Wind-US User's Guide - Keyword Reference, DEBUG

(Wind-US Documentation Home Page) (Wind-US User's Guide) (GMAN User's Guide) (MADCAP User's Guide) (CFPOST User's Guide) (Wind-US Utilities) (Common File User's Guide) (Wind-US Installation Guide) (Wind-US Developer's Reference) (Guidelines Documents)

(Introduction) (Tutorial) (Geometry and Flow Physics Modeling) (Numerical Modeling) (Boundary Conditions) (Convergence Monitoring) (Files) (Scripts) (Parallel Processing) (Keyword Reference) (Test Options)

DEBUG - Developer debug options

DEBUG number [mode]

This keyword is intended for use during code development and testing. The various debug options and modes are described below. For each option, the subroutines referencing that option are listed in parentheses. If mode is not described for a given debug option, it should be set to 1 to activate that option. If mode is omitted, a default value of 1 is used.

Unless noted otherwise, these options only apply to unstructured grids.

number Description

1 Generate grid and solution files for post-processing by Ensight. (evrzon, US_hdzwrt)

*2*

Obsolete. Replaced with the WRITE WALLDISTANCE keyword. Original description was as follows:

Write distance to the nearest wall for each grid point into the .cfl file as the variable WallD. This option applies to both structured and unstructured grids. (InitWallDst, walldfil)

*3*

Obsolete. Replaced with the RESIDUAL_OUTPUT keyword. Original description was as follows:

Residual computation. (lpschmu, TURB_goldbergUS, TURB_shihkeUS, TURB_spalartUS, TURB_sstUS)

	mode		Result
	0		Compute residuals based on ΔQ
	1		Compute residuals based on right-hand-side
	2		Compute residuals based on right-hand-side normalized by cell volume

4 With the HLLE differencing scheme, replace the anti-diffusive terms with Lax-like dissipation reduction. The size of the dissipation reduction term is scaled by mode / 100. (hlle, US_HLLE, US_HLLEpt, US_HLLERL)

7 Write the zonal maximum and minimum values of the dependent variables, and the maximum and minimum vertex value of the pressure, into the .lis file. (US_Qcell2Qvrtx)

8 Split the edge data (.cge) file into mode files. (edgopn)

9 Write extra arrays containing cell volume, cell centroid coordinates, face center coordinates, and cell connectivity into the .cge file for debugging purposes. (edgopn, prerdinpt, US_PreUnstructRead, US_PreUnstructWrite)

10 Write detailed grid information and wall distance into the .lis file. For wall distances, this option applies to both structured and unstructured grids. (GlobalWallDist, MkGlblWallLst, US_mklines US_PreUnstructWrite, US_StencilFill, US_ZnlWallDist)

11 Multiplying factor for viscous-diffusion Jacobian. If DEBUG 11 is not used, a value of 4.0 is used in TURB_goldbergfluxUS, and 2.0*Dissip_Jcb in US_FluxJcbnDiss, where Dissip_Jcb is the Jacobian dissipation parameter set using the LAX DAMPING keyword. (TURB_goldbergfluxUS, US_FluxJcbnDiss)

*12*

Removed. Original description was as follows:

Maximum ratio of production to destruction allowed in Goldberg turbulence model. (TURB_goldbergsrcUS)

*13*

Removed. Original description was as follows:

Maximum ratio of destruction to production allowed in Goldberg turbulence model. (TURB_goldbergsrcUS)

14 When updating turbulence model values, write maximum turbulent viscosity to .lis file. (TURB_UpdateGoldberg, TURB_UpdateSA)

15 Not applicable in Wind-US. (NSzsolv, US_unstrpstinp)

16 For multi-species flows, write thermodynamic data table into the file Thermo_Tables. This option applies to both structured and unstructured grids. (chpstin1, propin)

17 When writing flow solution to .cfl file, for vertices at coupled surfaces with multiple boundary conditions (e.g., at corners), make the "coupled" boundary condition the lowest priority. (US_Qcell2Qvrtx)

