Tesi di laurea presso il Consorzio RFX

Il Consorzio RFX propone a studenti di corsi di laurea triennale e magistrale in Fisica e Ingegneria, tesi di laurea su diversi argomenti di studio, nell’ambito della ricerca sulla fusione.
Le tesi si svolgono presso i nostri laboratori sotto la supervisione di ricercatori del Consorzio e docenti dell’Università di Padova.
La tesi può essere sperimentale, teorica, numerica o compilativa.

Fisica

Tesi Triennali
proposte dal Consorzio RFX

Impurity Transport in RFX-mod Plasmas

Proponente: M.Gobbin /I.Predebon

Relatore Accademico:  da stabilire

Capogruppo: D. Terranova

Tipologia: Modelling, Teorica, Numerica, Compilativa

Abstract:

In fusion plasmas, impurities refer to elements other than the primary working gas. These impurities can arise from various sources, such as plasma-facing components, external injections, or residual gases. Their presence profoundly influences the plasma’s behavior, affecting energy confinement, radiation losses, and overall stability. Consequently, understanding and controlling impurity transport is essential for enhancing plasma performance.

Experiments conducted on the reversed field pinch device RFX-mod have revealed that impurities exhibit an outward flux, preventing them from penetrating the plasma’s inner core. However, these observations cannot be fully explained by first-principles theoretical models.

To address this, simulations using the Hamiltonian guiding centre code ORBIT could provide a valuable complementary perspective. The modeling work first requires a brief analysis of experimental data, focusing on magnetic modes, electron temperature, and density profiles.

The ORBIT simulations aim to reconstruct the magnetic topology of the RFX-mod plasmas under investigation and examine the diffusion of various impurity species (differing in atomic mass and charge). These simulations will account for collisions with the main gas particles and the effects of the radial electric field, enhancing our understanding of impurity dynamics in these plasmas.

Competenze necessarie per svolgere con successo la tesi:

Data della proposta: 20/11/2024

Evoluzione dei profili di temperatura e densità elettronica durante gli ELMs a TCV

Proponente/Relatore RFX: M. La Matina, M. Agostini
Relatore Accademico: M. Agostini
Capogruppo: L. Carraro
Responsabile di Programma: T. Bolzonella

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Abstract:

Gli ELMs sono instabilità che si verificano nei plasmi confinati in modo H, regime fondamentale per i
futuri reattori a fusione. La tesi si propone di studiare L’evoluzione temporale dei profili elettronici
prima, durante e dopo un ELM.
Il lavoro prevede l’analisi sperimentale dei dati raccolti dal tokamak TCV (Tokamak à Configuration Variable) per plasmi in modo H che presentano ELMs: come principale diagnostica verrà utilizzata la diagnostica spettroscopica Thermal Helium Beam (THB).

Competenze richieste (se necessarie): Programmazione di base (python)
Data della proposta: 22/11/2024

Studio dell’ottica del fascio di SPIDER con telecamere 2D

Proponente/Relatore RFX: M. Ugoletti, G. Emma

Relatore Accademico: M. Agostini
Capogruppo: L. Carraro

Responsabile di Programma: TBD

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Abstract:

Il fascio di ioni negativi di SPIDER quando interagisce con il gas di fondo in cui si propaga, emette luce che viene raccolta da un set di 15 telecamere visibili, che compongono il sistema tomografico di SPIDER.

Dal segnale raccolto da queste telecamere, che osservano il fascio da diverse angolazioni, è possibile studiare le caratteristiche del fascio stesso, come la sua divergenza e intensità.

Per la prima volta SPIDER ha operato con 320 fascetti (1/4 del fascio totale) e diversi parametri operazionali sono stati esplorati. L’ottica e l’omogeneità di questi fascetti sono due caratteristiche fondamentali per il funzionamento di MITICA, l’altro esperimento presente a RFX e prototipo 1 a 1 di un iniettore di fasci neutri di ITER.

Numerosi dati sono stati raccolti nella seconda metà del 2024 e si propone la caratterizzazione del fascio (divergenza, intensità, omogeneità) al variare dei parametri della macchina e con il confronto con le altre diagnostiche di fascio. Questo progetto permetterà di approfondire i processi fisici alla base di un fascio di ioni negativi e di acquisire competenze nell’analisi di dati sperimentali e della loro interpretazione.

Competenze richieste (se necessarie): Programmazione di base
Data della proposta:
25/11/2024

Studio dello spazio operazionale della diagnostica Thermal Helium Beam sul tokamak TCV

Proponente/Relatore RFX: M. Ugoletti

Relatore Accademico: M. Agostini

Capogruppo: L. Carraro

Responsabile di Programma: TBD

Tipologia: Sperimentale, Teorica, Analisi dati, Compilativa

Abstract:

Lo studio del bordo del plasma è cruciale negli esperimenti di fusione nucleare, poiché influenza sia la stabilità del plasma principale, sia le interazioni con le pareti del reattore, determinando l’efficienza globale del sistema. Tra le varie diagnostiche di bordo, il Thermal Helium Beam (THB) è una diagnostica molto importante perché permette di misurare simultaneamente sia la densità e la temperatura elettronica che le loro relative fluttuazioni e instabilità. Il THB che è stato installato sul tokamak TCV, per la prima volta, misura anche il riassorbimento dei fotoni dell’He, informazione fondamentale per la corretta stima dei parametri del plasma.

Si propone un lavoro di tesi di analisi dei dati sperimentali raccolti a TCV, con l’obiettivo di definire uno spazio operazionale che correli i parametri del plasma con l’assorbimento della radiazione e la quantità di elio iniettata per effettuare la misura. Attraverso lo studio e l’elaborazione dei dati, si cercherà di identificare le condizioni ottimali di funzionamento della diagnostica, migliorandone sensibilità e affidabilità. Questo progetto permetterà di approfondire i processi fisici che avvengono nel plasma edge e di acquisire competenze nell’analisi di dati sperimentali e nello sviluppo di modelli per la diagnostica di plasmi ad alta temperatura.

Competenze richieste (se necessarie): Programmazione di base

Data della proposta: 25/11/2024

Superconductivity for Sustainable Magnetic Fusion Development

Proponente/Relatore RFX: L. Piron

Relatore accademico: L. Piron, L. Salasnich
Capogruppo RFX: D. Terranova

Tipologia: Sperimentale, Teorica, Analisi dati, Compilativa

Abstract:

In tokamaks, the plasma, a fully ionized gas, is confined by magnetic fields. For the economics of fusion to be viable, reducing the energy consumption of the magnets is absolutely necessary. The energy consumption of a magnet is directly related to electrical resistance. Without resistance, electrical consumption would drop to zero, and, as an added benefit, no heat would be generated in the magnets. In this context, the use of superconducting magnets is a game changer: superconductivity, which allows current to flow without resistance at low temperatures, enables electromagnets to meet the demanding requirements of modern tokamaks.
During the thesis work the student will analyze the basic properties of superconducting materials and superconducting magnets in fusion energy applications, providing insights into the current state of the technology and identifying ongoing and future trends in research and development. The activity will be carried out under the joint supervision of Prof. Lidia Piron, who works in the field of plasma physics, and Prof. Luca Salasnich, who works in the field of superconductors and Bose-Einstein condensates.

Data della proposta: 13/01/2025

Tesi Magistrali
proposte dal Consorzio RFX

Laser-aided diagnostics on a negative ion source for fusion

Tutor RFX: R. Agnello, M. Barbisan

Academic supervisor: G. Serianni

Group leader: L. Carraro

Responsabile di Programma: Vanni Toigo

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Topic of thesis: negative ion sources, laser-plasma diagnostics

Abstract:

Laser-based diagnostics for plasmas are essential tools to characterize negative ion sources for fusion. In SPIDER, the full-scale source prototype for ITER, two techniques are routinely employed: the Cavity Ring-Down Spectroscopy (CRDS) and the Laser Absorption Spectroscopy (LAS).  CRDS is based on a high finesse optical trap in which a laser beam, travelling back and forth, photo-ionizes the negative ions (H-, D-) providing a measure of their line-integrated volumetric density. LAS measures the volumetric density of caesium atoms evaporated inside the source, one of the main ingredient required to optimize negative ion production. Both techniques require accurate calibration and constant monitoring due to their sensitivity and, especially, during the experimental campaigns, they are crucial to promptly provide the source operators with useful feedbacks to properly run the facility.

The careful monitoring of these techniques is therefore essential to guarantee a constant and reliable inflow of data to drive the experimental campaigns.

The thesis project is structured as follows:

  • Preliminary study/training
  • Literature research. Study of laser-plasma interaction. Laser safety course. Study of the results obtained in the previous experimental campaigns.
  • Detailed numerical studies. Model for Cavity Ring-Down Spectroscopy and Cs absorption spectroscopy.
  • Experimental activities
  • Calibration and alignment of optical diagnostics. Characterization of the signals in vacuum. Participation in the experimental campaign. Support in data analysis.
  • Comparison with other diagnostics. Possible application of the collisional radiative model for caesium.

Required skills: familiarity with laboratory equipment and any programming language.

Proposal date:  06/02/2025

Development of a mm-wave interferometric system for a negative ion source for fusion

Tutor RFX: R. Agnello

Academic supervisor: G. Serianni

Groupleader: R. Pasqualotto

Responsabile di Programma: D. Marcuzzi

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Topic of thesis: mm-wave interferometry, negative ion sources

Abstract:

Heating Neutral Beam (HNB) injectors are crucial for fusion energy research, consisting of a plasma source, an electrostatic accelerator, and a gas neutralization system. At the Neutral Beam Test Facility (NBTF) in Padova, the SPIDER and MITICA experiments are advancing HNB development for ITER. In SPIDER, plasma discharges are generated by eight RF drivers, but their behavior is influenced by electromagnetic interactions, magnetic fields, and cesium dynamics, affecting negative ion production.

To investigate these effects, the MINION experiment, a compact radiofrequency-driven negative ion source, has been developed. Its goals are to enhance plasma confinement in SPIDER and establish a single-driver testbed for plasma studies. A key diagnostic for MINION is a millimeter-wave interferometer, offering non-intrusive, high-precision plasma density measurements with excellent time resolution. However, implementing this technique requires numerical modeling, wave propagation studies, and experimental validation.

This thesis will focus on:

  1. Theoretical studies on mm-wave propagation, interferometry, and plasma diagnostics.
  2. Conceptual design of the interferometer, including waveguide layout, antenna placement, and numerical modeling using COMSOL and CST.
  3. Experimental validation, involving microwave component characterization, interferometer assembly, and plasma measurements under varying conditions.

The project aims to improve wave-plasma interaction understanding and support SPIDER’s plasma confinement optimization.

Required skills: familiarity with laboratory equipment and any programming language.

Proposal date:  06/02/2025

Towards global boundary turbulence simulations in reversed field pinch configurations

RFX Supervisor: M. Giacomin

Academic supervisor: M. Giacomin

Head of the RFX research group: M. Zuin

Leader of the RFX research program: L. Marrelli

Tipologia: Modelling, Teorica, Numerica, Compilativa

Abstract:

RFX-mod is a toroidal device with minor radius a = 0.46 m and major radius R0 = 2 m located at Consorzio RFX, Padua, which confines plasmas by means of magnetic fields.  RFX-mod is optimized to operate in the so-called reversed field pinch (RFP) magnetic configuration, where the toroidal component of the toroidal magnetic field reverses its direction at the plasma boundary. After a significant machine upgrade, RFX-mod2 is now equipped with new diagnostics that allow for a deeper characterization of plasma turbulence. Given the complexity of the magnetic field in the boundary region, often characterized by a significant degree of chaoticity, a deep understanding of turbulence dynamics can only be attained by complimenting the experimental analysis with the results of global turbulence simulations. However, the reversal of the toroidal magnetic field poses several numerical challenges, which will be addressed in this thesis project.

The main goal of this project is to approach the first-of-its-kind global turbulence simulation in the boundary region of a RFP plasma by using the GBS code (M. Giacomin et al, Journal of Computational Physics 463 (2022) 111294). GBS is a three-dimensional turbulence code that evolves two-fluid equations in the boundary region of magnetic fusion devices in arbitrary magnetic configuration, which has been recently extended with a three-dimensional Poisson and Ampère solver to unable the simulation of the reversal region. Building on previous tests, the student will perform the first turbulence simulation of a RFP plasma of a reduced RFX-mod2. The main numerical challenges will be identified and potential solutions will be tested. Eventually, the results of this first simulation will be qualitative compared to already existing experimental data from a RFX-mod discharge.

Required skills: Basic knowledge of computational physics may help, but it is not mandatory

Proposal date: 06/02/2025

Ricostruzioni dei profili radiali di temperatura elettronica in RFX-mod (nei regimi a confinamento migliorato) dalle misure di Thomson Scattering e SXR

Proponente/Relatore RFX: P. Franz

Relatore accademico: L. Giudicotti

Capogruppo RFX: L. Carraro

Responsabile di Programma RFX: L. Marrelli

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Abstract:

The temperature profiles of electrons in quasi-single helical (QSH) states in RFX-Mod present clear transport barriers. The study of these profiles is mainly conducted using Thomson scattering, which has a time resolution limited by laser pulse repetition frequency (100 Hz). A two-foil SXR spectrometer allowed this study to be extended. Temperature profiles can be measured up to a frequency of a few kHz, and the entire temporal evolution of the electron temperature during a QSH cycle can be determined. A mapping technique, originally developed for Thomson temperature profiles, has also been applied to profiles measured with SXR diagnostics. In particular, the temperature gradients linked to the presence of transport barriers are reconstructed and analyzed in a helical reference system, consistently with the equilibrium of the underlying plasma. A clear difference in the behavior of the temperature gradient is observed between the ascending phase of the QSH and the saturated (or flattop) phase.

The proposed thesis subject is the statistical application of the method to the available QSH database. More specifically, analysis of the greatest number of Te profiles during the QSH phases, choosing the cases in which both types of profiles are available (from the TS and the SXR), will be performed. The target is the establishment and characterization of three types of electron temperature profiles: MH (without transport barriers), DAx (transport barrier with two magnetic axes) and SHAx (transport barrier with a single magnetic axis). Moreover, the study will allow to “calibrate” the Te profile from the SXR to the TS profile, which is much more resolved in space. The work will then extend the Te from SXR data analysis to follow the temporal evolution of Te profiles and gradients with a resolution of a few kHz.

The adaptation/ development/optimization of the already available software tools, will be an important element of the proposed work.

Competenze necessarie per svolgere con successo la tesi:

Good aptitude to data analysis work. Knowledge of computer language programming (Python, IDL) required.

