- Email: [email protected]
Design of ECH launcher for KSTAR advanced Tokamak operation
Fusion Engineering and Design 151 (2020) 111395
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Design of ECH launcher for KSTAR advanced Tokamak operation a,
a
a
a
a
T a
Mi Joung *, Minho Woo , Jongwon Han , Sonjong Wang , Sunggug Kim , Sanghee Hahn , Dongjea Leeb, Jonggu Kwaka, Robert Ellisc a b c
National Fusion Research Institute, Daejeon, Republic of Korea Ulsan National Institute of Science and Technology, Ulsan, Republic of Korea Princeton Plasma Physics Laboratory, Princeton, United States
A R T I C LE I N FO
A B S T R A C T
Keywords: KSTAR ECH Launcher Launcher control Advanced tokamak operation
For the advanced tokamak operation in Korea Superconducting Tokamak Advanced Research (KSTAR), up to 6 MW of electron cyclotron heating (ECH) is planned. The ECH launcher should be able to operate in steady state and serve as a central/off-axis heating and current drive, q-profile control, and control of the neoclassical tearing mode (NTM) and sawtooth. The water cooled launchers have been upgraded to include continuous operation with precise, high-speed real-time control of the beams by integration with the plasma control system (PCS) for supporting the experiments, in particular for suppressing NTM. In this study, the launcher was successfully operated for 78 s with about 650 kW EC power, where the temperature of the coolant, with a flow rate of 9.6 lpm, is saturated and approximately 2 °C. The launcher control system had a maximum control cycle of 5 kHz and a motor latency of about 10 ms, and the poloidal scanning speed of the beam was 20°/s. Additionally, for the higher localization of power deposition, the curvature and size of the mirrors in the launcher were optimized. This paper presents the ECH launcher design and its control system with the operation results.
1. Introduction Since the first plasma operation of Korea Superconducting Tokamak Advanced Research (KSTAR), electron cyclotron heating (ECH) has played a key role in various experimental results such as ECH pre-ionization [1], ECH-assisted startup [2], a plasma rotation study [3,4], an impurity transport study [5], a high poloidal beta operation, and a long pulse operation [6]. The main purpose of KSTAR is to explore the physics and technologies of high performance steady-state operations that are essential for International Thermonuclear Experimental Reactor (ITER) research and future fusion reactors [7]. To achieve this goal, totally 28 MW of external heating power was planned. The main heating systems are planned to provide 12 MW until year 2019 and 6 MW until year 2022 by neutral beam injection (NBI) and ECH, respectively, before the major upgrade of KSTAR to prepare long pulse and high beta discharges. All ECH systems are required to have a power capability to handle 1 MW RF power for 300 s which is the maximum pulse length of the KSTAR plasma. As the KSTAR operation capability has been improved, the launcher also has been gradually upgraded from the initial design for 500 kW, 2 s pulse toward higher power compatibilities, a longer pulse, and faster movement in real-time [8,9]. In order to achieve the
⁎
ultimate goal to withstand a continuous-wave (CW) of 1 MW, an activewater cooling system is necessary. However, this constrains flexibility and the mobility of the mirror in terms of the speed and the coverage of the mirror steering. A wide range of beam coverage beneficially widens the range of possible operation scenarios and applicable plasma parameters. Moreover, accurate movement of the highly localized beam was crucial in controlling magneto-hydrodynamic (MHD) instabilities such as the neoclassical tearing mode (NTM) [10–12] and sawtooth [13]. This paper discussed the design of the ECH launcher and controller, as well as the results of the long pulse operation in KSTAR. Section 2 describes the launcher design for the steady state and high performance operation, including the optimization of the steering range and the beam diameter. Section 3 explains the launcher control system and its operation results. The present status and future work are discussed in the Conclusion. 2. Launcher design Fig. 1 shows a layout of the 6-MW ECH system. Among the six1 MWclass 300 s gyrotrons, two 105/140-GHz dual frequency gyrotrons are under operation and two identical gyrotrons are being procured. The other two units will be 170 GHz. Each KSTAR ECH system consists of an
Corresponding author. E-mail address: [email protected] (M. Joung).
https://doi.org/10.1016/j.fusengdes.2019.111395 Received 31 May 2019; Received in revised form 25 October 2019; Accepted 1 November 2019 0920-3796/ © 2019 Elsevier B.V. All rights reserved.
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 1. Layout of the 6 MW ECH system.
injected corresponding to the requested deposition position by two kinds of steering: rotation of the entire steerable mirror assembly and rotation of the steerable mirror about the stationary axis holding the mirror in the assembly. The combination of these two rotation controls permits a 40° poloidal and ± 20° toroidal scanning range. In the previous design, all launcher components were passively cooled by the thermal radiation. However, for the steady state operation, two mirrors that are directly affected by the RF beam had been upgraded with active cooling. The temperature of the residual parts in the launcher design, including the waveguide, does not increase, prior to the melting point. This is because the effects from the RF loss and plasma are not large enough; the launcher is far from the plasma and in the shadow of the other object. The straight corrugated waveguide can be also affected by RF beam. However, in the ideal case, the theoretical loss of HE11 mode at the waveguide is minimal - less than 1% per 100 m at all of the anticipated operating frequencies. The temperature increase can be easily calculated by Ohmic loss at the isolated 2.3 m launcher waveguide via the heat capacity equation:
individual high voltage power supply, a transmission line (TL), and a launcher with a 1 MW-class 300 s gyrotron. The four 105/140 GHz gyrotrons were procured from GYCOM in Russia. The frequency of each gyrotron can be selected based on the operation scenario. The TL is composed of the evacuated corrugated waveguide components in the 63.5 mm inner diameter. This includes the waveguide switch, pump-out tee, RF gate valve, bellows, DC break, and several kinds of miter-bends, e.g., the power monitor, arc detector, and polarizer-set. All waveguide components was delivered by General Atomics (GA) in the USA. Theoretical loss of the TL with seven miter-bends in the length of approximately 65 m is less than 8% for both frequencies. All TL components were designed for the steady-state operation at 1 MW of power, with each operating frequency. KSTAR has two, nearly identical, single channel launchers. The initial launchers were designed and fabricated in collaboration with the Princeton Plasma Physics Lab. (PPPL). Its basic mechanism is almost identical to the DIII-D ECH launcher [14]. The beam angle can be changed in ± 20° poloidally and toroidally by the variation of the steerable mirror’s tilt and rotation angles. As KSTAR operation’s capability has improved, the launcher also has been gradually upgraded. It was initially designed for 500 kW, a 2 s pulse and upgraded toward higher power, longer pulses, and fast movement in real-time [8,9]. The ultimate goal of the handling power for a single launcher is to withstand a CW of 1 MW. For this purpose, the radiative cooling of the initial design is insufficient and active water cooling is required. However, this constrains flexibility and mirror mobility with respect to the speed and coverage possible in steering the mirror. Additionally, in the limited space, the beam diameter at the plasma should be minimized for the highly localized beam. Moreover, the 6 MW ECH was recently assigned to the large middle port for installation of six launchers. Pivot positions of the ECH launchers should be determined in a manner that achieves a high current drive for the advanced tokamak operation.
Q = c⋅m⋅ΔT, Eq.
(1)
where Q is the RF loss, c is the specific heat, and m is the mass of waveguide. The launcher waveguide was specially made from the Nickel coated stainless steel (SS) to increase the mechanical strength and to decrease the eddy current induced by changes to the magnetic field. Note that all other transmission line components are made from aluminum. When 1 MW power is transmitted for 100 s, the SS waveguide with c = 510 J/kg⋅K rises approximately 3.7 K. This indicates that the active cooling is not needed. However, there are some points where the temperature increases further, e.g., where the higher order mode or the gap between two waveguides appears. Therefore, additional temperature monitoring may be required. Fig. 3 shows the actively cooled fixed and steerable mirror models. The design was based on interpretations of the operation results from the passively cooled mirrors, including the preliminary analysis regarding the active water-cooling mirrors, seen in [10]. The absorbed power of the mirror heated by the RF beam varies according to the material of the mirror. Copper alloy is a good candidate for low RF power absorption and sufficient thermal conductivity. In terms of thermal stress, Copper Chromium Zirconium (CuCrZr) may be better than pure copper. However, it is not trivial to braze with the stainless steel. Oxygen-free high thermal conductivity (OFHC) copper was utilized as the reflective surface of the two mirrors. The surface of the fixed mirror is only affected by the RF beam; however, the steerable
2.1. Launcher structure and cooling performance The KSTAR ECH launcher consists of a long waveguide, a first fixed mirror and a second steerable mirror, as shown in Fig. 2. The highpower RF beam emitted from the gyrotron is transmitted to the waveguide of the launcher through the transmission line. There is no vacuum RF window in the launcher as well as in the TL. With regard to vacuum condition, the launcher and the TL are connected directly to the tokamak. The RF beam radiated from the waveguide is reflected at the first-fixed mirror and the second-steerable mirror. The RF beam is 2
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 2. (a) The installation picture and (b) the flange at the Bay-O port, and (c) the layout inside the ECH launcher.
coefficient of 1 W/cm2K. When the fixed mirror with a 0.75 mm thick copper sheet brazed to the 316 L SS block is considered, the temperature increased to 157 °C. Similarly, assuming the heat flux of 30 W/cm2 from the plasma, the temperature rise of the copper steering mirror was 30 °C higher than the fixed mirror and the surface temperature of the mirror with thin copper on the SS block was increased to 215 °C in the simulation. Based on this analysis, the fixed mirror and the steerable mirror were made from the copper block brazed with SS for the steady state operation. The mirrors expect to be stable with regard to temperature rise, however additional thermal stress analysis may be required. The thermal strength of the mirrors would be increased if the copper is replaced with CuCrZr. Until recently, a shutter was used to protect the mirrors from the plasma when the RF beam was not required. Currently, the shutter has been removed to reduce the beam diameter by increasing the mirror size. This allows an easier connection of the cooling tube to the mirrors. The cooling circuit design of the fixed mirror was chosen by considering the inlet and outlet positions in the limited space and the cooling area. For the steerable mirror, the inlet and outlet are located near the steering pivot to minimize the deformation of the bellows in the whole poloidal and toroidal steering ranges. The bellows were changed from short welded bellows to long formed bellows in order to disperse the stress and expand the steering range as shown in the Fig. 2(c). The length and the smallest bending radius of the formed bellow made of SUS316 L are 400 mm and 50 mm, respectively. The designed maximum pressure of the formed bellows is 30 kg/cm2. It was tested up to 15 bar and was typically used at 4 bar. The cross section of the cooling circuit inside the mirror is rectangular with area 7 × 4.5 mm2 and
Fig. 3. Models of (a) the fixed mirror and (b) the steerable mirror.
mirror is affected both by RF and plasma. From the Ref. 15, the temperature increase to approximately 102 °C at the copper fixed mirror was deduced by a finite element thermal analysis performed in ANSYS. Here, the peak heat flux was estimated to be 84 W/cm2 corresponding to the absorption power of 0.2% from an 800 kW RF beam and film 3
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 4. The temperature differences of the inlet and outlet coolants of the mirrors in KSTAR operation for (a) shot #21003 and (b) shot #21706.
8 × 6.5 mm2 for the fixed mirror and for the steerable mirror, respectively. Fig. 4 shows the temperature differences of the outlet and inlet of the coolants on the two mirrors when the RF beams are injected with the pulse length and the power of (a) 12.78 s and approximately 540 kW (shot #21003) and (b) 77.9 s and approximately 650 kW (shot #21706). Due to the long path and the slow flow rate, these differences have a time delay of 18 s and 40 s, for the steerable mirror and the fixed mirror, respectively. The flow rate at the fixed mirror was approximately 4 lpm and less than 9.8 lpm at the steerable mirror. The temperature difference does not precisely reflects the temperature rise, because the inlet temperature of the antenna coolant is not well regulated. The inlet temperature of these selected two results relatively well regulated. However, the gradual decrease and increase of the temperature differences at the end of the pulses can be seen, due to changes of the inlet coolant temperature. Nevertheless, the absorbed power at the mirror can be estimated by the simple correction of the data. In shot #21003, the absorbed powers of the fixed and the steerable mirror are approximately 0.4 kW and 1.1 kW, respectively. In shot #21706, these are 0.52 kW and 1.137 kW by assuming that the temperature difference is saturated for a CW operation. By adding the lost data to the fixed mirror in shot #21003, the results of two shots were similar and reasonable. The results are summarized in the Table 1. In the [15], the absorption rate of the RF power by the mirror was 0.08–0.11%. When the plasma current is 1 MA, the heat flux from the plasma was estimated to be 20–30 W/cm2, respectively. These results are in a good agreement with the estimations in the [15], considering the different plasma current.
cooling tubes caused a strong resistance and narrow rotation angles. Accordingly, the cooling circuit was changed to incorporate the long bellows as shown in the Fig. 2(c). This provided the same operation ranges as the first design. The available scanning range can be changed not only by rotation ranges of the mirror, but also by the pivot position of the launcher. Additionally, the launcher vertical position zpivot can affect the EC current drive efficiency, which is an important factor in q-profile control and MHD instability mode control. Fig. 5 shows EC driven currents as a function of zpivot calculated by the ray tracing code TORAY-GA. It was obtained by injecting 105 GHz ECH beam into the typical plasma, with a plasma current of 400 kA at the toroidal magnetic field of 1.792 T. As the zpivot is away from the mid-plane, the total EC driven current is slightly increased. However, the EC driven current is sensitive to the deposition position of ECCD, zECCD in vertical or rECCD (or ρECCD) in radial. In the case of zpivot = 0, the normalized radius ρECCD is slightly outside compared to ρECCD at the other pivot positions when the ECCD deposition position zECCD=+/- 300 mm. This leads the decrease of the driven current. Due to diffraction, the beam is not easy to approach near the plasma core. The peak current density of ECCD J_max shown in Fig. 5(b) shows that the driven current is more localized at the large zpivot. This calculation is not in perfect agreement with the previous study on NTM suppression [16]. However, the results should not be identical, as the plasma is different. Nevertheless, the qualitative results showing that the EC current drive is better at the large zpivot are the same. Considering the limitations, such as the size of the equatorial port allocated to ECH and the passive stabilizer installed in the vessel, the zpivot of ± 350 mm was decided. This position is slightly increased from the first design with zpivot = 300 mm. This increase takes into account the current drive efficiency and the mirror steering coverage. In order to install six launchers in this Bay-O port, the existing two launchers should be installed vertically and symmetrically. Installing the other four launchers in the rest space requires careful consideration. To solve this problem, a 2-channel launcher was introduced as shown in Fig. 2(b). The configuration and operation principle of the 2-channel launcher are the same as the present launcher. The accessible region by
2.2. Accessible range of the beam As described above, the beam angle is changed through the combination of the two types of rotations of the steerable mirror. In the first design without active cooling, the beam could be injected in the range of 50–90° from the vertical axis poloidally and -20 to+20° from the launcher axis toroidally. However, the cooling circuit interferes with the free motion of the mirror. In fact, the short bellows and long rigid
Table 1 The estimated absorption rates by RF power and the heat flux from the plasma at two mirrors in the KSTAR operation. Shot
Plasma current (pulse length)
RF power (pulse length)
@ fixed mirror
@ steerable mirror
Flow rate
Measured absorbed power
Calculated absorption rate
Flow rate
Measured absorbed power
Calculated heat flux from plasma
∼ 0.5 kW (added lost power) 0.52 kW
0.0926%
9.8 lpm
9.09 W/cm2
0.08%
9.6 lpm
1.1 kW (peak @ 1.7 °C) 1.137 kW (peak @ 1.826 °C)
#21003
∼ 400 kA (16.07 s)
540 kW (12.78 s)
4.0 lpm
#21706
400 kA (77.94 s)
650 kW (77.9 s)
4.1 lpm
4
9.35 W/cm2
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
increase the beam divergence: increasing the distance from the end of the waveguide to the fixed mirror and using the convex fixed mirror. The first case is similar to ITER upper launcher [19]. However, the size of the two mirrors would need to be larger than the present launcher boundary. Using the convex mirror is more effective in the small space. The beam radiated from the waveguide can be represented by a Gaussian beam with a beam waist of ω0 at the end of the waveguide. The radius of beam waist ω0 = 0.322D, where D is the inner diameter of the waveguide [20]. The beam radius ω(z) along the beam path length z is,
ω(z) = ω0 1 + (
λz 2 ) πω02
(2)
where λ is the wavelength of the RF beam. This beam diameter would change after each reflection by the curved mirrors following Gaussian Beam Optics. After reflection by the mirror, the new beam radius ω(z) can be expressed as, 2
2
⎛ ω(z) ⎞ = 1 + ⎛ z − z1 ⎞ ⎝ ω1 ⎠ ⎝ d1 ⎠ ⎜
⎟
⎜
⎟
(3)
where ω1 and z1 are the beam waist radius and the axial waist position before the mirror, and d1=πω12/λ. Finally, the beam diameter of about 60 mm at the resonance layer of Rres = 1.8 m corresponding to the beam path length of about 1240 mm is obtained (Fig. 7 and 8). The average power density of the EC beam was increased by approximately 1.7 times. However, the width of the second mirror was limited by launcher structure. It is approximately 1.7 times of the beam width. Thus, the transmission loss is unavoidable at the second mirror. 3. Launcher control system 3.1. Steering actuators of the mirror The antenna actuator consists of two sets of a motor and an encoder connected to each other in the gear assembly. For fast and accurate control, the pneumatic motor was replaced by a DC motor (Faulhaber, 3272-024CR). As the motor driver was upgraded to a higher current of 50 A, the motor controlled by pulse width modulation (PWM), with the frequency of 20 kHz, is sufficient for the wide ranges of speed and torque. The irregular resistive torque is caused by the mechanical wearing or imperfection of the driving assembly. The absolute magnetic encoder (Posital, UCD-S101B-1212) which conveys the mirror position, is connected to the motor with a 3:1 gear ratio. The absolute magnetic encoder is also connected to the leadscrew, which connects to the mechanical assembly of the mirror. The absolute magnetic encoder with a 24 bit multi-turn resolution provides a resolution of 0.002 °/ count for the mirror angle. It was enveloped by the iron cylinder to avoid a kind of chattering by a high electro-magnetic field.
Fig. 5. (a) The EC current drive and (b) the maximum current density of EC as a function of the vertical position of the launcher.
the beam is represented by the hatched area in the Fig. 6. For the lower launcher, the beam can be deposited from -350 mm to 500 mm in the vertical direction at the major radius R = 1.8 m corresponding to the beam angle from 90 to 50 degrees from the vertical axis. The upper launcher is the opposite. For the toroidal angle, the mirror can be changed from -20 to 20°, but due to the poloidal limiter installed in the vessel, the beam cannot be injected to the counter current direction. This limitation will be resolved by moving the launcher inward. Such a modification has the benefit of minimizing the beam diameter. However, the electromagnetic stress caused by plasma disruption, as well as the thermal and impurity effects from the plasma should be reanalyzed.
3.2. Launcher controller To drive the motor in the real time by Plasma Control System (PCS), the launcher controller has been developed based on a field-programmable gate array (FPGA) (Altera EP1C6Q240C8 with a 27 MHz clock). As the EC power deposition position can be easily changed with the highly localized power by the fast control of the launcher mirror, ECCD is well known as a powerful tool for the suppression and control of MHD instability modes. This is especially true for NTM, which results in a deterioration of plasma performance and occasionally plasma disruption. For NTM control, EC power should be applied to the precise mode location before the mode destroys the plasma. In addition to the fast detection of the mode growth and location, the response of the mirror is also critical in the control sequence. Moreover, during the long pulse discharge, the role of ECH can be changed, for example from off-
2.3. Beam diameter The EC beam diameter in KSTAR has been emphasized because ECH power is marginal to control MHD instabilities. However, the long distance from the last mirror to the plasma, as well as the limited mirror size, which was originally designed for 84 GHz, made it difficult to decrease the beam diameter. In the previous design of the launcher, the minimum size of the EC beam at the plasma center was 80 mm in diameter at 140 GHz [17]. This diameter is larger than the typical island size of NTM in KSTAR plasma [18]. With the limited power, such a broad EC beam is not favorable to suppress NTM. However, in the given space, in order to decrease the beam diameter, increasing the beam divergence could not be avoided. Two methods were investigated to 5
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 6. Accessible area by the hatching of the two launchers (a) in the top view and (b) in the side view.
axis current drive to central heating. The quick movement of the mirror can increase the flexibility of the operation. The block diagram of the launcher control system is shown in Fig. 9. The PCS sends the setpoints, which have been converted to the two encoder values, in real time to the FPGA through the Heating Integrated Control System (HICS). The HICS has a time synchronization function and acts as a bridge between two different controllers: the launcher controller with a serial interface and the PCS with a reflective memory (RFM). The PCS operator can choose how to define the setpoints, which are the vertical target positions at the resonance layer Rres corresponding to the given Bt. The operator can also choose a the tracking parameter, such as normalized poloidal magnetic flux psi, or normalized radius rho, to calculate the setpoints. In this case, the real-time equilibrium reconstruction data (rtEFIT) is involved. The setpoint (target position) is converted to the mirror angles by an algorithm with or without plasma in the PCS. In the case of no plasma or vacuum, mirror angles are calculated with the beam vector from the antenna pivot position to the target position. When the plasma is present, mirror angles are calculated with a real time ray tracing code (real-time
Fig. 7. The traces of the beam radius of 140 GHz and 105 GHz along the beam path where arc length ‘0′ means the end of the launcher waveguide.
Fig. 8. Beam shapes simulated by Code V at the vertical target position of (a) z = -300 mm and (b) z = 0 mm at Rres = 1.8 m when the two mirrors have toroidal curvatures. 6
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 9. KSTAR ECH launcher control system.
TORBEAM [21] with the rtEFIT data with the assumed density and temperature profiles. These angles are calculated at one given toroidal angle during the entire pulse duration and converted to encoder values. The converted encoder values are transferred to the HICS with the cycle of 20 kHz through the RFM, and then to the FPGA in the individual launcher controller of each gyrotron with the serial communication rate of 5 kHz. The FPGA is the main element that operates the EC antenna actuator. FPGA compares the present value of the encoder at a rate of 20 kHz with the commended encoder value from PCS at a rate of 5 kHz. The rate which the FPGA drives the motor is 20 kHz. The control cycle of the launcher controller is 20 kHz; however, the command value was limited to 5 kHz by the serial communication between the HICS and the launcher controller. Fig. 10 shows the PCS control results of the two launcher mirrors with Rres = 1.728 m and perpendicular injection. The latency caused by the inertia of the motor is about 10 ms. As shown in the signals of speed 4 in EC2 launcher and speed 3 in EC3 launcher, the mirror was vibrated or overshot at the high speed. In order to decrease these phenomena, a reverse current braking with opposite voltage is applied to the motor control algorithm. Braking was optimized at speed 3 in EC2 and at speed 2 in EC3. In both cases, the beam scan speed was approximately
Fig. 11. (a) The result of vertical position control of the mirror via the qtracking algorithm in the PCS. (b) Expanded view for t = 2.3 – 3 s.
20 °/s in the beam angle and about 50 cm/s in the vertical length. The misalignment was within about ± 2 mm. The asymmetry between the upward and downward directions shown in the EC3 launcher is a result of the different load. The different load varies according to vacuum pressure, misalignment, wear-out, and so on. Fig. 11 shows the result of the mirror control for q-tracking in real time by rtEFIT in the KSTAR experiment. After starting the q-tracking algorithm at 1.5 s, the new mirror position commands are calculated from the q-value in rtEFIT with geometrical offset, wherein the vertical position z of the mirror was tracking to z = zq=2 – 5 cm along Rres = 1.92 m. Since the latency of the motor response was about 10 ms, the error, or position difference between the command and the returned value of the mirror, was almost equal to the faster variation of the command value. This fluctuating error can be seen from 2.5 s to 3.7 s and at the end of the shot from 5.5 s onward. Except for these unusual periods, the mirror was aligned well to the requested value within 2 cm. Fig. 10. The results of the vertical position scan for the (a) EC2 and (b) EC3 launcher with pre-programmed control in the PCS where speed 1, 2, and 3 mean driving motor PWM duty of approximately 25%, 37%, 51%.
4. Conclusions The design of the ECH launcher for KSTAR’s long pulse, high7
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
performance operation were summarized. The rising temperature of the two water-cooled mirrors was saturated, not exceeding 2 °C for a 78 s long pulse operation in the tokamak experiment. With the cooling pipes included, the steerable mirror allows a beam angle range from 0–40° in the poloidal direction and ± 20° in the toroidal direction. The vertical pivot position of the launcher was optimized to have the highest driving current. Consequently, the beam diameter was minimized up to 60 mm in diameter at the plasma center through the implementation of two curved mirrors. The vertical beam position can be controlled by the FPGA-based controller in real time with a latency of 10 ms and a speed of 20 °/s, which corresponds to the vertical scan speed of 50 cm/s. With these specifications, the launchers operated successfully in a 78 s long pulse operation. Because the poloidal limiters installed in both sides of ECH port constrains the available toroidal beam angle of ECCD, the radial pivot position of the launcher will soon be moved closer to the plasma. For the next four ECH systems, the launcher should be designed as a 2-channel launcher to save the space. However, these will have the same configuration and cooling circuits as the launcher described herein. The improved ECH launcher system is expected to contribute qprofile control and MHD instability mode control to the advanced tokamak operation.
Nucl. Fusion 57 (2017) 066040. [5] Joohwan Hong, S.S. Henderson, Kimin Kim, C.R. Seon, Inwoo Song, H.Y. Lee, Juhyeok Jang, Jae Sun Park, S.G. Lee, J.H. Lee, Seung Hun Lee, Suk-Ho Hong, Wonho Choe, Modification of argon impurity transport by electron cyclotron heating in KSTAR H-mode plasmas, Nucl. Fusion 57 (2017) 036028. [6] Y.M. Jeon, et al., Distinctive features of KSTAR stationary high poloidal beta scenario, talk, 16th International Workshop on H-Mode Physics and Transport Barriers, St. Petersburg, Russia, Sept., 2017, pp. 13–15. [7] G.S. Lee, J. Kim, S.M. Hwang, et al., The KSTAR project: an advanced steady state superconducting tokamak experiment, Nucl. Fusion 40 (2000) 575. [8] M. Joung, M.H. Woo, J.H. Jeong, S.H. Hahn, S.W. Yun, W.R. Lee, Y.S. Bae, Y.K. Oh, J.G. Kwak, H.L. Yang, Real time MHD mode control using ECCD in KSTAR: plan and requirements, AIP Conference Proceedings 506 (2014) 1580. [9] R. Ellis, Y.S. Bae, J. Hosea, M. Joung, D. Miller, W. Namkung, H. Park, Development of Steady-State Mirrors for the KSTAR ECH Launchers, Proceedings, 2013 IEEE Symposium on Fusion Engineering (2013) article No. 6635461. [10] M. Reich, K. Behler, A. Bock, L. Giannone, M. Lochbrunner, M. Maraschek, E. Poli, C. Rapson, J. Stober, W. Treutterer, ASDEX Upgrade team, ECCD Based NTM Control at ASDEX Upgrade, 39th EPS Conference &, 16th Int.Congress on Plasma Physics, 2012 P1.076. [11] E. Kolemen, A.S. Welander, R.J. La Haye, N.W. Eidietis, D.A. Humphreys, J. Lohr, V. Noraky, B.G. Penaflor, R. Prater, F. Turco, State-of-the-art neoclassical tearing mode control in DIII-D using real-time steerable electron cyclotron current drive launchers, Nucl. Fusion 54 (2014) 073020. [12] A. Isayama, G. Matsunaga, T. Kobayashi, S. Moriyama, N. Oyama, Y. Sakamoto, T. Suzuki, H. Urano, N. Hayashi, Y. Kamada, T. Ozeki, Y. Hirano, L. Urso1, H. Zohm, M. Maraschek, J. Hobirk, K. Nagasaki, JT-60 team JT60U, Neoclassical tearing mode control using electron cyclotron current drive and magnetic island evolution in JT-60U, Nucl. Fusion 49 (2009) 055006. [13] T.P. Goodman, F. Felici, O. Sauter, J.P. Graves, the TCV Team, Sawtooth pacing by real-time auxiliary power control in a tokamak plasma, Phys. Rev. Lett. 106 (2011) 245002. [14] R. Ellis, J. Hosea, J. Wilson, R. Prater, R. Callis, Design of a dual high-power, long pulse, steerable ECH launcher for DIII-D, AIP Conference Proceedings 595 (2001), p. 318. [15] R. Ellis, J. Hosea, Additive manufacturing of steady-state mirrors for the KSTAR ECH launchers, Proceedings, 2015 IEEE Symposium on Fusion Engineering (2015) article No. 7482419. [16] Y.S. Park, Y.S. Hwang, Simulation of ECCD optimization for neoclassical tearing mode suppression in KSTAR, Fusion Eng. Desig. 83 (2008) 207–210. [17] S.G. Kim, M. Joung, J.W. Han, S.J. Wang, D.S. Kim, M.S. Choi, Design of a steerable launcher for the ECH system on KSTAR, EPJ Web of Conferences 157 (2017), p. 03022. [18] Kyungjin Kim, Yong-Su Na, Minhwa Kim, Y.M. Jeon, K.D. Lee, J.G. Bak, M.J. Choi, G.S. Yun, S.G. Lee, S. Park, J.H. Jeong, L. Terzolo, D.H. Na, M.G. Yoo, KSTAR Team, Experiment and simulation of tearing mode evolution with electron cyclotron current drive in KSTAR, Curr. Appl. Phys. 15 (2015) 547–554. [19] M.A. Henderson, R. Heidinger, D. Strauss, R. Bertizzolo, A. Bruschi, R. Chavan, E. Ciattaglia, S. Cirant, A. Collazos, I. Danilov, F. Dolizy, J. Duron, D. Farina, U. Fischer, G. Gantenbein, G. Hailfinger, W. Kasparek, K. Kleefeldt, J.D. Landis, A. Meier, A. Moro, P. Platania, B. Plaum, E. Poli, G. Ramponi, G. Saibene, F. Sanchez, O. Sauter, A. Serikov, H. Shidara, C. Sozzi, P. Spaeh, V.S. Udintsev, H. Zohm, C. Zucca, Overview of the ITER EC upper launcher, Nucl. Fusion 48 (2008) 054013. [20] R.L. Abrams, Coupling losses in hollow waveguide laser resonators, IEEE J. Quantum Electron. 838 (1972) Vol. QE-8. [21] E. Poli, A.G. Peeters, G.V. Pereverzev, TORBEAM, a beam tracing code for electroncyclotron waves in tokamak plasmas, Comput. Phys. Commun. 136 (2001) 90–104.
Declaration of Competing Interest None. Acknowledgements This work was supported by National R&D Program funded by the Ministry of Science and ICT of Korea under the KSTAR project References [1] Y.S. Bae, J.H. Jeong, S.I. Park, M. Joung, J.H. Kim, S.H. Hahn, et al., ECH preionization and assisted startup in the fully superconducting KSTAR tokamak using second harmonic, Nucl. Fusion 49 (2008) 0022001. [2] M. Joung, Y. Gorelov, S. Park, J.H. Jeong, Y.S. Bae, H.L. Yang, J.H. Kim, S.H. Han, J.G. Kwak, J. Lohr, Second harmonic 110 GHz ECH-assisted start-up in KSTAR, EPJ Web of Conferences 32 (2012) 02012. [3] J. Seol, S.G. Lee, B.H. Park, H.H. Lee, L. Terzolo, K.C. Shaing, K.I. You, G.S. Yun, C.C. Kim, K.D. Lee, W.H. Ko, J.G. Kwak, W.C. Kim, Y.K. Oh, J.Y. Kim, S.S. Kim, K. Ida, Effects of Electron Cyclotron Resonance Heating induced internal kink mode on the toroidal rotation in the KSTAR Tokamak, Phys. Rev. Lett. 119 (2012) 195003. [4] Y.J. Shi, J.M. Kwon, P.H. Diamond, W.H. Ko, M.J. Choi, S.H. Ko, S.H. Hahn, D.H. Na, J.E. Leem, J.A. Lee, S.M. Yang, K.D. Lee, M. Joung, J.H. Jeong, J.W. Yoo, W.C. Lee, J.H. Lee, Y.S. Bae, S.G. Lee, S.W. Yoon, K. Ida, Y.S. Na, Intrinsic rotation reversal, non-local transport, and turbulence transition in KSTAR L-mode plasmas,
8
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Design of ECH launcher for KSTAR advanced Tokamak operation a,
a
a
a
a
T a
Mi Joung *, Minho Woo , Jongwon Han , Sonjong Wang , Sunggug Kim , Sanghee Hahn , Dongjea Leeb, Jonggu Kwaka, Robert Ellisc a b c
National Fusion Research Institute, Daejeon, Republic of Korea Ulsan National Institute of Science and Technology, Ulsan, Republic of Korea Princeton Plasma Physics Laboratory, Princeton, United States
A R T I C LE I N FO
A B S T R A C T
Keywords: KSTAR ECH Launcher Launcher control Advanced tokamak operation
For the advanced tokamak operation in Korea Superconducting Tokamak Advanced Research (KSTAR), up to 6 MW of electron cyclotron heating (ECH) is planned. The ECH launcher should be able to operate in steady state and serve as a central/off-axis heating and current drive, q-profile control, and control of the neoclassical tearing mode (NTM) and sawtooth. The water cooled launchers have been upgraded to include continuous operation with precise, high-speed real-time control of the beams by integration with the plasma control system (PCS) for supporting the experiments, in particular for suppressing NTM. In this study, the launcher was successfully operated for 78 s with about 650 kW EC power, where the temperature of the coolant, with a flow rate of 9.6 lpm, is saturated and approximately 2 °C. The launcher control system had a maximum control cycle of 5 kHz and a motor latency of about 10 ms, and the poloidal scanning speed of the beam was 20°/s. Additionally, for the higher localization of power deposition, the curvature and size of the mirrors in the launcher were optimized. This paper presents the ECH launcher design and its control system with the operation results.
1. Introduction Since the first plasma operation of Korea Superconducting Tokamak Advanced Research (KSTAR), electron cyclotron heating (ECH) has played a key role in various experimental results such as ECH pre-ionization [1], ECH-assisted startup [2], a plasma rotation study [3,4], an impurity transport study [5], a high poloidal beta operation, and a long pulse operation [6]. The main purpose of KSTAR is to explore the physics and technologies of high performance steady-state operations that are essential for International Thermonuclear Experimental Reactor (ITER) research and future fusion reactors [7]. To achieve this goal, totally 28 MW of external heating power was planned. The main heating systems are planned to provide 12 MW until year 2019 and 6 MW until year 2022 by neutral beam injection (NBI) and ECH, respectively, before the major upgrade of KSTAR to prepare long pulse and high beta discharges. All ECH systems are required to have a power capability to handle 1 MW RF power for 300 s which is the maximum pulse length of the KSTAR plasma. As the KSTAR operation capability has been improved, the launcher also has been gradually upgraded from the initial design for 500 kW, 2 s pulse toward higher power compatibilities, a longer pulse, and faster movement in real-time [8,9]. In order to achieve the
⁎
ultimate goal to withstand a continuous-wave (CW) of 1 MW, an activewater cooling system is necessary. However, this constrains flexibility and the mobility of the mirror in terms of the speed and the coverage of the mirror steering. A wide range of beam coverage beneficially widens the range of possible operation scenarios and applicable plasma parameters. Moreover, accurate movement of the highly localized beam was crucial in controlling magneto-hydrodynamic (MHD) instabilities such as the neoclassical tearing mode (NTM) [10–12] and sawtooth [13]. This paper discussed the design of the ECH launcher and controller, as well as the results of the long pulse operation in KSTAR. Section 2 describes the launcher design for the steady state and high performance operation, including the optimization of the steering range and the beam diameter. Section 3 explains the launcher control system and its operation results. The present status and future work are discussed in the Conclusion. 2. Launcher design Fig. 1 shows a layout of the 6-MW ECH system. Among the six1 MWclass 300 s gyrotrons, two 105/140-GHz dual frequency gyrotrons are under operation and two identical gyrotrons are being procured. The other two units will be 170 GHz. Each KSTAR ECH system consists of an
Corresponding author. E-mail address: [email protected] (M. Joung).
https://doi.org/10.1016/j.fusengdes.2019.111395 Received 31 May 2019; Received in revised form 25 October 2019; Accepted 1 November 2019 0920-3796/ © 2019 Elsevier B.V. All rights reserved.
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 1. Layout of the 6 MW ECH system.
injected corresponding to the requested deposition position by two kinds of steering: rotation of the entire steerable mirror assembly and rotation of the steerable mirror about the stationary axis holding the mirror in the assembly. The combination of these two rotation controls permits a 40° poloidal and ± 20° toroidal scanning range. In the previous design, all launcher components were passively cooled by the thermal radiation. However, for the steady state operation, two mirrors that are directly affected by the RF beam had been upgraded with active cooling. The temperature of the residual parts in the launcher design, including the waveguide, does not increase, prior to the melting point. This is because the effects from the RF loss and plasma are not large enough; the launcher is far from the plasma and in the shadow of the other object. The straight corrugated waveguide can be also affected by RF beam. However, in the ideal case, the theoretical loss of HE11 mode at the waveguide is minimal - less than 1% per 100 m at all of the anticipated operating frequencies. The temperature increase can be easily calculated by Ohmic loss at the isolated 2.3 m launcher waveguide via the heat capacity equation:
individual high voltage power supply, a transmission line (TL), and a launcher with a 1 MW-class 300 s gyrotron. The four 105/140 GHz gyrotrons were procured from GYCOM in Russia. The frequency of each gyrotron can be selected based on the operation scenario. The TL is composed of the evacuated corrugated waveguide components in the 63.5 mm inner diameter. This includes the waveguide switch, pump-out tee, RF gate valve, bellows, DC break, and several kinds of miter-bends, e.g., the power monitor, arc detector, and polarizer-set. All waveguide components was delivered by General Atomics (GA) in the USA. Theoretical loss of the TL with seven miter-bends in the length of approximately 65 m is less than 8% for both frequencies. All TL components were designed for the steady-state operation at 1 MW of power, with each operating frequency. KSTAR has two, nearly identical, single channel launchers. The initial launchers were designed and fabricated in collaboration with the Princeton Plasma Physics Lab. (PPPL). Its basic mechanism is almost identical to the DIII-D ECH launcher [14]. The beam angle can be changed in ± 20° poloidally and toroidally by the variation of the steerable mirror’s tilt and rotation angles. As KSTAR operation’s capability has improved, the launcher also has been gradually upgraded. It was initially designed for 500 kW, a 2 s pulse and upgraded toward higher power, longer pulses, and fast movement in real-time [8,9]. The ultimate goal of the handling power for a single launcher is to withstand a CW of 1 MW. For this purpose, the radiative cooling of the initial design is insufficient and active water cooling is required. However, this constrains flexibility and mirror mobility with respect to the speed and coverage possible in steering the mirror. Additionally, in the limited space, the beam diameter at the plasma should be minimized for the highly localized beam. Moreover, the 6 MW ECH was recently assigned to the large middle port for installation of six launchers. Pivot positions of the ECH launchers should be determined in a manner that achieves a high current drive for the advanced tokamak operation.
Q = c⋅m⋅ΔT, Eq.
(1)
where Q is the RF loss, c is the specific heat, and m is the mass of waveguide. The launcher waveguide was specially made from the Nickel coated stainless steel (SS) to increase the mechanical strength and to decrease the eddy current induced by changes to the magnetic field. Note that all other transmission line components are made from aluminum. When 1 MW power is transmitted for 100 s, the SS waveguide with c = 510 J/kg⋅K rises approximately 3.7 K. This indicates that the active cooling is not needed. However, there are some points where the temperature increases further, e.g., where the higher order mode or the gap between two waveguides appears. Therefore, additional temperature monitoring may be required. Fig. 3 shows the actively cooled fixed and steerable mirror models. The design was based on interpretations of the operation results from the passively cooled mirrors, including the preliminary analysis regarding the active water-cooling mirrors, seen in [10]. The absorbed power of the mirror heated by the RF beam varies according to the material of the mirror. Copper alloy is a good candidate for low RF power absorption and sufficient thermal conductivity. In terms of thermal stress, Copper Chromium Zirconium (CuCrZr) may be better than pure copper. However, it is not trivial to braze with the stainless steel. Oxygen-free high thermal conductivity (OFHC) copper was utilized as the reflective surface of the two mirrors. The surface of the fixed mirror is only affected by the RF beam; however, the steerable
2.1. Launcher structure and cooling performance The KSTAR ECH launcher consists of a long waveguide, a first fixed mirror and a second steerable mirror, as shown in Fig. 2. The highpower RF beam emitted from the gyrotron is transmitted to the waveguide of the launcher through the transmission line. There is no vacuum RF window in the launcher as well as in the TL. With regard to vacuum condition, the launcher and the TL are connected directly to the tokamak. The RF beam radiated from the waveguide is reflected at the first-fixed mirror and the second-steerable mirror. The RF beam is 2
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 2. (a) The installation picture and (b) the flange at the Bay-O port, and (c) the layout inside the ECH launcher.
coefficient of 1 W/cm2K. When the fixed mirror with a 0.75 mm thick copper sheet brazed to the 316 L SS block is considered, the temperature increased to 157 °C. Similarly, assuming the heat flux of 30 W/cm2 from the plasma, the temperature rise of the copper steering mirror was 30 °C higher than the fixed mirror and the surface temperature of the mirror with thin copper on the SS block was increased to 215 °C in the simulation. Based on this analysis, the fixed mirror and the steerable mirror were made from the copper block brazed with SS for the steady state operation. The mirrors expect to be stable with regard to temperature rise, however additional thermal stress analysis may be required. The thermal strength of the mirrors would be increased if the copper is replaced with CuCrZr. Until recently, a shutter was used to protect the mirrors from the plasma when the RF beam was not required. Currently, the shutter has been removed to reduce the beam diameter by increasing the mirror size. This allows an easier connection of the cooling tube to the mirrors. The cooling circuit design of the fixed mirror was chosen by considering the inlet and outlet positions in the limited space and the cooling area. For the steerable mirror, the inlet and outlet are located near the steering pivot to minimize the deformation of the bellows in the whole poloidal and toroidal steering ranges. The bellows were changed from short welded bellows to long formed bellows in order to disperse the stress and expand the steering range as shown in the Fig. 2(c). The length and the smallest bending radius of the formed bellow made of SUS316 L are 400 mm and 50 mm, respectively. The designed maximum pressure of the formed bellows is 30 kg/cm2. It was tested up to 15 bar and was typically used at 4 bar. The cross section of the cooling circuit inside the mirror is rectangular with area 7 × 4.5 mm2 and
Fig. 3. Models of (a) the fixed mirror and (b) the steerable mirror.
mirror is affected both by RF and plasma. From the Ref. 15, the temperature increase to approximately 102 °C at the copper fixed mirror was deduced by a finite element thermal analysis performed in ANSYS. Here, the peak heat flux was estimated to be 84 W/cm2 corresponding to the absorption power of 0.2% from an 800 kW RF beam and film 3
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 4. The temperature differences of the inlet and outlet coolants of the mirrors in KSTAR operation for (a) shot #21003 and (b) shot #21706.
8 × 6.5 mm2 for the fixed mirror and for the steerable mirror, respectively. Fig. 4 shows the temperature differences of the outlet and inlet of the coolants on the two mirrors when the RF beams are injected with the pulse length and the power of (a) 12.78 s and approximately 540 kW (shot #21003) and (b) 77.9 s and approximately 650 kW (shot #21706). Due to the long path and the slow flow rate, these differences have a time delay of 18 s and 40 s, for the steerable mirror and the fixed mirror, respectively. The flow rate at the fixed mirror was approximately 4 lpm and less than 9.8 lpm at the steerable mirror. The temperature difference does not precisely reflects the temperature rise, because the inlet temperature of the antenna coolant is not well regulated. The inlet temperature of these selected two results relatively well regulated. However, the gradual decrease and increase of the temperature differences at the end of the pulses can be seen, due to changes of the inlet coolant temperature. Nevertheless, the absorbed power at the mirror can be estimated by the simple correction of the data. In shot #21003, the absorbed powers of the fixed and the steerable mirror are approximately 0.4 kW and 1.1 kW, respectively. In shot #21706, these are 0.52 kW and 1.137 kW by assuming that the temperature difference is saturated for a CW operation. By adding the lost data to the fixed mirror in shot #21003, the results of two shots were similar and reasonable. The results are summarized in the Table 1. In the [15], the absorption rate of the RF power by the mirror was 0.08–0.11%. When the plasma current is 1 MA, the heat flux from the plasma was estimated to be 20–30 W/cm2, respectively. These results are in a good agreement with the estimations in the [15], considering the different plasma current.
cooling tubes caused a strong resistance and narrow rotation angles. Accordingly, the cooling circuit was changed to incorporate the long bellows as shown in the Fig. 2(c). This provided the same operation ranges as the first design. The available scanning range can be changed not only by rotation ranges of the mirror, but also by the pivot position of the launcher. Additionally, the launcher vertical position zpivot can affect the EC current drive efficiency, which is an important factor in q-profile control and MHD instability mode control. Fig. 5 shows EC driven currents as a function of zpivot calculated by the ray tracing code TORAY-GA. It was obtained by injecting 105 GHz ECH beam into the typical plasma, with a plasma current of 400 kA at the toroidal magnetic field of 1.792 T. As the zpivot is away from the mid-plane, the total EC driven current is slightly increased. However, the EC driven current is sensitive to the deposition position of ECCD, zECCD in vertical or rECCD (or ρECCD) in radial. In the case of zpivot = 0, the normalized radius ρECCD is slightly outside compared to ρECCD at the other pivot positions when the ECCD deposition position zECCD=+/- 300 mm. This leads the decrease of the driven current. Due to diffraction, the beam is not easy to approach near the plasma core. The peak current density of ECCD J_max shown in Fig. 5(b) shows that the driven current is more localized at the large zpivot. This calculation is not in perfect agreement with the previous study on NTM suppression [16]. However, the results should not be identical, as the plasma is different. Nevertheless, the qualitative results showing that the EC current drive is better at the large zpivot are the same. Considering the limitations, such as the size of the equatorial port allocated to ECH and the passive stabilizer installed in the vessel, the zpivot of ± 350 mm was decided. This position is slightly increased from the first design with zpivot = 300 mm. This increase takes into account the current drive efficiency and the mirror steering coverage. In order to install six launchers in this Bay-O port, the existing two launchers should be installed vertically and symmetrically. Installing the other four launchers in the rest space requires careful consideration. To solve this problem, a 2-channel launcher was introduced as shown in Fig. 2(b). The configuration and operation principle of the 2-channel launcher are the same as the present launcher. The accessible region by
2.2. Accessible range of the beam As described above, the beam angle is changed through the combination of the two types of rotations of the steerable mirror. In the first design without active cooling, the beam could be injected in the range of 50–90° from the vertical axis poloidally and -20 to+20° from the launcher axis toroidally. However, the cooling circuit interferes with the free motion of the mirror. In fact, the short bellows and long rigid
Table 1 The estimated absorption rates by RF power and the heat flux from the plasma at two mirrors in the KSTAR operation. Shot
Plasma current (pulse length)
RF power (pulse length)
@ fixed mirror
@ steerable mirror
Flow rate
Measured absorbed power
Calculated absorption rate
Flow rate
Measured absorbed power
Calculated heat flux from plasma
∼ 0.5 kW (added lost power) 0.52 kW
0.0926%
9.8 lpm
9.09 W/cm2
0.08%
9.6 lpm
1.1 kW (peak @ 1.7 °C) 1.137 kW (peak @ 1.826 °C)
#21003
∼ 400 kA (16.07 s)
540 kW (12.78 s)
4.0 lpm
#21706
400 kA (77.94 s)
650 kW (77.9 s)
4.1 lpm
4
9.35 W/cm2
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
increase the beam divergence: increasing the distance from the end of the waveguide to the fixed mirror and using the convex fixed mirror. The first case is similar to ITER upper launcher [19]. However, the size of the two mirrors would need to be larger than the present launcher boundary. Using the convex mirror is more effective in the small space. The beam radiated from the waveguide can be represented by a Gaussian beam with a beam waist of ω0 at the end of the waveguide. The radius of beam waist ω0 = 0.322D, where D is the inner diameter of the waveguide [20]. The beam radius ω(z) along the beam path length z is,
ω(z) = ω0 1 + (
λz 2 ) πω02
(2)
where λ is the wavelength of the RF beam. This beam diameter would change after each reflection by the curved mirrors following Gaussian Beam Optics. After reflection by the mirror, the new beam radius ω(z) can be expressed as, 2
2
⎛ ω(z) ⎞ = 1 + ⎛ z − z1 ⎞ ⎝ ω1 ⎠ ⎝ d1 ⎠ ⎜
⎟
⎜
⎟
(3)
where ω1 and z1 are the beam waist radius and the axial waist position before the mirror, and d1=πω12/λ. Finally, the beam diameter of about 60 mm at the resonance layer of Rres = 1.8 m corresponding to the beam path length of about 1240 mm is obtained (Fig. 7 and 8). The average power density of the EC beam was increased by approximately 1.7 times. However, the width of the second mirror was limited by launcher structure. It is approximately 1.7 times of the beam width. Thus, the transmission loss is unavoidable at the second mirror. 3. Launcher control system 3.1. Steering actuators of the mirror The antenna actuator consists of two sets of a motor and an encoder connected to each other in the gear assembly. For fast and accurate control, the pneumatic motor was replaced by a DC motor (Faulhaber, 3272-024CR). As the motor driver was upgraded to a higher current of 50 A, the motor controlled by pulse width modulation (PWM), with the frequency of 20 kHz, is sufficient for the wide ranges of speed and torque. The irregular resistive torque is caused by the mechanical wearing or imperfection of the driving assembly. The absolute magnetic encoder (Posital, UCD-S101B-1212) which conveys the mirror position, is connected to the motor with a 3:1 gear ratio. The absolute magnetic encoder is also connected to the leadscrew, which connects to the mechanical assembly of the mirror. The absolute magnetic encoder with a 24 bit multi-turn resolution provides a resolution of 0.002 °/ count for the mirror angle. It was enveloped by the iron cylinder to avoid a kind of chattering by a high electro-magnetic field.
