[12trR]2(μt/μ)μ
(1365)

where:

  • v is the local velocity magnitude in the local frame of reference. For initial conditions, v is the specified initial turbulent velocity scale.
  • C μ and β* are model coefficients.
  • fv1 is a Damping Function.
For the Reynolds stress model, a fourth choice allows you to enter the dissipation rate ε directly but have the Reynolds stresses R derived from the turbulent kinetic energy k using the relation:
Figure 15. EQUATION_DISPLAY
R=23kI
(1366)

where I is the identity matrix.

Boundary Layer Initialization

When convergence is slow, or when the initialization of the boundary layer is the main consideration (as in internal flows), you can initialize flow in the near-wall region only using the local relative velocity ur, the wall-parallel non-dimensional velocity u+ and the y+ value:

RANS Model Transported Variable Calculation
Spalart-Allmaras model ν˜
Figure 16. EQUATION_DISPLAY
max[1du+/dy+1,0]=ρν˜fv1μ
(1367)
K-Epsilon model k
Figure 17. EQUATION_DISPLAY
(1du+/dy+1)εκνfu
(1368)
ε
Figure 18. EQUATION_DISPLAY
1max(17,y+)ur4κν
(1369)
(Lag) Elliptic Blending model k As for K-Epsilon model.
ε
φ As specified for region initialization.
K-Omega model k As for K-Epsilon model.
ω
Figure 19. EQUATION_DISPLAY
εβ*fβ*k
(1370)

where k and ε are calculated as for K-Epsilon model.

Reynolds Stress model R
Figure 20. EQUATION_DISPLAY
23kI
(1371)

where k is calculated as for K-Epsilon model.

ε As for K-Epsilon model.

where: