Observation of thermal events on the plasma facing components of Wendelstein 7-X

Sitjes, A. Puig;Gao, Y.;Jakubowski, M.;Drewelow, P.;Niemann, H.;Ali, A.;Moncada, V.;Pisano, F.;Ngo, T. T.;Cannas, B.;Sleczka, M.

2019-01-01

---

Abstract

Long pulse operation of present and future magnetic fusion devices requires sophisticated methods for protection of plasma facing components from overheating. Typically, thermographic systems are being used to fulfill this task. Steady state operation requires, however, autonomous operation of the system and fully automatic detection of abnormal events. At Wendelstein 7-X (W7-X), a large advanced stellarator, which aims at demonstrating the capabilities of the stellarator line as a future fusion power plant, significant efforts are being undertaken to develop a fully automatic system based on thermographic diagnostics. In October 2018, the first divertor-based experimental campaign has been finished. One of the goals of this operation phase (named OP1.2) was to study the capabilities of the island divertor concept using an uncooled test divertor made of fine-grain graphite tiles. Throughout this campaign, it was possible to test the infrared imaging diagnostic system, which will be used to protect the actively water-cooled plasma facing components (PFCs) during the steady-state operation in the next experimental campaign. An overview of the most relevant thermal events on the PFCs that were detected in OP1.2 using this system are presented. This includes events that limited operation during the campaign, like baffle hot spots and divertor overloads, events that are potentially critical in steady state operation like leading edges, events caused by the ECRH and NBI heating systems (shine-through hot spots and fast particle losses) and other events which are a common source of false alarms like surface layers. The detected thermal events are now part of an important and extensive image database which will be used to further automate the system by means of computer vision and machine learning techniques in preparation for steady-state operation, when the system must be able to detect dangerous events and protect the machine in real-time