Fig. 5. (a) The EC current drive and (b) the maximum current density of EC as a function of the vertical position of the launcher.
the beam is represented by the hatched area in the Fig. 6. For the lower launcher, the beam can be deposited from -350 mm to 500 mm in the vertical direction at the major radius R = 1.8 m corresponding to the beam angle from 90 to 50 degrees from the vertical axis. The upper launcher is the opposite. For the toroidal angle, the mirror can be changed from -20 to 20°, but due to the poloidal limiter installed in the vessel, the beam cannot be injected to the counter current direction. This limitation will be resolved by moving the launcher inward. Such a modification has the benefit of minimizing the beam diameter. However, the electromagnetic stress caused by plasma disruption, as well as the thermal and impurity effects from the plasma should be reanalyzed.
3.2. Launcher controller To drive the motor in the real time by Plasma Control System (PCS), the launcher controller has been developed based on a field-programmable gate array (FPGA) (Altera EP1C6Q240C8 with a 27 MHz clock). As the EC power deposition position can be easily changed with the highly localized power by the fast control of the launcher mirror, ECCD is well known as a powerful tool for the suppression and control of MHD instability modes. This is especially true for NTM, which results in a deterioration of plasma performance and occasionally plasma disruption. For NTM control, EC power should be applied to the precise mode location before the mode destroys the plasma. In addition to the fast detection of the mode growth and location, the response of the mirror is also critical in the control sequence. Moreover, during the long pulse discharge, the role of ECH can be changed, for example from off-
2.3. Beam diameter The EC beam diameter in KSTAR has been emphasized because ECH power is marginal to control MHD instabilities. However, the long distance from the last mirror to the plasma, as well as the limited mirror size, which was originally designed for 84 GHz, made it difficult to decrease the beam diameter. In the previous design of the launcher, the minimum size of the EC beam at the plasma center was 80 mm in diameter at 140 GHz [17]. This diameter is larger than the typical island size of NTM in KSTAR plasma [18]. With the limited power, such a broad EC beam is not favorable to suppress NTM. However, in the given space, in order to decrease the beam diameter, increasing the beam divergence could not be avoided. Two methods were investigated to 5
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 6. Accessible area by the hatching of the two launchers (a) in the top view and (b) in the side view.
axis current drive to central heating. The quick movement of the mirror can increase the flexibility of the operation. The block diagram of the launcher control system is shown in Fig. 9. The PCS sends the setpoints, which have been converted to the two encoder values, in real time to the FPGA through the Heating Integrated Control System (HICS). The HICS has a time synchronization function and acts as a bridge between two different controllers: the launcher controller with a serial interface and the PCS with a reflective memory (RFM). The PCS operator can choose how to define the setpoints, which are the vertical target positions at the resonance layer Rres corresponding to the given Bt. The operator can also choose a the tracking parameter, such as normalized poloidal magnetic flux psi, or normalized radius rho, to calculate the setpoints. In this case, the real-time equilibrium reconstruction data (rtEFIT) is involved. The setpoint (target position) is converted to the mirror angles by an algorithm with or without plasma in the PCS. In the case of no plasma or vacuum, mirror angles are calculated with the beam vector from the antenna pivot position to the target position. When the plasma is present, mirror angles are calculated with a real time ray tracing code (real-time
Fig. 7. The traces of the beam radius of 140 GHz and 105 GHz along the beam path where arc length ‘0′ means the end of the launcher waveguide.
Fig. 8. Beam shapes simulated by Code V at the vertical target position of (a) z = -300 mm and (b) z = 0 mm at Rres = 1.8 m when the two mirrors have toroidal curvatures. 6
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
Fig. 9. KSTAR ECH launcher control system.
TORBEAM [21] with the rtEFIT data with the assumed density and temperature profiles. These angles are calculated at one given toroidal angle during the entire pulse duration and converted to encoder values. The converted encoder values are transferred to the HICS with the cycle of 20 kHz through the RFM, and then to the FPGA in the individual launcher controller of each gyrotron with the serial communication rate of 5 kHz. The FPGA is the main element that operates the EC antenna actuator. FPGA compares the present value of the encoder at a rate of 20 kHz with the commended encoder value from PCS at a rate of 5 kHz. The rate which the FPGA drives the motor is 20 kHz. The control cycle of the launcher controller is 20 kHz; however, the command value was limited to 5 kHz by the serial communication between the HICS and the launcher controller. Fig. 10 shows the PCS control results of the two launcher mirrors with Rres = 1.728 m and perpendicular injection. The latency caused by the inertia of the motor is about 10 ms. As shown in the signals of speed 4 in EC2 launcher and speed 3 in EC3 launcher, the mirror was vibrated or overshot at the high speed. In order to decrease these phenomena, a reverse current braking with opposite voltage is applied to the motor control algorithm. Braking was optimized at speed 3 in EC2 and at speed 2 in EC3. In both cases, the beam scan speed was approximately
Fig. 11. (a) The result of vertical position control of the mirror via the qtracking algorithm in the PCS. (b) Expanded view for t = 2.3 – 3 s.
20 °/s in the beam angle and about 50 cm/s in the vertical length. The misalignment was within about ± 2 mm. The asymmetry between the upward and downward directions shown in the EC3 launcher is a result of the different load. The different load varies according to vacuum pressure, misalignment, wear-out, and so on. Fig. 11 shows the result of the mirror control for q-tracking in real time by rtEFIT in the KSTAR experiment. After starting the q-tracking algorithm at 1.5 s, the new mirror position commands are calculated from the q-value in rtEFIT with geometrical offset, wherein the vertical position z of the mirror was tracking to z = zq=2 – 5 cm along Rres = 1.92 m. Since the latency of the motor response was about 10 ms, the error, or position difference between the command and the returned value of the mirror, was almost equal to the faster variation of the command value. This fluctuating error can be seen from 2.5 s to 3.7 s and at the end of the shot from 5.5 s onward. Except for these unusual periods, the mirror was aligned well to the requested value within 2 cm. Fig. 10. The results of the vertical position scan for the (a) EC2 and (b) EC3 launcher with pre-programmed control in the PCS where speed 1, 2, and 3 mean driving motor PWM duty of approximately 25%, 37%, 51%.
