Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Tokamak devices aim to magnetically confine a hydrogen plasma at sufficiently high pressure to achieve net energy production from nuclear fusion of light isotopes. Predictive modeling and optimization is crucial for reliable operation of tokamak reactors, like ITER and DEMO.Exploitation of a fusion reactor requires a reliable stationary operating point, maximizing the reactor performance, while maintaining a safe margin from physics instabilities and engineering limits. Active control of plasma profiles aims to improve confinement and MHD stability. Of particular interest are rational values of the safety factor q, where instabilities can be triggered, like sawteeth and neoclassical tearing modes (NTMs). Hybrid and advanced scenarios prolong the pulse duration by shaping the q profile.Fast and reliable access to the desired operating point needs to be ensured during the plasma current ramp-up phase. An optimized ramp-down strategy is critical for safe and fast termination. Navigating the plasma state through a stable envelope within the operating space demands proper understanding of the non-linear plasma dynamics, including the impact of actuators like heating, current drive, gas injection and plasma shaping.By combining fast numerical schemes with reduced physics models, the RAPTOR transport solver achieves rapid simulation of plasma profile dynamics, including current diffusion and transport of heat and particles. A range of transport models is available, varying from empirical to first-principles-based. A new stationary state solver is developed for rapid plasma scenario optimization. Coupled to a neural network surrogate transport model, the ITER hybrid scenario is optimized at different plasma currents, densities and pedestal heights and for varying heating mix. We show how the electron cyclotron current drive (ECCD) deposition can be optimized to shape the q profile, aiming for a maximum fusion gain Q, while maintaining q>1, avoiding sawtooth oscillations. We find the optimum plasma current for maximizing Q and describe the operating ranges enabled by 20MW and 40MW of EC power.ASDEX Upgrade (AUG) allows to study advanced scenarios with elevated q profile in reactor-relevant conditions. Fast formation of an elevated q profile is achieved through heating during the ramp-up, and maintained stationary by off-axis neutral beam injection (NBI) and central counter-current ECCD. Inter-discharge modeling in RAPTOR is applied to optimize the NBI onset time and ECCD deposition, aiming for an early stationary phase, while avoiding the onset of NTMs. We demonstrate how trajectory optimization allows to obtain and sustain the desired q profile, including an early qmin
António João Caeiro Heitor Coelho