|
|

Disadvantages of Complicated Cell Descriptions
Cell descriptions containing many parentheses, unions, and complements should be avoided for several reasons. They require more computational effort on the part of MCNP and other transport codes, they can lead to long polygonalization times or approximate 3D representations in Moritz, and can be difficult for a human to understand. The latter point can apply to others attempting to understand your model or to yourself a year or two later. Conversion to solid body geometry is more likely to succeed with simple cell descriptions.
When tracking particles through the geometry, every time the particle crosses a surface contained in the description of the current cell, the cell description is evaluated to determine if the particle has entered another cell. A longer description leads to more cell surface crossings and description evaluations. A long and complicated description increases the computer time for each evaluation. Because the Moritz 2D and ray traced plots are made using a similar ray tracing technique, the same comments apply to those plots.
The following are relevant excerpts from the MCNP manual (Version 4C, LA–13709–M, 18 December 2000):
Be cautious about making any one cell very complicated. With the union operator and disjointed regions, a very large geometry can be set up with just one cell. The problem is that for each track flight between collisions in a cell, the intersection of the track with each bounding surface of the cell is calculated, a calculation that can be costly if a cell has many surfaces. (page 1–15)Discussing use of complement operators that introduce surfaces not bounding the cell in question:
1. Avoid excessively complicated cells. MCNP runs faster when the problem geometry is made up of many simpler cells rather than fewer more complicated cells. (page 3–11)
2. Avoid adding unneeded surfaces to the geometry description of a cell through poor use of the complement operator. The extra surfaces make the problem run slower and may (page 3–11)
Using the complement operator can destroy some of the necessary conditions for some cell volume and surface area calculations by MCNP. (page 2–10)
Even though surfaces 1 and 2 do not physically bound cell 4, using the complement operator as in this example causes MCNP to think that all surfaces involved with the complement do bound the cell. Even though this specification is correct and required by MCNP, the disadvantage is that when a particle enters cell 4 or has a collision in cell 4, MCNP must calculate the intersection of the particle's trajectory with all real bounding surfaces of cell 4 plus any extraneous ones brought in by the complement operator. This intersection calculation is very expensive and can add significantly to the required computer time. (page 2–10)
Moritz can usually process and display complicated models with hundreds of cells in seconds. In rare cases, one or a few very complicated cells require a significant time to process. The processing time and the possibility that of a bad or no 3D representation of a cell increases with the length and complexity of the cell description. Use of a number of simpler cells is preferable to single complicated cell; this approach should also reduce Monte Carlo transport time. The polygonalization § of the troubleshooting chapter addresses methods of dealing with such cases.
| White Rock Science http://www.whiterockscience.com/wrs.html 505 672 1105 |
|
|