Improvements in the fast vertical control systems in KSTAR, EAST, NSTX and NSTX-U

Transcription

1 1 PPC/P8-17 Improvements in the fast vertical control systems in KSTAR, EAST, NSTX and NSTX-U D. Mueller 1, N.W. Eidietis 2, D. A. Gates 1, S. Gerhardt 1, S.H. Hahn 3, E. Kolemen 1, L. Liu 5, J. Menard 1, Y.S. Park 4, S.A. Sabbagh 4, S.W. Yoon 3, and B.J. Xiao 5 1 Princeton Plasma Physics Laboratory, Princeton Univ., Princeton, N.J , U.S.A. 2 General Atomics, San Diego, CA, , U.S.A. 3 National Fusion Research Institute, Daejon, Korea 4 Columbia University, New York, NY, 10027, U.S.A. 5 Institute of Plasma Physics, Chinese Academy of Sciences, Hefei Anhui, China contact of main author: mueller@pppl.gov Abstract. The control of the vertical position in tokamak plasmas is essential for exploring the operational control space. In particular, the vertical control system must be capable of responding to transients as quickly as possible and controlling plasmas near the vertical stability limit. The design of the coils systems constrains the ultimate response frequency that is available, but in order to realize the potential of the coil design it is essential to have adequate diagnostics of the position and a feedback control system capable of that. 1. Introduction The realization of a wide variety of plasma shapes with varying plasma current density profiles places challenging demands on the vertical control system in a tokamak. In particular the bootstrap current scales quadratically with elongation at fixed normalized βn = βtab T /Ip where I p is the plasma current, B T is the toroidal field, and β T = P/(BT 2 /2µ0) and P is the volume averaged plasma pressure. The plasma elongation is controlled by the action of coils that generally act to produce a field shape with an index of curvature that becomes closer to vertical instability as the elongation is increased. In devices such as the Korean Superconducting Tokamak Advanced Research (KSTAR), The Experimental Advanced Superconducting Tokamak (EAST) and ITER, the superconducting coils are separated from the plasma by conducting structures that increase the response time of the plasma to changes in the coil currents. Fast control of the plasma vertical motion, essential for stable operation at high elongation and disruption avoidance, can be accomplished with coils internal to the vacuum vessel in these devices. Each of the four devices, NSTX, EAST, KSTAR and NSTX-U, uses its own version of the General Atomics (GA) plasma control system (PCS)[1],[2],[3],[4] to control the coil power supplies with feedback on plasma parameters. The shape control systems in these tokamaks are generally based upon equilibrium analysis that does not lend itself well to fast control and in particular cannot produce a reliable time derivative of the vertical position which is necessary for fast response. The simple analysis of standard integrated magnetic signals to

2 2 PPC/P8-17 Figure 1. The geometry of the EAST and KSTAR devices showing the locations of the control coils and the sensors discussed in the text. produce a vertical position (z) signal can be incapable of yielding a derivative term (dz/dt) with sufficient signal-to-noise ratio to be adequate for control. Furthermore, for the two existing superconducting tokamaks, the slow response of the plasma to the distant superconducting coils necessitates the introduction of coils internal to the vacuum vessel in order to provide faster control of vertical instabilities. 2. Discussion The difference of the signal from up-down symmetric pairs of voltage loops can be used to provide a better signal for the dz/dt in the PD control loop of z. The NSTX control system used the analog difference of a pair of voltage loops, with 2 khz low-pass filters to remove signals due to n=1 plasma instabilities, to provide the dz/dt signal. This signal was used in the control of the difference in the rectifier power supplies for up/down difference in a pair of outer PF coils. Use of this improved signal-to-noise ratio signal realized an improvement in vertical control that allowed elongation (κ) to be increased by about 0.2.[5] NSTX-U will have 6 pairs of voltage loops instrumented to provide dz/dt for feedback control. Use of the RMP saddle coils with 7 khz power supplies for vertical control will be tested for use in vertical control if the use of the slower rectifier supplies on the poloidal field coils proves inadequate at high elongation with high internal plasma inductance (li). Analysis of these signals indicate they will provide an excellent representation of dz/dt for NSTX-U.[6] The KSTAR control system uses coils internal to the vacuum vessel to provide control of vertical instabilities and has instrumented 2 pairs of voltage loops, whose positions are indicated in Figure 1, to provide a fast dz/dt that have better signal to noise than the derivative of signals derived from flux loops or Mirnov coils as can be seen in Figure 2. Relay feedback[7] was used to optimize the vertical PD control gains. The use of the voltage loop

3 3 PPC/P8-17 Figure 2. The derivative of the fast z signal derived from integrated magnetics sensors shown in black is exhibits much more noise that the dz/dt term from the difference in a pair of up-down symmetric voltage loops shown in blue. pairs in the control system has been successfully tested, and proven to have better signal to noise ratio than is provided by the derivative of integrated signals. Insufficient operational experience exists so far to quantify the expansion of the available operational space with the improved control response. The EAST control system has similarly been modified to include two pairs of loop voltage difference signals as indicated in Figure 1. Furthermore, the internal coil power supply has been modified to use voltage control rather than current control for faster response. The improvement in the signal to noise ratio for dz/dt based on the loop voltage pairs compared that derived from the derivative of the z position derived from the integrated magnetics is shown in Figure 3. Figure 3. EAST data showing the dz/dt signal derived from the loop voltage signals compared to that from the derivative of z derived from the integrated magnetics signals.

