Kevin Clarno

• A commercial camera records Cherenkov light for reactor power monitoring. • Video-derived blue intensity is synchronized with reactor console data. • Nonlinear calibration corrects sensor saturation at high reactor power. • Cherenkov-derived estimates reduce high-power noise versus detectors. • Optical monitoring provides a spatially integrated complementary signal.

Digital twin technologies, which couple physical systems with their virtual representations for monitoring, prediction, and analysis, have advanced significantly in many engineering domains. However, their deployment in nuclear systems remains limited due to stringent requirements on safety, validation, and system-level integration. Existing efforts often focus on individual components such as high-fidelity simulation or data-driven models, while practical implementations that unify these elements into an accessible and operational framework remain scarce. In response, this work presents the Nuclear Twins Website (NTW), a web-based digital twin platform developed for the NETL TRIGA reactor. The NTW is designed as a centralized and integrated hub that unifies data, simulation, and user interaction. The platform integrates five core components: core configuration analysis, operational data access, a natural-language-based data query interface, an interactive reactor simulator, and a high-fidelity prediction module. By connecting historical data, physics-based models, and user interaction, the NTW enables intuitive exploration, training, and analysis of reactor behavior. Rather than introducing a standalone software tool, this work demonstrates a structured approach for deploying nuclear digital twins that emphasizes integration, accessibility, and validation. The NTW provides a scalable foundation for bridging high-fidelity reactor simulations with practical operational and research workflows.

The Molten Salt Reactor Experiment (MSRE), operated at Oak Ridge National Laboratory in the 1960s, was a pioneering effort to demonstrate MSR technology and remains a cornerstone for MSR research. This study presents a full-resolution multiphysics model of the MSRE, resolving the entire reactor vessel and internal structures without porous-media or axisymmetric approximations. The workflow couples Monte Carlo neutron–photon transport (Serpent 2) with conjugate heat transfer and turbulent flow simulation (GeN-Foam within the foamForNuclear platform) on unstructured meshes, iterated to convergence with temperature and density feedback. Predicted effective multiplication factor and power fractions show good agreement with benchmark data, and velocity profiles in the volute, annulus, and core passages are compatible with experimental measurements and previous simulations within known uncertainties in geometry and operating conditions. Where design parameters are uncertain, the model demonstrates the ability to infer missing data from experimental observations, highlighting the potential for high-fidelity simulation techniques to support reactor start-up procedures and digital-twin applications. The model reproduces major experimental flow trends in the lower plenum while revealing three-dimensional structures that axisymmetric surrogates cannot capture. Some discrepancies remain, largely attributable to input uncertainties and limitations of RANS models under strong adverse pressure gradients, suggesting future extensions to LES turbulence modeling.

Molten Salt Reactors (MSRs) promise significant advantages over traditional light water reactors, including enhanced safety and improved fuel efficiency. Among MSR designs, those using fueled fluoride salts have the most extensive operational history and are the focus of this study. In these systems, understanding and controlling salt chemistry is essential to reactor safety and performance. Challenges include preventing the release of volatile uranium species, avoiding unintended metallic uranium deposition, and managing fission product speciation as burnup progresses. These issues impact heat exchanger fouling, component corrosion, and potential criticality fluctuations, which define operational limits. Given the complexity of operating MSRs’ chemical and isotopic inventory, accurate simulation is challenging, particularly with limited experimental data. This work introduces a simulation surrogacy methodology to simplify the reactor system by grouping chemically similar species based on available data, balancing computational efficiency with fidelity. The approach integrates SCALE reactor physics framework for modeling fuel salt depletion and fission product generation with the Molten Salt Thermochemical Database-Thermochemical (MSTDB-TC) and Thermochimica for Gibbs energy minimization to predict species behavior. The methodology is demonstrated for a simple MSR design at key points in its lifetime, addressing operational challenges and safety concerns. Key data gaps, including radiological, neutronic, and quantitative limitations, are identified. Although focused on fueled fluoride salts, this methodology is adaptable to other reactor designs, providing a versatile framework for advancing MSR chemistry simulations. While fundamentally exploratory due to current database limitations, this methodology provides a systematic framework for advancing MSR chemistry modeling and identifying critical data needs.

