Avoiding Large Polygonalization Times

---

Although it is a powerful method, it requires much more time than other methods. The processing time increases quickly with the number of objects in the cell description. Consider the orange air cell shown in wireframe in this figure The cell lies between an outer sphere and an inner sphere enclosing a target model and excludes a ring of spheres introduced as tally volumes. As each of the tally spheres is added to the collection of triangles, the sphere’s triangles are tested for intersection with all the triangles added to the cell so far, resulting in additional analysis time for each new component. Often complicated cells that take very long to polygonalize with this method result from the exclusion of many objects in a cell description, such as when defining air space around a complicated model. Exact polygonalization of such cells, besides taking a long time, is usually not useful for visualization of the model. For the example shown, an alternate cell description using only the outer sphere would suffice because the embedded spheres are themselves polygonalized and the additional surfaces from the polygonalization of the orange cell are redundant. The algorithm successfully polygonalizes the air surrounding the coffee model shown at left, but the processing time is 15 times greater than the next most expensive cell. The coffee model with the air cells invisible is shown in the figure below right.

A number of settings are available to prevent algorithm from attempting to polygonalize cells that would require a significant amount of time:

• The 3D Options property page contains an option to bypass polygonalization of cells that are not visible in 3D. 3D visibility can be set with QT and ST commands in a command file or as embedded C MORITZ comments in an MCNP input file. An invisibility density can be defined on the 3D Options page. Cells with a density less than that value are set to invisible in 3D. The setting should be useful for excluding air cells.
• A limit on the number of surfaces and bodies in cell description that the algorithm will attempt to process is set on the 3D Options property page. Cells with descriptions that exceed the limit are not processed.
• Moritz monitors the time elapsed during the algorithm. If the time exceeds the limit specified on the 3D Control property page3D Options property page, Moritz will abandon polygonalization of the cell or cell part using the triangles algorithm. If the limit is < 0, the test is not made.
• Alternate cell descriptions can be used for cells that can be adequately represented with a simple 3D shape.
• The polygonalization time depends on the number of triangles representing the constituent surfaces and bodies. This number depends on the tessellation parameters that can be changed on the 3D Style property page. Reducing the values results in fewer triangles and quicker polygonalization.
• The subdivision process may lead to very small triangles and extra work in further subdivision of the small triangles. The 3D Advanced Options property page contains a size below which triangles are not subdivided. The size is measured by the sum of the lengths of the 3 sides.
  

The combinatorial triangles algorithm is used, by default, as the method of last resort for surface geometry. The user can adjust the order in which the algorithm is used with respect to some other methods.  The other methods may result in quicker polygonalization but may or may not result in a poorer representation.

Processing time and the possibility of bad or no 3D representation of a cell increases with the complexity (the number of unions, complements, and parentheses) and length of a cell description. Using a number of simpler cells instead will usually give better results and performance in Moritz (for the 2D plotting as well as the 3D) and better performance in the Monte Carlo calculations.

If many cells are added for importance splitting, consider saving the original model, before the importance layers are added, for the 3D visualization.

White Rock Science http://www.whiterockscience.com/wrs.html 505 672 1105 email