Monte Carlo neutron transport

Monte Carlo is a high precision simulation method, broadly used both in research and industry
in many different areas. In the nuclear field, it is applied to simulate neutron transport, and the
high accuracy solutions produced are often used as a reference for comparison with
deterministic codes. However, in order to obtain good statistics and low variance results, an
extremely high number of neutrons must be simulated. This is particularly true in complex
geometries, e.g. a full reactor core, and it results in large computational times and resources
utilization.

The most limiting process in Monte Carlo neutron transport is neutron tracking and collisions
sampling. Typically, tracking is simulated with ray-tracing, the standard method, or by deltatracking,
introduced by Woodcock in 1965 (Woodcock et al., 1965). Some codes, i.e. Serpent,
use a combination of the two routines. However, in the past years several different
methodologies have been formulated and evaluated; some of those are Weighted Delta
Tracking (Morgan & Kotlyar, 2015), Direct Numerical Sampling (Brown & Martin, 2003),
Delta Tracking with arbitrary sampling cross section (Legrady et al., 2017), and hybrid
methods which combine two or more of the listed methods (Rehak et al., 2019). Some of these
algorithms were proven to perform better than the standard consolidated methods. Abundant
literature on tracking algorithms is available, and it will constitute the main body of the
project's literature review.

In the frame of this project, at first an extensive literature review on existing tracking
methodologies will be carried out. Then, the most promising ones will be implemented in the
code SCONE. SCONE is an in-house Monte Carlo code, developed as part of two PhD projects
in the Nuclear Group at Cambridge University. Due to its modular structure, it is particularly
adapt to testing and modification. Several existing tracking routines can be tested in SCONE
and their performance compared.
Following, a novel method, which has never been applied for Monte Carlo neutron transport,
can be investigated. This method relies on a single kernel model for the sampling of outgoing
particles, angles and energies after a collision. The algorithm includes the creation of material
specific tables, which collect all the possible collision outputs and the respective weighted
probabilities. Those tables must be created once, before the start of the neutron transport
simulation, for each material in the geometry. The new algorithm can be implemented in
SCONE, and validated against analytical or experimental benchmarks as well as compared
with existing methods. The precision and efficiency of the new method can be evaluated in
terms of uncertainties produced and computational time. This algorithm may potentially
accelerate Monte Carlo neutron tracking, in geometries where a limited number of materials
are present.

This project will contribute to the field of high-precision nuclear simulations through the
implementation of new, possibly improved Monte Carlo neutron tracking techniques. If
successful, such algorithms can accelerate Monte Carlo simulations, broadening their range of
application.