Laka Foundation

Publication Laka-library:
Nucleon (1994)

AuthorNuclear Research Inst.REZ
Date1994
Classification2.39.0.00/01 (CZECH REPUBLIC - GENERAL)
Front

From the publication:

COMPARISON OF RESULTS OF MCNP 4.2 MONTE CARLO CODE WITH EXPERIMENTAL DATA OR 
COMPUTATIONAL RESULTS OF SOME DETERMINISTIC APPROXIMATE CODES

1. INTRODUCTION

An important way how to meet the contemporary demands of nuclear facilities 
operating organizations on safety and economic efficiency is the development 
and utilization of methods and codes based on the non-approximate form of the 
Boltzmann transport equation. The Monte Carlo codes belong to the category of 
such tools for the solving of transport problems. They make possible a detailed 
specification of the required neutron features of examined areas of the nuclear 
assemblies in a more detailed and more optimum way than it is possible with other 
numerical methods. The Monte Carlo code enables to form exact foundations for the 
planning of nuclear reactor operation and for the development of effective codes 
based on the approximate form of the transport equation. Moreover. MC codes can 
serve for testing the accuracy of some other numerical methods.

One of the best MC codes, the well-known Los Alamos MC code MCNP 4.2 [I], 
has been exploited by the author since 1991, when it was obtained from RSIC in 
Oak Ridge. This code is often referred as " the most rigorous analytical tool". 
A large number of nuclear reactor physics tasks ([3-6]. etc.) were run during this 
time period and many interesting and surprising results were acquired.

A small part of them is presented in this paper to show a portion of a very wide 
problem spectrum which can be treated by MCNP 4.2 code. The main reason for 
doing it is to point out the importance of conservative application of the 
deterministic approximate reactor codes.

Before presenting selected solved problems, it may be suitable to quote words of 
a Monte Carlo authority. Dr. E. M. Gelbard from Argonne National Laboratory, 
presented in a panel discussion [2]:

"But for one thing Monte Carlo could be used to benchmark the whole computational 
method. That is, we do slowing down calculations. We get group constants. We 
compute fluxes in cells. We group average, and in the group averaging we have to 
compute resonance escape probabilities and Dancoff factors, and when we are 
finished we end up with a very complicated problem. Let us say we have a lattice that 
is facing a water channel, and we have to make assumption about Dancoff factors. 
Every stage of that calculation could be benchmarked by Monte Carlo techniques."

This publication is only available at Laka on paper, not as pdf.
You can borrow the publication or request a copy. When we're available, this is possible for a small fee.