Modernising neutron spectrum unfolding for fusion applications

Neutron spectrum unfolding with activation foils is a technique uniquely suited to measuring the neutron spectra of fusion reactors due to its passive nature. However, more research is needed to consolidate its use as a potential diagnostic tool for future fusion power plants, as there is much room for optimization for its use in nuclear fusion applications. This PhD scrutinised this technique thoroughly and improve upon it where possible. Separating the project into two parts, this thesis first studied the mathematics of unfolding problems, exploring how its characteristics changes with the degree of determination. Gaps in the rigour and suitability of the existing algorithms were identified, and to bridge these gaps, new unfolding algorithms were created, and they were shown to perform well on some example synthetic and real-world data. These algorithms leverage modern advances in linear algebra packages and improvement in gradient optimization algorithms, producing versatile algorithms that are quick to converge and allows for uncertainties to be analytically propagated. These new algorithms, along with some commonly used existing unfolding algorithms, were implemented into a Python module known as the unfoldingsuite. Then, the process of designing activation foil neutron spectrum unfolding experiments was examined. A deficiency was identified, namely, there is currently no standardised method of selecting foils. Therefore, a procedure of choosing activation foils was created and formalised into the foil selector framework, in which the accuracy and precision of the expected unfolded neutron spectrum were simultaneously maximised, using a discrete multi-objective optimization algorithm. This framework was then implemented as a Python program known as the foil selector program, which was used to select foils for an experiment at a high-energy neutron beamline. The results of this experiment was presented to examine the strengths and shortcomings of the foil selector program, and of activation foil neutron spectrum unfolding in general. Directions of future development of the foil selector framework and program were identified as a result. Keywords:foil selection, discrete multi-objective optimization, unfolding algorithms, fusion neutronics, inverse problems.