Wind-US 3.0 --> Wind-US 4.0 Changes

These notes describe changes up to and including version 4.108.

The keyword links below point to the Wind-US 3.0 documentation. The 4.0 documentation will be available soon.

• Wind-US Code
  • Physical Modeling
  • Turbulence Modeling
  • Chemistry
  • Boundary Conditions
  • Numerics
  • Parallel Processing
  • Miscellaneous
  • Keywords
  • TEST Options
  • DEBUG Options
  • Bug Fixes
  • Non-Backward Compatibility
  • Code Structure
• Makefiles
• Scripts
• CFPOST
• MADCAP
• Utilities
• libcfd
• ADF
• PVM
• Known Bugs
• Feature Requests

Many of the changes from Wind-US 3.0 to 4.0 were enhancements to the thermal models, multi-species models, and turbulence models.

The changes in Wind-US, and in the associated tools and utilities, are described in more detail in the following sections. Changes between previous releases are also available, as follows:

• Wind-US 2.0 --> Wind-US 3.0 Changes
• Wind-US 1.0 --> Wind-US 2.0 Changes
• Wind 5.0 --> Wind-US 1.0 Changes
• Wind 4.0 --> Wind 5.0 Changes
• Wind 3.0 --> Wind 4.0 Changes
• Wind 2.0 --> Wind 3.0 Changes
• Wind 1.0 --> Wind 2.0 Changes

Wind-US Code

The following changes have been made to Wind-US. A number in parentheses following the description (e.g., 4.xxx) is the specific IVMS version in which the change was introduced.

Physical Modeling

• Several variable turbulent Prandtl number models are now available for use on structured grids. (4.107)

Turbulence Modeling

• The Menter SST k-ω turbulence model stress limiter coefficient can now be adjusted by the user. (4.107)
• The Menter SST k-ω turbulence model has been extended to optionally include net vorticity transport (NVT) modifications as described in AIAA Paper 2014-2086. See the descriptions of test options 34, 38, 39, 41, 44, 45 below. (4.84)
• The Menter BSL k-ω turbulence model has been added for both structured and unstructured grids. (4.107)
• The turbulence values on inflow boundaries are now enforced every cycle when using the Menter BSL and SST k-ω turbulence models. Previously, these boundaries were initialized this way but then assumed constant during subsequent cycles. Transient outflow along such an inflow plane could contaminate the boundary with values from the interior. (4.80)
• Several k-ε turbulence models have been added for structured grids: (4.107)
  • Nagano-Tagawa-Tsuji (NTT)
  • Abe-Kondoh-Nagano (AKN)
  • So-Zhang-Speziale (SZS)
  • So-Sarkar-Gerodimos-Zhang (SSGZ)
  • Craft-Tech
• New optional source term corrections are available for use with any of the structured grid k-ε turbulence models. (4.107)
  • Pope vortex stretching correction.
  • Chen-Kim extra strain correction.
• New compressibility correction options are available for use with any of the structured grid k-ε turbulence models. (4.107)
  • Aupoix
  • Craft
  • Suzen
  The format of the user-specified compressibility correction has changed.
• The k-ε linearization option has been removed. All structured grid k-ε models now use an approximate linearization, which was the previous default. (4.107)

Chemistry

• The .chm chemistry file is now parsed using some of the same routines used for the .dat file. This should make the code less picky about exact spacing and case. (4.44,4.75)
• Implicit boundaries may be used with the CHEMISTRY FINITE RATE FIXED mode. (4.44)
• TEST 76 - Use air C_p for implicit chemistry terms. (4.44)
• Added N2 real gas capability via the CHEMISTRY THERMO9 keyword. (4.45)
• Added ability to write total enthalpy to the .cfl file via the WRITE TOTALS keyword. (4.45)
• The initialization/reinitialization process now tries to make use of more known information to allow restarts from runs with any chemisty model to any other chemistry model, including real gas. (4.45)
• Multi-phase capability has been added, including condensation, evaopration, and agglomeration. These features are available for testing and validation only, not production use. Contact NPARC support for the current keyword documentation. (4.59,4.98,4.99)
• For finite rate chemistry, added the ability to turn off reactions in specified zones, but still compute the convection and mixing. See the NOREACT keyword below. (4.68)
• The point implicit method can now be used with chemistry. (4.97)
• For TEST 92 < 0, perform point implicit integration for approximate chemistry source terms where the number of subiterations is |test(92)|. (4.97)
• A variable turbulent Schmidt number capability is now available for use on structured and unstructured grids. The model is merely that of a constant zonal value. (4.107)
• Added enhancements to support the latest Thermodynamics Application Programming Interface (TAPI). (4.87) This capability remains experimental. The new keywords listed in chem/chinpt.f90 are:

  TAPI tapi_engine_name [tapi_engine_input]
    [ SPECIES { AIR | spnam1 spfrac1 ... spnamN spfracN } ]
    [ FIXED_ALL ]
    [ FIXED [zone_selector] ]
    [ REACTING [zone_selector] ]
  

• Added VISCOSITY table lookup capability for C02 gas.