18 Neglect the diffusion Jacobian from the implicit operator in the unstructured Spalart-Allmaras turbulence model. (TURB_diffuseSA)

*19*

Obsolete. Replaced with the FIXER THRESHOLD keyword. Original description was as follows:

Threshold the solution values when a cell exceeds the specified min/max rather than performing an average with the surrounding cells. See the FIXER keyword. (US_AverageQ)

20 Not applicable in Wind-US. (GRD_UpdateMet)

22 Not applicable in Wind-US. (loadqbc, lpmgu, stagepstinp, US_unstrpstinp)

23 For the Rusanov (Lax) scheme, explicitly set the damping factor for strong shocks to 0.01 × mode. The default is to set this damping factor to 0.5 (i.e., equivalent to mode = 50). (US_Lax, US_LaxRL)

*24*

Removed. Original description was as follows:

The convergence criterion (see the CONVERGE LOAD keyword) is set to 1 / mode. The default convergence criterion is 0.01. (LoadCnvrg)

25 Prevents an error exit when both CFL# AUTOMATIC and a time step in seconds are specified in the same zone. Note though that this shouldn't happen, unless both CFL# AUTOMATIC and CFL MODE 1 or 2 are erroneously used in the same zone. (pstinp)

27 Not applicable in Wind-US. (US_unstrpstinp)

28 Not applicable in Wind-US. (US_unstrpstinp)

30 Output lines created for the IMPLICIT Gauss-Seidel line solver to a FieldView "particle path" file for visualization. This includes all the lines in the grid (i.e. those generated by cfpart and those generated internally in the solver). (US_mklines)

31 Don't stop execution when errors are found in cell volumes or face normals. (US_CheckPreInfo)

32 Minimum line length to be used when creating lines for line Gauss-Seidel solver. The default value is 10. (US_mklines)

Alters how the local time step is computed from the specified CFL number. When activated a constant time step is used along "lines" of an unstructured grid. (US_TimeStep)

	mode		Result
	<0		Use the average time step along the lines.
	>0		Use the minimum time step along the lines.

For a description of what lines are and how they are used, see the IMPLICIT UGAUSS LINE option.

Use an alternate procedure for computing the friction velocity u_τ for WALL FUNCTION boundary conditions.

	mode		Method
	0		Use the unstructured grid u_τ routine.
	≠0		Use the structured grid u_τ routine instead.

Note also that if mode<0, the near wall damping terms in the Spalart-Allmaras turbulence model will be disabled. This DEBUG option may only be used for a perfect gas. (TURB_sasrcUS, US_utau)

38 When implicit boundary conditions are being used (see IMPLICIT BOUNDARY), use explicit boundary conditions at outflow boundaries. (US_rlupw)

39 Compute distance to the nearest wall for each grid point the "old" way (i.e., on a zonal basis). This option applies to both structured and unstructured grids. (MkGlblWallLst, walldfil)

40 Activate the fictitious gas law Mach scaling as an alternative to using FIXER. (gas1, gas4)

*41*

Obsolete. Replaced with the IMPLICIT BOUNDARY ON keyword. Original description was as follows:

When implicit boundary conditions are being used (see IMPLICIT BOUNDARY), also treat coupled boundaries implicitly. Requires a parallel run with as many processors as zones, and with direct communication between workers (see the .mpc file communication command). (US_bcalloc, US_implCpl, US_ImplStencil, US_InitDq, US_lhsvis, US_rlupw, US_TransferDq, wbnd2)

*42*

Obsolete. Became the default setting in version 3.138. Original description was as follows:

Solve the Navier-Stokes and turbulence equation sets concurrently in one pass even though they are not coupled sets. This option applies to unstructured grids only. (lpgrp, lpmgu, lpns, lpschmu, setzon, TURB_sstUS, TURB_goldbergUS, TURB_shihkeUS, TURB_spalartUS, TURB_realkeUS, US_DqLimit)

