This work investigates the relationship between porosity morphology and the effective thermal conductivity of ceramic nuclear fuel. To this end, a numerical framework based on Fast Fourier Transform (FFT) homogenization is employed to simulate three-dimensional porous media featuring both closed and interconnected pore networks. The structure generation and voxelization are carried out using Mérope, a newly developed software at CEA designed for realistic microstructure modeling. As an initial step, the influence of voxel resolution and Representative Volume Element (RVE) size is assessed to ensure statistical representativity of the simulations while optimizing computational cost. Subsequently, numerical predictions of effective thermal conductivity are compared against analytical models such as Maxwell and Loeb, highlighting their respective applicability and limitations under conditions relevant to nuclear fuel. A parametric analysis is then performed to quantify the influence of key geometric features, including pore size, anisotropy and grain size distribution. Particular attention is given to the confinement thickness of intergranular porosity, characterized by a newly introduced parameter, delta. Based on this analysis, a corrective factor incorporating morphological effects, especially the impact of delta, is proposed and integrated into the Loeb model, improving its predictive accuracy and alignment with numerical results, particularly for interconnected porosity structures. Finally a Bayesian optimization algorithm is employed to generate synthetic 3D microstructures whose 2D slices statistically match real experimental images. The morphological agreement is evaluated using statistical tests such as the Kolmogorov-Smirnov and chi-squared tests. Subsequently, the thermal conductivity of these structures is calculated and compared with both the classical Loeb model and its modified version. The proposed approach provides a robust workflow for inverse design and realistic thermal modeling of porous nuclear ceramics.
Questo lavoro indaga la relazione tra la morfologia della porosità e la conducibilità termica efficace dei combustibili nucleari ceramici. A tal fine, viene impiegato un ambiente di simulazione numerica basato sull'omogenizzazione tramite la Trasformata Veloce di Fourier (FFT) per simulare mezzi porosi tridimensionali contenenti sia reti di pori chiusi che interconnessi. La generazione e la voxelizzazione delle microstrutture sono effettuate con Mérope, un software recentemente sviluppato presso il CEA per la modellazione realistica di microstrutture. Come primo passo, si valuta l'influenza della risoluzione voxel e della dimensione dell'Elemento di Volume Rappresentativo (RVE) per garantire la rappresentatività statistica delle simulazioni e ottimizzare i tempi di calcolo. Successivamente, le predizioni numeriche della conducibilità termica efficace sono confrontate con modelli analitici, tra cui quelli di Maxwell e Loeb, evidenziandone l'applicabilità e i limiti. Viene quindi condotta un'analisi parametrica per quantificare l'influenza di caratteristiche geometriche chiave, come dimensione dei pori, anisotropia e distribuzione dimensionale dei grani. Particolare attenzione è rivolta allo spessore di confinamento della porosità intergranulare, descritto da un nuovo parametro introdotto, delta. Sulla base di tale analisi, si propone un fattore correttivo che tiene conto degli effetti morfologici, con particolare enfasi su delta, da integrare nella formula di Loeb; questo fattore ne migliora l'accuratezza predittiva e l'accordo con i risultati numerici, specialmente per le strutture che presentano porosità interconnessa. Infine, viene utilizzato un algoritmo di ottimizzazione bayesiana per generare microstrutture sintetiche 3D le cui sezioni 2D corrispondono statisticamente a immagini sperimentali reali. L'accordo morfologico viene valutato mediante test statistici come Kolmogorov-Smirnov e chi-quadro. Successivamente, la conducibilità termica di tali strutture viene calcolata e confrontata con il modello di Loeb classico e la sua versione modificata. L'approccio proposto fornisce un flusso di lavoro robusto per la progettazione inversa e la modellazione termica realistica di ceramici porosi per applicazioni nucleari.
Numerical analysis of the effects of porosity morphology on the thermal conductivity of ceramic nuclear fuels
MATTIUZ, ALESSIO
2024/2025
Abstract
This work investigates the relationship between porosity morphology and the effective thermal conductivity of ceramic nuclear fuel. To this end, a numerical framework based on Fast Fourier Transform (FFT) homogenization is employed to simulate three-dimensional porous media featuring both closed and interconnected pore networks. The structure generation and voxelization are carried out using Mérope, a newly developed software at CEA designed for realistic microstructure modeling. As an initial step, the influence of voxel resolution and Representative Volume Element (RVE) size is assessed to ensure statistical representativity of the simulations while optimizing computational cost. Subsequently, numerical predictions of effective thermal conductivity are compared against analytical models such as Maxwell and Loeb, highlighting their respective applicability and limitations under conditions relevant to nuclear fuel. A parametric analysis is then performed to quantify the influence of key geometric features, including pore size, anisotropy and grain size distribution. Particular attention is given to the confinement thickness of intergranular porosity, characterized by a newly introduced parameter, delta. Based on this analysis, a corrective factor incorporating morphological effects, especially the impact of delta, is proposed and integrated into the Loeb model, improving its predictive accuracy and alignment with numerical results, particularly for interconnected porosity structures. Finally a Bayesian optimization algorithm is employed to generate synthetic 3D microstructures whose 2D slices statistically match real experimental images. The morphological agreement is evaluated using statistical tests such as the Kolmogorov-Smirnov and chi-squared tests. Subsequently, the thermal conductivity of these structures is calculated and compared with both the classical Loeb model and its modified version. The proposed approach provides a robust workflow for inverse design and realistic thermal modeling of porous nuclear ceramics.File | Dimensione | Formato | |
---|---|---|---|
2025_07_Mattiuz_Thesis_01.pdf
non accessibile
Dimensione
2.96 MB
Formato
Adobe PDF
|
2.96 MB | Adobe PDF | Visualizza/Apri |
2025_07_Mattiuz_Executive_Summary_02.pdf
non accessibile
Dimensione
1.82 MB
Formato
Adobe PDF
|
1.82 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/240175