Student: Jose Ferradas University of Geneva & CERN MDT
Transcript of Student: Jose Ferradas University of Geneva & CERN MDT
Student: Jose Ferradas University of Geneva & CERN MDT
Supervisors: Paolo Ferracin Lawrence Berkeley National Lab.
Carmine Senatore University of Geneva
Ezio Todesco CERN TE-MSC-MDT
27/03/2020
MDT Engineering Meeting – Scientific topics in the COVID-19 era
Main authors (In alphabetical order):H. Bajas, M. Bajko, L. Bianchi, L. Brouwer, B. Castaldo, M. Duda, S. Emami, F. J. Mangiarotti, M. Guinchard,
S. Izquierdo, J.V. Lorenzo, J.C. Perez, E. Ravaioli, E. Tapani Takala and G. Vallone
Jose Ferradas TroitinoMDT Seminar
The main objective of the doctoral research is to perform an exhaustive analysis of themechanical behavior of a Nb3Sn superconducting magnet during a quench.
The topic is tackled from three different sides:
1. Numerical simulation of the magnet mechanics during a quench using Finite ElementAnalysis.
2. Analysis of the experimental data collected during magnet cold tests’ for CERN magnets
Mechanical instrumentation
Electromagnetic instrumentation
3. Characterization of the conductor electro-mechanical limits at University of Geneva.
Study of the cable behavior from a single wire experiment. It includes:
Strands’ electro-mechanical characterization under transverse loads
Strands’ electro-mechanical characterization under axial loads
Stress-Strain measurements
PREFACE
Introduction to the Thesis’ Topic
Jose Ferradas TroitinoMDT Seminar
The main objective of the doctoral research is to perform an exhaustive analysis of themechanical behavior of a Nb3Sn superconducting magnet during a quench.
The topic is tackled from three different sides:
1. Numerical simulation of the magnet mechanics during a quench using Finite ElementAnalysis.
2. Analysis of the experimental data collected during magnet cold tests’ for CERN magnets
Mechanical instrumentation
Electromagnetic instrumentation
3. Characterization of the conductor electro-mechanical limits at University of Geneva.
Study of the cable behavior from a single wire experiment. It includes:
Strands’ electro-mechanical characterization under transverse loads
Strands’ electro-mechanical characterization under axial loads
Stress-Strain measurements
PREFACE
Introduction to the Thesis’ Topic
SEMINAR 1
Jose Ferradas TroitinoMDT Seminar
The main objective of the doctoral research is to perform an exhaustive analysis of themechanical behavior of a Nb3Sn superconducting magnet during a quench.
The topic is tackled from three different sides:
1. Numerical simulation of the magnet mechanics during a quench using Finite ElementAnalysis.
2. Analysis of the experimental data collected during magnet cold tests’ for CERN magnets
Mechanical instrumentation
Electromagnetic instrumentation
3. Characterization of the conductor electro-mechanical limits at University of Geneva.
Study of the cable behavior from a single wire experiment. It includes:
Strands’ electro-mechanical characterization under transverse loads
Strands’ electro-mechanical characterization under axial loads
Stress-Strain measurements
PREFACE
Introduction to the Thesis’ Topic
SEMINAR 2
Jose Ferradas TroitinoMDT Seminar
The main objective of the doctoral research is to perform an exhaustive analysis of themechanical behavior of a Nb3Sn superconducting magnet during a quench.
The topic is tackled from three different sides:
1. Numerical simulation of the magnet mechanics during a quench using Finite ElementAnalysis.
2. Analysis of the experimental data collected during magnet cold tests’ for CERN magnets
Mechanical instrumentation
Electromagnetic instrumentation
3. Characterization of the conductor electro-mechanical limits at University of Geneva.
Study of the cable behavior from a single wire experiment. It includes:
Strands’ electro-mechanical characterization under transverse loads
Strands’ electro-mechanical characterization under axial loads
Stress-Strain measurements
PREFACE
Introduction to the Thesis’ Topic
SEMINAR 2
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
New high-field Nb3Sn accelerator magnets arepushing the boundaries of magnet design andquench protection towards new limits.
