Logo: to the web site of Uppsala University

This work introduces a comprehensive and statistically robust framework for quantifying model parameter uncertainties in nuclear fuel performance predictions, ensuring that all sources of discrepancy (e.g., measurement noise, parameter variability, and model inadequacy) are systematically captured. Parameter uncertainties are essential inputs to safety analyses and are typically obtained through calibration; however, standard Bayesian calibration often underestimates uncertainty when the underlying model cannot fully reproduce experimental observations. To overcome this limitation, we employ a hierarchical probabilistic formulation in which model parameters vary according to distributions governed by both local experiment-specific and global hyperparameters, thereby explicitly accounting for unmodeled variability and structural deficiencies in the model. This paper demonstrates the practical implementation of this approach using an efficient Metropolis-Hastings-within-Gibbs sampling to calibrate a fission gas release model in the fuel performance code TRANSURANUS against Westinghouse's commercial database. The resulting posteriors provide uncertainty estimates that credibly span all code-to-measurement discrepancies, remain robust across linear and nonlinear models, and directly support safety evaluations by offering defensible uncertainty margins. This approach establishes a generalizable pathway for uncertainty quantification in fuel performance modeling and enhances confidence in predictive simulations to inform high-consequence decisions.

Fuel performance modeling involves complex, nonlinear systems with uncertain parameters that require rigorous calibration against experimental data to improve their prediction potential. In the presence of model bias, standard Bayesian calibration, typically combining Gaussian Process surrogates with Markov Chain Monte Carlo algorithms, may underestimate parameter uncertainty. This study compares standard Bayesian calibration with a hierarchical approach that moderates posterior uncertainty reduction by inflating parameter uncertainty, thereby reducing overconfidence in inferred parameters when model bias is present. Using Gaussian Process surrogates of the OpenFOAM Fuel BEhavior Analysis Tool solver to accelerate computationally intensive evaluations and Morris' screening for dimensionality reduction, we apply both calibration methods to a synthetic fuel performance dataset. The synthetic data are constructed to be representative of realistic fuel performance measurements while retaining known true parameter values, enabling an objective assessment of calibration accuracy and uncertainty quantification. Results show that hierarchical calibration outperforms standard techniques in the presence of systematic error, accurately recovering the true parameter value and consistently encompassing the synthetic experimental observations. The proposed framework provides a robust approach for quantifying uncertainties in complex fuel performance models. The paper also discusses challenges related to calibration performance, convergence diagnostics, and algorithm tuning, and outlines future enhancements, including advanced Markov Chain Monte Carlo methods and treatment of time-dependent outputs.

The pp -> W-+/-(-> mu(+/-) nu(mu)) X cross-sections are measured at a proton-proton centre-of-mass energy root s = 5.02TeV using a dataset corresponding to an integrated luminosity of 100 pb(-1) recorded by the LHCb experiment. Considering muons in the pseudorapidity range 2.2 < eta < 4.4, the cross-sections are measured differentially in twelve intervals of muon transverse momentum between 28 < p(T) < 52 GeV. Integrated over p(T), the measured cross-sections are sigma(W+) (->mu+ (nu) over bar mu) = 300.9 +/- 2.4 +/- 3.8 +/- 6.0 pb, sigma(W-) (->mu- (nu) over bar mu) = 236.9 +/- 2.1 +/- 2.7 +/- 4.7 pb, where the first uncertainties are statistical, the second are systematic, and the third are associated with the luminosity calibration. These integrated results are consistent with theoretical predictions. This analysis introduces a new method to determine the W-boson mass using the measured differential cross-sections corrected for detector effects. The measurement is performed on this statistically limited dataset as a proof of principle and yields m(W) = 80369 +/- 130 +/- 33 MeV, where the first uncertainty is experimental and the second is theoretical.

Fuel performance codes, such as Transuranus, predict fuel behavior and are used to ensure the safe operation of nuclear reactors. These codes are moderately time-consuming and affordable in many applications but may be limited in others, primarily when many fuel rods must be evaluated simultaneously. This work presents how the temporal neural network techniques, Temporal Convolutional Networks, and a Fourier Neural Operator can be combined to form a deep heterogeneous joint architecture as a surrogate model for fuel performance modeling in time-critical situations. We train the model using realistic power histories and corresponding outputs generated using the fuel performance code Transuranus. The ultimate result is a surrogate model for use in time-critical situations that take milliseconds to evaluate for thousands of fuel rods and have a mean test error of unseen data around a few percent.