Boundary Conditions

• Implicit boundaries may be used with the CHEMISTRY FINITE RATE FIXED mode. (4.44)
• Two new methods for specifying wall temperature are available. These include the WALL TEMPERATURE FIXED keyword and TEST 33, both of which are described in greater detail in the sections below. The old TTSPEC keyword may still be used to specify wall temperature. However, this method has the limitation that the TTSPEC data does not get split when decomposing the domain. (4.49,4.58)
• The turbulence values on inflow boundaries are now enforced every cycle when using the Menter BSL and SST k-ω turbulence models. Previously, these boundaries were initialized this way but then assumed constant during subsequent cycles. Transient outflow along such an inflow plane could contaminate the boundary with values from the interior. (4.80)
• Improvements to the screen boundary condition for high solidity screens when near choke (high loss). (4.81,4.92)
• For arbitrary inflows, a HOLD_FLOWRATE option has been added. This option will hold the momentum and temperature at the inflow to the values already in the solution file. Not tested for USURFACE boundaries. (4.55,4.72)
• For arbitrary inflows, a HOLD_MASS option has been added. This option will add a total pressure increment to the inflow boundary and iterate it until the desired mass flow is obtained. Not implemented for USURFACE boundaries. (4.83,4.94)

Parallel Processing

• The section on parallel processing has been updated to reflect how the code currently operates, especially with regards to MPI.

Numerics

• The DQ LIMITER NXAIR algorithm was modified to work with any gas model and to use factors other than 0.5. This limiter seeks to ensure that: density remains positive, mass fractions remain non-negative, the total energy (without heat of formation) remains positive, and the velocity is not allowed to grow so large that the sensible energy becomes negative. The advantage of the NXAIR limiter is that, if only a few points are having trouble within a block, one does not have to slow down the solution in the entire block to keep things going. (4.76)
• Several Runge-Kutta schemes are now available. For a complete list, see the STAGES keyword below. (4.41)

Miscellaneous

• Increased support for up to 9999 zones. (4.61)

Keywords

The following keywords have been added, modified, or deleted. For more information, see the more detailed list of changes below, or the Keyword Reference section in the Wind-US User's Guide

• Special Characters:
  • A continuation character '\' may be added at the end of lines in the .dat file in order to break up a single keyword entry over multiple lines. Note that '/' or '!' at the beginning of lines still denote comments and '!' may be used for trailing comments. (4.38)
  
  
• New keywords:
  • BC [LIMITER] [OFF | ON ] [PRINT ON | OFF]
    This keyword controls whether the DQ LIMITER is applied to coupled boundaries in structured zones. The default setting for the bc limiter is OFF. When activated, the default PRINT setting is ON. (4.102)
  • K-E SOURCE { OFF | NONE | POPE [COEF pope_value] | CHEN-KIM [COEF chenkim_value] }
    This keyword is used to control optional source term corrections for all structured k-ε turbulence models. Default setting for the source term is OFF. The default value for the Pope coefficient is 0.79, and the default value for the Chen-Kim coefficient is 0.06. For compressible flows, each term is multiplied by an exponential damping function with coefficient β, which is specified through the K-E COMPRESSIBILITY keyword. (4.107)
  • SST STRESS [LIMITER] COEFFICIENT value
    This keyword can be used to adjust the stress limiter coefficient used by the SST turbulence model. It limits the ratio of shear stress to turbulent kinetic energy to the value prescribed. This coefficient is applied to all zones. The default value is 0.31, sometimes referred to as Bradshaw's coefficient. (4.107)
  • VARIABLE TURBULENT PRANDTL NUMBER ...
    This keyword can be used to activate the variable turbulent Prandtl number capability for structured grids. The available models include constant zonal values, algebraic models, and differential models based on the transport of a thermal scalar variance. The differential models require use of a two-equation turbulence model, while the other Prandtl models will work with any turbulence model. The default setting is to use the freestream turbulent Prandtl number, as specified by the PRANDTL keyword. (4.107)
  • VARIABLE TURBULENT SCHMIDT NUMBER ...
    This keyword can be used to activate the variable turbulent Schmidt number capability for both structured and unstructured grids. The current model is merely that of a constant zonal value. The default setting is to use the freestream turbulent Schmidt number, as specified by the SCHMIDT keyword. (4.107)
  
  
• Modified keywords:
  • ARBITRARY INFLOW
    • HOLD_FLOWRATE can now be used to specify an inflow flow rate. This method will hold the momentum and temperature at the inflow to the values already in the solution file. Keyword must appear on a line before the ZONE keyword. Not tested for USURFACE boundaries. (4.55,4.72)
    • HOLD_MASS value can now be used to specify an inflow flow rate [lbm/s]. This option will add a total pressure increment to the inflow boundary and iterate it until the desired mass flow is obtained. The frequency with which the total pressure increment is adjusted is the same as that specified for outflow MASS FLOW conditions. Keyword must appear on a line before the ZONE keyword. Not implemented for USURFACE boundaries. (4.83,4.94)
    • HOLD_CURRENT will compute characteristics based on the current (frozen) inflow values. This feature was actually added to version 3. Keyword must appear on a line after the ZONE keyword. (3.159)
    • Note that HOLD_FLOWRATE and HOLD_MASS are control functions, like HOLD_TOTALS. These must appear on a line before the ZONE keyword.
    • Note that the ARBITRARY INFLOW FREESTREAM documentation should be updated to include HOLD_FLOWRATE and HOLD_MASS.
    • Note that HOLD_CURRENT is a flow condition specification like UNIFORM. It must appear on a line after the ZONE keyword.
  • CHEMISTRY
    • FINITE RATE
      • NOREACT zone_selector is a new option to turn off reactions in specified zones, but still compute the convection and mixing. The zone selector is required. (4.68)
  • DQ LIMITER syntax has changed to:

    DQ [LIMITER] [OFF | [ON | NXAIR | POINTWISE]
                        [DRMAX drmax] [DTMAX dtmax] [DPMAX dpmax]
                        [RELAX dqrlx] [PRINT ON | OFF]]
    

    POINTWISE is the new and more descriptive name for NXAIR. PRINT is a new option to enable/disable the printing of limiter messages. The default is still to print the messages.
    The default setting of the coupled boundary limiter has changed and is now OFF. It can be activated with the BC LIMITER keyword, described above. (4.102)
    See additional remarks on changes to the numerics above.
  • K-E COMPRESSIBILITY syntax has changed to:

    K-E COMPRESSIBILITY { OFF | NONE | SARKAR | WILCOX |
                          CRAFT | AUPOIX | SUZEN | USER coef value }
    

    where coef is one of the following keywords

        { ALPHA1 | ALPHA2 | ALPHA3 | ALPHA4 |
          BETA_POPE | BETA_CHEN-KIM |
          MT01 | MT02 | AUPOIX1 | AUPOIX2 }
    

    Multiple coefficients may be specified by repeating coef value pairs after the initial USER option. The predefined coefficient sets are as follows:

      OFF|NONE = ( All coefs 0.0 )
      SARKAR   = ( ALPHA1  1.0 )
      WILCOX   = ( ALPHA1  1.5 MT02    0.25 )
      CRAFT    = ( ALPHA1  2.5 ALPHA2  2.5  ALPHA4 5.0 BETA_POPE 75.0 MT01 0.20 )
      SUZEN    = ( ALPHA1  1.0 ALPHA2  0.4  ALPHA3 0.2 )
      AUPOIX   = ( AUPOIX1 0.9 AUPOIX2 0.32 )
    

    Or in tabular form as:

    Coef	OFF	SARKAR	WILCOX	CRAFT	SUZEN	AUPOIX
    ALPHA1	0.00	1.00	1.50	2.50	1.00	0.00
    ALPHA2	0.00	0.00	0.00	2.50	0.40	0.00
    ALPHA3	0.00	0.00	0.00	0.00	0.20	0.00
    ALPHA4	0.00	0.00	0.00	5.00	0.00	0.00
    BETA_POPE	0.00	0.00	0.00	75.00	0.00	0.00
    BETA_CHEN-KIM	0.00	0.00	0.00	0.00	0.00	0.00
    MT01	0.00	0.00	0.00	0.20	0.00	0.00
    MT02	0.00	0.00	0.25	0.00	0.00	0.00
    AUPOIX1	0.00	0.00	0.00	0.00	0.00	0.90
    AUPOIX2	0.00	0.00	0.00	0.00	0.00	0.32

    The SARKAR and WILCOX options are the same as before and NONE remains the default setting. CRAFT, SUZEN, and AUPOIX are new, and the USER syntax has changed. The SUZEN settings are the same as those used in the compressible SST model. Note that the USER option only adjusts the individual coefficients specified. Thus it may be used after loading one of the predefined sets in order to make minor adjustments. The coefficients used will be printed in the .lis file. (4.107)
  • STAGES, which controls the Runge-Kutta schemes now has the following options: (4.41)

    mode	Stages	Order	Description
    0	1	1	Backward Euler method
    1	4	2	Jameson style low-storage Runge-Kutta
    2	6	3	Williamson's low-storage Runge-Kutta
    3	2	3	Van Leer-Tai-Powell optimal smoothing method
    4	3	3	Van Leer-Tai-Powell optimal smoothing method
    5	3	3	Wray's minimal storage Runge-Kutta
    6	9	4	Wray's minimal storage Runge-Kutta
    7	2	2	???-style Runge-Kutta

    mode	#Coefs	Default Coefs
    0	0	n/a
    1	4	1/4, 1/3, 1/2, 1
    2	6	0, 1/3, -5/9, 15/16, -153/128, 8/15
    3	2	1/3, 1
    4	3	0.1481, 0.40, 1
    5	5	1/4, 8/15, 0, 5/12, 3/4
    6	9	0.13777497, 0.41883005, 0.38832895, 0.88635766, 0.14633970, -1.0364937, 0.0037218696, -0.087624810, 0.32383451
    7	2	1/2, 1

