WI

\

  

U

&

IIWIIWHH'IWWWIWIWWIHHHHM

 

3007

This is to certify that the
thesis entitled

RADIATION RESISTANT SUPERFERRIC MAGNETS FOR
FRAGMENT SEPARATORS

presented by

JONATHAN DAVID DELAUTER

has been accepted towards fulfillment
of the requirements for the

MS. degree in Physics

 

 

we

 

Major Professor’s Signature
I (J 1241 / C

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

LIBRARY
MiCI‘ligdl’. State
University

 

 

 

--—u—u_.-o-o-o-o-o-o-o-.-o-o--u—o-o-o-o-o-a-o-n-g-o-o-o-o-u-c-o-o—o-o—c-o—o-¢-¢-u-c-—---..

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/Date0ue.indd-p.1

RADIATION RESISTANT SUPERFERRIC MAGNETS FOR FRAGMENT
SEPARATORS

By

Jonathan David DeLauter

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Physics and Astronomy

2006

ABSTRACT

RADIATION RESISTANT SUPERFERRIC MAGNETS FOR FRAGMENT
SEPARATORS

By

Jonathan David DeLauter

Several proposed accelerator facilities will be capable of providing intense beams
on targets that are in close proximity to superconducting magnets. These magnets will
have to operate in high-radiation environments. This work examines a metal-oxide
insulated version of the standard CICC (Cable—in-Conduit Conductor) as a proposed
solution for radiation levels directly following the target, and examines a cyanate-ester as
a proposed solution for lower radiation areas in the separator. A small superferric dipole,
similar to one previously constructed with conventional epoxy-potted coils, has been
fabricated with metal-oxide CICC and, separately, with cyanate ester. Both technologies

have been tested and results are reported.

Copyright by
JONATHAN DAVID DELAUTER
2006

ACKNOWLEDGEMENTS

The author would like to thank A. Zeller for advice, instruction, help, patience,
providing unlimited opportunity and helping me follow it through; J. C. DeKamp for
instruction, patience, and for sharing his skills and experience; Chen-yu Gung and others
at MIT Plasma Science and Fusion Center for sharing their magnet facility and their help;
B. Sherrill and F. Marti for volunteering their time on my committee; D. Pendell for
welding help and expertise; J. Bierwagen, S. Bricker, and J. Gonzales for project advice;
J. Savory and D. Arthur Brown for labor and project advice; T. Hudson and others at BPS
Machine Shop for machining instruction and advice. Thanks to Mom, Dad, and Mike for

their support.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES .......................................................................................................... vii
1 INTRODUCTION ...................................................................................................... l
1.1 Radiation ............................................................................................................. 2
l .2 Cyanate-ester ....................................................................................................... 4
1.3 Metal-Oxide Insulated Cable-in-Conduit Conductor (MOCICC) ...................... 6
2 THE CYANATE-ESTER DIPOLE .......................................................................... 10
2.1 Design of the Cyanate-ester Dipole .................................................................. 10
2.2 Construction of the Cyanate-ester Dipole ......................................................... 17
2.3 Testing of the Cyanate-ester Dipole ................................................................. 21
2.4 Simulation of Quench in the Cyanate-ester Dipole .......................................... 23
2.5 Discussion of the Cyanate-ester Dipole ............................................................ 28
3 THE METAL OXIDE CICC DIPOLE ..................................................................... 30
3.1 Design of the MOCICC Dipole ........................................................................ 30
3.2 Constructing the metal-oxide CICC dipole magnet .......................................... 41
3.3 NSCL Test Results ............................................................................................ 48
3.4 October 2006 MIT Test Results ........................................................................ 50
3.5 Analysis: ........................................................................................................... 58
3.6 Further Research ............................................................................................... 6O
4 CONCLUSIONS ....................................................................................................... 62
5 APPENDICES .......................................................................................................... 63
5.1 Cyanate-Ester Winding Instructions ................................................................. 63
5.2 Cyanate-Ester Winding Photo Flow-Chart ....................................................... 65
5.3 Gandalf Input File for MOCICC Dipole; Operation in 4 W/m Heating ........... 66
6 BIBLIOGRAPHY ..................................................................................................... 67

LIST OF TABLES

Table 2-1: Wire, coil, and magnet parameters for the cyanate-ester dipole. .................... 12
Table 2-2: Problems encountered during the winding of the cyanate-ester coils ............. 19
Table 3-1: Magnet and Wire Parameters for the MOCICC Dipole .................................. 38

vi

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

LIST OF FIGURES

1-1: Diagram of a fragment separator. The fragment separator is composed mainly
of dipole and quadrupole magnets. In high radiation conditions the magnets that
succeed the target have the most exposure. Radiation levels drop significantly
further down the beamline 1n the separator due to the energy lost in beam dumps
through the dipole magnets. ........................................................................................ 8

1-2: Metal oxide insulated cable-in-conduit conductor. The outer and inner conduits
are 316 stainless steel, about 1 mm thick each. The white layer is MgO and is about
2 mm thick. The entire cross section is 1 cm x 1 cm. The inner conduit contains

strands of copper/NbTi. .............................................................................................. 9

2-1: This shows the QUENCH [Ref. 12] simulation of the quench voltages (for
various currents) in the coil during a quench. The insulation around the wire was
rated for 689 V (between two strands). These results demonstrate that the operating
current used, which was near 54 A, would not damage the coil through arcing if a

quench were to occur. ............................................................................................... 13
2-2: Examples of a Stycast® epoxy-potted coil (in back) and a CTD-422 cyanate-
ester—potted coil (in front). ........................................................................................ 14
2-3: The field lines calculated with POISSON [Ref. 13] for one symmetric quarter
of the cyanate-ester dipole. The units for the x and y axis are cm. .......................... 15
2-4: The short sample current limit guaranteed by the manufacturer (Supercon,

Shrewsbury, MA) is plotted along with the field (at the conductor) and required
current calculated by POISSON [Ref. 13]. Their intersection, about 50 amps, is the
estimated short sample current of the magnet [Ref. 4]. ............................................ 16

2-5: The complete magnet assembly hangs from the Dewar lid ready to be tested.
Visible are the lead connections to the Dewar leads, magnet iron, coils, G-lO
spacers, and voltage taps. Partially visible are the coils shims, solder-connected
leads, and solder-connected lead fixture. .................................................................. 20

2-6: Shown here is the ratio of applied current to short sample current (50 A) at the
time of quenching [Ref. 4]. ....................................................................................... 22

2-7: The current measured in the cyanate-ester dipole during a quench, matched to
simulation data from the code QUENCH [Ref. 12]. The best fits were for a
transverse heat propagation velocity ratio of around 1.4 % (g=0.014). Inductively
coupling the iron (denoted as ‘with iron’ above) in the simulation did not
significantly vary the result. Simulations fit to Stycast®-potted coils have a
transverse heat propagation velocity of about 7%. ................................................... 24

vii

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

2-8: The voltage measured in the cyanate-ester dipole during a quench, matched to
simulation data from the code QUENCH [Ref. 12]. The best fits were for a
transverse heat propagation velocity ratio of around 1.4 % (g=0.014). Inductively
coupling the iron (denoted as ‘with iron’ above) in the simulation did not
significantly vary the result. Voltages during the quench did not approach levels

that would damage the coils. ..................................................................................... 25
2-9: The first derivative was calculated for the measured current and for the
QUENCH-simulated current [Ref. 12]. The first derivative of current corresponds
well to the voltage measurements. ............................................................................ 26
2-10: The second derivative of current with respect to time for the quench data

measured, and for two simulations from QUENCH [Ref. 12]. Inductively coupling
the iron (denoted ‘with iron’) in the QUENCH [Ref. 12] input did not significantly
vary the result. ........................................................................................................... 27

3-1 continued: GANDALF-simulated [Ref. 18] time evolution and equilibrium of
pressure, massflow, and temperature. A 4 W/m (representative of the 150 W into
the first RIA quadrupole) radiation-heat load was used. The reason for the drop-off
at x=3.6 is because the conduit ends there; the liquid helium is exiting the conduit.

1) Liquid helium flow pressure reaches equilibrium around 4.6 atm (at a distance of
x=1.8 m down the length of the conduit). II) Liquid helium mass flow reaches
equilibrium at 4.87 kg/s (at a distance of x=1.8 m down the length of the conduit).
III) Conductor temperature along the length of the conduit at times 0.1 s (A), 1.0 s
(B), 4.0 s (C). The temperature profile does not change after t=4.0 5; curve c
represents the equilibrium temperature profile along the conduit. ........................... 35

3-2 continued: GANDALF [Ref. 18] simulation of the propagation of a quench. The
quench was induced by a 400 W/m, 0.01 3 heat pulse distributed from x=l.7 m to
x=1.8 m in the center of the 3.6 m conduit of the MOCICC dipole. The conduit has
a helium flow of 5 g/s at 5 atm. The quench lasts only 0.01 s, so the time on the
graph essentially represents the time following the quench. I) The power-supply
current was deactivated along an exponential decay, with time constant 0.1 s. H)
Voltage between the conductor and ground (the outer stainless steel jacket) peaks at
about 1 V in this quench; there is only about 2 k] of stored energy in the magnet.

III) The temperature at 0.013 (the end of the heat pulse) was not dangerous to the
magnet. ...................................................................................................................... 37

3-3: The short-sample current limit (for the individual wire strand) is represented by
the intersection (about 290 A at about 2.5 T maximum field on conductor) of the
circle points, POISSON-simulated [Ref. 13] fields for given currents, and the
diamond points, the critical current of the amount of NbTi material that is in the wire
used (NbTi data from an Otokompu table of NbTi material properties). Also shown
are the guaranteed, typical, and actual current-field specs provided by the
manufacturer, Supercon. ........................................................................................... 39

viii

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

Fig.

3-4: The magnetic field lines for the top-right quarter (symmetry) of the MOCICC
dipole cross-section, calculated by POISSON [Ref. 13], for 40 kAOtums. The x and
y axes represent the x and y directions and the units are cm. ................................... 40

3-5: A steel winding arm pivots inside the winding-form cylinders to wind the coil
into shape. The cylinders have a 2” outer diameter and are welded onto the plate
beneath them. During winding, tack welds were placed occasionally to maintain the
shape. ........................................................................................................................ 44

3-6: The MOCICC dipole hangs from the NSCL Dewar lid, ready to be tested. On
the left is the lead connection between the two coils. ............................................... 45

3-7: The joint that connects the coils in series is comprised of the solder wire bundle
surrounded by two clamshell halves made of an outer stainless steel jacket with a
Macor® insert to prevent the conductor from making contact with ground. Seen
here is a cutaway: one of the two clamshell halves. ................................................. 46

3-8: The completed lead connection joint is seen here, secured to the magnet iron by
a post. The two clamshell halves were welded in place around the soldered bundle.
................................................................................................................................... 47

3-9: The MOCICC dipole hangs under the NSCL dunking Dewar lid, ready to be
tested. The magnet leads have been connected to the Dewar leads. ........................ 49

3-10: MOCICC dipole magnet ready to be inserted into MIT’s Dewar. The large
Dewar leads are helium gas cooled. .......................................................................... 52

3-11: Close-up view of the lead connections. The short magnet lead on the right is
connected to the Dewar lead by some clamped Rutherford superconducting cable,
which quenched during the testing. Labels have been applied that correspond to the
diagram in Fig. 3-12. ................................................................................................. 53

3-12: Diagram of the test instrumentation inside the Dewar. V_1a and V_le are
voltage taps on the left and right (respectively) connections between the Dewar leads
and the magnet leads. V_lc and V_ld are voltage taps a nearly-co-wound coil for
measuring induced voltage. V_lb is a voltage tap to measure the magnet voltage.
The liquid helium level sensor instrumentation is labeled V_Za through V_2d. On
the right hand side of the diagram are the liquid helium levels needed to reach the
specified heights in the Dewar; a height between 40% and 50% was desired .......... 54

3-13: The instrumentation setup at MIT. Data were recorded by a Macintosh
computer (bottom left), a Yukagawa LR8100 8-channel data recorder (top), and
three digital voltmeters (top-left). ............................................................................. 55

3-14: Measurements during the ramp-up of the magnet (ramp #5 according to Fig.
3-15) from the voltage taps V1_a and V1_e (labeled according to Fig. 3-12), the

ix

bridge connections between the coil leads and the Dewar leads.) V1_e begins to
generate resistive voltage shortly before the quench (the quench occurred after 3.5 s)
and V1_a is very noisy during the entire ramp. The magnitude of the noise on V1_a
is much higher than the resistive voltage on V1_e. It is likely that the noise on V1_a
was generated by wire movement, and was the cause of the premature quench of the
magnet. The ramp rate for this test was 700 A18. The other tests, at various ramp
rates, showed similar results. .................................................................................... 56

Fig. 3-15: Quench history of the MOCICC dipole test at MIT. The magnet could not be
ramped to its full sustainable current due to problems with the connection between
the Dewar leads and the magnet. By increasing the ramp rate, the magnet current
could be pushed farther before heat propagating from the quench in the lead
connection would initiate a quench in the magnet. Error in the measurement due to
sampling rate is shown with thick, solid error bars. For the last four measurements,
the measurement was read from voltmeters, which did not agree with the
measurement read by the computer. This discrepancy is represented by light, dotted
error bars. .................................................................................................................. 57

1 INTRODUCTION

High intensity accelerators like GSI (Gesellschaft f'Lir Schwerionen-forschung,
Darmstadt, Germany) [Ref. 1], and the proposed Isotope Science Facility (ISF), a
revision of the Rare Isotope Accelerator (RIA) [Ref. 2], require superconducting magnets
to operate in high radiation environments. The radiation leads to problems with neutron
heating and long-term radiation damage, which existing superconducting magnet designs
cannot address. A high radiation environment is the target-in area in front of the
fragment separator, which consists mainly of dipole, quadrupole, and sextupole magnets
(Fig. 1-1). The magnets function to bend and focus the trajectory of the beam. Fragment
separator magnets are superferric (iron-dominated and superconducting) due to needs for
high magnetic field magnitude, uniformity, and aperture. This work examines cyanate-
ester and metal-oxide insulated cable in conduit conductor (CICC) technologies as

solutions for developing radiation tolerant (radiation resistant) magnets.

1.1 Radiation

The radiation problem can be divided into nuclear heating and long-term radiation
damage. Radiation exiting the target can be composed of ions, neutrons, electrons,
muons, and gamma rays. Magnet coils can be shielded from gamma rays and ions, but
shielding is not effective for neutrons. The flux of neutrons dissipates energy in the form
of heat into the material. The energy and heat distributed depends on the energy of the
neutrons and the mass of the material. For this reason the cold mass, the mass of material
that is being cooled by the liquid helium, must be kept as small as possible.

The first quadrupole coil in the triplet that follows the target in the RIA fragment
separator has a calculated dose rate, for some combination of beam and target, of 1 Gy/s
from a 400 kW heavy ion beam [Ref. 3]. With an assumed operation time per year of 107
5 (about 3000 hr), there is a yearly dose of 107 Gy (10 MGy). A sextupole magnet
precedes the first quadrupole triplet, and another triplet follows (Fig. l-l), all of these
will be subject to similarly high radiation levels and need an inorganic radiation resistant
technology [Ref. 4]. Beyond these magnets and after the beam dump in the first dipole,
the beam power drops to at most 10 kW. It is in this area that the cyanate-ester would be
applicable. 10 kW is only a fraction of the original 400 kW beam, but organic materials
have a factor of 200 more radiation sensitivity than inorganic [Ref. 5]. Thus, the organic
radiation dose expected in these areas is at most 5 MGy, making it suitable for the
cyanate-ester material.

081 has a similar configuration, with the expected dosage in the first coil of 14

MGy/yr, or a total dose of 280 MGy [Ref. 6]. The later magnets in the G81 fragment

separator, for reasons comparable to RIA, are likely to have radiation levels low enough
for the use of cyanate-ester.

Over a long period of time (one to many years), depending on material and
exposure rate, radiation can damage coil materials. Organic materials are especially
sensitive to this radiation because they are composed of molecular chains, which have
low binding energies (compared to a metal lattice for example) and can be split easily by
radiation [Ref. 7]. Hydrogen bonds are especially sensitive to this type of radiation and
cyanate esters generally have about half as many hydrogen bonds as normal epoxies.
Damage due to radiation can radically change the physical properties of these materials,
and in practice leads to irreparable weaknesses. Strong Lorentz forces can exacerbate
these weaknesses, and shorting or quenching of a coil can occur through failure of the

epoxy or electrical insulation.

1.2 Cyanate-ester

The majority of the beam energy is dumped into the first few magnets in the
fragment separator, and past these magnets the radiation levels are much lower, although
still hazardous to epoxies normally used for magnets. In these areas a radiation resistant
organic compound called cyanate—ester may be suitable [Ref. 5]. The cyanate-ester has
radiation tolerance much greater than standard epoxy, with equivalent compression
strength, but lower shear strength [Ref. 8]. It was proposed as the solution for the lesser
radiation areas at GSI and RIA. A cyanate-ester, CTD-422, formulated by CTD
(Composite Technology Development, Lafayette, CO), for use in fusion research, was
used to wind a radiation resistant dipole for the purpose of researching the use of cyanate-
ester for coil winding.

CTD’s 400 series cyanate-ester has been irradiated and compared with standard
epoxies [Ref. 8]. It was found that the compressive strength of epoxies starts to break
down above 2.0*1022 neutrons*m‘2, whereas the compressive strength of the cyanate-
esters tested did not show significant degradation up to 5.0"‘1023 neutrons*m'2. The
cyanate-ester’s shear strength is comparable to epoxy and degrades similarly in radiation.
Thus, cyanate-ester can be used as a more radiation-resistant substitute for epoxy in
designs that reduce shear and keep the insulation in compression [Ref. 7].

Factors especially important to consider in the design of an epoxy or cyanate-ester
magnet are the stored energy, transverse heat propagation velocity, thermal conductivity,
quench voltages and other factors that affect quenching. For the cyanate-ester magnet,

the radiation tolerance of the materials was an important consideration.

A small superconducting magnet must be capable of releasing its stored energy
before the coil voltage rises to a dangerous point where it can are between turns, shorting
and destroying the coil. The transverse heat propagation velocity is important because it
is the primary factor in how quickly a local quench will spread through the coil. When a
local quench occurs it is important to quench the rest of the coil as quickly as possible so

that high voltages do not build up at the local quench location.

1.3 Metal-Oxide Insulated Cable-in-Conduit Conductor (MOCICQ

The proposed technology uses cable in conduit conductor (CICC) technology with
a metal-oxide ceramic as an intertum insulator, an example of which can be found in Fig.
1-2. Metal-oxide-insulated CICC (MOCICC) have the advantage that all of the materials
used (see material radiation tolerance table) have radiation tolerances greater than or
equal to that of the superconductor (about 50108 Gy). These coils are expected to operate
in the high radiation environments of the fragment separator.

There are few technologies with radiation tolerance on the order of 108 Gy, and of
them only the resistive mineral-insulated-conduit (MIC) is scalable. Research done on
internally anodized aluminum MOCICC found that current densities of up to 70 A/mm2
were achieved and it was estimated that current densities of up to 100 A/mm2 were
possible with more R&D [Ref. 5]. However, those coils were small and no way has been
found to scale the technology to larger magnets. The first quadrupole in the separator
receives radiation as high as 2 Gy per second [Ref. 5].

Resistive mineral—insulated-conduit (MIC) magnets are well established and have
been used in PSI and LANL [Ref. 9, Ref. 10]. This technology has been used for 30
years, but has low current densities on the order of 2 A/mmz. The low current densities
make the resistive version of this technology suitable only for where high current density
is not required, like dipole magnets [Ref. 3]. Quadrupole magnets, which need high
current density, will require a different technology. Superconducting MOCICC has
current densities of 43 A/mm2 established, and about 100 A/mm2 expected. This is one
to two orders of magnitude better current density than the MIC, with similar radiation

tolerance.

