The LHC energy: past and future XXIX Workshop on Recent Advances in Particle Physics and Cosmology, Saturday, 16 April 2011 Mike Koratzinos, CERN

Introduction The LHC has been operating in 2010 and 2011 at 3.5TeV per beam, half its design energy of 7TeV per beam We will review the constraints leading to this operating energy and provide an outlook for the years to come. Sergio Bertolucci yesterday already gave the landscape. M. Koratzinos Acknowledgements: This work would not be possible without the help of a great number of colleagues, amongst which are Andrzej Siemko, Ruediger Schmidt, Arjan Verweij, Zinour Charifoulline, Knud Dahlerup-Petersen, Reiner Denz, Bob Flora, Howie Pfeffer, Jim Strait, Hugues Thiesen, Francesco Bertinelli, Christian Scheuerlein, George Kourkafas, Marta Bajko, Christian Giloux, Jerome Feuvrier, Andrea Apollonio, Pawel Dupert, Per Hagen, Bernard Dubois, Olivier Fournier and others.

The Chamonix workshops The last couple of years, the subject of beam energy is discussed in the Chamonix workshop, where all aspects of possibly raising the energy are discussed (physics motivation, measurements, simulations, effect of a possible thermal runaway, etc.). Soon after, a decision is taken as to what the energy should be for the year. M. Koratzinos

The safe energy of the LHC The reason behind the decision about the operating energy of the LHC are linked to risk rather than technical constraints. The main circuits of the LHC have no problem attaining their design current* which gives the design energy. However, there is a risk of an incident which is energy dependent. *Bar training quenches M. Koratzinos

The 19 September 2008 incident We know what is unsafe: Calorimetric analysis of test PLI3.a2 done on 15 September 2008 yields the following value for the excess resistance of the splice: R 23R3-16R3, excessive = 234 ± 15 nΩ

Powering scheme of high current circuits The LHC synchrotron is divided into 8 sectors. Each sector has three high current (13000A) circuits (RB, RQD, RQF) powered by one power converter each (24 high power converters total) In the case of the RB circuits, 154 magnets are powered in series (around 50 for RQ) This necessitates connecting the s/c cable from one magnet to its neighbour in situ in the LHC tunnel (  interconnects or splices) M. Koratzinos

Energy stored in the magnets At the LHC, the stored energy is very high. Such an accelerator commands respect The energy stored in the LHC magnets at the design energy is 9.4GJ – similar to the kinetic energy of a Nimitz class aircraft carrier (80000 tons) traveling at 35 knots (battle speed) [Even the energy of the beams is large: 360MJ – like the kinetic energy of a speeding train]

Busbar copper stabilizer In case of a quench, the current needs a way to pass through as it cannot go though the superconducting cable For this reason, the busbars contain copper stabilizer of sufficient diameter M. Koratzinos

High current splices The main circuits of the LHC (RB, RQD, RQF) have about splices. Out of these there are: interconnect splices and magnet splices Interconnect splices are not protected by diodes and in the case of a problem all the current of the circuit passes through them Nominal interconnect splice resistance: At cold: 300pΩ At warm (300K): 10μΩ For the LHC to operate safely at a certain energy, there is a limit to how big a splice resistance can be M. Koratzinos

Copper stabilizer continuity busU-profile bus wedge SolderNo solder Bad joint busU-profile bus wedge SolderNo solder Remnants of solder/oxidation busU-profile bus wedge SolderNo solder Slightly better joint Good joint

What can create an incident? The greatest danger for a 19 September 2008-type accident is 1.An interconnect with a discontinuity in the copper stabilizer coupled to a quench in the vicinity (1) alone or (2) alone is not enough to create an accident So, to define a safe energy of the LHC, we need to know: 1.How good/bad the copper joints are 2.How many quenches we will get 3.The relationship between resistance and thermal runaway (derived from a simulation) M. Koratzinos

nOhms and uOhms The 19 September 2008 incident happened due to a interconnect whose resistance at cold was 1000 times more than specification – 300nOhms at cold Subsequent analysis and review showed another failure mode: even if the resistance at cold was nominal, a copper discontinuity coupled with a length of ‘bare’ cable could burn out in case of a quench, as the current is forced to go through the (narrow xsection) s/c cable. Relationship between length of cable and resistance (at warm now) is about 1cm = 14uOhms at warm M. Koratzinos

Resistance at cold A measuring campaign during 2009 (so-called Maya pyramid campaign) revealed that all resistances at cold do not pose a problem. The highest resistance measured was 3nOhms, low enough so that it can never quench the cable. The next question had to do with if such a resistance would deteriorate with time. M. Koratzinos

s/c interconnects: evolution with time The plots on the left provide the answer: no deterioration seen on the worst joints during 2010 Noisy board Zinur Charifoulline M. Koratzinos

Resistance at warm – non-invasive measurements A heroic campaign took place in 2008, where 5 out of 8 sectors were measured at warm (RB circuit) An added complication is that a single interconnect cannot be measured; instead a whole busbar segment is measured (around 40m long) where a defective joint is about 2% of the value of the busbar segment M. Koratzinos Busbar segment at warm: 2mOhm Typical large defect: 50uOhms

Warm non-invasive measurements M. Koratzinos : Interconnect splices : Voltage taps Spread (uOhms) Highest Excess resistance A12 RB A34 RB A45 RB A56 RB A67 RB13.651

Resistance at warm – invasive measurements Unfortunately the resolution of the non-invasive method (around 10-15uOhms) is not good enough to get a good idea of the underlying distribution of splice resistances in the machine For this reason, we used the non-invasive method to ‘point’ to possible excessive resistances. M. Koratzinos The joints were then opened and measured very accurately (invasive method) [error less than 1uOhm]. The distribution thus measured is used to derive the underlying distribution of bad splices in the machine – an exponential fits the cumulative distribution of the data well.

The fitted distribution Our knowledge comes from a SMALL and most probably BIASED sample of 134 joints (out of 10000) measured accurately in Out of these 23 are above 20uOhms. Distribution is shown below. Joints could possibly deteriorate with time. Analysis by Jim Strait + MK No. of measurements: 134 – in this plot: 23 Extrapolating an (unknown) distribution is always risky! The sample is from RB only, 5 sectors M. Koratzinos

Quenches in 2010 One of the pleasant outcomes of this year’s running, is that we have observed to date zero unintentional beam-induced quenches. This is due to – A well behaving, reproducible machine – An BLM system that intervened in time and prevented magnet quenches due to beam losses. If we were sure that we will get zero quenches next year, running at ANY energy would be safe from the splices point of view. However, non-beam-induced quenches have been seen (spurious quenches related to EM interference during a fast power abort, etc.). For next year, we simply do not know how many quenches we are going to get. Therefore in what follows, I have left this as a free parameter. M. Koratzinos

Quench statistics 2010 This is a rudimentary attempt to look at how many quenches we had in 2010 (thanks to H. Reymond). The PM system is interrogated for cases where – QPS signal present: RB U_QS0> 100mV for 3 successive measurements – High current: I_MEAS>5000A This gave a total of 27 events. Manual inspection of those revealed that 9 were not real quenches (no heaters fired) and 3 more quenches were identified. The categories identified were: – Trip during PGC1: 11 quenches – Provoked quench: 3 quenches – Power cut: 2+3 quenches – Other: 2 quenches In 2010 we had about 20 quenches of the RB circuits above 5000A, due to various reasons. M. Koratzinos

“Giraffe” plots Safe max. defect is the rear foot of the giraffe. Strange shape comes from combination of gaseous Helium heating and heating through the busbar A. Verweij, “Probability of burn-through of defective 13 ka splices at increased energy levels”, Chamonix 2011

A method to present risk One way to present how safe it is to run at a specific energy is to calculate how many quenches we need per year before reaching a pre-defined incident probability level.  I have used 0.1% as the acceptable incident probability level (one incident every 1000 years), but other numbers can also be used. In this framework, if the number of quenches we can “afford” is large, then operation is safe, and we would only need to re-asses the situation when this number of quenches is reached. If, however, the number of quenches we can “afford” is less than one, getting a single quench would put us above our pre-defined ‘risk’ level. M. Koratzinos

Types of events We can have the following types of quench events: – Prompt quench of the interconnect region (beam induced) [this is very unlikely since the adjacent magnet has a quench threshold which is 10 5 lower] – Busbar propagation quench where the interconnect will quench from the heat propagated through the busbar from the (hot) diode within a few seconds – Gaseous Helium propagation quench where the interconnect will quench ~20 seconds later from heat transferred through gaseous Helium of a quadrupole magnet: Very unlikely New calculation M. Koratzinos

No. of quenches to reach P incident =0.1% Energy (TeV), T dump (s) Allowed R excess RB(µΩ) Allowed R excess RQ (µΩ) No. of bad joints above R excess RB No. of bad joints above R excess RQ No. of quenches RB to reach P accident =0.1% No. of quenches RQ to reach P accident =0.1% Prompt quenches – 2 RB and 4 RQ joints affected 3.5, 50s , 50s [4.0,68s]72 [56] [2] [0.6] 4.5, 68s , 68s Busbar propagation quenches 1 RB or 2 RQ joints affected 3.5, 50s 90 any ∞ 4.0, 50s [4.0,68s] 74 [55] [2]0.910 [2]6 4.5, 68s , 68s 44? Gaseous He propagation quenches (RB) – 3 RB and 8 RQ joints affected 3.5, 50s115any , 50s [4.0,68s]98 [75]any0.03 [0.3]033 [4] 4.5, 68s63any , 68s Preliminary = old calculation (RRR=100) = new calculation + = will increase M. Koratzinos

On the safe energy of 2011 The most stringent limits come from prompt, beam induced, quenches, considered unlikely. – if we are confident we will get no prompt quenches, we can ignore the ‘prompt’ quench category. The most relevant limits come from the ‘bus propagation’ quenches, calculations on this are new this year. Some calculations (for 5TeV or for the RQ) have not been updated with the latest information – will be finished shortly. 4TeV with an extraction time of 50 seconds gives us some margin for all types of quenches, therefore 4TeV operation cannot be ruled out from the information we have about the state of the machine. However, limits are more stringent than for 3.5TeV operation. Therefore, running at a higher energy needs to be balanced against pushing for higher luminosity this year, that might result in a larger number of quenches than in M. Koratzinos Decision to run at 3.5TeV for 2011

Is there something we can do to be able to run at a higher energy in 2012? Since 2012 will be a year with LHC operations, it makes sense to try and see if something can be done that fits in a period of (an expended perhaps) shutdown, that might allow us to run at 5TeV for A promising possibility is the ‘Thermal Amplifier’ idea. It is a qualification tool that would allow us to run at the highest possible energy with near- zero risk. The CSCM (as it is officially called) is now an official CERN project M. Koratzinos

Why Thermal Amplifier? The knowledge of the state of the copper stabilizer joints in the main circuit is very poor: – Not all sectors have been measured – Different sectors can have very different joint quality – Time degradation has not been studied and is a worry The current safe energy analysis is based on a lot of assumptions – mostly pessimistic – this is to counteract our lack of knowledge of the copper stabilizer joints We should not underestimate the engineering complexity of such a project: There is a series of engineering challenges to be solved before we can put such a method into production. Proof of principle successfully performed in Conceptual work continues to simplify the project. M. Koratzinos

Thermal amplifier principle The thermal amplifier uses a pulse of high current (order 6000A) to selectively warm up bad splices in a sector. Current flows through the diodes, so very little energy is stored in the circuit (= safe). It is a direct measurement of a thermal runaway at the exact conditions of a joint. Temperature of operation: around 20K 2.7kA pulses at 41K The GREEN voltage contains a 50uOhm defect. The BLUE and RED voltages are across perfect joints RB: a typical bad joint has excess resistance of 2% - if we warm it up, its resistance grows by ~200 times – easy to detect! REAL DATA M. Koratzinos

Thermal amplifier proof of principle A series of tests were performed in September 2010 in the Block 4 test facility (special thanks to M. Bajko, C. Giloux, J. Feuvrier). A defective joint (about 50uOhms) and a series of perfect joints were monitored as high current (2000A to 4000A) passed through the copper at temperatures varying from 20K to 40K. What was measured agreed with simulations (program of A. Verweij) and valuable experience was gained. Comparison of simulation (blue) and real data (red). Current in this test (bottom) was 3200A. Initial temperature was 42K M. Koratzinos

The CSCM project February 2011: Project approved. The core team has taken the first decisions regarding some important parameters. An external review will be performed early autumn Period of the tests: xmas technical stop First estimates for the time needed for the tests: 12 weeks. We are working on streamlining this. M. Koratzinos A word of caution: the Thermal amplifier is not a ‘carte blanche’ to run at 5TeV; it will tell us however with some certainty what is the maximum safe energy of the LHC, or it will give us precise probabilities of an incident in case of a quench. It will always be up to the CERN management to decide at what level of risk, if any, the LHC should run

Splice consolidation During an extensive campaign will take place so that all 10,000 splices will all be opened up, inspected, repaired if necessary and a shunt fitted. This new design will allow us to completely forget copper continuity problems. This will allow the LHC to operate with a minimal number of training quenches at around 6.5TeV M. Koratzinos

The crystal ball Although the future is tricky to predict, regarding the LHC energy the future could be something like the following (this is a purely a personal ‘best bet’): – 2011: operation at 3.5TeV – 2012: operation at 5TeV, if lucky, 3.5TeV if unlucky – 2015: operation at 6.5TeV, then over a period of a couple of years slowly reach the design energy of 7TeV M. Koratzinos

Conclusions The condition of the copper stabilizer around the main circuit interconnects limits the maximum safe energy of the LHC Given the (lack of) knowledge of the excess resistance of the copper stabilizer joints, running at any energy carries a certain risk. A lot more has been learned on the subject since the accident of 19 September 2008 The “Thermal Amplifier” idea can qualify the machine to the highest safe energy with minimal risk – this is planned for Xmas 2011 and if we are lucky 2012 can be a 5+5TeV run M. Koratzinos

End