Stability of cable-in-conduit superconductors for Large Helical Device

The stability of cable-in-conduit superconductors has been experimentally investigated as part of a poloidal field coil program for the Large Helical Device (LHD) project. A new conductor was designed and fabricated, focusing on the stability. As a result of a zero-dimensional stability analysis, it was found that the conductor had a high stability, 5*10/sup 5/ J/m/sup 3/, at the design condition of 20.8 kA and 6.5 T. Current transfer performance after partial quenching has been investigated by using a short sample of the conductor for the poloidal field coil. The effects of the current transfer among the strands on the conductor stability are discussed.<<ETX>>


I. INTRODUCIION
Large Helical Device (LHD) is a heliotron/torsatron type fusion experimental device and its construction is progressing as a 7 year project which began in 1990. In 1991, we started the construction of one of the poloidal field coils, named Inner Vertical Coil (IV-coil) [l]. From the research and development programs, we decided to design the conductor focused on reliability and stability. In this paper, we deal with stability of the conductor used for the IV-coil.
Stability of cable-in-conduit conductors has been theoretically studied by using zero-and/or one-dimensional models [2,3]. Here we include the effects of normal propagation and current transfer in the transverse direction of a conductor. In the excitation tests of the demonstration coil (TOKI-PF), the conductor was found to be unexpectedly unstable [4]. In this coil, the strands were insulated with formvar of 11 vm thickness in order to reduce the coupling losses. We confirm that there are some problems concerning the insulation. First, the formvar insulation reduces the heat transfer coefficient effectively. Second, rapid commutation of current may lower the stability. The quench of a multi-strand cable may originate from the normal transition of some portions of the strands. Vysotsky et al. pointed out that the current transfer from a strand with the normal zone to the adjacent strand occurred rapidly when using insulation or high resistive matrices, and the quench current of the adjacent strand could not reqch the DC critical one [5].
In the experiment presented here, the current transfer in the transverse direction was studied using a short sample of Manuscript received August 24,1992 the conductor for the IV-coil. A partial normal zone was generated by a resistive heater instead of an inductive heater and the current distribution was monitored by pick-up coils.

CONDUCTOR DESIGN
Main parameters of the new conductor are listed jn calculated one, which confirms no damage of the strands. The critical current was extrapolated to be 62 kA at 6.5 T which is these parameters in consideration o f t Figure 2 shows the calculated margin, In dimensional model proposed by L Minervini [6] was applied. Magnetic fi the operating current, bp. The results in decreases rapidly when the operating current excee The limiting current was, therefore, estimated to approximately 20 kA. The margin, however, keeps rather hi value, 5x 105 Jln?, at the designed point.  corresponds to the recovery current which is observed for poo4 boiling type conductor. Figure 6 shows the time evolution of voltages and field changes, AB, at 18.2 kA (L,p/Ie0.49) where the normal zone shrank. When the voltage generation was initiated, the field at the positions of PC07, P C l l and PC24 decreased, since the field produced by the initial transported current has a positive sign. These pickup coils were located near the strands attached to the heater (see Fig. 4). So, the field decrease indicates that the current in the strands with the normal zone was transferred to adjacent strands which were far from these pickup coils. Though the heater input continued for 1 sec, the voltage vanished in about 0.2 sec. This can be explained as follows: the current with the normal zone dropped down to nearly zero and then the voltage vanished., Although the current of the adjacent strands increased, they were still superconducting. The data also show that AB saturated after the vanish of the voltage, which indicates that the distribution of the transport current has been changed due to the generation of the partial normal zone. Figure 7 shows the time evolution where the normal zone propagated at 21.3 kA (10$I,y0.57). The change of AB seems to be composed of two phenomena. AB for the initial 0.08 sec (period I in the figure) was almost similar to the previous data at 18.2 kA. AB a t P C l l decreased and then saturated. This means that only the strands near the heater 513 quenched in the period I. In the subsequent period (I1 in the figure), AB at P C l l and PC13 decreased and the others increased along with the increase of the voltage signals. The distribution of AB indicates that the current w the transverse direction.     Fig. 8.

Time [sec]
We calculated AB by using a simplified model as shown in Fig. 8 in order to confirm the current transfer process. In this model, the current in the region 1 with the width of AX drops down to zero and this current is added to the region 2 of 1 mm width. The width of 1 mm simulates a row of strands. The current density in the region 3 is assumed to be constant. The calculated results are shown in Fig. 9. AB at P C l l decreases and AB at PCO9 and PC15 increase with increasing AX. AB at PC13 decreases a s AX>2 mm. These tendencies were consistent with the experimental results. In period I, AX was evaluated to be 2-3 mm. In Fig. 7, all AB's approached zero again from 0.45 sec, which indicates that the current density became uniform since all the strands turned into normal states.
It should be noted that the stability boundary appeared at bp/Ic0.5. At 21.3 kA, the normal propagation started after the commutation. The reason may be that the current in the adjacent strands increased and finally exceeded the critical current. The strands over the critical current must have quenched in a long region if the magnetic field is constant. If considering a two-strand cable, it is clear that the stability boundary is Iop/Ic=0.5. In the experiment, the phenomenon similar to the two-strand system may occurred. Vysotsky et al. reported that quench current of the adjacent strand cannot reach the critical one if the duration of commutation is rapid, the order of 1 msec [5]. The duration was, however, about 100 msec in our sample. The joint resistance between strands may be effectively reduced because of no insulation of strands.
The experimental results suggest that the critical current should be more than twice as large as the operating one in regard to a cable-in-conduit conductor. The IV-coil has the critical current of three times. Therefore, we expect high stability in the operating condition.

V. CONCLUSIONS
Stability of the conductor for LHD poloidal coils was experimentally investigated concentrating on partial quenching using a resistive heater installed in the conduit. The summary of the results is shown below.
(1) Propagation of normal zone was observed when the operating/critical current ratio exceeded 0.5. In the case of bp/Icc0.5, the current in the quenching strand dropped down to zero and the adjacent strand could carry the superconducting current. In the case of Iop/Tc>0.5, the current in the adjacent strand seemed to exceed the critical one and then the transverse and longitudinal propagation of the normal zone progressed.
(2) The experimental results showed t h a t i t t o o k approximately 100 msec to commutate the current fully from the quenching strand to the adjacent one. This duration was much larger than Vysotsky's observations, which may be related to no insulation of strands.
(3) In the actual design of IV-coil, the operating/critical current ratiois set to be 0.33. High stability margin is, therefore, expected in regard to the partial quenching.