HIGH POWER HELICON ANTENNA DESIGN FOR DIII-D. R.C. O NEILL General Atomics San Diego, California, USA

Transcription

1 HIGH POWER HELICON ANTENNA DESIGN FOR DIII-D R.C. O NEILL General Atomics San Diego, California, USA oneill@fusion.gat.com M.W. BROOKMAN, J.S. DEGRASSIE, B. FISHLER, H. GRUNLOH, M. LESHER, C.P. MOELLER, C. MURPHY, R. PINSKER, M. SMILEY, J. F. TOOKER, H. TORREBLANCA General Atomics San Diego, California, USA A. NAGY Princeton Plasma Physics Laboratory Princeton, NJ, USA Abstract A new current drive system is being designed and fabricated for the DIII-D tokamak to drive current in high beta discharges, using electromagnetic helicon waves. The High Power Helicon Antenna (HPHA) is intended to couple up to 1 MW of power into DIII-D plasmas at a frequency of 476 MHz. The antenna assembly consists of 30 individual, inductively coupled antenna modules fastened to six internally water-cooled inconel back-plates which are mounted to the DIII-D vacuum vessel wall. The end modules of the antenna array have special design features to provide RF stripline feed attachment points. This high power traveling wave antenna array is expected to have higher off-axis current drive efficiency than 80 kev neutral beam injection or outside-launch electron cyclotron current drive. A 100 W, 12-module antenna was developed and tested in DIII-D, and demonstrated sufficient coupling at low power. A prototype antenna module was also fabricated and a high Q of 1267 was obtained, which allows for a predicted coupling of up to 80% into the plasma for the antenna assembly. The HPHA assembly is designed to minimize and survive disruption forces, thermal loading from RF losses and plasma heat loads, and vacuum vessel bake (350C). Each of the antenna modules is protected from plasma interaction using molybdenum Faraday shield rods, with the plasma facing surfaces of the Faraday shield rods coated with boron carbide to enhance thermal resistance and minimize high-z impurity generation. A single folded stripline provides a 180 phase shift between pairs of end module connections for driving the four poloidal straps. This single mono RF coaxial stripline design optimizes the allowed in-vessel space, ease of installation, structural rigidity and RF tuning ability. Multiphysics finite element analyses are performed to optimize the geometric shapes of the various components of the antenna assembly. A description of the mechanical design and analyses of the HPHA and stripline assembly is presented. 1. INTRODUCTION: PHYSICS MOTIVATION Design [1] studies have shown that steady-state tokamak reactors can benefit greatly from efficient non-inductive current drive in the mid-radius region, ~ 0.5, where is the normalized minor radius. The so-called helicon wave has been identified as a candidate to efficiently drive plasma current in this region [2]. The helicon is a fast magnetosonic wave, often called a whistler wave [3], and its frequency is in the range of frequencies occupied by lower hybrid waves, referred to as the LHRF. In this regime the helicon wave interacts with electrons of high parallel velocity to drive plasma current, so that trapping in the tokamak magnetic field variation can be less for these electrons, than for those that resonate with electron cyclotron waves. In reactor-relevant plasma scenarios, the damping of the helicon wave is weak enough in the outer radial regions to allow good penetration, yet sufficient to be absorbed and drive current at the desired off-axis location. Current drive (CD) using the helicon is lacking an experimental basis to date because in previous experiments [4] the single-pass electron damping was too weak, the electron pressure being too low. However, previously current drive with fast waves in the medium to high ion cyclotron range of frequencies, below the LHRF, was demonstrated in DIII-D [5] at frequencies up to 120 MHz. The present helicon effort will extend this CD investigation up to the LHRF regime, specifically 476 MHz, obtaining strong total first-pass damping by utilizing high electron pressure target plasmas. A crucial aspect of this regime is the challenge of an appropriate wave-launching structure. To obtain electron Landau damping, waves with an index of refraction parallel to the static magnetic field greater than unity (n > 1) must be launched. Such waves are radially evanescent in the vacuum region just in front of an antenna. The radial decay rate at a given n increases with frequency, so the coupling per unit area of antenna is relatively weak in the LHRF. A necessary condition for a suitable antenna is that sufficient coupling to the plasma be achieved at the helicon frequency. The wave launcher must supply the necessary power to the plasma to obtain the desired CD at a voltage on the antenna below breakdown limits. 1

2 An innovative traveling wave antenna (TWA) of the comb-line variety [6] has been selected for the DIII-D helicon launcher. The efficacy of the TWA to launch fast waves has been previously demonstrated in the JFT- 2M tokamak in Japan [7]. Subsequently, a very low-power helicon 12-module TWA (~0.4 kw) was designed and installed in DIII-D. Operation of the low-power antenna assembly at 476 MHz demonstrated that plasma conditions could be achieved in front of the antenna assembly. The desired fast helicon wave was launched and traveled toroidally along the line of modules via mutual inductive coupling between the modules and resulted in RF power being coupled into the plasma [8]. With the reactive coupling from module to module, input and return power connections were required only on each end of the antenna array. As with the low-power helicon antenna, the High Power Helicon Antenna (HPHA) is comprised of a series of resonant modules of the same geometry as the low-power antenna modules [9]. A conceptual design of the HPHA is being installed in DIII-D is shown in Figure 1. The HPHA system overview is presented along with the design and analysis of the antenna assembly. Fig. 1A. In-vessel View of High-Power Helicon Antenna Assembly with Protective Graphite Tiles Located in DIII-D Vessel 2.SYSTEM OVERVIEW: Figure 1B. Major Subassemblies of the High-Power Helicon Antenna Mounted on the Vacuum Vessel Wall and in Ports 2.1 Antenna Assembly The antenna assembly consists of thirty antenna modules, six water cooled back plates mounted from the vacuum vessel wall, two RF stripline assemblies and two cm diameter coaxial 50 vacuum RF feedthroughs (Fig. 1B). The antenna assembly is located poloidally above the tokamak midplane. This location was selected to achieve the maximum current drive and to minimize the interference with the vessel ports. To minimize the thermal load on the antenna array, the array is recessed approximately 4 cm from the flux surface defined by the DIII-D bumper limiter tile surfaces and 3 mm behind the protective tile surface surrounding the antenna array. The antenna array length is 1.7 m, necessitating faceting of each back plate in which the modules are mounted so that the array of modules follows the contour of the flux surface. Each stripline assembly toroidal length is 0.5 m. The RF power is injected from either feed to enable driving current in either toroidal direction Modules and Back Plates The antenna consists of twenty eight center-modules (Fig. 2) and two end-modules (Fig. 3) for a total of thirty modules. Each module is 5 cm wide (toroidal direction) by 8 cm deep (major radial direction) by 21 cm tall (poloidal direction) and consists of two pedestals, two capacitive straps, four capacitor plates, seventeen Faraday Shield (FS) rods, two end plates, two side walls, sixteen RF tuning rods, and one bottom plate (Fig. 2). All modules have two internal straps (capacitor straps) mounted on two pedestals Each pedestal has geometrical features to accept RF probes and thermocouples. The end-modules have two extra pairs of smaller straps connected to the pedestals that operate as 15 Ohm impedance RF feeds into or out of the end modules (Fig. 3). These four feeds connect to the 15 inner conductor of the stripline assembly via four ports: P1, P2, P3, P4. Each pair of the four ports (P1-P2 & P3-P4) are intended to deliver equal amounts of RF current to the antenna module with the current 180 out-of-phase in relation to the adjacent port. All modules are made of copper-chromium-zirconium (CuCrZr) for electrical conductivity and mechanical strength with the exception of the FS rods and fasteners. The sixteen tuning rods provide tuning capability of each antenna strap to operate at the desired frequency of 476 MHz once all the modules are mounted onto the 2

3 Fig. 2. Exploded View of Antenna Module Assembly Showing Various Components Fig. 3. End-module with the Sidewall Removed Showing Internal Components backplates. The seventeen FS rods are made of titanium- zirconium-molybdenum (TZM) to withstand the high radiative heat loads of the plasma. The plasma-facing surfaces of the Faraday shield rods are coated with boron carbide (~ 25 m thick) to enhance thermal resistance and minimize high-z impurity generation. The thirty modules are grouped in six sets of five modules that are mounted to six inconel 625 water-cooled faceted backplates, each 30 cm wide by 3 cm deep by 21 cm tall (Fig. 4). The back plates are segmented to minimize thermal expansion stresses during bake-out of the vacuum vessel. A pedestal block and bracket for each plate provides the mechanical and thermal connection between the back-plate and the vessel wall. The back-plate, pedestal block and bracket are made of Inconel to reduce the plasma disruption-induced currents. Each module is tilted toroidally by 15 to align the FS rods with the static magnetic field lines of the nominal DIII-D target plasma, and are tilted poloidally by 19 match the plasma shape. Fig. 4. Backplate Assembly with End Module & Two Internal Modules Stripline Assembly The high and low angle folded striplines consist of a two-layer inner stripline conductor and a single outer conductor housing, and are electrically connected to the 150 and coaxial feedthroughs via a 25 coax (Figures 5 and 6). The design of the stripline was determined by the available ports of the vacuum vessel, allowable volume adjacent to the vacuum vessel wall and the required rf performance. Each stripline is supported at the feedthrough, at the four ports end module ports (P1, P2, P3 & P4), and at four quarterwavelength stubs connected to the outer conductor. The four quarter-wavelength stubs allow mechanical support of the inner stripline conductor to resist the plasma disruption loads, eddy current forces and vacuum vessel bake out loads as well as providing heat transfer paths from the strip line to the water cooled vessel wall. The copper-plated 316 stainless steel outer conductors of the stripline assemblies are secured with brackets that straddled the port openings, with mounting blocks at the quarter-wavelength stubs, and at the area adjacent to the end modules with angle brackets. The quarter-wavelength stub lengths are adjustable for tuning, and are clamped to the outer housing stubs via a sliding clamp mechanism so each stub can be tuned independently of one another. There are two stubs per side of the layered stripline one inboard, one outboard for a total of four stubs per folded stripline. The stripline layer impedance is reduced in three steps from the stripline input to the end modules, 50 to 25 to 15 Ohms. To 3

4 Fig. 5. Stripline Assembly Inner and Outer Conductors Fig. 6: Stripline Assembly Details accommodate the different electrical length of each layer of the stripline to the four ports of the end module and to insure a 180-degree phase shift between the four ports of the end module, a sawtooth pattern, i.e. notches, have been incorporated into the split of each leg of the stripline. This allows adjustment of the electrical length from the stripline input to the 4 ports. The upper portion of the striplines consist of solid CuCrZr while the lower 15 sections are made of copper-plated solid Inconel 625 to reduce disruption forces. The quarter-wavelength stubs are made of CuCrZr providing the necessary heat transfer mechanism from the various stripline areas, through the outer conductor walls to the water-cooled vacuum vessel wall. Both analytic and COMSOL analyses indicate most of the RF heating occurs in the higher impedance CuCrZr regions of the stripline, with averages ranging between ~4-7 W/cm 2, while the lower 15 region of the inner stripline only yielded ~0.5 W/cm 2 of heating RF Transmission Line System A block diagram of the RF transmission line is shown in Figure 7 with an overview of the entire helicon system. The klystron output transitions from a WR2100 rectangular waveguide to a 23 cm diameter, 50 coaxial transmission line ~80 meters long. The single 23 cm diameter, 50 coax terminates at a centralized switching network located on a mezzanine situated outside the DIII-D radiation shield wall. The switching network divides the single transmission line into two 15 cm diameter 25 coaxial transmission line antenna feeds and returns of ~ 20 m long. The switches allow for remote switching of the antenna feed and return between experiments to determine the bi-directional antenna/plasma-coupling efficiency. The RF power that flows through the antenna and exits the opposite end is routed to a water-cooled dummy load #1 located outside the torus hall. The transmission line always terminates into a dummy load #1 regardless of the direction of RF power flow through the antenna. The dummy load #1 acts as an end-of-line dump for non-coupled power from the antenna. The two 15 cm, 25 coaxial lines from the switching network to the vacuum vessel feedthroughs contain electrical isolation DC breaks for personnel protection and bellows to allow for the thermal expansion and contraction of the DIII-D vessel RF Power Source Fig. 7. RF Transmission Line Schematic The 1.2 MW 476 MHz RF power source for the antenna consists of a Stanford Linear Accelerator Center (SLAC) klystron system previously used for the Stanford accelerator (Figure 8). The klystron system consists of the klystron, a dummy load, high voltage snubber, circulator, high voltage power supply and transformer. The 2 MW high voltage transformer, also supplied by SLAC, provides the 80 kv to feed the high voltage power supply. A high voltage snubber unit serves to protect the klystron from high voltage faults originating in the power supply and the stored energy in the high voltage cables connecting the power supply to the klystron. The circulator 4

5 directs the generated rf power from the klystron into WR2100 waveguide and coax transmission line, and directs reflected rf power from the antenna and transmission line into a rf absorbing waveguide dummy load #2 to protect the klystron from damage. 2.2 Design Analyses COMSOL Multiphysics was used for the optimization of the RF design of the antenna assembly and striplines while ANSYS was utilized in the thermal and mechanical analyses of the antenna assembly. Based on specialized Fig. 8. SLAC Klystron System physics codes [10], the antenna was designed to resonate at 476 MHz which coincides with maximum power level of the SLAC klystron with a bandwidth of 3 MHz. The antenna assembly and striplines were designed to operate at 1 MW for 5 seconds and to withstand the RF and plasma heat loads, as well as the DIII-D disruption and baking loads Antenna Resonance Frequency Each of the thirty modules were designed to inductively couple with the adjacent modules. With the excitation of a single antenna module, two resonant frequencies are generated (Figure 9). The lower resonant frequency induces current to flow in the same poloidal direction on the straps (desired), whereas the higher frequency induces current to flow in opposite direction (Figure 10). With two or more modules adjacent to each other, the resonant frequencies are shifted, an increase of approximately 8 MHz for the lower resonant frequency and a decrease of approximately 33 MHz for the upper frequency maximizing the total induced current in the straps at 476 MHz. Subsequent modules cause smaller frequency shifts of ~2 MHz or less. With coupling of the power from the one module to the next, the analysis indicates that the RF power coupling efficiency is 95 % between the modules. With a calculated Ohmic loss of 1% or 10 kw/mw, the design is to couple 40 kw/mw to the plasma from each module, as the traveling-wave power decays along the modules. Fig. 9. Computed single module resonant frequencies. Fig. 10. Induced currents for two center-modules End Antenna Module RF Losses A COMSOL RF loss profile of an end module was generated (Figure 11) since the end module will incur the highest power input. The loss distribution analysis shown in Figure 11 was for a 1% RF loss of the 1 MW input power (10 kw) with a plasma load of 50 W/cm 2. To allow for some safety margin in the design of the module, a conservative loss assumption of 2.5% of the 1 MW input power coupled with 10-second operation (12-minute repetition rate) was assumed. A preliminary basic thermal analysis of the end module showed large amounts of thermal energy on the module pedestal/strap/capacitor plates due to the RF loss and the plasma heat load. The RF losses contributed approximately 30% of the total power deposited on the module. Further detailed analyses indicated that the antenna module reaches a peak temperature of 460 o C after the 10 second pulse, well above the recommended service temperature for the CuCrZr material. Since RF loss is temperature dependent, it is crucial that the maximum temperature of the antenna module be kept as low as feasible. For CuCrZr material, the RF loss at 350 o C is 1.4 times larger than at room temperature and for the TZM Faraday shield rods, the RF losses are 1.6 times higher than at room temperature. The RF analyses with the assumed 10 second operation led to unreasonably high module component temperatures and RF losses. Of particular interest was the CuCrZr finger region of the module that attaches to the TZM rods, which ratcheted to a temperature of 475 o C after 4 cycles. 5

6 Fig. 11. End-module RF loss distribution for 1 MW of input power. With the manufacturing and assembly of the antenna modules, the CuCrZr material is work hardened, and overaged and annealed during brazing resulting in an overall lower yield strength of 200 MPa versus a typical 310 MPa for virgin CuCrZr material at room temperature. The yield strength of the material is further compromised with operations above 400 o C. Given the RF loss analyses, plasma loading and decreased material strength, an administrative operation limit of 5 second operation (12 minute repetition rate) of the antenna system is being implemented to insure the antenna modules RF heating are controlled, and the material structural strength properties are maintained Stripline RF Losses Fig. 12. Low-angle stripline RF loss map for 1 MW of input power. Fig. 13. Low angle stripline temperature rise after 4 operational cycles A portion of the 1 MW RF power that drives the antenna is dissipated in the stripline. Analytic calculations and COMSOL analysis indicate the majority of the RF heating occurs in the higher impedance (50 and 25 ) regions of the striplines and with the ends of the ¼ wavelength support stubs with approximately power loss of 5 W/cm 2 and 7 W/cm 2 respectively (Figure 12). The upper 50 and 25 regions of the stripline and the ¼ wavelength support stubs are made of CuCrZr for thermal conductivity and to allow for heat dissipation into adjacent components of the stripline and the vacuum vessel wall. The lower 15 stripline parts are made of copper-plated Inconel 625 which lowers the electrical conductivity of the overall assembly and ultimately will reduce the plasma disruptioninduced current in the overall assembly and the disruption forces torques imparted into the assembly. Though the Inconel parts have lower electrical conductivity and a lower thermal coefficient, the parts do have a higher heat capacity and allow for the surface RF current with the copper plating. With the 5 W/cm 2 loading on the 50 impedance section, the resulting ratcheting temperature rise of this section is approximately 110 o C from 30 o C ambient after 7 operation cycles (Figure 13) Though the ¼ wavelength support stubs experience the highest RF loss load, the ends of the stubs are conductively cooled via CuCrZr blocks that are attached to the water cooled vacuum vessel wall (Figure 14). The temperature rise of the ¼ wavelength support stubs is less than 20 o C due to the efficient cooling Thermal Analysis of the End Antenna Module and Stripline 6

7 A finite-element (FW) thermal analysis of the stripline inner conductor and the end module was performed using SolidWorks Simulation with temperature-dependent RF loads and a plasma radiative loading of up to 50 W/cm 2. This analysis of this stripline and end module together was performed since it was determined that the highest RF energy loss occurs at the feed locations of the stripline to the end module and the end module could experience the additional plasma loading due to the fast ions generated from the adjacent 210 o DIII-D neutral beam injector. Increases in temperature of the stripline inner conductor and the end module early in plasma discharges greatly magnified the RF losses due to the temperature rise to ~350 C which increases the RF loss by ~40% in the CuCrZr components and ~60% in the TZM Faraday Shield Rods. With the administrative operation limit of 5 second operation (12-minute repetition rate) of the Fig. 14. ¼ Wavelength Support Studs with CuCrZr Conducting Blocks antenna system, the FE analysis indicates the CuCrZr material will reach a temperature of 340 C and the TZM Faraday Shield Rods will reach a temperature of 407 C on the TMZ rods (Figure 15). With these temperatures, minimum RF losses are tolerable and the structural integrity of the CuCrZr material is maintained. After each plasma shot the stripline inner conductor and the antenna end module return to temperatures less than 50 C (Figure 16). Fig. 