43 Applies the slope limiter (see TVD) to the gradients used in the viscous operator (in addition to the required application in the inviscid flux). Since the form of the viscous operator has been changed to something less sensitive, this option may be removed in the future. (US_SlopeLimitVenkat)

45 Not applicable in Wind-US. (bcstdrd)

*46*

Obsolete. Replaced with the RHS VISCOUS keyword. Original description was as follows:

Viscous flux correction method. (TURB_diffuseSST, US_FaceConCA, US_FaceConFBnd, US_FaceGradGG)

	mode		Method
	0		No correction
	1		Original correction method
	2		Only use normal variation
	3		Only use normal component

The default is no correction if GRADIENTS LEAST_SQUARES was specified, and the original correction method otherwise.

47 Not applicable in Wind-US. (lpschmu, US_TimeStep)

48 A solution limiter, similar to that in the CFL3D code, which adjusts all of the flow variables (not just density and pressure) to maintain realizability. This option has seen limited success beyond the default mode. (US_DqLimit, US_GS_FullStore, US_LGS_FullStore, US_UpdateNS)

49 Use full linearization of turbulence model. In addition, if mode < 0, and the coefficient matrix is not diagonally dominant, stability will be enhanced by multiplying the relevant diagonal terms by 1.05. (TURB_lhsdt, TURB_sasrcUS)

50 Not applicable in Wind-US. (US_rhsvis)

51 For GRADIENTS LEAST_SQUARES, activate least squares based on face neighbors. (US_LSstencil)

53 For the Roe scheme, get interface flux using same routines as in structured solver. (This option is currently disabled, and will cause an abort if used in conjunction with RHS ROE.) (US_rhsupw, US_rlupw)

Used to alter the viscous wall boundary condition used with the SST TURBULENCE MODEL. The objective of this option is to reduce the model's near-wall grid sensitivity, but thus far only a marginal improvement has been obtained. (TURB_sstboundUS, TURB_SSTf3, TURB_SSTfixK, TURB_SSTmut, TURB_sstsrcUS, TURB_sstUS, US_tdbcg)

In the description that follows, note that the analytic smooth-wall asymptotic boundary condition as the wall normal distance, s→0 is given by: ρω→6μ_L/(β(Re*s)²). The quantity ds is the distance to the first point off the wall, and all equations are written in terms of non-dimensional variables.

	mode		Method
	0		Menter's original viscous wall boundary condition: ρω=106μ_L/(β(Reds)²) [Menter, F. R. (1994) "Two Equation Eddy-Viscosity Turbulence Models for Engineering Applications," AIAA Journal, Vol. 32, No. 8, pp.1598-1605]
	>0		Hellsten's modified viscous wall boundary condition: ρω=1.56μ_L/(β(Reds)²) [Hellsten, A. (1998a) "On the Solid Wall Boundary Condition of ω in the k-ω Type Turbulence Models," Report B-50, Helsinki University of Technology]
	-1		Hellsten's modified viscous wall boundary condition based upon a surface roughness formulation: ρω=(ρu_τ)² (S_R/μ_L)* S_R=(50/k_s⁺)² k_s⁺=2.4\|y⁺\|^0.85 [Hellsten, A. (1998b) "Some Improvements in Menter's k-ω SST Turbulence Model," AIAA Paper 1998-2554]
	-2		An omega wall function boundary condition for viscous walls: ρω= (g)ρω_vis + (1-g)ρω_log ρω_vis=6μ_L/(β(Reds)²)* ρω_log=ρu_τ/(0.41(β^)^0.5Reds) (u_τ)²=(g)(u_τ0)² + (1-g)(u_k)² (u_τ0)²=ν_Lu_tan /(Reds)* (u_k)²=(β^)^0.5k* g=exp(-Rey/11) Rey=(k)^0.5Reds/ν_L* [Knopp, T., Alrutz, T., and Schwambornm D. (2006) "A Grid and Flow Adaptive Wall-Function Method for RANS Turbulence Modelling," J. Computational Physics, Vol.220, pp.19-40] [Popovac, M. and Hanjalic, K. (2007) "Compound Wall Treatment for RANS Computation of Complex Turbulent Flows and Heat Transfer," Flow Turbulence and Combustion, Vol.78, pp.177-202] [Durbin, P. A. (2009) "Limiters and Wall Treatments in Applied Turbulence Modeling," Fluid Dynamics Research, Vol.41, pp.1-18] The production of turbulent kinetic energy in the near-wall region is also modified: P_k=ρ(u_τ)⁴Re/(0.41μ_Ly⁺) [Popovac, M. and Hanjalic, K. (2007) "Compound Wall Treatment for RANS Computation of Complex Turbulent Flows and Heat Transfer," Flow Turbulence and Combustion, Vol.78, pp.177-202]

