Non-classical radiation transport in stochastically heterogeneous materials :: SPE RESST

Non-classical radiation transport in stochastically heterogeneous materials


Computer simulation methods are now widely used in the development of nuclear technologies and resource-saving radiation processes as well as in resolving of fundamental and applied problems of radiation and reactor materials science. The common problem for all these applications is the need to calculate the radiation fields with an adequate consideration of the structural features of the radiation propagation environment. Therefore, the development of computer experiment relevant methods, algorithms and software tools and increasing of their computing efficiency are an integral part of R&D in modern radiation physics and technology.

In many practical cases, the propagation media have an experimentally unresolved internal heterogeneity, which nevertheless drastically affects the radiation transport inside them. Typical examples of such media are:

	subjects of industrial electrophysical and gamma-irradiation processes [1, 2007],

	disperse absorbers for reactors and radiation shielding,

	TRISO-fuels of pebble-bed high-temperature gas-cooled reactors (HTHRs) with double heterogeneity,

	turbulent flows of coolant and media with heterophase density fluctuations,

	atmospheres (cloud cover) of the Earth and planets.

In the theory of radiative transfer, their stochastic consideration as random media is accepted. For some of them (e.g., binary random mixtures), there exist statistically accurate twofold Monte Carlo methods [30, 2010] of radiation transport modeling. However, their computational efficiency is low and insufficient for solving many practically important problems.

In order to increase it, it is promising to use the concept of nonclassical (anomalous) transport phenomena in the structurally complex media, which is being intensively developed in modern statistical physics. On its basis, for materials whose structure is characterized only statistically, a new radiation transport modeling algorithm has been developed. In its framework, a total macroscopic cross section for the interaction of radiation with a material is represented by an ergodic random process along the mean free path of the particles, and the generalized radiation propagation kernals averaged over statistical ensemble of realizations of a random environment are introduced into the random material model.

This allowed us to build a one-time Monte Carlo method of solution of the generalized Boltzmann equation (Ed. Larsen, 2007) which describes nonclassical transport in media with spatial correlations of unresolved heterogeneity. Its application resulted in the performance gain of the CERN Geant4 based Monte Carlo code RaT 3.1 [19, 2010] by 1-2 orders of magnitude and opened the possibility of adequate modeling of the effects of non-classical transport of neutrons and photons in dispersive absorbers (see Figure 1) and new radiation-shielding materials.

Fig. 1 - Spatial distributions of helium production by thermal neutrons in the NSC KIPT developed dispersed neutron-absorbing material (grains of boron carbide in a pyrolytic matrix) Monte Carlo calculated in the approximation of an atomically homogenized material (1), by the double Monte Carlo method (2) and by the developed nonclassical one-time Monte Carlo method (3)