Metal oxide CICC allows forced—flow of liquid helium through and among the
conductor strands, which results in excellent conductive and convective cooling and heat
removal. In addition, MOCICC is easy to incorporate in a warm-iron design because it
does not need an external helium vessel. A warm-iron design cryogenically/thermally
separates the coil mass from the cold mass such that the liquid helium only cools the coil
mass. The warm—iron design is optimal because the iron absorbs 10 kW of radiation heat
regardless of the temperature; trying to remove this much heat at 4K is prohibitively
expensive.

This MOCICC technology is scalable, the insulation and stainless steel inner and
outer jackets occupy a large fraction of the CICC cross section in this work, leaving only
a small fraction for the conductor, but higher current densities can be achieved by
increasing the size of the overall cross section without increasing the thickness of the
steel or MgO layers. This would provide a higher ratio of conductor to cross sectional
area, but it would increase the operating current as well. Higher operating current
requires more refrigeration; Le. 10 kA current leads require 30 Uhr of liquid helium, 50
kA leads require 150 Uhr of liquid helium, and these heat loads would be present even
when the beam was off [Ref. 11]. An operating current limit of 10 kA was preferred for

this work, so the cross section was limited to about 10 mm x 10 mm.

Quadrupole Dipole

  

   

it

 
  
  

 

Sextupole .
Extremely High \\
Radiation Area Very High X
Radiation Area High

Radiation Area

Fig. 1-1: Diagram of a fragment separator. The fragment separator is composed mainly of dipole
and quadrupole magnets. In high radiation conditions the magnets that succeed the target have the
most exposure. Radiation levels drop significantly further down the beamline in the separator due to
the energy lost in beam dumps through the dipole magnets.

 

Fig. 1-2: Metal oxide insulated cable-in-conduit conductor. The outer and inner conduits are 316
stainless steel, about 1 mm thick each. The white layer is MgO and is about 2 mm thick. The entire
cross section is 1 cm x 1 cm. The inner conduit contains strands of copper/NbTi.

2 THE CYANATE-ESTER DIPOLE
2.1 Design of the Cyanate-ester Dipole

The standard method for building superconducting fragment separator magnets
used at NSCL is to wind several thousand turns of superconducting wire and bind them
with an high-strength epoxy like Stycast® (Emerson & Cuming, Billerica, MA). This
method of building coils is cost-effective, fast, leads to high current density, and is well
understood. However, even the lower radiation areas at GSI and RIA are too hazardous
for coils made with this epoxy. In this work a radiation resistant cyanate-ester from
CTD’s 400 series, specifically CTD-422, was used to pot two 2200-tum dipole coils; the
dimensions for the coils and wire can be found in Table 2-1. CTD’s 400 series was
chosen because it maintains high compression-strength up to 50 MGy. The wire
insulation was polyimide and also had high radiation tolerance of 108 Gy [Ref. 8].

Prior to construction, simulations were done using Wilson’s code: QUENCH
[Ref. 12] in order to ensure the safety of the coils during operation (Fig. 2—1). The coil
was random wound and the wire insulation was rated for 500 V per 0.0254 mm (0.001
in.). The thickness of insulation between two strands was 0.035 mm, which translated to
a 689 V insulation strength. Quench voltages exceeding this could damage the coils;
with a safety factor included, quench voltages below 600 V were sought.

The size and shape of the cyanate-ester coil were derived from the two small,
superferric dipoles used at NSCL to steer the beam just outside the K1200 cyclotron. An
example of both the cyanate-ester and the Stycast® coil are shown in Fig. 2-2. Winding
methods and fixtures were already available, and the cyanate-ester coil could easily be

compared to an identical NSCL epoxy coil [Ref. 4].

10

The field from the coil was simulated using the code POISSON [Ref. 13] (Fig.
2-3). Using various current inputs the field magnitude was calculated using POISSON.
A plot of the magnetic fields and the applied current necessary to generate them is shown
in Fig. 2-4. Above 2.5 T (highest conductor field) the iron saturated and the current vs.
field curve raised much slower (i.e. doubling the current produced only 0.7 T more field).
This curve intersected the manufacturer guaranteed short sample current curve at about
50 amps, indicating that 50 amps was the short sample current for the magnet. Because
applying more current would not achieve an appreciably higher field, and because
increasing the amount of copper would improve the coil’s safety during a quench, the

copper to superconductor ratio was set at 7:1.

11

Table 2-1: Wire, coil, and magnet parameters for the cyanate-ester dipole.

 

 

System Parameters

Wire 0.445 mm diameter (bare)
0.48 mm diameter (insulated)
7:1 Cu:SC

Coil 2200 turns
25.4 x 25.4 mm cross section

Magnet 203 mm long iron

4.29 H at 51 A
5.6 kJ stored enerov at 51 A

12

Comparison of Simulated Quench Voltages for Various
Applied Currents in the Cyanate Ester Dipole

 

 

 

 

 

   
   

 

 

 

 

 

 
 
     
   

 

 

  
  

       

 

1200
1000~~———~ ———e~~————e——————-*—~-————e—~-~
<>27amp
Cl 45amp
800 -54amp
9 X90amp
3’ +108amp
53> 600 __ rm— -—--—»—————-~ ~ — ——- -~~—-~
'6
>
400-» e*-~ev~e »e
200 _-.ullllllltllllllllllll!’llllllIllllllltlluhfi..." - _ T— -
59! X I """Hllm
+ 00000 :si'i'ullc!"-=sm.u.
O t..'?i§:::e e 9 e o O 9 00°®0§0wo . U
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time (s)

Fig. 2-1: This shows the QUENCH [Ref. 12] simulation of the quench voltages (for various currents)
in the coil during a quench. The insulation around the wire was rated for 689 V (between two

strands). These results demonstrate that the operating current used, which was near 54 A, would not
damage the coil through arcing if a quench were to occur.

13

 

Examples of a Stycast® epoxy-potted coil (in back) and a CTD-422 cyanate-ester-potted coil

(in front).

Fig. 2.2

14

O—l

 

Short Sample for Cyanate-Ester Dipole

 

 

 

 

 

 

 

 

 

 

 

225
200 ~ 7 eeeeeeeeee F
I/’
175x15- 7 7 7/7 77~ 5,,ch 7 ~
,/
1 50 ,‘”‘—“’“_”‘ + Manufacturer "
/' Guaranteed Limit

A 125 it —I—-applied current ~
g /
T 100 — 77

75 «

50 «

25 ~ 7-7 7

0 I I I I I I I I I I I I I I I I l I
O 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5

B(T)

Fig. 2-4: The short sample current limit guaranteed by the manufacturer (Supercon, Shrewsbury,
MA) is plotted along with the field (at the conductor) and required current calculated by POISSON
[Ref. 13]. Their intersection, about 50 amps, is the estimated short sample current of the magnet

[Ref. 4].

16

2.2 Construction of the Cyanate—ester Dipole

Magnet construction began with development of a coil winding method.
Superconducting NbTi wire with 0.48mm diameter was wrapped around a winding form
to build the coils; Table 2-1 lists the wire, coil, and magnet parameters in detail. The
wire was drawn through a bath filled with warm CTD-422 as it was being wound.
Difficulties encountered during the winding process led to electrical shorts inside the coil
and other problems (Table 2-2). More detail about this process can be found in Appendix
5.1 and 5.2 (winding procedure and flow diagram). After exactly 2200 turns had been
wound, fiberglass tape was cut to fit and then wrapped around the coil in the winding
form. Form bars were then clamped to the form and the entire assembly was baked in a
computer controlled pizza oven at CTD’s recommended heat cycle.

The finished coils were installed into the magnet iron and shimmed in with
Kapton® (polyimide) tape. Sanding and filing the iron eliminated the sharp edges that
remained from the machining process. G-lO spacers were added in between the coil ends
and the iron (Fig. 2-5). The spacers served to hold the coils in position when the magnet
was not energized; once the magnet was energized, the field forced the coil toward the
iron, and friction then immobilized the coil.

The coil leads were soldered together such that the two coils were electrically in
series. The soldered leads were then fixed in place by clamping them to a post welded to
the iron (Fig. 2-5). To prepare for the testing, voltage taps were also soldered to the
leads.

In comparison to Stycast®, CTD-422 exhibited much greater workability and

ease of use. After the CTD-422 had cured, repairs could be made by applying the

17

cyanate-ester over the area that needed repair, and then curing again; the bond that the
cyanate-ester made to itself in repairs like this was better than the bond that Stycast®
made in similar repairs. CTD-422 may need a filler to improve the viscosity because the
edges of the coils were difficult to cure correctly. The solution for this problem used for

the dipole coils was to wrap fiberglass tape around the coil before curing it.

18

Table 2-2:

Problems encountered during the winding of the cyanate-ester coils

 

Problem

Solution

 

Loose turn on
inner face of coil

i. Repair by applying fiberglass
tape, painting with cyante-ester,
and curing again.

ii. Before starting the wind, wrap a
layer of fiberglass tape around
the inside of the winding form.

 

Loose turn on
outer face of coil

Machine better-fitting form bars

 

Rough Edges

Wrap Teflon®-coated strips around
comers to give the coils a smooth and
round shape

 

 

Electrical short
found near end
of coil

 

The lead wires are sensitive to damage,
check for abrasion, cracks, or
looseness/bad bonding.

 

l9

 

-L.:..I 1
"mm
. - l

“I lllfil \
~ n3:

500 amp Dewar
Leads

\u-n.

Coil lead to

I l . 7 "
Coil—> - _ Dewar lead

lam.
I

. Connection

uh

tum '

Series lead

connection —.

Lead
connection
fixture

manner

I

”pl!