Date: 06/02/2025

Developement of the Light Impurity Tomography diagnostic for RFX-mod2 experment

Proponente: M. Agostini, M. Ugoletti

Relatore Accademico: M. Agostini

Capogruppo: L. Carraro

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

The Light Impurity Tomography (LIT) is a spectroscopic diagnostic for measuring the poloidal distribution of the emissivity of low Z impurities in RFX-mod2 experiment. It will observe the emission along the poloidal angle of neutral hydrogen or of other low Z impurities that are present in the edge region.

The diagnostic consists of 7 cameras, installed in 7 different portholes around the poloidal direction, with changeable interference filters. The 7 cameras observe line-integrated signals, and the 2D map of the emissivity of the specific ion species selected by the interferential filter is obtained with a tomographic inversion.

The goal of the thesis is to optimize the position of the cameras for obtaining the best reconstruction of the emissivity. Simulation of the emissivity has to be performed, together with the synthetic data of the cameras. A tomographic algorithm will be developed to obtain the poloidal emissivity to be compared to the simulated one. By minimizing the errors between the simulation and the reconstruction, the best spatial position and orientation of the 7 cameras are determined.

Competenze richieste (se necessarie): Programmazione di base
Data della proposta:
12/01/2024

Deep Learning Models for real-time Fusion Device Data Compression Algorithms

Proponente/Relatore RFX: A. Rigoni Garola

Relatore Accademico: J. Pazzini (DFA)
Capogruppo: C. Taliercio

Abstract:

The recently enhanced RFX-mod2 experiment [2], located at Consorzio RFX in Padova, presents a set of distinctive prospects for the advancement and validation of cutting-edge ML and DL algorithms and techniques, for plasma control. RFX-mod2 operates as a multi-configuration device, generating plasmas across various magnetic configurations: tokamak, ultra-low-q, and reversed field pinch (RFP). This innovative experimental facility provides an exceptional venue for pursuing the outlined research project, featuring unparalleled diagnostic sensor arrays, machine flexibility, robust control infrastructure, and a large historical database of ten years of experiments with RFX-mod.

Among the peculiar features of RFX-mod2, it will provide a very high spatial resolution magnetic diagnostic with more than 1700 sensors [3], along with more than 200 actuator coils independently controlled [4]. These numbers are roughly ten times higher than those found on other existing fusion plasma devices, giving unprecedented full 3D plasma shape reconstruction and control capability, with the flexibility to develop and test a wide variety of control schemes.

However, the overall throughput required for the complete transfer of information from the sensors to the central control system cannot be handled in real-time. The compromise applied so far is a dual-channel acquisition: one channel for low-latency, low-bandwidth data acquisition, specifically designed for the control system, and a second channel for full-resolution data. The second channel takes advantage of the transient nature of the experimental setup by buffering the data locally and storing all the acquired raw data on the central acquisition server after the pulse. However, the useful information within the signals acquired by both channels is rich only for very short periods, resulting in large amounts of data that are mostly noise for the rest of the pulse. Additionally, most of the non-zero information signals can actually be modeled by a composition of known response functions.

The idea of the present project proposal would be to try the application of time series compression algorithms, specifically trained with the historical information acquired by the full length row signals in the RFX database, at the edge of the sensor devices.

The Thesis path could be organized in 6 main phases:

  • Familiarization with RFX MDSplus data management system
  • Study of existing time series compression algorithms ( with non ML approach )
  • Synthetic dataset elaboration targeted to a simplistic model fitting
  • Extraction of relevant pulses and data mining for a proper real data training input
  • DNN model construction in Tensorflow/Pytorch
  • Training and Validation on GPU for first performance assessment and latency estimation

As a reference of successful attempt already published we propose the following paper:

Deep Dict: Deep Learning-based Lossy Time Series Compressor for IoT Data

Jinxin Liu, Petar Djukic, Michel Kulhandjian, Burak Kantarci ( https://arxiv.org/abs/2401.10396 )

Data della proposta: 30/05/2024

Developement of the Light Impurity Tomography diagnostic for RFX-mod2 experment

Proponente/Relatore RFX: M. Agostini, M. Ugoletti

Relatore Accademico: M. Agostini

Capogruppo: L. Carraro

Responsabile di programma: TBD

Abstract:

The Light Impurity Tomography (LIT) is a spectroscopic diagnostic for measuring the poloidal distribution of the emissivity of low Z impurities in RFX-mod2 experiment. It will observe the emission along the poloidal angle of neutral hydrogen or of other low Z impurities that are present in the edge region.

The diagnostic consists of 7 cameras, installed in 7 different portholes around the poloidal direction, with changeable interference filters. The 7 cameras observe line-integrated signals, and the 2D map of the emissivity of the specific ion species selected by the interferential filter is obtained with a tomographic inversion.

The goal of the thesis is to optimize the position of the cameras for obtaining the best reconstruction of the emissivity. Simulation of the emissivity has to be performed, together with the synthetic data of the cameras. A tomographic algorithm will be developed to obtain the poloidal emissivity to be compared to the simulated one. By minimizing the errors between the simulation and the reconstruction, the best spatial position and orientation of the 7 cameras are determined.

Competenze richieste (se necessarie): Programmazione di base
Data della proposta: 25/11/2024

Experimental investigation of the edge and SOL operational boundaries

RFX Supervisor: A. Stagni

Academic supervisor: M. Giacomin

Head of the RFX research group: M. Zuin

Leader of the RFX research program: T. Bolzonella

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

Predicting, extrapolating and optimizing the heat and particle fluxes on material surfaces, so as to avoid damaging the first wall components, still represents a key challenge in view of safe and reliable operation of future large tokamak devices. In particular, the power and particle exhausts travelling radially towards the plasma-facing components are largely determined by the transport processes taking place in the scrape-off layer (SOL) region at the tokamak boundary. Given their inherently non-linear and turbulent nature, a common understanding of the physical mechanisms responsible for SOL radial transport has not been reached yet. Recent observations have highlighted the importance of the plasma conditions near the separatrix or last closed flux surface (LCFS), separating the confined region from the open field lines comprising the SOL. Leveraging these results, the separatrix operational space (SepOS) modelling framework has been developed for a description of SOL transport processes based on a reduced set of parameters. So far, this framework has been successfully exploited to predict the edge operational boundaries (e.g. L-H transition, density limit, etc.) on ASDEX-Upgrade, while work for its extension to other machines is underway. This thesis project is devoted to making progress in extending the SepOS framework to plasma discharges in the TCV tokamak (Lausanne, Switzerland), which is particularly suited for this task thanks to its great flexibility in the possible plasma configurations and scenarios. This work is structured as follows: the first part is devoted to getting familiar with the signals and working principles of the main SOL profile and fluctuation diagnostics, as well as analyzing their data within a wide range of TCV working scenarios. The outcome of this data analysis phase will be exploited to systematically estimate the relevant edge operational quantities and build a TCV-specific SepOS parameter space, discriminating several operational and disruptive boundaries. Finally, the results will be exploited to predict the operational point and qualitatively assess the SOL transport characteristics of next experimental and reactor-level devices, like DTT and ITER.

Previous experience (if necessary): Basic knowledge of Plasma Physics and fluid models. Knowledge of data analysis techniques and Python is welcome, but not necessary.

Date: 16/12/2024

Study of Cs transport in SPIDER plasma using a test-particle Monte Carlo code

RFX Supervisor: B. Segalini

Academic supervisor: M. Giacomin

Head of the RFX research group: M. Zuin

Leader of the RFX research program: D. Marcuzzi

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

Description of the thesis:

In negative ion sources such as SPIDER, surface processes of negative ion production are enhanced by Caesium evaporation in the plasma chamber. Tracking the Cs particle trajectories and predicting their distribution on the source walls and grids, hence, is a key task to optimise H- extraction and acceleration.

