Field of the Invention

The present invention relates to high temperature superconductor field coils. In particular, the invention relates to partially insulated coils.

Background

The challenge of producing fusion power is hugely complex. Many alternative devices apart from tokamaks have been proposed, though none have yet produced any results comparable with the best tokamaks currently operating such as JET.

World fusion research has entered a new phase after the beginning of the construction of ITER, the largest and most expensive (c15bn Euros) tokamak ever built. The successful route to a commercial fusion reactor demands long pulse, stable operation combined with the high efficiency required to make electricity production economic. These three conditions are especially difficult to achieve simultaneously, and the planned programme will require many years of experimental research on ITER and other fusion facilities, as well as theoretical and technological research. It is widely anticipated that a commercial fusion reactor developed through this route will not be built before 2050.

To obtain the fusion reactions required for economic power generation (i.e. much more power out than power in), the conventional tokamak has to be huge (as exemplified by ITER) so that the energy confinement time (which is roughly proportional to plasma volume) can be large enough so that the plasma can be hot enough for thermal fusion to occur.

WO 2013/030554 describes an alternative approach, involving the use of a compact spherical tokamak for use as a neutron source or energy source. The low aspect ratio plasma shape in a spherical tokamak improves the particle confinement time and allows net power generation in a much smaller machine. However, a small diameter central column is a necessity, which presents challenges for design of the plasma confinement magnet. High temperature superconductor (HTS) field coils are a promising technology for such magnets. Summary

The invention is set out in the appended claims.

Brief Description of the Drawinqs

Figure 1 is a schematic diagram of HTS tape;

Figure 2 is a schematic diagram of a wound FITS coil;

Figure 3 is a schematic diagram of a sectional FITS coil;

Figure 4 shows graphs of a simulation of the performance of a TF coil;

Figures 5 and 6 show graphs of simulations of the performance of respective exemplary TF coils;

Figures 7A and B show an exemplary partially insulating layer;

Figures 8A to E show a further exemplary partially insulating layer.

Detailed Description

In insulated and partially insulted FITS magnets, it is desirable to avoid the propagation of a local hotspot into a local thermal runaway, which can damage the coil in which it occurs due to differential thermal strain. This is normally achieved by an“active quench detection” method, which involves monitoring the coils and detecting the incipient hotspot using suitable means (eg: temperature, voltage taps, inductive pickup or strain) and then actively causing the whole magnet to quench (eg: by use of electrical heaters on the coils), thus dumping the energy stored in the magnetic field into the coil’s mass, or into a separate dump resistor outside the magnet cryostat. This results in a lower overall coil temperature and less differential strain than a localized thermal runaway, hence reducing or eliminating coil damage.

In a magnet comprising partially insulating coils it has been observed that a very rapid global quench of the entire coil can be achieved simply by turning off the power supply. The turn-to-turn resistance acts as a dump resistor embedded in the coil itself, and thus provides the heating to globally quench the coil. This eliminates the need for additional heaters. This method dumps all of the magnet’s stored energy into the coil windings. As the quench is very fast (less than a second) there is no time for significant amount of energy to be conducted away (ie: the quench is quasi-adiabatic),. In large magnets the stored energy can be enough to raise the temperature of the coils so quickly that it causes degradation to the fragile HTS layer in the tapes due to differential strain. Furthermore it can also cause some regions of a coil to get hotter than others, if those parts of the coil have reduced thermal capacity. The high terminal temperature (> 200K) and large variations in terminal temperature around a large coil are both potential causes of irreversible magnet degradation.

One solution is to increase the thermal capacity of the windings by adding electrically conductive padding or packing material, such as brass, copper or stainless steel, to the stacked tape cable from which the HTS coil is wound. When the partially insulated coil is actively quenched the current switches from the HTS layer to the stabilizing copper layer, and also share substantially into any other metal provided it is in good electrical contact to the tapes (eg: soldered between type-0 pairs). In this way more material can be provided to absorb the ohmic heating. Stainless steel has a very similar heat capacity to copper, but has greater mechanical strength and electrical resistivity, so it is a better choice as a stabilizer in a partially insulated coil (since current can share across turns and hence the electrical resistivity of the stabilizer is not a top priority).

However, adding packing metal reduces the coil’s current density, which is clearly undesirable in many applications where high current density is needed. One such application is the central column of the TF magnet in a spherical tokamak, where it is highly desirable to maximise the HTS current density to leave as much area available as possible for radiation shielding in the central column (which has a fixed diameter for a given tokamak major radius and aspect ratio). Adding packing material to increase heat capacity in the TF central column coil windings is therefore not feasible.

However, the return limbs of the coils of the TF magnet have no such restriction on space, and packing material can be freely added. However, doing this will not help if a coil turn’s stored energy is dumped uniformly around the turn during the quench - the regions without packing material will still experience the same temperature rise. It is therefore desirable to find a way to direct the stored energy of the magnet into the return limbs, which have been provided with sufficient packing material to absorb the majority of the coils stored energy whilst keeping their terminal temperature low enough. Since a large fraction of coil energy is directed to the limbs, the core windings also reach a lower terminal temperature despite having no added packing material, as they receive a smaller fraction of the total stored energy.

This desirable feature can be achieved by making the turn-turn resistance in the core part of the coil much larger (by several orders of magnitude) than that in the return limbs, as will now be described.

Figure 4 shows graphs of a simple thermal-electrical network model simulation of the performance of two turns of a TF coil in a small ST with major radius of 0.25m, split into a“core” (central column) leg and a return leg (limb). In each figure, the“core” (i.e. the central column part of the coil) is shown in blue or pink, and the return limb is shown in red or green. Where only one line is visible, the two lines coincide.

In the simplified model of two turns of a coil, the construction of a turn in the core and in the return limb is identical (each with 4 type-0 FITS pairs, with no packing, and a uniform turn-turn resistance of 10 ⁵ Ohms), but the core experiences a magnetic field which is twice that of the return limb. The coil is assumed to be conduction cooled to a remote heat sink and to be at a uniform temperature before ramping.

A global quench is induced at t = 0.8x10 ⁵ s by simply turning off the power supply, so that the transport current though the coil ceases. The coil’s inductance generates a voltage which continues to drive the current in each turn in a closed loop, shorting back to the start of the turn via the turn-turn resistance. This causes ohmic heating which reduces the critical current in the FITS. This process continues for a short period (typically a few seconds, depending on many factors) until the FITS in the turn quenches and generates sufficient voltage to eject its loop current into the metal stabilizer in the spiral path. Since this has a much higher resistance than the turn-turn resistance the turn’s magnetic field energy is rapidly converted to heat in the stabilizer and the turn temperature rises very fast (« 1 s). The duration of the two periods is slightly different in each turn in the coil, since each sees a different magnetic field, and has a slightly different perimeter length, amongst other differences. This means that the quench propagates radially in a cascade from turn-to-turn, like a wave breaking. The exact propagation wavefront shape depends on the turn-turn resistance, but this does not make a significant difference to the present disclosure.

Since the turn-turn resistance is constant around the loop, the quench dumps an equal amount of amount of energy per unit length into both parts of the turn, which rise to the same terminal temperature because they have the same amount of material per unit length. Note that the simple model does not account for the large increase in heat capacity of materials with temperature at cryogenic temperatures, so the absolute temperature values shown are not realistic. However, a more accurate calculation of the magnetic energy and coil enthalpy shows that the terminal coil temperature would be over 300 K. This is a rise of 288 K in less than 0.1 second, which would damage the HTS layers in the coils due to differential strain from sudden thermal expansion.

This issue can be resolved by adding packing tapes to the coil in the return limb section only, to increase the overall heat capacity of the coil, in addition to configuring the partial insulation such that the turn to turn resistance in the central column part of each turn is substantially more than the turn to turn resistance in the return limb. The latter modification will cause radial (turn to turn) current to flow preferentially in the return limbs, and hence cause more ohmic heating there, such that the majority of the magnet’s stored energy is dumped in the return limbs before the central column turns begins to shed radial current. The extra material in the limbs keeps the overall terminal coil temperature lower than the original design with uniform cable composition around the turn. More discussion of potential ways to achieve the tuned turn to turn resistance will be provided below, but first the results of such tuning will be shown.

Figure 5 shows the results for a coil in which:

• the number of HTS tapes in each turn is varied through the coil, so that that core section and the return limb section have similar critical currents (ie: lower magnetic field in the limbs means higher critical current per tape, so fewer tapes are needed to carry the transport current at the same fraction of critical current). Missing HTS tapes in the return limbs have been substituted with stainless steel to keep the heat capacity constant. In practice, additional stainless steel (e.g. as extra tapes) may be added in close proximity to the HTS to increase the heat capacity in the limbs..

• the turn to turn resistance of the core section is 10 ⁶ times the turn to turn resistance of the return limb section (10 Ohms vs 10 microOhms).

As can be seen from Figure 5, the temperature rise in the core is now insignificant compared to the temperature rise in the return limbs.

Metals other than stainless steel may also be used as packing material within the turns, to increase the heat capacity. Other materials having good heat conduction and high heat capacity may also be used, for example insulators, or metal-insulator transition materials such as vanadium oxide which act as insulators below a critical temperature, and act as conductors above that temperature. The packing material may be provided mainly between the HTS tapes of the turns, to ensure close thermal contact with the HTS. The choice of material will depend on the electrical and thermal design constraints of the coil, and will likely vary significantly between different coil designs - the important factor for this disclosure is that the parts of the coil with additional packing material have greater thermal capacity per unit length than the other parts of the coil.

Figure 6 shows the results for a coil similar to that of Figure 5, but with the turn to turn resistance selected to give approximately the same temperature rise in each section. This corresponds to turn-turn resistances of 0.5 Ohms and 0.2 milliOhms for the core and return limb sections respectively.

In each of figures 4 to 6, the quench is induced by shutting off the PSU current, causing a short delay as the turns warm slightly due to current flowing in loops returning via the low resistance radial (turn-turn) path, and then a temperature spike as HTS quenches and ejects current into the metal in the spiral path.

In the examples above, the structure of the core section and the return limb section has been substantially similar (with the replacement of one HTS tape pair with stainless steel). However, this need not be the case. The variation in resistance described above will adjust the amount of heat dumped into each section of the coil, but the temperature rise can be further controlled by increasing the heat capacity of each section - e.g. by selecting different materials (i.e. those with higher volumetric heat capacity, but similar structural/electrical properties) or increasing the amount of material in each section.

In the case of a TF coil, the return limbs have significantly less space constraint than the central column, so the return limb coils could be made with stainless steel strips packed between the HTS tapes so that they act as additional stabilizer and add extra heat capacity, or a similar integration of conductive, high heat capacity material to increase the total heat capacity of the return limbs.

While the above has been described in the context of a TF coil, it will be appreciated that the same principles apply to any partially insulated FITS coil - if the coil is asymmetric and notionally divided into two sections (connected in series), and there is a need to reduce the coils terminal temperature after an induced quench, by dumping more heat (per unit length) into the first section than would occur with uniform turn-turn resistance, then the first section is provided with a lower turn to turn resistance and additional heat capacity material than the second section. This need to dump more heat may come about, for example, due to design constrains allowing more heat capacity to be added to the first section than the second section, or due to an asymmetric magnetic field or strain profile on the coil, causing the second section to have a lower critical current.

Consider the case of a coil with a lower critical current during operation in a second section than in a first section, with uniform thermal capacity around the coil and uniform turn-turn resistance. During a situation in which current flows through the radial path (e.g. a quench or shutdown of the PSU), the coil will heat substantially evenly due to this radial current flow (neglecting any local hot-spot if the radial flow is due to a local quench). This will cause the second section to quench first - since that section will quench at a lower temperature. During the period where the second section has quenched and the first section has not, current in the spiral path in the second section will be flowing through resistive material - and so the second section will experience considerably more heating than the first section. This will result in a higher peak temperature in the second section (as seen for the“core” in Figure 4).

By adjusting the ratio of turn to turn resistance between the first and second section, a value can be chosen which ensures a shorter period between the quench of the first section and the quench of the second section, allowing the coil to heat more evenly, and reducing the peak temperature of the second section, and therefore the coil as a whole (the peak temperature of the first section will increase).

Therefore, even in situations where there is no ability to place additional thermal packing on one section of the coil, varying the turn-turn resistance as disclosed above can be advantageous - i.e. if there is a situation where a critical current during operation varies around the coil, a smoother temperature profile during quench or ramp down can be achieved by also varying the turn-turn resistance, such that regions of the coil with greater critical current have lower turn-turn resistance, and vice versa.

A TF magnet such as that discussed in the examples above is a good example of both reasons for varying the turn-turn resistance - if the central column comprises the same amount and kind of HTS conductor as the return limbs (i.e. there is no grading of the HTS), then the central column will have a reduced critical current during use due to the higher magnetic field (and additionally due to the higher strain, though this is a lesser effect in most TF magnets). In addition, space in the central column is much more limited than in the return limbs, so any additional packing material to increase heat capacity of the turns is preferably added in the return limbs.

It will also be appreciated that while the above has described two sections of each coil, the same would apply to three or more regions - e.g. for a full TF magnet with several coils connected in series, where each coil has a return limb and a core region (with the core regions of all the coils forming the central column), or for a coil having three sections each with a different turn to turn resistance (e.g. due to different constraints on the available space for packing material in each section).

The turn-turn resistance may be adjusted by the selection of metals for the cladding within the FITS cable and/or between turns of the coil. In typical FITS cables, this would be copper, but to allow for greater resistance other metals such as stainless steel may be used. Alternatively or additionally, the spacing between turns of the coil may be increased, resulting in a thicker (and hence more resistive) layer of metal between turns of the coil. Suitable materials that would also satisfy other engineering constraints (e.g. current density and structural stability) include germanium and other semi-conductors. A further alternative is to use normal (i.e. non-superconducting) metals for the insulation, but adjust the current path geometry using conventional insulation. In other words increase the distance over which current is forced to flow between turns.

The material between turns of the coil may include a partially insulating layer comprising a metal strip (or other electrically conducting strip) having“leaky insulation”, as shown in Figure 7A and B. The metal strip 901 is provided with a thin insulating coating 902 on at least the sides facing the HTS cables, where the insulating coating is removed or missing over windows (or“through holes”) 903 at intervals on each side of the metal strip. The windows can have any shape and can extend to the edges of the tape. The location of windows on either side of the metal strip are staggered, as shown in Figure 7B, which increases the resistance (compared to an uninsulated strip, or to a strip where the windows on each side were directly opposite each other) as the current must take a path 910 along part of the length of the metal strip.

By varying the spacing of the windows such that they are closer together in the return limbs and further apart in the core, the required difference in turn to turn resistance between the return limbs and core can be achieved. Further tuning may be achieved by using a different metal for the metal strip in the core compared to in the return limbs, or by varying other aspects of the geometry of the strip.

To allow for even further tuning, rather than a solid metal strip, a layer having several metal tracks may be used - effectively forming an insulating layer having conductive radial tracks disposed within it, where the spacing and length of the tracks determines the resistance of the partially insulating layer.

Figures 8 A to E shows an example leaky insulation layer. The leaky insulation layer comprises 5 layers, which are, in order:

• a first metal connection layer 161 1 ;

• a first insulating layer 1621 ;

• an electrically conducting layer 1630;

• a second insulating layer 1622;

• a second metal connection layer 1612. Figures 8 A to C show the layout of the first metal connection layer 161 1 , electrically conducting layer 1630, and second metal connection layer 1622 respectively. Figures 8 D and E are cross sections along the lines D and E in Figures 8 A to C.

The connection layer is present to facilitate attachment to FITS cables by soldering.

In contrast to the previous example where the electrically conducting layer is a continuous metal strip, in this example the electrically conducting layer is divided into several conductive regions. These regions come in two types. The square regions 1631 (though they may be any shape in practice) are connected by vias 1606 only to one of the metal connection layers. These regions do not affect the electrical properties of the partially insulating layer, but provide a thermal path through the respective insulating layer. By varying the size of these regions and the number of connections between them and the metal connection layer, the thermal properties of the partially insulating layer can be varied independently of the electrical properties.

The other regions 1632 each connect a window 1601 of the first insulating layer 1621 to a window 1602 of the second insulating layer 1622. The resistance between the windows can be controlled by varying the geometry of the regions 1632 - e.g. where the region 1632 contains a track 1633 which is elongate as shown in Figure 8B, increasing the width of the track would reduce the resistance between the windows, and increasing the length of the track (e.g. by providing a non-linear track, or by moving the windows) would increase the resistance between windows.

The windows 1601 in the first insulating layer are formed by drilled vias through the first connection layer and the first insulating layer, which are then plated with metal 1603 (or other electrically conductive material) to connect the first connection layer and the electrically conductive layer. The windows 1602 in the second insulating layer are formed by drilling a via 1602 through all of the layers, which is then plated with metal 1604 (or other electrically conductive material). To prevent a connection being formed to the first connection layer through the windows 1602 of the second insulating layer, the first connection layer is etched around the via 1602 to electrically isolate it, and an insulating cap 1605 is placed on the end of the via 1602 to ensure no bridging occurs due to soldering or contact with the FITS cable. As an alternative, the windows 1602 may instead be drilled from the other side of the partially insulating layer, such that they pass through the second connection layer, second insulating layer, and electrically conducting layer, and do not pass through (or do not pass completely through) the first insulating layer. As a further alternative, all the windows may be formed from vias which pass through all layers, with etching of the second connection layer and an insulating cap on the second connection layer being used for windows 1601 of the first insulating layer.

An advantage of using constructions such as those in figures 7 and 8 is that the electrical properties of the partial insulation can be varied substantially independently of the structural properties (e.g. thickness) and thermal properties (e.g. heat capacity and heat conduction). This allows the required tuning of the turn to turn resistance to be achieved, while also allowing variation of the thermal mass of each section and without significantly impinging on other design considerations (e.g. space in the central column).