  • MASS FLOW FREQUENCY parameter now also controls how often the ARBITRARY INFLOW HOLD_MASS condition is applied. No changes to keywords syntax.
  • TURBULENCE MODEL now supports the following new models:

      {BSL  | MENTER-BSL}                   [k-w]
      {NTT  | NAGANO-TAGAWA-TSUJI}          [k-e]
      {AKN  | ABE-KONDOH-NAGANO}            [k-e]
      {SZS  | SO-ZHANG-SPEZIALE}            [k-e]
      {SSGZ | SO-SARKAR-GERODIMOS-SPEZIALE} [k-e]
      {CRAFT-TECH}                          [k-e]
    

    The BSL k-ω model works with structured and unstructured grids. The other new k-ε models only work with structured grids. (4.107)
  • WALL TEMPERATURE FIXED will attempt to hold the temperature on a viscous wall at the current value. Do to the implementation approach, the value might drift a little. TEST 33 also holds the wall temperature constant, albeit using the conjugate heat transfer model. The older TTSPEC keyword is yet another means for specifying wall temperature. However, this method has the limitation that the TTSPEC data does not get split when decomposing the domain. (4.49,4.58)
  • WRITE will now output the following additional variables:
    • TOTALS outputs total pressure, total temperature, and total enthalpy. In the .cfl file, the variable names are Pt-frozen, Tt-frozen, and Ht-frozen, respectively. (4.45)
    • PITOT outputs pitot pressure (Pp-frozen). This option only works under the TAPI option for chemistry. (4.53,4.70)
    • PRANDTL outputs turbulent Prandtl number (Prt). (4.107)
  • VISCOSITY CO2 now provides table lookup capability for CO2 gas. Requires the use of additional data files.
  
  
• Deleted keywords:
  • The K-E LINEARIZATION keyword option has been removed. All structured k-ε models use an approximate linearization, which was the previous default. (4.107)

TEST Options

The following TEST options have been added, modified, or deleted. See the Test Options section for more detailed descriptions.

• New options
  
  
  • 33 - Hold the temperature on a viscous wall at the current value, using the conjugate heat transfer model. See also the WALL TEMPERATURE FIXED keyword. (4.49)
  • 34 - Coefficient for the streamwise vorticity term when using the SST turbulence model with the net vorticity transport (NVT) option. For details, see the description for TEST 44 below. (4.84)
  • 38 - Coefficient for the streamwise velocity gradient term when using the SST turbulence model with the net vorticity transport (NVT) option. For details, see the description for TEST 44 below. (4.84)
  • 39 - Used with the BSL and SST turbulence models to specify freestream sustainment terms. (4.80,4.84)

    mode	Result
    0	Sustainment terms are not used.
    1	Sustainment terms are used at all locations to preserve the freestream turbulence values, (akfree,aefree)=(k,ω)∞. These values are set using the FREE_K and FREE_OMEGA keywords.
    >1	Sustainment terms are used at all locations. However two different levels of turbulence may be preserved. In regions where Tt<1.3*Tt∞, freestream turbulence values are sustained. In regions where Tt>1.5*Tt∞, (e.g., jet plumes) engine turbulence values are sustained instead. In intermediate regions, the sustainment terms are linearly interpolated based upon the local total temperature. The engine turbulence values are computed by treating mode as one thousand times the intensity of turbulent kinetic energy relative to the freestream velocity.     akeng = 1.5*(0.01*(mode/1000)*U∞)2     aeeng = aefree Thus, using mode 1000 would result in a turbulence intensity of one percent.

  • 41 - Coefficient to weight the density gradient term when using the SST turbulence model with the net vorticity transport (NVT) option. For details, see the description for TEST 44 below. (4.84)
  • 43 - Specify an alternate value for σ_ω2, the diffusion coefficient used in the ω transport equation of the BSL and SST turbulence models. (4.84)

    mode	Result
    0	Use the default value of 0.856.
    n	Use a value of 0.001*n.

  • 44 - Controlls modifications of the turbulent kinetic energy source term (P_k=ν_t*Γ²) in the outer layer of the SST turbulence model. These changes are referred to as Net Vorticity Transport (NVT) modifications and are described in AIAA Paper 2014-2086. They are intended to incorporate sensitivity to thermal shear layers by the inclusion of density gradient terms. (4.84)

    In order to understand what this option does, the following definitions are needed:

    Variable	Description
    Ω=∇×V	Vorticity vector
    Ωs=(V/|V|)·Ω	Vorticity magnitude in streamwise direction. if TEST(34)≠0, then multiply Ωs by 0.001*TEST(34).
    D=(1/ρ)∇ρ×V	Cross product of density gradient and velocity vector. if TEST(41)≠0, then multiply D by 0.001*TEST(41).
    ududs=V·(du/dx,dv/dy,dwdz) /         sqrt[|V|2*|(du/dx,dv/dy,dwdz)|2]	A pseudo-streamwise velocity gradient term (though not a proper vector quantity). if TEST(38)≠0, then multiply ududs by 0.001*TEST(38).

