Day 1 – Field Quality of the 11-T Dipole

Introduction and Scope of the Review

11 T Project Overview and Status – M. Karppinen

Content

1. Project setup: Joint CERN-FNAL project, building upon >10 years of experience with Nb₃Sn coils, mainly from US HF magnet programs, as well as LHC experience at CERN.
2. R&D phase planning:
   1. Q1 2012: 2-m-long single-bore demonstrator at FNAL to prove quench performance and learn about coil magnetization and magnet protection.
   2. Q4 2012: 2-m-long 2-in-1 demonstrator at FNAL, emphasis on field quality.
   3. Q2/3 2013: 2-m-long 2-in-1 demonstrator at CERN, field quality and reproducibility.
   4. Q4 2014: 5.5-m-long 2-in-1 prototype, aperture 1 from FNAL, aperture 2 from CERN, cryostat assembly and cold testing at CERN. Demonstrate the scale-up and the feasibility of the Nb₃Sn technology for the LHC upgrades.
3. Status of design, tooling manufacture, production
   1. 2-m-long single-bore demonstrator at FNAL: design and coil-manufacture tooling completed. Coil #1 winding under way.
   2. 2-m-long 2-in-1 demonstrator: engineering design finished. Technical design under way. CERN coil-manufacture tooling design is under way.
4. Superconductor needs
   1. For single-bore demonstrator: 26 km.
   2. For 2-in-1 FNAL demonstrator: 34 km.
   3. For 2-in-1 CERN demonstrator: 51 km, of which 17 km of low-J_c strand for practice coils.
   4. For 5.5 m prototype: 240 km.
   5. For series production, assuming 24 off 5-m-long cold-masses and 4 spares for 3 points: some 71 km of cable (>3400 km of strand).

Discussion

• Luca: Is 165 MPa peak stress really safe for the Nb₃Sn coils? Mikko: Nb₃Sn coils have operated at higher stresses at Berkeley. Gijs: 165 MPa at what field? Mikko: At the ultimate design field of 12 T. Ezio: Hence at 11 T we can expect stresses below 150 MPa. Luca: Nonetheless we should check for potential degradation once we have cable.
• Lucio: So we cannot expect 1-in-1 demonstrator results before Chamonix in January 2012? Mikko: Most likely not.
• Bernhard Holzer asks to be continuously updated on new insights in the expected field quality of the 2-in-1 magnet. Mikko fully agrees.
• Jean-Phillipe: There are a lot of activities planned for 2012-13. Is it clear there will be enough resources in view of the long shutdown (LS1)? Luca: I am aware of this.
• Vittorio: What is the rational in the decision between two 5.5-m-long or one 11-m-long magnet? Is the decision already taken? Mikko: There would be a substantial investment at FNAL, if we were to make longer coils than 6 m.  The cost of the superconductor, going for 11-m-long coils increases the risk significantly. Two 5.5-m-long coils back-to-back in a common yoke may be envisaged. Luca: But the real issue is the sagitta, therefore 5.5 m is most likely preferable. Vittorio: 11 m is still the more compact solution. Mikko: About 0.5 m are lost in the 5.5-m solution due to the reduction of field in the end regions. Sasha confirms the number. Luca: But what is the additional orbit kick with 11-m-long magnets? Bernhard Holzer: This is covered in the talk, but for now, 11 m is clearly worse than two times 5.5 m, especially when we consider a configuration where the collimator goes in between the two 5.5-m-long magnets. Sasha adds that there is no fundamental technological limitation. However the present tooling at FNAL is limited to 6 m. Moreover, with present strand unit lengths often around 1 km, 11-m-long magnets would waste much more conductor as drop-off. Massimo: Having an angle between two 5.5-m-long magnets avoids moving many other elements in the ring. This might be the main point.

11 T 2-in-1 Magnet Design with Emphasis on Field Quality and Integration (TF Compensation and QPS) – B. Auchmann

Contents

1. Transfer function: The design matches the MB integrated dipole field at nominal current 11.85 kA. At about 6 kA the deviation of the 11 T field reaches its maximum, being 2.4 Tm stronger. The preferred solution to this problem is a 300-A trim power converter.
2. Field quality
   1. The geometric coil field is optimized for all multipoles to be < 1 unit at a 17-mm reference radius.
   2. The yoke is optimized for low variation of multipoles. The b₃ variation is < 1 unit, and the absolute value is ~7.5 units. This could be compensated by the coil in a future coil cross-section update. The b₂ variation from ~zero at injection to -12.5 units at nominal cannot be further reduced for the given beam distance.
   3. The |b₃| due to persistent currents with RRP 108/127 conductor can be kept below 20 units by a dedicated precycle (compatible with the present LHC precycle) and/or passive compensation by passive strands in optimized positions in the aperture.
   4. Interstrand coupling currents are expected to have a sizeable impact on the transfer function and the sextupole at 10 A/s ramp rates. Moreover, modeling shows that, without counter-measures, the magnet would quench during a fast power abort. For both reasons, the use of a stainless-steel core seems advisable. This should also help to reduce any decay and snap-back effects.
3. Magnet protection will require highly efficient protection heaters. Simulations indicate that the goal of two redundant heater circuits could be best reached with intra-layer heaters. This option should be studied in an SMC coil.

Discussion

• Ezio: Why would you not optimize, i.e., shift the transfer function to reduce the mismatch from 2.4 Tm at 6 kA to about 1.2 Tm at injection and nominal? Bernhard Holzer: Even 1.2 Tm would have to be corrected, and the problem would be shifted towards injection and high energy.
• Lucio on HQ data: Is this the first time we see snapback in Nb₃Sn magnets? Luca: It also depends of the conditions under which the measurements were taken. We will make dedicated measurements at CERN when the magnet arrives.
• Giorgio: Which will be the first magnet with a cored cable in the R&D program? Mikko: The 1-in-1 demonstrator will not have the core. The demonstrator will serve to validate the models. The 2-in-1 demonstrator is therefore the first to have a core.
• Ezio: Mirror magnets showed no degradation of performance with high ramp rates, independently of the presence of a core – this is not yet understood.
• Sasha:
  • Intra-layer quench heaters have been used before in Nb₃Sn magnets at FNAL.
  • Cored cable was used in three dipole magnets and it proved effective to reduce the eddy currents. Effectively in the mirror test no big difference between cored and uncored cable was observed.
  • The passive correction scheme may also include iron strips which have proven effective in the past.

Requests from LHC Optics – B. Holzer

Contents

1. A shorter magnet would change the design orbit by ~6.5 mm, a large number of magnets would need to be realigned.
2. Edge Focusing is negligible.
3. Sagitta:
   1. In the 11-m-long magnet the sagittal is 7.2 mm; in the 5.5-m-long magnet it’s 1.7 mm.
   2. Given the large expected b₃ (~100 units at 17 mm reference radius in the initial worst-case error table from February 2011), feed-down effects are a problem in the case of the 11-m-long version, if the b₃ is not compensated by standard arc spool piece correctors.
   3. Example: worst-case b₃ of 110 units (corresponding to an unrealistic precycle with reset current equal to injection current) in an 11-m-long magnet causes tune-shift of 0.031 per magnet. 0.0059 could be deemed acceptable.
4. MAD/SixTrack dynamic aperture studies for points 2 and 7
   1. |b₃| > 20 units at injection would lead to unacceptable reduction in dynamic aperture from 11 σ to 8 σ.
   2. Other multipoles, so far, are negligible.
   3. Local correction of b₃ via spool-piece correctors would be desirable and it can be demonstrated that the dynamic aperture can be fully recovered, if local b₃ spool-pieces are applied.
   4. Non-local correction via sextupole correctors, e.g., at Q10, could equally work but would require stronger correctors.
   5. At present the multipoles that are foreseeen at high field (i.e. luminosity case) are small enough and have no significant influence on the dynamic aperture.

Discussion

• Luca: wouldn’t a shift of the magnet by half the sagitta reduce the feed-down effect? Bernhard Holzer: Needs to be studied in detail. Massimo: Luca is right, this needs to be taken into account in the analysis.
• Ezio: Shifting the b₃-curve at injection by part-compensation in the coil would help. Bernhard Auchmann: This can only be effective if b₃ doesn’t change sign, i.e., if it isn’t already centered between +20 and –20 units.
• Massimo and Luca: One should expect similar effects from systematic b₂ at nominal current as from feeddown effects at low currents. Why is the systematic b₂ deemed negligible? Bernhard Holzer: To be checked in detail. Qualitatively spoken, the emittance at high energy is much smaller and for the same beam optics multipoles at high field have not such a strong influence.
• Mikko: To correct for persistent currents and snap back, we may need stronger spool-piece correctors, or one standard spool-piece for each 5.5-m-long magnet. To be seen with the results of the 1-in-1 demonstrator.
• Massimo: Only systematic field errors have been considered, not the random errors. Bernhard Holzer: As long as we don’t have better estimates, one could provide upper bounds from tracking studies. Massimo agrees.
• Massimo: For HL-LHC the first points to be equipped with 11-T dipoles might be 1 and 5. Therefore all computations should also be done for the ATS optics scheme. Conclusions might differ and b₂ and b₃ be more of an issue.
• Luca: Are there other effects that might affect the beam? Bernhard Holzer: The transfer function mismatch between MB and 11 T will be discussed in his Wednesday talk. Short story: we need a trim power converter.
• Ezio: Do we only need to worry about b₃? Bernhard Holzer: At the moment, the others can be neglected.
• Bernhard Holzer: Keep me updated on any new insights into the expected field quality! Mikko and Bernhard Auchmann confirm they will.

Superconductor: Parameters Space – Luca Bottura

Contents

1. Lessons learned from the NED cable-development
   1. Best performance was reached through optimization of the heat treatment.
   2. A technology transfer had to take place to obtain equivalent quality as a commercial product.
   3. In retrospect, the initial specifications were probably too demanding for stable performance. There is an intrinsic interplay of critical current density, filament diameter, and RRR.
   4. New targets for the 11 T would be:
      1. Sub-element diameter 30 μm.
      2. J_c (12 T, 4.2 K) of 2650 A/mm².
      3. Local RRR > 100.
2. Magnetization matters for field quality à we need ongoing R&D on strands to reduce sub-element diameter.
3. Questions for this review:
   1. Do we agree on specifications (strand, cable) for magnet R&D and production?
   2. Material lead-time is long (> 12 months). How do we manage/share the present stock?
   3. What is the procurement strategy beyond the magnet R&D?

Discussion

• Sasha: Lead time for OST, based on experience, is 15 months.
• Gijs: Where can we find low- J_c conductor within the next 3 months for practice-coil winding at CERN?
  • Luca: There is old 1.26-mm PIT strand from the CERN/Twente dipole, it needs to be drawn down a lot, with risk of breaking many sub-elements. Lucio: Is there not also 0.9-mm strand from Twente? Luca: Only very small quantities.
  • Luca: Is there any RRP 54/61 strand available in stock at FNAL? Sasha: No low-performance strand is available at FNAL – but he will address the question to the LARP members later that day. Gijs: Could there be left-overs from 600-m unit lengths for 5.5-m-long coils? Sasha: Not now. Lucio offers to formulate an official request to LARP via other channels. Sasha will first ask and give feedback to Lucio.
  • Luca: ITER strand could be an option, but has different mechanical properties. Lucio: RRP or PIT would be better.
• Ezio: Can we manage to go to 30 μm strand diameter? Luca: Yes, the question is what it does to J_c and RRR. Sasha: Emanuela’s talk contains the latest on 0.7-mm RRP 150/169.

