1. New TCSPM Design Outcome of HRMT-23: Robustness of HL-LHC jaws
F. Carra*, A. Bertarelli, L. Gentini, J. Guardia, M. Guinchard,
L. Mettler, E. Skordis
Special ColUSM: Material and design readiness for LS2 productions
02/05/2017
*With many contributions from colleagues of EN/MME, EN/STI, BE/ABP, EN/HE, HSE/RP, SMB/SC, EN/EA

2. Outline TCSPM design scenarios HRMT-23 recap
LHC and HL-LHC materials: simulations and benchmarking
Conclusions
02/05/2017
F. Carra

3. TCSPM design scenarios
TCSPM: low-impedance secondary collimator, jaw in MoGr
Design cases  Same as LHC TCS (CFC):
Slow losses
Nominal operation: 1h BLT
Accidental case: 0.2h BLT (10s)
Direct beam impact (incident)
Asynchronous beam dump: 8 full LHC bunches impact
Beam injection error: 288 SPS bunches impact
Note that FLUKA simulations for the asynchronous beam dump case on HL-LHC materials are not available. For a CFC secondary, this has always been less critical than BJE  Check from FLUKA
For a tertiary, we are using equivalent SPS bunches to assess the asynchronous beam dump error  Check from FLUKA at 7 TeV
For a tertiary: 1 LHC bunch
02/05/2017
F. Carra

4. Direct beam impact incidents
HRMT-23 Goal: determine consequences of failure scenarios affecting machine performance for LHC Run 2, Run 3 and HL-LHC
Failure Scenario
Beam Type
Beam Energy [TeV]
Intensity Deposit. [p+]
Beam Emittance [m]
RMS beam size [mm]
Injection Error
LHC Ultimate
0.45
4.9e13
3.5 1
Run 2 BCMS
3.7e13
1.3
0.61
HL-LHC
6.6e13
2.1
0.77
Injection Error (BIJ)
LIU BCMS
5.8e13
Asynchronous Beam Dump
BCMS Run 2 7
1.3e11
~0.5
Asynchronous Beam Dump (ABD)
2.3e11
~0.6
Demonstrate the viability of a low-impedance collimator solution
Address the issue of TCT robustness limit
Demonstrate the robustness of present carbon-based collimators (TCS, TCP) against injection failures with smaller emittances
02/05/2017
F. Carra

5. HRMT-23 experiment recap
Three separate complete jaws
TCSPM MoGr jaw with MoGr tapering
TCSPM CuCD jaw with MoGr tapering
TCSP CFC jaw with Glidcop tapering
Extensive instrumentation
126 strain gauges (4 MHz)
42 temperature probes (200 Hz)
60 optical fiber Bragg gratings (500 Hz)
LDV, water pressure sensor, high speed / HD camera, ultrasound probes, etc.
Stainless steel vacuum vessel (p > 10-3 mbar)
Be/CFC vacuum windows
Horizontal actuation stroke (H) 35 mm
Vertical actuation of tank stroke (V) +/-140 mm
02/05/2017
F. Carra

6. HRMT-23 experiment beam parameters
Test Runs: July 2015
Beam energy: 440 GeV
Bunch spacing: 25 ns
Protons/bunch: up to 1.32e11
1 to 288 bunches per pulse
Beam size (): 0.35 to 1 mm
Different impact positions
Total Pulses: 100 (excluding alignment)
Total Bunches: 8110 (excluding alignment)
Total Protons: ~ 1e15
Equivalence SPS/LHC in terms of damage for Asy. Dump (Inermet)  24 bunches
HL-LHC  48 bunches
HL-LHC injection error on MoGr:
5413 J/cm3
HRMT-23 max energy on MoGr:
5659 J/cm3 (+5%)
HL-LHC injection error on CFC:
2443 J/cm3
HRMT-23 max energy on CFC:
3158 J/cm3 (+29%)
12 April 2016
Linus Mettler - HL-LHC Collimator Tests at CERN

