NewNewNewNew LHC quench calculations and the LHC quench calculations and the LHC quench calculations and the LHC quench calculations and the luminosity limit for heavyluminosity limit for heavyluminosity limit for heavyluminosity limit for heavy----ion collisionsion collisionsion collisionsion collisions
R. Bruce*, S. Gilardoni, J.M. Jowett R. Bruce*, S. Gilardoni, J.M. Jowett
and
D. Bocian**, B. Dehning, A. Siemko
CERN
* also at MAX-lab, Lund University** also LARP Toohig Fellow
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 1
– Network model– Validation of the model– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
Motivation (1)Motivation (1)The LHC will run ~1 month/year with heavy ions. Nominal parameters:
07/01/2009 R. Bruce, D. Bocian, AP Forum 22
• Although the stored energy in the Pb82+ beam is much lower than in the proton beam, beam loss mechanisms peculiar to ions may limit luminosity. Most serious are:
– Collimation inefficiency
– Bound free pair production (BFPP)
Motivation (2)
• Important to predict the quench limit as accurately as possible to estimate the impact of these beam losses. Same holds true for proton losses.
• Earlier estimates of quench limit make simplifying assumptions about the distribution of beam losses or the thermal behaviour of magnets
07/01/2009 R. Bruce, D. Bocian, AP Forum 3
• To make more accurate estimates, these factors need to be accounted for
• Here we calculate the quench limit for a specific beam loss mechanism – BFPP – combining tracking, FLUKA shower simulations and a thermal network simulation of the heat flow in a magnet
Bound free pair production
• During Pb82+ operation in the LHC, electromagnetic interactions between colliding beams take place at IP:
– Bound Free Pair production (BFPP):
( )Cross section for (several authors)Bound-Free Pair Production (BFPP)
07/01/2009 R. Bruce, D. Bocian, AP Forum 4
Compare: σhadr=8 barn
( )
[ ]
1/2
- +1 2 1 21s ,
PP5 2
1 2
1 27
e e
has very different dependence on ion charges (and energy)
log
for
0.2
log
CM
CM
Z Z
Z Z
Z Z
A B
Z ZA BZ
+ → + + +
σ ∝ γ +
∝ =
≈
γ +
K
b for Cu-Cu RHIC
114 b for Au-Au RHIC
281 b for Pb-Pb LHC
We use BFPP values from Meier et al, Phys. Rev. A, 63636363, 032713 (2001), includes detailed calculations for Pb-Pb at LHC energy
• BFPP creates 1-electron ions with altered magnetic rigidity:
• These ions follow locally generated dispersion function dxfrom IP
BFPP at IP2δ=0.012
Magnetic rigidity change
Secondary Pb81+ beam emerging from IP and impinging on beam screen
07/01/2009 R. Bruce, D. Bocian, AP Forum 5
from IP
• Lost in localized spot where aperture Ax and δ satisfy
• Apart from significant luminosity decay, induced heating risks to quench superconducting magnets
S. Klein, NIM A 459459459459 (2001) 51
Beam screenBeam screen
Main Pb82+ beam
BFPP tracking
• Distribution leaving IP does not correspond to the bunch distribution, but to the distribution of collision points
• Spatial distribution in each plane is narrower by a factor
• As it propagates through the lattice, the distribution
07/01/2009 R. Bruce, D. Bocian, AP Forum 6
• As it propagates through the lattice, the distribution changes, in the same way as an unmatched beam at injection
• Beam size at a later point
• Tracking with matrix formalism, off-momentum optics calculated by MAD-X, analytical algorithm finds impact in MB.B10R2
• LHC optics 6.500 as reference case, comparison with 6.503 later
• At IP2: losses at s=378.9 m downstream in end of dispersion suppressor dipole, spot size around 0.5 m
• IP1 and IP5: losses in connection cryostat in missing dipole, less critical. Will focus on IP2.
Tracking
07/01/2009 R. Bruce, D. Bocian, AP Forum 7
IP2
Beam screen
Main Pb82+ beam
Secondary Pb81+ beam
Longitudinal Pb81+ ion distribution on screen
FLUKA shower simulation
• FLUKA simulation to estimate the heat load in the dispersion suppressor dipole at IP2
• impact coordinates of lost BFPP particles from tracking fed as starting conditions to FLUKA
• 3D model of LHC main dipole
07/01/2009 R. Bruce, D. Bocian, AP Forum 8
FLUKA modelreal magnet
Simulated power deposition
15
20P HmWêcm3L
8 bins
4 bins
2 bins
1 bin
Power deposition from FLUKA in the inner coil layer, averaged
over width of coil,
normalized with BFPP cross section and luminosity:
Ptot = σBFPP L Eparticle
Energy deposition longitudinally in hottest bin, different radial binnings.
88 bins in φ (cable), 5 cm longitudinal cell size
07/01/2009 R. Bruce, D. Bocian, AP Forum 9
in LHC design report: Quench limit=4.5 mW/cm3
Beam impact10-5 10-4 10-3 10-2 10-1 100 101 P HmWêcm3L
1300 1350 1400 1450z HcmL0
5
1032 bins
16 bins
Interpolation of power deposition
• However, now more accurate methods to estimate the quench limit exists – thermal network model (see later slides)
• Detailed map of power deposition in the coil needed
• Strand positions not compatible with R-φ mesh
07/01/2009 R. Bruce, D. Bocian, AP Forum 10
compatible with R-φ mesh used in FLUKA
• Interpolating the “best possible” FLUKA mesh
• Applying global scaling factor to compensate for insulation, helium space in cables etc.
• Mathematica program automatically generates network input from FLUKA output
Input to network simulation
• Combining detailed simulated energy deposition from “real beam loss” with thermal network model of magnet
• input to network model:W/m
07/01/2009 R. Bruce, D. Bocian, AP Forum 11
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 12
– Network model– Validation of the model– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
ModelModelllinging of quench levels of quench levels induced by steady state induced by steady state beambeam lossloss heat heat loadload
Thermodynamics of magnet structure
Network Model
Validation of the model
Steady state beam loss heat load simulation
07/01/2009 R. Bruce, D. Bocian, AP Forum 13
Steady state beam loss heat load simulation
More details:D. Bocian, B. Dehning, A. Siemko, Modeling of Quench Limit for Steady State Heat Deposits in LHC Magnets, IEEE Transactions on Applied Superconductivity, vol. 18, Issue 2, June 2008 Page(s):112 – 115; CERN-AB-2008-006, 2008;
D. Bocian, B. Dehning, A. Siemko, Quench Limit Model and Measurements for Steady State Heat Deposits in LHC Magnets, accepted for publication in IEEE Transactions on Applied Superconductivity, 2009
Thermodynamics of magnet structureThermodynamics of magnet structure
Heat transport in the cableHeat transport in the cable
Courtesy C. Scheuerlein
MB magnet – inner layerRutheford type cable
07/01/2009 R. Bruce, D. Bocian, AP Forum 14
NbTi + Cu
He(inside cable)
Cryogenic System
He(bath)
Collar
Yoke
Insulation
cable
1.9K / 4.5K
> 4.5K
Courtesy C. Scheuerlein
Thermodynamics of magnet structure Thermodynamics of magnet structure
Heat transport in the coil at 1.9KHeat transport in the coil at 1.9K
A heat transfer in the main dipole
Cold bore
07/01/2009 R. Bruce, D. Bocian, AP Forum 15
inner layer outer layer
Electrical insulation is the largestElectrical insulation is the largest thermalthermal
barrier at 1.9 K against coolingbarrier at 1.9 K against cooling
Thermodynamics of magnet structure Thermodynamics of magnet structure
Heat transfer in the Heat transfer in the magnet magnet coilcoil
07/01/2009 R. Bruce, D. Bocian, AP Forum 16
A sketch of the heat transfer in the magnet
at nominal operation (a) and at quench limit (b).