5'

Atari M

Hr I
A “kriubll

I

am t'btt’i'misj

G-10
spacer

N Voltage tap

 

Fig. 2-5: The complete magnet assembly hangs from the Dewar lid ready to be tested. Visible are the
lead connections to the Dewar leads, magnet iron, coils, G-10 spacers, and voltage taps. Partially
visible are the coils shims, solder-connected leads, and solder-connected lead fixture.

20

2.3 Testing of the Cyanate-ester Dipole

The magnet was suspended from the lid of NSCL’s dunking Dewar. The coil leads
were connected to Dewar leads rated for 500 amps. An NSCL 100 A power supply was
connected to the Dewar leads through a calibrated shunt that made accurate
measurements of current possible. The resistance of the shunt was known, so that the
current could be determined from the voltage. The Dewar was evacuated with a vacuum
pump and then flushed with helium gas (called pumping and purging) in order to clean it.
Liquid helium was then pumped in until the entire magnet was immersed.

During testing of the magnet the current, ramping voltage, and voltage across
each coil were recorded and the results have been reported in [Ref. 4]. From the current
measurements the quench history was tracked and graphed in Fig. 2-6. The magnet was
able to exceed the guaranteed short-sample current on several occasions because wire
manufacturers generally include a safety margin in their estimations.

The first three ramps were at 8 V and then the following ramps were at 3 V.
Ramps ten and eleven were again at the full 8 V, and these showed some ramp rate
dependence. While standard NSCL magnets do not exhibit ramp rate dependence, the
room-temperature thermal conductivity of CTD-422 is only 2/3 that of Stycast®. If the
difference in thermal conductivity also exists at 4 K, then the ramp rate dependence can
be explained. The higher ramp-rate led to higher eddy-current losses, which created
more heat [Ref. 4]. Poor thermal conductivity allowed heat buildup, which led to a

quench.

21

Quench History of the Cyanate Ester Dipole Test

 

 

 

80 _, L _- __ , , , , , , ELL”, .,, ,-,-Lm____L_~
+l/Ic percent

 

 

 

 

 

Percent of short sample
0)
O
l
l
l

 

 

 

40— - e +e+~-e
20~ --s--- ++e+~~
o r T

o 5 1o 15 20

Quench number

Fig. 2-6: Shown here is the ratio of applied current to short sample current (50 A) at the time of
quenching [Ref. 4].

22

2.4 Simulation of Quench in the Cyanate—ester Dipole

The data measured in the test experiment was compared to data simulated by
QUENCH [Ref. 12]. QUENCH simulates the voltage, current, temperature, stored
energy, and normal volume and has given reasonable comparisons to previous magnet
systems [Ref. 14].

Estimations of the transverse heat propagation velocity and the effects of iron
coupling were obtained by comparing the simulation data to the experimental; the results
were presented in [Ref. 4]. Fig. 2-7 shows current as a function of time and it is clear that
simulations with transverse propagation velocity ratios near g=0.015 (1.5 %) are the best
fits to the experimental current. This is considerably lower than Stycast®, which
normally fits well with a g=0.07 (7%) transverse propagation velocity ratio [Ref. 4, Ref.
14].

Voltage simulations and measurements are recorded in Fig. 2-8. Again, the
simulated 1.3% and 1.5% transverse heat propagation ratios are the best fits, and again
there is little difference whether the iron is taken into account in the simulation. The first
derivative of current was calculated, and matches well with the recordings for voltage
(Fi g. 2-9). Finally, the second time derivative of current is plotted in Fig. 2-10. The
second derivative of current is used to check the agreement of the first derivative of
current with the voltage; here, the inflection points and slope equal zero points are
accurate. In some of these simulations the better fit is with inductively coupled iron, but
the differences are not significant enough to say that the iron changes the pr0perties of

quench propagation [Ref. 4].

23

Current During Quench of the Cyanate-ester Dipole

 

 

 
 
 
   

 

 

 

 

 

 

 

 

 

60
El 1coilg=.013
0 With Iron, g=.015
5° ' + 1coil g=0.1
A 1coilg=.05
A 40 Measured e
O.
E
s.
E 30 - - e i e e 4
2
{=3
20 .. ~ a 7~
10 ~ a 7~_
+A +A Q
0 I I I
0 0.5 1 1.5 2

Time (s)

Fig. 2-7: The current measured in the cyanate-ester dipole during a quench, matched to simulation
data from the code QUENCH [Ref. 12]. The best fits were for a transverse heat propagation velocity
ratio of around 1.4 % (g=0.014). Inductively coupling the iron (denoted as ‘with iron’ above) in the

simulation did not significantly vary the result. Simulations fit to Stycast®-potted coils have a
transverse heat propagation velocity of about 7%.

24

Normalized Voltage in Cyanate-ester Dipole Quench

 

 

  
  
 
 

 

 

 

 

 

 

 

 

120
- ch1:Vnorm
Elf coil
100 _ OWIth Iron
+1coil g=0.1
- , Alcoilg=.05
80 ~7 — ‘ - 7- - L»--.
E
E; 60 «7» -—~«
_0. b
>
40 - —----——
a
20 ~ ‘ .043.-- ~
+6
0 . 0
0 0.5 1 1.5 2

Time (s)

Fig. 2-8: The voltage measured in the cyanate-ester dipole during a quench, matched to simulation
data from the code QUENCH [Ref. 12]. The best fits were for a transverse heat propagation velocity
ratio of around 1.4 % (g=0.014). Inductively coupling the iron (denoted as ‘with iron’ above) in the
simulation did not significantly vary the result. Voltages during the quench did not approach levels
that would damage the coils.

25

First Derivative of Current for Cyanate-ester Dipole

 

 

 

 

 

 

 

 

 

 

 

 

60
<1) T.5
,E 50 “ i i ’ 'i i ’i T WM‘C‘Q. * T T T “ “
*-: r u c
5404 H -5- mm .2, a:'..\ -22-
3 I; \‘
‘g 1.1 can
._ g a V i .. ,,_. 7'1 ”A-.. A v i ‘ t I..‘ . g g _
5 30 . (I, m
0 r' 1
‘8 " 03211
(D204 -L_/Jcc--z-, ~~ "’-~~e
.2 . ’ g
g 10 ,j ,1, -O—Measured ..
<1) r'hJ '.
O :7. .
27$ mica —8 -Single g=0.013 '
-‘—' 0 E (3:33" ‘-
u. Ll
- 13 'With Iron g=0.015
-10
0 0.5 1 1.5

Time (s)

Fig. 2-9: The first derivative was calculated for the measured current and for the QUENCH-
simulated current [Ref. 12]. The first derivative of current corresponds well to the voltage
measurements.

26

150

 

 

la \ —0— Measured

I I
100 ' l‘ m —D—Single Coil

- 0- 'Withlron g=0.015
,\z___- - It I - be I -, - I

 

l

  
    

 

 

 

507~ 7* ~

 

 

ll

Second Derivative of Current w.r.t. time

 

 

-100

 

0 0.5 1 1.5
Time (s)

Fig. 2-10: The second derivative of current with respect to time for the quench data measured, and
for two simulations from QUENCH [Ref. 12]. Inductively coupling the iron (denoted ‘with iron’) in
the QUENCH [Ref. 12] input did not significantly vary the result.

27

2.5 Discussion of the Cyanate-ester Dipole

The CTD-422 cyanate-ester is suitable for application in the RIA fragment
separator. Its advantages are its workability and higher radiation resistance than epoxy;
however, high heat loads due to radiation and protecting the coil in case of a quench are
of concern.

The low transverse quench propagation velocity exhibited by CTD-422 lessens its
ability to produce self-protecting coils with large stored energies. An alternative active
quench protection system is difficult to sustain in a high radiation environment due to the
relatively low radiation tolerance of electronics. Changing the types and amounts of
fillers in the cyanate-ester may improve the transverse quench propagation velocity [Ref.
4].

The low thermal conductivity of CTD-422 (compared to Stycast®) could cause
problems for the stability of the superconductor. The heat generated by radiation
absorption in the coil is a problem because in potted coils the conductor does not have
good thermal contact with the surrounding liquid helium. The heat deposited in the
conductor is approximately 2 mW/g, which is near the stability limit of NbTi . To
consider the coil safe from quenching from heat, it should be used in areas with at most 1
mW/g. Therefore the coils should be placed in locations where the radiation dosage will
be limited to 50 MGy or less [Ref. 4]. Also, the compression-strength of the cyanate-
ester begins to significantly degrade above 50 MGy [Ref. 8]

The radiation tolerance of the magnet system has not been tested directly. The
CTD-422 has been tested by itself [Ref. 15] and exhibits significant reductions in strength

at dosages above 50 MGy. The radiation tolerances of the conductor are known and are

28

significantly higher, about 50108 Gy [Ref. 16]. Because it is radiation resistant,
irradiation of the magnet system as a whole is impractical; long irradiation times are
necessary in order to notice any damage in these materials. The neutron radiation from
an irradiation reactor that might be used would have a peak energy of 10 MeV; however,
the average neutron energy at RIA would be above 20 MeV. In addition, because there is
copper in the coils, the induced radiation may lead to problems with radioactive

waste/hazard [Ref. 4].

29

3 THE METAL OXIDE CICC DIPOLE
3.1 Design of the MOCICC Dipole

A metal-oxide CICC dipole magnet was fabricated and tested to examine the
operating properties. The primary objectives in the design of this magnet were current
density and radiation tolerance.

The relationship between the amount of cold mass and absorbed nuclear heating
suggests thermal separation of the coils from the iron; however, the cold-iron design is
easier to fabricate for testing. For this work, a cold—iron model was developed for
practical reasons; however, all development on the cold-iron model applies directly to the
warm-iron model development. Forced-flow liquid helium through the inner conduit
keeps the cooling central to the superconductor, and vacuum outside the coils thermally
insulates them from the iron. The use of forced-flow cooling allows large amounts of
heat to be convectively removed from the coils. Flow in the inner conduit would be both
around the bundle and among the strands, creating a large surface area for conductive
heat removal. Flowing liquid helium cools so effectively that localized heat spikes can
be contained and cooled before they propagate to cause quenching.

The calculated heat load for the first quadrupole in the RIA fragment separator is
150 W in the coil mass [Ref. 17]. This heat load was simulated for the MOCICC dipole
with the code GANDALF [Ref. 18], which has produced accurate simulations for other
magnet systems [Ref. 19]. The simulation showed that with an input pressure of 5 atm, a
mass flow of 5 g/s, and a 4 W/m heat load uniformly distributed along the 3.7 m length of

conduit, the temperature rose 0.8 K [Fi g. 3-1].

30

In order to simulate the quench behavior of the MOCICC dipole, a 0.01 s, 400
W/m heat pulse was input, using GANDALF, to the conduit in the center of the 3.7 m
length. The resultant quench voltage was low (Fig. 3-2) and would not be damaging to
the coil.

An increased current density could have been achieved by increasing the amount
of conductor in the conduit. However, radiation heat load is proportional to density of
the material, so trading liquid helium for conductor would increase the density within the
conduit, and consequently increase the heat load. The ratio of the cross sectional area of
conductor to void in the inner conduit is called ‘fill factor’. The fill factor was set at 44%
because it was difficult to insert more conductor into the conduit without special tooling,
and because it represented a convenient compromise between current density and
radiation heat absorbance.

The relatively small amount of stored energy in the magnet at 10 kA was 2040 J,
which allowed a copper to superconductor ratio of only 2:1. Wire, coil, and magnet
dimensions and parameters can be found in Table 3-1. The short-sample current limit for
the wire used is about 290 A/wire or about 12 kA for the bundle.

The nominal maximum operating current for the magnet is estimated to be 10 kA
because the magnet cannot operate at short sample, and because movement of the wires
and subsequent heat generation are unknown for this 44% fill factor. It is important to
demonstrate that MOCICC conduit with 44% fill can handle 10 kA because the required
current to achieve 2.5 T field at the poletip of the first RIA quad was calculated to be 10

kA [Ref. 20].

31

The CICC conduit, manufactured by Tyco Thermal Controls (Trenton, Ontario
Canada), was composed of outer and inner 316-stainless-steel square tubing, with a layer
of dense MgO powder between them as the insulator (Fig. 1-2). The conduit
manufacturing process is proprietary, but presumably the volume between the tubes was
packed with MgO powder and then all were drawn into the small (1 cm) conductor
dimension seen in Fig. 1-2.

MgO was chosen as the insulator because, as a ceramic, it has high radiation
tolerance, and is effective as an electrical and thermal insulator [Ref. 21]. The dielectric
strength of the MgO was tested by attaching a high potential generating machine, or
Hipotter, to the CICC conduit; the positive lead was attached to the inner conductor, and
the negative lead to the outer jacket. Voltage breakdown occurred at 5 kV, but the arcing
occurred across the ends of the conduit (through air), and there was no evidence of arcing
internally through the MgO. The MgO is hygroscopic, so absorbance of water from air
and from condensate after cold-shockin g, can degrade the dielectric at room temperature.
This may cause problems in resistive magnets, but it does not affect operation at liquid
helium temperatures. Quench voltages are not expected to be over 1 kV, so 5 kV is a safe
margin as long as the ends are well insulated. Breakdown voltage was not tested after the
coils were assembled and connected in series, and all breakdown tests were done in air,
not liquid helium.

For financial and practical reasons the same iron yoke was used for the cyanate-
ester magnet and the MOCICC magnet. The available space in the iron allowed only four

turns in the coils. Another limitation in the number of turns was the lengths of conduit

32

available from the manufacturer. Only lengths of about 4 m were available for our
developmental work. At larger production scales, longer lengths would be available.
Using the code POISSON [Ref. 13], the maximum field on the coil for 10 kA of
current was calculated to be 2 T. The forces on the coil (per unit of depth) were also
calculated to be 6.19*103 N/m in the x-direction and 3.76*104 N/m in the y-direction
(axes and direction are defined in Fig. 3-4). These forces are supported in the coil by

welding adjacent turns together.

33

 

 

 

 

 

 

 

   

 

 

 

 

m —fi‘I‘-LIIIIIFFIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIHIII.‘
O — ._
+ T. :
m L2? _.
«O l: _.
O H ..
\03‘ —‘
A at _. 2
t6 3‘ :
C1: _, z
,3 i. _._.
:3 i T
U) '7' _.
m i 1
a - —
O I —l
C“ mE :
Ln :- —*
er I 2
: X: 1.8000E+OO —
"LllLllllLlllILIIJLLIIIIIIlllllllLlIllLILLl1111111“
O 2 4
t (s)
O
O
m0 FIIIIIIIITI—II—IFIIIIIIIIIIIIIUIIIIIIIIIIIIIIIIIII _‘
O‘RE _
' X: 1.8000E+00 -«
ta 4 a
A l— —
U) C - ‘-
UI 0‘ '" “
M - r- -j
v qr — .1
3 :_ _:
o _ a
r-I _ \ g
in o _ g
.— —I
m g _ .2
U)
0 i" cut
{5 q. — —t
2': ~ 2
l— ”-0.”; _
4111111111mrMT’i'I-liairirirrlriirmrLLLLirmiLrirr
0 2 4
t (S)

Fig. 3-1: GANDALF-simulated [Ref. 18] normal operation of the MOCICC dipole in a 4 W/m
(representative of the 150 W into the first RIA quadrupole) radiation-heat load.

34

 

t—IIIIIIIIIIIIIIIIIIIIIIIIIII]IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIILH’K
I-— —l

III

 

A P
5d .
V <f- Time (s)
A 100015—01
B 1.000E+00 ave-'3’"

/Bf

c 4.000E+00

T conductor
450

 

)JlllllLLlLlLl/IIIIIIIIIILL

 

IIIIITIIIIIIIIIIIIIII[IIIIII

 

 

 

V3 11 n 11
(\l H H H
5" IllllllllllllllllllllllllllllllllllllllllllllllilllllllLlllllJllllLll
0.0 1.0 2.0 3.0
x (m)

Fig. 3-1 continued: GANDALF-simulated [Ref. 18] time evolution and equilibrium of pressure,
massflow, and temperature. A 4 W/m (representative of the 150 W into the first RIA quadrupole)
radiation-heat load was used. The reason for the drop-off at x=3.6 is because the conduit ends there;
the liquid helium is exiting the conduit. I) Liquid helium flow pressure reaches equilibrium around
4.6 atm (at a distance of x=1.8 m down the length of the conduit). II) Liquid helium mass flow
reaches equilibrium at 4.87 kg/s (at a distance of x=1.8 m down the length of the conduit). III)
Conductor temperature along the length of the conduit at times 0.1 s (A), 1.0 s (B), 4.0 s (C). The
temperature profile does not change after t=4.0 5; curve c represents the equilibrium temperature
profile along the conduit.

35

.00

 

 

 

 

 

 

 

 

q. _IIIIIIIIIIIIIIITIII[IIIIIIIIIIIIIIIIIIIIIIIIIIII_,
OH: :
[:5 — Flow Path 4
i. It: ‘

17.— .T.
I Z
a 3 3
u 8:1 2
8 3 ‘_'
3., OJ, 7
H :l I
:3 ,_I
U :l :
._j _
I——‘ —
l—I -—<
i—I —4
I—[ —<
_ 1
: :
<:>”iii«LLrirlrilirrirrliiiriirrilrirLLririliiiririrr1
oo 2 4
t (s)
g:fiTrirlilrlrIIIIIrrrflrillfllrjifirirlrlFIIIIIIIITL
H; Flow Path :
._.’ .4
a II :
> I :
(1)0 I I
am i .

4.10". _"

H I
O 7i

> I‘.

*Q —

ET!

,_ I“ _

t l.

: I.
OL- '\ __
O“iiirmuirrrrrrirrlrirriirrrliriiriiirlirrrrrrri
OO 2 4

t (s)

Fig. 3-2: GANDALF [Ref . 18] simulation of the propagation of a quench in the 3.7 m conduit of the
MOCICC dipole.

36

 

 

 

 

JIIIIIIIIIIIIIIIIIIIIIIIIllllllll InIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
r- ' —d
O : . '1. :
o — f I _
.—. I i :
._ o ' _.
_ 3 I _
— I '- —
A — III ‘ I -
a: : ,l . :
a... _ . I ..
b l j _
$4 I l , I
o — i j —
4.1 Z l . I
g : I I I
T, O :_ I l .-
Fi If) - I .—
7: - r’ I e
U : I l 2
U — I ..
: A .1. I
I" “ j \A ‘
:-_ 'I-I'.’ \‘\ T.
_ -s-g \. _
E /.‘/I X :
-— f/’ \r .-
llLLl-«PPI‘ITIIIIIIIIIIJIIIllllllllllllllllllllllllllllll iHJ—lLLlllllllLll"
0.0 1.0 2.0 3.0

x (m)

Fig. 3-2 continued: GANDALF [Ref. 18] simulation of the propagation of a quench. The quench was
induced by a 400 W/m, 0.01 5 heat pulse distributed from x=1.7 m to x=1.8 m in the center of the 3.6
m conduit of the MOCICC dipole. The conduit has a helium flow of 5 g/s at 5 atm. The quench lasts
only 0.01 s, so the time on the graph essentially represents the time following the quench. I) The
power-supply current was deactivated along an exponential decay, with time constant 0.1 5. II)
Voltage between the conductor and ground (the outer stainless steel jacket) peaks at about 1 V in this
quench; there is only about 2 Id of stored energy in the magnet. III) The temperature at 0.015 (the
end of the heat pulse) was not dangerous to the magnet.