Their large stored energy and current densities posenew challenges for the community…
INTRODUCTION
Quench processes and superconducting magnets
Courtesy of L. Bottura
Courtesy of L. Bottura
Scaling of the energy per unit length with the bore
field
Maximum magnetic field for state of the art
superconducting accelerator magnets
Jose Ferradas TroitinoMDT Seminar
… but also the electro-mechanical limits of the conductor become a parameter of extremeimportance.
INTRODUCTION
Quench processes and superconducting magnets
60 80 100 120 140 160 180
0.0
0.5
1.0
1.5
2.0
2.5
F = 0.5 kN
F = 15 kN
unload to 0.5 kN
@ 4.2 K, 19 T
Elec
tric
fie
ld [
V/c
m]
Current [A]
Example of Nb3Sn PIT strain sensitivity for a round strand tested under transverse loads:
Group of Applied Superconductivity
University of Geneva
Jose Ferradas TroitinoMDT Seminar
The coupling of the above-mentioned characteristicsduring a quench is a case of special interest that addsnew complexity for the design of superconductingmagnets beyond 10 T.
INTRODUCTION
Quench processes and superconducting magnets
Thermo-mechanical stressesQuenchRisk of magnet damage / conductor degradation
The study of quench processes and their intrinsic
mechanics becomes essential!
We need tools capable of predicting the mechanical response of the magnet during a quench
Digression on thermal stresses: Thermal buckling in railway tracks
Source: [Business Insider]
Courtesy of L. Bottura
Jose Ferradas TroitinoMDT Seminar
At the same time one cannot forget that the accurate simulation of quench transients is areal multiphysics effort:
- Electromagnetic study ~ Safe voltage levels
- Thermal analysis ~ Safe temperature level
- Mechanical analysis ~ Safe stress levels
Important lessons from the past…
INTRODUCTION
Quench processes and superconducting magnets
This is the result of a chain of events triggered
by a quench in an LHC bus-bar
Courtesy of Marta Bajko
Damage in HL-LHC coil as a result of an
electrical fault
Courtesy of Paolo Ferracin
This is the result of a quench in
the pre series magnet during
its qualification test
Courtesy of Marta Bajko
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
Early 2000s – Development of the first APDL models in the other side of the ocean:
Lawrence Berkeley National Lab:
2003 – S. Caspi et al.: “Calculating Quench Propagation With ANSYS®”
2004 - P. Ferracin et al.: “Thermal, Electrical and Mechanical Response in Nb3Sn Superconducting Coils”
3D Coupled thermal-electric transient analysis. Conductor block with smeared material properties. Transition is defined asa jump in resistivity at Tcs. Not varying magnetic field was included.
Fermi National Lab:
2001 - Yamada et al.: “Quenches and resulting thermal and mechanical effects on epoxy impregnated
Nb3Sn high field magnets”
2002- Yamada et al.: “ 2D/3D Quench Simulation using ANSYS for Epoxy Impregnated Nb3Sn High Field
Magnets”
2003 -Yamada et al.: “ 3D ANSYS Quench Simulation of Cosine Theta Nb3Sn High Field Dipole Magnets”
Only the thermal part is solved in ANSYS. At the end of each time step, solution is stopped and the electric part is solvedwith implemented analytical formulas. Then the solving is restarted.
FE Modelling
The magnet community needs tools capable of predicting the mechanical response of the magnets during a quench…
Often, new is forgotten old!
Jose Ferradas TroitinoMDT Seminar
Later in Europe…
CERN:
2018 - Jose Vicente Lorenzo, Hugo Bajas et al.: “Quench Propagation Velocity and Hot SpotTemperature Models in Nb3Sn Racetrack Coils”
Sort of combines the two previous. 3D Coupled-physics transient analysis is solved. At the end of each time step,solution is stopped and coded analytical formulas add some of the quench physics that are not possible tomodel with commercial ANSYS. Varying magnetic fields and other phenomena can be included.