With this thesis work, the student will learn to use a 3D particle tracing Montecarlo code aimed at computing particle trajectories, taking into account many collisional processes, the influence of magnetic and electric fields and plasma interactions. The student will adapt the code to Caesium related processes, and study the influence of the main plasma parameters and source conditions on its distribution on the source surfaces, correlating it to negative ion production.

Previous experience (if necessary):

Date: 19/12/2024

Unsupervised Data Analysis in Machine Learning Using a Classification System for Breakdown Discharge Signals in the MITICA Experiment

Proponente/Relatore RFX: A. Rigoni Garola

Relatore Accademico: Andrea Rigoni Garola (Ingegneria), da definire (Physics of Data)
Capogruppo/Servizio: A. Rigoni Garola

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

The thesis aims to explore unsupervised data analysis methods in machine learning, focusing on the classification of breakdown discharge signals recorded during the MITICA experiment. By employing advanced encoding systems such as a convolutional autoencoder or a transformer model, the research intends to efficiently analyze and synthesize the data acquired from fast measurement systems. These methods will allow for the automatic isolation and characterization of breakdown current waveforms, offering a synthetic representative value to facilitate database comparisons with experimental test parameters and signals.

Background: MITICA, a full-scale prototype of the neutral beam injection (NBI) system for the ITER experiment in France, is a critical component of the international effort led by the RFX Consortium. The NBI system relies on a radiofrequency ion source, an accelerator, and a charge neutralizer to deliver high-energy neutral atoms. The acceleration system of MITICA uses an 8-stage continuous voltage multiplier capable of reaching a maximum grid voltage of 1 MV. To ensure reliable operation under pseudo-vacuum conditions, the system undergoes a conditioning process. This process involves gradually increasing the voltage to optimize material properties and achieve consistent voltage withstand capability. During the conditioning phase, numerous breakdown discharge events occur, where the entire current flows between the acceleration system electrodes and is grounded. These events, characterized by rapid current evolution (~20 μs), are measured through multiple sensors positioned at the electrodes and along primary current transmission lines. The conditioning phase typically spans a prolonged experimental period, encompassing hundreds of breakdown events.

Objective: The primary objective of this thesis is to develop and implement an unsupervised machine learning system to:

  1. Automatically analyze and classify breakdown current waveforms.
  2. Derive synthetic representative values characterizing each discharge.
  3. Enable efficient comparison of these values with the experimental database to correlate signal patterns with specific test parameters.

Methodology:

  1. Data Collection and Preprocessing:

    1. Gather high-resolution signals from the fast measurement systems during breakdown events.
    1. Normalize and preprocess the data to ensure consistency and remove noise artifacts.
  2. Model Development:

    1. Implement a convolutional autoencoder or transformer-based architecture to encode fixed-length signal representations.
    1. Train the model to minimize reconstruction error, ensuring accurate feature extraction from the signals.
  3. Feature Extraction and Clustering:

    1. Use the latent space representations from the model as inputs for clustering algorithms (e.g., k-means, DBSCAN) to group similar discharge events.
    1. Evaluate clustering results against known test parameters to identify patterns and anomalies.
  4. Synthetic Value Derivation:

    1. Develop a metric or index to represent each discharge waveform’s key characteristics based on the latent space features.
    1. Validate the metric by comparing it with the experimental database and observing its correlation with conditioning outcomes.
  5. Validation and Testing:

    1. Test the system on a separate set of experimental data to assess its robustness and scalability.
    1. Perform comparative analysis to benchmark the proposed method against traditional classification approaches.

Expected Outcomes:

  • A functional unsupervised machine learning pipeline for the automatic classification and analysis of breakdown discharge signals.
  • A synthetic metric to represent waveform characteristics, aiding in efficient database comparisons.
  • Insights into the relationship between breakdown event characteristics and experimental parameters during the conditioning phase of MITICA.

Significance: This thesis will contribute to the ongoing development of the MITICA prototype and the broader ITER project by providing advanced tools for data analysis and system optimization. The proposed machine learning approach could potentially be extended to other high-energy physics experiments requiring detailed signal analysis.

Timeline:

  • Month 1-3: Model development and initial training.
  • Month 4-5: Feature extraction and clustering.
  • Month 6: Thesis writing and defense preparation.

Competenze richieste (se necessarie): Advanced programming skills: Python, Tensor Flow
Data della proposta:
13/1/2025

Ensuring Data Integrity and Accountability in Scientific Research Using IPFS and Blockchain for Nuclear Fusion Experiment Data

Proponente/Relatore RFX: A. Rigoni Garola

Relatore Accademico: A. Rigoni Garola (Ingeneria), da definire (Physics of Data)
Capogruppo/Servizio: Andrea Rigoni Garola

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

This thesis investigates the integration of the InterPlanetary File System (IPFS) protocol and blockchain technology to enhance data distribution, integrity, and accountability in scientific and medical research. The focus is on data produced by the MDSplus software, a widely used tool for signal acquisition in nuclear fusion experiments. The research aims to validate the use of IPFS for secure and tamper-proof data distribution while proposing a blockchain-based methodology to maintain an unbroken chain of responsibility from data acquisition to analysis. The implementation and testing of a distributed system using a minimal Kubo server setup will provide insights into network and storage impacts, enabling efficient data sharing within the RFX Consortium and the broader research community.

Background: Modern scientific and medical research heavily relies on the analysis of large datasets. Ensuring data validity and accountability is paramount, particularly when autonomous artificial intelligence (AI) systems are involved. AI systems often operate as black boxes, posing challenges in explaining and verifying their decisions. This lack of transparency risks undermining the credibility of models used to interpret natural phenomena.

Maintaining data integrity across the entire lifecycle – from collection and experimentation to analysis – is critical. Distributed data systems, such as IPFS, offer promising solutions by providing decentralized storage and tamper-proof distribution. By integrating blockchain technology, it becomes possible to establish a chain of responsibility, ensuring that all participants in the data lifecycle are accountable.

Objective: The thesis aims to:

  1. Evaluate the use of IPFS for distributing scientific data produced by MDSplus.
  2. Propose a blockchain-based methodology to ensure accountability across the data lifecycle.
  3. Implement and test a minimal IPFS-based distributed system using a Kubo server.
  4. Assess the network and storage impact of the system for data dissemination within the RFX Consortium and international research institutions.

Methodology:

  1. System Setup:
    1. Deploy a minimal IPFS-based distributed system using a Kubo server to manage MDSplus-generated data.
    1. Configure the system for decentralized storage and efficient data retrieval.

  2. Blockchain Integration:
    1. Develop a methodology to integrate blockchain for tracking data provenance and accountability.
    1. Ensure the chain of responsibility is recorded from the system designer to the analyst.

  3. Performance Evaluation:
    1. Test the system’s performance under various scenarios, including large-scale data sharing and simultaneous access by multiple users.
    1. Analyze network bandwidth and storage requirements for distributed data sharing.

  4. Data Validation:
    1. Validate the system’s ability to prevent data tampering and ensure integrity.
    1. Simulate potential threats to assess the robustness of the IPFS and blockchain integration.