15. Stripline/End Module Temperatures After 5 Second RF Pulse Fig. 16. End Module Temperatures After 12 minute cool-down Disruption Induced Forces & Torques During disruption events in DIII-D, large currents may be induced into the antenna and stripline assemblies. The change of the poloidal magnetic flux will induce a current on the vessel wall and antenna components that will interact with the background toroidal magnetic field producing forces and torques on the antenna. A 2 MA DIII-D plasma current collapses within a few milliseconds, resulting in large torques and forces as the magnetic fields collapse. Several analyses have been performed to insure the antenna array and stripline assemblies survive such Fig. 17. Disruption induced current density distribution on the antenna 5.6 msec after a disruption event with a current collapse rate of 1 MA/msec. DIII-D disruption events. Currents are induced in the vacuum vessel wall, and in conducting components and assemblies attached to it. The intensity of these induced currents depends on the geometry, orientation to the toroidal and poloidal fields and the component or assembly material properties. For the antenna assembly, the largest current density occurs in the CuCrZr antenna module side walls, the TZM Faraday shield rods, and the Inconel back-plate assembly (Figure 17). These materials have electrical conductivities (S/m) of 4.65x10 7, 1.8x10 7 7

8 and 7.57x10 5 respectively, showing direct proportionality to the amount of induced current. Since the vessel wall is made of Inconel, its current contribution to the antenna is minimal. In the development of the design of the stripline inner conductor assembly, various analyses were performed to determine if the inner conductor assemblies could be fabricated entirely of Inconel to minimize the disruption forces and torques on the assembly, or be a hybrid of Inconel and CuCrZr materials to allow for efficient thermal transfer of energy from the RF losses in the assembly. An analysis of the forces and torques on the two design approaches was performed and the toroidal forces (F y) and poloidal forces (F z) were, on average, on the order of 200 N and 1000 N respectively, whereas the radial forces, on average, were ~ 3000 N. A comparison of the forces and torques on the stripline inner conductors, backplate assembly and the five antenna modules for the two designs was performed (Figure 18). The difference in forces and torques for the modules is very small; however, for the inner stripline the forces increased by approximately 2x and torques by 30x. With better thermal capabilities and reduced forces and torques, the hybrid material stripline inner conductor assembly was determined to be the optimal design. 3. Summary RADIAL FORCES: FX (N) SL (Inconel) + Antenna SL (CuCrZr/Inconel) + Antenna PEDESTAL BACKPLATE M1 M2 M3 M4 M5 SLIN (N) SLOUT (N) RADIAL TORQUE: TX (N-M) SL (Inconel) + Antenna SL (CuCrZr/Inconel) + Antenna PEDESTALBACKPLATE M1 M2 M3 M4 M5 SLIN (N) SLOUT (N) Figure 18. Maximum radial force (top) and radial torque (bottom) for an array of five modules and stripline inner conductor assembly A high power helicon antenna has been designed and is being installed at DIII-D with operation scheduled to begin in the autumn of The antenna coupling 1 MW of RF power at 476 MHz to the plasma is expected to drive current off-axis in high magnetic discharges with greater efficiencies than other current drive systems presently available on DIII-D based on theoretical predictions and previous experimental results. Engineering analyses were performed to corroborate the antenna s performance and its survivability in the DIII-D tokamak environment. ACKNOWLEDGEMENTS This material was based upon work supported by the U.S. Department of Energy, Office of Science, Office of Fusion Energy Sciences, using the DIII-D National Fusion Facility, a DOE Office of Science user facility, under Award No. DE-FC02-04ER DISCLAIMER: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. REFERENCES [1] Jardin, S.C. et al 1997 Fusion Eng. Des [2] Prater, R. et al 2014 Nucl. Fusion [3] Pinsker, R.I Phys. Plasmas [4] Pinsker, R.I AIP Conf. Proc [5] Petty, C.C. et al 2001 Plasma Phys. Control. Fusion [6] Moeller, C.P US Patent 5,289,509 [7] Ogawa, T. et al 2001 Nucl. Fusion [8] Pinsker, R.I. et al 2018 Nucl. Fusion [9] H. Torreblanca, et al., SOFT 2018 (2018). [10] R. Prater, et al., Nucl. Fusion 54, (2014). 8