University of Tampere (T. Salmi, A. Stenvall and J. Zhao):
2017 - J. Zhao et al., “Mechanical behavior of a 16 T FCC dipole magnet during a quench.,”2018 - J. Zhao et al., “Mechanical stress analysis during a quench in CLIQ protected 16 T dipolemagnets designed for the future circular collider.,”
In the framework of EuroCircol, the method combines 2D COMSOL simulations from CERN MPE with 2Dmechanical simulations in ANSYS.
Again recently in the US…
Lawrence Berkeley National Lab:
L. Brouwer – Developed special ANSYS User Defined Elements. They allow to incorporate intoANSYS APDL solver the most important physics of superconductivity! A fantastic tool, 2D for themoment.
FE Modelling
Jose Ferradas TroitinoMDT Seminar
Mechanical model
Thermal-electric model
Electro-magnetic
model
Reviving the idea of performing complete 2D and 3D analysis ofthe magnet mechanics during a quench, using commercialANSYS APDL at CERN.
The HL-LHC MQXF Low-β quadrupole magnet, our playgroundfield.
First, magnet protected with quench heaters (placed in the outer layer of the coils). Then, add other protection systems (CLIQ). Characterize the magnet response.
The method should combine the necessary ANSYS multi-physicsmodels. They will provide the input for the mechanical model.
The campaign on electro-mechanical characterization of Nb3Sn strands should determine the limits at which the magnet can operate safely.
Our approach
3D MQXF magnet mechanical model(G. Vallone, 2017)
FE Modelling
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
Nominal gradient: 132.6 T/m
Nominal current: 16470 A
Peak field in the conductor at Inom : 11.4 T
Aperture: 150 mm
Magnet outer radius: 630 mm
Stored energy at Inom: 1.17 MJ/m
Fx / Fy (per octant) at Inom = +2.47 / -3.48 MN/m
Fz (Whole magnet) at Inom = 1.17 MN
MQXF
MQXF Low-β quadrupole magnet
Courtesy of Giorgio Vallone
Mechanical concept: Aluminium shell pre-loaded with bladders.
First time for an accelerator magnet.
Jose Ferradas TroitinoMDT Seminar
Nominal gradient: 132.6 T/m
Nominal current: 16470 A
Peak field in the conductor at Inom : 11.4 T
Aperture: 150 mm
Magnet outer radius: 630 mm
Stored energy at Inom: 1.17 MJ/m
Fx / Fy (per octant) at Inom = +2.47 / -3.48 MN/m
Fz (Whole magnet) at Inom = 1.17 MN
MQXF
MQXF Low-β quadrupole magnet
Courtesy of Paolo Ferracin
Mechanical concept: Aluminium shell pre-loaded with bladders.
First time for an accelerator magnet.
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
ANSYS APDL Tools
Adds to the mechanical models alreadyavailable: multi-physics FE Models able toreproduce quench events.
Details in : [1] , [2]
Thermal-Electric (Quench Propagation) andMechanical models coupled together.
Inter-Filament Coupling Losses (IFCL)implemented. Further development possible.
Disclaimer: These tools have been fundamentallybuilt for the analysis of the magnet mechanics duringquench. They are not “another quench code”!
Quench
Heater Detail
The developed ANSYS APDL package:
[2] J. Ferradas Troitino et al, “On the magnet mechanics during a quench: 2D Finite element analysis
of a quench heater protected magnet”
[1] J. Ferradas Troitino et al, “3D Thermal-Electric Finite Element Model of a Nb3Sn Coil During a Quench”
Jose Ferradas TroitinoMDT Seminar
Modelling strategy
3D Thermal-Electric
quench simulation
E.M. forces from
2D/3D
Electromagnetic
model
Mechanical
model
2D/3D ROXIE
electromagnetic
model
Magnetic field on each conductor (Load Lines)
E.M. Forces
Temperature distribution
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
Overview on the multiphysics thermal-electric models
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
3D (1mm) MQXF Thermal-Electric symmetric model
Adiabatic boundary conditions
Direct coupled-physics elements
Surf./Surf. contact elements
Exploit the quadrupole symmetry.