37

Table 3-1: Magnet and Wire Parameters for the MOCICC Dipole

 

 

System Parameters
Wire 0.5 mm diameter (bare)
2:1 copper to superconductor ratio
Bundle 42 strands (14 x 3)

Coil Dimensions

Magnet

44% fill factor
4 turns
21 x 21 mm cross section

178 mm width, 192 mm length

310 x 310 x 180 mm outer dimensions

 

38

Short Sample of 0.5mm Wire
Data from Otokumpu NbTi 0.5mm wire filament table

 

 

   
  

 

 

 

 

 

400.00
+Jc times NbTi area I
35000 Manufacturer Guaranteed
X Manufacturer Typical
300 00 )K Manufacturer Actual
' — O— - POISSON MOCICC Dipole
250.00 4 -— - - , - -. . 2.. - -, L L.-- L __ L

Current (A)
N
O
O
8
l
l
l
l
l
l

 

 

 

150.oo~ ~ - —-v~ -
100.00 - _ c_-_ m_

- 1x2. -1”. c
50.00 X
0.00 r T l I I I I

0 2 4 6 8 10 12 14 16

Field (T)

Fig. 3-3: The short-sample current limit (for the individual wire strand) is represented by the
intersection (about 290 A at about 2.5 T maximum field on conductor) of the circle points,
POISSON-simulated [Ref. 13] fields for given currents, and the diamond points, the critical current
of the amount of NbTi material that is in the wire used (NbTi data from an Otokompu table of NbTi

material properties). Also shown are the guaranteed, typical, and actual current-field specs provided
by the manufacturer, Supercon.

39

 

 

 

 

 

 

 

 

 

 

Fig. 3-4: The magnetic field lines for the top-right quarter (symmetry) of the MOCICC dipole cross-
section, calculated by POISSON [Ref. 13], for 40 kAoturns. The x and y axes represent the x and y
directions and the units are cm.

40

 

3.2 Constructing the metal-oxide CICC dipole magnet

The construction of the magnet consisted of winding and bundling the wire,
inserting the wire into the conduit, finding the appropriate bend radius, developing a
method to bend the conduit into the desired shape, welding the coils, and installing the

coils in the yoke iron. Finally the leads of the two magnets were joined.

In order to create a wire bundle of appropriate length, the length of conduit
needed for the 4 turns of the coil was estimated. This was done by calculating the inside
and outside perimeter, averaging them, and multiplying by the number of turns. The
bundle was made 450 mm longer than the estimated length of the conduit, for a total
length of 4.0 m. The extra length was important to accommodate error in calculation and

defects in the ends of the bundle from the cabling process.

Three lengths of conductor were wrapped onto a winding device and then spun
into a small bundle that consisted of three wires. Fourteen of these 3-wire bundles were
created, and again attached to the winding machine. The fourteen bundles were then
spun (in the same direction as the 3-wire bundles were spun) into one large bundle of 42

wires.

The conductor wire was inserted into the conduit by hand, without the assistance
of machines. The wire bundle was inserted by pulling it on one end, via a steel cable, and
pushing it on the other end. The steel cable was connected to loops, on the strands at the
ends of the wire bundle, which were a byproduct of the winding process. To wind the
bundles, wire was wrapped around posts at opposite ends of the winder machine. The
wrap around these posts created the loops. To connect the steel cable to the wire bundle,

many of the bundle end-loops were cut and Teflon® tape was wrapped around the bundle

41

where they were cut. The 1/ 16” steel cable was fished through the conduit, threaded
through the six remaining bundle end-loops, and then the cable was fished back through
the conduit. The steel cable could then be used to pull the bundle wire into and through
the conduit. However, the high friction between the conduit wall and the copper in the
bundle led to problems. Sometimes, individual strands would move while the rest of the
bundle did not. Other times, individual strands would kink, creating a point of very high
friction that would prevent the bundle from moving. The most effective solution to these
problems was the combination of pushing the bundle in at one end while pulling on the
other end. The person pushing the bundle could monitor the uniformity of the pitch and

watch for other signs of sliding and bunching.

Scrap lengths of the CICC conduit were used to test the effects of the bend radius
on breakdown voltage. By bending the conduit with tighter and tighter radii, and then
testing the voltage breakdown between the inner and outer conduit, it was determined that
the integrity of the conduit’s electrical insulation deteriorated with bend radii of less than
19 mm (0.75 in.). It is believed that bending the tube with small radii causes internal
deformations that reduce the thickness or density of the insulation, leading to lower

breakdown voltages.

To bend the conduit into the coil shape, a bending form was fabricated using four
pieces of steel tubing welded to a steel plate (Fig. 3-5). A winding arm fit into the
sockets on the winding plate, and then the arm could torque the conduit into the necessary
shape and radii. During the winding process small tack welds were placed on adjacent
turns, to retain shape and reduce the elastic potential energy of the wound steel conduit.

Once the winding was complete, larger sections of weld were placed to secure the coils.

42

After the coils were complete they needed to be installed into the iron yoke.
Normally, coils would be shimmed into the iron to prevent movement, but in this case,
there was a bow in the sides of the coils (rather than being exactly rectangular, the coils
were very slightly oval shaped) as a result of the winding process. When the coils were
forced into place, friction due to compression of this bow held the coils in place. Once
the coils were properly positioned in the iron, small steel bars (Fig. 3-6) that ran from one
coil down to the coil beneath it were welded into place. These bars acted to restrain the

coils in the axial direction.

Once the coils were in place, the leads were connected together with a joint, such
that the coils were electrically in series. First, the wires from each magnet were soldered
together. An enclosure for the soldered leads was made by fabricating two clamshell
halves of a stainless steel jacket with an insulating layer of Macor® ceramic (Fig. 3-7).
To complete the joint, the clamshell halves were welded around the soldered leads and
around the CICC conduit. The joint was secured to the magnet iron by a post (Fig. 3-8).

The space between the square Macor® and the round soldered bundle allowed
liquid helium flow for cooling the joint (Fig. 3-7). There are some areas that the Macor®
insulation does not cover, where about 3 mm of liquid helium is the only insulation. The
3 mm of liquid helium should be enough to protect against the less than 1 kV quench
voltages expected because the dielectric strength of liquid helium at one atmosphere is
greater than 10 kV/mm [Ref. 22]. The test procedure for this magnet was to insert it into
a liquid helium bath; in a device designed for actual beamline there Would be liquid

helium flowing through the splice area.

43

is»???

.r. I...

. «PDQ-fl ”(1.1. _...._;..;r>/—

.32.?
. .

........H.,...H_W....n .2 ,

a

.1 .
,3...
p, .f

reaffirm” .... .

a... s...
.1; ..
a r
a. _
1.2.1...

is?
4 . .. _ .

a

Earp?” .

.; ..._ Saran...

 

.2

il into shape.

d the co

form cylinders to wrn

rnding-

de the w
linders have a 2” outer diameter and are welded onto the plate beneath them. During

ts

A steel winding am mm

Fig. 3-5

to maintain the shape.

lly

ding, tack welds were placed occasiona

The cy
win

Coil connection
joint

 

Fig. 3-6: The MOCICC dipole hangs from the NSCL Dewar lid, ready to be tested. 0n the left is the
lead connection between the two coils.

45

 

Fig. 3-7: The joint that connects the coils in series is comprised of the solder wire bundle surrounded
by two clamshell halves made of an outer stainless steel jacket with a Macor® insert to prevent the
conductor from making contact with ground. Seen here is a cutaway: one of the two clamshell
halves.

46

 

Fig. 3-8: The completed lead connection joint is seen here, secured to the magnet iron by a post. The
two clamshell halves were welded in place around the soldered bundle.

47

3.3 NSCL Test Results

The metal-oxide CICC dipole magnet was tested at the NSCL. The NSCL’s
available power supply, a 1200 A PowerTen (subsidiary of Elgar, San Diego, CA) power
supply, model P63C81200, limited the test to 1200 amps. The test was simple and served
mainly to confirm that MOCICC technology can be used to produce superconducting
magnets.

The complete magnet was mounted under a Dewar lid, where the magnet leads
were connected to 1000-amp-rated leads in the Dewar lid (Fig. 3-9). The Dewar lid and
magnet were lowered into the Dewar, which was then sealed, evacuated, subjected to a
pump-and-purge purification process, and then filled with liquid helium. It was
necessary to maintain the level/height of the liquid helium above the top of the magnet in
order to keep the superconducting coils immersed. Although MOCICC technology was
designed for forced-flow liquid helium, instead a liquid helium bath was used for this
test. The 1200 A NSCL power supply was connected to the 1000-amp-rated leads.

The magnet ramped to 1200 A at 1.2 V in about 1 minute in two tests. At this
current the magnet was stable and did not quench. The 1200 A current used in this test
was only a small fraction of the predicted sustainable current: 10,000 A. To gather more
information on the current density and other operating properties, another test was done
at the Massachusetts Institute of Technology’s (MIT) Plasma Science and Fusion Center

with a 10,000 A power supply and more extensive instrumentation.

48

11....5.
.._......
fl...

. ,n. .1... :1.............

48?.

a?

.11 .11314—12. fl:— xvi.»

2......

. I. :....-.r-.:.F..E

.5....._........_. .........=

44%—3‘25?“ .f... 5:

SS...

.............._........................ S

1 ,......_......_.......
. .-I......r...a..,5 r . . : .: , .
. {Ly—.13. . EWQ/m.fiym.:/U1fiwzc

.
.

.

 

The

le hangs under the NSCL dunking Dewar lid, ready to be tested.

rpo

The MOCICC d
magnet leads have been connected to the Dewar leads.

Fig. 3-9

49

3.4 October 2006 MIT Test Results

The MOCICC dipole was shipped to MIT, where it was installed into a Dewar
and prepared for testing. One of the magnet leads was connected to 10 kA helium gas-
cooled Dewar leads through a superconducting Rutherford cable lead bridge (Fig. 3-10).
The Rutherford Bridge was soldered to the magnet lead, and connected to the Dewar lead
by two copper blocks that bolted together to fix it in place (Fig. 3-11). The other lead
was simply clamped to the block. To prepare the Dewar for the test, it was cooled with
liquid nitrogen, evacuated, filled with liquid nitrogen, pumped free of liquid nitrogen, and
then filled with liquid helium. Nearly 300 L of liquid helium were needed to fill the
Dewar to the desired 50% level (Fi g. 3-12).