7. MoGr and CFC experimental results
MoGr on HL-LHC jaw survived the impact of several 288 b pulses with  down to 0.35 mm (peak energy density 5% higher than HL-LHC injection error)
CFC on LHC jaw survived the same impacts
Suggests that MoGr qualifies as an alternative to CFC in terms of robustness with a factor 5 to 10 gain in electrical conductivity
A hole in the TCSP Glidcop tapering was observed, while the two TCSPM jaw taperings, in MoGr, are visually unscathed  MoGr is a more robust option as a tapering material also for TCSP
MoGr after multiple impacts
MoGr tapering intact
CFC after multiple impacts
Damaged Glidcop tapering
Glidcop melting (red)
02/05/2017
F. Carra

8. CuCD experimental results
CuCD on HL-LHC jaw survived (with a limited surface scratch on the Cu coating) the impact of 24 b,  0.35 mm at 440 GeV, roughly equivalent to 1 LHC bunch at 7 TeV
At 48 b (~2 LHC 7 TeV bunches) the scratch is more severe, but the jaw appears globally undeformed, roughly equivalent to 1 HL-LHC bunch at 7 TeV
This would qualify CuCD as an superior material for TCT jaws (presently in Tungsten alloy). Local damage induced by Asynchronous Beam Dump could be compensated by jaw shift with 5th axis
Both MoGr taperings ok, no trace of high temperature on the BPM buttons
CuCD surface damage
24 b, σ 0.35 mm, impact 0.5 σ
MoGr tapering intact
144 b, σ 0.61 mm, impact 3.05mm
48 b, σ 0.35 mm, impact 0.5 σ
02/05/2017
F. Carra

9. HRMT-23 Post-irradiation – November 2016
HRMT-23 stored at b. 954, sealed, until November 2016
When the dose rate reached acceptable values (80 mSv/h at contact with the downstream flange), RP gave the go-ahead for the dismounting
Closed bunker needed: impossible to exclude a prior the risk of contamination. Dismounting done at b. 867-R-P58. Performed by S. Favre (EN-MME-MA) with support from E. Berthomé. K. Weiss and C. Saury from RP
Collective dose for the intervention: ~10 mSv. No contamination measured.
Jaws inserted into plastic bags and CR code applied on all, then registered in TREC
Then placed on a pallet, also registered in TREC as the container of the three items (HCPWPAB001-CR000002)
Tank closed, everything re-transported to b. 954
Many thanks to the colleagues involved in these operations! (EN/MME, EN/STI, EN/HE, HSE/RP, SMB/SC, EN/EA, BE/ABP, …)
02/05/2017
F. Carra

10. HRMT-23 Post-irradiation – April 2017
The disassembling of the tank, for the recuperation of the table and other delicate components (connectors, switch box, glasses, LVDTs, motors, ...) to be re-used in Multimat was cone in April 2017 at b. 109.
All the unrecovered equipment (Be windows, tank, pump) is treated as RP waste.
Testing program of the three jaws (first non-destructive, then destructive) launched – G.Gobbi working on this.
At this stage, preliminary conclusions given at the LMC in August 2015 (failure of Glidcop tapering, survival of MoGr tapering, survival of MoGr and CuCD absorbers to the respective failure scenarios) are confirmed
02/05/2017
F. Carra

11. Simulations Simulations with linear elastic models
HL-LHC intensity beam injection error simulated
For CuCD, asynchronous beam dump equivalent to 1 LHC bunch, same case as Inermet180
Simulation parameters CFC and MoGr:
288 bunches
6.4∙1013 total protons (HL-LHC)
440 GeV
7.2 µs impact duration
Beam sigma 1 mm
Emax,CFC ~ 60% LIU-BCMS; Emax,MoGr ~ 80% LIU-BCMS
Impact depth 5 mm
Elastic Constants
CFC
MoGr
CuCD
Ex [GPa]
2.8
7.1
160.5
Ey [GPa]
57.5
74.0
Ez [GPa]
93.0
Gxy [GPa]
3.5
4.3
75.1
Gxz [GPa]
6.4
Gyz [GPa]
10.6
31.3
nxy
0.11
0.13
0.07
nxz
0.10
nyz
0.19
Simulation parameters CuCD:
24 bunches
3.1∙1012 total protons (eq. 1 LHC bunch damage)
440 GeV
1.95 µs impact duration
Beam sigma 0.61 mm
Impact depth 3.05 mm
02/05/2017
F. Carra

12. Normal strains on CFC jaw
CFC simulation: BIJ
Reference values: M. Borg, Numerical Modelling and Experimental Testing of Novel Materials for LHC Collimators, master thesis
Factor of 2 in resistance between Y and Z  orthtropic
With this model, mostly critical is the high tensile strain in Y direction
Reference System
Normal strains on CFC jaw
Maximum over time
Simulation
[microstrain]
Reference value
Ultimate strain ex
2000
2600 ey
1400
850 ez
740
1800
ey on CFC
ey>1000
microstrain
02/05/2017
F. Carra

13. Normal strains on MoGr jaw
MoGr simulation: BIJ
Reference values: M. Borg, Numerical Modelling and Experimental Testing of Novel Materials for LHC Collimators, master thesis
Higher strains than CFC jaw, but similar ratio against reference ultimate strain
Again, Y is the most critical direction
Reference System
Normal strains on MoGr jaw
Maximum over time
Simulation
[microstrain]
Reference value*
Ultimate strain ex
4500
5200 ey
6500
2000 ez
1800
ey on MoGr
02/05/2017
F. Carra

14. CuCD simulation: ABD
Here the assessment of material failure is fairly easy: for impacts close to the surface, mechanism controlled by material melting
HRMT14 outcome: at these energy levels, the beam sigma contribution is very weak, what matters is the energy integral.
Damage recoverable by 5th axis
24 b, σ 0.35 mm, impact 0.17mm
144 b, σ 0.61 mm, impact 3.05mm
Significant difference between CuCD and Inermet. For estimating the precise factor of robustness gain, we would need to compare two identical scenarios in terms of sigma, impact depth and intensity. Rough estimation:
48 b, σ 0.35 mm, impact 0.17 mm
02/05/2017
F. Carra

15. Comparison numerical model / experimental measurements
Simulations
Conservative. Why?
CFC, graphite and MoGr are not linear in σ/ε. Energy dissipation related to internal friction  should be simulated as viscoelastic materials
Comparison numerical model / experimental measurements
Graphite dynamic test
𝐺 𝑡 = 𝐺 ∞ + 𝐺 0 𝑒 − 𝑡 𝜏
Model built for R4550 graphite1, ongoing for MoGr (more complex: transverse isotropy)
CuCD plasticity
CuCD: also non linear, because of plasticity of the matrix + also internal friction! (see next slide).
1L. Peroni, M. Scapin, F. Carra and N. Mariani (2013). Investigation of dynamic fracture behavior of graphite. In B. Basu, Damage assessment of structures X, Trans Tech Publications Inc, pp
02/05/2017
F. Carra

16. CuCD simulation: at intensity ~1 LHC bunch
HRMT shot #124: CuCD 24 bunches, s 0.61 mm, impact 5s
Pseudo-plasticity of the material taken into account!
EM disturbance
wave propagation
noise
simulation
CuCD block5 vertical strain back face
45.8 kHz
64.1 kHz
Numerical overestimation of the results when applying a linear elastic model is evident.
Plastic+ viscoelastic model needed for CuCD!
63.4 kHz
45.0 kHz
Why observed decrease in amplitude with time? Damping
31 January 2017
Federico Carra

17. MoGr simulation: lower intensity
HRMT shot #200: MoGr 24 bunches, s 0.6 mm, impact 5s
Orthotropic elasticity simulated up to now, must evolve into orthotropic viscoelasticity
wave propagation
noise
simulation
Dynamic tests at Polito for the construction of the strength model  launched
02/05/2017
F. Carra

18. When failure fails to predict failure...
Let’s have a look at the key ingredients of a simulation:
Equation of state ✔
Strength model –
Failure model x
Threshold 2 TCT: limit for spallation
Numerical failure does not mean functional failure
Example: CuCD failing locally (melted), but overall the structural resistance of the material is guaranteed  damage can be recovered with 5th axis
Fragmentation, spallation acceptable if the extent is limited and recoverable
But ANSYS does not distinguish between numerical and functional failure!
If we want to explore the acceptable damage, more sophisticated methods required  Autodyn SPH mesh, as we did for Inermet180
02/05/2017
F. Carra