MB – arc magnet,Tb=1.9 KHEAT FLOW
LIMITS
� heat flow barriers
- cable insulation
- interlayer insulation (MQM)
- ground insulation
- helium channel around cold bore (for temperatures above 2.16 K)
MQ – arc magnet,Tb=1.9 K
Thermodynamics of magnet structure Thermodynamics of magnet structure
Heat flow limitsHeat flow limits
07/01/2009 R. Bruce, D. Bocian, AP Forum 17
�bath temperature 1.9 K
- Transition HeII → HeI:helium channels are blocked = less effective heat evacuation due to the changing of heatevacuation path
�bath temperature 4.5K
- lower temperature margin (worst case: MQM0.45K)
- Helium channels does not play dominatingrole (heat conduction of He I and polyimide is the same order)
MQM – LSS magnet,Tb=1.9/4.5 K MQY – LSS magnet,Tb=4.5 K
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 18
– Network model• Model construction• Model of the superconducting cable and coils
– Validation of the model– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
ROXIE
magnet field
distribution,
temperature
margin
TECHNICAL
DRAWINGS
detailed magnet
coil geometry
OTHER
non beam induced
heat sources
Hysteresis losses
Eddy currents, etc.
A. Verweij
R. Wolf
Contribution to the quench level
is order of 1-2%
Network Model Network Model
Model ConstructionModel Construction
07/01/2009 R. Bruce, D. Bocian, AP Forum 19
HEAT FLOW
MODEL
MAGNET
QUENCH
LEVELS
FLUKA
beam loss profiles
Material properties
at low temperature
CRYODATA
MEASUREMENTS
model validation
Network Model Network Model
Model ConstructionModel Construction
07/01/2009 R. Bruce, D. Bocian, AP Forum 20
GROUND INSULATION
Courtesy G. Kirby
Network ModelNetwork Model
Model ConstructionModel Construction
07/01/2009 R. Bruce, D. Bocian, AP Forum 21
Network ModelNetwork Model
Helium in the Network Model
07/01/2009 R. Bruce, D. Bocian, AP Forum 22
The volumes occupied by helium in the magnet are considered as:
-the narrow channels,
-semi-closed volumes = inefficient inlet of fresh helium.
The steady heat load, heat up the helium in the semi- closed volumes:
-Helium temperature well above critical temperature at Tb=4.5K
- Critical helium temperature reached already below the calculated quench limit
Network ModelCable modelling
µ-channel
07/01/2009 R. Bruce, D. Bocian, AP Forum 23
Network ModelNetwork ModelCable modellingCable modelling
07/01/2009 R. Bruce, D. Bocian, AP Forum 24
Network ModelNetwork ModelCoil modelCoil modelllinging
07/01/2009 R. Bruce, D. Bocian, AP Forum 25
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 26
– Network model– Validation of the model
• Measurements at the CERN test facility– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
heat
VALIDATION
measured
quench current
END
MAGNET
EXPERIMENT
Heat source
- quench heaters
- inner heating apparatus
ValidationValidation of of thethe Network MNetwork Modelodel
07/01/2009 R. Bruce, D. Bocian, AP Forum 27
heat
predicted
quench currentHEAT SOURCE
MODEL
MAGNET
MODEL
MQY inner quench heater
3900
4000Ultimate current 3900 A
MQM magnets at 4.5 K
3700
3800
3900
4000
4100
4200
4300
4400
4500
4600
4700
25.00 30.00 35.00 40.00 45.00 50.00 55.00
P [mW/cm2]
I m
agne
t [A
]
MQM 627
MQM 677
MQMC 677
Ultimate current 4650 A
Nominal current 4310 A
ValidationValidation of of thethe modelmodel
MB magnet at 1.9K - Inner Heating Apparatus
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
I mag
net [
A]
measurements
simulation
07/01/2009 R. Bruce, D. Bocian, AP Forum 28
2900
3000
3100
3200
3300
3400
3500
3600
3700
3800
9.00 11.00 13.00 15.00 17.00 19.00 21.00 23.00
P [mW/cm2]
I m
agne
t [A
]
MQY 609
MQM 659
Nominal current 3610 A
MQY - outer quench heater
280029003000310032003300340035003600370038003900400041004200
25.00 30.00 35.00 40.00 45.00 50.00
P [mW/cm2]
I m
agne
t [A
]
MQY 609
MQM 659
Ultimate current 3900 A
Nominal current 3610 A
2000
100.00 150.00 200.00 250.00 300.00 350.00
P [mW/cm2]
MQ magnet at 1.9 K - Inner Heating Apparatus
2000
4000
6000
8000
10000
12000
14000
30 35 40 45 50 55 60 65 70 75 80
P [mW/cm2]
Imag
net
[A]
Measurements - middle
Measurements - head
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 29
– Network model– Validation of the model– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
Quench limit simulationsQuench limit simulations
Heat deposition map in the MB dipole magnet coil� heat deposition map for nominal LHC ion beam intensity was created by interpolation of FLUKA data to the cable strand coordinates from ROXIE(Roderik)
�heat deposition map
W/m
07/01/2009 R. Bruce, D. Bocian, AP Forum 30
Energy peak in the coil = 24.3 mW/mEnergy peak in the cold bore = 80.3 mW/m
ENERGY PEAK corresponds to the nominal LHC ion beam conditions (optics ver. 6.500)
�heat deposition map was implemented to Network Model
�the magnet current range from injection to ultimate values (761 A to 12840 A, nominal is 11850 A) was scanned by linear scaling of heat deposition map
Quench limit simulationsQuench limit simulations
Temperature map in the MB dipole magnet coil after heat load � temperature distribution for nominal LHC ion beam conditions, corresponding to 95% of loss energy peak in the coil (23.1 mW/m) and 95% loss energy peak in the coldbore (76.3 mW/m)
Ground insulationin the midplane
=
07/01/2009 R. Bruce, D. Bocian, AP Forum 31
Peak temperature rise in the coil ∆T= 2.0 KPeak temperature rise in the cold bore ∆T=1.4K
For nominal LHC ion beam conditions (beam optics ver. 6.500)
� quenching cable is located at the coil mid-plane
�this temperature map corresponds to nominalmagnet current (11850 A)
= heat flow barrier
Quench limit simulationsQuench limit simulations
Energy peak in the coil = 24.3 mW/m and in the cold bore = 80.3 mW/mENERGY PEAK corresponds to the nominal LHC ion beam conditions
07/01/2009 R. Bruce, D. Bocian, AP Forum 32
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 33
– Network model– Validation of the model– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
Extrapolation to other cases
• Redone tracking in v6.503, new FLUKA simulation
• Profile of power deposition in coil similar to v6.500 except global scaling factor. Scaling by integrated power:
0.020
0.025P HWêmL
0.020
0.025P HWêmL
07/01/2009 R. Bruce, D. Bocian, AP Forum 34
• Gives approximate margin of 15% to quench limit in v6.503
• Question: what has changed ?
100 200 300 400 500strand no.
0.005
0.010
0.015
6.503
6.500
100 200 300 400 500strand no.
