Discussion

Magnetic diagnostic measurements react only to the integrated magnetic field and are therefore most suitable for providing information regarding integral parameters. Nevertheless, even with the relatively limited set of diagnostics considered here, FP reconstructions based on magnetic signals successfully reproduce the NEMEC flux surface geometry over our database to within a few millimetres. Indeed, with only two signals, $\cos(m=1)$ and $\cos(m=2)/\cos(m=1)$, the recovery accuracy is 8 mm on-axis and drops to 1.5 mm at the edge. This should be sufficient for rapid visualization of discharges.

We reiterate that all reconstructions use only experimentally known inputs and require less than 10 ms on a 300MHz UltraSPARC workstation. This is fast enough to follow the evolution of the plasma during a discharge by performing a reconstruction every few milliseconds. The application to feedback control is obvious, since scalar parameters such as $W_{\mathrm{p}}$ can be evaluated much faster than the 3-D flux surface geometry. If necessary, the reconstructions could be further accelerated by parallelizing the (independent) Fourier coefficient recoveries.

Our analysis also shows that on W7-AS, with the exception of the poloidal field coil array and to some extent the $\cos(2\theta)$ coil, the available magnetic diagnostics are mainly sensitive to the plasma energy content. This poses a problem from the point of view of magnetic-based FP reconstructions for diagnostic systems lacking some type of poloidal field coil array, since the important second principal component in section 3.6.1 is unmeasurable. With information only on the plasma energy content (section 3.6.3), the reconstruction accuracy is quite limited and probably inadequate to be of practical use on the experiment. Indications for an ideal magnetic-only diagnostic setup point to numerous local measurements of the field which can be used to measure Fourier-like harmonics of the external field to reasonable accuracy, possibly combined with a diamagnetic measurement for an independent check of the energy content. This will be investigated further in chapter 5.

The most severe limitation to a scheme based on magnetic measurements is the lack of profile information, which restricts the accuracy with which the internal flux surfaces can be identified. If the magnetic information could somehow be supplemented by profile measurements, the technique could give far more accurate reconstructions assuming that the experimental data were internally consistent. This is elaborated on in chapter 4.