- Email: [email protected]
The generation of continuous fields: Future prospects
Journal of Magnetism and MagneticMaterials 11 (1979) 293-299 © North-Holland Publishing Company
THE GENERATION OF CONTINUOUS FIELDS: FUTURE PROSPECTS D. Bruce MONTGOMERY Francis Bitter National Magnet Laboratory, Massachusetts Institute o f Technology, 170 Albany Street, Cambridge, MA 02139, USA Received 20 September 1978 It is presently possible to generate a continuous field of 30 T by combiningan outer superconductingcoil with a 10 inner water-cooled coil. Using superconductorsalone it has been possible to generate 17.5 T and a 20 T cot has been proposed for construction in the nearterm future. Far beyond these accomplishments lie the measured upper-critical fields of the yet untried superconductors, some in excess of 50 T. This paper examines what we may expect in the future for superconducting and hybrid systems, and compares these truly continuous field systemswith systemsgenerating quasi-continuous long pulses.
1. Introduction
into independent nested solenoidal elements the stress can be limited to an arbitrary level by controlling the field generated by each sub-element [2,3]. The inner elements are more efficient of course, and the less field they can generate, the more must be generated in elements further out. Nonetheless, fields of arbitrary level could be generated. The limit then is not fundamental, but rather practical. Does the magnet grow too power consumptive, or become impossibly large and expensive. It is too idealistic to examine a 75 T continuous magnet without examining the current state-of-theart, and hence give some context and indication of how far one must extrapolate. We therefore approach future prospects through current status, and We do so first with superconducting coils and then resistive and hybrid systems. Finally, we examine the interim approach to 75 T, namely the quasi-static field systems.
The quest for magnetic field levels beyond that readily available at the time has been pursued for much of this century. The generation of continuous field, for example, progressed from the large ironmagnets of the 1920's t o the 10 T mark in the 30's, the 20 T mark in the early 60's and to the 30 T mark in the late 70's. Each increase was a major undertaking and it took significant time. Time is needed to develop and test concepts, and to marshal the major new commitment of resources. A recent National Research Council Panel [1] in addressing the future prospects for high field research examined the significance for research were a 75 T continuous field to be available. Fields of this magnitude are near the present limit for non-destructive capacitor-driven pulse coils. There appears no fundamental reason however, why fields of this magnitude could not be generated continuously were the justification sufficient to support the costs. There are three obvious design approaches which must be examined: fully resistive, fully superconducting and the hybrid approach combining the two. Common to all three approaches is the need to deal with magnetic stresses. The equivalent magnetic pressure at 75 T is 22 000 atmospheres. It should be remembered however, that magnetic stress is not a fundamental limit for solenoids. If a magnet is divided
2. Superconducting magnets 2.1. Present status
Superconducting magnets have come of age since the early 1960's when the first small volume nominal 10 T magnet appeared. The magnets listed in table 1 are representative of the current state-of-the-art. The 293
D. Bruce Montgomery / Generation of continuous fields
294
Table 1 State-of-the-art superconducting magnets Field
Bore
17.5 13.5 16.5 15.5 14.6 14.0
3 16 4 5.5 6.5 9.5
T T T T T Ta
Center cm cm cm cm cm cm
NRIM, Japan Lebedev, Moscow El, Moscow ORNL, NASA Carx~igie Mellon
a High resolution NMR.
advent of these reasonable volume fields above 10 T has made it possible to extend high field research to a number of centers. Prior to these developments, continuous fields of this level were available only in those central facilities having large power supplies. While fields of 10 T are easily reachable with superconducting magnets, the cost of producing fields substantially above that point rise rapidly. Fig. 1 gives an approximate cost curve for 5 cm bore coils. There is a factor of 5 rise between 10 T and 15 T and an additional factor of 3 rise is going to 17.5 T, the highest field generated to date. This rapid rise in cost has thus far limited the number of superconducting magnets generating fields above 14 T to approximately 25 magnets throughout the world. 400
As the capital investment in a superconducting magnet increases, so does the investment in support equipment and in the necessary operating cost to fully utilize it. As an approximation one might imagine an additional initial investment equivalent to the coil cost, and an annual operating cost for materials and people devoted to operations ranging from $25 K upwards depending on the System size and complexity. Failure to provide such operating expenses in the past has limited the full usefulness of many superconducting installations. Superconducting magnets have certain advantages beyond their independence from large central power supplies. In principle they can be put in persistant mode and hence are free from time variations. This plus the ability to carefully control the winding process can lead to extremely high homogeneity fields for NMR. Many NMR grade systems in the 6 to 8 T range are in use, but work is just beginning in the higher field ranges. An NMR system has recently been commissioned which produces 14 T in a 9.5 cm bore with an uncompensated homogeneity of 2 X 10 - s G over the central centimeter. This has been compensated to allow resolution of 0.4 Hz at 600 MHz. The magnet system cost exclusive of spectrometer was approximately $250 K. Further parameter improvements of NMR will undoubtedly result when multifilamentary Nb3Sn is available and when persistent current joints can be developed.