The magnet was powered by a Dynapower 10 kA, 20V supply, and was protected
by a quench protection circuit, which would trip the power supply at a voltage and time
delay set by the user. To record the test results, the Dewar instrumentation was
connected to both a Macintosh computer and a Yokogawa LR8100 8-channel data
recorder. In addition, three digital voltmeters measured power supply current, power
supply voltage, and induction coil voltage (respectively from top to bottom in Fig. 3-13).

The magnet, first ramped at 100 NS, quenched at 2750 A. The lead connection
denoted V1_a in Fig. 3-12 showed a great deal of noise (Fig. 3-14), probably from wire
movement, as the lead was not restrained well. Voltage spikes due to noise from wire
movement are common in accelerator magnets, and this was likely the cause of the
premature quenching. The other lead connection denoted V1_e in Fig. 3-12, had resistive

voltage, but on a smaller scale than the noise found in V1_a.

50

The quench protection circuit delay and the ramp rate were increased, and it was
found that by increasing the ramp rate, more current could be applied to the magnet coils
before the Rutherford bridge quench heat propagation would reach the magnet and
initiate a quench in the magnet. The results of this procedure are in Fig. 3-15 and the
maximum current attained was about 4320 A. A faster-sampling voltmeter read a peak
current of 4500 A, suggesting an uncertainty of about 180 A in the peak current

measurement.

51

FEE... E:

.... .... E...
.n‘a‘ianu—kx

......._

. .. ..
.. .
..

f.

‘ |~ .r..-4.I-14:.-:vn

 

s Dewar. The large Dewar leads

rted into MIT’

mse

le magnet ready to be

1P0

: MOCICC d'
are helium gas cooled.

Fig. 3.10

52

e

6
r
n

[1% r... .....

.49.”. 57 .. ..

45.....nffis

2.1.. ..!0 .

.........i.

 

which quenched during the

ble,
in Fig. 3—12.

ragram

ions. The short magnet lead on the right is connected to

f the lead connect
the Dewar lead by some clamped Rutherford superconducting ca

Close-up view 0

Labels have been applied that correspond to the d

Fig. 3-1 1
testing.

53

 

          

      

    

        

    
    

   

   

 

 

% % ,% % %

5 0. 3 2 7

«A. w % Hm... ,6...

m..m .m _m m
m . c C C C C
c LV+ .0. 0.__3 .2 2
2H 3 1V m . m .1” 2: n0!” m
5.kw, . . _ ._
Hawk.

6 be“ :d
..v.., .
mmb,d .mmM

_202 odo
ev[J_ cmw
H VV .ch
L .wsm

C60
. C
.hmm

a N
a

:2.
g
.

.
A

. E : . e
a a... .3...
 . _,

. .
;
xww.

 

30hm

  

 

300K: 3.
77K: 80.6 kOhm

LN thermirsto 4

               

 

300K: 84 Ohm
77K: 1.7 Ohm

LN thermistor

   

 

V_lc
__1b is a voltage
V_2a

1e are voltage taps on

V
lium leveb needed to reach the

a height between 40% and 50% was desired.

V_1a and

inside the Dewar.

the left and right (respectively) connections between the Dewar leads and the magnet leads.
The liquid helium level sensor instrumentation k labeled

2d. 0n the right hand side of the diagram are the liquid be
specified heights in the Dewar;

Diagram of the test instrumentation

and V_ld are voltage taps a nearly-co-wound coil for measuring induced voltage. V

tap to measure the magnet voltage.

wwwv

Fig. 3.12

M

 

Fig. 3-13: The instrumentation setup at MIT. Data were recorded by a Macintosh computer
(bottom left), a Yukagawa LR8100 8-channel data recorder (top), and three digital voltmeters (top-
left).

55

Voltage in MOCICC Dipole Coil-Lead-to-Dewar—Lead Bridge

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.0
O.5«+—~ ~~—v~— e —- L — -———
9
E, 0.0
(D
D)
:9
305- ee ee e ,,_ ee e ___ e
4.0 +V1_a
—Cl- 'V1_e
-1.5
0 1 2 3 4 5

Time (s)

Fig. 3-14: Measurements during the ramp-up of the magnet (ramp #5 according to Fig. 3-15) from
the voltage taps V1_a and V1_e (labeled according to Fig. 3-12), the bridge connections between the
coil leads and the Dewar leads.). V1_e begins to generate resistive voltage shortly before the quench
(the quench occurred after 3.5 s) and V1_a is very noisy during the entire ramp. The magnitude of
the noise on V1_a is much higher than the resistive voltage on V1_e. It is likely that the noise on
V1_a was generated by wire movement, and was the cause of the premature quench of the magnet.

The ramp rate for this test was 700 A13. The other tests, at various ramp rates, showed similar
results.

56

Quench History for Oct. MIT Test
Labels represent quench number

 

 

 

 

 

 

 

 

 

 

 

 

4500 l
- 9 :
4300-e in“ —'
g 10 $11
4100"*’ ’ TT—iiiiii
I
6
3900 ~- iw - e eeeee
8
A 7
5‘. 3700 - a“; J4 — - — e
9‘.» 4 5
33500— e e e eh
.C
0
§ 3300 e e e e _
0 3 020ms quench protection
3100 { delay .4
050 ms quench protection
2900 delay sampling error H
i2 050 ms quench protection
2700 I 1 delay discrepancy error A
2500 l T T l l T l
0 200 400 600 800 1000 1200 1400 1600
Ramp Rate (A/s)

Fig. 3-15: Quench history of the MOCICC dipole test at MIT. The magnet could not be ramped to
its full sustainable current due to problem with the connection between the Dewar leads and the
magnet. By increasing the ramp rate, the magnet current could be pushed farther before heat
propagating from the quench in the lead connection would initiate a quench in the magnet. Error in
the measurement due to sampling rate is shown with thick, solid error bars. For the last four
measurements, the measurement was read from voltmeters, which did not agree with the
measurement read by the computer. This discrepancy is represented by light, dotted error bars.

57

3.5 Analysis:

The successful ramp test at the NSCL demonstrated the first step in producing a
scalable superconducting MgO CICC magnet with radiation resistant materials. The
1200 A applied current was only a small fraction of the maximum sustainable current, but
the test was significant as proof-of—concept for the new technology, and established its
ability to carry an effective current of over 12 A/mmz.

The importance of the result is that the MgO CICC technology is scalable to

larger devices. The entire MgO CICC conduit can be scaled up in size, without changing

pm *4 1.333;;

the thickness for the MgO layer or the steel jacket. This would create more room for
conductor and reduce the fraction of the overall cross section that the non-conductor
components occupy. Other research into magnet technologies with radiation resistance at
the 108 Gy scale has produced small coils, with anodized aluminum or with ceramic-
sheet-insulated turns [Ref. 23], but these technologies cannot be straightforwardly scaled
up to the size necessary for fragment separators.

Scaling the size of the CICC to achieve higher current density has the
consequences of higher operating current and more absorbed radiation heat. Absorbed
radiation heat is related to the density of the material; therefore, exchanging liquid helium
for copper/NbTi leads to an increased heat load. Increasing the size of the CICC and
hence the size of the conductor would require an increased operating current in order to
achieve larger current density.

Flaws in the connection from the Dewar leads to the magnet leads prevented the
magnet from ramping to the full 10 kA desired. By increasing the ramp rate to

circumvent the bridge connection, a current of 4.3 kA was achieved. The 4300 A applied

58

current translates to an effective current of 43 A/mmz. Resistive mineral-insulated-
conduit, the only other scalable magnet technology with 108 Gy radiation tolerance, has
current density on the order of 2 A/mmz; therefore, 43 A/mm2 is a significant step

forward in radiation-resistant magnet research.

59

373.: 1 72“.“‘5" int _ a

3.6 Further Research

The MOCICC can be improved to achieve higher current densities. The
manufacturer of the conduit claims that the thickness of the outer stainless steel jacket,
the MgO insulation layer, and the inner stainless steel jacket can all be reduced to 0.5
mm. This would reduce the overall cross section, and greatly improve the effective
current density. This approach could lead to a current density of up to 170 A/mmz.
Another means of improving the current density would be to ‘co-draw’ the conductor
with the conduit. The conduit starts as larger tubing and is drawn down to the 1 cm size;
by inserting the conductor before this is done, a much greater packing factor can be
attained. The forces involved in the co-drawing process may damage the
superconducting filaments in the wire, so testing is necessary; a sample of co-drawn
MOCICC is awaiting testing at Brookhaven National Lab.

Changing the MOCICC insulation material may also improve its dielectric
performance. Spinel (MgAlZO4) is another radiation resistant ceramic that has a better
dielectric insulation, better radiation tolerance, and resulted in higher arcing voltages
when tested (the sample had 30% spine] and 70% MgO) [Ref. 5]. Spinel appears to be
less hygroscopic than MgO, but at liquid helium temperatures water would freeze and no
longer affect the dielectric strength.

Constructing the conduit with aluminum jackets, rather than stainless steel, might
also improve the radiation tolerance. Because of its lower density, the aluminum would
absorb less radiation heat. In tests, welding aluminum jacketed, A1203-powder insulated
MOCICC resulted in a significant, although non-damaging temperature rise. The A1203

also showed a lower breakdown voltage than MgO, but the arc voltage was still much

60

greater than the expected quench voltages [Ref. 5]. Unfortunately, the weld size for
aluminum is much greater, leading to a larger overall cross section, and a smaller
effective current density.

In order to test the stainless steel MOCICC at higher magnetic field strength, and
also to confirm the scalability of the technology, a large quadrupole is in development at
N SCL. The quadrupole will consist of multiple coils, and a standardized method of

connecting coils will be developed.

61

I'll}. .J'igrlll“

4 CONCLUSIONS

The technology now exists to build next generation fragment separators with
radiation resistant superconducting magnets. Metal~oxide insulated CICC magnets can
provide relatively high current densities with materials tolerant of radiation on the order
of 108 Gy. This will be useful for applications like quadrupoles, where the current
density requirements make resistive mineral-insulated magnets an unpleasant option.
Cyanate-ester potted magnets can be used in areas on the order of 107 Gy, and will be
useful for fragment separator magnets that are not directly following the target. Both

technologies may have further applications, such as magnets for fusion research.

