N. M. Greene
April 2002
ENDF/B-VI was released in 1989 and employs a collection of data formats referred to as the ENDF-6 formats. These formats use the same record structures as have been used in all previous versions, but make use of some complicated extensions to include new kinds of data that essentially made all of the codes that had been used to process ENDF/B obsolete. Since 1989, changes have been made to the formats to address new situations, though these changes have been modest relative to the change from ENDF-5 to ENDF-6.
ENDF/B-VII is being planned for release in 2004. As with previous versions, it will include new evaluations for the standards materials, in addition to several new evaluations for other materials. Some materials will have their evaluations extended to 150 MeV, and some new data types will be included. Along with the data release, the format changes that have been made to ENDF-6 since 1989 will be consolidated with additional changes that address some problems with ENDF-6 and also add some new capabilities.
Three major requirements for the ENDF-7 format are:
A special CSEWG subcommittee consisting of representatives from BNL, INEL, LANL, LLNL, ORNL, and the IAEA was formed to review format needs. This report is based on their recommendations and observations.
The consensus opinion is that the easiest and most logical way to satisfy the need for more digits on some terms is to simply use more than 80 characters in a record. The 80-character limit was imposed by punched cards, which are no longer in use. This means that we can use 12 or 15 digits for floating point numbers, 8 digits for the identifier or MAT (ZA-isomeric state) number, and 4 digits for the process identifier (MT number). There will be no change to the record structures or ordering of the records. This will minimize the impact on the processing codes, and, hopefully, also eliminate any need to increase these limits in the future.
The following are situations that are candidates for inclusion in the new format:
Several years ago, an effort was initiated to examine the existing ENDF/B formats to see if alternative schemes could be designed that would be easier for evaluators, processors, or other users to employ. The new format was designated "ENDF/C", and, unfortunately, was never fully designed or implemented. This task is being resurrected with a canvassing of the members of the ENDF-7 Format Subcommittee and some other interested parties to determine the characteristics that should be included in formats that are developed for future versions of ENDF/B. It should be emphasized that new formats are not being designed, or even seriously planned. This effort is simply attempting to ensure that we don't miss or ignore technological advances that could be used to our benefit, or that we consider new ideas that would make it easier to use the data that are put into ENDF/B files.
It is felt that the ENDF/B file should be an ASCII data transmittal file and should not employ special structures or database management systems, etc. It is duly recognized that an individual site or processing system can convert the file to any other kind of format or structure needed to satisfy special needs.
A "tagged" or "keyword" format is favored, because of several obvious advantages. There is no fixed format, and, hence, no limitation of number of digits. The file is easily readable by eye. New formats are easy to add (or ignore).
There are many reasons for ordering the data in an evaluation by nuclear reaction, and not by type of data, as is presently done. All of the data for an individual process (the point cross sections, the angular and energy distributions), etc., for all particles involved in a reaction, could be presented together, and not scattered in multiple artificial "files". It is recognized that some kinds of data, such as resonance parameters and nuclear decay chains, cannot be accommodated in this manner.
ENDF-6 has several places where the same kinds of data can be presented in different places. For example, discrete-level inelastic scattering can use File 4 or File 6 to specify angular distribution data. Likewise, many of the formats in File 5 can easily be accommodated in File 6 by simply describing a function that has isotropic angular variation or has the angular variation extracted from File 4. The photon production Files (12, 13, and 15), could easily be moved to Files 3 and 6, as could the photon interaction data in Files 23 and 27. It is believed that the elimination of many of these similar files and consolidating their data into a reduced number of places will make data processing easier and will result in a higher level of quality assurance.
There are more attractive schemes for managing evaluated cross-section data than are provided by the present ENDF/B files. However, there is no identifiable funding source for pursuing the development of a radically different file structure, along with the requisite new processing and utility programs. The task of looking ahead to the next generation of formats will continue to remain as a discussion item for the foreseeable future.