Design of accelerator magnets at CERN with Opera

Design of accelerator magnets
at CERN with Opera
Daniel Schoerling
with material of TE-MSC-MNC
June 3rd 2014
CERN
The Twenty-One Member States of CERN
Distribution of All CERN Users by Nationality (2014)
Convention: The Organization shall provide for
collaboration among European States in nuclear
research of a pure scientific and fundamental character,
and in research essentially related thereto. The
Organization shall have no concern with work for military
requirements and the results of its experimental and
theoretical work shall be published or otherwise made
generally available.
3
CERN accelerator complex
4
Daniel Schoerling
TE-MSC-MNC
6
Who are we within CERN?
TE-MSC-MNC Section:
CERN-wide support for accelerator Normal
Conducting Magnets (50000 tonnes, 5000
magnets, 52 persons).
• Design, construction, commissioning,
maintenance and upgrade of the normal
conducting (nc) magnets for present and
future CERN accelerators and beam lines
• Management of non-installed nc magnets
(spares and shelf-ready units)
• Develop knowledge and maintain excellence
in nc magnet technologies, in radiation
resistant magnets and in permanent magnets.
CTF3 Quadrupoles
SPS
CLIC Hybrid Quadrupole
7
What is required to bend particle beams?
Dipole magnets
By
N
x-axis
S
• Equation for normal (non-skew) ideal (infinite) poles: y=  r
(r = half gap height)
• Magnetic flux density: Bx = 0; By= b1 = const.
• Applications: synchrotrons, transfer lines, spectrometry, beam
scanning
8
What is required to focus particle beams?
Quadrupole magnets
By
S
N
x-axis
N
•
•
S
Equation for normal (non-skew) ideal (infinite) poles: 2xy=  r2
(r = aperture radius)
Magnetic flux density: Bx= b2y; By= b2x
9
What is required to correct chromatic aberrations?
Sextupole magnets
By
S
N
N
S
S
x-axis
N
•
•
Equation for normal (non-skew) ideal (infinite) poles: 3x2y y3 =  r3 (r = aperture radius)
Magnetic flux density: Bx= b3xy; By= b3(x2- y2)/3
10
For what do we use Opera?
SESAME
Proposal 1 (Xray diagnostic)
D03(INFRARED BEAMLINE)
SR-C03-D
Visab
SR-C0
le
2-D
04-D
SR
Opening
0.8*0.2m
line
Beam
SR-C
-C0
1-D
Openi
0.15*0
ng
.1m
0.15*0.15
m
Opening
0.15*0.1m
SR-C15-D
0.3*0.25
Opening
0.3*0.2m
Opening
SR-C07-D W16-ALS-WIG
y diag
2 (Xra
LER SR-C0
Opening
m
0.2*0.15
Opening
Prop
osal
16-D
Opening
1*0.2m
6-D
SR-C
nost
ic)
2m
SR
Openi
0.7*0.
ng
-C0
5-D
ng
Openi .2m
0.5*0
m
Opening
B CM
SM
0.3*0.25
QP
ol e
Triplet
SM
m
Opening
1 X 0.2m
QP
ol e
Triplet
0.2m
SM
Opening
1X
SM
Opening
0.2*0.15m
SR-C1
Opening
0.2*0.15m
SR-C
4-D
08-D
g
m
Openin
0.5*0.2
Opening
0.3*0.2m
m)
SR
EUV
(29.2
Opening
0.4*0.2m
-D
C09
SR-
12-D
-D
SR-C10
I11-POWD
D10
D0
I08-V
m)
UV (36.6
SR-C11-D
EIN CRYSTAL
Opening
1X
0.2m
SR-C
I07-PROT
D12
3-D
LOGRAPH
Y (36.4m)
Opening
0.2*0.15m
-C1
9-X
-SA
ER DIFFRACT
ION (31.3m)
X-W
AX
(29.
6m)
F-X
m)
4.3
S(3
AF
• Design of new magnets for new projects:
 Preparation of technical specifications
 Procurement in industry
• Improvement and upgrades of installed magnets
which were build up to 50 years ago:
 Many magnets are older than 40 years. The
influence on the magnetic field of required
modifications are predicted by using Opera.
 Prediction of the error distribution in installed
magnets to improve the machine
performance.
MedAustron
LIU: PS 2 GeV upgrade
ELENA
11
Design of new magnets
• Required prediction of the magnetic field
in the order of 10-5 in the gap region
(circa 100 x 50 mm), specifications are
often within ±2 x 10-4
• Relative accuracy of at least 1% for the
absolute field value
• For fast ramped magnets dynamic
simulations have to be performed
Functional
Specification
• Parameters:
Beam optics,
power, cooling,
vacuum,
integration,
transport, survey,
safety
Analytical Design
• Check for
feasibility
• Fix coil
parameters
• Check different
design options
2D Optimization
• Check different
cross-section
design
• Optimize field for
field homogeneity
• Depending on
magnet length
iteratively with 3D
optimization
3D Optimization
• Optimize
integrated field
homogeneity
Design Report &
Engineering
Specification
• Perform iterative
design with
mechanical
design
12
How a quadrupole is designed? I
• All interfaces and parameters are defined.
• Analytical design:
𝑁𝐼
𝐺 = 2𝜂𝜇0 2
𝑅
 Coil parameter definition (available power
converter, conductor, space, cooling
method).
