Contents

• Abstract
• Purpose and Background
• VIM in the DOE Nuclear Criticality Safety Program
• Primary Features
• Nuclear Data Treatments
• Variance Reduction and Edit Features
• Computing Platforms
• References
• Using this Website

Purpose and Background

VIM is a continuous-energy criticality, reactor physics, and shielding code. It solves the transport problem for neutrons or photons, including thermal neutron scattering effects, either in the eigenvalue mode or for photon or neutron fixed source. The development of VIM was initiated at Atomics International (Ref 1) by L. Levitt and has been continued since the 1970s in the Reactor Analysis Division (now Nuclear Engineering Division) at Argonne National Laboratory, by R. E. Prael, E. M. Gelbard (Ref 2), L. J. Milton, and R. N. Blomquist. The VIM code as presently implemented at Argonne features a flexible geometrical capability, neutron physics data closely representing the ENDF/B and JEF data from which it has been derived, and photon data based on MCPLIB (Ref 3). Special neutron physics capabilities in VIM include unresolved resonance probability tables, and direct treatment of resolved resonances described with Reich-Moore parameters. It has been extensively benchmarked, using both experiments and other accurate codes.

VIM Version 4.0 is available through the Radiation Safety Information Computational Center at Oak Ridge National Laboratory.

Top of page

VIM in the DOE Nuclear Criticality Safety Program

VIM is one of the three general purpose Monte Carlo criticality codes used for analyzing the criticality state of fissile systems in the Department of Energy nuclear complex. The Department relies on three distinct codes in order to avoid common mode failures in analysis, thereby permitting more robust analysis of nuclear systems through code diversity and inter-code comparisons. See the DOE Nuclear Criticality Safety Program and the USDOE Nuclear Criticality Safety Program: Criticality Safety Codes and Data

VIM is also a testbed for DOE's studies of Monte Carlo source convergence problems that can arise with loosely coupled systems (Refs 33 and 34). Slow convergence (or non-convergence) can be insidious because the analyst might see no evidence of a convergence problem, and the apparently precise eigenvalue is non-conservative. A stratified sampling method that improves the robustness of Monte Carlo source convergence has undergone preliminary testing in a developmental version of VIM (Ref 35). VIM is being applied to the convergence benchmarks under study by the OECD/NEA Expert Group on Source Convergence in Criticality Safety Analysis.

VIM has been used in a variety of ICSBEP criticality safety benchmarks (Ref 36), e.g., the U/Fe Benchmark Assembly (a high-leakage core very sensitive to the treatment of scattering in iron, HEU-MET-FAST-035), the U9 Benchmark Assembly (an intermediate-enriched uranium core reflected with DU, sensitive to the U-235 and U-238 data, IEU-MET-FAST-010), and a Be-reflected HEU system (HEU-MET-FAST-041).

Top of page

Primary Features

The original VIM geometry package was designed to permit a simple description of plate lattice critical experiments. All cells of identical characteristics, with plates, clad, and void defined by a particular combination of rectangular parallelepipeds, need be defined only once; the full assembly is then described by defining a rectangular lattice constructed from the basic cells. The combinatorial geometry package developed for the code SAM-CE (Ref 4) has been implemented in VIM and extended by adding specific geometrical descriptions of particular interest in reactor analysis. The use of combinatorial geometry permits a detailed description of complicated and irregular assemblies as well as a convenient means of simulating one-, two-, and three-dimensional analytical calculations with non-orthogonal interfaces. The above two geometrical techniques have been combined in VIM to provide options for the description of repeating hexagonal and rectangular lattices with the in-cell geometrical definition employing the full combinatorial geometry capability. In addition, an infinite, homogeneous medium option is available to provide an efficient capability for data testing and cross section methods evaluation.

Top of page

Nuclear Data Treatments

Neutron cross section definition in VIM is by composition-independent microscopic data sets produced by a separate set of cross section processing codes. These codes are independent of all other continuous-energy cross section codes, e.g. NJOY. Resonance and smooth cross sections are specified pointwise with linear interpolation to provide a continuous energy cross section description; unresolved resonances are described by the probability table method (Ref 5). The pre-computed probability tables embody the statistical properties of the evaluated resonance parameter distributions. Thus, at run time, the resonance self-shielding is accounted for by the process of sampling the tables in the naturally occurring spectrum. Other non-statistical methods provide only an average unresolved resonance cross section, ignoring the self-shielding effects.

The reaction types fission, elastic scattering, discrete level inelastic scattering, inelastic continuum scattering, and (n,2n) reaction are specifically defined, while "capture" is defined as the sum of remaining possible outcome of a neutron collision. Neutron trajectories and scattering are continuous in angle. Anisotropic elastic and discrete level inelastic scattering are described with uniform delta-mu mesh probability tables derived from ENDF/B data. Most features of the ENDF/B specification of secondary energy distributions are allowed. The present VIM Version 4, 5, and 6 and JEF-2.2 cross section libraries consist of nearly 200 nuclides or materials at 300 degrees (K) defined over the energy range 20 MeV to 1.E-5 eV. Some materials have data at higher temperatures.

It is possible to construct VIM material files which are exactly equivalent to multigroup macroscopic cross sections. ISOTXS or COMPXS files with PN scattering can be directly converted, as can CASMO printed output, into VIM Material Files using the cross section conversion code, ISOVIM (documented separately in Ref 6). Conversely, the VIM cross section edits can be converted to ISOTXS files for use in deterministic transport calculations. Diffusion coefficients, however, are not produced.

Photon cross sections are also defined by composition independent microscopic datasets in the energy range from 1 keV to 100 MeV. Coherent and incoherent scattering, pair production, and photoelectric cross section data are described by pointwise values with log-log interpolation. Photon heating numbers are specified pointwise with linear-log interpolation. Pair production is simulated by creation of a double-weighted photon of energy 0.511008 MeV, and production by fluorescence is treated explicitly. The Klein- Nishina distribution is sampled exactly (Ref 7) for secondary angle and energy during incoherent scattering events.

Top of page

Variance Reduction and Edit Features

VIM automatically calculates multiplication eigenvalues by analog, collision, and track length estimation, and optimum combinations of pairs of the three eigenvalue estimates (Ref 10) is provided for variance reduction. Experience with other codes (Ref 8) has shown that a three-combined estimate is not significantly better than the two- combined estimate. Both collision and track length estimation are used to provide reaction rate estimates by region, group, and/or nuclide, while group- and region-wise integrated fluxes are provided by track length estimation. Optionally, infinite dilution region-averaged microscopic reaction rate ratios may be obtained in a designated central region. Automatically accumulated track length estimates of isotopic reaction rates and fluxes are used to provide estimates of broad group macroscopic and microscopic cross sections over edit regions. All output quantities are provided with standard deviation estimates at the completion of a run.

Top of page

Computing Platforms

VIM runs on Solaris, HP, Linux, and Windows systems. The "Running VIM" section of this guide provides instructions on running the codes on Solaris. These ought to be applicable on HP and Linux systems. Instructions for Windows PC use are also provided in the same section. Previous versions have been run on CDC and Cray computers, as well as IBM mainframes.

Top of page

Using this Website

This User's Guide is in use at Argonne National Laboratory, and was developed using Netscape Navigator (UNIX version). It uses various HTML constructs, including tables, and includes many files that contain column sensitive information which cannot be displayed accurately (or even very helpfully) with proportional fonts. Accordingly, the pre-formatted portions of the text use a prescribed fixed width font (Courier). The UNIX Netscape Font Preference settings used were (1) Times (Adobe) variable width font, (2) Courier (Adobe) fixed width font, and (3) Use document-specified fonts, including Dynamic Fonts. Because of variation among browsers and computer types, users may need to explore their browser environment to improve the visual quality of the guide.

This guide was written for the UNIX version of VIM at ANL. A separate section has been added for the Windows version, but many of the details for running the various codes are provided in the original parts of the manual. It will probably be necessary for PC VIM users to refer to both parts of the guide.