2.2. Future possibilities 300 O9
n~ _1 ..J
200
o
lqo
lOT
[
,l
i
12.5T 15T 17.5T FIELD (TESLA) Fig. 1. Approximate cost of 5 em bore superconducting coils in the field range from 10 to 17.5 T.
In the near term, multifilamentary Nba Sn materials will be sufficiently well developed to provide an alternative for the Nb3Sn or V3Ga tapes now used in coils. This should lead to faster sweep times and to potentially higher homogeneity and to persistent mode coils. Also in the near term one can expect some extention of the maximum field generated by superconductors, if the very best V3Ga tapes are used, a field of 20 T can be generated in a 3.2 cm bore for a magnet cost approximately a factor of 23 times higher than at 17.5 T [4]. This would put the cost of such a magnet at about $700 K. In the far term one can speculate on the impact of applying higher critical field materials. There are,
D. BruceMontgomery / Generation of continuous fieMs \
\
\
\
8M \
\ \
7M
\ 6M
E IC ,¢C
O3 n,,¢ 5 M -J .J
"o
/
o4M E3
0
295
I
I0
I
,
,
I
I
i
~
20 30 FIELD (TESLA)
Pbu.,_~ I
I
40
3M
I
50
Fig. 2. L i m i t i n g w i n d i n g c u r r e n t d e n s i t y a s s u m e d for a d v a n c e d
superconducting cots. The limiting useful field for various developed and speculative superconducting materials are indicated. of course, materials with critical fields in the 50 T range. Experience with all materials developed to date suggests, however, that while the higher critical field materials obviously extend the maximum possible fields, they tend to have the same constant product of field times critical current as the lower field materials [5]. Thus any given wire will carry half the current at 20 T as it did at 10 T independent of what superconductor is used in the wire. If we make some reasonable assumptions about the cost and properties of future materials, we can project the cost of magnets above 20 T. We assume that: (1) consideration of coil protection limit overall coil current density to 10 X 103 A/cm2; (2) this limiting current density can be used up to 20 T only, and must be degraded at a constant H X/above that point as increasing fractions of the conductor must be occupied by the superconductor; (3) any portion of the winding at diameters greater than 40 cm must drop the limiting current density to 7.5 X 103 A/cm 2, again for protection considerations; (4) any new materials will cost the same as Nb3Sn scaled by the field to which it is exposed. These limiting assumptions are illustrated in fig. 2 together with the appropriate field ranges for the currently developed and for the speculative materials. The cost curve resulting from these assumptions is given in fig. 3 up to 30 T. The cost of a 30 T magnet is clearly very high, and in fact compares unfavorably with the cost of alternative generation
/
n
2M IM I
15
20
25
1
30
FIELD (TESLA) Fig. 3. Estimated cost of 3 cm bore superconducting coils for very high fields, using the current density limitations of fig. 2. Material cost are based on current Nb3Sn conductors sealed by the field to which they are exposed.
techniques such as hybrid systems. Fig. 3 for fields above 20 T further assumes the commercial development of materials which exist now only as small samples. Unfortunately, there is reason to expect that this commercialization may not happen because there are no markets of sufficient scale to warrent the very expensive development. Niobium-titanium and niobium-tin have been developed in response to the high energy, fusion, and rotating machine markets, none of which require or profit from field levels beyond 15 T. Research magnets, no matter how ambitious, would not compare with the scale of these applications. If we accept the basic assumptions that went into fig. 3 it appears that continuous field generation beyond 20 T would turn to other than superconducting systems, such as resistive or resistive/ superconducting hybrid systems. If fields must have special characteristics, such as high resolution or extraordinary spacial homogeneity, the superconducting approach might be carried further inspite of the unfavorable cost comparison. It is vital to also consider what might happen were the assumption of fig. 3 not to hold. Most influential would be a future violation of the constant H × / principle. This constancy is based on
296
D. Bruce Montgomery / Generation of continuous fields
defect-type flux-pinning and hence based more on experimental than fundamental arguments. Considerable effort has gone into maximizing current carryhag capacity in NbTi, NbaSn and VaGa. The prospect that fundamentally different pinning mechanisms will be discovered which apply to the higher field materials can not be ruled out, but the probability would not appear high. It must also be noted that the allowable overall current density in the magnets would not be altered by such a new mechanism. Overall current density is more determined by stability and protection considerations..However, the amount of superconductor required per unit coil cross section would diminish, and that would be expected to have an impact on the wire cost.
