1. NAME AND TITLE

DISKTRAN: Dose Calculations at Detectors from the End of a Cylinder Using DOT IV Scalar Flux Data.

2. CONTRIBUTOR

Oak Ridge National Laboratory, Oak Ridge, Tennessee.

3. CODING LANGUAGE AND COMPUTER

Fortran IV and Assembler; IBM (A). Fortran 77; Cray (B).

4. NATURE OF PROBLEM SOLVED

Fluxes and detector responses are computed at detectors in void regions using DOT IV calculated scalar fluxes on the end surfaces (disks) of a cylinder. Integral, as well as spectral responses, may be calculated. In addition 1-D angular dependent detector responses may be calculated by group at each detector location. Several response functions may be entered, either as 1-D or as 2-D ANISN cross section arrays.

5. METHOD OF SOLUTION

Fluxes at the detectors are obtained by subdividing the disk surface radially and azimuthally and summing contributions to each detector. Response functions are then folded with the fluxes to calculate detector responses.

6. RESTRICTIONS OR LIMITATIONS

The accuracy of the calculation probably increases with the order of the quadrature. Quadrature orders below S8 may give inaccurate results because with the few quadrature directions neither the constant flux nor interpolated flux option is likely to give a good representation of the continuous flux field. A given detector can see only one end of the cylinder. Contributions from the side surface are not calculated. Storage limitations require that there be enough space for at least one group of fluxes to reside in core.

7. TYPICAL RUNNING TIME

Running time depends mostly on the number of groups, the number of radial intervals (including subdivisions of those intervals), the number of detectors, and the number of angles used in the azimuthal angle spatial integration if there are off-axis detectors. Note that for the azimuthal angle integration, the number of angles may be constant and equal for all radial intervals or may vary from two angles in the inner radial intervals to many angles in the outer radial intervals when the spatial integration is performed based on equal-area subdivisions of the entire disk. The number of quadrature directions (because of the searching required to determine the directional flux for a given direction) and the number of responses (because of the extra multiplications and sums) also influence the running time, but their influences appear to be small.

For a sample case with 51 energy groups, 63 radial intervals (each subdivided into three subintervals for a total of 189 subintervals), nine on-axis detectors, 96 quadrature directions (S12), 10 detector responses, and 75 axial intervals, a total of 327,558 storage locations (for all groups in core) and 5.1 seconds of IBM 3033 CPU time were used. For off-axis detectors, the time should be roughly in proportion to the number of angles used in the azimuthal integration (or for the equal-area option, the average number of angles: the computed number of equal-area segments divided by the total number of radial interval subdivisions). For 40 angles this case would use about 3.4 minutes if all nine detectors were off axis. A time penalty is incurred if all group fluxes cannot be contained in the core, since the same searches will be conducted for each of the subgroupings. If the search time is small, there should be little effect on the total time. On the IBM 3033 the sample problems ran in less than 5 seconds. The same cases ran in less than 1 second on the Cray X-MP.

8. COMPUTER HARDWARE REQUIREMENTS

DISKTRAN runs on the IBM 3033 or Cray X-MP computers.

9. COMPUTER SOFTWARE REQUIREMENTS

For the IBM (A) version, either a Fortran H compiler or the VS Fortran compiler (specifying the 'LANGLVL(66)' option) is required to run DISKTRAN. An IBM assembler is also needed. The ABEND routine is used to write an appropriate error message and terminate the calculation. A system timing routine is needed to retrieve and pass time in hundredths of seconds to Function ICLOCK. ALOCAT is used to dynamically allocate an array and pass it, along with its dimension, to a subroutine supplied as an argument. The Cray (B) version uses the CFT compiler with FORTLIB under CTSS.

10. REFERENCE

Charles O. Slater, "DISKTRAN," ORNL Informal Documentation (1986 and 1988).

"Important Modifications for Cray Users," Informal Notes (8/88).

11. CONTENTS OF CODE PACKAGE

Included are the referenced document and one (1.2MB) DOS diskette containing the source codes, sample input and output.

12. DATE OF ABSTRACT

August 1988, revised October 1990.

KEYWORDS GAMMA-RAY; KERNEL; NEUTRON; SPATIAL INTEGRATION

Important Modifications for Cray Users

August 1988

The Cray version is functionally equivalent to the IBM version. Fortran routines have been substituted for the assembler language routines, and some Fortran 77 statements were introduced to allow it to run under CTSS using the CFT compiler with FORTLIB.

The input differs slightly from the IBM input listed earlier in this document. The 0$ array for NBUF has been omitted from the Cray version and replaced by an additional entry on the 1$ array. After NREG Cray users should specify: