Quench Analyses of the MIT 1.3-GHz LTS/HTS NMR Magnet

Abstract

The MIT 1.3-GHz LTS/HTS NMR magnet is currently under development. The unique features of this magnet include a 3-nested formation for an 800-MHz REBCO insert (H800) and the no-insulation (NI) winding technique for H800 coils. Because when it is driven to the normal state, an NI REBCO magnet will respond electromagnetically, thermally, and mechanically that may result in permanent magnet damage, analysis of a quenching magnet is a key aspect of HTS magnet protection.

We have developed a partial element equivalent circuit method coupled to a thermal and stress finite element method to analyze electromagnetic and mechanical responses of a nested-coil REBCO magnet each a stack of NI pancake coils. Using this method, quench simulations of the MIT 1.3-GHz LTS (L500)/HTS (H800) NMR magnet (1.3G), we have evaluated currents, strains, and torques of H800 Coil 1 to Coil 3 and L500, and center fields of 1.3G, L500, and H800. Our analyses show H800 is vulnerable to mechanical damage.

I. Introduction

The 1.3-GHz LTS/HTS NMR magnet is currently being developed at the Francis Bitter Magnet Laboratory, Plasma Science and Fusion Center, MIT [1],[2]. An 800-MHz REBCO insert (H800), consisting of 3 nested coils, is wound with the no-insulation (NI) winding technique [3],[4]. The reference [3] showed that the NI winding technique made NI coils self-protecting against overcurrent, made the magnet compact and mechanically robust. The NI winding technique has a potential to generate a high magnetic field. However, the electromagnetic and mechanical responses of an NI magnet composed of nested NI coils are complex [5]–[7], especially when one of the NI coils is driven to the normal state. For example, a large hoop stress and an unbalanced torque will occur [8], mechanically damaging the magnet. An excessive strain due to a large hoop stress leads to peeling of the REBCO layer from Hastelloy substrate, which in turn degrades the critical current [9]. We believe that quench analysis of a nested NI coils during fault mode is an important component of magnet protection.

We have developed a quench analysis based on an equivalent circuit method coupled with thermal and stress finite element method. We have investigated electromagnetic and mechanical responses of the MIT 1.3-GHz LTS/HTS NMR magnet when one of H800 3 nested coils, Coils 1, 2, 3, quenches. Note that each H800 coil is a stack of NI double-pancake (DP) coils. Here, by quench we mean one entire H800 coil is driven to the normal state—in reality, a quench is initiated only in one DP coil. In this paper, we present simulation results of maximum azimuthal currents, conductor strains, and torques of H800 Coil 1 to Coil 3, and center fields of 1.3G, L500, and H800.

II. Simulation Method and Model

A. MIT 1.3-GHz LTS/HTS Magnet

The MIT 1.3-GHz LTS/HT NMR magnet (1.3G) consist of a 500-MHz LTS (L500) and H800. Three H800 coils, Coil 1 to Coil 3, are composed, respectively, of 26, 32, and 38 double-pancake (DP) coils, thus a total of 96 DP coils and 192 single-pancake (SP) coils.

Tables I and II list the specifications of H800 and L500 magnets, respectively.

TABLE I.

Specifications of H800 Magnet

	Coil 1	Coil 2	Coil 3
Total 0023 DP coils	26	32	38
# turns/pancake	185	121	95
Winding i.d. [mm]	91.0	150.7	196.9
Winding o.d. [mm]	119.1	169.2	211.3
Winding overall height [mm]	323.7	392.1	465.7
SS overband radial build [mm]	7	5	3
Inductance [H]	2.43	3.08	3.71
Total inductance [H]		20.8
Operating current [A]		251.3
Cooling		Liquid Helium

TABLE II.

Specifications of L500 Magnet

Inner diameter [mm]	241
Overall diameter [mm]	780
Overall height [mm]	1422
Inductance [H]	152

The width and thickness of REBCO tapes are 6 mm and 76 µm, respectively. It is supposed that the critical current of REBCO tape is 150 A at 77 K, self-field. The lift factor depending on temperature and magnetic field is given by SuperPower Inc.

B. Equivalent Circuit Model

In the equivalent circuit, each SP coil is represented by a parallel circuit of a characteristic resistance and an inductance with the resistance of REBCO layer and stabilizer, as shown in Fig. 1 [10]. The H800 magnet is represented by 192 element parallel circuits. The L500 is represented by an inductance of 152 H, paralleled by a persistent current switch (PCS) of an open resistance of 6 Ω. The L500 and the H800, in series, are connected to a DC power supply. Finally, the equivalent circuit used in our analysis is shown in Fig. 2, in which the shunt resistance is 5 mΩ.

Fig. 1.

Element equivalent circuit of single pancake coil. It consists of a coil inductance L, the REBCO layer resistance R_sc, the matrix resistance R_mt, and the turn-to-turn contact resistance R_ct.

Fig. 2.

Equivalent circuit of 1.3-GHz LTS/HTS magnet system containing power supply. The H800 magnet is represented by 192 element equivalent circuits. The L500 magnet has a persistent current switch (PCS). The variables R_st, R_sc, R_mt, I_ct, and I_L differ in each SP coil.

III. Quench Simulation Results

We performed quench simulations to investigate the electromagnetic phenomenon after quench of some SP coils. The equivalent circuit analysis is coupled with thermal and stress analyses. In the stress analysis, the overband effect is included. The detail of the coupling analysis is described in [11].

A. Coil 1 Quench

1). Top SP Coil Quench

In this quench analysis in which a quench is initiated at the top SP coil in Coil 1, we first investigate the electromagnetic and stress behaviors of H800 Coil 1 to Coil 3. In the quench simulation, it was assumed that the top SP coil (Coil 1–1; the 2^nd number designates the SP coil counted from the top of Coil) quenched at t = 1 s, and operating current started to decrease with at a rate of 0.006 A/s.

Figs. 3a and 3b show maximum azimuthal currents vs. time plots: respectively, Coil 1 to Coil 3; L500; and top Coil 1 6 SP coils. From Fig. 3a, we note that Coil 1 maximum azimuthal current increases quickly to 420 A, while those in Coils 2 and 3 slightly increase, as L500 current remains constant. In Fig. 3b, Coil 1–1 azimuthal current quickly decreases from 251.3 to 0 when Coil 1–1 quenches. At the same time, Coil 1–2 current rises to 410 A, induced by a large drop in Coil 1–1 current. Note that currents are induced in SP coils neighboring to Coil 1–1.

Fig. 3.

Simulation results in case of top SP coil quench in Coil 1; (a) maximum azimuthal current, (b) azimuthal current in Coil1, (c) center field, and (d) maximum strain.

Fig. 3c shows the center fields vs. time plots: total (1.3G); H800; and L500. During quench, 1.3G center field remains at 30.7 T as the 1.3G keeps its total flux constant by inducing extra currents within H800. Fig. 3d presents the maximum strains vs. time plots in Coil 1 to Coil 3. An instant after quench, Coil 1 strain reaches 0.7%, large enough to peel the REBCO layer from Hastelloy substrate. Maximum strains in Coils 2 and 3 remains essentially at a safe level of 0.4%.

2). Entire Coil 1 Quench

Although the entire Coil 1 quenching at the same time is highly unlikely, we simulate when all 52 SP coils in Coil 1 quench. However, the entire Coil 1 becoming normal in a short period of time (< 1 s) is not impossible.

Fig. 4 shows simulation results, all vs. time. Fig. 4a maximum azimuthal currents: L500; Coil 1 to Coil 3; Fig. 4b center fields of: 1.3G, L500, H800; Fig. 4c maximum conductor strains: Coil 1 to Coil 3; and Fig. 4d maximum torques in clockwise and counterclockwise directions: Coil 1 to Coil 3. From Fig. 4a, we observe that the largest maximum azimuthal current is induced in Coil 2, while the center field generated by H800 drops and as does 1.3G field (Fig. 4b). From Fig. 4c we note that conductor strain in Coil 2 is large enough for peeling. After Coil 1quench, torque is generated in Coil 1 to Coil 3. Torque directions are opposite on neighboring coils (Fig. 5), and the differences are approximately 10 Nm (Fig. 4d).

Fig. 4.

Simulation results in case of entire Coil 1 quench; (a) maximum azimuthal current, (b) center field, (c) maximum strain, and (d) maximum torque.

Fig. 5.

Torque on each SP coil in Coil 1 at t = 2 s. Counterclockwise direction is positive in graph.

B. Coil 2 Quench

We also simulated the entire Coil 2 quenching. Fig. 6 shows simulation results, all vs. time; 6a, 6b, 6c, and 6d are respectively maximum azimuthal currents, center fields, maximum strains, and torques.

Fig. 6.

Simulation results in case of entire Coil 2 quench; (a) maximum azimuthal current, (b) center field, (c) maximum strain, and (d) torque.

With a Coil 2 quench, the Coils 1 and 3 currents increase almost instantly, and then Coil 1 current decays, while Coil 3 current remains essentially constant. Immediately after the quench, the center field slowly decreases with Coil 1 current. Increased strains in Coils 1 and 3 result in REBCO layer peeling that in turn degrades their critical currents. As for torque, Coil 1 to Coil 3 are all under unbalanced torques.

C. Coil 3 Quench

Here, we present simulation results with Coil 3 quenching at t = 1 s, all vs. time. Fig. 7a, 7b, 7c, and 7d are respectively maximum azimuthal currents, center fields, maximum strains, and torques.

Fig. 7.

Simulation results in case of entire Coil 3 quench; (a) maximum azimuthal current, (b) center field, (c) maximum strain, and (d) torque.

Induced currents are large in Coil 2 and small in L500. Although the total magnetic field remains almost constant, L500 field slightly increases within a few seconds. Coils 1 and 2 strains are large, and so are torques in Coils 2 and 3.

Only when Coil 3 quenches, L500 current increases because L500, shunted by a PCS, is closest to Coil 3.

D. Discussion

The most important result of our analyses is that in every case maximum conductor strain exceeds 0.5%, a maximum safe level. Large strains peel off the REBCO layer from Hastelloy, degrading critical current [12]. However, this critical-current-degradation phenomenon is not included in our analysis.

Unbalanced torques as well as large hoop stress are challenging issues in H800. Although the NI winding technique makes H800 self-protecting against overheating, our results show that it makes H800 vulnerable to mechanical damage.

IV. Conclusion

We investigated H800 in the MIT 1.3-GHz LTS/HTS NMR magnet for its mechanical vulnerability to a quench. Irrespective of which coils among 96 DP coils (192 SP coils) in H800, a pancake coil quench induces a large current in the adjacent pancake coils. A large induced current generates a large hoop stress and strain that peels the REBCO layer from the Hastelloy substrate that in turn degrades critical current. Our analyses presented here clearly demonstrate that of two H800 protection issues, thermal and mechanical, mechanical is more challenging.