4LH05
1
Design of a HTS Solenoid for a Gyrotron Magnet Upgrade. S. Sanz, P. Abramian, J. Calero, A. Fernández, L. García-Tabarés, J. L. Gutierrez, J. Lucas, E. Rodríguez, I. Rodríguez, F. Toral, and C. Vázquez
Abstract—The design of a HTS cryogen-free solenoid for a gyrotron magnet upgrade is presented. This gyrotron is used to set up and warm the plasma in an experimental fusion device at Ciemat. The solenoid will be wound with BISCCO-2223 tapes. A two stage Gifford-McMahon cryocooler will be used to cool down the magnet. A significant operation cost will be saved as the present solenoid is a Nb-Ti magnet cooled down in a helium bath supplied with mobile tanks. The main requirement of the magnet is to reproduce the original magnetic profile in the axis of the solenoid with a peak of 2 T in a 150 mm diameter warm bore. The magnet consists of several stacked double-pancake coils. The outer radii, the axial positions and the number of windings have been optimized by means of a genetic algorithm. Afterwards, the cryostat design is also described, including the heat leakage calculation. Finally, the winding techniques and the results of the tests of a prototype coil are also reported in this paper.
at 20 K. It is a cryogen-free device, cooled down by a GiffordMcMahon cryocooler. II.
BASIC CONFIGURATION
The magnet is made of double pancakes (see Fig. 1). Aluminium disks insulated with 25 µm Kapton foils are inserted between the coils to stabilize the magnet from the thermal point of view while providing the specified field profile. Copper disks have been discarded because of the lower enthalpy of aluminium, that allows to reduce the cooling time, one of the main issues of a cryogen-free facility.
Index Terms—Bismuth compounds, Cryogenics, Genetic algorithms, Gyrotrons, High-temperature superconductors.
I. INTRODUCTION
T
HIS paper deals with the design of a solenoid for a gyrotron magnet upgrade. A gyrotron [1] is a microwave source commonly used to start-up and heat the plasma in experimental fusion devices. It consists of an electron gun, a resonant cavity and a collector for the electrons. The magnetic field profile along the solenoid axis yields the electron synchrotron frequency, which is matched to the resonance of the cavity. The system under study is a GYCOM gyrotron used in the TJ-II stellarator at Ciemat [2]. A Nb-Ti solenoid produces a uniform magnetic field of 2 T in a 150 mm warm bore. The magnet is cooled down at 4.2 K by continuous supply of expensive liquid helium and nitrogen in mobile tanks. The new magnet is made of BISSCO-2223 tapes operating Manuscript received August 29,2006. S. Sanz is with CIEMAT (phone: +34-913466690; fax: +34-913466068; email:
[email protected]). P. Abramian, A. Fernández, J. L. Gutierrez, E. Rodríguez, I. Rodríguez, F. Toral and C. Vázquez are also with CIEMAT, 28040 Madrid, Spain (e-mail:
[email protected],
[email protected],
[email protected],
[email protected],
[email protected],
[email protected],
[email protected]). J. Calero and L. García-Tabarés are with CEDEX, 28037 Madrid, Spain (
[email protected],
[email protected]). J. Lucas is with Elytt Energy, 28046 Madrid, Spain (
[email protected]).
Fig. 1. Solenoid configuration as a stack of double pancake coils (light color) alternating with aluminium spacers (dark). The first, last and medium spacers have the drills for the copper braid thermal connection.
The main advantage of this arrangement is that the solenoid is made of several independent coils. Therefore, it is possible to test them before the final assembly. Coil splicing is possible after assembly, because both current leads are in the outer diameter of the magnet. III. MAGNETIC FIELD PROFILE OPTIMIZATION One of the main requirements is to reproduce accurately the magnetic field profile of the original magnet along the solenoid axis, because it will be assembled in the same cavity.
4LH05 region.
Magnetic field shape 2
Magnetic field (T)
2
1.5
IV. CRYOSTAT DESIGN
1
0.5
0 0
200 400 600 800 Distance alog the magnet axis (mm)
1000
The refrigeration of the solenoid is achieved by conduction, with a so-called cold finger. The coldest tip of the cryocooler is linked to the magnet by means of copper braids. A thin indium foil is placed between the copper braid terminal and the cold finger, in order to reduce as much as possible the thermal contact resistance.
Fig. 2. Magnetic field profile required along the axis of the solenoid (continuous line) and computed field for the new magnet (crosses). -4
Absolute error (T)
4
x 10
Error in the magnetic field with the proposed solution
2 0 -2 -4 -6 0
200 400 600 800 1000 Distance along the magnet axis (mm)
Fig. 3. Magnetic field error computed in each specified point, for one of the possible configurations. The magnet consists of 16 coils in this case, but a similar field quality can be obtained with less coils.
The magnet configuration allows to obtain easily this profile by modifying the spacer thickness and the outer radius of the coils. However, it is much more convenient to use the thicknesses as the only optimization parameter, because two equal diameter coils can interchange their position if they have different performance. The magnetic field depends on the coil position, and it yields the critical current requirement for that coil. There are a large number of optimization parameters, and also a great number of solutions producing the same magnetic field profile (see Fig. 2 and 3). A classical algorithm, based on the local behaviour of the objective function through its derivatives, tends to fall into the nearest local minimum. A better approach is the so-called genetic algorithm. This heuristic method searches the minimum by imitating the species evolution mechanism. Each trial solution is represented by its genome. New solutions are produced by mechanisms analogous to crossover and mutation. Differential evolution [3], a variant of this procedure has been used. The main difference with a basic genetic algorithm is that the genome is based on continuous variables, contrary to the basic genetic algorithm, where the genome is a binary fixed length number. The solution depends on the required field quality and the critical current of the tape at the working temperature. It has an important impact on the total tape length and therefore on the magnet cost. A maximum error of 0.1% is specified, but it is under study to relax this requirement out of the cavity
Fig. 4. Cross section of the magnet in the cryostat.
The selected Leybold 7/25 cryocooler has 25 W of cooling power in the first stage and 7 W in the second stage. A two stages cryocooler was chosen to intercept most of the heat leakage at an intermediate temperature, because it is thermodynamically much more efficient. The cryostat thermal design is mainly based on the reduction of the different heat transfer modes (see Fig. 4 and 5). The suspension system consists of thin wall, high thermal resistivity glass fiber composite tubes to reduce the heat leakage through the supports. On the other hand, a vacuum insulation solution has been adopted to decrease the heat transfer by convection. Finally, an intermediate radiation shield thermally anchored to the first stage of the cryocooler is used to reduce the radiated heat. To improve the efficiency of the shield, MLI (Multi Layer Insulation) pre-made blankets supplied by the company Jehier will be wrapped around. They consist of 30 perforated layers of mylar film aluminized on both sides separated by an insulating net. The inner surfaces of the vacuum chamber and the radiation shield are mechanically polished in order to reduce the emissivity. Special care has been taken in the design of the current
4LH05 leads. A resistive solution has been adopted with a thermal interception at the intermediate temperature. An HTS current lead would be interesting for higher currents. V. WINDING TECHNIQUES Initially, some small coils have been wound to develop the best winding techniques for HTS double pancake coils. A winding machine specially suited to double pancake coils has been used. It has two independent reels each containing half of the wire length each, that can counter rotate around the coil mandrel, which can be moved up and down to avoid any torsion on the brittle HTS tape. A 0.2 mm thick G-11 glass fiber sheet is placed between layers, as electrical insulator. The coils are wet impregnated with epoxy resin Araldit 2011, suitable for cryogenic temperatures. Vacuum impregnation has been discarded because it is more expensive and complicate, and no mechanical degradation has been detected for the moment. Several materials have been studied for the mould fabrication: PVC, POM, aluminium, carbon steel and stainless steel. Thermoplastic moulds are not stiff enough. POM does not need a release agent, but it has a very high thermal
3 damaged, even curing at a relatively low temperature as 80ºC. Aluminium is metallic and can damage the tapes, as it is difficult to be Teflon-sprayed. Finally, carbon steel and stainless steel are stiff and can be protected by a Teflon spray. A slot for the layer jump was machined in the first mandrels, but the tape was damaged during the mould releasing. As the position of the tape in the layer jump is not important for the field quality, the best solution is to allow the tape to follow its natural bending and then clamp it with adhesive Kapton tape. The winding process starts with the cleaning of the tape with alcohol. The spools are Teflon-sprayed to release smoothly the tape. Once the lower layer is wound, the resin is spread on the upper surface. Afterwards, a G11 sheet is placed, and the upper layer is wound. Finally, the upper surface is painted with resin, the mould is closed, and cured in a oven. The final choice was, however, grinded carbon steel plates (see Fig. 6), because of the deformation of the stainless steel ones during machining. The flatness of the coils is very important, because the final assembly is made by piling up many elements. This discards Teflon-spraying because the tolerances obtained by the grinding will be lost.
Fig. 5 Estimation of the heat leakage through the cryostat. Several different configurations have been studied, including stainless steel and G-11 glass fibre supports. For comparison, the same configurations have been studied without radiation shield.
expansion coefficient compared to the tapes. The coils can be
4LH05 TABLE I LARGE SIZE COILS MAIN CHARACTERISTICS Coil
SB13
SB14
Tape manufacturer Mean critical current value, 77K, s.f. (A) Mean n-value Thickness (mm) Width (mm) Inner diameter (mm) Turns in upper layer Turns in lower layer Turns in layer jump Coil critical current 77K, s.f (A)
EHTS 85.6 29 0.28 4.11 185 200 200 0.5 36.65
EHTS 85.2 29 0.28 4.11 185 205 205 0.5 36.3
Coil critical current 40 K, s.f. (A)
126 A
-
4 the splice between coil and current leads was 65 K. Figure 9 shows the measurements on the lower layer of the coil, the one wound first during its manufacturing.
Fig. 8. Cryostat insert with the final size coil SB-13 ready for testing.
Fig. 6. Mould configuration for HTS double pancake coils manufacturing. The surfaces are grinded to get tight tolerances in the final assembly.
No degradation due to the winding process was found in the electrical tests carried out in a liquid nitrogen bath at different temperatures. VI. PROTOTYPE COIL TESTS Two final size coils were wound and tested in liquid nitrogen (see Fig. 7 and Table I). The measured critical current degradation is due to the magnetic field. A simplified 7 parameter model is used to fit the critical current dependence on magnetic field and temperature [4], [5].
Fig. 9. Measurements on SB13 coil in the small cryostat. The critical current corresponds to 1.52•104 µV.
VII. CONCLUSIONS The design of a HTS solenoid and its cryostat for a gyrotron magnet upgrade application is presented. The manufacturing techniques have been improved by the fabrication of smaller coils. Finally, two large size prototype coils have been successfully fabricated and tested at different temperatures, both in a liquid nitrogen bath and in a specially-developed cryostat with a cryocooler.
Fig. 7. Final size coil SB-13 being tested in liquid nitrogen.
A smaller cryostat has been designed and fabricated to test the prototype coils with the cryocooler (see Fig. 8). The main difference with the final one is that it has stainless steel supports instead of glass fiber ones, and no warm bore is present. The cryostat size allows to test only two coils. The measurements at 77 K shows a critical current quite similar to the one obtained with the model previously mentioned (36.7 A). However, the critical current measured at 40 K is different than the prediction from the same model (155 A). The likely cause is the poor refrigeration of the cryostat current leads. In fact, during the tests, the temperature close to
REFERENCES [1] [2] [3] [4] [5]
V.Flyagin, V.Gaponov, M.Petelin. “The gyrotron”. Microwave theory and techniques (1977). Vol. 25. 514 – 521 [Online] Available: http://www-fusion.ciemat.es/New_fusion/en/TJII/presentacion.shtml Differential Evolution (DE) for Continuous Function Optimization (an algorithm by Kenneth Price and Rianer Storn) [Online]. Available: http://www.icsi.berkeley.edu/~storn/code.html V.Hussennether, M.Leghissa, H.W.Neumüller. “Current Capacity of Ag/Bi-2223 Wires for Rotating Electric Machinery”, Presented at 7th EUCAS 2005, Vienna S.Sanz et al. “Characterization of HTS Coils for a Gyrotron Update Application” Presented at International Cryogenic Engineering Conference ICEC 21, Prague, 2006