  5. Impact Analysis:
    1. Evaluate the feasibility and scalability of the proposed system for widespread use within the RFX Consortium and other research institutions.
    1. Analyze the potential for adoption in other fields of scientific and medical research.

Expected Outcomes:

  • A functional prototype of a distributed data system for MDSplus using IPFS and blockchain.
  • A poc methodology for maintaining accountability and data integrity.
  • Comprehensive performance metrics on network and storage impacts.
  • Recommendations for deploying similar systems in other scientific domains.

Significance: This research addresses the critical need for secure, transparent, and accountable data management systems in scientific research. By combining the strengths of IPFS and blockchain, the proposed solution ensures data integrity while fostering trust in AI-driven research outcomes. The findings will benefit RFX and nuclear fusion research providing an initial framework applicable to other domains.

Timeline:

  • Month 1-2: Literature review and system requirements analysis.
  • Month 3-4: Blockchain integration and methodology development.
  • Month 5: System testing and performance evaluation.
  • Month 6: Thesis writing and defense preparation.

Competenze richieste (se necessarie): Advanced programming skills: Go, Python, IPFS basics
Data della proposta:
13/1/2025

Drift-kinetic corrections to global mode stability in fast-particle dominated plasmas

Proponente/Relatore RFX: L. Pigatto

Relatore accademico: T. Bolzonella

Capogruppo RFX: D. Terranova

Responsabile di Programma RFX: T. Bolzonella

Tipologia: Numerica, Teorica, Modellistica, Compilativa

Abstract:

Global stability of a Tokamak plasma, in the framework of ideal MHD, roughly depends on the toroidal current density and pressure gradients. The same ingredients play a crucial role in the performance (confinement, fusion production) of the same plasma, with desirable configurations often being close to stability thresholds. Assessing this proximity to unstable regimes is important during the design of operational points in future experiments.

Auxiliary heating systems, such as Neutral Beam Injectors (NBI), introduce particle populations with different energies with respect to the plasma thermal bulk. These so-called hot particles can trigger instabilities themselves as well as interact with global ideal MHD modes. When non-ideal contributions come into play, extended MHD models need to be used and validity of such models needs to be considered. In this work a drift-kinetic MHD hybrid model will be applied, through the linear stability code MARS-K, to assess the global stability of a NBI heated plasma. In this peculiar case the neutral beams provide most of the plasma pressure and are designed to achieve beam-target fusion. The thesis will start with studying the beam characteristics and evaluating the applicable physics model, implementing then simulations with increasing complexity.

Competenze necessarie per svolgere con successo la tesi: Ambienti Python e Matlab, corsi di base della Laurea Magistrale in Fisica

Data della proposta: 31/01/2025

Toroidal flow physics and control in fusion plasmas

Proponente/Relatore RFX: L. Pigatto

Relatore accademico: T. Bolzonella

Capogruppo RFX: D. Terranova

Responsabile di Programma RFX: T. Bolzonella

Tipologia: Numerica, Teorica, Modellistica, Compilativa

Abstract:

The velocity field of particles in a fusion plasma is rather important for stability on many scales, from turbulence to global MHD instabilities. Understanding and possibly controlling the momentum sources is thus important and non-trivial at the same time.

Non-axisymmetric (3D) magnetic fields are often used in fusion experiments with different purposes. For example, to correct unwanted components on the equilibrium field, to tailor the edge magnetic topology or to control Magneto-Hydro-Dynamic instabilities. These 3D fields introduce both electromagnetic and neo-classical torques, usually braking the plasma flow in particular near the edge. If this braking is often undesired, it can be also exploited as a method for tuning plasma rotation. This thesis wants to explore the penetration of non-axisymmetric magnetic fields in a wide range of fusion plasmas. Starting from the circular, high aspect ratio device RFX-mod2 and moving then to reactor-relevant models for the European demonstration power plant. The quasi-linear MHD code MARS-Q will be used to simulate toroidal rotation evolution with 3D fields, comparing very different regimes. The results of such a study are relevant both to the RFX-mod2 scientific program and to DEMO design activities

Competenze necessarie per svolgere con successo la tesi: Ambienti Python e Matlab, corsi di base della Laurea Magistrale in Fisica

Data della proposta: 31/01/2025

Linear analysis of tearing mode stability in visco-resistive magneto-hydrodynamic Type: theoretical

RFX Supervisor: P. Zanca

Academic supervisor: L. Piron

Head of the RFX research group: D. Terranova

Leader of the RFX research program: L. Marrelli

Tearing mode are fundamental instabilities in magnetic fusion. They resonate in the plasma, namely a magnetic surface inside the plasma exists where the wave front of the instability has the same helical pattern of the equilibrium magnetic field lines. This surface is called resonant surface. The growth of the tearing mode is associated to the development of a current sheet at the resonant surface. The linear analysis of their stability is solved by the computation of the so-called delta-prime , a quantity which represents the magnitude and phase of the current sheet.

A recent work developed a novel method to compute  for a case in which the current sheet is produced by an external error field into a tearing-mode-stable plasma. The model considered single-fluid magneto-hydrodynamic with resistivity and viscosity. We propose to adapt this method to the case of a plasma with no error-field applied, but intrinsically unstable to tearing modes. The results will be compared to those of standard theories.

Previous experience (if necessary):
Date:
06/02/2025

Ingegneria

Tesi Triennali
proposte dal Consorzio RFX

Analysis of the maintenance status of the Medium Voltage Circuit Breakers of the RFX experiments Medium Voltage Board and proposal of revamping

RFX Supervisor: L. Zanotto
Academic supervisor: N. Marconato
Head of the RFX research group: A. Maistrello
Leader of the RFX research program: L. Marrelli

Tipologia: Sperimental, Theoretical, Numeric, Compilative

Abstract:

The thesis is about the analysis of the maintenance status of the medium voltage circuit breakers of the RFX experiments: the analysis shall identify which kind of maintenance is needed and what revamping actions are necessary to make the system able to operate in the coming years.

Previous experience (if necessary): None
Date: 26/03/2024

Analisi dell’anello di regolazione della componente continua di corrente degli inverter dell’esperimento MITICA

Proponente/Relatore RFX: L. Zanotto / M. Dan

Relatore Accademico:  N. Marconato
Capogruppo: A. Maistrello

Responsabile di Programma: D. Marcuzzi

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Abstract:

la tesi è la continuazione di precedenti lavori di tesi effettuati sull’argomento; si tratta di continuare l’analisi del funzionamento dell’anello di controllo in retroazione della componente continua di corrente iniettata dagli inverter di potenza che forniscono alimentazione alle griglie di accelerazione dell’esperimento MITICA. Lo scopo della tesi consiste nel creare un modello di simulazione del sistema di regolazione e di validarlo con i dati sperimentali, per poi verificare se sono possibili ottimizzazioni dei parametri di regolazione.

Data della proposta:  07/05/2024

Superconductivity for Sustainable Magnetic Fusion Development

Proponente/Relatore RFX: L. Piron

Relatore accademico: L. Piron, L. Salasnich
Capogruppo RFX: D. Terranova

Tipologia: Sperimentale, Teorica, Analisi dati, Compilativa

Abstract:

In tokamaks, the plasma, a fully ionized gas, is confined by magnetic fields. For the economics of fusion to be viable, reducing the energy consumption of the magnets is absolutely necessary. The energy consumption of a magnet is directly related to electrical resistance. Without resistance, electrical consumption would drop to zero, and, as an added benefit, no heat would be generated in the magnets. In this context, the use of superconducting magnets is a game changer: superconductivity, which allows current to flow without resistance at low temperatures, enables electromagnets to meet the demanding requirements of modern tokamaks.
During the thesis work the student will analyze the basic properties of superconducting materials and superconducting magnets in fusion energy applications, providing insights into the current state of the technology and identifying ongoing and future trends in research and development. The activity will be carried out under the joint supervision of Prof. Lidia Piron, who works in the field of plasma physics, and Prof. Luca Salasnich, who works in the field of superconductors and Bose-Einstein condensates.

