UC Berkeley

At its core, research represents an attempt to break from the "this is the way we have always done it" paradigm. This idea is evidenced from the start in this research effort by the problem formulation to develop a new way to generate synthetic debris that mimics the samples that would be collected for forensics purposes following a nuclear weapon attack on the U.S. or its allies. The philosophy is also demonstrated by the design methodology used to solve the synthetic debris problem, using methods not commonly applied to nuclear engineering problems. Through this research, the bounds of what is deemed possible in neutron spectral shaping are moved ever so slightly.

A capability for the production of synthetic debris and fission products was developed for the National Ignition Facility (NIF). Synthetic debris has historically been made in a limited fashion using sample doping techniques since the cessation of nuclear weapons testing, but a more robust alternative approach using neutron spectral shaping was proposed and developed by the University of California-Berkeley and Lawrence Livermore National Laboratory (LLNL). Using NIF as a starting source spectrum, the energy tuning assembly (ETA) developed in this work can irradiate samples with a combined thermonuclear and prompt fission neutron spectrum (TN+PFNS). When used with fissile foils, this irradiation will produce a synthetic fission product distribution that is realistic across all mass chains.

To design the ETA, traditional parametric point design approaches were discarded in favor of formal optimization techniques. Finding a lack of suitable algorithms in the literature, a metaheuristic-based optimization algorithm, Gnowee, was developed for rapid convergence to nearly globally optimum solutions for complex, constrained engineering problems with mixed-integer and combinatorial design vectors and high-cost, noisy, discontinuous, black box objective function evaluations. Comparisons between Gnowee and several well-established metaheuristic algorithms are made for a set of continuous, mixed-integer, and combinatorial benchmarks. These results demonstrated Gnoweee to have superior flexibility and convergence characteristics over a wide range of design spaces.

The Gnowee algorithm was implemented in Coeus, a new piece of software, to perform optimization of design problems requiring radiation transport for the evaluation of their objective functions. Currently, Coeus solves ETA optimization problems using hybrid radiation transport (ADVANTG and MCNP) to assess design permutations developed by Gnowee. Future enhancements of Coeus will look to expand the geometries and objective functions considered to those beyond ETA design.

Coeus was used to generate an ETA design for the TN+PFNS application on NIF. The design achieved a reasonable match with the objective TN+PFNS and associated fission product distributions within the size and weight constraints imposed by the NIF facility. The ETA design was built by American Elements, and initial validation tests were conducted at the Lawrence Berkeley National Laboratory's 88-Inch Cyclotron. These experiments used foil activation and pulse height spectroscopy to measure the ETA-modified spectrum. Additionally, pulse height spectroscopy measurements were taken as the ETA was built-up component-by-component to measure the impact of nuclear data on the ability to model the ETA performance. Some initial analysis of these results is included here.

Finally, an integral validation experiment on NIF was proposed using the Coeus generated ETA design. A scoping study conducted by LLNL determined the proposed experiment and ETA design are within NIF facility limitations and current radio-chemistry capabilities. The study found that the proposed ETA experiment was "low risk," has "no show stoppers," and has a "reasonable cost." All that is needed is a sponsor to close the last funding gap and bring the experiment to fruition.

This research broke with the current sample doping approach and applied neutron spectral shaping to design an ETA that can create realistic synthetic fission and activation products and improve technical nuclear forensics outcomes. However, the ETA presented in this research represents more than a stand alone point design with a limited scope and application. It is proof of a concept and the product of a unique capability that has a wide range of potential applications.

This research demonstrates that the concept of neutron spectral shaping can be used to engineer complex neutron spectra within the confines of physics. There are many possible applications that could benefit from the ability to generate custom energy neutron spectra that fall outside of current sources and methods. The ETA is the product of a general-purpose optimization algorithm, Gnowee, and design framework, Coeus, which enables the use of Gnowee for complex nuclear design problems. Through Gnowee and Coeus, new ETA neutronics designs can be generated in days, not months or years, with a drastic reduction in the research effort required to do so. Most importantly, the new designs could be for applications completely disconnected from the current research and do not have to even be ETA designs at all. The capability of Gnowee and Coeus have the potential to greatly improve the design process in many fields of nuclear engineering.