A decade-long operation of the Korean Superconducting Tokamak Advanced Research (KSTAR) has contributed significantly to the operation of superconducting tokamak devices and the advancement of tokamak physics which will be beneficial for the ITER and K-DEMO programs. Even with limited heating capability, various conventional as well as new operating regimes have been explored and have achieved improved performance. As examples, a long pulse high-confinement mode operation with and without an edge-localized mode (ELM) crash was well over 70 and 30 s, respectively. The unique capabilities of KSTAR allowed it to improve the capability of controlling harmful instabilities, and they have been instrumental in uncovering much new physics. The highlights are that the L/H transition threshold power is sensitive to the resonant magnetic perturbation (RMP) and insensitive to non-resonant magnetic perturbation. Co-Ip offset rotation dominated by an electron channel predicted by general neoclassical toroidal viscosity theory was confirmed. Improved heat dispersal in a divertor system using three rows of rotating RMP was demonstrated and predictive control of the ELM-crash with a priori modeling was successfully tested. In magnetohydrodynamic physics, validation of the full reconnection model (i.e. q0  >  1 right after the sawtooth crash) and self-consistent validation of the anisotropic distribution of turbulence amplitude and flow in the presence of the 2/1 island with theoretical models were achieved. The turbulence amplitude induced by RMP was linearly increased with the slow RMP coil current ramp-up time (i.e. the magnetic diffusion time scale). The Dα spikes (i.e. ELM-crash amplitude) was linearly decreased with the turbulence amplitude and not correlated with the perpendicular electron flow. In the turbulence area, a non-diffusive ‘avalanche’ transport event and the role of a quiescent coherent mode in confinement were studied. To accommodate the anticipation of a higher performance of the KSTAR plasmas with the increased heating powers, a new divertor/internal interface with a full active cooling system will be implemented after a full test of the new heating (neutral beam injection II and electron cyclotron heating) and current drive (CD) (Helicon and lower hybrid CD) systems. An upgrade plan for the internal hardware, heating systems and efficient CD system may allow for a long pulse operation of higher performance plasmas at βN  >  3.0 with f bs ~ 0.5 and Ti  >  10 keV.

A dual-frequency microwave imaging reflectometry system was commissioned to measure both coherent and turbulent electron density fluctuations in KSTAR plasmas. Imaging of the density fluctuations is achieved with an array of 16 vertically aligned detectors and two X-mode probe beam frequencies (tunable over 78–92 GHz between plasma discharges). The system provides the capability of fluctuation measurements with poloidal wavenumbers (kθ) up to ∼3 cm−1 at the maximum sampling rate of 2 MHz. Following extensive laboratory tests, the system was further tested with known coherent density fluctuations during the precursor oscillation of the m/n = 1/1 internal kink mode. The phase information of the reflected beam was compared with the precursor oscillation of the electron temperature measured by an electron cyclotron emission (ECE) radiometer. Density fluctuation levels (δne/ne) at two radial positions separated by the inversion radius (inside and outside) were comparable to temperature fluctuation levels (δTe/Te) from ECE signals. Subsequently, two correlation analysis methods were applied to turbulent fluctuation measurements in a neutral beam heated L-mode plasma to determine the mean poloidal rotation velocities of density fluctuations at two radial positions. The measured mean poloidal velocities were ∼8.4 km s−1 at r/a ∼ 0.6 and ∼5 km s−1 at r/a ∼ 0.7 in the clockwise direction, which differed by 1–2 km s−1 with the projected poloidal velocities from the toroidal rotation velocity measured by charge exchange recombination spectroscopy.

Multiple (two or more) flux tubes are commonly observed inside and/or near the q = 1 flux surface in KSTAR tokamak plasmas with localized electron cyclotron resonance heating and current drive (ECH/CD). Detailed 2D and quasi-3D images of the flux tubes obtained by an advanced imaging diagnostic system showed that the flux tubes are m/n = 1/1 field-aligned structures co-rotating around the magnetic axis. The flux tubes typically merge together and become like the internal kink mode of the usual sawtooth, which then collapses like a usual sawtooth crash. A systematic scan of ECH/CD beam position showed a strong correlation with the number of flux tubes. In the presence of multiple flux tubes close to the q = 1 surface, the radially outward heat transport was enhanced, which explains naturally temporal changes of electron temperature. We emphasize that the multiple flux tubes are a universal feature distinct from the internal kink instability and play a critical role in the control of sawteeth using ECH/CD.

The central safety factor (q0) during sawtooth oscillation has been measured with a great accuracy with the motional Stark effect (MSE) system on KSTAR and the measured value was However, this measurement alone cannot validate the disputed full and partial reconnection models definitively due to non-trivial off-set error (~0.05). Supplemental experiment of the excited m  =  2, m  =  3 modes that are extremely sensitive to the background q0 and core magnetic shear definitively validates the ‘full reconnection model’. The radial position of the excited modes right after the crash and time evolution into the 1/1 kink mode before the crash in a sawtoothing plasma suggests that in the MHD quiescent period after the crash and before the crash. Additional measurement of the long lived m  =  3, m  =  5 modes in a non-sawtoothing discharge (presumably ) further validates the ‘full reconnection model’.