    The resulting effect on the turbulence production can then be described as:

    mode	Result
    0	Use default SST Pk formulation, based on vorticity: Γ=|Ω|.
    1	Use modified SST Pk formulation, based on the vector sum of Ω and D: Γ=|Ω+D|. Note that Ω+D=(1/ρ)*∇×(ρV) is related to the curl of momentum.
    2	Use modified SST Pk formulation, based on the stress tensor and momentum gradient. Γ2=[(1/ρμt)τijd(ρui)/dxj].
    3	Use modified SST Pk formulation, based on the amplitude sum of vorticity and density gradient term: Γ2=|Ω|2+|D|2.
    4	Reserved for future use. Use modified SST Pk formulation, based on the amplitude sum of vorticity and density gradient term, and scaled by the streamwise velocity gradient term: Γ2=(|Ω|2+|D|2)*(1+ududs).
    5	Reserved for future use. Use modified SST Pk formulation, based on the amplitude sum of vorticity and density gradient term, less the streamwise vorticity, and scaled by the streamwise velocity gradient term: Γ2=(|Ω|2+|D|2-(Ωs)2)*(1+ududs). This value is forced to remain non-negative.

    Note that the above correction is only applied in the outer layer of the model. In the inner layer, the production term is computed from vorticity: Γ=|Ω|. The same F1 blending function used to switch model coefficients between inner and outer layer values is used to switch between the inner and outer layer Γ values.
  • 45 - Activates some limiters recommended by Menter after the initial development of the the BSL and SST turbulence models. (4.84)

    mode	Result
    0	Deactivate these extra limiters.
    1	Limit the production of ω to be less than 20 times the dissipation of ω. Also limit the magnitude of the cross-diffusion term to 10 times the dissipation of ω while maintaining the sign.

  • 76 - Use air C_p for implicit chemistry terms. (4.44)
  
  
• Modified options
  
  
  • 92 - Added a new mode. For test(92)<0, perform point implicit integration for approximate chemistry source terms where the number of subiterations is |test(92)|. (4.97)
  • 188 - Added new mode 2. Controls how structured-grid SA/SST/BSL turbulence model data is passed across coupled zones. (4.110)

    mode	Result
    0	Passing of turbulence variables across the boundary is weighted by the fraction of boundary-normal velocity to the total magnitude.
    1	Passing of turbulence variables across the boundary is simply based on whether the flow is in/out of domain. No velocity weighting is done.
    2	Turbulence variables on coupled boundaries remain frozen at current values.

  
  
• Deleted options
  
  
  • No changes to report.
  
  

DEBUG Options

The following DEBUG options have been added, modified, or deleted. See the description of the DEBUG keyword for more detailed descriptions.

• New options
  
  
  • No changes to report.
  
  
• Modified options
  
  
  • No changes to report.
  
  
• Deleted options
  
  
  • No changes to report.
  
  