SC Cable Development at FNAL – E. Barzi, F. Nobrega, and A. Zlobin

Contents

1. The result of the first phase of cable development was a 40-strand, 14.7-mm-wide cable with 1.269 mm mid-thickness and 0.79˚ keystone angle. The cable is first fabricated as rectangular; it is then keystoned in a second operation with or without intermediate annealing.
2. The second phase of cable development was based on feedback from winding trials.
   1. An additional intermediate annealing step required revisiting the size of the rectangular cable.
   2. The cable’s mechanical stability needed to be optimized.
3. Reducing the cable mid-thickness from 1.27 mm to 1.25 mm reduces I_c only marginally. At 1.23 mm the degradation was measured to be 7.6%.
4. Intermediate annealing improves the rectangular cable’s mechanical stability and slightly reduces I_c degradation in the keystoned cable.
5. The development of tooling to make keystoned cable in a single pass is under way.
6. The new RRP 150/169 is an option for the 11T dipole program. It performs as well as the 108/127 baseline design, is more stable, and has larger RRR.
7. Coil winding has shown that the following measures reduce the risk of popped strands and enable the use of higher winding tension:
   1. More compaction of the cable.
   2. Application of ceramic binder to the cable before winding the ends to hold strands together.
   3. Re-optimized coilends to reduce torsion of the cable.

Discussion

• Gijs: Does the RRP 150/169 strand show better stability also at 1.9 K? Emanuela: Theoretically it does, it has smaller effective filament diameter, same J_c, and higher RRR than 108/127. Experimental data for 150/169 and comparison at 1.9 K will be available soon. Luca: Was the strand magnetization measured yet? Emanuela: Not yet.
• Luca: Could FNAL send 10-20 m of the strand to CERN for analysis? Emanuela: They will send some length of 0.7 mm RRP 150/169.
• Gijs: Can we allow for popped strands? Fred: Great care is taken during winding. The term “out-of-lay” may be more appropriate, as they can simply be pushed back into place.
• Luc: The reduction of cable mid-thickness could also reduce RRR – has this been measured? Emanuela: The stability has been measured, which is the most important test that includes the RRR effect.
• Lucio: Is the production of cored cable a lot more difficult? Emanuela: So far, only rectangular cored 40-strand cable was produced without problems. The core was slightly narrow and a wider one will be procured. Sasha: The experience with cored cable at FNAL includes ~300 m cable used in 3 dipole magnets HFDA02-04 and two unit lengths (~150 m) of cable for TQ coils (one coil was tested in 2010). So far, only rectangular cored cable was produced without problems. The core was slightly narrow and a wider one will be procured. Sasha: The experience with cored cable at FNAL goes back to 3 dipole coils and two unit lengths for TQ coils. Lucio: Why doesn’t LARP use cored cable? Emanuela: The main challenges are 1) the optimum width of the core – too wide and too narrow both cause problems, and 2) the centering of the coil within the cable.
• Lucio: What is the thickness of the core? Sasha: 25 μm. The additional compaction of the cable not only makes it mechanically more stable, but will also help to accommodate the core.
• Lucio recommends not to use up all margins and avoid a design were the cable is marginally stable or marginally free from degradation. If a core has to be used, then better accept to change the cross-section design to accommodate it.
• Giorgio: Whether or not to redesign the coil cross-section for the core will probably be the most important decision for the next year.
• Luca: Is the annealing step really necessary? Sasha: For technical reasons the cable is produced in two passes, first rectangular, then keystoned. Annealing between the steps helps to reduce degradation during the second step and improves the overall mechanical stability of the cable. But in principal a cable can also be done without annealing.

Cable Production, Strategy, and Material Flow – A. Ballarino

Contents

1. Today’s “standard” wires in RRP and PIT technologies are 0.7 mm RRP 108/127 and 1 mm PIT. R&D wire is being developed for the final application.  A first delivery of 0.7 mm PIT wire (44 mm filament size) is expected to be at CERN in the next few weeks. For meeting the quantity requirements for coils assembly, the next orders will specify the available 0.7 mm diameter wires.
2. What we have today:
   1. 27 km of “standard” RRP delivered in July 2011.
   2. 20 km of “standard” RRP expected for December 2011/January 2012.
3. Ongoing price enquiry:
   1. 45 km from OST, of which 30 km of 36 μm filament diameter, and 15 km of 32 μm.
   2. 45 km of 0.7 mm PIT with 44 μm filament diameter.
   3. In total: 90 km expected for November 2012.
4. Strategy for procurement of wire for the R&D program:
   1. 200 km of “standard” wire to be delivered by end 2012.
5. Future orders, including series production:
   1. Large quantities of “standard” wire.
   2. Smaller quantity of R&D conductor to push for “final” specifications.

Discussion

• Lucio: Discussion after Luc’s talk.

SC Cable Development at CERN – Luc-René Oberli

Contents

1. 7 cabling runs were performed at CERN to optimize cable parameters for a 40-strand cable with RRP 108/127 strands.
2. 3 runs were dedicated to develop a 15.1-mm-wide cable. The transposition pitch was reduced in steps to 80 mm. The resulting cable’s mechanical stability was improved, but it was less flexible and “highly nervous” or “springy”. I_c degradation was below 1.7% in four extracted strands.
3. Good mechanical stability was obtained with 14.7-mm-wide cable with 100 and 95 mm twist pitch and nominal mid-thickness. Average I_c degradation on six extracted strands was 2.6%, with a maximum of 6%.
4. To further increase cable stability and reduce degradation, cabling with nominal mid-thickness but slightly lower keystone angle (0.65˚ instead of 0.79˚) is suggested.