3. Resistive and hybrid magnets 3.1. Present status
For the foreseeable future, continuous fields above the 20 T level will be generated by resistive water-cooled magnets, or resistive inserts boosted by external superconducting coils. These boosted, or hybrid, systems are a good marriage because they place the resistive elements on the inside where the field is high and the power requirements lowest, and the superconductor on the outside where the field is low enough but the power requirements highest. The current state-of-the-art for resistive and hybrid systems is given in table 2. Table 2 indicates that resistive magnets must have power supplies of 5.0 MW capacity, and hybrid magnets of 2.5 MW capacity to offer simple field magnitude advantages over superconducting magnets. There are certain advantages to each type, Table 2 State-of-the-art of resistive and hybrid magnets Field
Bore
Power
Type
30 T 25 T 20 T 23.5 T 19.5 T 18.5 T 16 T
3.2cm 3.2 cm 3.2cm 3.2 cm 5.4 cm 3.2cm 5.4 cm
10 MW 5 MW 2.5 MW 10 MW 10 MW 5 MW 5 MW
hybrid .... " resistive " " "
of course, and in overlapping areas choices are made depending on local facilities or special field requirements. In central facilities where power supplies exist, the hybrid concept can be readily applied to boost any field level limited by the available power. They can also be used to allow multiple experiments by reducing the power necessary for a given field, releasing power supplies for parallel operation. A hybrid magnet to generate 30 T with 10 MW power requires an investment of approximately $300 K for a 7.5 T boost superconducting coil and another $200 K for installation and a suitable closed-cycle refrigeration system. If 30 T were desired and no facilities existed, one must examine the total capital cost. The total cost will depend on the field generated by the power supply. If 8 T is chosen for the superconductor, as being a modest technology undertaking, 10 MW of power would be required, calling for an investment of approximately $5 M for power supplies and $0.5 M for the superconducting system. If one instead chooses 12 T for the superconductor as being a reasonable step forward, the superconducting system cost increases to approximately $1.7 M, but the power requirement drops to $3.3 M. The total cost for the latter case is thus about 10% lower. Once given a central power supply, however, it can be time shared with a number of magnets in a cost effective way. Water cooled magnets are relatively inexpensive, a 5 MW 18.5 T repeat magnet costing about $20 K to construct. A large central facility such as the National Magnet Laboratory has 24 high field magnets time sharing a central 10 MW power supply which can be subdivided into four 2.5 MW units. Two projected hybrid systems will extend the maximum available field to 30 T, and allow two 25 T magnets to run simultaneously. 3.2. Future possibilities
An ambitious near term goal for hybrid systems might be to extend an NMR environment beyond that reachable with superconductors. For example, a field of 25 T in a 10 cm bore hybrid magnet could be generated with 4.0 MW's of power and a 15 T superconducting boost coil. The field could be
D. Bruce Montgomery / Generation o f continuous fields
stabilized by series regulation and stabilizer coils fed-back from the field, with a final stage stabilization by means of a thin superconducting shield made from a high critical field material. Resistive magnets at 15 T have been homogenized to 1 part in 104 over a 6 cm sphere without shim coils. Running cost for such a system will be high, however, unless the magnet can be rapidly brought to proper homogeneity on separate runs. A 75 T hybrid magnet would be a longer term very ambitious goal. The present record continuous field of 30 T [6] could be extended to much higher fields if the necessary very large power supplies were available. Utilizing heat transfer rates and copper alloys reinforced with interleaved steel sheets, both presently used in the 30 T hybrid magnet, one can extrapolate power requirements to a field level of 75 T. Fig. 4 shows such an extrapolation, with the amount of reinforcing progressively increased as the field increases. Were such a coil, having an outer diameter of 1 m, to be operated without an external superconducting booster section, it would require 165 MW continuously to generate 75 T. Even with an ambitious 20 T superconducting booster coil, 100 MW's would still be required. While such a 75 T system is technically feasible, its cost clearly would be high. A 100 MW power
supply of adequate stability would cost about $50 M. A 20 T booster coil of 1 m inner diameter~is estimated to cost an additional $25 M at a minimum. Power costs for a 100 MW supply would run $5000/h at $0.05/kW h. The largest continuous dc supply suitable for research magnets now in existance is 50 MW (Princeton Plasma Physics Laboratory) and is used for their nuclear fusion machines. The NML and Grenoble each have 10 MW supplies, the largest devoted to high field research applications. We can expect some progress in the intermediate range between the 30 T now possible and 75 T, which is feasible but certainly a long range goal. Moving from 30 to 35 T is certainly feasible, by simply increasing the present 30 T hybrid booster field from 7.5 T to 12.5 T. An investment of approximately $1.5 M would be required. If one were able to use the Princeton 50 MW supply for a research magnet, it is possible that a field of 50 T could be achieved for an investment of approximately $5 M in a 10 T booster superconducting coil to surround a 40 T, 50 MW watercooled insert.
4. Long pulse systems 4.1. Present status
200 /
/RESEnVE
150
//,o,
, ,,, I00
I
0
297
i
/
25T 50T CENTRAL FIELD
/
/
t..B.,D / 2or
i
75T
Fig. 4. Power supply requirements for 3 cm bore resistive and hybrid magnets. The two hybrid curves are for 10 T and 20 T booster field respectively.