Bug Fixes

• ARBITRARY INFLOW
  • HOLD_TOTALS or HOLD_CHARACTERISTICS
    • For input files missing the underscore in HOLD_TOTALS or HOLD_CHARACTERISTICS, the parser would match HOLD to the first available keyword. Improved the robustness of the parser to prevent this from happening. (4.109)
  • XYZ_RANGE or RTZ_RANGE
    • The error message for bad min/max box was printing the same min/max values for all three coordinate directions. (4.109)
    • Failed to check if the number of ranges exceeds MXBCRG. Data could overwrite adjacent memory leading to random errors. (4.109)
  • Specified k-ω turbulence values along structured-grid ARBITRARY INFLOW boundaries (either UNIFORM, IJK_RANGE, XYZ_RANGE or RTZ_RANGE) was broken with update 4.80. The volume was still initialized properly, but the boundary applied freestream turbulence. Now, the proper inflow turbulence is applied for the given inflow type. (4.109)
• CHEMISTRY
  • EQUILIBRIUM LIU-VINOKUR
    • Values on freestream boundaries were improperly computed. (4.51)
    • In some cases the model was not gracefully handling the temperature going outside the tabulated data. (4.63)
  • DIFFUSION EFFECTIVE-BINARY
    • A constant used in the chemistry routines to compute the effective binary difusion was found to have the wrong value. (4.66)
  • TEST 69
    • Changes were made to the species diffusion terms to ensure mass conservation when solving all of the species conservation equations with TEST 69 nonzero. (4.66)
  • TEST 71 5
    • The calculation of entropy for TEST 71 5 was modified to be consistent with the specific heat (C_p) and static enthalpy (h) for temperatures below the curve fit minimum temperature. (4.67)
  • Wilke's mixture transport properties failed if any viscosity was zero (e.g., electrons). (4.69)
  • The relaxation factor used for iterating the total to static conversion for chemistry cases was changed so that the iteration loop could reach convergence for higher Mach number conditions. (4.88)
  • The global reaction rate is now limited to prevent negative concentration and improve robustness. (4.89)
  • A possible numerical overflow could result when any of the species mass fractions was zero for reacting flows. (4.91)
• DQ LIMITER
  • Changes in the treatment of the near wall terms in 4.108 to improve mass conservation for bleed and blowing caused ITAC Test Case 4 (NXAir Limiter) to fail, but could be overcome by lowering the CFL number. These terms have been reformulated to be more robust during startup transients. (4.110)
• HISTORY
  • Use of sequenced structured grids in zones using the time history command was disabled because it resulted in an immediate core dump when reading the grid. (4.109)
• LOADS
  • Print frequency for the LOADS command was not working as expected. It now uses the iteration count to determine whether to print. (4.22)
  • LOADS output for the massflow of individual species, reported in the .lis file as "mdot of [spec_name], ILOAD: [mdot_spec] [iload]" was incorrectly computed as a vector magnitude rather than a linear sum of the mass fluxes in the x-, y-, and z-directions. (4.90)
  • LOADS output for the massflow of individual species, reported in the .lis file as "mdot of [spec_name], ILOAD: [mdot_spec] [iload]" was missing 1 speci (with test 69 0) with structured grids. (4.109)
• SCREEN
  • Fixed the internal energy limiter that should only have been active for screens. It was missing some heat of formation terms. (4.86)
• SOLVER-STAGES
  • The ability to change structured grid sequencing was disabled because it resulted in a memory allocation error. (4.109)
• TURBULENCE
  • The SST turbulence values on inflow boundaries were only set during initialization, and then assumed constant during subsequent cycles. Transient outflow along such an inflow plane could contaminate the boundary with values from the interior. Boundary values are now enforced every cycle. (4.80)
  • The SST transition model has been corrected for a deficiency in the diffusion terms near the centerline of an axisymmetric jet that inhibited the propagation of turbulent kinetic energy to the axis on structured grids. The baseline and compressible SST models were fixed in (3.142), but somehow the transition model was overlooked. (4.104)
• WALL TEMPERATURE EQUILIBRIUM [EMISSIVITY]
  • Values entered for the emissivity must be in the range [0.0;1.0]. (4.107)
• VISCOSITY
  • Sutherland viscosity values on unstructured inflow boundaries were significantly different than the interior. Other flow properties like temperature were ok, and Wilke viscosity worked fine. The value stored in the boundary array should not have affected solution accuracy, but looked bad when visualizing the results. (4.109)
• Misc
  • Unintialized pointers were being passed to the ASSOCIATED intrinsic, which is illegal in Fortran. Unintialized pointers were also being passed to ALLOC which also checks the pointer status via the ASSOCIATED intrinsic. All such pointers are now nullified before use. (4.109)

Non-Backward Compatibility

A few changes are not backward-compatible. I.e., some input data (.dat) files and multi-processing control (.mpc) files used with Wind-US 3.0 may need to be changed for use with Wind-US 4.0. The known non-backward-compatible changes are listed briefly below.

• The format for the USER option to the K-E COMPRESSIBILITY CORRECTION keyword has changed. (4.107)

Code Structure

• No changes to report.

Makefiles

• Some makefiles have been updated and others added for newer system architectures. (4.6,4.57,4.71,4.73,4.77,4.104)
• The make wipe_clean option in the tools makefile deleted the tools-dev/bin/execselect.pm script needed by some of the tools. It is now retained. (1.45)

Scripts

• The default remote shell mode has changed from rsh to ssh. (2.0.96)
• The default multi-processor mode has changed from PVM to MPI. If MPI is not compiled, then PVM will be the default. (2.0.96)
• The wind script that launches Wind-US now checks your .dat file for syntax errors before the run begins. This is especially useful when submitting to a queue. (2.0.96)
• If MP mode is set to MPI, the scripts now make sure the executable on the master is compiled with MPI before starting the job. Otherwise the code will issue an "MPI unable to initialize, running in serial mode" error during startup and proceed in serial mode. (2.0.99) When compiling the code with MPI, make sure to do the following:
  • Set USE_MPI=YES in Makefile.configure.
  • Set MPILIBS in the appropriate Makefile.include.SYSTEM.SYSTEM_CPU.* file.
• The Linux PBS job script that is created used bash instead of the user's shell. The Wind-US environment variables are not loaded if the user has a different default shell. The script now runs in the user's default shell. (2.0.99)
• The entire section on scripts needs to be updated to reflect how the code currently operates, especially with regards to new command line options.

CFPOST

The following changes have been made to CFPOST. The features below are available in version 4.23.

• Added the following new variables:

  variable	definition
  THETAinc	Incompressible BL momentum thickness
  Mt	Turbulent Mach Number, sqrt(2*k)/a
  k+	Non-Dim Turbulent Kinetic Energy, k/utau^2
  epsilon+	Non-Dim Dissipation Rate, epsilon*mul/(rho*utau^4)
  turbtime	Turbulent time scale, k/epsilon
  uu ,vv ,ww ,uv ,uw ,vw	Reynolds Stresses
  uu+,vv+,ww+,uv+,uw+,vw+	Reynolds Stresses / utau^2
  S	Strain Rate Tensor magnitude, sqrt(2*Sij*Sij)
  Sxx,Sxy,Sxz	Strain Rate Tensor components
  Syx,Syy,Syz	Strain Rate Tensor components
  Szx,Szy,Szz	Strain Rate Tensor components
  Wxx,Wxy,Wxz	Rotation Rate Tensor components
  Wyx,Wyy,Wyz	Rotation Rate Tensor components
  Wzx,Wzy,Wzz	Rotation Rate Tensor components