Material properties as function of
temperature and magnetic field.
Conductor simulated as a block with
smeared properties.
IFCL included
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
Adiabatic boundary conditions
3D (1mm) MQXF Thermal-Electric full model
Same characteristics. Geometry
expanded.
Allows the study of failure cases and hot
spot effect.
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
The model could be validated using the extensive experimentalcampaign for MQXFS magnets:
Quench Integral (QI) tests
QI ~ Less than 10% difference w.r.t. experimental data in all tests
S5 – Low RRR
Measured QI is lower than simulations (Also
for other codes!)
S6b – Low RRR ~ S5
Lowest QI measured up to date.
High measured resistance for the coils with Bundle Barrier: very fast
discharge.
(Not included in the model)
S4 – Closest to REF
One HF strip not present in the
experimental case. This explains the fastest
discharge in the model.
Validated against experimental measurements Simulation results
obtained under the
model assumptions.
An “error bar” should be
always considered!
[2] J. Ferradas Troitino et al, “On the magnet mechanics during a quench: 2D Finite element analysis
of a quench heater protected magnet”
ANSYS APDL Tools
[3] H. Bajas “MQSXFS5 Test Report, EDMS: 2165441”
Jose Ferradas TroitinoMDT Seminar
The model could be validated using the extensive experimentalcampaign for MQXFS magnets:
Quench Heaters (QH) tests : MQXFS4
Simulation results
obtained under the
model assumptions.
An “error bar” should be
always considered!
Delay precision ~ Around 1 ms at Inom
Thermal behaviour matched
Validated against experimental measurements
[2] J. Ferradas Troitino et al, “On the magnet mechanics during a quench: 2D Finite element analysis
of a quench heater protected magnet”
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
It also gives an insight on how the reference conductor propertiesrepresent the different magnets produced:
Quench Heaters (QH) tests : All magnet tested
Simulation results
obtained under the
model assumptions.
An “error bar” should be
always considered!
Validated against experimental measurements
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
3D (Full length) MQXF Thermal-Electric full model
Full 3D coil geometry including coil ends
Topology simplified: QH are not included.
The delays to quench for each turn are
extracted from the 2D, and imposed here.
In view of 2D results, where the structure
remains cold, just the coil is simulated.
Adiabatic boundary conditions
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
3D propagation also extensively validated (ASC 2018)
QPV in a cable experiment
Simulation of a real training quench
[1] J. Ferradas Troitino et al, “3D Thermal-Electric Finite Element Model of a Nb3Sn Coil During a Quench”
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
Overview on the mechanical models
ANSYS APDL Tools
Jose Ferradas TroitinoMDT Seminar
2D & 3D MQXF Mechanical model
PLANE82 – structural 2D quadratic
Surf./Surf. contact elements (CONTA172/TARGE169).
Plane stress assumption, verified against 3D models and
measurements.
Exploit the quadrupole symmetry. Also full one available.
Linear elastic material properties.
Coil simulated as a block with smeared properties (isotropic).
Adds:
Half-magnet length, longitudinal
symmetry B.C.
Longitudinal loading
Courtesy of Giorgio Vallone
ANSYS APDL Tools
[4] G. Vallone et al., “Summary of the Mechanical Performances of the 1.5 m Long Models of the Nb3Sn
Low-β Quadrupole MQXF”
Jose Ferradas TroitinoMDT Seminar
Very good agreement along the experimental campaign, both 2D and 3D.
Assembly and magnet powering
Loading
Cool-Down
PoleShell
Mechanical models also deeply validated
ANSYS APDL Tools
[5] G. Vallone et al., "Modelling Coil-Pole Debonding in Nb3Sn Superconducting Magnets for Particle
Accelerators”
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
In last presentation, we showed the 2D analysis of MQXF magnet in nominal operation conditions…
Just small reminder of what we learnt here!