While continuous fields are necessary for many experiments and always desirable, certain experiments can be carried out in shorter times. Pulse times in the order of one second could represent an approach for experiments which benefit from continuous fields. These long pulses can be powered by various techniques. In the University of Amsterdam facility [7] a 5.6 MW controlled rectifier supply switches directly onto the mains. A field of 40 T can be produced for 0.1 s in a 2 cm bore. Similar parameters have been obtained at the NML by SCR switching of the fully-excited rotating dc generator output [8]. At Toulouse a third approach [9] uses a 1.25 MJ capacitor bank and a crow-bar circuit to produce an 0.1 s rise and a 1 s fall to a peak of 40 T in a 2.5 cm bore. These coils operate adiabatically and generally utilize precooling with liquid nitrogen or, in the
298
D. Bruce Montgomery / Generation of continuous fields
Amsterdam case, liquid neon, to limit the temperature rise. Recool times of present devices are the order of 30 min, but increased attention to obtaining high repetition rates could reduce this to the 5 min range. Because the pulse rate is limited, particular attention must be paid to multiple channel data collection and sophisticated diagnostics. This must be an important part of any limited time scale facility and can dominate the cost of experimental equipment. 4.2. Future possibilities
Long pulse facilities are likely to represent the best chance to extend present experiments well above 30 T over the next decade. A national resource greatly enhancing this possibility is the increasing use of large pulse supplies for fusion experiments throughout the community. The NML, for example, is installing a 200 MW, 200 MJ pulse supply for the Alcator fusion experiment. It is of interest to examine what such a supply can offer to long pulse experiments in small volumes. Precooled magnets which depend on thermal
tOOT
75T
~
W =4000kgms
50T
251" -
-O
I
I
inertia are limited by the product of magnet current density squared times the pulse time (1-2r). The larger the scale of the magnet for a given field, the lower the resultant current density and hence the longer the pulse can be without over heating. The larger the scale, the larger the energy source must be of course, but the longer the pulse can be held. This is illustrated by fig. 5. The figure illustrates two scales, namely magnet weights of 500 kg and 4000 kg, and gives the field achievable at various pulse times, subject to a given temperature limit. The scale curves also represent constant energy requirements. We note that 50 T can be generated for 1 s if a coil has a mass of 4000 kg and if a 200 MJ energy source is available. A peak power of 200 MW would be required. The curve also indicates that 75 T could be held for 0.5 s with the same temperature rise but 400 MW peak power supply would be required. It is of interest to note also that the resistive continuous field coils in fig. 2 could also be driven on a pulse basis. The magnets are more complex than the LN 2 precooled coils, and require large coolant pumps, but fields can be held for times determined by the energy storage of the supply rather than the heat capacity of the coils. The cost of the long pulse magnets themselves is relatively low. A 50 T 1 s magnet is estimated to cost about $100 K. A large pulse supply involving a rotating alternator capable of delivering 200 MJ at a 200 MW peak power level would cost approximately $10 M. The long pulse magnets, particularly utilizing existing fusion program power supplies, represent the least expensive entree into the field range between 30 T and 75 T. Realization of these ambitious pulse coils will represent a major magnet design challenge for the next few years. This program will pave the way for truely continuous fields in this range through technology exploration and through opening up significant physics opportunities.
t
0.5 I.O 1.5 T SECONDS Fig. 5. Field versus pulse time for 5 cm bore LN2-cooled long-pulse coils. The two curvesare for constant magnet weights and stored energies. The peak resistive powers required can be estimated by dividing the stored energy by the pulse time.
Acknowledgements
The author is indebted to Paul Schwartz for his comments on the future of superconducting magnets and to Robert Weggel for his comments concerning hybrid and long pulse systems.
D. Bruce Montgomery / Generation o f continuous fields
References [1 ] Natl. Res. Council, Am. Acad. Sci., Panel Rept. Research Opportunities in High Magnetic Fields, Chairman, Seymour Keller, October 1978. [2] D. Bruce Montgomery, Solenoid Magnet designs, (Wiley, New York, 1969) p. 127. [3] P. Carden and A.M. Collins, J. Phys. E, 7 (1974) 750. [4] P.S. Schwartz, W.D. M~kiewicz, C.H. Rosner, Physica 86-88B (1977) 1084.
299
[5] D. Dew-Hughes, "Superconducting Materials for LargeScale Applications", Adv. Cryogenic Eng. 22 (1975) 316. [6] M.J. Leupold, R. Weggel, Y. Iwasa, 6th Int. Magnet Technology Conf., Bratislava (1977). [7] F.A. Muller, L.W. Roeland and R. Gersdorf, Proc. 3rd Int. Conf. on Magnet Technology, Hamburg (May 1970). [8] H.C. Praddaude and S. Foner, Production of Quasistatic High Magnetic Fields by Switching Low Voltage DC Generators, IEEE Magnetics, September 1975. [9] S. Askenazy, G. Carrere and J. Marquez, Physique sous Champs Magnetiques Intense (CNRS Paris 1975) p.357.
