Thermohydraulic modelisation of a steam generator and chemical species propagation
Steam generators are essential components of nuclear reactors whose main function is heat exchange. The chemical species present in steam generators are the cause of many parasitic phenomena (clogging, fouling, sludge deposition, etc.). Numerical simulation of species transport, taking into account the migration of chemical species and exchanges between species, both intra- and inter-phase, will allow a better understanding and better management of these problems. Numerical resolution of species transport systems presents real difficulties, in particular the management of the appearance and total disappearance of certain species, high void rates, as well as rapidly excessive calculation times.
While relying on the new code for nuclear components developed at STMF, the thesis will address the following three main scientific issues:
• Upstream, the analysis of numerical methods allowing in particular the management of evanescence, as mentioned above, and thermo-hydraulic modeling at high void rates. For this, we will rely on the PolyMAC and PolyVEF numerical schemes, already implemented in the component code.
• The physical modeling of a steam generator in the new component code, via the addition (in C++) of correlations specific to steam generators, the completion of the state laws already available, etc..
• The determination of the major chemical species to be transported, in order to be able to take into account both thermo-hydraulics and chemistry. The algorithmic coupling between thermo-hydraulics and chemistry, taking into account feedback, being the long-term objective.
While benefiting from the existing parallelization of the component code, the thermo-hydraulic and chemical modeling will be done taking into account the constraints on computation times.
Improving the predictivity of large eddy simulations using machine learning guided by high-fidelity simulations
This thesis aims to explore the application of machine learning techniques to improve turbulence modeling and numerical simulations in fluid mechanics. More specifically, we are interested in the application of artificial neural networks (ANNs) for large eddy simulation. The latter is a modeling approach that focuses on the direct resolution of large turbulent structures, while modeling small scales by a subgrid-scale model. It requires a certain ratio of total kinetic energy to be resolved. However, this ratio may be difficult to achieve for industrial simulations due to the high computational cost, leading to under-resolved simulations. We aim to improve the latter by focusing work along two main axes: 1) Using ANNs to build generic sub-mesh models that outperform analytical models and compensate for coarse spatial discretization; 2) Training ANNs to learn wall models. One of the main challenges is the ability of the new models to generalize correctly in configurations different from those used during training. Thus, taking into account the different sources and quantification of uncertainties plays a vital role in improving the reliability and robustness of machine-learned models.
Topologic optimization of µLED's optical performance
The performance of micro-LEDs (µLEDs) is crucial for micro-displays, a field of expertise at the LITE laboratory within CEA-LETI. However, simulating these components is complex and computationally expensive due to the incoherent nature of light sources and the involved geometries. This limits the ability to effectively explore multi-parameter design spaces.
This thesis proposes to develop an innovative finite element method to accelerate simulations and enable the use of topological optimization. The goal is to produce non-intuitive designs that maximize performance while respecting industrial constraints.
The work is divided into two phases:
Develop a fast and reliable simulation method by incorporating appropriate physical approximations for incoherent sources and significantly reducing computation times.
Design a robust topological optimization framework that includes fabrication constraints to generate immediately realizable designs.
The expected results include optimized designs for micro-displays with enhanced performance and a methodology that can be applied to other photonic devices.
Characterization of motor recovery in stroke patients during a BCI-guided rehabilitation
Brain-computer interfaces (BCIs) make it possible to restore lost functions by allowing individuals to control external devices through the modulation of their brain activity. The CEA has developed a BCI technology based on the WIMAGINE implant, which records brain activity using electrocorticography (ECoG), along with algorithms for decoding motor intentions. This technology was initially tested for controlling robotic effectors such as exoskeletons and spinal cord stimulation devices to compensate for severe motor impairments. While this initial paradigm of substitution and compensation is promising, a different application potential is now emerging: functional recovery through BCI-guided rehabilitation. Current literature suggests that BCIs, when used intensively and in a targeted manner, can promote neural plasticity and, in turn, improve residual motor abilities. In particular, ECoG-based implanted BCIs could offer significant therapeutic outcomes. The objective of this thesis is therefore to assess the potential of CEA's BCI technology to enhance patients' residual motor functions through neural plasticity.
This work will be approached through a rigorous and multidisciplinary scientific methodology, including a comprehensive review of the scientific literature, the setup and execution of experimentations with patients, the algorithmic development of tools for monitoring and analyzing patient progress, and the publication of significant results in high-level scientific journals.
This PhD is intended for a student specializing in biomedical engineering, with expertise in signal processing and the analysis of complex physiological data, as well as experience in Python or Matlab. A strong interest in clinical experimentation and neuroscience will also be required. The student will work within a multidisciplinary team at CLINATEC, contributing to cutting-edge research in the field of BCIs.
Advancing Semantic Representation, Alignment, and Reasoning in Multi-Agent 6G Communication Systems
Semantic communications is an emerging and transformative research area, where the focus shifts from transmitting raw data to conveying meaningful information. While initial models and design solutions have laid foundational principles, they often rest on strong assumptions regarding the extraction, representation, and interpretation of semantic content. The advent of 6G networks introduces new challenges, particularly with the growing need for multi-agent systems where multiple AI-driven agents interact seamlessly.
In this context, the challenge of semantic alignment becomes critical. Existing literature on multi-agent semantic communications frequently assumes that all agents share a common understanding and interpretation framework, a condition rarely met in practical scenarios. Misaligned representations can lead to communication inefficiencies, loss of critical information, and misinterpretations.
This PhD research aims to advance the state-of-the-art by investigating the principles of semantic representation, alignment, and reasoning in multi-AI agent environments within 6G communication networks. The study will explore how agents can dynamically align their semantic models, ensuring consistent interpretation of messages while accounting for differences in context, objectives, and prior knowledge. By leveraging techniques from artificial intelligence, such as machine learning, ontology alignment, and multi-agent reasoning, the goal is to propose novel frameworks that enhance communication efficiency and effectiveness in multi-agent settings. This work will contribute to more adaptive, intelligent, and context-aware communication systems that are key to the evolution of 6G networks.
Enhancing Communication Security Through Faster-than-Nyquist Transceiver Design
In light of the growing demand for transmission capacity in communication networks, it is essential to explore innovative techniques that enhance spectral efficiency while maintaining the reliability and security of transmission links. This project proposes a comprehensive theoretical modeling of Faster-Than-Nyquist (FTN) systems, accompanied by simulations and numerical analyses to evaluate their performance in various communication scenarios. The study will aim to identify the necessary trade-offs to maximize transmission rates while considering the constraints related to implementation complexity and transmission security, a crucial issue in an increasingly vulnerable environment to cyber threats. This work will help identify opportunities for capacity enhancement while highlighting the technological challenges and adjustments necessary for the widespread adoption of these systems for critical and secure links.
Numerical simulation of the impact between immersed structures in a compressible liquid using immersed boundary type approaches.
Many industrial systems involve structures immersed in dense fluids. Examples include the submarine industry, or, more specifically, certain 4th generation nuclear reactors using coolant fluids such as sodium or salt mixtures. The effect of the interaction of the surrounding fluid on the contact forces between structures is a phenomenon of primary importance, particularly during accidental transient scenarios that can generate large displacements of structures whose residual integrity must be demonstrated for safety purposes.
In the context of this thesis, we are particularly interested in modeling the rapid impact of a structural fragment immersed in a fluid against a wall, resulting, for example, from an explosive phenomenon in a nuclear reactor vessel cooled by sodium. In this context, the sodium, modeled as a compressible fluid, is treated numerically using a volume-finite approach. The reactor's internal structures are treated using a finite-element approach. In order to deal with large structural displacements and possible fracturing, “immersed boundary” techniques are used for fluid-structure interaction.
The aim of this thesis is to define an innovative numerical method to better simulate the fluid film between two structures that come into contact in this context. Initially, we will focus on identifying the physical characteristics of the flow at the level of the fluid film (compressibility, viscosity, etc.) that have the greatest influence on the kinematics of the structures. Secondly, the main challenge of this thesis will be to improve current numerical methods in order to represent the flow characteristics of the fluid film as accurately as possible.
The proposed thesis will be carried out at CEA Saclay, in close collaboration with the EM2C laboratory at CentraleSupélec, within the environment of the Université Paris-Saclay. The PhD student will be immersed in a team with recognized expertise in transient simulations of fluid-structure interaction.
Monte Carlo methods for sensitivity to geometry parameters in reactor physics
The Monte Carlo method is considered to be the most accurate approach for simulating neutron transport in a reactor core, since it requires no or very few approximations and can easily handle complex geometric shapes (no discretisation is involved). A particular challenge for Monte Carlo simulation in reactor physics applications is to calculate the impact of a small model change: formally, this involves calculating the derivative of an observable with respect to a given parameter. In a Monte-Carlo code, the statistical uncertainty is considerably amplified when calculating a difference between similar values. Consequently, several Monte Carlo techniques have been developed to estimate perturbations directly. However, the question of calculating perturbations induced by a change in reactor geometry remains fundamentally an open problem. The aim of this thesis is to investigate the advantages and shortcomings of existing geometric perturbation methods and to propose new ways of calculating the derivatives of reactor parameters with respect to changes in its geometry. The challenge is twofold. Firstly, it will be necessary to design algorithms that can efficiently calculate the geometric perturbation itself. Secondly, the proposed approaches will have to be adapted to high-performance computing environments.
Impact of power histories on the decay heat of spent nuclear fuel
Decay heat is the energy released by the disintegration of radionuclides present in spent fuel. Precise knowledge of its average value and range of variations is important for the design and safety of spent fuel transport and storage systems. Since this information cannot be measured exhaustively, numerical simulation tools are used to estimate the nominal value of decay heat and quantify its variations due to uncertainties in nuclear data.
In this PhD, the aim is to quantify the variations in decay heat induced by reactor operating data, particularly power histories, which are the instantaneous power of fuel assemblies during their residence in the core. This task presents a particular challenge as the input data are no longer scalar quantities but time-dependent functions. Therefore, a surrogate model of the scientific computing tool will be developed to reduce computation time. The global modeling of the problem will be carried out within a Bayesian framework using model reduction approaches coupled with multifidelity methods. Bayesian inference will ultimately solve an inverse problem to quantify uncertainties induced by power histories.
The doctoral student will join the Nuclear Projects Laboratory of the IRESNE institute at CEA Cadarache. He/she will develop skills in neutron simulation, data science, and nuclear reactors. He/she will be given the opportunity to present his/her work to various audiences and publish it in peer-reviewed journals.