Discussion

• Lucio suggests to present the cable parameters in a CERN/FNAL uniform format to facilitate comparison.
• Luca underlines that a higher keystone angle under-compacts the outer edge and over-compacts the inner edge, therefore a lower keystone angle seems appropriate.
• Sasha: Although the present cable doesn’t seem to have stability issues, both labs continue to explore the parameter space. Lucio: First explore the parameter space, find the optimum, and then worry about a redesign of the cross-section. Don’t put yourself in a corner just in order to preserve the cross-section. 
• Sasha agrees and adds that the question of an updated cross-section might arise anyways due to any of the following reasons:
  • different compaction and/or keystone angle,
  • introduction of the core,
  • cable expansion during reaction (so far HQ values were used: 1% rad. and 3% azim.),
  • update of the insulation scheme, or
  • the need for part-compensation of the field quality.
• Sasha: CERN should make cabling trials with longer cable length, as some problems will only materialize in longer runs. In addition, the production of cored cable should be studied also at CERN.
• Sasha: FNAL has sent 114/127 cable to CERN for testing and cross-calibration of measurement methods. Has it been measured? Could CERN also send cable for these purposes? Luca was not informed of this, but strongly agrees with the procedure. Mikko will pass on the cable to Luc and Luc will send CERN cable to FNAL.

General Discussion

• R&D-phase superconductor procurement
  • Lucio: Amalia’s talk shows the best that CERN can do. What are FNAL’s plans?
  • Sasha:
    • January/February 2012 FNAL will receive 45 km of RRP 108/127, and
    • another 45 km is scheduled for August 2012.
    • When FNAL receives FY2012 funding, i.e., probably around May 2012, they will start procurement of another 100 km for FY2013, representing the FNAL contribution to the 5.5-m-long prototype.
  • Mikko: That leaves FNAL with just enough conductor for four coils for the 2-m-long 2-in-1 magnet, and for one  5.5-m-long coil in 2012 – no contingency.
  • Giorgio: To speed things up, FNAL therefore will procure 108/127 for the next big order.
  • Lucio: Clearly the first 5.5-m-long coils don’t need to be made from the final conductor.
  • Luca: We are running in continuous reserve mode. Whenever something goes wrong we run out of conductor. Need additional action for 2-m-long and 5.5-m-long practice coils!
  • Sasha agrees and adds that in the future we should plan for larger margins, since, as the unit lengths will go from 250 m of cable to 650 m, the waste/drop-off will be larger. Hence the need to try and procure more conductor.
  • Lucio: Why not order at CERN the 200 km with an option for an additional large amount – still keeping, of course, the 100 km from FNAL. Luca/Amalia: That is the plan.
• Low-J_c conductor for practice coils
  • Lucio: Have we ever seen that really we cannot have a few joints in Nb₃Sn conductors? At least for practice coils?
  • Sasha: Will ask LARP for RRP 54/61 strand that they would be willing to donate/lend.
  • Lucio: Nonetheless we should check if ITER strand could not be used as a practice-coil conductor.
• Spare unit length for 2-m-long FNAL 2-in-1 demonstrator
  • Luca: What has been decided between CERN and FNAL?
  • Mikko: CERN sends one unit length. Luc is ready to produce it.
  • Sasha agrees. If not needed, it will be returned.
  • Luca:  To avoid shipments, the proposal is to send a short length of cable produced according to the final dimensions (to be communicated by FNAL), store the full length at CERN, and send it only if needed. Lucio and Sasha agree.
• Conductor development to reach “final” specifications
  • Luca: We must still be aiming for smaller filament diameters. As we try to push the machine for higher performance, we should also try to push the vendors. If we don’t try now, we will never get it.
  • Giorgio: Would we have to plan for additional model magnets in order to use new conductor?
  • Luca: Maybe a short single-bore model for validation, but not more.
  • Lucio: All the mechanical structure will be available. We could even live with different strands in different points of the machine, especially if the improved conductor does not arrive in time for LS2.
  • Lance: We must state clearly our priorities. Without stating that we absolutely need better conductor, other clients will get priority.
  • Sasha: We are saying that we can survive and RRP 150/169 would improve the situation. Ideally the filament size would be < 20 μm , so the b₃ minimum would be outside the operational field.
  • Lance maintains that without a clear consensus, other priorities will win over our needs.
  • Lucio: It is necessary (not just desirable) to improve. But not at the cost of RRR or J_c!
  • Sasha: Time is an additional parameter. We go ahead with RRP 108/127 as long as there is no alternative! Lucio/Giorgio/Mikko strongly agree.
  • Luca: The November LARP workshop should come up with a statement including the insights from beam dynamics. RRP 108/127 introduces a level of complication to the project, as well as reproducibility issues, that make the machine more complex to operate. Therefore we need to push for R&D towards 30 μm filaments.
• Cored cable
  • Giorgio: Do we push for a cored cable for the first 2-in-1 model? We should rather start earlier than later.
  • Sasha: FNAL is now working to validate a cable design with 20 μm additional compaction. A core would therefore not necessarily require a new cross-section. Maybe at a 2-in-1 review in January the topic could be discussed. But first we need more tests and, in particular, the 1-in-1 demonstrator results.
  • Mikko: Adding the core in a more compacted cable should be step 1, so we can go ahead as planned. In parallel both institutes work to improve the cable. Based on test results and the final cable parameters, the cross-section can be redesigned at a later point.
  • Lucio agrees, but warns that if improved cable parameters exist, those should be used for an updated cross-section. Don’t stick to old parameters for convenience or speed.
  • Giorgio: The end of the discussions appears to be that the first 2-in-1 demonstrator will have a cored cable with unchanged cross-section.
  • Mikko: Yes, provided that the cabling tests for the more compacted and cored cable are positive.
• 11-T impact on beam dynamics
  • Lucio: After Bernhard Holzer’s talk is seems that if we keep the magnetization-induced sextupole within 20 units, we are in business. We can build more margins on these preliminary results.
  • Massimo: The ATS scheme for HL-LHC needs to be checked in points 1 and 5.
  • Lucio: ATS eats the margin at top energy, at injection it is less demanding.
  • Massimo: There are still many details, the higher orders need to be checked.
  • Gijs: If the b₃ curve is shifted to balance the sextupole between injection and nominal, that would eat margins for ATS at high fields.
• Definition of 11-T project success
  • Bernardo: What is the final goal of the magnet? How do you define success?
  • Lucio: It is a success when there is stable operation at 11 T. Short-sample is not the primary goal. The margin we need is thermal margin for the beam.
  • Sasha: The goal is to reach >11 T , with no retraining to go back to 11 T. If there is enough margin, we should start quenching around 11 T, not below.
• Cable procurement beyond 2012
  • Lucio: For Sasha and Luc: how will the procurement continue beyond 2012? We need a plan for that. Sasha agrees. Lucio hopes that in November there will be an updated plan.

