Solver-Numerics

• Change to second order spatial discretization as the default method for the pressure based solver.

  • The second order discretization scheme will provide improved solutions compared to the first order scheme used in previous releases. However, cases may take more iterations to converge and/or need changes to the solver settings for optimal convergence.

  • Previously setup cases are not affected and will retain the old default. New cases will use the updated default method.

• Change in default method of boundary limiting.

  • The new default boundary gradient limiting procedure improves solutions, particularly for cases with coarse meshes near boundaries. It also improves convergence by avoiding out of bound values during iterations. To revert to pre-FLUENT 14 code behavior, use the following rpvar command:

    (rpsetvar ‘recon/bc-minmax-id-new 1)

Solver-Meshing

• Several dynamic mesh algorithms related to remeshing and smoothing have been improved. These changes can result in slightly different meshes for dynamic mesh simulations that can effect the solution.

• The polyhedra conversion algorithm has been improved. Using the same mesh as a starting mesh, the polyhedra conversion might produce a slightly different polyhedra mesh.

• The quality based mesh smoothing (in the Smooth/Swap menu) has been improved and might return meshes of better quality.

Turbulence

• The new default near-wall treatment for the Spalart-Allmaras turbulence model is now the enhanced wall treatment with the Low-Re damping option enabled. The Low-Re damping option has been removed from the GUI. To revert to FLUENT 13 settings, first turn off the enhanced wall treatment for the Spalart-Allmaras model via the /define/models/viscous> sa-enhanced-wall-treatment? text command.

  A new text command is then available that allows you to turn the Low-Re Damping on or off: /define/models/viscous> sa-damping?

• Improvements have been made to scale-resolving turbulence simulations employing an underlying one- or two-equation RANS model (i.e. SAS or DES) and using a synthetic turbulence generator at an inlet or at a RANS/LES interface. Results may vary from previous releases.

• Rough wall treatment has been improved for epsilon-equation based turbulence models to avoid reduction in effective roughness when the near-wall mesh is refined. This is the new default treatment. Set the following rpvar command to false to return to pre-FLUENT 14 code behavior.

  (ke-rough-wall-treatment-r14? #f)

• The implementation of the Delayed DES (DDES) shielding function, f_d (Equation 4–228 in the Theory Guide), has been optimized in the SST and Realizable k-ε Detached Eddy Simulation (DES) models to provide effective shielding. The constant was changed from 8 to 20. With this change, DDES is now the recommended shielding function for the SST k-omega model with Delayed DES enabled and is used by default.

• The calculation of SAS-specific terms at periodic boundary conditions has been corrected and will yield improved model behavior.

Heat Transfer

• For the shell conduction model at T-junctions formed with 2 walls, the heat-conduction treatment has been corrected and will yield improved results.

• Postprocessing Wall Function Heat Transfer Coefficient (WFHTC) has been corrected. FLUENT no longer reports a value of zero for WFHTC on adiabatic walls. The previous behavior can be recovered with the following rpvar command.

  (rpsetvar 'wf/zero-wfhtc-on-adiabatic-walls? #t)

Reacting Flow

• The diffusion for the spark model is now limited to cells in close proximity to the spark region specified. This results in a more realistic prediction of spark propagation. Historically, the spark model would affect diffusion throughout the flow domain, and the new treatment only affects diffusion around the location of the spark.

Discrete Phase Model

• Movement and deformation of sliding, moving, and deforming meshes are now considered during the particle tracking. This improves the accuracy of particle tracks when particles are reflected from moving walls, especially in cases without wall boundary layers. Results may vary from previous releases. This effect can be disabled by using the following scheme commands:

  (rpsetvar 'dpm/consider-transient-mesh-movement? #f)

  (check-mesh-interpolate-in-time)

• A boiling rate equation for multi-component particles has been introduced, which has been derived consistently with the existing vaporization and boiling models in ANSYS FLUENT. This boiling rate replaces the rate equation used previously for the multicomponent particle boiling regime. The documentation has been updated in the Theory Guide. This change cannot be reversed through an rpvar.

