ERRORJ: Covariance Processing Code System, Version 2.3.
NJOYCOVX
Reactors Physics Research Group, Japan Atomic Energy Agency, 2-4, Shirane, Shirakata, Tokai-mura, Ibaraki, 319-1195, Japan through the OECD NEA Data Bank, Issy-les-Moulineaux, France.
Fortran 77; PC Windows/Linux (P00526MNYCP03).
ERRORJ produces multigroup covariance matrices from ENDF/B-6 format following mainly the methods of the ERRORR module in NJOY94.105. Thus, all the function of ERRORR is available in ERRORJ. ERRORJ produces a COVFIL format file to store multigroup covariance data which is converted to a COVERX format file by the conversion program NJOYCOVX which is included. The COVERX format is proposed as a standard file of multigroup covariance data in the FORSS system.
ERRORJ Version 1 and NJOYCOVX were developed by K. Kosako in 1999. In September 2002, the average cosine processing of the elastic scattering angle was modified to calculate relative covariance correctly by Go Chiba of Japan Nuclear Cycle Development Institute. In the Version 2.2 release, non-diagonal elements of covariance matrices are calculated in the resonance energy region. An option for high-speed calculation was implemented and the perturbation amount was optimized in a sensitivity calculation. The effect of resonance self-shielding on the covariance of multigroup cross sections can be considered. It is possible to read a compact covariance format proposed by N. M. Larson.
Expands processing for the covariances of resolved and unresolved resonance parameters;
Processes average cosine of scattering angle and fission spectrum;
Accepts input of multigroup constants with various forms (user input, GENDF, etc.);
Outputs files with various formats through utility NJOYCOVX (COVERX format correlation matrix, relative error and standard deviation);
Uses a 1% sensitivity method for processing of resonance parameters.
With respect to the previous version of ERRORJ 2.2.1, the treatment of the self-shielding effect was corrected in this version. Note that point-wise cross section data must be used for the calculation with the self-shielding treatment. On the other hand the group-wise data is necessary when the correlation between different nuclides is considered. In other words, it is impossible to consider both the correlation and the self-shielding effect in the current code. The author of ERRORJ, Go Chiba, therefore suggests that the self-shielding effect should be evaluated without considering the correlations between different nuclides.
Additional features of version 2.3 of ERRORJ:
- In May 2010, two bugs were found in the FORTRAN source. This information was documented/reported to the author.
In general the JENDL 3.2 and 3.3 covariances were found to be very complete. Still some further improvements may be needed to resolve remaining problems. For example, several zero or very low standard deviations were observed at the energies around keV for the following isotopes:
- U235(tot), (elast.), (n,f), (n,gamma) : below ~20 keV
- U-238(n,f) between ~500 eV - ~100 keV
- Pu-239(n,f) and (n,gamma) ~keV
- Pu-240(n,gamma) ~keV
- Pu-241(n,f) and (n,gamma) < ~200 eV
For some isotopes (e.g. U-235, Pu-241) this happens at the energy regions where the cross section values are not small.
Running time is case dependent.
ERRORJ 2.3 has been run on LINUX and Windows operating systems.
ERRORJ 2.3 was developed on a PC with Linux. A Fortran 77 compiler is required to compile the codes; no executables are included with the package.
a) included in document:
Go Chiba, ERRORJ – A Code to Process Neutron-Nuclide Reaction Cross Section Covariance Version 2.3, Japan Atomic Energy Agency, JAEA-Data/Code 2007-007 (March 2007).
b) background information:
“Preparation of a Covariance Processing System for the Evaluated Nuclear Data File, JENDL,” PNC PJ9037 97-001 (1997), PNC TJ9037 98-001 (1998), JNC TJ9440 99-003 (1999).
K. Kosako, Covariance Data processing Code: ERRORJ, Specialist's meeting of reactor constants, Tokai, February 22-23, 2001.
R. E. MacFarlane and D. W. Muir, The NJOY Nuclear Data Processing System, Version 91, LA-12740-M (October 1994).
ERRORJ 2.3 is distributed on a CD which includes the document referenced in 10.a in PDF format and a compressed Linux tar file which includes source code, data files for sample cases, test input and output.
May 2004, September 2005, January 2011, August 2012.
KEYWORDS: COVARIANCE MATRICES; MULTIGROUP CROSS SECTION PROCESSING BASED ON ENDF/B; SENSITIVITY ANALYSIS