2D MECHANICS OF A QH PROTECTED MAGNET
Jose Ferradas TroitinoMDT Seminar
2D MECHANICS OF A QH PROTECTED MAGNET
Case study: Nominal current training quench
15 ms Detection + Validation time example
o Inner layer pole turn quenched(Highest peak field conductor)
Magnet protected with outer layer QH:
o Reference MQXF conductor parameters.o Results: 34 MIITS → Peak T during the disch. = 300 K
Adiabatic estimation of hot spot temp.
RRR =150Cu2SC = 1.2
Ref. Crit. Surf. Param. (5% degrad. cabling)
Jose Ferradas TroitinoMDT Seminar
2D MECHANICS OF A QH PROTECTED MAGNET
2D Mechanical results: Stress evolution
Focus on the pole region (The hottest block in this case):
o Additional azimuthal compression to the inner layer pole turn block
(Compared to Cool down). The plot below shows the interface between coil and pole.
Singularities coming from non-linear contact definition.
Better to study the stresses inside the block: (Next slide)Selected
Elements
Jose Ferradas TroitinoMDT Seminar
2D MECHANICS OF A QH PROTECTED MAGNET
2D Mechanical results: Stress evolution
Quench (End of the discharge) to Cool Down difference:
o Contours = End of the discharge – Cool down
Azimuthal
40 to 30 MPa additionalcompression to the poleblock !
Remember: To complete the study, we also performed a parametric study for analyzing the influence of quench position, conductor parameters, protection parameters, etc!
Jose Ferradas TroitinoMDT Seminar
2D MECHANICS OF A QH PROTECTED MAGNET
Three main conclusions
This was a reference case at nominal current, just protected with QH. Conservative.
For it, under the modelling assumptions, we could learn:
The peak stress during a quench in the coil block is not defined by the hotspot temperature. Rather than that, it is governed by an averagetemperature in the coil blocks:
Dissipated energy
In global terms, a performed parametric study shows that the mechanicalresponse for a quench at nominal current does not change if the designparameters are kept within the established tolerances.
Therefore, this response would be almost the same for all magnets tested up to now!
The quench transient adds an additional azimuthal compression to thecoil, due to the fast thermal expansion, which cannot be neglected.
Most of magnets tested up to now include a dump resistor (part of the energy extracted) or CLIQ (not yet studied): The exact peak stress value
will come later, just keep in mind the physics for now!
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
Model Validation
Can we validate these results?
Jose Ferradas TroitinoMDT Seminar
Model Validation
Jose Ferradas TroitinoMDT Seminar
MQXFS Mechanical Instrumentation: MME Mechanical Measurements Lab.
MQXFS magnets are instrumented with Strain Gaugesand Optic Fibers in the winding poles and Al Shell.
Strain Gauges were the baseline configuration formech. measurements in the magnet.
Synchronized FBG were added for validation purposesstarting with MQXFS5, now becoming also a baselineprocedure for MME Mechanical meas. lab.
An extensive campaign on mechanicalinstrumentation validation was performed, proving theaccuracy of the system down to 1.9 K.
The mechanical instrumentation was then used tovalidate the mechanical models and to obtainessential information from the magnet tests.
… nevertheless, the instrumentation has never beenused to study what happens during a quench.
The reason: quench is an extremely fast transient. Themechanical instrumentation was conceived from thebeginning for steady-state! We never intended to usethem for transitory phases! For example: temperaturecompensation issues!
Model Validation
[6] M. Guinchard et al., "Techniques of mechanical
measurements for CERN Applications and Environment"
[7] A. Chiuchiolo et al., "Strain Measurements With Fiber Bragg Grating Sensors in the Short Models of the HiLumi
LHC Low-Beta Quadrupole Magnet”
[8] M. Guinchard, L. Bianchi., "Mechanical Strain Measurements Based on Fiber Bragg Grating Down to Cryogenic Temperature – Precision
and Trueness Determination"
Jose Ferradas TroitinoMDT Seminar
Our new analysis indicate that we must re-consider these statements
The MQXFS “Quench” model can give important information on the system’s temperaturedistribution during a quench… and how temperature can affect the measurements!
