Analysis and concept development towards MHD measurement capability of a magnetic diagnostic system on DEMO

This work reports on the progress obtained during the EuroFusion FP9 work-program (2020–2025) towards developing an overall concept for the system design for the DEMO magnetics, focusing specifically in this contribution on the use of magnetic sensors for the measurement of coherent low-frequency magnetic instabilities with a relatively low toroidal mode number. These instabilities, such as Neoclassical Tearing Modes, can be very detrimental to the stability and performance of a tokamak discharge, and need to be suitably and efficiently detected and then controlled for the safety of the discharge. We find that the currently foreseen equi-spaced, and thus low-resolution toroidal distribution of the DEMO magnetic sensors does not allow a sufficiently correct detection of toroidal modes higher than |n| = 1 for the 4-sectors and |n| = 3 for the 8-sectors arrangements. This is one toroidal mode lower than thetheoretical maximum given by the Nyquist value, which is valid only in the idealized case of no noise nor measurement uncertainties in all the input signals being analysed. This problem occurs simply because the unavoidable electrical noise and the different sources of measurement and processing uncertainties may cause the phase shift between adjacent sensors to become ≥ π. This situation can be improved by adding a relatively small number of high-resolution toroidal sensors. We initially propose to position these additional sensors very close to the nominal location of the 4-sectors and 8-sectors arrangements so as to facilitate their potential in-vessel installation. The detection capabilities improve significantly, allowing to obtain correct measurements up to |n| = 10 in essentially all cases analysed and that are relevant for DEMO. The currently foreseen poloidal distribution of the DEMO magnetic sensors suffers much less from issues linked to the phase shift between adjacent sensors being close to π, this only becoming apparent for |m| ≥ 10 for the in-vessel and for |m| ≥ 20 for the ex-vessel poloidal arrays. This could allow inferring concurrently toroidal mode numbers higher than those obtained with the low-resolution toroidal arrays (|n| = 1 and |n| = 3) only if one single poloidal mode is associated to a single toroidal mode n at the resonant q-surface and no additional poloidal harmonics are present at the Last Closed Flux Surface due to toroidicity and ballooning effects and only if the different sources of noise and measurement and processing errors can be kept reasonably low. The addition of the high-resolution toroidal sensors significantly improves the concurrent detection of toroidal and poloidal mode numbers, which is now feasible up to |n| = 7 in essentially all cases analysed. In terms of analysis methods that could be used for the spatial decomposition of the magnetic signals in DEMO, the SparSpec 1D (and 2D) methods suffer much less than other methods from the limitations due to the phase shift between adjacent signals being close to π and can then in principle handle relatively better the different sources of noise and errors in the input signals. The V5ν0 version of the SparSpec-1D algorithm has specifically been developed to correctly treat such perturbations to the exact input signals. It is therefore proposed to add, for future activities, the SparSpec-V5ν0 method to the analysis suite for DEMO to test the resilience of the measurement performance vs. such signal perturbations.