4 4 PPC/P8-17 Figure 4. Comparison of plasma current, elongation, IVC current, feedback terms and z position in KSTAR as a function of time for two discharges. In shot 8870 shown in black, the z reference in the IVC control loop was zero. For shot 8872, the z-position reference was adjusted beginning at 3 s to achieve IVC current ~0 and z p ~ 0 after 4 s in plasmas with otherwise identical control. The implications of this discrepancy is discussed in the text. In each device, the ability of the fast vertical control is affected by the slower shape control so the target values for the fast and slow control should be compatible in order to provide good control. In Figure 4 an example of a DND plasma in KSTAR for which the z request in the fast control loop was adjusted to match the z provided by the slower shape control is shown in blue. This results in more symmetrical plasma with higher kappa than the discharge shown in black without such an adjustment. Small differences in the vertical position derived from simple, but fast systems demanded for vertical control and that derived from the shape based on real-time equilibrium analysis (rtefit)[8] used for control of the superconducting coils are almost unavoidable. If that difference exceeds some value, the internal coils will be controlled to their maximum value and the fast control response will be lost. A remedy for this conflict is to employ a high-pass fast filter on the error in the z measurement, the target value minus the measurement, used in the control of the internal coils. This solution uses the fast internal coil response to operate only on the high frequency vertical movement of the plasma while the slower superconducting coils control slower motion. This avoids the necessity of empirically tuning the z reference from shot-to-shot as agreement between the z signal and the shape control changes.

5 5 PPC/P8-17 Figure 5. The plasma vertical motion was ramped by controlling DRSEP shown in the second frame which permitted zp to be ramped 0.1 m while maintaining fast vertical stability and avoided large current in the internal coil (IC). This remedy has been employed in the EAST PCS system and has proved reliable to avoid saturation of the internal coil current while providing fast vertical control. The high-pass filter used in the EAST discharge shown in Figure 5 maintained stable control of the fast motion of the plasma while keeping the internal coil s current near zero while the shape control of DRSEP was used to move the plasma vertical position under control of the superconducting coils. 3. Summary Use of the voltage loop-based sensors for dz/dt in NSTX, KSTAR and EAST was successful in improving the signal to noise ratio of dz/dt used for vertical control and allowed operation closer to the vertical stability limit with an increase in achievable elongation in NSTX of 0.2. Use of multiple loop voltage pairs on NSTX-U will be commissioned in upcoming experimental campaigns. The use of a high-pass filter in the fast control loop was successful in both the superconducting tokamaks to eliminate the contention between control from the internal coils and the slower, but more powerful superconducting PF coils while maintaining control of fast transients with the internal coils. 4. Acknowledgements This research was supported by U.S. DOE contract DE-AC02-09CH11466.

6 6 PPC/P8-17 References: [1] FERRON, J.R., et al., A flexible software architecture for tokamak discharge control systems, Proc. Of the 16th IEEE/NPSS Symposium on Fusion Engineering, Champaign, IL Vol. 2, (1995) 870. [2] GATES, D., et al., Control system development plan for the National Spherical Torus Experiment, IEEE Trans. Nucl. Sci., 47, (2000) 222. [3] XAIO, B.J., et al., EAST plasma control system, Fusion Engineering and Design 83 (2008) , doi: /j.fusengdes [4] HAHN, S.H., et al., Plasma control system for Day-One operation of KSTAR tokamak, Fusion Engineering and Design 84 (2009) , doi: /j.fusengdes [5] GATES, D.A., et al., Progress towards steady state on NSTX, Nucl. Fusion 46 (2006) S22-S28, doi: / /46/3/s04. [6] GERHARDT, S.P., et al., Magnetics For Equilibrium Reconstruction and Realtime Plasma Control in NSTX-Upgrade, Review of Scientific Instruments 20th Topical Conference Proceedings on High-Temperature Plasma Diagnostics (2014) (in press). [7] HAHN, S.H., et al., Progress and improvement of KSTAR plasma control using model-based control simulators, Fusion Engineering and Design 89 (2014) [8] FERRON, J.R., et al., Real time equilibrium reconstruction for tokamak discharge control, Nucl. Fusion 38 (1998)