Day 2 – Interface 11-T Dipole with Cold Collimation

Introduction and Scope of the Review

Scope of the Meeting: The Framework of the Collimation Project – R. Aßmann

Contents

1. Phase 1 collimation was developed over the past 9 years. It solves the collimation problem for nominal beam:
   1. Better than design cleaning efficiency.
   2. Additional gain from lower than specified beam losses.
2. Additional collimators for the DS region of points 1/2/3/5/7/8 was requested already in 2003, but deferred by management.
3. Going beyond nominal luminosity with HL-LHC will require protection, in particular of Q8 and Q10 in the DS.
4. Work ahead:
   1. Specify scenario for collimation (proton and ion intensities, lifetimes, luminosity, including HL-LHC optics parameters).
   2. Define collimator locations and gaps.
   3. Simulate/optimize collimation efficiency (Six-Track), energy deposition (FLUKA), jaw-temperature and cooling including accident cases (ANSYS), and RF impact on impedance and trapped modes (RF tools).
   4. Study a collimator design, iterate, and finally freeze the design.

Discussion

• Luca: How long will the above program take? Ralph: The team will need reinforcement. Once the design is frozen ~2-3 years. The first step is the definition of the collimation scenario. That might take ~1/2 year. Lucio: So we have ~2years, until the end of LS1, to bring technical solutions for the 11-T magnet and the collimator together.
• Ralph: We need a clear baseline for HL-LHC. Lucio: How does ATS affect collimation? Bernhard Holzer: From talk to talk he sees more IRs to be equipped with collimation and 11-T magnets. HL-LHC will make a big difference in IRs 1, 5, and 8. The specifications for the 11 T will change dramatically! The results of Tuesday’s talk were for IRs 2, 3, and 7. Ralph: IRs 1 and 5 might well be the most important! Fredy says that we can equip 6 IRs in LS2. Lucio: Of course if there was only one collimator per DS we could equip more IRs.
• Alessandro: Has it been decided yet whether the collimators will be 1- or 2-sided? Ralph: The design for IR3 (the canceled collimation upgrade in IR3 during LS1) was 1-sided. Ralph prefers 2-sided, but the decision will be a scientific decision based on the best arguments. The other question to be decided is whether to have one or two collimators per DS.

Collimation Needs and Boundaries – A. Rossi

Contents

1. The present collimation system reached 99.995% collimation efficiency with 50% smaller gaps than nominal.
2. Off-momentum particles are at the origin of losses in the dispersion suppressor. These particles are generated by particle-matter interaction in collimators, and by collisions at IPs. The first bending dipole after the LSS acts as a spectrometer. Superconducting magnets act as halo dump – dipoles for ions and quads for protons.
3. HL-LHC requires catching losses in the DS regions: Even if such losses do not quench the magnets, they may cause damage if the magnet is permanently operated close to the  quench limit.
4. Simulations show that a factor 10 to 15 improvement in cleaning efficiency can be achieved adding DS collimators.
5. DS collimators are needed in up to 6 IRs:
   1. IRs1 and 5 for proton luminosity.
   2. IRs 2, 1, and 5 for ion luminosity
   3. IRs 3 and 7 for proton and ion intensity.
   4. Priority: DS 1, 2, and 5, then 3 and 7, then 8.

Discussion

• Ralph: The full list of IRs and their priority is subject to ongoing studies.
• Alfredo thinks the ion analysis is too simple. The present analysis assumes ion fragments keep the momentum per nucleon of the original projectile. Furthermore it considers only fragments close in mass to the original ion. Realistic simulations have shown that there is a significant spread in the fragment momentum per nucleon, particularly for light fragments. Many more fragments can therefore be close to the nominal rigidity of the machine and therefore there might be additional losses. Ralph: We know from measurements that the model captures the most important effects.
• Discussion on 1-sided/2-sided collimators for ions/protons – see also last point of discussion after Ralph’s talk.
• Luca: It appears that warm magnets, especially the warm D1 magnets, get a large dose. Lucio: Davide should look into this, find out which magnets are really getting the worst dose.

