For certain applications it is necessary to create statistically distributed inhomogeneities, e.g. to generate a bimodal distribution of objects. The Match Solid Volume Fraction (SVF) Distribution option allows to create such structures with an inhomogeneous solid volume distribution. Objects are shifted and rotated to optimally match a SVF distribution given by a Gaussian random field. This takes place after the generation step. Consider that Periodicity is checked automatically for all directions. This is a prerequisite for the Match Solid Volume Fraction algorithm.

|
Note!
- Using this feature, overlap cannot be removed during structure generation.
- As the SVF distribution is matched after generating the objects fitting the Stopping Criterion, the resulting values can differ from the inserted parameters for the Stopping Criterion. This is because the structure is generated with overlap and the amount of overlap can change when Fibers are shifted.
- The distribution for Center defined in the Fiber Options tab will also be affected strongly by matching the SVF distribution after generating the grain structure with the specified center distribution. The difference between the Center Distribution and the Match SVF Distribution is explained below.
- It is recommended to generate the structure without using the Center Distribution option under the Fiber Options tab when using this algorithm. The objects will be moved afterwards matching the given SVF distribution not considering the center distribution. If you need inlet and outlet, add them later either with GadGeo - Edit Domain to avoid cropped objects or with ProcessGeo -Embed (see the ProcessGeo handbook).
- It is possible to reuse an existing stochastic field describing the desired SVF distribution. For this, first generate the structure with the mode Allow Overlap and then use the Match SVF distribution algorithm in GadGeo. Open the GadGeo section by selecting Model → GadGeo from the menu bar. Select Algorithms from the first pull-down menu and then Match SVF Distribution from the second.
- For large structures, the feature needs a large amount of memory as a volume field is loaded in GeoDict additional to the structure file.
|
The SVF distribution can be controlled through the Match SVF Distribution dialog accessible via the Edit … button.
In the Match SVF Distribution dialog, the following parameters can be set:
Stopping Criterion: Define the Error Bound to determine when the SVF matching should be stopped. When the normalized relative change of solid volume fraction is smaller than the given error bound, the algorithm is stopped.
Distribution Coarsening Factor: The SVF distribution is matched using a coarser version of the created Gaussian random field. The value must be small enough to resolve the heterogeneity of the structure. A larger value leads to a shorter run-time. If the Distribution Coarsening Factor is set to 1, the Gaussian random field is generated with the same resolution as the grain structure.
For a structure with a domain size of 2048x1024x512 voxels and a voxel length of 1µm a coarsening factor of 32 leads to a random field of 64x32x16 voxels with a voxel length of 32µm.
To resolve the correlation length with a whole number of voxels, the coarsening factor should be chosen as a divisor of the correlation length. For example, if the correlation length is set to 128 voxels and the coarsening factor is set to 32, 128/32 = 4 voxels remain to resolve one feature of the random field.
The correlation length is recommended to be a divisor of the domain size. Otherwise, the runtime increases significantly.
Distribution Mode: For distribution select Gauss or Gradient from the pull-down menu. The following parameters are different for these two options.
- Gauss:
- Correlation Length: The correlation length is a measure for the inhomogeneity of the material. The parameters for X-, Y- and Z-Direction determine the correlation length of the Gaussian random field, which defines the SVF distribution. It is recommended to select a divisor of the domain size in the corresponding direction. For a constant SVF in one direction choose the corresponding domain size as correlation length.
- Distribution Standard Deviation: Define the relative standard deviation of the created SVF distribution. Relative means relative to the given object solid volume percentage. A larger value leads to a larger difference between the regions of low and high density. Using the relative standard deviation instead of the absolute standard deviation allows using the same Gaussian field for different solid volume percentages.

- Gradient:
- Use Relative Position: If checked, the left column values from 0 to 1 correspond to locations in the structure. In the Z-direction, the value 0 is at the origin and the value 1 is at end of the domain. If not checked, the values in the left column correspond to absolute values in the given unit.
- Distribution Table: The right column assigns relative density values at the locations defined in the left column, e.g. the value 10 means that there are five times more objects at Z = 1 than at Z = 0.1, with a density value of 2. The object density increases and decreases smoothly between the given locations in the Z-direction.
- Number of Rows: The number of rows in the distribution gradient can be increased or decreased to enter as many pairs of position and density as desired.
- Load/Save: The density distribution can be saved to and loaded from a .txt file.

Example - Gauss
In the following example curved fibers are generated in a domain of 600x600x100 µm. For the first structure Allow Fiber Overlap is checked and no SVF distribution is matched. The structure of the second picture was generated with the same parameters except for matching the SVF distribution shown below.
A volume field describing the used gaussian random field is saved in the result folder after generating a fiber structure matching an SVF distribution. Select File → Load Volume Field … from the menu bar to load it or choose the Data Visualization tab in the Result Viewer of the result file (*.gdr) and click Load *.gvf. In the Loading volume file dialog click OK.
For the above example, the resulting gaussian random field is shown in 2D view from all three directions.
Viewed from X-direction a high correlation in Z-direction is observed, as the structure size in this direction equals the Correlation Length Z. In Y-direction a more inhomogeneous SVF distribution was generated, as the domain size in Y-direction (600 µm) is 12 times the Correlation Length Y (50 µm).
Viewed from Y-direction a similar observation can be obtained. Still the distribution in Z-direction is very homogeneous, whereas it has low correlation in X-direction.
Viewed from Z-direction, the structure is very inhomogeneous, due to small Correlation Length values for X and Y (50µm) compared to the structure size in these directions (600µm).
Example - Gradient
In the following example observe how the fibers are distributed corresponding to the given density distribution.
The fibers in the first example on the left have the highest density in the center of the Z-axis. Towards both ends of the domain in Z-direction the fibers become fewer. This corresponds to the density values 0 for the relative positions Z=0 and Z=1 and a density of 1 for the relative position Z=0.5.
The second example on the right shows a density distribution in Z-direction with 4 rows. A density of 20 for position Z=0 leads to twice as many fibers near the top as fibers in the lower third with a density value of 10 for position Z=0.7. Towards the center and the bottom, the fibers become fewer for a density value of 0.
Difference Gradient VS Density Distribution
This feature is different from the Density Distribution for Center. The Match SVF Distribution is a post-processing step, and the fibers are moved until they match the given solid volume fraction distribution.
For the center distribution (in the example below shown on the left), the fiber centers are placed according to the given distribution, which is much faster and does not change the orientation of the fibers but leads to a different solid volume fraction distribution and the distribution is not as smooth as with Match SVF Distribution, which is shown on the right in the example below.
|