Criticality experiments are experiments where fissile material is arranged in a way that generates meaningful and measurable neutron multiplication. These experiments can be a subcritical bare sphere of material or can scale all the way up to a full scale demonstration of a reactor. Criticality experiments are typically low power (a few kW and below), although burst experiments can have instantaneous powers of many GW for a very short amount of time. A typical criticality experiment is used to measure the reactivity of the system or other nuclear parameters. At NCERC, we often use the term “criticality experiments” over “critical experiments” as “critical” implies a k_eff of 1.0, whereas the experiments we perform can range from very subcritical to critical and above prompt supercritical. Each of the critical assemblies at NCERC use different means of reactivity control and various detectors so that important parameters such as reactivity and delayed critical configurations can be measured with high accuracy.

The main sponsor for NCERC experiments is the Nuclear Criticality Safety Program (NCSP). NCERC (and LACEF before that) has performed many benchmark critical and subcritical experiments which are now part of the International Criticality Safety Benchmark Evaluation Project handbook (ICSBEP) (OECD-NEA, 2022). The ICSBEP benchmarking process is extremely rigorous; each benchmark includes dimensions of each component, material compositions, temperature information, isotopics for fuel and other relevant components, and uncertainties for each of these values. These uncertainties are used to estimate the uncertainty in the experiment, the uncertainty in the benchmark model, and biases between the experiment and benchmark model. Due to their high quality, these benchmarks are crucial for validating nuclear data and ensuring nuclear criticality safety. Major nuclear data libraries such as ENDF/B-VIII.0, JEFF-3.3, and JENDL-4.0 make extensive use of criticality benchmarks for validating nuclear data and benchmarking the performance of the nuclear data library compared to other libraries (van der Marck, 2006, 2012; CHIBA et al., 2011; SHIBATA et al., 2011; Brown et al., 2018; Plompen et al., 2020). In fact, nuclear data libraries are sometimes adjusted to match these benchmarks, for example, nubar for ²³⁹Pu and ²³⁵U respectively were adjusted to match k_eff of Jezebel and Godiva in ENDF/B-VIII.0 (Peterson, 1953; Peterson and Newby, 1956; LaBauve, 2002; Favorite, 2016; Brown et al., 2018; Chadwick et al., 2018; Kimpland et al., 2021).

Recently, NCERC has performed a number of experiments related to nuclear energy and advanced reactors, some of which will be covered in Section 4. Given NCERC’s unique capabilities, experiments can be performed on relatively accelerated timelines, and often without major changes to the facility. This allows NCERC to fill an interesting niche for performing criticality testing on fuels and components, and even small scale, low power demonstrations of advanced reactor designs.

In 2012, the Demonstration Using Flattop Fission (DUFF) experiment was performed (Poston, 2013). This experiment used the Flattop assembly with a heatpipe inserted in the center of the core attached to a Stirling engine. This experiment was the predecessor to the KRUSTY/Kilopower experiment, which was performed in late 2017 and early 2018.

The KRUSTY/Kilopower experiment was a demonstration of a space nuclear reactor being developed with NASA. The experiment itself was a collaboration between NASA, NCSP, and NNSA. The motivation was to ensure high reliability by minimizing moving parts as much as possible, while also maximizing the power to weight ratio. The final experiment used an HEU core about the size of a roll of paper towels, surrounded by a BeO reflector, on the Comet critical assembly (seen in Figure 10). The HEU core was new and designed specifically for this experiment and manufactured at Y-12. Eight heatpipes were also connected to the core, which were cooled by Stirling engines and simulated Stirling engines. To simulate space conditions, the core, heatpipes, and Stirling engines were enclosed in a large vacuum chamber.

Weeks of tests were performed to bring the system critical and measure reactivity of components. The final test included a 28 h “high power” test, averaging 3 kW of power and an average core temperature of 800°C while undergoing many transients (startup, loss of cooling, reactivity insertions, shutdown, etc.). These tests showed the capability for the system to load-follow (match power to a desired electrical load) with no operator input. Much more has been written about this experiment in a special issue of Nuclear Technology, along with other publications (McClure et al., 2020b; Poston et al., 2020a; Gibson et al., 2020; McClure et al., 2020a; Sanchez et al., 2020; Grove et al., 2020; Poston et al., 2020b, Poston et al., 2020c; Stolte et al., 2022). The KRUSTY experiment has been approved and has been included as a benchmark in the 2021 ICSBEP Handbook and will be useful for validating Be nuclear data.

Hypatia was an experiment conducted in early 2021 that was focused on performing a heated critical experiment with yttrium-hydride (YH_x). YH_x has been proposed by many as a potential neutron moderator in micro-reactors, due to its high hydrogen density and low hydrogen loss even at high temperatures (Trellue et al., 2021). However, until the Hypatia experiment, there had never been a critical experiment performed with YH_x.