Abstract Regression is a fundamental problem in statistics and machine learning, where the true output is a continuous and stochastic function of the input. Regression methods model the output variables using training datasets composed of inputoutput pairs derived from real-world scenarios. Inputs are broadly categorized as known or unknown, where unknown inputs are often represented as additive noise. Most existing models rely on two key assumptions: Gaussian residuals and homoscedasticity. While foundational, these assumptions often limit the applicability of traditional machine learning methods. To address these limitations, this study employs deep neural networks, leveraging the Universal Approximation Theorem, which ensures their ability to approximate any continuous function given sufficient hidden units. Conventional neural network-based regression methods typically use MSE or MAE as their objective functions, adhering to Gaussian or uniform residual assumptions. However, such methods remain constrained by homoscedasticity and fixed residual probability density functions. Inspired by classification tasks, this study proposes an innovative approach that discretizes continuous regression outputs into bins and models the probability of each bin. The framework utilizes Softmax as the activation function in the output layer and cross-entropy as the objective function, akin to multi-class classification. A comparative analysis against linear regression, Gaussian processes, and conventional MSEbased DNNs is conducted on synthetic datasets with varying complexity and combined cycle power plant dataset. Results demonstrate that the proposed Softmax-based DNN outperforms aforementioned regression methods by effectively estimating conditional probability density functions without having to assume Gaussian or homoscedasticity residuals. The Softmaxbased DNNs pave the way for a new class of regression models that can more effectively handle the complexities of real-world data.

<ns5:p>This technical report aims to provide energy developers and policymakers with information and preliminary analyses on the potential for Texas coal plant sites to be repurposed for nuclear power. Investigation into coal-to-nuclear (C2N) has shown that constructing a nuclear reactor on the site of a retired coal plant has both economic and environmental benefits. The data presented includes operational details of the coal power plants, the presence of nearby hazards, geological and hydrological data, and population statistics. This information was gathered for 19 coal powered electricity generation sites in Texas. Thirteen of the sites assessed have no hazards or other factors that would disqualify them from hosting a Small Modular Reactor (SMR). Of these, 11 sites are also suitable for a Light Water Reactor (LWR). The smaller size and power output of SMRs makes these additional 2 sites possible, even near a population center. The remaining 6 coal plant sites would require more specific on-site analysis or potential adjustments to the reactor design to be considered for licensing. Information supporting these findings are presented here in detail and are intended to give interested parties easy access to relevant C2N conversion information. These results can be utilized to make informed decisions on this strategy for nuclear development and its benefits to grid stability and baseload power in Texas.</ns5:p>

In this analysis we examine the economic costs and benefits of installing dozens of small modular reactors at recently retired coal power plants in Texas to determine the viability of a grid or “program-level” approach to nuclear power plant planning in the United States. Previous studies have indicated that utilizing stranded infrastructure assets at retired coal power plants, known as the “coal-to-nuclear” transition, could greatly reduce the amount of time and capital required to build just a single commercial nuclear plant. A discounted cash flow analysis was created using data from regional electricity markets, coal-to-nuclear studies, and other industry sources to estimate the potential value of SMR projects. The analysis includes multiple scenarios to account for varying project sizes, changes in technology learning rates, and recently implemented energy tax credits. Results indicate that increasing the rate of learning has only a minimal lowering effect on the Levelized Cost of Electricity (LCOE), as both the learning rate and LCOE quickly plateau. The most significant cost reductions were enabled by tax credits and coal-to-nuclear cost enhancements that decreased the LCOE to a competitive range of $43–52/MWh. However, our work finds that program-level benefits will likely be the result of cost sharing and risk modularization rather than direct improvement in metrics like LCOE and net present value as those are still smaller in comparison. • Economic analysis to determine viability of program-level nuclear plant planning. • SMRs can be financially attractive replacements for recently retired coal power plants. • Cost savings from reusing coal power plant assets are especially significant. • Detailed cash flow analysis includes federal energy tax credits and learning rates. • Resulting LCOE and NPV indicate competitiveness with renewable energy generation sources.

