Integrated modelling of plasma scenarios for JET and JT-60SA Tokamaks: validation and predictions for future JT-60SA experiments

To accelerate the realisation of fusion energy, the Broader Approach (BA) agreement was signed between the European Atomic Energy Community (Euratom) and Japan in 2007. As part of this agreement, the Satellite Tokamak Program (STP) included the construction of the superconducting tokamak JT-60SA in Naka, Japan, which achieved its first plasma on the 23rd of October 2023. JT-60SA aims to address key physics and engineering challenges essential for the development of future demonstration power plants and to support the exploitation of ITER by mitigating its operational risks and studying advanced plasma scenarios. Through JT-60SA experiments, scientists will gain knowledge on how to operate a superconducting tokamak. Integrated modelling is essential for predicting and interpreting the behaviour of thermonuclear plasmas by simultaneously simulating a variety of physical phenomena that occur across different regions, each with its own distinct scales and geometries. This approach can support the development of plasma scenarios to achieve the desired experimental goals. The work presented in this thesis aims to contribute to the scientific exploitation of JT-60SA by supporting through integrated modelling the scenario development in view of its second operational phase (OP2), expected to start in 2026. This will be the first operational phase in which high heating power and enhanced diagnostics will be available. The JT-60SA target scenarios will be developed in stages, starting with lower plasma currents, magnetic fields, and auxiliary heating power. The present work provides the first prediction, performed using advanced 1.5-dimensional transport codes, of the performance expected during the second operational phase of JT-60SA. The models used for JT-60SA predictions are validated against experiments conducted at JET in 2023, aiming to develop a scenario with dimensionless parameters as close as possible to those envisaged for JT-60SA. The aim of the experiments was to establish a high-β scenario comparable to JT-60SA Scenario #4.2 (advanced inductive) and Scenario #5.1 (non-inductive steady-state), while exploring the MHD limits at various magnetic field strengths and NBI power levels. This experiment provided a unique opportunity to test different transport models that have not been extensively employed in scenarios with relatively high values of βN, allowing to evaluate their performance in such conditions. However, it is important to note that JET lacked the advanced current drive and current density profile tailoring capabilities that will be available in JT-60SA, namely the high energy off-axis Negative Neutral Beam Injection (N-NBI) and Electron Cyclotron Resonance Heating (ECRH), making it challenging to replicate the exact conditions expected in JT-60SA. In fact, the JET scenario analysed in this thesis relied exclusively on NBI heating, as no ECRH power was installed. Furthermore, the JET shaping capabilities were constrained, with an elongation (κ) limited to below 1.6 and a triangularity (δ) below 0.4, compared to the higher shaping parameters of JT-60SA scenarios, which features κ ∼ 1.9 and δ ∼ 0.47. The modelling has been performed with the JINTRAC suite of codes, coupled to both semi-empirical (i.e. Bohm/gyro-Bohm and CDBM), and reduced first-principle transport models (i.e. QuaLiKiz and TGLF), on three JET pulses at different magnetic field (1.7, 2.0 and 2.4 T) and with varying Neutral Beam Injection power (13.5, 16 and 19 MW, respectively). Overall, the modelled plasma kinetic profiles and time traces exhibit good agreement with the experimental data across all magnetic field levels, with the majority of the modelled profiles falling within the experimental error bars. Among the transport models employed, the Bohm/gyro-Bohm transport model exhibits the best agreement, consistently providing an upper boundary for the prediction of temperatures, βN, total plasma energy, and neutron rate. In contrast, the CDBM model consistently provides a lower boundary for these key plasma parameters, with experimental results consistently falling within the range predicted by the two models. QuaLiKiz consistently overpredicts the electron density peaking across all magnetic fields, whereas this behaviour is not observed in TGLF at 1.7 T and 2.0 T, and only begins to emerge at 2.4 T, although to a lesser extent than QuaLiKiz. The JET DTE3 experimental campaign in 2023 presented a valuable opportunity to evaluate the performance of the advanced high-β scenario at 2.4 T and 19 MW in a D-T plasma mixture. However, when conducted in D-T, the pulse failed to achieve stable performance due to the presence of deleterious MHD activity at the end of the ramp-up phase, which had not been observed in the reference deuterium pulse. In order to avoid this, the NBI power was reduced to 16 MW to achieve a stable deuterium shot. The discrepancy observed can be attributed to the higher sputtering yield in D-T compared to D, which results in a modified radiation profile and lower temperature peaking. These changes influence the evolution of the central safety factor (q0), the normalised beta (βN), and the associated MHD activity, ultimately impacting the plasma stability. In a subsequent deuterium experimental campaign, the ramp-up phase was adjusted to replicate the βN and q0 evolution observed in the unstable