This debug option simultaneously controls whether the eddy viscosity is computed using the strain rate rather than vorticity, which is not frame invariant.

	mode		Method
	0, >1		Use vorticity.
	<0, 1		Use strain rate.

55 Uses the turbulent shear stress tensor instead of vorticity in the production term of the SST TURBULENCE MODEL on unstructured grids. (TURB_sstsrcUS)

56 Not applicable in Wind-US.

57 Freeze the unstructured grid slope limiter after mode iterations. (US_SlopeLimit)

59 To improve convergence, bleed boundaries in unstructured grid zones are normally handled explicitly for the first 100 cycles, then included as part of the implicit operator thereafter. Using this debug option disables the implicit treatment of the bleed boundaries. (US_lhswall)

Applies the viscous operator to a known solution for the purpose of assessing accuracy of the unstructured solver via the method of manufactured solutions. (lpschmu, US_rhs_solve)

[Cary, A., Dorgan, A., and Mani, M. (2009) ``Towards Accurate Flow Predictions Using Unstructured Meshes,'' AIAA Paper 2009-3650.]

61 Write the unstructured solver linear system (subiteration) convergence to the list output (*.lis) file. (TURB_ILU, TURB_pntGS, US_GS_FullStore, US_ILU_FullStore)

62 Not applicable in Wind-US.

63 Not applicable in Wind-US.

Modifies how the local min/max in the solution is determined for the purpose of TVD VENKAT slope-limiting in the unstructured solver. (US_SlopeLimitVenkat)

	mode		Method
	0		Default mode, which only uses the face neighbors and is likely what is required for TVD.
	1		The node-neighbor mode, which includes all the cells that share a node with the cell of interest. This larger search space generally results in a larger acceptable range of solution values. The node-neighbor mode should allow second-order accuracy (i.e. a linear function will not be limited on an arbitrary mesh). However, it may not enforce TVD.

65 Write log of memory allocation/deallocation to the file memdebug.lis. This option applies to both structured and unstructured grids. (MemDebug, prerdinpt, solver_main)

66 Used to test the unstructured solver gradient calculator. (lpschmu, US_SetGradQLS)

67 Print cell data for testing the unstructured solver. (lpschmu)

68 Turn off the inviscid terms for the purpose of testing the viscous operator. This option applies to unstructured grids. (US_rhs_solve)

69 For the unstructured solver, generate lines based on Jacobian weight. (TURB_implUS, TURB_linearRes, US_impl)

70 For the unstructured solver, compute an L2-norm of the difference between left and right states over all cell faces for the Navier-Stokes equation set. Print this norm to the list output (.lis) file. (US_rhs_solve)

71 Not applicable in Wind-US.

72 In the unstructured solver, convert Roe flux to advection. (US_RoeRL)

73 In the unstructured solver, neglect the roll source terms. (US_add_rotate)

74 In the unstructured solver, use the exact supersonic corner flow to test the numerics. (lpschmu, sins, US_bcfil, US_exact_corner_soln)