• 2D Opera design with the aim to have only a B2
component:
N−1
B𝑟,𝑘 𝑟0 sin nφ𝑘
1
𝑘=0
 Optimize overal shape of the magnet, in
particular by using a pole shape 2xy= r2 and
tangential line with length s
0.5
Br in T
B𝑛 𝑟0
2
=
𝑁

0
0
45
90 135 180 225 270 315 360
-0.5
-1
Angle  in deg
s
13
How a quadrupole is designed? II
1.00E-04
Integrated b6 (r = 27 mm)
• 3D Opera design
 Optimize the integrated field.
 To be performed iteratively with 2D
design. End effects usually cause a
“drop-off” that means a negative
multipolar component.
 If the magnet is long enough a positive
multipolar component can be added in
2D to minimize or avoid the endchamfer.
0.00E+00
-1.00E-04 0
5
10
15
20
25
-2.00E-04
-3.00E-04
-4.00E-04
-5.00E-04
-6.00E-04
Chamfer Height in mm
• Mechanical Design
 Produce a 3D CATIA model of the
magnet
 Produce 2D functional drawings which
contain all functional dimensions and
tolerances (“ISO language”)
14
Comparison with measurements I
• High level of confidence in the magneto-static simulations with Opera since many
years. Numerous magnets are designed without means for after-production field
quality correction!
• Excellent prediction of magnetic length and field quality.
• The small differences between simulation and magnetic measurements come from:
 Mechanical errors.
 Uncertainty of the BH curve of the used material.
 Uncertainty of the magnetic measurements.
• If means for field quality correction are foreseen this is due to extremely tight
requirements (for example for ring dipole magnets).
dB/B [10-4]
y = 24 mm
x [mm]
y=0
Simulation (dashed)
Measurement (solid)
Measurement before shimming
Courtesy MedAustron
15
Comparison with measurements II
Challenge
• Excellent and repeatable field quality at very low field.
Solution
• Selection of high permeability electrical steel M270-50 A HP
diluted with 1 mm thick stainless steel to increase the magnetic
induction in the iron and to avoid working in the highly nonlinear
area of the BH-curve.
Tangential magnetic
Simulations & Measurements
• 2D model with adjusted current (3 x higher), 3D model with
adjusted current (3 x higher) and no packing factor, 3D model with
packing factor of 33%, sliced 3D model.
Normal magnetic
• Good prediction of field homogeneity and magnetic length
• Poor prediction of the TF (problem of the material model?)
No dilution
Remanent
effects
1.00
Normalized TF
1.05
0.95
Linear decrease:
Not in the model
Dilution 1:2
Saturation
effects
3D isotropic at the center
2D anisotropic
3D sliced
NMR measurements at center
0.90
0.85
0.80
0
0.1
0.2
0.3
0.4
Field in the aperture [T]
0.5
0.6
16
Analysis of existing magnets I
Upgrade requirements
• Long life-time and versatile use of accelerators at
CERN require excellent prediction of the field
distribution in accelerator magnets.
• Tests and magnetic measurements are often too
time consuming and expensive.
Improvements of machines
• PS magnet with 6 current circuits and non-linear
iron makes analytical prediction difficult.
17
Analysis of existing magnets II
Dipole, Focusing
x 10
0
Sextupole, Focusing
-4
Figure-of-eight
W-PFW
ΔS [Tm-2/A]
B [T/A]
-0.4
0.04
-0.8
N-PFW
W+N-PFW
0.02
0
-1.2
-0.05
x 10 -3
3
-0.03
-0.01
0.01
X [m]
0.03
-0.02
-0.05
0.05
Quadrupole, Focusing
2
O [Tm-3/A]
G [Tm-1/A]
1
-0.01
0.01
X [m]
0.03
0.05
x 10-3
0.5
• The exact
knowledge of the
field distribution
allows to control
crucial machine
parameters
-0.5
-1
0
-0.05
-0.03
Octupole, Focusing
1.5
1
x 10-3
• Fits were produced
with Opera of
multipoles up to
octupole
-0.03
-0.01
0.01
X [m]
0.03
0.05
-0.05
-0.03
-0.01
0.01
X [m]
0.03
0.05
18
Analysis of existing magnets III
• 2D & 3D calculation including Gaussian
distribution of the position of the coils and the
shape of the iron with many DOFs per magnet
(OPERA)
• 2D: 1000 models per magnet type and current
level have been calculated:
 1-2 d with DSS licenses, before 10 d
• 3D: 1000 models have been calculated:
 Deformation of the mesh was implemented
in Opera (new feature), possible with
mosaic mesh
 Re-create every model and re-solve it
(tetrahedral mesh only)
19
Summary & Outlook
•
Opera is our most important software tool for normal-conducting accelerator magnet design.
•
Very good support available!
•
Opera was used to design numerous magnets with solid and laminated yokes (packing factor
95-97%). Magneto-static simulations and measurements predict precisely the magnetic field
quality.
•
The interface is relatively user-friendly.
•
Mesh generator improved drastically over the last years but further improvements would be
appreciated.
•
Solution times of magneto-static and especially transient simulations can be extremely long,
which is prohibitive for many calculations!
•
Parallel-processing available but licensing scheme makes it unattractive to run many models
on many CPUs.
•
Obtaining highly accurate results with less solution time would be beneficial (for example
higher order elements?).
•
Very closed software with limited interface capabilities.
•
Multi-physics capabilities were limited, but seem to improve.
•
Easy to use and fast hysteresis solver would be very beneficial (we have set-up a
measurement lab for electrical steel and try to understand hysteresis in accelerator magnets).
20