1. Long term behavior and high MIITs test in the LARP program
GianLuca Sabbi – LBNL
WP3 Technical Meeting – May 23, 2019

2. LARP Model Magnet Reference Chart
Technology Development
Large aperture quadrupoles
HQ and LQ Mirrors
Long quadrupoles
HQM
WP3 Technical Meeting, May 23, 2019

3. Fabrication and Test Database
Test facilities: LBNL (11 tests); BNL (2 tests); FNAL (28 tests); CERN (8 tests, funded by CERN)
(#): includes coil exchanges with previously used coils, full reassembly with same coils, or pre-load adjustments
(*): includes contributions from FNAL GARD program; (**): includes contributions from LBNL GARD program
For this presentation:
Long term reliability assessment is based on TQS03 models (where focused studies were performed) and additional data from SQ, LR, HQ and mirror tests
Quench temperature limits assessment is based on HQ02a/a2/b and TQS01c studies
WP3 Technical Meeting, May 23, 2019

4. Long term reliability: TQS03 studies
A series of tests was performed on TQS03 (coil 30 to 33) to investigate robustness and long term reliability in several key areas:
Performance after multiple pre-load and thermal cycles
Performance under/after high coil stress (pre-load and operation)
Average cold preload: 120 MPa (TQS03a & d); 160 MPa (TQS03b); 200 MPa (TQS03c)
Performance after a large number of powering cycles (1000)
Overview of TQS03 load cycles:
After cool-down (0 kA)
12 kA (91% SSL)
200 MPa
200 MPa
Ref: H. Felice et al., IEEE Trans. Appl. Supercond., vol. 21, no. 3, June 2011.
WP3 Technical Meeting, May 23, 2019

5. Validation of TQS03 Mechanical Analysis
Comparison with strain gauge measurements: shell, axial rods and coil poles
Ref: H. Felice et al., IEEE Trans. Appl. Supercond., vol. 21, no. 3, June 2011.
WP3 Technical Meeting, May 23, 2019

6. TQS03 Quench Performance
Four tests performed at CERN in TQS03 models with different pre-load
Balance speed of training and consistent plateau after thermal cycle vs. degradation
Increasing pre-load in a/b/c gives more stable but slightly lower plateau (93/91/88%)
Degradation is permanent: TQS03d with lower pre-load does not recover initial level
Good performance in a broad range of azimuthal stress with peak (calculated) up to 260 MPa
91%
93%
WP3 Technical Meeting, May 23, 2019

7. TQS03 cycling test at CERN
Performed 1000 cycles from 6 kA to 11 kA with 50 A/s ramp rate
Control quenches every ~150 cycles
No change was observed in mechanical parameters or quench levels
Cycling test was performed after four pre-load/thermal cycles, including studies of performance under high stress, and more than 70 quenches
Control quenches are stable within ±25 A
H. Bajas, M. Bajko, G. DeRijk, H. Felice, A. Milanese et. al
WP3 Technical Meeting, May 23, 2019

8. LARP Magnets reliably above 88%
Notes for “Quenches” column: tests performed with the same coils are grouped in dotted boxes; sums show thermal cycles.
WP3 Technical Meeting, May 23, 2019

9. Degradation and Memory after Thermal Cycles
LARP data on performance changes after a thermal cycle is limited
Few tests included two thermal cycles: priority on exploring variants and improving performance
Each thermal cycle included training at different temperatures, protection studies etc.
Available data mostly at 4.5K due to test infrastructure limitations until the later phase of LARP
No indication of performance degradation after a thermal cycle from the available data:
Data from thermal cycles that included re-assembly or pre-load changes with same coils also shows no indication of performance degradation (except in cases where high stress or high MIITs studies were performed between cycles)
In previous slide: SQ02a  SQ02b; LRS01  LRS02; TQS03e  TQC03E; HQ02a  HQ02b
No indication of detraining after a thermal cycle, but relevant data is even more limited since only cos2q quads in shell structure are applicable (in addition to the above issues)
WP3 Technical Meeting, May 23, 2019

