PROJECTS

Monte Carlo Neutron Collimator & Shielding Simulation

NSE 233 Final Project — Computational Physics

Executive Summary

This research evaluates the efficacy of two common nuclear materials—Lead (Pb) and High-Density Polyethylene (HDPE)—as shields in a 2D neutron collimation system. Using a custom 2D Monte Carlo particle transport simulation written in Python, I tracked 300,000 neutron histories per material.

The study demonstrates that while both materials deliver equivalent collimated beam transmission fractions (~0.0044), HDPE displays a vastly superior capability for structural shielding. HDPE absorbed nearly 100 times more neutrons inside the shielding blocks and deposited approximately 3 times more energy than the lead shield. This makes HDPE the superior material for collimator shielding, protecting secondary reactor and detector systems.

Problem Statement & Geometry

The objective is to collimator-align isotropic, uniform neutrons escaping a rectangular source, forcing them into a narrow beam, while absorbing neutrons heading into the peripheral shielding blocks.

We define a coordinate system where (0,0) is the center-back of the neutron source. Below is the layout of the boundaries and shielding zones:

+---------------------------------------------------------------+ (System Upper Limit: y_s/2 + w_coll) | | | +---------------------------------------------+ | | | | | | | SHIELDING ZONE 1 (Top Shield) | | | | | | | SOURCE +--------+ +--------+ DETEC | | ZONE | | COLLIMATOR CHANNEL | | TOR | | (0,0) | | (Free space path) | | | | +--------+ +--------+ | | | | | | | SHIELDING ZONE 2 (Bottom Shield) | | | | | | | +---------------------------------------------+ | | | +---------------------------------------------------------------+ (System Lower Limit: -y_s/2 - w_coll) 0 ----- x_s --- x_coll ----------------------------- x_det (Boundary)

The boundaries are defined algebraically:

The Monte Carlo Transport Method

A history-based particle tracker simulates individual neutrons. Every neutron is initiated with uniform coordinates within the source zone and a random isotropic emission angle \(\mu \in [0, 2\pi]\).

The interaction length \(l\) (the distance traveled between successive collisions) is computed via rejection sampling. In the shielding medium, the probability density distribution of the travel path is governed by the total macroscopic cross section:

P(l) = Σ_t * exp(-Σ_t * l)
where Σ_t is the total macroscopic cross section (Σ_t = Σ_s + Σ_a)

At each collision coordinate, a Monte Carlo decision step determines the interaction type:

Comparison & Findings

Below is the tabular data compiled from simulating 300,000 neutron histories for both High-Density Polyethylene and Lead:

Parameter HDPE (High-Density Polyethylene) Lead (Pb)
Total Neutrons Simulated 300,000 300,000
Fraction Exiting Collimator Channel (C) 0.00436 0.00435
Fraction Absorbed in Shielding 0.06348 0.00124
Total Count Absorbed in Shielding 41,015 430
Fraction Leaking via Sides 0.49808 0.51167
Fraction Exiting Back of Source 0.26081 0.25283
Avg. Energy Deposited per Source Particle 1.57362 MeV 0.41667 MeV

The comparison highlights that lead performs poorly as a neutron shield. Lead is dense and effective against gamma rays, but its cross-section for neutron absorption and scattering is low. Neutrons traverse the lead shield with fewer collisions, escaping with higher kinetic energies.

Conversely, HDPE contains high concentrations of hydrogen and carbon atoms. Light hydrogen nuclei are highly efficient at slowing down neutrons (elastic moderation). As a result, neutrons inside the HDPE shield undergo a massive number of thermalizing collisions, which spreads out the spectrum and results in 100 times higher absorption and 3 times higher energy deposition.

Simulation Plots & Output

Below are the generated spectral distributions and computation behaviors extracted directly from the python code outputs:

Execution Performance

The python codebase was designed modularly and includes both serial (linear) and parallelized loops (using multiprocessing pools) to speed up the 300,000 history calculations.

Interestingly, the code exhibits a performance quirk: for HDPE, the linear loop executes in ~20 seconds, whereas parallelized overhead doubles the runtime to ~40 seconds. However, when simulating Lead (where shielding interactions are sparse and particles escape or exit the system much faster), the parallel loop is significantly faster.

Overall, the project successfully bridges nuclear reactor physics concepts with computational algorithms, validating that light, hydrogenous materials like HDPE are critical components in collimator and collimator-shielding systems.

Return to Main Portfolio