THE GENERATION OF CONTINUOUS FIELDS: FUTURE PROSPECTS D. Bruce MONTGOMERY Francis Bitter National Magnet Laboratory, Massachusetts Institute o f Technology, 170 Albany Street, Cambridge, MA 02139, USA Received 20 September 1978 It is presently possible to generate a continuous field of 30 T by combiningan outer superconductingcoil with a 10 inner water-cooled coil. Using superconductorsalone it has been possible to generate 17.5 T and a 20 T cot has been proposed for construction in the nearterm future. Far beyond these accomplishments lie the measured upper-critical fields of the yet untried superconductors, some in excess of 50 T. This paper examines what we may expect in the future for superconducting and hybrid systems, and compares these truly continuous field systemswith systemsgenerating quasi-continuous long pulses.
1. Introduction
into independent nested solenoidal elements the stress can be limited to an arbitrary level by controlling the field generated by each sub-element [2,3]. The inner elements are more efficient of course, and the less field they can generate, the more must be generated in elements further out. Nonetheless, fields of arbitrary level could be generated. The limit then is not fundamental, but rather practical. Does the magnet grow too power consumptive, or become impossibly large and expensive. It is too idealistic to examine a 75 T continuous magnet without examining the current state-of-theart, and hence give some context and indication of how far one must extrapolate. We therefore approach future prospects through current status, and We do so first with superconducting coils and then resistive and hybrid systems. Finally, we examine the interim approach to 75 T, namely the quasi-static field systems.
The quest for magnetic field levels beyond that readily available at the time has been pursued for much of this century. The generation of continuous field, for example, progressed from the large ironmagnets of the 1920's t o the 10 T mark in the 30's, the 20 T mark in the early 60's and to the 30 T mark in the late 70's. Each increase was a major undertaking and it took significant time. Time is needed to develop and test concepts, and to marshal the major new commitment of resources. A recent National Research Council Panel [1] in addressing the future prospects for high field research examined the significance for research were a 75 T continuous field to be available. Fields of this magnitude are near the present limit for non-destructive capacitor-driven pulse coils. There appears no fundamental reason however, why fields of this magnitude could not be generated continuously were the justification sufficient to support the costs. There are three obvious design approaches which must be examined: fully resistive, fully superconducting and the hybrid approach combining the two. Common to all three approaches is the need to deal with magnetic stresses. The equivalent magnetic pressure at 75 T is 22 000 atmospheres. It should be remembered however, that magnetic stress is not a fundamental limit for solenoids. If a magnet is divided
2. Superconducting magnets 2.1. Present status
Superconducting magnets have come of age since the early 1960's when the first small volume nominal 10 T magnet appeared. The magnets listed in table 1 are representative of the current state-of-the-art. The 293
D. Bruce Montgomery / Generation of continuous fields
294
Table 1 State-of-the-art superconducting magnets Field
Bore
17.5 13.5 16.5 15.5 14.6 14.0
3 16 4 5.5 6.5 9.5
T T T T T Ta
Center cm cm cm cm cm cm
NRIM, Japan Lebedev, Moscow El, Moscow ORNL, NASA Carx~igie Mellon
a High resolution NMR.
advent of these reasonable volume fields above 10 T has made it possible to extend high field research to a number of centers. Prior to these developments, continuous fields of this level were available only in those central facilities having large power supplies. While fields of 10 T are easily reachable with superconducting magnets, the cost of producing fields substantially above that point rise rapidly. Fig. 1 gives an approximate cost curve for 5 cm bore coils. There is a factor of 5 rise between 10 T and 15 T and an additional factor of 3 rise is going to 17.5 T, the highest field generated to date. This rapid rise in cost has thus far limited the number of superconducting magnets generating fields above 14 T to approximately 25 magnets throughout the world. 400
As the capital investment in a superconducting magnet increases, so does the investment in support equipment and in the necessary operating cost to fully utilize it. As an approximation one might imagine an additional initial investment equivalent to the coil cost, and an annual operating cost for materials and people devoted to operations ranging from $25 K upwards depending on the System size and complexity. Failure to provide such operating expenses in the past has limited the full usefulness of many superconducting installations. Superconducting magnets have certain advantages beyond their independence from large central power supplies. In principle they can be put in persistant mode and hence are free from time variations. This plus the ability to carefully control the winding process can lead to extremely high homogeneity fields for NMR. Many NMR grade systems in the 6 to 8 T range are in use, but work is just beginning in the higher field ranges. An NMR system has recently been commissioned which produces 14 T in a 9.5 cm bore with an uncompensated homogeneity of 2 X 10 - s G over the central centimeter. This has been compensated to allow resolution of 0.4 Hz at 600 MHz. The magnet system cost exclusive of spectrometer was approximately $250 K. Further parameter improvements of NMR will undoubtedly result when multifilamentary Nb3Sn is available and when persistent current joints can be developed.
2.2. Future possibilities 300 O9
n~ _1 ..J
200
o
lqo
lOT
[
,l
i
12.5T 15T 17.5T FIELD (TESLA) Fig. 1. Approximate cost of 5 em bore superconducting coils in the field range from 10 to 17.5 T.
In the near term, multifilamentary Nba Sn materials will be sufficiently well developed to provide an alternative for the Nb3Sn or V3Ga tapes now used in coils. This should lead to faster sweep times and to potentially higher homogeneity and to persistent mode coils. Also in the near term one can expect some extention of the maximum field generated by superconductors, if the very best V3Ga tapes are used, a field of 20 T can be generated in a 3.2 cm bore for a magnet cost approximately a factor of 23 times higher than at 17.5 T [4]. This would put the cost of such a magnet at about $700 K. In the far term one can speculate on the impact of applying higher critical field materials. There are,
D. BruceMontgomery / Generation of continuous fieMs \
\
\
\
8M \
\ \
7M
\ 6M
E IC ,¢C
O3 n,,¢ 5 M -J .J
"o
/
o4M E3
0
295
I
I0
I
,
,
I
I
i
~
20 30 FIELD (TESLA)
Pbu.,_~ I
I
40
3M
I
50
Fig. 2. L i m i t i n g w i n d i n g c u r r e n t d e n s i t y a s s u m e d for a d v a n c e d
superconducting cots. The limiting useful field for various developed and speculative superconducting materials are indicated. of course, materials with critical fields in the 50 T range. Experience with all materials developed to date suggests, however, that while the higher critical field materials obviously extend the maximum possible fields, they tend to have the same constant product of field times critical current as the lower field materials [5]. Thus any given wire will carry half the current at 20 T as it did at 10 T independent of what superconductor is used in the wire. If we make some reasonable assumptions about the cost and properties of future materials, we can project the cost of magnets above 20 T. We assume that: (1) consideration of coil protection limit overall coil current density to 10 X 103 A/cm2; (2) this limiting current density can be used up to 20 T only, and must be degraded at a constant H X/above that point as increasing fractions of the conductor must be occupied by the superconductor; (3) any portion of the winding at diameters greater than 40 cm must drop the limiting current density to 7.5 X 103 A/cm 2, again for protection considerations; (4) any new materials will cost the same as Nb3Sn scaled by the field to which it is exposed. These limiting assumptions are illustrated in fig. 2 together with the appropriate field ranges for the currently developed and for the speculative materials. The cost curve resulting from these assumptions is given in fig. 3 up to 30 T. The cost of a 30 T magnet is clearly very high, and in fact compares unfavorably with the cost of alternative generation
/
n
2M IM I
15
20
25
1
30
FIELD (TESLA) Fig. 3. Estimated cost of 3 cm bore superconducting coils for very high fields, using the current density limitations of fig. 2. Material cost are based on current Nb3Sn conductors sealed by the field to which they are exposed.
techniques such as hybrid systems. Fig. 3 for fields above 20 T further assumes the commercial development of materials which exist now only as small samples. Unfortunately, there is reason to expect that this commercialization may not happen because there are no markets of sufficient scale to warrent the very expensive development. Niobium-titanium and niobium-tin have been developed in response to the high energy, fusion, and rotating machine markets, none of which require or profit from field levels beyond 15 T. Research magnets, no matter how ambitious, would not compare with the scale of these applications. If we accept the basic assumptions that went into fig. 3 it appears that continuous field generation beyond 20 T would turn to other than superconducting systems, such as resistive or resistive/ superconducting hybrid systems. If fields must have special characteristics, such as high resolution or extraordinary spacial homogeneity, the superconducting approach might be carried further inspite of the unfavorable cost comparison. It is vital to also consider what might happen were the assumption of fig. 3 not to hold. Most influential would be a future violation of the constant H × / principle. This constancy is based on
296
D. Bruce Montgomery / Generation of continuous fields
defect-type flux-pinning and hence based more on experimental than fundamental arguments. Considerable effort has gone into maximizing current carryhag capacity in NbTi, NbaSn and VaGa. The prospect that fundamentally different pinning mechanisms will be discovered which apply to the higher field materials can not be ruled out, but the probability would not appear high. It must also be noted that the allowable overall current density in the magnets would not be altered by such a new mechanism. Overall current density is more determined by stability and protection considerations..However, the amount of superconductor required per unit coil cross section would diminish, and that would be expected to have an impact on the wire cost.
