polito.it
Politecnico di Torino (logo)

Bayesian optimization of the CHERAB code for the reconstruction of the D-α camera view in ST40 tokamak

Raffaele Piergallini

Bayesian optimization of the CHERAB code for the reconstruction of the D-α camera view in ST40 tokamak.

Rel. Fabio Subba, Matteo Moscheni. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Energetica E Nucleare, 2023

[img]
Preview
PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (6MB) | Preview
Abstract:

The next generation of nuclear fusion devices aims to achieve and sustain significantly improved plasma performance compared to existing tokamaks. This requires reliable quantification of plasma properties, power, and particle exhaust during the design phase and diagnosis during operations. Moreover, unprecedented levels of neutron irradiation are expected in pilot plants, which pose operational and servicing challenges. In this context, synthetic diagnostics are valuable tools for the fusion community. They support the interpretation of experimental data, which is inherently challenging, and enable careful diagnostic design and integrated analyses for capturing essential plasma features. The D-α camera, for example, can infer properties of the neutral deuterium distribution in the chamber, making it useful for diagnosing the edge plasma, particularly in detached divertor regimes. This work, conducted in collaboration with Tokamak Energy Ltd, utilizes the 3D Monte-Carlo inverse ray-tracing software CHERAB to simulate the 2D perspective of a synthetic D-α camera. The primary objective is to extract valuable neutral parameters all throughout the reactor chamber from these simulations. In order to speed up the computational time, the D-α emission source is created using simplified analytical models for both the core and edge plasma. These models involve numerous free parameters, necessitating their optimization to accurately match experimental results. Due to the inherent characteristics of the Monte-Carlo code, the optimization process necessitates the utilization of a derivative and gradient-free method. Specifically, Bayesian optimization process is employed for this purpose. In the two plasma shots examined, a notable level of qualitative agreement has been observed, although achieving precise quantitative agreement poses a significant challenge due to the inherent three-dimensional nature of the ST40 neutral emission originating from the wall.

Relatori: Fabio Subba, Matteo Moscheni
Anno accademico: 2023/24
Tipo di pubblicazione: Elettronica
Numero di pagine: 55
Soggetti:
Corso di laurea: Corso di laurea magistrale in Ingegneria Energetica E Nucleare
Classe di laurea: Nuovo ordinamento > Laurea magistrale > LM-30 - INGEGNERIA ENERGETICA E NUCLEARE
Ente in cotutela: Tokamak Energy Ltd (REGNO UNITO)
Aziende collaboratrici: Tokamak Energy Ltd
URI: http://webthesis.biblio.polito.it/id/eprint/29811
Modifica (riservato agli operatori) Modifica (riservato agli operatori)