10. LARP “Technology Development” Tests
Long term reliability studies need to be decoupled from performance limitations due to design or fabrication flaws
WP3 Technical Meeting, May 23, 2019

11. Protection limits studies
Goal: characterize how the quench temperature affects subsequent magnet performance
Approach:
Correlate with key physical quantities affecting magnet performance (hot spot temperature, stress etc.)
Adjust protection system parameters to achieve a controlled increase of the Quench Integral  I2(t) dt
Correlate with subsequent magnet quench levels (detraining or permanent degradation)   *
I(t)
Delay
(var.)
Decay (~const.)
WP3 Technical Meeting, May 23, 2019

12. General procedure for High MIITs testing
Performance characterization before and after high MIITs is the key challenge
Desired conditions:
Reproducible quenches at the highest field location (inner layer pole turn) as close as possible to SSL
High detraining threshold (and fast training/retraining)
A PERFECT MAGNET!
WP3 Technical Meeting, May 23, 2019

13. Quench Limits Studies: HQ02 Models
HQ02 = second-generation high field quadrupole incorporating critical improvements in the coil design, fabrication and QA processes
Coil fabrication by LBNL and BNL, assembly at LBNL
First assembly was tested twice at FNAL with different setups
HQ02a: kA current limit, 1.9K-4.5K temperature
HQ02a2: No current limit but 2.2K-4.5K temperature
Based on the HQ02a2 feedback, the coil pre-load was increased in the HQ02b assembly by ~20 MPa, and tested at CERN
HQ02b: No current limit and 1.9K-4.5K temperature
The combined tests provided a comprehensive performance characterization including extensive quench protection studies
WP3 Technical Meeting, May 23, 2019

14. HQ02 (a, a2, b) Quench Performance
HQ02a: At 4.5 K, above 80% in two quenches, but slow training beyond that
HQ02a2: training rate: 30 A/quench; 98% SSL at 4.5K after training at 2.2K
HQ02b: training rate: 230 A/quench; 95% SSL at 1.9 K with 200 MPa pre-load
HQ02b
HQ02a-2
HQ02a
WP3 Technical Meeting, May 23, 2019

15. HQ02 Quench Integral vs. Temperature
Quench integral vs. temperature calculation by QMIITs & QLASA (T. Salmi, V. Marinozzi)
Material properties from MATPRO (G. Manfreda et al.) using individual coil parameters
Estimates based on HQ test data are K above adiabatic heat balance (TASC# )
WP3 Technical Meeting, May 23, 2019

16. HQ02a2 Quench Integral Study
98% SSL
10.7 MIITs
96% SSL
-120A
12.4 MIITs
~9 MIITs
-430A
~9 MIITs
HQ02a2 reached 98% 4.5K with limiting quenches at high field (pole turn)
Slow training above 15 kA and detraining at the MIITs level, ~150K
Detraining and slow retraining deemed incompatible with degradation studies
WP3 Technical Meeting, May 23, 2019

17. HQ02b Quench Integral study
Increased pre-load resulted in less detraining and faster retraining after High MIITs quenches
After multiple quenches in the range of MIITs ( K) degradation of the quench plateau was less than 2% (from 95% to 93% at 4.3K, without full retraining due to schedule constraints)
WP3 Technical Meeting, May 23, 2019

18. Analysis of 24 MIITs Quench
Comparison between 24 MIITs spot heater quench #18 and verification #20 at 4.3K
HQ02b-18
Value
Current
6.0
Coil 17
Segment
A9A10
Field [T]
5.1
Q.I. [MIITs] 24
Tmax [K]
383
HQ02b-20a
Value
Current (kA)
15.38
Coil 17
Quench segment
A9A10
Field A9A10 [T]
12.1
Iq/Iss (4.3K)
0.93
XS Ic
meas.
Additional retraining and 4.3K verification would have been needed to demonstrate permanent degradation or provide a lower constraint
WP3 Technical Meeting, May 23, 2019

