Nuclear fuel failures during the reactor operation happen quite often in the world. The theoretical part of this thesis is dedicated to the most common causes of nuclear fuel failures. It describes failure mechanism and corrective actions. The unfavorable trends in nuclear fuel behavior are prevented by suitable method of nuclear fuel monitoring. Some of them may affect the safety of the power plant. For example, the fuel assembly bow affects the function of rod cluster control assembly. Another part, which describes inspection methods, is devoted to inspection and repair of nuclear fuel. The thesis concentrates on the Temelin NPP, where there was implemented post-irradiation inspection program for checking compatibility between Westinghouse's fuel assemblies and water chemistry of reactor VVER. This program continues even after the change of nuclear fuel supplier. Practical part of this thesis is dedicated to proposal of a new method of fuel assembly bow measurement for Temelin NPP based on ultrasound. This proposal is supported by measurement on the experimental device for detection of spacer grid position developed by Research Centre Rez.

In collaboration with Westinghouse Electric AB this project aims to study the possibilities with using machine learning methods to predict fuel failure in a Boiling Water Reactors (BWRs). The main objective has been to create a dataset consisting of both empirical measurements and simulated samples from a physics model and evaluate different machine learning algorithms, that use these datasets to predict fuel defects. The simulated data is created using a physics model derived from the ANS-5.4 standard which allows for good control over specific parameter values. Three machine learning algorithms were deemed fit for this type of problem and used throughout the project: Random Forest (RF), K-Nearest Neighbor (KNN) and Neural Network (NN). Both classification and regression type problems have been assessed. All three methods showed good results for the classification problems, where the goal was to predict if there was a fuel failure or not. All models reached an accuracy above 97% and performed well, the RF model had the highest overall, with an accuracy of 98.2 %. However, the NN method made the fewest false negative predictions and can therefore be seen as the best model for this purpose. For the regression, problems with the aim of predicting escape rates, both the RF and KNN had similar promising results with very small errors overall. Yet, there is a slight increase in errors when predicting higher escape rates for both models. This is most likely due to the available data being of mostly low escape rates. The NN did not perform well with this problem, the predictions having large error for both low and high escape rates, a possible explanation is the lack of data. To improve the results, and create even better models, the empirical measurements need to contain more information such as defect location and fuel failure size, also an increase in the number of samples taken at fuel failure operation would be valuable.

<p>Presently there is a clear trend of increasing demands on in-pile performance of nuclear fuel. Higher target burnups, part length rods and various fuel additives are some examples of this trend. Together with an increasing demand from the public for even safer nuclear power utilisation, this implies an increased focus on various experimental, preferably non-destructive, methods to characterise the fuel.</p><p>This thesis focuses on the development and experimental evaluation of such methods. In its first part, the thesis presents a method based on gamma-ray spectroscopy with germanium detectors that have been used at various power reactors in Europe. The aim with these measurements is to provide information about the thermal power distribution within fuel assemblies in order to validate core physics production codes. The early closure of the Barsebäck 1 BWR offered a unique opportunity to perform such validations before complete depletion of burnable absorbers in Gd-rods had taken place. To facilitate the measurements, a completely submersible measuring system, LOKET, was developed allowing for convenient in-pool measurements to be performed.</p><p>In its second part, the thesis describes methods that utilise in-pile measurements. These methods have been used in the Halden test-reactor for determination of fission gas release, pellet-cladding interaction studies and fuel development studies.</p><p>Apart from the power measurements, the LOKET device has been used for fission gas release (FGR) measurements on single fuel rods. The significant reduction in fission gas release in the modern fuel designs, in comparison with older designs, has been demonstrated in a series of experiments. A FGR database covering a wide range of burnup, power histories and fuel designs has been compiled and used for fuel performance analysis. The fission gas release has been measured on fuel rods with average burnups well above 60 MWd/kgU. The comparison between core physics calculations (PHOENIX-4/POLCA-7) and the in-pool measurements of thermal power indicates that the nodal power can generally be predicted with an accuracy within 4% and the bundle power with an accuracy better than 2%, expressed as rms errors.</p><p>In-pile experiments have successfully simulated the conditions that occur in a fuel rod following a primary debris failure, being secondary fuel degradation. It was concluded that massive hydrogen pick-up takes place during the first few days following the primary failure and that a pre-oxidized layer does not function as a barrier towards hydriding in an environment with a very high partial pressure of hydrogen. Another series of in-pile experiments clearly indicate that increased UO₂ grain size is an effective way of suppressing fission gas release in LWR fuel up to the burnup level covered (55 MWd/kgUO₂).</p>