19. Conclusions
Jaws tested in HRMT-23 performed well (MoGr surviving energy density of HL-LHC injection error and CuCD surviving asynchronous beam dump)
TCSP Glidcop tapering melted, all TCSPM taperings in MoGr remained intact
Non destructive tests ongoing and destructive testing on irradiated blocks foreseen → to examine damage, amount of plasticity, loss of thermophysical properties due to Cu / CD debonding, etc.
What if the beam σ was higher than expected in the HRMT23 shots?
(Almost) irrelevant for CuCD  high energetic impact with melting, damage depends more on the integral of energy in the impacted volume than on the energy peak. Also, CuCD was tested up to intensities 6 times higher than the nominal scenario
Relevant for the onset of damage for MoGr and CFC, but:
Several high-intensity shots on the jaws  statistics
Hard to believe that the functional damage on the jaw can be more relevant than what we had on CuCD, with melting involved…
The same scenarios will be tested again in Multimat, with new BTV/BPKG system, as well as solutions such as coating, monitoring and actuation system, FCC materials, …
02/05/2017
F. Carra

20. Conclusions
Simulations with elastic material models are over-conservative, since they ignore plasticity, damping, and other nonlinear phenomena observed in laboratory tests
As a consequence, the local failure predicted for CFC and MoGr even with larger σ than HL- LHC scenarios were not observed in HRMT-23
Nonlinear models required, to be built with dynamic tests:
Inermet180, Molybdenum, R4550 graphite: completed with Politecnico di Torino
MoGr: launched at Politecnico di Torino
CuCD and CFC: to be launched
Complemented by high-temperature measurements at MME MechLab  new machine equipped for that
If one wants to examine the fragmentation process  SPH method with Autodyn needed
Fracture modelling becomes more and more important transitioning towards future high intensity accelerators (not only HL-LHC, but also FCC)
FLUKA Asynchronous beam dump simulations on CuCD, MoGr, CFC needed. Simplified case is sufficient (bunches impacting in the same spot) to compare CuCD robustness with that of Inermet180 and to verify that the scenario is less relevant than beam injection error on MoGr and CFC
02/05/2017
F. Carra

21. Thanks for your attention

22. Backup slides

23. CFC simulation: temperature
Maximum temperature reached on the CFC jaw: 685 ˚C
Maximum temperature in the instrumented points ~170 ˚C
Temperature at the impact
Beam
02/05/2017
F. Carra

24. CFC simulation: shockwaves
The frequency of the shockwaves to be acquired ranges from 3 kHz (longitudinal) to 45 kHz (transversal); maximum strains to be acquired ~3000 microstrain
Jaw flexural oscillation has a frequency of ~80 Hz and a maximum amplitude of ~1 mm
Transverse velocity of the external surface ~5 m/s
300 ms
Reference System
02/05/2017
F. Carra

25. CFC simulation: Glidcop support
The compressive stress provoked by the impact on the Glidcop plate exceeds the material yield strength
Equivalent plastic strain is ~0.17%
This compressive state remains once the jaw comes back to room temperature  permanent deflection induced ~50 mm
Jaw permanent deflection
Beam
02/05/2017
F. Carra

26. MoGr simulation: temperature
Maximum temperature on MoGr ~1200˚C, much lower than melting point (~2590˚C)
The temperature peak of 1290˚C is found on the Aluminium tapering  local melting
Another calculation performed with Glidcop tapering still shows T>Tmelt  alternative solution to be found (proposal: MoGr tapering)
On the strain gauge closer to the beam impact, T~360˚C > Tlim
OK if this occurs only at the very last shot: the strain gauge will register the dynamic phenomenon before being detached by the incoming thermal wave
This model uses aluminium tapering and ultimate strain from different grade
Tmax<Tmelt,MoGR
Al tapering
melting
Temp (˚C)
02/05/2017
F. Carra

27. MoGr simulation: shockwaves
Maximum temperature on MoGr ~1200˚C, much lower than melting point (~2590˚C)
The frequency of the shockwaves ranges from 2.5 kHz (longitudinal) to 50 kHz (transversal); maximum strains around 5000 microstrain
Jaw flexural oscillation frequency of ~125 Hz and a maximum amplitude of ~2.1 mm
Transverse velocity of the external surface ~12 m/s
20 ms
This model uses aluminium tapering and ultimate strain from different grade
Reference System
02/05/2017
F. Carra