Orbit Correction Issue – B. Holzer

Contents

1. As a repetition of Tuesday’s talk: the expected change of the design orbit between Q8 and Q10 due 11-m-long 11-T magnets is expected to be around 6.5 mm. This value can be much improved by splitting the magnet in two 5.5-m-long magnets with an angle between them.
2. Transfer-function deviation of the 11-T dipole w.r.t. the MB:
   1. If the deviation of the transfer function remains uncorrected, it would constitute an orbit bump of ±15 mm.
   2. After correction, ±1.5 mm would remain. However, we would use 42% of the available corrector strength just to fix this problem.

Discussion

• Bernhard Holzer: Up until now only IRs 2 and 7 have been studied. IRs 1 and 5 will be studied, together with HL-LHC ATS optics in the months to come.

• Bernhard Holzer: The information that the collimator might be installed either before or after Q10 is news. The impact on optics will need to be studied in detail. At some point we have to converge!
• Ralph: Also we need a decision whether the collimator is put in between two 5.5-m-long dipoles, or before/after the two dipoles. The former option might be less efficient than the latter. This is not necessarily a showstopper, but needs to be studied.
• Lucio: Are 6.5 mm within the range of alignment jackets? Vittorio Parma: This needs to be checked. It might be just at the limit.
• Ezio: We must be sure that the effect of the deviating transfer function is really important before we proceed to add trim power converters. Luca: Added complication due to 11-T dipoles is indeed a drawback. Bernhard Holzer: We should try to solve the problem at the source, and avoid patching solutions that eventually use up all available margins. Ezio: Also, as discussed on Tuesday, the transfer-function might be shifted. Ralph: It is preferable that the transfer-function mismatch is biggest at 3.5 TeV where the beam is already smaller. Bernhard Holzer agrees: If you shift the curve, you have to pay for the improvement at injection.
• Lucio reiterates Tuesday’s conclusion that in terms of field quality we are in business, as far as the present optics is concerned. Ezio: Do not forget that the random component of multipoles needs to be considered in future analyses!

Cryocatcher Prototype at GSI – P. Puppel

Contents

1. A single charge exchanged ion in SIS100 could cause desorption of >20000 particles – via further charge exchange reactions a self-amplifying effect evolves, which is called dynamic vacuum. Therefore ion catchers are installed with low-desorption surface material. In SIS100 the ion catcher is made of high-purity copper with a gold surface and a nickel layer in between, like in SIS18.
2. The cryocatcher is needed for stripped ions only; hence there are only catchers at the inner side of the ring.
3. The surrounding cold chamber at 5 K acts as cryopump. The catcher itself has to be at higher temperature (77 K in case of the prototype) to prevent gases from freezing out on its surface.
4. A stainless-steel bellow is the thermal insulation between the 77-K thermal shield and the 5-K vacuum chamber. The catcher itself is directly connected to the thermal shield.
5. In tests, the catcher showed an unexpected correlation between the desorption yield and the energy of incident ions. Whereas at room-temperature the desorption decreases with increasing ion energy (sclaes with the electronic energy loss in the material), at cryogenic temperatures the desorption rate increases with energy.
6. For constant ion energy, no dependency of the desorption yield on the catcher temperature was observed between 32 K and 94 K.
7. The work on the final SIS100 cryocatcher specification is under way.

Discussion

• Alessandro: What is the thermal load on the catchers? Patrick: 2 W on average, up to 50 W during very short peaks for the catcher with the highest load during slow extraction.
• Vittorio Parma: What about failure scenarios like a jaw getting stuck? Patrick: The cryocatcher has no moving parts.
• Luca: Have other materials than gold been investigated? Patrick: There had been a whole measurement campaign a few years ago for SIS18, with the result that gold-copper showed the lowest desorption yield.
• Alessandro: Why were lower temperatures than 77 K studied? Patrick: The working temperature in SIS100 will be closer to 50 K. The study was carried out to check if there are critical temperatures.
• Vincent: Do you rely on beam conditioning (beam scrubbing)? Is the increase in pressure due to a desorption rate of 1000 a problem? Patrick: They do not rely on beam scrubbing. The trapping power of the cold surfaces is very high, so those desorption yields are not a problem. On the long run beam scrubbing might help.
• Vincent. What is the impact angle of the ions? Patrick: Zero degrees, i.e., perpendicular. The cryocatcher is installed precisely to intercept most ions perpendicularly. At grazing incidents, the desorption yield does not scale with the ion energy.
• Lucio: Is there no impact on impedance? Patrick: This was studied by Olvier Boine-Frankenheim. The impact should be low enough.
• Lucio: Are there no showers onto the adjacent elements? Patrick: This was studied with FLUKA, especially for the neighboring superconducting quads. Lucio: How about showers due to grazing incidents on the vacuum chamber? Patrick: The cryocatcher’s jaw is tilted away from the beam axis in order to avoid grazing incidents.
• Alessandro: What is the temperature of the neighboring quads? Patrick: All elements are cold, around 10 K, with the exception of the vacuum chamber that is heated by eddy-current effects.
• Lucio: The SIS100 is a machine that needs to be at cryogenic temperatures for the vacuum. The decision to go for a superconducting magnet system could therefore be argued with the fact that cryogenics was needed anyways!

CERN Mechanical Pre-Study on Cold Collimation - A. Bertarelli

Contents

1. The initial idea to install a cold collimator in IR3 during LS1 was abandoned in July 2010 for a solution based on a warm collimator. The preliminary design of the cold collimator was developed by D.D. Ramos and C. Mucher.
2. The preliminary requirements for the DS collimators were:
   1. Two jaws per collimator (because of back-scattering and positive Δp/p of ions). This proved to be a major complication for the design.
   2. 1-m-long tungsten jaws.
   3. A heat load of 40 W steady-state and 200 W maximum during a 10-s transient.
   4. A jaw stroke of 25 mm.
3. The collimator jaws would have to work above 80 K. To evacuate 200 W the only possibility is to rely on the line E (50-65 K), and to heat the helium to >80 K.
4. Design-optimization (saving space) is possible, especially if the expansion loops and busbar routing could be integrated with the 11-T dipoles.
5. Main issues and potential showstoppers:
   1. The jaw operation temperature would be 100-130 K. Beam-vacuum operation at 100 K could prove to be a showstopper – GSI experience?
   2. Sufficient cryo-pumping surfaces at ~3 K need to be provided, with a thermal shield towards the >80 K jaws.
   3. Tungsten brittleness at 100 K.
   4. The reliability of the active parts for the cooling circuit (valves, etc.).
   5. Possible additional heating from RF impedance.
   6. Issues with moving/sliding parts in UHV cryogenic environment.
6. Alternative solutions like a warm collimator should be studied.

Discussion

• Lucio: Why are sliding RF contacts a problem? Isn’t that the same as now? Alessandro: The sliding contacts produce dust that has an impact on the vacuum. Additionally, these sliding contacts, unlike other similar contacts such as PIMs, undergo both longitudinal and radial movement, thus increasing the possibility of getting stuck during cold operation.
• Rob: Is cryo-pumping at 3 K sufficient? Vincent: This needs to be studied in more detail.
• Vittorio Parma strongly agrees that warm collimators should not be off the table. Paolo: Warm collimators would have the huge advantage of separate component assemblies. This is important also for the spares. There would have to be, however, two different types of 11-T coldmasses to accommodate lyras on both sides. Vittorio Parma: And the shuffling module to by-pass the warm collimator requires space. Lucio: Figuring this out is the job of the integration working group.
• Lucio: Couldn’t the actuator to move the jaws be a potential showstopper? Alessandro: He would consider it to be a design issue, not a showstopper. Ralph: If a jaw gets stuck at top energy this constitutes an obstruction to the beam and we cannot inject anymore. With a cold collimator, a repair would take 3 months. This is a subject for risk-analysis and might well be a showstopper!
• Luca: Would a cold collimator require special alignment mechanisms? Ralph: Vertical alignment is not really an issue. Alignment w.r.t. the coldmasses should suffice. Vittorio Parma: For horizontal alignment, would you put BPMs? Alessandro: Not a good idea to have BPMs in cold collimators. The main problem may come from the BPMs’ cables, which are back-filled with gas. This may lead to outgassing when the cables are moved. Fatigue issues may also arise.
• Vittorio Boccone: The 50-W average and 200-W peak loads are very challenging. They are based on simulations of IRs 3 and 7. The situation at IRs 1 and 5 are unkown. Alfredo: and even for IRs 3 and 7 there might be large error margins. Ralph: In addition those calculations were done for the nominal optics, not for HL-LHC. Values may go further up.
• Rob: Can the suitability of tungsten in a cold collimator not be quantified? Lucio: Stefano Sgobba could help answer this. Does tungsten have to be baked out? Ralph: Other materials may lead to longer jaws. Alfredo: Maybe iridium could be an option.

Heat Deposition Pre-Evaluation – V. Boccone

Contents

1. Calculations exist only for IR 7, 7 TeV horizontal loss scenario (collimation Phase II upgrade). The results  were rescaled to the LHC ultimate luminosity and 1h beam lifetime. Under those assumption:
   1. The maximum steady-state power is below 50 W for a DS collimator with two 1m tungsten jaws.
   2. The DS-collimator reduces the total power deposited in the arc by a factor of 10-12 for protons, and by a factor of 3 for ions.
   3. The DS-collimator reduces by a factor of 15 the peak power on the downstream magnets.

2. Which is the total power deposited in the DS collimator in the other cases? Answers to this question must include:
   1. IRs 1 and 5: Definition of the collimation scheme and study of collision debris down to the DS;
   2. IR 2: Definition and study of a realistic collimation scenario for ions;
   3. IRs 3 and 7: Definition of the collimation scheme and study of the impact of a realistic vertical and horizontal scenario.
3. Uncertainties from physics models and simulation must be taken into account. Moreover, the effect of imperfection can lead to an additional factor 10 of uncertainties. Is this number appropriate?

Discussion

• Lucio: How does a 2x smaller emittance influence the results? Ralph: Emittance doesn’t play a direct role. The collimation scrapes away the halo. It does not matter how small the rest of the beam is. However, the collimators can come closer to a smaller beam, which would mean more power on the collimators.
• Lucio: We now know the machine much better than two years ago. Can the facor 10 for uncertainties not be reduced? Ralph: The cleaning efficiency still varies by a factor 5 as a consequence of very small effects. And the factor 5 in the machine should be considered a success! Vittorio Boccone: We do not know the 7 TeV machine yet. Alfredo: Maybe that today a factor 10 is too much, but do not forget that there are many sources of errors in the analysis!

Interface of the 2x5.5m 11-T Cold Masses – M. Karppinen

Contents

1. Open questions for the coldmass integration include:
   1. The two 5.5-m-long magnets could be aligned straight or at an angle.
   2. The busbars need to be routed around the collimator, be it a cold or a warm solution. Splices need to be foreseen and expansion loops need to be designed in a compact fashion.
   3. The heat exchanger needs to be integrated.
   4. 600-A powering for potential spool-piece correctors and 300-A powering for transfer-function trim needs to be provided.
   5. A dedicated QPS system is needed.
2. Open questions for the integration of cold collimators:
   1. Minimum space requirement.
   2. Operation requirements.
   3. Access and maintenance requirements.
   4. Need for a temperature screen and/or vacuum separation.

Discussion

• Lucio: The collimator length can only be defined after further studies. The warm solution is still a very serious alternative.
• Ralph: First we need the scenarios for HL-LHC, to see for what losses, intensities, …, we need to prepare. The solution must be robust also for ATS.
• Luca: Aligning the two 5.5-m-long magnets at an angle is probably the best solution. Magnetic forces shouldn’t play a role if the collimator goes in between the magnets. Forces from pressure-waves might need to be studied.

Integration in the Cryostat – V. Parma

Contents

1. The CCFS (Cold Collimator Feasibility Study) Working group has only just started to meet. It will
   1. Analyze configurations for collimator/11-T-magnet integration.
   2. Identify potential showstoppers (vacuum, cryogenics, machine protection, alignment).
   3. Identify needs for R&D.
   4. Provide final recommendations and a draft timeline for the project.
2. Preliminary study of 3 assembly options:
   1. The following options have been proposed and will be investigated:
      1. An 11-m-long magnet next to the collimator.
      2. Two magnets with 5.3 m magnetic length side-by-side, next to the collimator.
      3. Two magnets 5.3 m magnetic length with the collimator in between.
   2. Option iii, which is favored by beam-physics, requires the most space for interconnects, thus limiting the available space for the collimator.
3. Cryostating of any of the above options should not be an issue with the available technologies at CERN.
4. In IR 2, the injection line creates very different conditions from the other IRs. Displacing a number of magnets would be especially inconvenient there. Special tooling and procedures will be needed.
5. Powering of trims and possibly spoolpiece correctors by 300 and 600 A lines must be studied. Spare spool circuits might be useable, although their availability needs to be checked with the list of non-conformities. Changes to the DFBAs, and local feedthroughs need to be studied.

Discussion

• Lucio: Interconnections between dipoles cannot be more compact than 0.5 m. But interconnects between coldmasses in the same cryostat could – see triplets. Paolo: The expansion loops could be implemented differently. Spools, if needed, should go on the side of the expansion loops. Do we need two expansion loops for two 5.5-m-long coldmasses?
• Lucio: Could MgB₂ be used for current leads or feedthroughs? To be checked with Amalia and Luca.
• Jean-Philippe (per eMail on the same day): In points 1, 2, 5, and 8, contrary to points 3 and 7, the 6-kA cables are in the line N. This needs to be addressed: either the 6-kA lines need to be removed from the line N, or new interconnections with 6-kA connection boxes could be required.

Cryogenic Margin and Operation Issues – R. van Weelderen

Contents

1. Cryogenic requirements for 11-T coldmass and cryostat:
   1. Continuity of heat exchanger.
   2. Minimum 15 l and maximum 26 l pressurized helium per meter coldmass.
   3. Longitudinal hydraulic impedance equivalent to a 50-mm diameter smooth pipe for cooldown, warmup, and quench.
   4. Free helium section >60 cm² for heat conduction.
   5. Radial helium channels equivalent to MB. 98% packing factor of collar and yoke laminations should suffice.
2. The 11-T magnets are about 25% shorter than MB. They should accumulate 25% less heat and the cryogenic margin is 25% higher, i.e., about 8-11 W. The additional 5 W from beam losses, which are currently considered for HL-LHC with DS collimators, can therefore easily be accommodated.
3. A 50 - 150 W heat load is expected on the collimator jaws. On the 1.9 K line, this corresponds to 14 times the MB heat load, which constitutes an excessive load. On the 5 – 20 K lines it corresponds to 1.6 times the MB equivalent heat load and could just be accommodated – to be studied. For vacuum considerations, the preferred temperature range for cold collimator jaws is 80 - 140 K. At these temperatures the heat load corresponds to 2 times the MB equivalent heat load. This could be realized by a heated bypass of the 55 – 65 K thermal-screen cooling line.

Discussion

• Lucio: It is too early to rule out cooling via the 5-20 K lines.
• Mikko: Do you have resources for thermal modeling of the 11-T dipole? Rob: I have asked for resources. Lucio: And you have them! (Six-month extension of the fellow concerned.) Mikko agrees to supply the design details necessary for a thermal evaluation of the design a.s.a.p.

Possible Vacuum Issues – V. Baglin

Contents

1. A warm-collimator solution was developed for IR 3, featuring:
   1. A bakeable collimator.
   2. Vacuum sectorization to allow for “fast” exchange.
   3. A port to insert an RF ball (PIM test).
   4. Access to the vacuum system to boost the pumping speed.
2. The preliminary cold-collimator design features none of the above, but it is shorter.
3. An unbaked collimator shows >100 times more outgassing than a baked collimator. Experience from RHIC, where one beamline had a baked, an the other had an unbaked collimator, shows that beam-intensity in the unbaked aperture was limited by pressure rises. The other beamline didn’t show vacuum instabilities.
4. Moving collimator jaws against RF contacts produce outgassing. The effect needs to be studied more and minimized.
5. The only available cooling power to accommodate 50-150 W heat load from the collimator is the 50-70 K cooling line. The zone around 80 K must be avoided to avoid CO₂ condensing on the jaws. As a consequence, we need an operating temperature above 90 K.
6. Detailed studies, theoretically and experimentally must be carried out to define and validate a cold collimator solution.

Discussion

• Lucio: What is the impact of a collimator exchange on the vacuum? Vincent: It would mean venting 2800 m in the cold-collimator case, unless vacuum valves are installed. The warm collimator would have to have vacuum valves since they are required on all warm-cold transitions and therefore exchanging a warm collimator means venting 20 to 60 m.
• Lucio: A bakeable cold collimator should be even better than a warm collimator? Vincent: Being able to bake a cold collimator requires again vacuum sectorization. With that in place, you can bake before cooling the collimator down.
• Ralph: The production of dust by movement must be studied in detail for the DS regions.