Presently there is a clear trend of increasing demands on in-pile performance of nuclear fuel. Higher target burnups, part length rods and various fuel additives are some examples of this trend. Together with an increasing demand from the public for even safer nuclear power utilisation, this implies an increased focus on various experimental, preferably non-destructive, methods to characterise the fuel. This thesis focuses on the development and experimental evaluation of such methods. In its first part, the thesis presents a method based on gamma-ray spectroscopy with germanium detectors that have been used at various power reactors in Europe. The aim with these measurements is to provide information about the thermal power distribution within fuel assemblies in order to validate core physics production codes. The early closure of the Barsebäck 1 BWR offered a unique opportunity to perform such validations before complete depletion of burnable absorbers in Gd-rods had taken place. To facilitate the measurements, a completely submersible measuring system, LOKET, was developed allowing for convenient in-pool measurements to be performed. In its second part, the thesis describes methods that utilise in-pile measurements. These methods have been used in the Halden test-reactor for determination of fission gas release, pellet-cladding interaction studies and fuel development studies. Apart from the power measurements, the LOKET device has been used for fission gas release (FGR) measurements on single fuel rods. The significant reduction in fission gas release in the modern fuel designs, in comparison with older designs, has been demonstrated in a series of experiments. A FGR database covering a wide range of burnup, power histories and fuel designs has been compiled and used for fuel performance analysis. The fission gas release has been measured on fuel rods with average burnups well above 60 MWd/kgU. The comparison between core physics calculations (PHOENIX-4/POLCA-7) and the in-pool measurements of thermal power indicates that the nodal power can generally be predicted with an accuracy within 4% and the bundle power with an accuracy better than 2%, expressed as rms errors. In-pile experiments have successfully simulated the conditions that occur in a fuel rod following a primary debris failure, being secondary fuel degradation. It was concluded that massive hydrogen pick-up takes place during the first few days following the primary failure and that a pre-oxidized layer does not function as a barrier towards hydriding in an environment with a very high partial pressure of hydrogen. Another series of in-pile experiments clearly indicate that increased UO2 grain size is an effective way of suppressing fission gas release in LWR fuel up to the burnup level covered (55 MWd/kgUO2).

La reproduction expérimentale de transitoires thermiques accidentels de type Accident par Perte de Réfrigérant Primaire (APRP) en laboratoire a permis d’observer la fragmentation du combustible fortement irradié lorsque la gaine se déforme sous l’augmentation de la température. Ces fragments de petites tailles peuvent se relocaliser dans le ballon voire être éjectés hors du crayon cas de rupture de gaine. La zone High Burnup Structure (HBS) des combustibles fortement irradiés est la plus susceptible de se fragmenter et d’être relocalisée par sa position en périphérie de pastille. Pour expliquer ce phénomène, l’hypothèse retenue est que le transitoire provoque une surpression dans les bulles HBS ce qui mène à la décohésion des joints de grains et à la fragmentation. Cette thèse a pour but de développer un critère de fissuration mécanique du combustible pour mieux comprendre le comportement des bulles HBS lors des conditions thermiques APRP. Ce travail se base sur une méthode une méthode micromécanique en trois étapes : i) la représentation qui permet de caractériser la microstructure de la zone HBS (leurs dimensions, leur fraction volumique, et la pression interne). Deux sources d’informations seront utilisées : les observations expérimentales provenant de disques ou de pastilles de combustible irradiés à fort taux de combustion et d’outils numériques(avec Alcyone-Caracas [JSB+14]) / Under Loss Of Coolant Accident(LOCA) transients conditions, the high irradiated fuel is fragmented in small sizes fragments who can be relocated in the balloon, or being ejected out of the fuel rod if the latter burst. This work focuses on the pellet rim, where bubbles density increases owing to a higher irradiation level. Usually the hypothesis used to explain fuel fragmentation during transient is grain cleavage induced by over pressurized fission gas bubbles, located at the grain boundary. The aim of this study is to define a macroscopic fragmentation model based on a micro mechanical approach to have a better understanding of the fuel mechanical behaviour at lower scale : size and volume fraction of fragments. This PhD introduces a stepwise micromechanical method based on three steps : i) firstly, we detail how to model the HBS microstructure including pressurized porosities, based on experimental or numerical data and define a representative volume element (RVE)