<ns3:p>This technical report aims to provide energy developers and policymakers with information and preliminary analyses on the potential for Texas coal plant sites to be repurposed for nuclear power. Investigation into coal-to-nuclear (C2N) has shown that constructing a nuclear reactor on the site of a retired coal plant has both economic and environmental benefits. The data presented includes operational details of the coal power plants, the presence of nearby hazards, geological and hydrological data, and population statistics. This information was gathered for 19 coal powered electricity generation sites in Texas. Thirteen of the sites assessed have no hazards or other factors that would disqualify them from hosting a Small Modular Reactor (SMR). Of these, 11 sites are also suitable for a Light Water Reactor (LWR). The smaller size and power output of SMRs makes these additional 2 sites possible, even near a population center. The remaining 6 coal plant sites would require more specific on-site analysis or potential adjustments to the reactor design to be considered for licensing. These findings exhibit the potential for cost effective nuclear development to benefit grid stability and provide baseload power to Texas.</ns3:p>

Monte Carlo-transport codes are designed to simulate the complex neutron transport physics associated with nuclear systems. Numerous approximations must be made by modern simulation codes, and many have been well validated. This article investigates an impactful simulation error caused by an approximation made in many Monte Carlo-transport codes. The approximation that target-at-rest is valid for neutrons at energies 400 times that of the thermal energy of the target particle is found to be inaccurate in certain scenarios. This paper identifies such cases, notably TRISO fuel and instances where fuel infiltrates the pores of graphite in Molten Salt Reactors. When threshold values are too small, resonance escape probabilities can deviate by as much as 1% per resonance, forming a baseline defect. The following are the authors recommendations: incorporate the threshold value as a user-defined variable and conduct convergence studies to ensure that the chosen threshold value is sufficiently high.

<ns3:p>Background Molten salt reactors, and other types of circulating, liquid fueled, nuclear reactors contain a certain amount of gas entrained in their liquid nuclear fuel. This gas induces an effect on the nuclear and dynamical behavior of the reactor as a whole. Gas voids respond to variation in temperature and pressure differently than liquids. When the gas voids in the reactor working fluid expand, the nuclear fuel is pushed from the core. Likewise, when the gas voids contract, more nuclear fuel enters into the core. Methods This paper examines the interplay of gas void fraction and reactivity in a molten salt reactor, and attempts to elucidate the dynamical response of the void fraction and the reactivity of the system to perturbation in system temperature, pressure, and gas quantity. A theory is presented that aims at describing the relationship between reactivity and gas behavior. This theory is then applied to the Molten Salt Research Reactor (MSRR) design, a facility currently under construction at Abilene Christian University campus. Results A result of this paper is the temperature and void fraction parameterized gas coefficients of reactivity for the Molten Salt Research Reactor. Conclusions The presence of voids accounts for 5-30% of the total temperature coefficient of reactivity, demonstrating their non-trivial contribution. Additionally, the study emphasizes the importance of considering gas content in MSR physics, especially in the context of pressure transients and system reactivity during pump trips. The initial system pressure, particularly in designs like the MSRR operating at sub-atmospheric pressures, is crucial due to its influence on reactivity changes during rapid pressure increases.</ns3:p>

The International Reactor Physics Benchmark Experiment’s MSRE benchmark provides zero-power critical validation data that has been used to assess the accuracy and consistency of SCALE, MCNP, and Serpent for design and licensing of a modern MSR. The codes were used to model the benchmark and a wide range of variations in the geometry, nuclear data library, code versions, and problem specifications. Most of these cases demonstrated excellent consistency, within two standard deviations of the stochastic error, for the reactivity and associated reactivity worth of the variation. However, the codes over-predicted the initial criticality of the MSRE by 3%, which is larger than the provided experimental uncertainty. It was shown that care must be taken to ensure that the bound scattering treatment is consistent with comparing codes. It was also shown that for the MSRE, the use of the ENDF/B-VII.0 library introduced a significant increase in the reactivity (>200 pcm). Code predictions for two coefficients of reactivity (fuel and isothermal) were validated with data from the MSRE and were well within the experimental uncertainty, thus providing confidence in each code to provide accurate data for safety analysis calculations. These results provide quantifiable estimates of the computational variation one could expect due to the choice of any one of these codes, or a particular nuclear data library, for both initial criticality and reactivity coefficients.