Data della proposta: 13/01/2025

Tesi Magistrali
proposte dal Consorzio RFX

High-frequency modelling of the Acceleration Grid Power Supply of SPIDER experiment

RFX Supervisor: A. Ferro

Academic supervisor: P. Bettini

Tipologia: Modelling, Teorica, Numerica, Compilativa

Abstract:

SPIDER is the test-bed of the negative ion beam source of the ITER Heating Neutral Beam Injectors (NBI). The Acceleration Grid Power Supply (AGPS)provides negative dc voltages up to-96 kV to the SPIDER acceleration grids, and dc currents up to 75 A. Frequent arc breakdowns occur between the acceleration grids, due to the short gap required by the beam optics. They represent short-circuits at the AGPS output, which cause voltage collapse and high frequency voltage and current oscillations, which stress the AGPS and the other items connected at the same potential. A thesis is proposed, aiming at studying the propagation of these voltage transients along the AGPS output cable, and the voltage fluctuations occurring on the AGPS. The work includes the development of a high-frequency model of the AGPS components, considering the stray inductances and capacitances, estimated analytically, from test reports or through measurements on the field. The results of the model will be compared with the available measurements to provide a first validation, and will give useful inputs to define possible improvements of the plant.

Competenze necessarie per svolgere con successo la tesi:

Data della proposta: 06/02/2025

Analysis of the performance of the Toroidal Power Supply of RFX-mod2 after the upgrade

Proponente/Relatore RFX: F. Santoro

Relatore accademico: P. Bettini

Capogruppo RFX: E. Gaio

Responsabile di Programma RFX: S. Peruzzo

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Abstract:

The Toroidal Power Supply system of RFX-mod is composed of an ac/dc thyristor converter, rated for 3 kV and 16 kA, which feeds 12 identical sectors, divided into 2 groups. Each sector has a dc capacitor bank that feeds an IGCT-based four-quadrant converter, rated for 3 kV and 6 kA. The power supply proved to be reliable and flexible, but in the past experimentation of RFX-mod it resulted the need of more stringent performance in terms of latency, precision of the controlled current and synchronism between the two power supply groups of the two sectors, which were initially not foreseen.

In view of the upgraded RFX-mod2, is under design an enhancement of the Toroidal Power Supply, which consists in a reconfiguration of the circuit that could allow overcoming the problems of synchronism between groups and improving precision of the control system.

After the modification, the Toroidal Power Supply will be tested in order to assess the effective improvement of the performances in comparison with the previous configuration. Aim of the thesis will be to contribute to the preparation and conduction of the tests, analyzing the results also with the aid of circuit simulations and comparing them with measurements from the past shots.

Competenze necessarie per svolgere con successo la tesi: Basic concepts of electrical engineering, MATLAB or Python, Simulink.

Data della proposta: 31/03/2023

Prototype of the new Magnetic Energy Storage and Transfer system: experimental tests and model validation

Proponente/Relatore RFX: A. Maistrello

Relatore accademico: P. Bettini / N. Marconato
Capogruppo RFX: A. Maistrello

Responsabile di Programma RFX: S. Peruzzo

Tipologia: Sperimentale, Teorica, Numerica, Compilativa

Abstract:

The Magnetic Energy Storage and Transfer system (MEST) is a novel concept to supply the SuperConducting (SC) coils based on Superconducting Magnetic Energy Storage (SMES). This innovative technology allows the avoidance of high active power peaks required from the grid and the huge reactive power demand usually associated with the use of thyristor converters.

A small-scale prototype of the MEST is being developed as a proof of principle to explore the feasibility, analyse the performance and identify possible unexpected issues. The thesis is focused on the verification and completion of the electrical models already developed against the final detailed design, the analysis of the experimental results of the tests on the first module of the MEST prototype and the validation and possible improvement of the models.

Competenze necessarie per svolgere con successo la tesi: Basic concepts of electrical engineering and power electronics, MATLAB or Python, Simulink

Data della proposta: 05/03/2024

Design and specification for the construction of the cryogenic, electrical, and cooling penetrations for the DTT vacuum vessel

Proponente: M. Dalla Palma

Capogruppo RFX: A. Rizzolo

Relatore Accademico: P. Sonato

Tipologia: Modelling, Teorica, Numerica, Compilativa

Argomento della tesi:

DTT is one of the largest superconducting tokamak under construction with the mission to get scientific and technological proofs of power exhaust in prospect of the first nuclear fusion power plant [1, 2]. The 5.5MA maximum plasma current, 6T toroidal magnetic field at the plasma center, and 2.19m plasma radius make DTT a flexible and compact facility for testing D-shaped plasmas with different configurations of heat load spreading.

The mechanical systems of DTT are designed and integrated analysing interfaces consistently with machine operating states including plasma operation, disruptions, baking, seismic event, testing, and maintenance.

Multi-purpose ports are designed for the DTT vacuum vessel to house in each port a combination of more than one of the following three types of systems:

  • diagnostics probing the plasma;
  • auxiliary plasma heating systems (ECRH, ICRH, neutral beam injection with shielding plates in the port duct);
  • services (divertor and first wall cooling tubes, vacuum pumping including cryolines, in-vessel coil feeders, cables for sensors embedded into the in-vessel components).

In particular, the use of multi-purpose ports introduces the need to integrate one port bellows (compensating relative displacements between the vacuum vessel and the cryostat) and more service bellows (compensating relative displacements between the services and the cryostat) in the same port.

The allocation of services, diagnostics, and auxiliary plasma heating systems is defined, but the design of supports and displacement compensation systems has to be developed. Interfaces between the vessel ports and in-port systems have been analysed in order to address the structural integrity verification and the heat transfer analysis in particular during baking and plasma operation.

The design parameters resulting from this analysis and verification activity will be used to prepare suitable technical specifications for the procurement of the vacuum vessel and the cryostat of DTT.

The proposed activity could include CAD design, numerical calculations, finite element analyses, verification of the simulation results, design of welded joints, preparation of the technical specification for the procurement, interactions and follow-up with possible suppliers.

References

[1] R. Ambrosino, “DTT – divertor tokamak test facility: A testbed for DEMO,” Fusion Engineering and Design, vol. 167, p. 112330, 2021

[2] R. Martone, R. Albanese, F. Crisanti, A. Pizzuto, P. Martin Eds.. “DTT Divertor Tokamak Test facility Interim Design Report, ENEA (ISBN 978-88-8286-378-4), April 2019 (“Green Book”)”

Data della proposta: 18/06/2024

Analyses for the integration of seismic isolation dampers in the DTT facility Analisi per l’integrazione di isolatori e dissipatori sismici nella macchina DTT

Proponente: M. Dalla Palma

Capogruppo RFX: A. Rizzolo

Relatore Accademico: P. Sonato

Tipologia: Modelling, Teorica, Numerica, Compilativa

Argomento della tesi:

DTT is one of the largest superconducting tokamak under construction with the mission to get scientific and technological proofs of power exhaust in prospect of the first nuclear fusion power plant [1, 2]. The 5.5MA maximum plasma current, 6T toroidal magnetic field at the plasma center, and 2.19m plasma radius make DTT a flexible and compact facility for testing D-shaped plasmas with different configurations of heat load spreading.