3. Resistive and hybrid magnets 3.1. Present status
For the foreseeable future, continuous fields above the 20 T level will be generated by resistive water-cooled magnets, or resistive inserts boosted by external superconducting coils. These boosted, or hybrid, systems are a good marriage because they place the resistive elements on the inside where the field is high and the power requirements lowest, and the superconductor on the outside where the field is low enough but the power requirements highest. The current state-of-the-art for resistive and hybrid systems is given in table 2. Table 2 indicates that resistive magnets must have power supplies of 5.0 MW capacity, and hybrid magnets of 2.5 MW capacity to offer simple field magnitude advantages over superconducting magnets. There are certain advantages to each type, Table 2 State-of-the-art of resistive and hybrid magnets Field
Bore
Power
Type
30 T 25 T 20 T 23.5 T 19.5 T 18.5 T 16 T
3.2cm 3.2 cm 3.2cm 3.2 cm 5.4 cm 3.2cm 5.4 cm
10 MW 5 MW 2.5 MW 10 MW 10 MW 5 MW 5 MW
hybrid .... " resistive " " "
of course, and in overlapping areas choices are made depending on local facilities or special field requirements. In central facilities where power supplies exist, the hybrid concept can be readily applied to boost any field level limited by the available power. They can also be used to allow multiple experiments by reducing the power necessary for a given field, releasing power supplies for parallel operation. A hybrid magnet to generate 30 T with 10 MW power requires an investment of approximately $300 K for a 7.5 T boost superconducting coil and another $200 K for installation and a suitable closed-cycle refrigeration system. If 30 T were desired and no facilities existed, one must examine the total capital cost. The total cost will depend on the field generated by the power supply. If 8 T is chosen for the superconductor, as being a modest technology undertaking, 10 MW of power would be required, calling for an investment of approximately $5 M for power supplies and $0.5 M for the superconducting system. If one instead chooses 12 T for the superconductor as being a reasonable step forward, the superconducting system cost increases to approximately $1.7 M, but the power requirement drops to $3.3 M. The total cost for the latter case is thus about 10% lower. Once given a central power supply, however, it can be time shared with a number of magnets in a cost effective way. Water cooled magnets are relatively inexpensive, a 5 MW 18.5 T repeat magnet costing about $20 K to construct. A large central facility such as the National Magnet Laboratory has 24 high field magnets time sharing a central 10 MW power supply which can be subdivided into four 2.5 MW units. Two projected hybrid systems will extend the maximum available field to 30 T, and allow two 25 T magnets to run simultaneously. 3.2. Future possibilities
An ambitious near term goal for hybrid systems might be to extend an NMR environment beyond that reachable with superconductors. For example, a field of 25 T in a 10 cm bore hybrid magnet could be generated with 4.0 MW's of power and a 15 T superconducting boost coil. The field could be
D. Bruce Montgomery / Generation o f continuous fields
stabilized by series regulation and stabilizer coils fed-back from the field, with a final stage stabilization by means of a thin superconducting shield made from a high critical field material. Resistive magnets at 15 T have been homogenized to 1 part in 104 over a 6 cm sphere without shim coils. Running cost for such a system will be high, however, unless the magnet can be rapidly brought to proper homogeneity on separate runs. A 75 T hybrid magnet would be a longer term very ambitious goal. The present record continuous field of 30 T [6] could be extended to much higher fields if the necessary very large power supplies were available. Utilizing heat transfer rates and copper alloys reinforced with interleaved steel sheets, both presently used in the 30 T hybrid magnet, one can extrapolate power requirements to a field level of 75 T. Fig. 4 shows such an extrapolation, with the amount of reinforcing progressively increased as the field increases. Were such a coil, having an outer diameter of 1 m, to be operated without an external superconducting booster section, it would require 165 MW continuously to generate 75 T. Even with an ambitious 20 T superconducting booster coil, 100 MW's would still be required. While such a 75 T system is technically feasible, its cost clearly would be high. A 100 MW power
supply of adequate stability would cost about $50 M. A 20 T booster coil of 1 m inner diameter~is estimated to cost an additional $25 M at a minimum. Power costs for a 100 MW supply would run $5000/h at $0.05/kW h. The largest continuous dc supply suitable for research magnets now in existance is 50 MW (Princeton Plasma Physics Laboratory) and is used for their nuclear fusion machines. The NML and Grenoble each have 10 MW supplies, the largest devoted to high field research applications. We can expect some progress in the intermediate range between the 30 T now possible and 75 T, which is feasible but certainly a long range goal. Moving from 30 to 35 T is certainly feasible, by simply increasing the present 30 T hybrid booster field from 7.5 T to 12.5 T. An investment of approximately $1.5 M would be required. If one were able to use the Princeton 50 MW supply for a research magnet, it is possible that a field of 50 T could be achieved for an investment of approximately $5 M in a 10 T booster superconducting coil to surround a 40 T, 50 MW watercooled insert.
4. Long pulse systems 4.1. Present status
200 /
/RESEnVE
150
//,o,
, ,,, I00
I
0
297
i
/
25T 50T CENTRAL FIELD
/
/
t..B.,D / 2or
i
75T
Fig. 4. Power supply requirements for 3 cm bore resistive and hybrid magnets. The two hybrid curves are for 10 T and 20 T booster field respectively.
