New winding techniques for superconducting dipole magnets

A single-phase power transformer with windings made of T/sub c/ superconducting tapes was fabricated and tested. The tapes are made of Bi-2223 imbedded in a silver alloy matrix. The primary and secondary windings are made of double pancake coil. The rated voltages of the windings are 220/110 V. Four pancake type coil windings are stacked on the core. The inner two act as the primary winding. Fundamental characteristics are obtained through short circuit and no load tests. The overload capability and characteristics are investigated.

We made a preliminary investigation of the applicability of high-T/sub c/ superconducting parallel conductors to pancake coils. For the sake of a uniform current distribution and low AC loss, the constituent strands need to be transposed so as to be inductively equivalent with each other. We adopted an interdisk transposition where the strands are not transposed inside a single-pancake coil but only at the joint between the pancake coils. The fabrication process is simple. We have only to fabricate the same double-pancake coils and connect the strands individually with transposition outside the winding. We searched theoretically for the optimum transposition in the case of 3-strand and verified the theoretical result by using small test coils wound with NbTi 3-strand parallel conductors for convenience.

Recently various high temperature superconducting (HTS) conductors have been commercialized in the power applications. While a BSCCO becomes tape or wire, others coated conductors (CC) are tapes. An HTS magnet utilized with these conductors can be fabricated as a form of pancake winding or layer winding. In the case of HTS magnet using a thin and rectangular tape, the pancake winding method is satisfied in the practical cases because of its suitable features. However, one of the disadvantages of this winding method is increasing brands and splices. From this point of view, we propose a layer winding method using an HTS tape and fabricated HTS layer coils. To verify the feasibility of the proposed winding method, critical current and over-current tests are investigated in liquid nitrogen and the results are compared with those of pancake coils. In addition, operating characteristics on the condition of various winding tensions are analyzed.

Morse and Rubenstein did not investigate the induced currents when they published a solution of the problem of the diffraction of waves by ribbons and slits. The authors of the present paper regretted this omission because the diffraction pattern of the scattered wave is directly dependent upon these induced currents and it is very easy to predict the pattern from the distribution of induced current, but not vice versa. Moreover, the distribution of induced current near a bounding edge is a matter of great interest in itself and is likely to serve as a guide in other problems which have not yet been solved. The first two Sections of the paper are devoted to extending the analysis developed by Morse and Rubenstein in terms of Mathieu functions, so as to make possible the calculation of the induced currents. Curves are given which show the distribution of these currents across ribbons whose width is 2λ/π and 4λ/π. The necessary analytical development was very difficult and cumbersome and the evaluation of the results was very troublesome and exacting: the senior author wishes to disclaim any share in this portion of the work, beyond bringing that part of the problem to the attention of the junior author.The senior author had previously obtained an expression for the currents induced in a half-plane and had derived the curve of this distribution. It differed little from the curves derived from the Mathieu function solution for a plane of finite width, thus showing that the disturbance at one edge is very little dependent on the existence of a second parallel edge, provided it is more distant than about ½λ It became clear that the disturbance due to an edge may be considered to be substantially contained within a width of, say, ⅛λ.If this conclusion is accepted as general there is no further necessity to use Mathieu functions in this problem: they are extremely involved and poorly tabulated, and it is advantageous to dismiss them after they have served a useful turn.The third Section of the paper is devoted to predicting the diffraction pattern from the currents induced in the ribbon and to discussing the effect which their distribution will have on the shape of the pattern. It is found that the patterns calculated by Morse and Rubenstein are not compatible with the current distributions derived from the extension of the Morse and Rubenstein analysis. The discrepancy must be due to the inclusion of insufficient terms of an infinite series in the process of numerical evaluation. In evaluating the curves of current distribution given in the paper, the junior author has included as many terms as the published tables of Mathieu functions will permit and is confident that they are adequate. But since there was no direct method of deciding whether the higher degree of accuracy of evaluation was to be credited to the Morse-Rubenstein patterns or to the current distributions obtained by the junior author, the two authors were led to consider, very closely and generally, the form of current distribution which must obtain to satisfy the essential condition of the problem. This condition is that the current distribution must be such as to produce an electric force which is constant in magnitude and phase all across the ribbon, thereby equilibrating the field of the wave incident normally on it.In Section 7 it is shown, by very simple methods, which are in no way dependent upon either the Mathieu solution or the half-plane solution, that the surface field can be constant over the major portion of the width of a wide ribbon only if the distribution consists of a uniform co-phased density together with a second and quadrature distribution which is virtually concentrated in a filament located at each edge, the strength of each filament being equal to the current in a strip of width 0.16λ of the uniform density, and thus independent of the width of the ribbon. It is shown, moreover, that if the uniform co-phased density in fact becomes not uniform very near the edges there must be regions, near the edges, over which this density is greater in the one region and less in the other region than the mean density over the width, in such a way that the mean density is unchanged by the lack of uniformity near the edges. But all these properties had already been inferred and derived from the distribution evaluated from the Mathieu functions and for the half-plane. These evaluated distri butions have thus been found to possess the characteristics which are now found to be inherently necessary. In fact this apparently very difficult problem turns out to be a simple one if only it is attacked by examining that distribution which must obtain to produce a uniform force across the ribbon at its surface, in contrast to deriving the currents which must be induced by a given incident field.Once this fact had been recognized, it became apparent that the problem of a very narrow ribbon could now be evaluated, though this is not practicable by the Mathieu process because the limiting values of these functions have not yet been tabulated for small arguments. Section 8 gives the solution for a narrow ribbon, valid within the range such that we may write Y0(z)≃2/π(log z+y).When the ribbon is wide it is obvious that the echoed field must be proportional to width. A curve is given in which the echoed field per unit width is plotted as a function of width, expressed as a multiple of a wavelength: it is substantially constant provided the width of the ribbon exceeds about 0.2λ It is of great interest to find that a result which is obvious for a wide ribbon is still substantially correct even when the width is only ⅕λ; this result is believed to be novel.It is hoped that the paper throws considerable light on the mechanism of the induction of current in a ribbon; it certainly shows the reason why the phase of this current must necessarily change across the width: an effect which was to be expected from general experience of comparable problems, where it had but emerged in the final result of the analysis without disclosing the physical necessity which causes it to exist.Section 10 discusses the case when the direction of propagation of the wave is parallel to the plane of the ribbon. It includes curves showing the distribution of induced currents.An Appendix (Section 12) contains details of the solution for the half-plane, and several formulae are worked out which are required for various computations used in the main part of the paper.

The dimensioning of Winding Packs and allocation of materials in DEMO PF coils is proposed based on the requirements defined by the EUROfusion 2018 DEMO baseline. The coils are pancake-wound and use two types of forced flow cable-in-conduit conductors. The conductor which uses NbTi has a high void fraction but has no separated low-impedance cooling channel. On the other hand, the conductor made using Nb3Sn has a separated low-impedance cooling channel and therefore low void fraction in the remaining cable space. Both the conductors have a rectangular cross-section. The design addresses the dimensioning of the Winding Pack (WP) based on the electromagnetic field calculations and stress analysis. The amount of superconductor, copper, and cable void fraction in the conductor are determined by the coil current, magnetic field and assumed temperature margin. The amount of structural material in the conductor jacket depends mainly on the fatigue stress corresponding to the pulsed operation of these coils. For PF1 and PF6 coils, the comparison of NbTi and Nb3Sn design options suggest a more efficient allocation of structural and superconducting material in the latter case. The WP dimensions do not significantly depend on the winding technique (pancake vs layer winding).

An eight-pancake Nb-Ti dipole magnet, with bent up ends, called D-108B has been built and tested. This magnet is a Nb-Ti version of a Nb/sub 3/Sn magnet designed to produce a 10-tesla dipole field in a 40 mm diameter aperture. The pancack design is used for the heavy 12,000 ampere Nb/sub 3/Sn cable because of the mechanical difficulty in winding such a heavy cable into the conventional nested cylindrical shell configuration with a 2'' inner winding diameter. The Nb-Ti version operates at 1.8K, in He II, has superconducting cable half as thick as the Nb/sub 3/Sn cable, and operates at half the operating current: 6000 A rather than 12,000 A at 10 tesla. Both magnets are approximately one meter long. D-10B was tested from January 26 to February 2, 1983 and reached short-sample performance in both He I and He II after moderate training. The central field at 4.3K is 7.0 (+- 0.1) tesla, and at 1.8K is 9.1 (+- 0.2) tesla. Ramp rate sensitivity and cyclic heating data were also measured.

Skin depth is an important parameter that quantifies depth of inspection of eddy current (EC) nondestructive testing. The formula which states that the skin depth is inversely proportional to the square root of the product of frequency, permeability, and conductivity has been widely used for calculating skin depth for many decades. It has been noticed that the true skin depth is less than that obtained by the formula when the EC is excited by the ac current in a pancake coil. In-depth and convincible explanation of this discrepancy has not been published. This paper studies this problem by finite-element analysis and concludes that the difference is due to two reasons. One is that the EC associated with pancake coil diffuses with increasing depth, whereas there is no diffusion effect of EC associated with uniform plane field which is the assumption of the conventional formula. The other reason is that the ECs induced by wire segments symmetric about the coil axis cancel each other and the cancellation becomes more severe with increasing depth due to the diffusion of EC. This paper also investigates the influence of each of coil parameters on skin depth. Then, a new formula for estimating skin depth is proposed.

Techniques for high field generation have been advancing. No-insulation (NI) rare-earth barium copper oxide (REBCO) pancake coils enable high field generations with high thermal stability. One of the major challenges in REBCO magnet development is the screening current reduction. The screening current is induced in the REBCO tape by the change of the field perpendicular to the REBCO tape surface. It deteriorates the field, sometimes together with mechanical stability. Since the behaviors of the screening current are clear by simulations, it is important to consider the screening current effect in the design stage to reduce it or control the coil performances. In this paper, we explore the effectiveness of the non-energized coil (hereafter called “dummy coil”) to modify the screening current distribution. The dummy coils are wound employing the NI winding technique. The electromagnetic behaviors of the insert NI REBCO magnet that consists of 12 stacked single pancake coils with a few dummy coils are investigated through the electromagnetic simulation with the finite element method (FEM). The results show that the dummy coils can suppress the variance of the current density distribution, although there is a trade-off with respect to field reduction.

A set of small inner bore (8.5 mm) single pancake coils was wound from multifilament Bi(2223)/Ag tapes using the `wind and react' (W&R) technique. W&R means sintering heat treatment of the Bi(2223)/Ag tape wound to a pancake coil. During the winding procedure, the tape was insulated by a mixture of Al2O3 or ZrO2 powders with varnish. After the first sintering step (841 °C/50 h), the printed insulation was easily removed and the tape with a clean surface was subjected to intermediate deformation by rolling. After the final heat treatment (841 °C/150 h), the coils were impregnated by epoxy resin. Electrical measurements of individual pancakes certified the good quality of interturn insulation. Some variations of transport currents among the pancakes were measured. The selected coils were connected in series or parallel and current distribution and I-V curves of individual coils were measured. Comparison of both connections is discussed in detail. The results of a small magnet arranged by connection of five single pancake coils with partial compensation of the radial field component is also presented. The compensation of the radial field component allows us to increase the magnetic field in the bore axis by 215%. Further lowering of temperature to 65 K allows us to generate a magnetic field nearly five times higher than in the case of the non-compensated winding at 77 K.

The current distribution, the magnetic field and the AC loss are numerically studiedfor pancake coils with up to 200 turns, taking into account the real thickness ofthe conductor and the magnetic interaction between the turns. All calculationsassume the critical state model with a constant critical current density. The resultsshow a slab-like current distribution, with the absence of magnetization currentsdue to the radial magnetic field. Consequently, longitudinal striation, with orwithout twisting, cannot reduce the self-field AC loss of this system. Finally,an analytical formula has been deduced for the AC loss at current amplitude:Im = Ic,where Ic is the critical current, for large mean coil radius and any number of turns.

Based on a 2-D magnetohydrodynamic model, high-current vacuum arcs (HCVAs) under axial magnetic fields (AMFs) with consideration of current density distribution at the cathode are simulated and analyzed. The current density distribution at the cathode used in the simulation is obtained by processing the image of cathode spots from experiments. In order to study the influence of current density distribution at the cathode on vacuum arc characteristics, HCVA is simulated under two kinds of current density distribution at the cathode. Then, for HCVA simulation in a half-sinusoidal period, the current density distribution at the cathode and the AMF delay are both taken into account. The simulation results show that the nonuniform current density at the cathode makes the values of the arc plasma parameters in the interelectrode region larger and their radial distribution more nonuniform than the uniform current density at the cathode. The more uniform current density at the cathode and the stronger AMF at the moments in the second 1/4 cycle make the values of the arc plasma parameters in the central region smaller and their radial distribution broader than those in the first 1/4 cycle. In addition, the simulation results will also be compared with the experimental results.

The current distribution in a HTS coil winding which uses a wide superconducting tape reflects the history of the current sourcing, the local inhomogeneity and especially the local distribution of the radial component of the magnetic field inside the winding. We measured the radial component of the magnetic field in a rectangular frame around the winding cross-section using a thin Hall probe. From the mapped data of radial component, the current distribution in the cross-section of the winding was determined using an inverse calculation relating the known mapped data and unknown distribution of the current's circumferential component in the coil. The Tikhonov method of regularization was applied in the calculation. The exact current distribution in the winding cross-section was approximated by a dense mesh of discrete points. The results are expressed for different coil currents during a gradual ramping up and down.

No-insulation(NI) Rare-Earth Barium Copper Oxide (REBCO) pancake coils can generate high magnetic fields and have the high thermal stability against normal-state transition due to allowing currents to radially bypass a local normal-state region through the turn-to-turn contacts. Despite these merits, no turn-to-turn insulation causes a charging delay. To solve this charging delay problem, an NI REBCO coil wound with multi-bundled (MB) conductors was proposed. As an experimental result, a charging delay was improved because the inductance per tape of MB coil is smaller than that of single-tape NI coil. Meanwhile, the current behaviors of MB coils are complicated and has not been clarified. We have investigated the current and thermal behaviors of MB REBCO coils during charging and against normal-state transition, comparing with and without turn-to-turn insulation. As the result, it is shown that the MB coil can be operated stably, and no pessimistic temperature rise is observed. However, the operating current of MB coils with turn-to-turn insulation may concentrate on one tape when the normal-state transition occurs. Therefore, MB coils without turn-to-turn insulation seems to be desirable.

The single pancake HTS coil (inner diameter (ID) 85 mm, outer diameter (OD) 96.5 mm, 35 turns) was investigated under axial pressure up to 12 tons in liquid nitrogen to test the influence of axial pressure on critical properties in a central plane of designed HTS double pancake–based 3 T solenoid magnet for magnetic refrigeration machine. It was shown that 15-μm thick Al2O3 oxidized coating of the aluminum disk (which was used as a heat-conducting structure) is stable (in terms of breakdown voltage and mechanical strength) with such a loading. Critical current of pancake coil does not change with axial compression up to 30 MPa pressure and starts to decrease irreversibly if P > 30 MPa. Designed axial pressure level in 3 T magnet does not exceed this permissible boundary.

The screening current–induced magnetic field (SCIF) brings bad effects on the field quality of REBCO high-temperature superconducting (HTS) magnet, which is mainly due to its specific large aspect ratio structure. In this paper, numerical simulation and experiments were conducted to investigate SCIF in a non-insulated HTS DP coil firstly. Then, a heating flux method was proposed to help eliminate the effect of screening current in a GdBCO HTS double pancake (DP) coil. A heating plate was mounted under the DP coil and used to generate a heating flux, which could increase the temperature and decrease critical current of the HTS coil in a short time. The decreasing of critical current could reduce the margin left for the screening current and eliminate the effect of SCIF further more. Heating flux experiments were performed to verify its effects on eliminating SCIF in the DP coil at the end. The results show that the SCIF could be reduced obviously by the method of heating flux. The research provides an important reference to eliminate SCIF in the HTS magnet.