Improving Mapping Quality

When a coupled analysis is executed, System Coupling executes a mapping process, using data from source-mesh locations to calculate data for target-mesh locations. When mapping is performed, the process runs to completion and generates data regardless of any existing issues that may impact the quality of the mapping. Use the following best practices to help ensure the accuracy of the mapping and the resulting data transfer:

Review mapping diagnostics.

Mapping diagnostics written to the Mapping Summary of the Transcript/Log file provide insight into the ultimate accuracy of the data transfers. They show the percentage of the target and source mesh locations that were successfully mapped, providing insight into the ultimate accuracy of the data transfers.

  • Target-side diagnostics:

    • If less than 100% of the target-side locations were mapped, then System Coupling is implementing a method of generating data on target locations according to the type of mapping used for the transfer.

      This is often caused by differences in the geometry. For example, mapping might be prevented because the target geometry does not overlap/intersect the source geometry, or there might be a gap (too much distance) between the source and target geometries.

      For more information, see Generating Data for Target Locations.

    • If between 99% and 100% of the target-side locations were mapped, then this is shown in the diagnostics as >99% and all other values are rounded to the nearest whole number.

  • Source-side diagnostics:

    • If less than 100% of the source-side locations were mapped, then some target data are not being used.

      In conservative mapping, this means that the quantity transferred is either not fully conserved and/or is conserved only in areas where the mesh overlaps. However, source-side mapping of less than 100% typically occurs only in cases of profile-preserving mapping where the source mesh is significantly finer than the target mesh.

    • If between 99% and 100% of the source-side locations were mapped, then this is shown in the diagnostics as >99% and all other values are rounded to the nearest whole number.

Use similar participant geometries.

Use geometries that are as similar as possible on all participant regions on the coupling interface. For example:

  • Minimize the gap between the source and target geometries, if possible.

  • Ensure that there is overlap/intersection between the source and target geometry regions where data will be transferred.

    • For data transfers using profile-preserving mapping to a target surface, you may address areas of non-overlap by changing the value of the UnmappedValueOption setting to Extrapolation. With this setting, target data is smoothly extrapolated onto the non-overlapping sections of the target surface.

  • Ensure that all source and target regions on the coupling interface have the same reference state.

Take advantage of mapping controls.
  • By default, System Coupling checks the quality of intersection between the sides of a coupling interface. An interface-level StopIfPoorIntersection setting allows you to specify that System Coupling stops and generates a setup error when intersection quality falls below a specified level, which you can change using the PoorIntersectionThreshold setting.

  • If gaps cannot be eliminated for surface transfers (for example, a structural geometry is in its manufactured state and a fluid geometry is in its deformed state), mapping accuracy may be improved by adjusting the AbsoluteGapTolerance and/or the RelativeGapTolerance.

  • For surface transfers, ensure that the surface mesh for each participant has area vectors which point out of the volume region associated with that participant. You can visualize these area vectors in EnSight using the area__EV variable. In the following rare situations, the area vector might point into the volume region rather than out:

    • The surface region comes from a Fluent fluid-porous interface.

    • The surface region comes from the solid side of a conjugate heat transfer (CHT) interface in Fluent or CFX.

    • The mesh does not satisfy usual Ansys mesh ordering conventions. This should not happen with meshes generated by Ansys meshing products but may be possible if the mesh was generated using an external tool.

    If any of these situations occurs, you may improve the mapping by setting the FaceAlignment setting to:

    • OppositeOrientation (same as default ProgramControlled option except when MAPDL shells are present)

    • AnyOrientation (may fail if the geometry has thin bodies)

    • SameOrientation (if you know the area vectors are directed inward)

  • For surface transfers to one side of an MAPDL shell region, if the geometry also has thin bodies (for example, thin airfoils constructed from shells), you may need to inspect the area vectors (area__EV) in EnSight to determine which way they are oriented, and then set FaceAlignment to either SameOrientation (if vectors are oriented into the shell) or OppositeOrientation (if vectors are oriented out of the shell).

  • For surface transfers to both sides of an MAPDL shell region, System Coupling will give accurate mapping results only for conservative transfers to the shell region and profile-preserving transfers from the shell region.

Use similar participant meshes.

Use meshes that are as similar as possible on all participant regions on the coupling interface. For example:

  • Refine meshes to the degree feasible. Coarse meshes can cause and/or exacerbate mesh gaps or dissimilarities, which in turn can have an adverse effect on mapping. This is especially important for geometries with meshes on curved surfaces.

  • Ensure that the meshes of the source and target regions of the coupling interface are suitable for the problem being solved. In general, source and target mesh resolutions should be as similar as is possible. Variances should be guided primarily by the participant physics and the type of data being transferred. For example:

    • For data transfers of intensive variables (that is, which use a profile-preserving mapping algorithm), a target-mesh resolution that is similar to or finer than the source-mesh resolution is recommended.

      When a profile-preserving algorithm is used, the transferred quantity is better preserved when the target mesh has either the same number or more nodes than the source mesh. This prevents features that are resolved on the source mesh from being lost in the transfer to the coarser target mesh.

    • For data transfers of extensive variables (that is, which use a conservative mapping algorithm), a target mesh that is similar to the source mesh is recommended, though a target mesh that is slightly coarser than the source mesh is acceptable.

      When a conservative algorithm is used, there is no benefit (from a mapping point of view) in the target mesh having more nodes than the source mesh. The conservative nature of the transfer ensures that the quantity is preserved even when transferred to a coarser target mesh.

      Note, however, that a finer target mesh may be beneficial from a physics-modeling point of view. Mesh resolution should be adjusted as needed to obtain an accurate solution for the participant physics.

Use geometries that are similarly oriented.

Ensure that participant geometries have similar orientations. Significant differences may cause the intersection between coupling interface sides to be insufficient for accurate mapping between them. System Coupling provides the following ways to minimize orientation-based issues:

  • Determine when orientation issues affect mapping.

  • Improve the intersection between participant geometries.

    Use System Coupling's geometry transformation functionality to define transformation reference frames.


Note:  In most cases, reference-state disparities must be addressed as part of the geometry and/or participant setup. For thermal-electromagnetic problems, however, they can be addressed by System Coupling during the coupled analysis setup.