D-T shot. This was achieved by modifying the electron density to influence temperature peaking, successfully reproducing the same MHD activity of the unstable D-T shot. Unfortunately, the optimisation of the ramp-up phase could not be implemented in D-T due to limited experimental time. However, to evaluate the potential performance achievable in D-T at higher NBI power, JINTRAC extrapolations were performed, extrapolating the D pulse to a D-T mixture and increasing the NBI power. These simulations were conducted to estimate the achievable βN during the flat-top phase, assuming the deleterious MHD activity could have been avoided through proper optimisation of the ramp-up phase. Extrapolating D shots to a D-T mixture remains a significant challenge due to the differences in transport behaviour arising from the varying isotope mass and the distinct sputtering yields; accurately modelling the latter would necessitate complex coupled core-SOL (scrape-off layer) simulations. The JINTRAC integrated modelling framework was subsequently employed to simulate the ramp-up and flat-top phases of the baseline and hybrid scenarios envisioned for the second operational phase of JT-60SA (OP2). These simulations provide valuable insights into the feasibility of achieving the target plasma parameters, laying the foundation for future physics studies within the JT-60SA Experiment Team. Due to the good agreement achieved with the JET validation, as well as its relatively low computational demand, the Bohm/gyro-Bohm semi-empirical transport model has been chosen to predict JT-60SA scenarios. The hybrid scenario was modelled at both 3.7 MA/2.28 T and 2.7 MA/1.7 T with an auxiliary heating power of 19 MW by scaling down the plasma current, magnetic field, density, and power from the reference METIS simulation at 3.5 MA/2.28 T with 37 MW of heating power. Optimisation of the ramp-up phase with respect to METIS was necessary to achieve a safety factor profile above one with a low magnetic shear region at lower power. The optimisation indicates that a slower current ramp-up in the initial phase (0.5 to 6 s), combined with earlier NBI injection and a more rapid density increase, is required to maintain qmin > 1. The scenario was evaluated at three different values of the Greenwald density fraction (fGW = ne/nGW = 0.8, 0.6, 0.4). Results indicate that a hybrid-like q profile can be sustained at fGW = 0.8 and 0.6 for the 3.7 MA case and at fGW = 0.8 for the 2.7 MA case. At lower densities, hollow current density profiles and reversed q-profiles occur due to the significant off-axis penetration of the Negative-NBI. A solution was explored involving the use of only the upper injector unit. At 2.7 MA, with fGW = 0.4 and 19 MW, values of βN = 3 are predicted, suggesting that exploring the high-β regime of the advanced inductive scenario could be feasible even during the initial research phase. The baseline scenario envisaged for Operational Phase 2 (OP2) at 4.6 MA/2.28 T (fGW = 0.4) has been modelled under various assumptions regarding the normalised critical pressure gradient (αcrit) in the pedestal, which is directly linked to its height. A scan of this parameter has been conducted based on results from the EPED1 code, suggesting a higher temperature pedestal than that predicted by METIS using scaling laws. Results show that increasing the pedestal temperature at a fixed density leads to an increase in both the stored thermal energy and the confinement enhancement factor (H98). Moreover, the core plasma kinetic profiles scale proportionally with the pedestal increase, maintaining a consistent gradient in the core region. A confinement factor of H98 = 1.07 is predicted at pped = 20 kPa, with a βN = 1.8, typical of the baseline scenario, and a stored energy of around 10 MJ. These results underscore the importance of an accurate model for predicting both density and temperature in the pedestal, which is essential for reliably assessing the performance of the scenario. The thesis is organised as follows: • In Chapter 1 the main approaches to fusion are presented. A brief history of the JET and JT-60SA tokamaks is provided, alongside a detailed description of their main characteristics. A comparison between the two devices highlights how JET has contributed to the development of JT-60SA scenarios, and how JT-60SA will build on results of JET to further develop non-inductive steady-state operations. • Chapter 2 outlines the theoretical background of the transport simulations performed in this work. A description of the models and of the JINTRAC integrated modelling framework is also provided. • In Chapter 3 the JET programme in support of JT-60SA is discussed. Experimental results are summarised, and the validation of various transport models across three pulses at different magnetic field strengths is explored. Additionally, extrapolations to higher NBI power in Deuterium-Tritium (D-T) scenarios are presented. • Chapter 4 focuses on predictive modelling for the second operational phase (OP2) of JT-60SA. The optimisation of the ramp-up phase of the hybrid scenario is analysed, with particular attention to the impact of plasma current ramp rate, density ramp velocity, and NBI heating timing on the safety factor profile. The baseline scenario is also investigated, with emphasis on the effects of pedestal assumptions on the predicted performance. • In Chapter 5 the conclusions of the work are drawn, and potential future research directions are outlined.