For example, looking at an extreme case in terms of hot-spot (Inner pole turn quenched at Inominal,only QH protected):
After 0.5 s the magnet is fully discharged and the adiabatic “quench” model
predicts a temperature below 20 K where the SG/FBG are placed!
This should be very conservative since Helium cooling is neglected
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
What does the model predicts as pole and shell response during a quench?
For the sake of understanding, a generic model result is shown below. We introducethe concept of the “Quench delta”.
Preload
Cool Down
Powering
*Adiabatic, not in scale!
Quench
Delta in Stress due to Quench
The “Quench delta” definition: A GENERIC MODEL RESULT
Preload
Cool Down
Powering
Quench
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
What does the experimental measurements show?
An example: Strain Gauges and FBG measurements are plot together for MQXFS5,where both systems are available.
The agreement between both systems is almost perfect!
FBG and SG show the same delta!
But different behavior during re-
cooling!
Delta in Stress due to Quench
Difficult to see, but two lines per
color!
Model Validation
One coil selected
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
For the experimental signals. Essential to distinguish 3 different phases:
Keep in mind: Two systems (FBG 10x sensitive to temperature)
1. Assembly and magnet powering: Measurement system designed for this. (Scale of minutes)
2. Quench: Very fast transitory phase (Scale of milliseconds). Thanks to high. freq. acquisition
and based on the modelling result where the pole stays cold during the discharge, we can
use this data.
3. Re-cooling: Slower transitory phase (Scale of tenths of seconds). Definitely, the measurement
system is not designed for this: presence of high temperature effects. Not really trustful.
Model Validation
(1)
(2)
(3)
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Perfect agreement between systems
for the range of interest !(Green)
SG in black
FBG in magenta
TAKEOVER:
The nature of both system is
completely different. The fact that
they show the same delta confirms a
physical effect!
As we said, we neglect the re-cooling transitory phase. It is an
interesting topic for a separated meeting! The Mechanical
Measurements Lab. is currently studying this effect.
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Just a last point confirming our analysis:
The response of the compensator is coherent with the physics behind the transient.
It comes back to the cool down value after the quench in our time range of interest, just
the effect of magnetic field is removed! .
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
The excellent agreement between SG and FBG has been verified for:
1. Different quenches2. Measured strain (Compensated and not-compensated)3. Different acquisition systems
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
How both experimental and numerical results compare to each other?
Strain analysis during MQXFS5 Quench Integral tests
Magnet is ramped to a selected current level, and all OL-QH arefired with nominal powering parameters
Cleanest case for Experimental-Model comparison (Nouncertainty on quench location, no hot-spot, etc.)
FEM model catches accuratelythe behavior. Difference in theorder of less than 5 MPa.
Important to note that the thermalexpansion coefficient plays a bigrole in the results. Up to date thereare not dedicated measurementsfor our HL-LHC magnets.
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Measured and simulated delta in the pole confirms an extra compression forthe coil due to the thermal expansion during quench.
The shown delta concerns the pole, the only estimation of what the coil seescomes from the model!
Pole Az. Delta: 41 MPa / Pole Az. Stress @Quench = 154 MPa / Coil Az. Peak = 145 MPa
For the QI test at Nominal Current:
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Measurements on different magnets confirms the model outcome:
The mechanical response during a quench (for a coil block!) is not defined by the hotspot temperature. Rather than that, it is governed by an average temperature in thecoil blocks:
Dissipated energy / Coil enthalpy
Model Validation
Different magnets, different conductors (different QI) : Same energy stored, same mechanical response!
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Results of the 3D model (not shown in thepresentation) are consistent with the presented2D cases !Indeed, these QI tests can be seen as a pure 2Dcase of uniform coil temperature!
The model cannot be validated using real trainingquench data: Uncertainty on quench positionversus the mechanical instrumentation one.