The JET neutron camera is used to monitor a 2D profile of the neutron emission from the plasma, using 19 sightlines with plastic scintillators suited for measuring neutrons from the D + T → n + ⁴He (DT) reaction. This paper describes an independent, first-principles physics method for estimating the volume integrated DT neutron yield with the neutron camera. This is performed for a selection of shots from the two recent DT campaigns at JET, the DTE2 and DTE3 JET campaigns. It covers the data reduction methods from a light yield calibration of the scintillators to treatment of pile-up, which is prevalent during high yield DT experiments. Several models of the camera geometry are used to account for scattering and transmission effects in the neutron transport. The neutron yield is estimated using models of the neutron emission profile, which are fitted to measurement data. The neutron yield estimates from this method are compared to corresponding estimates from the JET fission chambers. Our estimates with the neutron camera are on average 34% and 41% higher than the fission chamber estimates for DTE2 and DTE3, respectively. The reasons for the discrepancies between the two systems are presently unknown and prompt further investigation. In this paper, we detail the methods used to reach the neutron yield estimate from the neutron camera, along with their strengths, weaknesses, and potential points of failure. This method is an evolution of an earlier work that estimated the deuterium–deuterium neutron yield using similar methods.

Nuclear power is a key electricity source, with light water reactors being the most common type. Their fuel typically consists of uranium dioxide pellets stacked in zirconium alloy cladding tubes. The purpose of the fuel is to produce heat and act as a barrier against releasing radioactive material. During operation, heat and radiation cause thermomechanical changes that can lead to fuel failure if not controlled. Thus, the nuclear industry needs efficient fuel performance codes with well-quantified uncertainties to predict fuel rod behavior. This thesis focuses on improving inverse uncertainty quantification and surrogate modeling for efficient fuel performance predictions.

Inverse uncertainty quantification is essential because fuel performance codes require calibrated model parameters to ensure that predictions match measurements. However, standard calibration methods often underestimate uncertainties due to unaccounted-for uncertainty sources, such as model inadequacy. Therefore, this thesis presents how unknown sources of uncertainty can be accounted for in calibration using Markov Chain Monte Carlo (MCMC) by assuming a variability in the calibration parameters. Two methods are presented, and both are based on MCMC, requiring numerous samples to converge. Hence, Gaussian Process (GP) surrogate modeling is used in place of the code to provide the calibration methods with inexpensive estimates of code responses. The first method is derivative-based, relying on differentiated GPs. While it effectively calibrates cladding oxidation, it struggles with fission gas release. Consequently, a more accurate method based on MH-within-Gibbs sampling is presented that successfully calibrates fission gas release.

The presented calibration methods use GP surrogate models to efficiently calibrate model parameters with inexpensive scalar estimates of code predictions. Beyond calibration, surrogate modeling is also essential when time-dependent predictions are needed for numerous fuel rod irradiations simultaneously. For example, if time-dependent predictions of fuel rod behavior are required in core optimization or core monitoring, calculation time can become a limiting factor. Therefore, this work also investigates neural network architectures for temporal data based on Temporal Convolutional Neural Networks (TCNs) and Fourier Neural Operators (FNOs) designed to model numerous fuel rods with varying irradiation histories. These networks accurately predict the behavior of thousands of fuel rods within seconds, significantly improving computational efficiency.

A unifying and accurate model to predict critical heat flux (CHF) over a wide range of conditions has been elusive since wall boiling research emerged. With the release of the experimental data utilized in the development of the 2006 Groeneveld CHF lookup table (LUT), by far the most extensive public CHF database available to date (nearly 25 000 data points), the development of data-driven prediction models over a large parameter space in simple geometry (vertical, uniformly heated round tubes) can be achieved. Furthermore, the popularization of machine learning techniques to solve regression problems has led to more advanced tools for analyzing large and complex databases.

This work compares three machine learning algorithms to predict the CHF database. For each selected regression algorithm (nu-Support vector, Gaussian process, and neural network), a set of optimized hyperparameters is applied. Among the investigated algorithms, the neural network emerges as the most effective, achieving a CHF predicted/measured factor standard deviation of 12.3%, three times less than that of the LUT. In comparison, the Gaussian process regression and the nu-support vector regression achieve a standard deviation of 17.7%, about two times lower than the LUT. Hence, all considered algorithms significantly outperform the LUT prediction performance. The neural network model and training methodology are designed to prevent overfitting, which is confirmed by data analyses of the CHF predictions. Finally, future development directions (including data coverage, transfer learning, and uncertainty quantification) are discussed.

Small Modular Reactors (SMRs) are gaining popularity due to several economic and safety benefits, as well as their desirable load-following capability, which allows them to complement intermittent renewable energy sources. However, the impact of load-following operation on fuel performance remains under-explored. This study employs the fuel performance code, TRANSURANUS, to assess the effects of such operations on Pressurised Water Reactor (PWR) fuel performance, given the design similarities between many SMRs and large Light Water Reactors (LWRs). Hypothetical load-following operations were simulated by varying linear heat rate levels, neutron flux, and coolant flow rate in accordance with a parametrised load-following operation. Most fuel performance parameters remained below safety limits. However, in some cases, the PCI-SCC model (SPAKOR) predicted cladding cracks and failures that were not observed during regular reference operation. A sensitivity analysis of load-following parameters indicated minimal deviations in fuel performance, apart from the PCI-SCC related issues.