19. Summary of HQ02 High MIITs Studies
Detraining vs. MIITs (from HQ02a & HQ02b):
Not a fundamental issue or a focus of the high MIITs study, but it negatively affects the high MIITs study due to time required for retraining (in order to recover baseline quench level or assess permanent degradation)
Higher preload in HQ02b resulted in better training performance with less detraining due to high MIITs and faster retraining compared to HQ02a
Degradation vs. MIITs (from HQ02b):
Relative degradation:
Less than 2% after multiple high MIITs quenches up to 24 MIITs (380 K)
From 95% SSL at the start of the study to 93% at the end (4.3K)
Absolute degradation
A9A10 segment performs above 93% following 24 MIITs (380 K) quench
Hot spot temperature from experimental current/voltage data was ~50K higher
WP3 Technical Meeting, May 23, 2019

20. TQS01 Models TQS01a TQS01b TQS01c Coils 5,6,7,8 7,8,14,15 5,7,8,15
Test
4.5K
(LBNL)
4.5K (LBNL)
4.5K & 1.9K (FNAL)
Max.
current (A)
(4.4K)
(3.2K)
9932 (4.4K)
9600 (4.4K)
10521(1.9K)
Iss (%)
87% (4.4K)
82% (4.4K)
80% (4.4K)
80% (1.9K)
TQS01a quenches near gaps in bronze pole
Limited by degradation due to high axial strain
Pole gaps
Spikes in coil axial strain
(cool-down & excitation)
Bronze pole
(TQS01/b/c)
Ref: S. Caspi et al., IEEE TASC 17, no.2, p , June 2007
WP3 Technical Meeting, May 23, 2019

21. TQS01c Training and Quench Studies
Training at 4.5K and 1.9K
High MIITs study
All limiting quenches in coil 15 A6-A8 (pole turn)
Temp. dep.
8.7
9.5
Quench limits study conditions and main results:
Quench integral was increased from 6 to 9 MIITs by increasing heater and extraction delays
At the ~8 MIITs level: initial ±3% variation, then a +3% plateau and then some degradation
Stronger and irreversible degradation of 7.4% (25%) after the 8.7 (9.5) MIITs quenches
Ref: G. Ambrosio et al., TQS01c Test Report, Fermilab TD , May 2007
WP3 Technical Meeting, May 23, 2019

22. TQS01c Degradation vs. Temperature
The hot spot temperatures are calculated from the quench integral in the adiabatic approximation using QuenchPro (P. Bauer, TD ) using the relevant material fractions, RRR, and coil field
Main results and conclusions:
No significant changes below 340K; reversible variations of a few percent in the 370K-400K range; large irreversible degradation above 400K-450K
Results are consistent with HQ study and support selecting a 350 K maximum temperature
Ref: G. Ambrosio, WAMSDO Workshop Proceedings, CERN
WP3 Technical Meeting, May 23, 2019

23. Summary Long term reliability:
LARP optimized models demonstrated very robust behavior with respect to high pre-load, multiple assemblies, quench training and powering cycles
However, any design or fabrication flaws -- conductor, coil fabrication, mechanical assembly etc. -- can easily lead to degradation in magnet performance and these effects can exhibit a progressive behavior
Quench Limits:
Test performed in HQ and TQ did not show performance degradation below 350 K (including some margin to account for uncertainties in performance characterization and hot spot temperature estimate)
HQ tests indicate that the magnet can perform above 90% SSL after experiencing hot spot temperatures approaching 400 K
Quench limit studies are very demanding on facilities and resources, however, a more complete characterization program would be a worthwhile investment to reduce the present uncertainties
WP3 Technical Meeting, May 23, 2019