75 In the unstructured solver, use a prescribed pressure profile to test the numerics (GRD_zsolv)

Bitmap used to activate the ADF memory buffer for common files. (openf, psinpt)

	bit		File
	1		Grid File (.cgd)
	2		Solution File (.cfl)
	3		Time Data File (.tda)

Bitmap that controls how the boundary flux in the unstructured solver will be computed. If the flags are 0, then the boundary flux is imposed by the boundary condition value. If it is non-zero, then the flux is computed using the chosen interior scheme with the boundary value and cell value forming the left and right states. (US_lhswall, US_rlupw)

	bit		Boundary Type
	1		Bleed
	2		Freestream, Arbitrary Inflow
	3		Outflow
	4		Frozen

78 Use the older (and significantly slower) octree test for computing wall distance. (GlobalWallDist)

80 Compute the wall distance by solving a partial differential equation. Only for evaluation purposes with the unstructured solver. (axturb, evrzon, evsolv)

Modify the behavior of unstructured grid cells that violate the GRID LIMITER ANGLE setting. (US_FaceGradGG, US_rhsvisFace, US_SetGradQ)

	mode		Method
	1		Zero the gradient in cells that violate the slope-limiter. This yields solutions that are locally first-order inviscid and thin-layer viscous.
	2		Use thin-layer viscous in cells with grid angles greater than 120 degrees.
	3		Use thin-layer viscous in cells where the normal is in the opposite direction to the vector between the cell-centers.
	4		Reynolds stress approximations drop back to eddy viscosity in cells with grid angles greater than 120 degrees.

Activate the Dacles-Mariani curvature correction to the Spalart-Allmaras TURBULENCE MODEL. This correction is a simpler alternative to the standard rotation term. It reduces the eddy viscosity in regions where vorticity exceeds strain rate, such as in vortex core regions where pure rotation should not produce turbulence. In thin shear layers, where vorticity and strain are very close, this correction should have minimal effect. (TURB_sasrcUS)

[Dacles-Mariani, J., Zilliac, G. G., Chow, J. S., and Bradshaw, P. (1995) "Numerical/Experimental Study of a Wingtip Vortex in the Near Field," AIAA Journal, Vol. 33, No. 9, pp. 1561-1568]

83 Apply the cross-diffusion limiter of Langtry to the unstructured grid SST TURBULENCE MODEL. For evaluation by developers only. (TURB_sstboundUS, TURB_sstsrcUS)

84 Use a second-order advective flux with the unstructured SST TURBULENCE MODEL. (TURB_sstfluxUS)

85 Not applicable in Wind-US.

86 Revert to the old cell to vertex calculation for unstructured solutions. This refers to the creation of the cfl (node-based solution) from the tda (cell-based solution). Many improvements have been made to which variables are used and which boundaries have preference, in order to create "better looking" cfl files. (US_Qcell2Qvrtx)

87 Not applicable in Wind-US.

89 Limit the maximum orthogonal angle deviation for unstructured grid wall cells. The mode value is the maximum allowed angle (in degrees) between the face-cell center line and face normal. When this limit is exceeded, the face-center is moved to lie on the face-normal vector that passes through the cell-center. (evrzon)

90 Not applicable in Wind-US.

107 Compute the cell centroid for unstructured grids from the average of the nodal coordinates.

125 Not applicable in Wind-US.

200

This option exercises the Liu-Vinokur equilibrium air CHEMISTRY tables, generates some Tecplot files, and then quits. It is used to compare the internal tables to the plots published by Liu & Vinokur and allows one to look at the behavior when conditions depart the tables. (liu3, solver_main, lvtest)

[Liu, Y. and Vinokur, M. (1989) "Equilibrium Gas Flow Computations. I. Accurate and Efficient Calculation of Equilbrium Gas Properties," AIAA Paper 89-1736]

Last updated 1 Apr 2016