62

[’15 I!!!" ‘

5 APPENDICES
5.1 Cyanate-Ester Winding Instructions

The first step to wind the cyanate-ester coils was to cover the aluminum winding
form (the mold that the wire wraps around to give the coil the correct shape) with
Teflon® tape. The form was then held together using steel bolts into threaded holes in
the bottom plate of the form. Next, the coil-winding machine was setup by assembling
the parts, setting the desired wire tension, and bolting the complete form onto the coil-
winder. The end of the wire was pulled out from the wire spool, stripped of insulation,
and soldered to a thicker wire called the front current lead. A four-inch long piece of
fiberglass woven sheath (about .5” diameter) was inserted onto the front lead just past the
groove in the magnet form and the lead was clamped there using a Teflon®-tape-coated
binder-clip at the edge of the form. A strip of Kapton® was inserted above it to keep the
coil wires from cutting through the fiberglass and shorting to the lead.

The cyanate-ester was warmed for about an hour at 40° C, and then it was then
mixed up by adding 60 parts of hardener to each 100 parts of cyanate ester resin. Next
the cyanate-ester was painted generously onto the fiberglass sheath and around the inside
of the form. The winding process was then started slowly and with caution so as not to
catch the wire on the clip clamping the front lead. After approximately 500 turns around
the form the coil wires served to clamp the lead and so the clip was removed. When the
winding completed at 2200 turns the coil wire was stripped and a back lead similar to the
front was soldered on. After this, fiberglass tape, matched to the height of the coil, was
measured to the inside perimeter plus about four inches. The tape was wrapped around

the coil, starting just underneath the back lead. As the tape completed one turn around

63

[5523mm

the inside perimeter of the coil a slit was made such that the back lead would be able to

fit through. The tape was wrapped the rest of the way and painted with the cyanate-ester.

Then form bars were clamped into the form to compress the coil wires. Following
this, the coil was put into a pizza oven connected to a computer controlled power source.
The cyanate-ester needed a special heating cycle to cure and the computer was set to
cycle the oven to the cure cycle: Ramping the temperature to 90 deg. C and soaking for 4

hours then ramping the temperature to 177 deg. C and soaking for 4 hours.

64

run Amman!“

5.2 Cyanate—Ester Winding Photo Flow-Chart

  
 
  
  
   

Prepare
form

by applying
Teflon tape .
around each

part. I ’
7 -

Strip wire and
solder on
front lead.
Slide about
4" of
fiberglass
sheath onto

lead. I .

Clamp the
lead near
front edge of
form. Coat
clamp with
Teflon tape to
keep from
snagging the

wire as it
winds. 1

Coat fiberglass
sheath and fill
lead groove
with cyanate-
ester. Paint
generous coat
of epoxy onto
inside edges of
form.

      
   
 

 

 

fl

 

 

Begin winding with caution to not catch wire on
clamp. After ~500 turns the clamp may be
removed.

 

 

 

Starting beneath

 

' f wipe away

 

When winding
is finished
strip tail wire
and solder tail
lead. Measure
a fitting
fiberglass tape
to perimeter
length plus
~4”.

 
  
  
  
  

the tail lead,
wrap the
fiberglass tape
and coat with
cyanate-ester.
Slit the last 4"
of the tape and
poke the lead
through the slit.

  

  
  
  

  

form bars.

 
 

the excess
epoxy and put
in oven for
curing.

 
 
      

65

5.3 Gandalf Input File for MOCICC Dipole; Operation in 4 W/m Heating

MOCICC 4W_m heating normal operation

&INDATA
NELEMSz 100, XLENGT= 3.6576, ITYMSH= 1,
NELREF= 50, XBREFI= 1.46, XEREFI= 2.2,
ICHFUN= 0,
AHEH = 0.0, AHEB = 1.976E—5,
DHH _ 0.0, DHB = 0.001,
PHTHB = 0.0, PERFOR= 0.0,
ICBFUN= O.
ISC 31, ASC = 2.77E—6,
E0 = 1.0E-4, NPOWER= 30, EPSLON=—0.99,
IST = 1, AST = 5.53E—6, RRR = 100.0,
IJK 13, AJK = 1.90E-5,
IIN 21, AIN = 2.89E-5,
PHTC = 0.055, PHTJ = 0.0178, PHTCJ = 0.0107,
INTIAL= 2,
PREINL= 5.0E+5, TEMINL= 4.2, MDTINL= 5.0E-3,
IOPFUN= 0, IOPO = 1.0E+4, TAUDET=0.01,
TAUDUM=0.1,
IBIFUN= 0, BISS = 1.0, BOSS = 2.0,
IQFUN : 0:
Q0 = 4, XQBEG = 0.1, XQEND = 3.6,
TAUQ = 10.0,
TEND = 10, PSTEP = 1.0, GSTEP = 0.1,
STPMIN= 1.0E-3, STPMAX= 0.1,
METHOD: 0,
ISTORP= 1, IRESTA= 0, MONITR = 1,
&END

66

6 BIBLIOGRAPHY

Ref. 1: Available: http://www.gsi.de/zukunftsproiekt/index e.html.

Ref. 2: Sherrill, B. 8., “Overview of the Rare Isotope Accelerator Project,” in NINB 204, 2003, pp
765-770.

Ref. 3: A. F. Zeller, V. Blideanu, R. M. Ronningen, B. M. Sherrill, R. Gupta, "Radiation Resistant
Magnets for the RIA Fragment Separator," Particle Accelerator Conference, 2005. PAC 2005.
Proceedings of the , vol., no.pp. 2200- 2202, 16-20 May 2005.

a ”vim-'1‘! if- nun

Ref. 4: A. F. Zeller, J. C. DeKamp, and J. DeLauter, “A radiation resistant dipole,” IEEE
Transactions on Applied Superconductivity, vol. 15, no. 2, pp. 1181-1184, June 2005.

Ref. 5: A. F. Zeller, J. C. DeKamp, J . DeLauter, and A. Ghosh, “Metal oxide CICC for radiation
resistant magnets,” Advances in Cryogenic Engineering -ICM C, vol. 52, pp. 575-581, 2006 American
Institute of Physics.

Ref. 6: M. Winkler, M. Svedentsov, K.-H. Behr, H. Geissel, H. Iwase, G. Moritz, C. Muhle, H. Weick,
"Radiation Resistant Quadrupole Magnet for the Super-FRS at FAIR," IEEE Transactions on
Applied Superconductivity, vol.16, no.2pp. 415- 418, June 2006.

Ref. 7: R. Reed and D. Evans, “Insulation Systems for Muon Collider,” Report to A. Zeller
(unpublished), Cryogenic Materials, Inc. April, 2000.

Ref. 8: J. Schultz, “Radiation resistance of fusion magnet materials,” 20‘” IEEE/NPSS Symposium on
Fusion Engineering, pp. 423- 426, 14-17 Oct. 2003.

Ref. 9: A. Harvery and S. A. Walker, “High current density coils for high radiation environments,”
IEEE Trans. Nucl. Sci, vol. NS-l6, pp. 611-613, 1969.

Ref. 10: D. George, “Magnets with mineral-insulated coils at SIN,” in Proc. 0f MT-5, pp. 719-722,
1975.

Ref. 11: American Magnetics, available: http://www.americanm__agnetics.com/.

Ref. 12: M. Wilson, “Computer simulation of the quenching of a superconducting magnet,” RHEUM
151, unpublished, 1968.

67

Ref. 13: Poisson Superfish, Los Alamos Accelerator Code Group, Los Alamos National Laboratory.

Ref. 14: A. F. Zeller, R. Zink, J. Wagner, S. Hitchcock, J. C. DeKamp, and A. Balaint, “Testing of
large aperture superferric quadrupoles,” IEEE Transactions on Applied Superconductivity, vol. 9, no.
2, pp. 693-696, June 1993.

Ref. 15: K. Bittner-Rohrhofer, P. Rosenkranz, K. Humer, H. W. Weber, J. A. Rice, P. E. Fabian, and
N. A. Munshi, “Characterization of reactor irradiated organic and inorganic hybrid insulation
systems for fusion magnets,” Advances in Cryogenic Engineering, vol. 48A, pp. 261-268, 2002.

Ref. 16: M. Sawan, P. Walstrom, “Superconducting magnet radiation effects in fusion reactors,”
Fusion Technology, vol. 10, p. 741, 1986.

Ref. 17: R. Gupta, M. Anerella, M. Harrison, W. Sampson, J. Schmalzle, R. Ronningen, and A.
Zeller, “Radiation resistant HTS quadrupoles for RIA,” IEEE Transactions on Applied
Superconductivity, vol. 15, no. 2, pp. 1148-1151, June 2005.

Ref. 18: Gandalf v.2.2, Cryosoft, 2001.

Ref. 19: J. Baldzuhn, H. Ehmler, A. Hoelting, K. Hertel, C. Sborchia, L. Genini, T. Schild, “Coil tests
and superconductor code calculations for the stellarator W7-X coils,” Cryogenics, vol. 46, issues 7-8,
pp. 507-516, July-August 2006.

Ref. 20: A. Zeller, private communication, Nov. 2006.

Ref. 21: A. Harvey, “Experience with the LAMPF Mineral-Insulated Magnets,” MT-6, Bratislava
551 (1977).

Ref. 22: S. W. Schwenterly, “Design and testing of electrical insulation for superconducting coils,”
Adv. In Cryo. Eng., vol 33, pp. 271-281, 1988.

Ref. 23: A. F. Zeller, J. C. Dekamp, "Radiation resistant magnet R&D at the NSCL," Proceedings of
the Particle Accelerator Conference, 2003. PAC 2003., vol.1, no.pp. 161- 163 Vol.1, 12-16 May 2003.

68

 

R

u11111111111111ij111111u