Description

The Method pull down list lets you specify the particular TFI method to apply to domains when the Initialize button is pressed. Blocks will always be initialized using the Standard method. On the other hand, the choices when initializing domains are the following:

Available TFI methods.
The Initialization pull down list allows you to specify the particular TFI method to use when initializing a structured domain.
  • Standard: It applies blending functions based on the relative spacing of the grid points on the boundaries (Ref. 22). The TFI algorithm is applied independently to the (x,y,z) coordinates of a grid. Because the blending functions are arc length based, the resultant grids usually have satisfactory clustering. Standard TFI is the third method after Parametric and Parametric Fit TFI that Fidelity Pointwise tries to apply by default to create the initial surface grid for domains. Standard TFI is the only method used for blocks.
  • Linear: It applies the TFI algorithm to the (x,y,z) coordinates of the grid, but the blending functions are no longer related to the distribution of grid points on the boundaries. This method is rarely better than Standard TFI but is provided for comparison and historical purposes.
  • Parametric: It can be used when all of the grid points on the boundaries of a grid are on the same database entity. In this method, the basic TFI algorithm is applied to the parametric database coordinates (u,v). Once the parametric coordinates (u,v) are computed at each grid point, the physical coordinates (x,y,z) are calculated from direct evaluation of the database entity. The result is a grid that lies exactly on the database entity. If you apply Parametric TFI to grids that do not reference one and only one database entity, you will be warned and the grid points will not be updated. Parametric TFI is the first method that Fidelity Pointwise tries to apply by default to create the initial surface grid for domains.
  • Polar: It is best suited for grids intended to be shaped like a body of revolution. Selecting this method opens the Polar Axis Selection frame where you can define the axis, shown below. Refer to the table for information on the definition of the current axis. Current Axis definition for the Polar initialization method.
    The current axis definition is displayed in a table when the Polar method is selected.

    Press Begin to expand the Polar Axis Selection frame. As shown below, there are three methods for defining the axis: Curve, 2 Points, and Point and Direction.

    Polar axis defined by curve
    A polar axis can be defined by curve, 2 points or point and direction.

    The default Curve option expects you to pick a curve in the Display window or in the List to use as the axis. If the curve is not linear, the endpoints will be used to define the axis. The 2 Points option allows you to pick points in the Display window or enter them directly into the XYZ text field. Point and Direction allows you to pick or enter a point and then select a primary axis direction through that point or specify a non-primary axis direction by entering directly into the Vector text field.

    Delete Last Point allows you to delete selected or entered points when using either the 2 Points or Point and Direction axis specification options. Clear Axis allows you to start the axis specification process over from the beginning by clearing the currently defined axis. Press End to confirm your change and the information in the Current Axis table will be updated accordingly.

    The (x,y,z) grid point coordinates on the boundaries are then transformed to polar coordinates (r, alpha, z') relative to the axis of revolution. The TFI algorithm is applied to each of the polar coordinates. Finally, the polar coordinates are transformed back to Cartesian coordinates. Polar is not available for blocks.

  • Parametric Fit: It is similar to Parametric TFI in that it will produce grids that lie on the database shape. As is the case for Parametric TFI, the grid points will be on a single database surface entity. However, it is much more general than Parametric TFI because the boundary points do not have to lie on a single database entity. Two less stringent conditions must be met before Parametric Fit TFI may be applied:
    1. A certain percentage of the grid's boundary points must reference database entities of any kind.
    2. All of the grid's boundary points must lie within the extent box of at least one database surface entity.

    If the grid passes the two tests above, then the grid's boundary points are projected (using closest point) onto each of the candidate (according to condition 2 above) database surfaces. The projection that results in the smallest cumulative movement of the boundary points determines the database entity to use for Parametric TFI. The grid's boundary points are projected onto the entity, and the resulting (u,v) coordinates are used to drive Parametric TFI. Note that the projection of boundary points is only a temporary operation; the original coordinates of the boundary points are restored.

    Parametric Fit TFI is the second method after Parametric TFI that Fidelity Pointwise tries to apply by default to create the initial surface grid for domains. For Parametric Fit TFI to be applied automatically to a new domain, a third condition must be satisfied:

    1. The total distance projected from the grid's boundary points to the candidate surface must be less than a tolerance.

Demonstration