4. Conclusions The design of the ECH launcher for KSTAR’s long pulse, high7
Fusion Engineering and Design 151 (2020) 111395
M. Joung, et al.
performance operation were summarized. The rising temperature of the two water-cooled mirrors was saturated, not exceeding 2 °C for a 78 s long pulse operation in the tokamak experiment. With the cooling pipes included, the steerable mirror allows a beam angle range from 0–40° in the poloidal direction and ± 20° in the toroidal direction. The vertical pivot position of the launcher was optimized to have the highest driving current. Consequently, the beam diameter was minimized up to 60 mm in diameter at the plasma center through the implementation of two curved mirrors. The vertical beam position can be controlled by the FPGA-based controller in real time with a latency of 10 ms and a speed of 20 °/s, which corresponds to the vertical scan speed of 50 cm/s. With these specifications, the launchers operated successfully in a 78 s long pulse operation. Because the poloidal limiters installed in both sides of ECH port constrains the available toroidal beam angle of ECCD, the radial pivot position of the launcher will soon be moved closer to the plasma. For the next four ECH systems, the launcher should be designed as a 2-channel launcher to save the space. However, these will have the same configuration and cooling circuits as the launcher described herein. The improved ECH launcher system is expected to contribute qprofile control and MHD instability mode control to the advanced tokamak operation.
Nucl. Fusion 57 (2017) 066040. [5] Joohwan Hong, S.S. Henderson, Kimin Kim, C.R. Seon, Inwoo Song, H.Y. Lee, Juhyeok Jang, Jae Sun Park, S.G. Lee, J.H. Lee, Seung Hun Lee, Suk-Ho Hong, Wonho Choe, Modification of argon impurity transport by electron cyclotron heating in KSTAR H-mode plasmas, Nucl. Fusion 57 (2017) 036028. [6] Y.M. Jeon, et al., Distinctive features of KSTAR stationary high poloidal beta scenario, talk, 16th International Workshop on H-Mode Physics and Transport Barriers, St. Petersburg, Russia, Sept., 2017, pp. 13–15. [7] G.S. Lee, J. Kim, S.M. Hwang, et al., The KSTAR project: an advanced steady state superconducting tokamak experiment, Nucl. Fusion 40 (2000) 575. [8] M. Joung, M.H. Woo, J.H. Jeong, S.H. Hahn, S.W. Yun, W.R. Lee, Y.S. Bae, Y.K. Oh, J.G. Kwak, H.L. Yang, Real time MHD mode control using ECCD in KSTAR: plan and requirements, AIP Conference Proceedings 506 (2014) 1580. [9] R. Ellis, Y.S. Bae, J. Hosea, M. Joung, D. Miller, W. Namkung, H. Park, Development of Steady-State Mirrors for the KSTAR ECH Launchers, Proceedings, 2013 IEEE Symposium on Fusion Engineering (2013) article No. 6635461. [10] M. Reich, K. Behler, A. Bock, L. Giannone, M. Lochbrunner, M. Maraschek, E. Poli, C. Rapson, J. Stober, W. Treutterer, ASDEX Upgrade team, ECCD Based NTM Control at ASDEX Upgrade, 39th EPS Conference &, 16th Int.Congress on Plasma Physics, 2012 P1.076. [11] E. Kolemen, A.S. Welander, R.J. La Haye, N.W. Eidietis, D.A. Humphreys, J. Lohr, V. Noraky, B.G. Penaflor, R. Prater, F. Turco, State-of-the-art neoclassical tearing mode control in DIII-D using real-time steerable electron cyclotron current drive launchers, Nucl. Fusion 54 (2014) 073020. [12] A. Isayama, G. Matsunaga, T. Kobayashi, S. Moriyama, N. Oyama, Y. Sakamoto, T. Suzuki, H. Urano, N. Hayashi, Y. Kamada, T. Ozeki, Y. Hirano, L. Urso1, H. Zohm, M. Maraschek, J. Hobirk, K. Nagasaki, JT-60 team JT60U, Neoclassical tearing mode control using electron cyclotron current drive and magnetic island evolution in JT-60U, Nucl. Fusion 49 (2009) 055006. [13] T.P. Goodman, F. Felici, O. Sauter, J.P. Graves, the TCV Team, Sawtooth pacing by real-time auxiliary power control in a tokamak plasma, Phys. Rev. Lett. 106 (2011) 245002. [14] R. Ellis, J. Hosea, J. Wilson, R. Prater, R. Callis, Design of a dual high-power, long pulse, steerable ECH launcher for DIII-D, AIP Conference Proceedings 595 (2001), p. 318. [15] R. Ellis, J. Hosea, Additive manufacturing of steady-state mirrors for the KSTAR ECH launchers, Proceedings, 2015 IEEE Symposium on Fusion Engineering (2015) article No. 7482419. [16] Y.S. Park, Y.S. Hwang, Simulation of ECCD optimization for neoclassical tearing mode suppression in KSTAR, Fusion Eng. Desig. 83 (2008) 207–210. [17] S.G. Kim, M. Joung, J.W. Han, S.J. Wang, D.S. Kim, M.S. Choi, Design of a steerable launcher for the ECH system on KSTAR, EPJ Web of Conferences 157 (2017), p. 03022. [18] Kyungjin Kim, Yong-Su Na, Minhwa Kim, Y.M. Jeon, K.D. Lee, J.G. Bak, M.J. Choi, G.S. Yun, S.G. Lee, S. Park, J.H. Jeong, L. Terzolo, D.H. Na, M.G. Yoo, KSTAR Team, Experiment and simulation of tearing mode evolution with electron cyclotron current drive in KSTAR, Curr. Appl. Phys. 15 (2015) 547–554. [19] M.A. Henderson, R. Heidinger, D. Strauss, R. Bertizzolo, A. Bruschi, R. Chavan, E. Ciattaglia, S. Cirant, A. Collazos, I. Danilov, F. Dolizy, J. Duron, D. Farina, U. Fischer, G. Gantenbein, G. Hailfinger, W. Kasparek, K. Kleefeldt, J.D. Landis, A. Meier, A. Moro, P. Platania, B. Plaum, E. Poli, G. Ramponi, G. Saibene, F. Sanchez, O. Sauter, A. Serikov, H. Shidara, C. Sozzi, P. Spaeh, V.S. Udintsev, H. Zohm, C. Zucca, Overview of the ITER EC upper launcher, Nucl. Fusion 48 (2008) 054013. [20] R.L. Abrams, Coupling losses in hollow waveguide laser resonators, IEEE J. Quantum Electron. 838 (1972) Vol. QE-8. [21] E. Poli, A.G. Peeters, G.V. Pereverzev, TORBEAM, a beam tracing code for electroncyclotron waves in tokamak plasmas, Comput. Phys. Commun. 136 (2001) 90–104.
Declaration of Competing Interest None. Acknowledgements This work was supported by National R&D Program funded by the Ministry of Science and ICT of Korea under the KSTAR project References [1] Y.S. Bae, J.H. Jeong, S.I. Park, M. Joung, J.H. Kim, S.H. Hahn, et al., ECH preionization and assisted startup in the fully superconducting KSTAR tokamak using second harmonic, Nucl. Fusion 49 (2008) 0022001. [2] M. Joung, Y. Gorelov, S. Park, J.H. Jeong, Y.S. Bae, H.L. Yang, J.H. Kim, S.H. Han, J.G. Kwak, J. Lohr, Second harmonic 110 GHz ECH-assisted start-up in KSTAR, EPJ Web of Conferences 32 (2012) 02012. [3] J. Seol, S.G. Lee, B.H. Park, H.H. Lee, L. Terzolo, K.C. Shaing, K.I. You, G.S. Yun, C.C. Kim, K.D. Lee, W.H. Ko, J.G. Kwak, W.C. Kim, Y.K. Oh, J.Y. Kim, S.S. Kim, K. Ida, Effects of Electron Cyclotron Resonance Heating induced internal kink mode on the toroidal rotation in the KSTAR Tokamak, Phys. Rev. Lett. 119 (2012) 195003. [4] Y.J. Shi, J.M. Kwon, P.H. Diamond, W.H. Ko, M.J. Choi, S.H. Ko, S.H. Hahn, D.H. Na, J.E. Leem, J.A. Lee, S.M. Yang, K.D. Lee, M. Joung, J.H. Jeong, J.W. Yoo, W.C. Lee, J.H. Lee, Y.S. Bae, S.G. Lee, S.W. Yoon, K. Ida, Y.S. Na, Intrinsic rotation reversal, non-local transport, and turbulence transition in KSTAR L-mode plasmas,
8