There was a large effort at LANL to develop YH_x samples large enough to work in a critical experiment (Trellue et al., 2022). Once the YH_x samples were produced, they were sealed and shipped to NCERC. The Hypatia core was a combination of HEU disks (C-Disks), graphite pieces, electric heaters, and YH_x samples, all surrounded by Be as a reflector and moderator.

During this test, the YH_x samples were heated to roughly 300°C and reactivity was measured as a function of temperature. This was the first experiment to demonstrate and measure the positive temperature coefficient of reactivity for YH_x, meaning that as the temperature of the YH_x increased, the reactivity also increased. This effect has been seen before in ZrH_x in the Topaz reactor (Buden, 1993), and was predicted based on new thermal scattering laws for YH_x in ENDF/B-VIII.0 (Zerkle and Holmes, 2017; Zerkle et al., 2021; Brown et al., 2018) and new experimental data on YH_x scattering (Mehta, 2020; Mehta et al., 2022). The positive temperature coefficient of reactivity for YH_x may have design implications for advanced reactors seeking to utilize this moderator (Ade et al., 2022). Additional information on the Hypatia experiments can be found in publications by Trellue and Cutler (Trellue et al., 2021, Trellue et al., 2022; Cutler et al., 2022b).

The Measurement of Uranium Subcritical and Critical (MUSiC) experiment was performed in 2020 and 2021 at NCERC using the “Rocky Flats Shells” (McSpaden et al., 2019a; Weldon et al., 2021, Weldon et al., 2020; Darby et al., 2021; McSpaden et al., 2021). Figure 11 is a photo of the experimental setup. The goal of the experiment was to perform high quality measurements of HEU systems with many detector types, from very subcritical to critical. Since the Rocky Flats Shells are nesting HEU hemishells, systems of different size spheres of HEU were made by varying which shells were used in the experiment. The MUSiC experiment included some measurements with a²⁵²Cf source, some with a pulsed deuterium-tritium neutron generator, and some with no external neutrons. Similar subcritical measurements are often made at NCERC, but this was the first time at NCERC a system like this was measured from subcritical to critical. Two MUSiC critical experiments will be submitted as benchmarks to the ICSBEP Technical Review Group in 2023. Other subcritical experiment benchmarks will follow in the years to come. Ultimately, this data will be used by nuclear data evaluators to validate ²³⁵U nuclear data in the fast energy region, which is extremely important for many advanced reactor designs. Additionally, due to uncertainties in the original “Lady Godiva” benchmark, MUSiC may become the new standard for ²³⁵U validation (Bess et al., 2013; Hutchinson et al., 2022b).

From 2015 to 2019, a series of experiments in collaboration with the Japan Atomic Energy Agency (JAEA) was performed, specifically focused on lead void coefficients of reactivity. The JAEA was interested in created an Accelerator Driven System (ADS) for spent fuel transmutation. Three configurations were built and operated, a Plutonium system (also known as Jupiter), an HEU system, and an effectively Intermediate Enriched Uranium (IEU) system (Fukushima et al., 2018, Fukushima et al., 2019). The “IEU” system combined natural and HEU Jemima plates to create an “effective enrichment” of 22.9 weight percent. Each of the systems had a large amount of lead (Pb), and multiple experiments were performed where the amount of lead in the system was varied while keeping the fuel concentration constant. This allowed for measurements of the void coefficient of reactivity of lead for each system. The HEU/Pb and IEU/Pb systems used the Jemima plants for fuel, and the Pu system (Jupiter) used ZPPR plates. There was also a Jupiter configuration which utilized ZPPR plates with higher ²⁴⁰Pu content. The HEU/Pb system had a negative lead void coefficient of reactivity, but the IEU/Pb had a positive lead void coefficient; this behavior was predicted in MCNP simulations, but it highlights the importance of demonstration experiments to prove that models are performing correctly (Goda et al., 2015; Goda et al., 2019, Thompson et al., 2021b). Work is ongoing to submit these experiments to the ICSBEP as benchmarks (McSpaden et al., 2019b, McSpaden et al., 2020; Amundson et al., 2020).

Experiments Underpinned by Computational Learning for Improvements in Nuclear Data (EUCLID) is a LANL Laboratory Directed Research & Development project focused on identifying and resolving compensating errors in nuclear data (Hutchinson et al., 2022a; Neudecker et al., 2022a; Neudecker et al., 2022b; Clark et al., 2022; Kleedtke et al., 2022; Rising and Clark, 2022). As part of this project, an experiment series at NCERC was designed using machine learning tools. This experiment was specifically focused on ²³⁹Pu scattering using two different configurations: one that maximizes neutron leakage, and one that minimizes leakage. Because of the large differences in leakage, there are large differences in sensitivities to both elastic and inelastic scattering. Even for a well known isotope such as ²³⁹Pu, different nuclear data libraries (ie. ENDF/B-VIII.0 and JEFF-3.3) have very different ²³⁹Pu elastic and inelastic scattering cross sections. This experiment should assist in resolving some of the issues between nuclear data libraries for ²³⁹Pu and this methodology of experiment design will be used for other experiments in the future.