While continuous fields are necessary for many experiments and always desirable, certain experiments can be carried out in shorter times. Pulse times in the order of one second could represent an approach for experiments which benefit from continuous fields. These long pulses can be powered by various techniques. In the University of Amsterdam facility [7] a 5.6 MW controlled rectifier supply switches directly onto the mains. A field of 40 T can be produced for 0.1 s in a 2 cm bore. Similar parameters have been obtained at the NML by SCR switching of the fully-excited rotating dc generator output [8]. At Toulouse a third approach [9] uses a 1.25 MJ capacitor bank and a crow-bar circuit to produce an 0.1 s rise and a 1 s fall to a peak of 40 T in a 2.5 cm bore. These coils operate adiabatically and generally utilize precooling with liquid nitrogen or, in the
298
D. Bruce Montgomery / Generation of continuous fields
Amsterdam case, liquid neon, to limit the temperature rise. Recool times of present devices are the order of 30 min, but increased attention to obtaining high repetition rates could reduce this to the 5 min range. Because the pulse rate is limited, particular attention must be paid to multiple channel data collection and sophisticated diagnostics. This must be an important part of any limited time scale facility and can dominate the cost of experimental equipment. 4.2. Future possibilities
Long pulse facilities are likely to represent the best chance to extend present experiments well above 30 T over the next decade. A national resource greatly enhancing this possibility is the increasing use of large pulse supplies for fusion experiments throughout the community. The NML, for example, is installing a 200 MW, 200 MJ pulse supply for the Alcator fusion experiment. It is of interest to examine what such a supply can offer to long pulse experiments in small volumes. Precooled magnets which depend on thermal
tOOT
75T
~
W =4000kgms
50T
251" -
-O
I
I
inertia are limited by the product of magnet current density squared times the pulse time (1-2r). The larger the scale of the magnet for a given field, the lower the resultant current density and hence the longer the pulse can be without over heating. The larger the scale, the larger the energy source must be of course, but the longer the pulse can be held. This is illustrated by fig. 5. The figure illustrates two scales, namely magnet weights of 500 kg and 4000 kg, and gives the field achievable at various pulse times, subject to a given temperature limit. The scale curves also represent constant energy requirements. We note that 50 T can be generated for 1 s if a coil has a mass of 4000 kg and if a 200 MJ energy source is available. A peak power of 200 MW would be required. The curve also indicates that 75 T could be held for 0.5 s with the same temperature rise but 400 MW peak power supply would be required. It is of interest to note also that the resistive continuous field coils in fig. 2 could also be driven on a pulse basis. The magnets are more complex than the LN 2 precooled coils, and require large coolant pumps, but fields can be held for times determined by the energy storage of the supply rather than the heat capacity of the coils. The cost of the long pulse magnets themselves is relatively low. A 50 T 1 s magnet is estimated to cost about $100 K. A large pulse supply involving a rotating alternator capable of delivering 200 MJ at a 200 MW peak power level would cost approximately $10 M. The long pulse magnets, particularly utilizing existing fusion program power supplies, represent the least expensive entree into the field range between 30 T and 75 T. Realization of these ambitious pulse coils will represent a major magnet design challenge for the next few years. This program will pave the way for truely continuous fields in this range through technology exploration and through opening up significant physics opportunities.
t
0.5 I.O 1.5 T SECONDS Fig. 5. Field versus pulse time for 5 cm bore LN2-cooled long-pulse coils. The two curvesare for constant magnet weights and stored energies. The peak resistive powers required can be estimated by dividing the stored energy by the pulse time.
Acknowledgements
The author is indebted to Paul Schwartz for his comments on the future of superconducting magnets and to Robert Weggel for his comments concerning hybrid and long pulse systems.
D. Bruce Montgomery / Generation o f continuous fields
References [1 ] Natl. Res. Council, Am. Acad. Sci., Panel Rept. Research Opportunities in High Magnetic Fields, Chairman, Seymour Keller, October 1978. [2] D. Bruce Montgomery, Solenoid Magnet designs, (Wiley, New York, 1969) p. 127. [3] P. Carden and A.M. Collins, J. Phys. E, 7 (1974) 750. [4] P.S. Schwartz, W.D. M~kiewicz, C.H. Rosner, Physica 86-88B (1977) 1084.
299
[5] D. Dew-Hughes, "Superconducting Materials for LargeScale Applications", Adv. Cryogenic Eng. 22 (1975) 316. [6] M.J. Leupold, R. Weggel, Y. Iwasa, 6th Int. Magnet Technology Conf., Bratislava (1977). [7] F.A. Muller, L.W. Roeland and R. Gersdorf, Proc. 3rd Int. Conf. on Magnet Technology, Hamburg (May 1970). [8] H.C. Praddaude and S. Foner, Production of Quasistatic High Magnetic Fields by Switching Low Voltage DC Generators, IEEE Magnetics, September 1975. [9] S. Askenazy, G. Carrere and J. Marquez, Physique sous Champs Magnetiques Intense (CNRS Paris 1975) p.357.