Results
from Recent Conferences Addressing the Monte Carlo (MC) Methods
(note
by the NEA Secretariat)
During
the last two years five conferences, cosponsored by the OECD/NEA, have
taken place during which particle transport MC methods had an eminent place
in the programme or were the subject of the conference itself. These are:
M&C'99,
Madrid, Spain; ICRS9, Tsukuba, Japan; PHYSOR'2000,
Pittsburgh, USA, SNA'2000, Tokyo, Japan, and
MC2000,
Lisbon, Portugal.
Brief
summaries are accesible through the respective hyperlinks above describing
the specific MC aspects presented in these. Each of these conferences has
shown that the Monte Carlo method is used more widely than ever before.
The reasons are:
-
Computer
architectures have evolved that are particularly well suited to increase
the speed of MC codes in problem solving.
-
It removes
unnecessary model simplification
-
Powerful
biasing schemes and well established, widely known computer codes are used,
statistical analysis has been further developed, the method has become
user-friendlier; in short it has matured a lot.
These
recent conferences have confirmed that use is made now in large areas of
applications. Particularly intensive is the use made in radiation physics,
diagnostics in material identification, material science, radiological
and medical applications. Another area of wide use is deep penetration
of radiation into matter and radiation shielding, including intermediate
particle energies applications. Criticality safety is a field were MC methods
are used as a standard analysis tool. It is particularly suitable for calculating
the integral parameter k-eff describing the level of criticality, however
the convergence of loosely coupled systems is still a challenging problem.
The
use of MC in the area of nuclear power has undergone an important evolution.
Notable are the extensions to compute burnup in reactor cores, and full
core neutronic simulations. The aspects concerned with material or geometry
perturbation are starting to be successful after a long development period.
First results from sensitivity analysis with MC have been presented that
are promising, but still timid. Adjoint MC is being used more widely now.
In
many aspects of NPP simulation the MC method is still not applicable and
its use would require much further development. Deterministic methods will
continue to play an important complementary role. We can predict a symbiosis
of stochastic and deterministic methods (including coupled and hybrid methods)
for many more years.
Two
examples of difficulty:
-
Most
MC codes provide variance of stochastic nature, not uncertainty due to
modelling. Besides perturbation and sensitivity, general uncertainty analyses
need to be further developed and integrated into MC if it is to be widely
used for engineering and safety applications. Deterministic methods used
today have developed more general uncertainty analysis modules than MC
codes.
-
Reactor
core transient simulations require coupling of 3D neutronics with thermal
hydraulics, fuel behaviour and eventually with structural mechanics. Success
in this area has been achieved with deterministic methods. Monte Carlo
codes have to go through a long development and validation phase before
this can be achieved. Also computing times would at presently available
computing power still be prohibitive.
In order
to meet the increased interest and needs of the nuclear community, several
training courses are organised every year, during which code users learn
how to carry out efficiently modelling with MC. Several hundred, mostly
young persons, were trained during the last few years.
In
conclusion the Monte Carlo method has proven to be very successful, in
particular for radiation transport problems. Its use will increase further
in particular if methods developments are pursued. In order to foster such
developments this topic should continue to be on the agenda at international
conferences and a specific series of MC conferences is justified and should
be maintained.
E.
Sartori, December 2000
Last reviewed: 27 May 2011