NCERC is currently partnering with Westinghouse to potentially perform a small scale demonstration of their eVinci microreactor. A portion of the eVinci core will be placed on the movable platen of Comet, and surrounded by graphite fueled with the CNPS fuel. The setup will be designed to accommodate measurements in the temperature range of 20–1,200°C and will contain materials to be tested. The current plan includes 1/M approach to criticals for the different configurations, component critical tests (testing reactivity of various components), and heated tests with various core configurations. This data will be used by Westinghouse to validate their simulation codes, reduce uncertainties, and to help support their safety case to their regulator. It is to be determined whether this data will stay proprietary or will become an official benchmark in the ICSBEP or IRPhEP handbooks (OECD-NEA, 2022, OECD-NEA, 2019).

The KRUSTY/Kilopower experiment summarized in Section 4.1 is a great example of a low power demonstration that can be performed at NCERC. However, one downside to the KRUSTY experiment was neutron activation; even with shielding around the core, the room remained unusable for other experiments for months. Given the high demand for the NCERC facility, being unable to operate key critical assemblies for over a month is not currently an option. As a lesson learned from KRUSTY, experiments in the future will have to be lower power, which often necessitates electric heating as opposed to nuclear heating.

There is still an enormous amount of information that can be gained from low power demonstrations, such as reactor physics parameters, system reactivity, reactivity of components, temperature coefficients, flux distributions, neutron spectra via activation foils, leakage spectra, multiplicity, shielding performance, control rod/drum worth, and much more. These are exactly the types of measurements being planned for the Westinghouse eVinci demonstrations.

Today many of these parameters are simulated without adequate validation data, especially for novel systems such as advanced reactors that often plan to use new coolants, fuels, moderators, or other materials. Additionally, there are even fewer validation experiments at the high temperatures many of these reactors are planning to operate at. It is important we make high quality measurements of these systems and materials so that when these systems are actually built, they operate as they are designed.

Nuclear data is extremely important for advanced reactors, and in some cases, the quality of the nuclear data today is still lacking. One rather important example that has gotten a lot of attention is the ³⁵Cl (n,p) cross section at high energies. As was shown by Oak Ridge National Laboratory, TerraPower, and Moltex Energy, the ³⁵Cl (n,p) has a huge impact on the reactivity of Molten Chloride Fast Reactors (Bostelmann et al., 2020; Kolos et al., 2022; Taylor et al., 2022). In fact, changes in recent nuclear data libraries have resulted in differences of thousands of pcm. Additionally, uncertainty/covariance data for ³⁵Cl (n,p) in ENDF/B-VIII.0 is missing above ∼1 MeV, meaning some analysis techniques which rely on covariance data to estimate nuclear data uncertainties will not attribute any nuclear data uncertainty to this reaction. In this way, covariances can be just as important as the underlying cross sections. This topic has already generated multiple papers and a request to the NEA Nuclear Data High Priority Request List.

This is an example of a material and energy range that could be targeted with a well designed critical experiment at NCERC. In fact, NCERC recently completed the Chlorine Worth Study (CWS) experiment (Cutler et al., 2022a) - this experiment was focused on the reactivity worth of chlorine at low energies, specifically related to aqueous chloride operations. While this CWS experiment is not very helpful for validating the ³⁵Cl (n,p) cross section in the fast energy region, another measurement could be designed to test just that. As was mentioned before, many experiments at NCERC are designed specifically to validate nuclear data.

Another relevant example of an experiment designed to validate nuclear data is the Critical Unresolved Region Integral Experiment (CURIE). CURIE was originally designed to validate the unresolved resonance region (URR) of ²³⁵U (Cutler et al., 2018). Around the same time CURIE was taking place, the Nuclear Data Machine Learning team lead by Denise Neudecker at LANL found some rather large discrepancies in nuclear data for ¹⁹F (Neudecker et al., 2020). CURIE used the large copper reflectors form ZEUS and teflon (PTFE) moderator plates to create an energy spectrum highly sensitive to the URR of ²³⁵U, making CURIE also highly sensitive to ¹⁹F in this energy region (Cutler et al., 2020; Thompson et al., 2021a). Because of this, CURIE is currently being used by members of the Cross Section Evaluation Working Group (CSEWG) and the International Nuclear Data Evaluation Network (INDEN) to validate and improve ²³⁵U, ¹⁹F, and ^63,65Cu nuclear data (Trkov and Capote, 2022). Work is ongoing to submit this experiment as an ICSBEP benchmark.

Many other experiments have been performed at NCERC specifically focused on nuclear data validation, including the TEX HEU and TEX Pu series, the Zeus series, and others (Percher and Norris, 2019; Percher and Norris, 2020, Norris and Araj, 2021; Mosteller et al., 2004; Mosteller, 2005; Hayes and Sanchez, 2006). These experiments are detailed in the NCERC Ten Year Papers and other publications (Hutchinson et al., 2021b; Thompson et al., 2021b; Goda et al., 2021; Hayes et al., 2021; Sanchez et al., 2021).