0.005
0.010
0.015
6.503 scaled by 1.2
6.500
9.5
10.0
mx
6.503
6.500
2000
3000
4000
bx HmL
6.500
Difference in optics• Phase advance after IP2 changed from v6.500 to v6.503
• Spot size larger - dependent on off-momentum β (calculated from starting conditions at IP2)
off-momentum β off-momentum µ
07/01/2009 R. Bruce, D. Bocian, AP Forum 35100 200 300 400 500 600 700s HmL
0.005
0.010
0.015
0.020
0.025
x HmL
6.503
6.500
300 320 340 360 380 400 420 440s HmL0
200
400
600
800
1000bx HmL
6.503
6.500
100 200 300 400s HmL8.5
9.0
6.503
100 200 300 400 500 600 700s HmL
10006.503
off-momentum β
central BFPP orbit
IP2 IP2impact
Squeeze, 5 TeV: preliminary result
• Tracking + FLUKA simulation for different β* and 5 TeV
=0.
5m
=0.
5m
=1.
1m
=5
m
=10
m
=0.
5m
,orb
.bu
mp
0.8
1.0
PêPquench
07/01/2009 R. Bruce, D. Bocian, AP Forum 36
6.50
07
Te
Ve
q.b* =
0.5
6.50
37
Te
Ve
q.b* =
0.5
6.50
37
Te
Ve
q.b* =
1.1
6.50
37
Te
Veq.b
*=
6.50
37
Te
Ve
q.b*=
6.50
35
Te
Ve
q.b* =
0.5
6.50
07
Te
Veqb* =
0.5
m,
0.0
0.2
0.4
0.6
Possible alleviation methods• with orbit bump we could gain >factor 5:
– possible to introduce orbit bump around BFPP impact
– particles lost at second dispersion max, with larger off-momentum β
– nominal orbit shifted by 2-3.8 mm depending on optics
• cold collimators (R.W. Assmann et al):
– could be installed at a later stage
6σ envelope at IP2, no kicks
07/01/2009 R. Bruce, D. Bocian, AP Forum 37
– could be installed at a later stage around IPs taking ion collisions
• at 5 TeV we gain a factor 3.5:
– lower field gives higher quench limit
– lower energy per BFPP particle
– larger geometric emittance gives larger spot size
– cross section only weakly energy dependent
• Increase of β*: not desired
6σ envelope at IP2, 4 kicks
Simulation uncertainties
• BFPP cross section: ~20%
• Changes in the optics (e.g. beta beating) could change the spot size: ~10%
• Network model: ~ 30%.
07/01/2009 R. Bruce, D. Bocian, AP Forum 38
• On top of this, uncertainty on the energy deposition from the FLUKA simulation, could in worst case be a factor 2. Dominating uncertainty for this specific beam loss but could be less in other cases.
• Background and introduction: Bound Free Pair Production
• Tracking: Distribution at the IP and at impact
• FLUKA simulation of shower in magnet
• Thermal network simulation (Dariusz)
– Thermodynamics of magnet structure
– Network model
OutlineOutline
07/01/2009 R. Bruce, D. Bocian, AP Forum 39
– Network model– Validation of the model– Steady state beam loss heat load simulation
• Comparison between optics version
• Simulation uncertainties
• Summary
Summary
• To make a detailed calculation of the quench limit for a main dipole due to a specific beam loss mechanism (BFPP during LHC Pb82+ ion operation), we have combined
– particle tracking,
– a FLUKA shower simulation of the heat load in a single magnet and
– a thermal network simulation of the heat flow in the magnet
07/01/2009 R. Bruce, D. Bocian, AP Forum 40
• BFPP creates one-electron Pb81+ ions at the IP, which follow an off-momentum orbit and are lost in the dispersion suppressor in the case of IP2.
• At nominal performance, the estimated heat load is expected to be very close to, and possibly above, the quench limit.
• For this loss distribution, the quench limit is a factor ~2 higher than calculated in LHC report 44 and LHC design report
• Possible alleviation methods include orbit bumps and cold collimators
Acknowledgements
We would like to thank people who have helped during the course of this work:
• A. Ferrari, M. Magistris and the rest of the FLUKA team
• E. Todesco, M. Lamm, G. Ambrosio
07/01/2009 R. Bruce, D. Bocian, AP Forum 41