• For multicomponent particles, the true boiling temperature is used to limit the Langrangian wall film model. Previously, the minimum of the component boiling points was used. The user cannot change this selection.

• In the DPM energy balance, the latent heat is computed consistently in the droplet and Lagrangian film models. Previously, the film model always used a constant latent heat value. The user cannot revert to the old method.

• Improvements to the droplet Vaporization Law numerics result in a more accurate vaporization history. As a result of the improved accuracy, computed trajectories may be longer compared with FLUENT 13.0. In addition, computational time may increase compared to FLUENT 13.0 if the computed vaporization time is longer. The change can be reverted by issuing the following commands in sequence:

  (rpsetvar ‘dpm/limiting-time-algorithm? #f)

  (rpsetvar ‘dpm/minimum-vapor-fraction-new 0.01)

  (dpm-parameters-changed)

• The Multicomponent Law numerics have been revised to speed up the computation. When importing case files from previous versions, you will need to disable Coupled Heat-Mass Solution for Multicomponent droplets to take advantage of the increased computational speed. This setting is found on the Numerics tab of the of the Discrete Phase Model dialog box.

• Several changes have been made to the Lagrangian wall film model that lead to more consistent evaporation of the wall film for pure and multi-component wall films. In addition, splashing of droplets has been improved to consider only one sampling from the cumulative probability density function of the underlying size distribution. These changes cannot be reversed.

Eulerian Multiphase Models

• The expression for ‘b’ in the Luo breakage kernel model in Table 2.1: "Luo Model Parameters" of the Population Balance Manual has been changed by a scaling factor, β^-1, where β=2.047. A domainvar, ‘pb/luo-beta-factor, has been introduced to make this factor user-modifiable using the following scheme command:

  (domainsetvar <pb-domain-id> 'pb/luo-beta-factor <value>)

  The FLUENT 13 behavior can be recovered by issuing the preceding command with <value>=1.

Acoustics

• Ffowcs Williams-Hawkings solver: reception time calculation is improved by interpolating the emitted timestep signal between the receiver timesteps covered by the received signal.

UDF Programming Interface

• Node unions replaced with node SVARs.

  • Two node union data members n1 and n2 in node_struct have been replaced by SV_N_TMP_0 and SV_N_TMP_1. SV_N_TMP_2 is also available if needed. Unlike previous versions, UDF developers will need to allocate/deallocate this storage in order to use the following node union macros:

    • NODE_MARK (uses SV_N_TMP_0)

    • NODE_RVAL1 (uses SV_N_TMP_0)

    • NODE_VISIT (uses SV_N_TMP_1)

    • NODE_RVAL2 (uses SV_N_TMP_1)

    For your convenience, two macros (ALLOCATE_NODE_SVAR and DEALLOCATE_NODE_SVAR) have been added to facilitate allocating this storage. For example, in order to use NODE_MARK, you would use the commands:

    ALLOCATE_NODE_SVAR(SV_N_TMP_0)

    DEALLOCATE_NODE_SVAR(SV_N_TMP_0)

  • Many node union macros such as NODE_VISIT and NODE_MARK have been used for flagging the nodes, so it is not really necessary to use a node union variable to do it. For your convenience, 3 new macros have been added. Please use CLEAR_NODE_VISITED to initialize a node flag, SET_NODE_VISITED to mark a node, and NODE_IS_VISITED to check the node status. You may also use function Clear_Node_Flags (domain, NODE_VISITED_FLAG) to initialize all nodes in the domain, and use Exchange_Node_Flags (domain, NODE_VISITED_FLAG) to exchange node flags in parallel.

• For multiphase simulations, the linearized mass transfer UDF is now used by default. To revert to the previous behavior, use the TUI command solve/set/expert and enter no at the Linearized Mass Transfer UDF? prompt. Alternatively, you can use the following scheme command:

  (rpsetvar ‘mp/mt/udf/linearized? #f)