Explore CTU's Digital Library

Access research publications, theses and academic works

Faculty of Civil Engineering Faculty of Mechanical Engineering Faculty of Electrical Engineering Faculty of Nuclear Sciences and Physical Engineering Faculty of Architecture Faculty of Transportation Sciences Faculty of Biomedical Engineering Faculty of Information Technology Klokner Institute Masaryk Institute of Advanced Studies Central Library Acta Polytechnica Acta Polytechnica CTU Proceedings CTU Publishing Activity

Recent Submissions

Long-term monitoring of concrete structures (temperatures and volumetric changes)
Dlouhodobé sledování betonových konstrukcí (teploty a objemové změny)
(Czech Technical University in Prague) Němčic, Vít; Vítek, Jan; Kolísko, Jiří; Coufal, Robert; Klusáček, Ladislav
Increasing demands on load-bearing capacity, durability and other performance properties of structures, together with economic requirements and principles of sustainable construction, are leading to the use of new materials and new technologies. Concrete is one of the building materials that has a high potential for modifying its properties. Depending on the requirements on elements or structures, a variety of concrete properties can be achieved by choosing suitable raw materials and selecting an appropriate technology. The issue of massive concrete bridge structures and monitoring of their deformation is an important area. Optimum materials must be sought for such structures, as conflicting demands are often placed on these structures. On one hand, a reduction of temperatures from heat of hydration, requires a limited amount of cement and on the other hand, acceleration of the construction process, rapid increase of concrete strength and early demoulding are required. The aim of the PhD thesis was to summarize the research results in the area of monitoring the evolution of temperatures induced by heat of hydration and in the area of volume changes in massive concrete structures.
Validation of an edge turbulence code in the Compass tokamak
(Czech Technical University in Prague) Mácha, Petr; Seidl, Jakub; Stegmeir, Andreas; Tichý, Milan
The thesis validates and compares fluid models of turbulence in the edge plasma of the COMPASS tokamak. The three-dimensional GBS and GRILLIX codes and the reduced DIFFLOW ESEL-like model are compared with experimental measurements. The models reproduce plasma profiles and fluctuations well, with GRILLIX providing the best description of blob dynamics and DIFFLOW enabling rapid GPU calculations. The observed discrepancies highlight the importance of boundary conditions and plasma-wall interaction modeling. The thesis also describes the spontaneous formation of a transport barrier in helium plasma on the GOLEM tokamak.
Effect of Protective Coating on High Temperature Corrosion of Zr Alloys
Vliv ochranného pokrytí na vysokoteplotní korozi Zr slitin
(Czech Technical University in Prague) Celbová, Lucie; Kratochvílová, Irena; Lovecký, Martin; Novotný, Radek
This thesis investigates high-temperature steam oxidation of nuclear fuel cladding ZIRLO coated with polycrystalline diamond (PCD), magnetron-sputtered Cr (CrPVD), cold-sprayed Cr (CrCS), and double layers (ZIRLO/PCD/CrPVD, ZIRLO/PCD/CrCS) under LOCA-relevant conditions (900°C/30 min1000°C/30 min) and provides an explanation of the carbon protective effect in the ZIRLO/PCD system. The thesis structure includes an introduction to the motivation for accident tolerant fuel coatings, a literature review on Zr alloy corrosion, chromium and PCD coatings, experimental methods, results, discussion, and conclusions. At the explored conditions, PCD offers modest protection but suppresses breakaway oxidation. Absolute weight gains for ZIRLO/PCD/CrPVD and ZIRLO/PCD/CrCS exceed those of Cr-only coated counterparts. Carbon gasifies during the ramp to 1000°C, yet ZIRLO/PCD/CrPVD significantly softens the aggressive breakaway observed in ZIRLO/CrPVD. A comprehensive Raman study of ZrO2 scales formed on ZIRLO under steam oxidation at LOCArelevant transient conditions (900°C/30 min1000°C/30 min) is presented. The work spatially maps the quasielastic scattering (QES) tail in zirconia scales and reveals its two different sources, resolves the distributions of defective tetragonal-like (140 cm1) and classic tetragonal (260 cm1) modes, reports the 800 cm1 mode in steam-corroded ZrO2, and demonstrates widespread boson peak occurrence in an industry-relevant nuclear oxide system. The 140 cm1 peak is identified as a softened vibration of oxygen-deficient tetragonal-like Zr coordination. Secondary-phase particle is captured modulating the vacancy distribution and creating diffusion shadow with reduced defect signatures when in particular conditions. Analysis of previously published