The mechanical systems of DTT are designed and integrated analysing interfaces consistently with machine operating states including plasma operation, disruptions, baking, seismic event, testing, and maintenance.

The DTT facility will be supported by a basement designed between the torus complex and the ground foundation in the tokamak hall. The integration of isolation dampers in the basement is under evaluation as a seismic protection in order to dynamically separate the torus complex from the foundation, then the amount of energy that is transferred to the tokamak during an earthquake is reduced significantly. 

The study of the isolation system includes the identification of the type of bearing like elastomeric pads, sliding plates or inverted pendulums to support the weight of the structure and to provide some level of energy dissipation, typically in the form of hysteretic damping.

Equivalent static and seismic spectra analyses can be performed using the finite element method and with acceleration spectra reduced with respect to the normalised response of ground motion. Moreover, a CAD model can be developed to verify the system integration and interfaces.

The activity will be carried out in collaboration with the DTT Building Team.

References

[1] R. Ambrosino, “DTT – divertor tokamak test facility: A testbed for DEMO,” Fusion Engineering and Design, vol. 167, p. 112330, 2021

[2] R. Martone, R. Albanese, F. Crisanti, A. Pizzuto, P. Martin Eds.. “DTT Divertor Tokamak Test facility Interim Design Report, ENEA (ISBN 978-88-8286-378-4), April 2019 (“Green Book”)”

Data della proposta: 18/06/2024

Analysis of the performance of the Toroidal Power Supply in view of the upgrade for RFX-mod2

Proponente/Relatore RFX: F. Santoro

Relatore accademico: P. Bettini

Capogruppo RFX: A. Maistrello

Responsabile di Programma RFX: L. Marrelli

Tipologia: Modelling, Teorica, Numerica, Compilativa

Abstract:

The Toroidal Power Supply system of RFX-mod is composed of an ac/dc thyristor converter, rated for 3 kV and 16 kA, which feeds 12 identical sectors, divided into 2 groups. Each sector has a dc capacitor bank that feeds an IGCT-based four-quadrant converter, rated for 3 kV and 6 kA.

In the past experimentation of RFX-mod the power supply proved to be reliable and flexible, but more stringent performance in terms of latency, precision of the controlled current and synchronism between the two power supply groups, initially not foreseen, resulted necessary for improving the plasma performance and the control of instabilities.
In view of the upgraded RFX-mod2, an enhancement of the Toroidal Power Supply is under design. It consists in a reconfiguration of the circuit that will allow overcoming the problems of synchronism between groups and precision of the control system.

After the modification, the Toroidal Power Supply will be tested in order to assess the effectiveness of the improvement, comparing the performance with the previous configuration.

Aim of the thesis is the design of a circuital model to analyze the operation of the system with the new configuration, to tune the control parameters and to assess the performance, by means of simulation. The results will be compared with measurements from the past shots. Depending on the thesis period, it may be possible to participate in the preparation and conduction of the tests.

Competenze necessarie per svolgere con successo la tesi: Electrical engineering, power electronics, MATLAB or Python, Simulink.

Data della proposta: 15/04/2024

Riprogettazione dei dispositivi elettronici per l’iniettore di neutri diagnostico di RFX-mod2

Proponente: M. Barbisan, A. Maistrello, R. Cavazzana, M. Bigi

Relatore Accademico: Da determinare a seconda del corso di laurea dello studente

Capogruppo: L. Carraro

Tipologia: Modelling, Teorica, Numerica, Sperimentale

Argomento della tesi:
Nell’ambito del progetto NEFERTARI, finanziato del PNRR, le diagnostiche dell’esperimento RFX-mod2 (Consorzio RFX, Padova) sono fase di manutenzione ed aggiornamento. Tra di esse vi è l’iniettore di neutri diagnostico (DNBI), un dispositivo che produce e inietta un fascio di atomi H/D ad alta energia (50 keV) all’interno del plasma di RFX-mod2. La luce emessa dall’interazione tra fascio e plasma, una volta analizzata dalle diagnostiche CXRS ed MSE, permetterà di misurare grandezze fisiche quali le concentrazioni di varie impurità, il flusso di ioni, la temperatura ionica e intensità e direzione del campo magnetico.

Il DNBI, installato nel 2005 dal Budker Institute for Nuclear Physics (Novosibirsk, Russia), necessita di importanti opere di manutenzione e aggiornamento, tra le quali si annoverano:

  • la sostituzione del sistema di controllo ed acquisizione dati.
  • la riprogettazione dell’elettronica per la generazione dei 50 kV necessari ad accelerare il fascio.
  • La riprogettazione dell’High Voltage Deck (HVD) e dei dispositivi in esso contenuti; con HVD si intende la gabbia di Faraday che ospita al potenziale di 50 kV le alimentazioni della sorgente di ioni del DNBI.
  • Una revisione generale dei cablaggi, per diminuire il rischio di interferenze elettromagnetiche causate da RFX-mod2 e dal DNBI stesso, e per ridurre i rischi a cose e persone legato a potenziali guasti.

Il lavoro di tesi consiste nell’affiancare attivamente ricercatori e tecnici coinvolti in queste attività. Potrà essere richiesto di partecipare a test elettrici sul DNBI. Obiettivo finale del lavoro di tesi è supportare la stesura delle specifiche tecniche per l’acquisto di componenti e dispositivi elettrici necessari alla rimessa in funzione del DNBI.

Competenze richieste (se necessarie):

Data della proposta: 30/09/2024

Electromagnetic modelling and design of a multi gap passive stabilizing shell for Reversed Field Pinch experiments

Relatore RFX: D. Abate, L. Marrelli

Relatore Accademico: N. Marconato

Tipologia: Modelling, Teorica, Numerica, Compilativa

Abstract:

The present thesis focuses on the electromagnetic modelling and design of a multi gap solution for the passive stabilizing shell of magnetic confinement fusion experiments in reversed field pinch (RFP) configurations. The modelling activity aims at studying the effects of multiple gaps in the passive stabilizing shell surrounding a RFP plasma, by means of electromagnetic models with different level of complexity. The effect of different number and shape of gaps will be investigated in terms of plasma stability and engineering complexity, by developing dedicated numerical codes based on FEM/BEM approach. The selected multi gap solution will then be analysed in terms of engineering design and manufacturing processes.

Competenze richieste: Elettrotecnica, Elettrotecnica computazionale, conoscenza base di linguaggi di programmazione e ambienti di sviluppo (es.: COMSOL, ANSYS, MATLAB)

Data della proposta: 16/12/2024

Unsupervised Data Analysis in Machine Learning Using a Classification System for Breakdown Discharge Signals in the MITICA Experiment

Proponente/Relatore RFX: A. Rigoni Garola

Relatore Accademico: Andrea Rigoni Garola (Ingegneria), da definire (Physics of Data)
Capogruppo/Servizio: A. Rigoni Garola

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

The thesis aims to explore unsupervised data analysis methods in machine learning, focusing on the classification of breakdown discharge signals recorded during the MITICA experiment. By employing advanced encoding systems such as a convolutional autoencoder or a transformer model, the research intends to efficiently analyze and synthesize the data acquired from fast measurement systems. These methods will allow for the automatic isolation and characterization of breakdown current waveforms, offering a synthetic representative value to facilitate database comparisons with experimental test parameters and signals.

