Using e-beam mapping to detect coil misalignment in NCSX

The primary object of the e-beam mapping simulation program on NCSX is to develop requirements for the hardware and machine capabilities necessary for the actual e-beam mapping experiments. The magnetic flux surface configuration was constructed using a numerical code, based on the Biot-Savart law, to calculate the magnetic field components and trace the field line trajectory many times around the torus. Magnetic surfaces are then mapped by recording the field line intersections with toroidal cross-sections of the magnetic system, much as in an actual e-beam mapping experiment. In the course of these calculations, a catalog of many hundreds of vacuum magnetic configurations was compiled, each with varying sensitivity to the coil displacements. The NCSX coils were designed to provide good magnetic surfaces at high beta with significant bootstrap current. The coil set includes separately powered modular, toroidal field, and poloidal field coils, and can produce a wide range of magnetic configurations. Many of the vacuum field configurations with low order rational surfaces have finite, stellarator-symmetric islands present. Nevertheless, despite the presence of these islands, configurations have been found which will allow, we believe, the identification of modular and poloidal field coil displacements of < 0.5 mm. There was generally less sensitivity to toroidal field coil displacement, and a novel approach of energizing a subset of the toroidal field coils at higher current is proposed. By using half of the toroidal field coils, at twice the current, it is possible to detect alignment errors of less than approximately 1 mm. These results assume that the spatial resolution of the e-beam mapping apparatus is of order 5 mm, a previously achieved result for the luminescent rod method. We have also investigated the possibility of performing the initial e-beam mapping (and possibly start-up) studies in NCSX using two or fewer power supplies for the coils in the magnetic system (18 MC + 6 PFC + 18 TFC). There is a potential advantage of minimizing the complexity and cost of the initial mapping phase. More importantly, reducing the number of power supplies reduces uncertainties in the field mapping from current regulation or measurement of the currents. We find that a single source powering all modular coils in series produces good magnetic surfaces, however, without the most sensitive ι = 0.5 rational surface. With the addition of current in the toroidal field coils, the ι = 0.5 resonance can be introduced and the previous sensitivity to coil displacements of order 1 mm is recovered. Future work will focus on developing methods to identify specific types of coil misalignments. (author)