Per accelerare la realizzazione dell'energia da fusione, nel 2007 è stato firmato l'accordo Broader Approach (BA) tra la Comunità Europea dell'Energia Atomica (Euratom) e il Giappone. Nell'ambito di tale accordo, il Programma Tokamak Satellitare (STP) ha incluso la costruzione del tokamak superconduttore JT-60SA a Naka, in Giappone, che ha raggiunto il suo primo plasma il 23 ottobre 2023. JT-60SA mira ad affrontare le principali sfide fisiche e ingegneristiche essenziali per lo sviluppo di futuri impianti dimostrativi di energia da fusione e a supportare lo sfruttamento di ITER mitigandone i rischi operativi e studiando scenari avanzati di plasma. Attraverso gli esperimenti su JT-60SA, gli scienziati acquisiranno conoscenze su come operare un tokamak superconduttore. La modellizzazione integrata è fondamentale per prevedere e interpretare il comportamento dei plasmi termonucleari, simulando simultaneamente una varietà di fenomeni fisici che si verificano in diverse regioni, ciascuna con proprie scale e geometrie. Questo approccio può supportare lo sviluppo di scenari di plasma per raggiungere gli obiettivi sperimentali desiderati. Il lavoro presentato in questa tesi mira a contribuire allo sfruttamento scientifico di JT-60SA, supportando tramite modellizzazione integrata lo sviluppo degli scenari in vista della sua seconda fase operativa (OP2), prevista per il 2026. Sarà la prima fase operativa in cui saranno disponibili alta potenza di riscaldamento e diagnostiche avanzate. Gli scenari target di JT-60SA saranno sviluppati in fasi, partendo da correnti di plasma, campi magnetici e potenza di riscaldamento ausiliario inferiori. Il presente lavoro fornisce la prima previsione, effettuata utilizzando codici di trasporto avanzati 1.5-dimensionali, delle prestazioni attese durante la seconda fase operativa di JT-60SA. I modelli utilizzati per le previsioni su JT-60SA sono stati convalidati rispetto agli esperimenti condotti su JET nel 2023, con l'obiettivo di sviluppare uno scenario con parametri adimensionali il più vicino possibile a quelli previsti per JT-60SA. Lo scopo degli esperimenti era stabilire uno scenario ad alto β comparabile agli scenari #4.2 (induttivo avanzato) e #5.1 (non induttivo stazionario) di JT-60SA, esplorando i limiti MHD a diverse intensità di campo magnetico e livelli di potenza NBI. Questo esperimento ha offerto un'opportunità unica per testare diversi modelli di trasporto, che non erano stati ampiamente utilizzati in scenari con valori relativamente alti di βN, consentendo di valutarne le prestazioni in tali condizioni. Tuttavia, è importante notare che JET mancava delle capacità avanzate di guida di corrente e modellizzazione del profilo di densità di corrente che saranno disponibili su JT-60SA, come il Negative Neutral Beam Injection (N-NBI) off-axis ad alta energia e il riscaldamento per Risonanza Ciclotronica Elettronica (ECRH), rendendo difficile replicare esattamente le condizioni previste per JT-60SA. Infatti, lo scenario JET analizzato in questa tesi si è basato esclusivamente sul riscaldamento NBI, poiché non era installata alcuna potenza ECRH. Inoltre, le capacità di shaping di JET erano limitate, con un'elongazione (κ) sotto 1.6 e una triangularità (δ) sotto 0.4, rispetto ai parametri di shaping più elevati degli scenari di JT-60SA, che prevedono κ ∼ 1.9 e δ ∼ 0.47. La modellizzazione è stata effettuata con la suite di codici JINTRAC, accoppiata a modelli di trasporto sia semi-empirici (Bohm/gyro-Bohm e CDBM) che ridotti di primo principio (QuaLiKiz e TGLF), su tre impulsi di JET a diversi campi magnetici (1.7, 2.0 e 2.4 T) e con vari livelli di potenza NBI (13.5, 16 e 19 MW, rispettivamente). Nel complesso, i profili cinematici del plasma modellizzati e le tracce temporali mostrano un buon accordo con i dati sperimentali a tutti i livelli di campo magnetico, con la maggior parte dei profili modellizzati all'interno delle barre di errore sperimentali. Tra i modelli di trasporto impiegati, il modello di trasporto Bohm/gyro-Bohm presenta il miglior accordo, fornendo costantemente un limite superiore per la previsione delle temperature, βN, energia totale del plasma e tasso di neutroni. Al contrario, il modello CDBM fornisce costantemente un limite inferiore per questi parametri chiave del plasma, con risultati sperimentali che rientrano costantemente nell'intervallo previsto dai due modelli.