Background: MITICA, a full-scale prototype of the neutral beam injection (NBI) system for the ITER experiment in France, is a critical component of the international effort led by the RFX Consortium. The NBI system relies on a radiofrequency ion source, an accelerator, and a charge neutralizer to deliver high-energy neutral atoms. The acceleration system of MITICA uses an 8-stage continuous voltage multiplier capable of reaching a maximum grid voltage of 1 MV. To ensure reliable operation under pseudo-vacuum conditions, the system undergoes a conditioning process. This process involves gradually increasing the voltage to optimize material properties and achieve consistent voltage withstand capability. During the conditioning phase, numerous breakdown discharge events occur, where the entire current flows between the acceleration system electrodes and is grounded. These events, characterized by rapid current evolution (~20 μs), are measured through multiple sensors positioned at the electrodes and along primary current transmission lines. The conditioning phase typically spans a prolonged experimental period, encompassing hundreds of breakdown events.

Objective: The primary objective of this thesis is to develop and implement an unsupervised machine learning system to:

  1. Automatically analyze and classify breakdown current waveforms.
  2. Derive synthetic representative values characterizing each discharge.
  3. Enable efficient comparison of these values with the experimental database to correlate signal patterns with specific test parameters.

Methodology:

  1. Data Collection and Preprocessing:

    1. Gather high-resolution signals from the fast measurement systems during breakdown events.
    1. Normalize and preprocess the data to ensure consistency and remove noise artifacts.
  2. Model Development:

    1. Implement a convolutional autoencoder or transformer-based architecture to encode fixed-length signal representations.
    1. Train the model to minimize reconstruction error, ensuring accurate feature extraction from the signals.
  3. Feature Extraction and Clustering:

    1. Use the latent space representations from the model as inputs for clustering algorithms (e.g., k-means, DBSCAN) to group similar discharge events.
    1. Evaluate clustering results against known test parameters to identify patterns and anomalies.
  4. Synthetic Value Derivation:

    1. Develop a metric or index to represent each discharge waveform’s key characteristics based on the latent space features.
    1. Validate the metric by comparing it with the experimental database and observing its correlation with conditioning outcomes.
  5. Validation and Testing:

    1. Test the system on a separate set of experimental data to assess its robustness and scalability.
    1. Perform comparative analysis to benchmark the proposed method against traditional classification approaches.

Expected Outcomes:

  • A functional unsupervised machine learning pipeline for the automatic classification and analysis of breakdown discharge signals.
  • A synthetic metric to represent waveform characteristics, aiding in efficient database comparisons.
  • Insights into the relationship between breakdown event characteristics and experimental parameters during the conditioning phase of MITICA.

Significance: This thesis will contribute to the ongoing development of the MITICA prototype and the broader ITER project by providing advanced tools for data analysis and system optimization. The proposed machine learning approach could potentially be extended to other high-energy physics experiments requiring detailed signal analysis.

Timeline:

  • Month 1-3: Model development and initial training.
  • Month 4-5: Feature extraction and clustering.
  • Month 6: Thesis writing and defense preparation.

Competenze richieste (se necessarie): Advanced programming skills: Python, Tensor Flow
Data della proposta:
13/1/2025

Ensuring Data Integrity and Accountability in Scientific Research Using IPFS and Blockchain for Nuclear Fusion Experiment Data

Proponente/Relatore RFX: A. Rigoni Garola

Relatore Accademico: A. Rigoni Garola (Ingeneria), da definire (Physics of Data)
Capogruppo/Servizio: Andrea Rigoni Garola

Tipologia: Sperimentale, Teorica, Modellistica, Compilativa

Abstract:

This thesis investigates the integration of the InterPlanetary File System (IPFS) protocol and blockchain technology to enhance data distribution, integrity, and accountability in scientific and medical research. The focus is on data produced by the MDSplus software, a widely used tool for signal acquisition in nuclear fusion experiments. The research aims to validate the use of IPFS for secure and tamper-proof data distribution while proposing a blockchain-based methodology to maintain an unbroken chain of responsibility from data acquisition to analysis. The implementation and testing of a distributed system using a minimal Kubo server setup will provide insights into network and storage impacts, enabling efficient data sharing within the RFX Consortium and the broader research community.

Background: Modern scientific and medical research heavily relies on the analysis of large datasets. Ensuring data validity and accountability is paramount, particularly when autonomous artificial intelligence (AI) systems are involved. AI systems often operate as black boxes, posing challenges in explaining and verifying their decisions. This lack of transparency risks undermining the credibility of models used to interpret natural phenomena.

Maintaining data integrity across the entire lifecycle – from collection and experimentation to analysis – is critical. Distributed data systems, such as IPFS, offer promising solutions by providing decentralized storage and tamper-proof distribution. By integrating blockchain technology, it becomes possible to establish a chain of responsibility, ensuring that all participants in the data lifecycle are accountable.

Objective: The thesis aims to:

  1. Evaluate the use of IPFS for distributing scientific data produced by MDSplus.
  2. Propose a blockchain-based methodology to ensure accountability across the data lifecycle.
  3. Implement and test a minimal IPFS-based distributed system using a Kubo server.
  4. Assess the network and storage impact of the system for data dissemination within the RFX Consortium and international research institutions.

Methodology:

  1. System Setup:
    1. Deploy a minimal IPFS-based distributed system using a Kubo server to manage MDSplus-generated data.
    1. Configure the system for decentralized storage and efficient data retrieval.

  2. Blockchain Integration:
    1. Develop a methodology to integrate blockchain for tracking data provenance and accountability.
    1. Ensure the chain of responsibility is recorded from the system designer to the analyst.

  3. Performance Evaluation:
    1. Test the system’s performance under various scenarios, including large-scale data sharing and simultaneous access by multiple users.
    1. Analyze network bandwidth and storage requirements for distributed data sharing.

  4. Data Validation:
    1. Validate the system’s ability to prevent data tampering and ensure integrity.
    1. Simulate potential threats to assess the robustness of the IPFS and blockchain integration.

  5. Impact Analysis:
    1. Evaluate the feasibility and scalability of the proposed system for widespread use within the RFX Consortium and other research institutions.
    1. Analyze the potential for adoption in other fields of scientific and medical research.

Expected Outcomes:

  • A functional prototype of a distributed data system for MDSplus using IPFS and blockchain.
  • A poc methodology for maintaining accountability and data integrity.
  • Comprehensive performance metrics on network and storage impacts.
  • Recommendations for deploying similar systems in other scientific domains.

Significance: This research addresses the critical need for secure, transparent, and accountable data management systems in scientific research. By combining the strengths of IPFS and blockchain, the proposed solution ensures data integrity while fostering trust in AI-driven research outcomes. The findings will benefit RFX and nuclear fusion research providing an initial framework applicable to other domains.

Timeline:

  • Month 1-2: Literature review and system requirements analysis.
  • Month 3-4: Blockchain integration and methodology development.
  • Month 5: System testing and performance evaluation.
  • Month 6: Thesis writing and defense preparation.

Competenze richieste (se necessarie): Advanced programming skills: Go, Python, IPFS basics
Data della proposta:
13/1/2025

Contatti per richieste e informazioni

Gli studenti interessati ad una delle tesi disponibili possono contattare
il coordinatore delle attività di tesi Prof. Leonardo Giudicotti (leonardo.giudicotti@unipd.it).

Seguici su: