Search NASA⌕ Search

Engineering topics

Favorite, Jeffrey A

Publications and source records attributed to Favorite, Jeffrey A.

Identification of Unknown Interface Locations in a Source/Shield System Using the Mesh Adaptive Direct Search Method

The Levenberg-Marquardt (or simply Marquardt) and differential evolution (DE) optimization methods were recently applied to solve inverse transport problems. The Marquardt method is fast but convergence of the method is dependent on the initial guess. While it has been shown to work extremely well at finding an optimum independent of the initial guess, the DE method does not provide a global optimal solution in some problems. In this paper, we apply the Mesh Adaptive Direct Search (MADS) algorithm to solve the inverse problem of material interface location identification in one-dimensional spherical radiation source/shield systems, and we compare the results obtained by MADS to those obtained by Levenberg-Marquardt and DE.

42 ENGINEERING↗

A Brief User's Guide for PARTISN. Revision 1

This document is intended as a very basic primer on the use of the PARTISN multigroup discrete-ordinates transport code for neutron criticality and time-independent neutron leakage problems. One-dimensional (r only) spherical and two-dimensional (r-z) cylindrical geometries and only the most basic options are discussed. It is assumed that the Nuclear Data Interface (NDI) is the source of nuclear data, that volume sources are isotropic, and that external boundaries are vacuums. The location and meaning of useful information in the PARTISN output is discussed. A sample input deck for a cylindrical problem that stacks a leakage calculation and a k eff calculation is listed. The method for inputting cross sections by hand (e.g., for analytic benchmark comparisons) is presented.

97 MATHEMATICS AND COMPUTING↗

User-Friendly Surface Flux Tallies in MCNP

Monte Carlo surface flux tallies use a fixed cosine value that defines the grazing range and a fixed cosine value to score with instead of a grazing cosine. In MCNP, these cosines are hard-wired, but making them user inputs would allow users to diagnose and overcome difficulties associated with grazing in surface flux (F2) tallies. This paper describes a simple means of allowing these cosines to be user inputs. A new special tally treatment, MUC, is used; it has two entries, the cosine cutoff and the substitute cosine. Nine subroutines were modified and one global variable, dimensioned 2 x the number of tallies, was added. These modifications will allow MCNP to compute analytic benchmarks for surface flux tallies.

97 MATHEMATICS AND COMPUTING↗

Monte Carlo surface flux tallies

Particle fluxes on surfaces are difficult to calculate with Monte Carlo codes because the score requires a division by the surface-crossing angle cosine, and grazing angles lead to inaccuracies. We revisit the standard practice of dividing by half of a cosine 'cutoff' for particles whose surface-crossing cosines are below the cutoff. The theory behind this approximation is sound, but the application of the theory to all possible situations does not account for two implicit assumptions: (1) the grazing band must be symmetric about 0, and (2) a single linear expansion for the angular flux must be applied in the entire grazing band. These assumptions are violated in common circumstances; for example, for separate in-going and out-going flux tallies on internal surfaces, and for out-going flux tallies on external surfaces. In some situations, dividing by two-thirds of the cosine cutoff is more appropriate. If users were able to control both the cosine cutoff and the substitute value, they could use these parameters to make accurate surface flux tallies. The procedure is demonstrated in a test problem in which Monte Carlo surface fluxes in cosine bins are converted to angular fluxes and compared with the results of a discrete ordinates calculation.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

A hybrid differential evolution/Levenberg-Marquardt method for solving inverse transport problems

Recently, the Differential Evolution (DE) optimization method was applied to solve inverse transport problems in finite cylindrical geometries and was shown to be far superior to the Levenberg-Marquardt optimization method at finding a global optimum for problems with several unknowns. However, while extremely adept at finding a global optimum solution, the DE method often requires a large number (hundreds or thousands) of transport calculations, making it much slower than the Levenberg-Marquardt method. In this paper, a hybridization of the Differential Evolution and Levenberg-Marquardt approaches is presented. This hybrid method takes advantage of the robust search capability of the Differential Evolution method and the speed of the Levenberg-Marquardt technique.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