data reveals that in the case of PCD coated ZIRLO exposed to 400°C steam, carbon diffuses into ZrO2 grain boundaries under the vacancy flux present at the metal/oxide interface. Numerical modeling based on Fromholds theory and carbon-induced potential landscapes reproduces the weight gain experimental values under the assumption of anisotropic carbon transport. Two approaches for achieving sub-parabolic kinetics in the Fromholds framework are discussed.The PCD layer in the ZIRLO/PCD system strongly modulates hydrogen behavior. It retains excess hydrogen at lower temperatures while releasing it as hydrocarbons at high temperatures, resulting in significantly lower overall hydrogen concentration compared to bare ZIRLO at high temperatures (despite similar corrosion rates). However, ultimately, the corrosion rate reduction in the ZIRLO/PCD systems prior to transition is attributed to carbon diffused and bonded to ZrO2 grains, not the PCD layer itself. The absence of protection at high temperatures results from rapid ZrO2 growth limiting carbon transport into the oxide. An optimal balance between vacancy flux and carbon diffusion, dependent on temperature and surface sp2 carbon concentration, is expected. Carbon bonded in between layers of fine-grained ZrO2 (which are parallel to the direction of oxide growth) is expected to promote the protective behavior. Future work should identify conditions optimizing these structures that do not compromise the overall integrity of the resulting system and should be concerned with the behavior of such pre-oxidized PCD coated samples at high temperatures ( 1000°C) with the emphasis on carbon evolution in such environments.
Advanced methods for tokamak heat load studies
Pokročilé metody studia tepelné zátěže tokamaku
(Czech Technical University in Prague) Čaloud, Jakub; Tomešová, Eva; Pokol, Gergö; Gerardin, Jonathan
This thesis is focused on the studies of heat loads on the tokamak first wall by advanced experimental and simulation methods. Heat flux represents one of the key factors that could limit the performance of future fusion reactors, therefore, the plasma-facing components must be optimized with respect to their thermal resilience and operational reliability. The work presents the design of selected components of the COMPASS-U tokamak, based on predictive simulations and design of a thermal diagnostic system utilizing the Fiber Bragg Grating temperature sensors. Furthermore, the work investigates transient heat loads caused by the impact of runaway electrons through the analysis of experimental data and a novel multiphysics workflow of simulation tools.
Moment Invariants and Invariant Neural Networks
Momentové invarianty a invariantní neuronové sítě
(Czech Technical University in Prague) Košík, Václav; Flusser, Jan; Kohout, Josef; Šikudová, Elena
This doctoral thesis addresses image representation and recognition methods that are invariant to image transformations, with emphasis on image moments and invariant neural architectures. First, it extends the theory of blur invariants from graylevel images to color and multispectral data by introducing new cross-channel invariants that exploit dependencies between channels. These invariants provide complementary information that is not present in standard single-channel invariants and significantly improve recognition performance. The theory is further extended to a more general degradation model combining spatial blur with linear channel mixing, which leads to new invariants for multispectral images that are jointly invariant to both effects. The proposed methods are derived theoretically in the spectral domain, transformed into stable moment-domain formulations, and validated experimentally. Their Python implementation in the moment domain is shared on GitHub together with all data and code needed to reproduce the experiments. The thesis also investigates invariance in deep learning without relying on data augmentation. It proposes a hybrid architecture combining a conventional convolutional neural network with an invariant branch, where their predictions are fused according to the estimated deformation intensity of the input. A rotationally invariant architecture based on the Radon transform is introduced and shown to provide strong and robust recognition performance. The thesis further examines limitations of harmonic networks in the discrete domain and implements modifications improving their practical invariance. Finally, new 3D rotation invariants based on Appell moments are proposed. Their construction is enabled by the quasi-monomial property of 3D Appell polynomials, which is proven in the thesis. The experiments demonstrate a strong recognition performance of these invariants, as well as their possible use in neural networks.