However, since we have shown that the model isvalidated for the presented cases, the numericalpredictions must apply as well for the trainingquenches!
The yellow star is the result from the simulation ofa training quench in MQXFS5!
MQXFS5 Training
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Extremely interesting information from training quenches! And in this case, the 3D simulation shows the importance of the Hot Spot in quench
mechanics!
Analysis will be published soon…!
STAY TUNED!
Model Validation
MQXFS5 Training MQXFS5 QI tests
Jose Ferradas TroitinoMDT Seminar
SEMINAR 1 - OUTLINE
1.- Introduction – Quench processes and superconducting
magnets
6.- Model validation: Mechanical behavior during quench
3.- MQXF – The low-β quadrupole for HL-LHC
4.- 2D & 3D ANSYS APDL Tools
5.- 2D FE Analysis of a quench heater protected magnet
2.- Finite element modelling for the analysis of the
magnet mechanics during a quench
7.- Conclusions
Jose Ferradas TroitinoMDT Seminar
SEMINAR CONCLUSIONS
Main outcome
First of all, this is the result of a joint effort, which profits from the expertise of each team. Thanks to all!
The essential collaboration between MDT, MME and TF made possible this analysis!
Numerical simulations show that, during a quench, the thermal and
electro-magnetic transients generate new peak stresses in the coils.
These new stresses build-up on top of the magnet pre-load.
In the case of MQXF, they remain below 175 MPa for the cases
studied up to now. Further cases may need to be studied.
The extensive set of data from mechanical measurements acquired
during quench tests, can be used to validate the model outcome.
This last experimental data combines different measurement systems
(electrical SG and optical FBG), which show an almost perfect
agreement.
Last verifications are being performed, but with the data available
up to the date, experimental results confirm the model predictions.
Jose Ferradas TroitinoMagnet Design and Technology
Long experience with the mechanical models: Validated 2D
and 3D models
In 2D, different strategies to model the connection between the
pole and the coil
1. “Separation/no-sliding”
Separation allowed with very high friction always
2. “Glued”
Bonded contact
3. Only for quench simulations, a third option: “Low friction” quench model:
Separation allowed with very high friction during current ramp
Separation allowed with 0.2 friction during quench
Results are identical to case 1 (Only changes during quench)
Different strategies for coil material properties can also be
adopted: Linear elastic / Bilinear
Quench Modelling Tools
3D & 2D Mechanical Models in ANSYS APDL
Jose Ferradas TroitinoMagnet Design and Technology
Quench Modelling Tools
3D & 2D Mechanical Models in ANSYS APDL
UltimateNot Powered
The unloading mechanism in ANSYS:
Jose Ferradas TroitinoMagnet Design and Technology
“Glued”
Bonded contact
Quench Modelling Tools
3D & 2D Mechanical Models in ANSYS APDL
Ultimate
Ultimate
Jose Ferradas TroitinoMagnet Design and Technology
Quench Modelling Tools
3D & 2D Mechanical Models in ANSYS APDL
“Separation/no-sliding”
Separation allowed with very high friction always
Ultimate
Ultimate
Jose Ferradas TroitinoMDT Seminar
Our new analysis indicate that we must re-consider these statements
Additionally, below 50 K thermal strain for titanium is rather constant
At the end of the quench
discharge and at the position of
the SG/FBG:
The difference in strain using
the mechanical strain or the
total one (including thermal) is
less than 10 µstr in the ANSYS
model.
-2.00E-03
-1.80E-03
-1.60E-03
-1.40E-03
-1.20E-03
-1.00E-03
-8.00E-04
-6.00E-04
-4.00E-04
-2.00E-04
0.00E+00
0 50 100 150 200 250 300 350
Linear Expansion [m/m]
NIST
ANSYS
Under these assumptions and material properties, we claim that the
study of the measured compensated strain (or even the non-compensated
one) is meaningful and must be considered!
Model Validation
Jose Ferradas TroitinoMDT Seminar
Model and experimental results
Model Validation
5 µStr