  Note that the magnitude of the rotation rate tensor, sqrt(2*Wij*Wij), is the same as the vorticity magnitude, W=sqrt(wx**2+wy**2+wz**2).
• Variables may now be negated by preceeding the name by a minus sign (ie, -y, -u, -uv, -rho*uv). Should work for all variables. The minus sign is applied in SI units before any optional ORIGIN, SCALE, unit conversion, or normalization.
• Added operators SINH, COSH, TANH to CALCULATE FUNCTION.
• Fixed bad unit conversion for rho*k and rho*epsi.

MADCAP

• No changes to report.

Utilities

• The cfaverage utility was incorrectly computing RMS values when using the two-step ACCUMULATE and FINALIZE process with AFILE. AFILE should now only be used during FINALIZE. The new JOIN mode can be used to combine accumulated files together. Comment lines may now begin with [/!#] in column 1. Previously only supported [/]. Trailing comments are still not supported. (1.13)
• The cfcnvt utility now allows Overflow files to be either single or double precision. (1.60)
• The cfpart utility will now attempt to repair errors in the unstructured grid header information output by Pointwise. (1.12) Pointwise corrected this bug (SPR 17249) in version 17.3R1.
• The resplt utility now defaults to the Perl version of the utility. To use the old Fortran version, run the following command:

  resplt -exe
  

  The Fortran version is no longer being updated, and is already missing many features available in the Perl version. There are some differences in whitespace in the generated output files. (1.15)
• The Madcap production library was modified to allow up to 9999 zones. (2.26)
• The Tecplot Loader was modified to read the IBLANK variable from the grid file to use for variable blanking. (1.5)

libcfd

• Output the variable name when reporting problems reading variables. (2.179)
• Updated CFUNIT routine to handle the units of new flow variables used by Wind-US. (2.180)

ADF

• No changes to report.

PVM

• No changes to report.

Known Bugs

• MPI
  • The SPAWN command only seems to work intermittently within the same run when using SGI MPI. For the cycles where it fails, it reports "Warning: WINDUS: Spawn: Unable to spawn process." Tested on NAS using comp-intel/2013.5.192 with mpi-sgi/mpt.2.11r13. Appears to be a bug with the SGI MPT. Intel MPI works fine. (GRC, GRC)✔
• Wind-US
  • LOADS output for unstructured grids prints 0 mass flow for all species. There is currently no logic to support integrating species mass flow on unstructured grids. (GRC, unassigned) This feature will be reconsidered in the next release.
  • Certain old .cfl solution files are not working correctly. (ITAC, unassigned) Need an example.
  • Coupling and transfer of flow data between a hybrid mix of structured and unstructured zones has not been validated. Use at your own risk. (Boeing, unassigned) MADCAP is the only tool available for establishing the coupling information, and it crashes when updating the grid file. This feature will be reconsidered in the next release.
  • ARBITRARY INFLOW RTZ_RANGE for an arbitrary axis will initialize the full 360 degrees regardless of the theta values entered. This is a limitation carried over from version 3.0. The .lis file does report a warning message. (GRC, unassigned)
  • ARBITRARY INFLOW XYZ_RANGE and RTZ_RANGE do not echo inputs to the .lis file. This limitation is carried over from version 3.0. Would this be useful or simply clutter the .lis file? (GRC, unassigned)
• Pointwise
  • The zone header data output to unstructured Wind-US grid files contains an incorrect number of faces. See cfpart bugs below. (GRC, Fixed in Pointwise 17.2R3C4)
  • The bleed boundary condition data written to unstructured grid files is a single srfbc number, for example, 151 rather than point-by-point values containing the Wind-US bleed BC value of Zone_c=15 and a separate bleed region number of BndryNo_c=1. Structured grids work fine. Must use MADCAP (not GMAN) to reset these "unknown" boundaries. (GRC, Reported to Pointwise as Case ID: 74-16743)
  • Pointwise exports unstructured grids in meters, regardless of the units specified. Use GMAN to check and/or reset the grid units. (GRC, Pointwise Bug PW-19113)
• GMAN
  • GMAN should NOT be used to couple to unstructured grid zones. It does not output the proper point-by-point boundary data structures and will corrupt your grid file. Use MADCAP instead. (GRC/Boeing, unassigned)
  • GMAN should NOT be used to set bleed boundaries in unstructured grid zones. It does not output the proper point-by-point boundary data structures and will corrupt your grid file. Use MADCAP instead. (GRC/Boeing, unassigned)
• MADCAP
  • MADCAP consistently crashes when updating the changed boundary condition data to either structured or unstructured grid files (originally output by Pointwise, if that matters). "Command terminated by signal 11." (GRC using latest source on 2015-01-22, unassigned)
• cfcnvt
  • When appending a large (many zone) grid file to a smaller (few zones) one, cfcnvt will begin appending each zone but eventually runs out of memory and crashes. Appending the smaller grid to the larger works fine. (GRC, unassigned)
• cfpart
  • IPAR(9) value of 0 is written to the global header of the grid file. Wind-US then fails to recognize groups that are defined in the file. Must use ADF Viewer to manually correct this value to 1 after using cfpart. (GRC, Boeing to investigate)
  • Unstructured grids output from Pointwise contain errors in the header information. cfpart will refuse to work with these files. Pointwise corrected this bug (SPR 17249) in version 17.3R1, and cfpart version 1.12 will attempt to repair the grid header information. If these remedies are insufficient, the manual work-around involves using ADF Viewer to modify the header info, as described below.
    • cfpart will stop with an error of the form:

         Step Face Connectivity
          nIntrrFaces_z =     14900292
          nIntFaces     =     13385134
          nIntFaces .ne. nIntrrFaces_z
          program terminated in FaceCnctvty
      

      Manually compute: nIntrrFaces_z - nIntFaces = 1515158 (for the example above)
    • Open the grid in adfviewer.
      • Click the + to the left of the Zone 1 branch to expand it.
      • Click on idat_cff in the expanded list. It should be just below Zone 1.
      • The integer data will appear in the lower right panel labeled Node Data. Note the value of the third integer, 15218134 for this example.
      • Subtract the number computed above from this third integer, 15218134 - 1515158 = 13702976, and replace the third integer with this new value.
      • Once the value has been replaced, be sure to hit the 'modify' button in order to update the file.
      • Exit adfviewer. You may delete the backup file when prompted.
    • Rerun cfpart. It should complete successfully.
• gpro
  • When rotating a (structured) grid with gpro, the boundary condition data is not written to the output grid file. This limitation was intended to force the user to re-apply coupling to the new grid, but it is unclear why other boundary types are not maintained. GMAN is the recommended tool to use for grid transformations such as rotation. Documentation should be updated accordingly. (AEDC, Documentation updated)✔
• CFPOST
  • Use of groups causes a segmentation fault. (GRC using latest source on 2015-03-03, unassigned)
  • The chemistry reference state is no longer compatible with what Wind-US now uses. (GRC, unassigned)
  • Are total pressure and temperature for chemistry cases still computed the same as Wind-US? (GRC, unassigned)

Feature Requests

• Wind-US
  • Ability to specify default units (ie, SI) in the .dat file, much like CFPOST does.
  • For 2D/axi grids (kdim=1), force sequencing in k-direction to be 0 rather than stop with error.
  • Arbitrary inflow with zone_selector capability rather than having to specify each individual zone.
  • Easier way to prescribe profiles on arbitrary inflow planes.
  • Wall distance computed even along zone edges so that damping functions could be applied at inflows.
  • Constant heat flux boundary condition.
  • Fix simple turbulence model outflow extrapolation along coupled structured-zone boundaries.
  • Make DDES available for structured grids.
  • Higher-order zone coupling for DES/LES simulations.
  • On-the-fly accumulation of turbulence statistics (ie, u'v') for DES/LES without the need to spawn or checkpoint the solution.
  • Allow trailing comments after a line continuation character.
• Scripts
  • Update tools.login and tools.profile to not override an existing TOOLSROOT setting. See how cfd.login and cfd.profile work.
• Makefiles
  • Standardize existing makefiles to use a consistent set of default paths for lua, cgns, etc. Perhaps use new environment variables for LUAROOT, CGNSROOT, etc.
  • Indicate (even within makefile comments) what compiler version the preset compiler flags are for. New makefiles are often copied from older ones, yet the compiler flags might not be updated for each compiler.
  • Create a configure script for compiling source.
• CFPOST
  • Make derivative operators (ie, d/dx) available in the CALCULATE function.
  • Make derivative variables (ie, vorticity) work with unstructured grids.
  • The CFPOST chemistry capability (which was based on the Wind-US routines) needs to be updated to use the same reference state as Wind-US. A flag is needed in the solution file or a similar ADJUST keyword must be added to CFPOST. Must be careful processing older solutions. Perhaps the stored BCFD/Wind-US version number could be used to differentiate.
  • Add ability to specify the surface normal direction of subsets as:
        SUBSET range1 range2 range3 [DIR {+|-}{I|J|K}]
    This would allow the list or Plot3d output file to be written in standard i-j-k order rather than the order demanded by the default right-hand-rule specification of subsets.
  • Extend the CFPOST cut at {x|y|z} command to include an additional units specifier (in, ft, mm, ...). It currently assumes the cut value is in default length units.
    • This can be accomplished by specifying the default length units immediately before the cut command. Then reset the units for the desired output variables.

         grid test.cgd
         solution test.cfl
         zone 1
           sufrace k 1
         units inch        ! default length units for the cut command
         cut at x 1.00     ! cut at x=1.0[in]
         units si          ! default units for the output variables
         variables x;u     ! x[m],u[m/s]    x will be 0.0254
         genplot output cut.gen overwrite
         quit
      

Last updated 27 May 2016