LHC Luminosity Prospects

After a two week break from physics runs, the Large Hadron Collider is once again clocking up proton-proton collisions at a steady rate. The break included a few days of machine development time, a planned technical stop for routine maintenance, Van der Mere scans to study the beam size and distribution and finally alignments for the TOTEM and ALFA experiments.

Now they are back to the process of building up the luminosity. The current fill has 768 bunches matching previous runs except that they are now injecting 108 bunches at a time compared to the previous scheme using 72 bunch injections. This is important because the next step up to 912 bunches will require the 108 injections and that is now expected for the next run later today.

The 912 bunch step is likely to take the luminosity through another milestone by peaking at more than 1000/μb/s. This is the official target for 2011 luminosity but it will be exceeded further as they step up through 1056, 1200 and then 1380 bunches in the next couple of weeks. These last three steps will require 144 bunch injections taking the injection systems to new limits. In the last few days they already injected 1308 bunches during a scrubbing run, but without ramping to full energy for collisions.

(Update 23-May-2011: A record luminosity of 1100/μb/s was reached on the third fill with 912 bunches per ring this morning.)

With the current plans it should be possible to reach luminosity of at least 1500/μb/s for the long run from June to October during which there will be 124 days allocated to proton-proton physics runs. So how much further will it be possible to increase the luminosity and when? As part of the machine development time they tried out some processes that could potentially increase luminosity further. Possibilities for higher luminosity without updating the hardware are the following

  • Increase of bunch intensity
  • Improved emittance
  • Better Squeeze
  • More bunches
Let’s look at each of these in turn

Increase of bunch intensity: The proton bunches currently circulating start out with 115 billion protons in each. This is the “nominal” bunch intensity that was originally planned for the collider with its present hardware. However, it is possible to inject an “ultimate” intensity of 170 billion protons into each bunch from the SPS. The luminosity increases in proportion to the bunch intensity in both beams so the potential luminosity increase is a factor of (170/115)2 = 2.2.  It had been anticipated that the beams would become too unstable at these levels but already last year they found that reaching nominal intensity was easier than expected. During the latest machine development time they injected a few ultimate intensity bunches to see how they perform. Once again they found that there was no problem with the head-on beam-beam tune shifts that they had expected to be a limiting factor, so the way is open to a further increase of luminosity.

The beam operators are being publicly cautious  about when this could happen. Paul Collier who heads the team said in a message on the LHCPortal forum that “The studies with very intense bunches are for the future – no plans to push higher bunch intensities into operation for the moment.”

Improved emittance: The emittance \epsilon is related the size of the beam w = \sqrt{\epsilon \beta} Transversely emittance should be 3.75μm but with better than expected performance  it has been typically 2μm to 3μm. During the machine development tests they were able to reduce this to 1.5μm. If they can do this for injections of full trains it represents a further increase of luminosity.

Despite the cautious words there are signs that increased intensity and improved emittance could be introduced into the physics runs this year taking luminosities up to “a few” times the 1000/μb target figure. Whether this happens and to what level will depend on how smooth operations go during the first few weeks of the long run with 1380 bunches.

Better Squeeze: The squeeze is a process of focusing the beams into a smaller cross section as they pass the collision points inside the detectors. The current squeeze is represented by a figure of β* = 1.5m but a tighter squeeze may be possible. The nominal squeeze is 0.55m but this will not be possible until the full 7TeV beam energy is available after the long shutdown. After that further improvements are still possible. As Paul Collier explained, the minimum β* is limited by the aperture of the triplets and also by the chromatic aberrations introduced by the very tight squeeze. Another test during the machine development period looked at a new method of ATS injection which is a novel and compicated scheme to bypass the strength limit of the existing lattice sextupoles and allow a much smaller β*. Values of around 0.15m should be possible with existing hardware.

These developments are definitely not for this year. Moving to a tighter squeeze by any means will require a long process of resetting collimators and building up luminosity again in steps. Too much time would be lost to make this worthwhile for current runs but in 2012 a smaller β* might be used.

More bunches: The last route to increased luminosity is more bunches. To exceed 1380 bunches per beam they must use a lower bunch separation time of 25ns compared to the 50ns spacing currently in use. Since they travel at virtually the speed of light this smaller spacing places the bunches just 7.5m apart. Such a change might require a larger crossing angle to avoid parasitic collisions as bunches come close 7.5m away from the desired interaction points. This would again mean resetting the collimators. Furthermore, it is not known if the vacuum and cryogenics can yet cope with the higher intensities that twice as many bunches would imply, especially if ultimate luminosity bunches are being used as well. So realistically this is not likely to happen this year either, however, we can expect some tests of 25ns spacing this year during further scheduled machine development times and a switch to 25ns for 2012 is a possibility if the tests go well.

Altogether this could take luminosities up to a remarkable 10000/μb/s next year, if everything works out. 5000/μb/s is probably a more realistic target. Further down the line there will be upgrades allowing for yet higher intensities and a better squeeze. Peak luminosities of 50000/μb/s are anticipated after a few years. At some point the pile-up from collisions coming too frequently to be separated will be a problem and the luminosities will have to be limited. Pile-up is already expected to be a challenge for ATLAS and CMS if futher luminosity increases are introduced this year. Even then it is worth pushing the peak luminosity higher because it will make it possible to sustain the maximum desired luminosity during much of the length of a long run. This is already being done for LHCb and ALICE.

With so much data the potential for new discoveries and precision measurements over the coming years is huge. In the shorter timespan of the next two or three months we can already anticipate some great results. There are two conferences this summer that are especially worth following. The Physics at LHC 2011 conference starting on June 6th will be an opportunity to show results using at least 200/pb of data collected up to now. That is significantly more than the 40/pb already analysed in detail. On July 21st another meeting for the International Europhysics Conference on High Energy Physics could already be presenting first results from as much as 2000/pb worth of collisions. However, if there are any really important discoveries forthcoming there will be a longer process of analysis, checking and consultation before going public. Rumors can of course be based on what they don’t show straight away.

33 Responses to LHC Luminosity Prospects

  1. JollyJoker says:

    1500/ub/s is 16/fb in 124 days, unless I’m mistaken. Multiply by the Hubner factor and you get about 6-8/fb this year, right? Triple the luminosity for 2012 and guess a bit less than double the days, giving maybe 40/fb in 2012.

  2. Luboš Motl says:

    “2056, 1200 and then 1380” – is that some new kind of ordering of integers? Or what did I exactly misunderstand?

    • JollyJoker says:

      Presumably 2056 is supposed to be 1056.

      If you remember, we had some discussion on your blog back in January on whether the difference between one and two years of 3,5TeV running would be significant. Given the guesses in my post above on integrated luminosity, the worst case 115GeV SM Higgs would give a roughly 3,5 sigma signal by the end of 2011 and seven sigma by the end of 2012.

      Any guesses on whether the Standard model will be rejected at five sigma a year and a half from now? 🙂

    • Philip Gibbs says:

      fixed, thanks

  3. carla says:

    I count 119 days left for physics runs after 7th June and there are 3 technical stops left. If it takes a week to get things running as they were after each technical stop, then we’re left with 98 days.

    This year is obviously a combination of collecting at least 1/fb and machine development looking at the calendar and I’d be very surprised if they manage to collect more than 5/fb with the remaining days left.

  4. Bill K says:

    I give up – – in the conference poster, why is everybody throwing things at V838 Monocerotis??

    • Lawrence B. Crowell says:

      I was sort of wondering the same thing! I like the guy trying to drive a ball at Monocerotis. Hell of a damned golf course that is. Since a putt is low energy I guess this course has parr of one.

    • Philip Gibbs says:

      Don’t know, but worth zooming in for more details

    • Ulla says:

      They throw protons? After all it is a proton smasher? 🙂

    • Philip Gibbs says:

      If you click on the picture and zoom in you will see that the balls being used in 11 different sports are each marked with a different fermion from the standard model. One is missing.

      The structure is ATLAS and there is another lab-coated couple having a pont-neuf style kiss at the side.

      I feel I’m missing something in the symbolism.

  5. felix says:

    Hi Philip, will the LHC continue to increase luminosity during breaks from June to late October? Or will they settle down once in June?

    • Philip Gibbs says:

      I think they will find it hard to resist pushing the luminosity a little further by some means as discussed in this post, but they wont do anything that requires a long setup and will probably have plenty of routine tasks scheduled for the MD time. We will have to wait and see.

  6. algernon says:

    well, it seems the first fill with 912 bunches (#1799) only produced 850/ub/s at peak – that’s what they claim on ATLAS’ website right now.

    This is basically just a little bit better than the old scheme with 768 bunches and quite far from 1000/ub/s… I guess some tweaking is needed?
    Or maybe they were overestimating lumi before the technical stop? I’ve read they intended to tweak the estimates too, since ATLAS and CMS used to give different values while they really should be similar.

    • Philip Gibbs says:

      If ATLAS had trusted their previous figures they would have recorded a peak luminosity of 906 so they have already compensated for that.

      I think there must have been something wrong on this fill. Perhaps they had some bunches in the wrong buckets or worse than usual emittance, it can happen. The morning report might tell us.

      Hopefully the next fills will do better.

    • Philip Gibbs says:

      The second fill was better at 940/μb/s for a new record. Now they are well over twice the Tevatron luminosity. Some plots on the morning reports seemed to indicate that some bunches were producing about half the normal rate. I’m guessing there were problems with emittance blow-up in places. They dumped one time this morning for that reason.

      It may prove hard for them to reach 1000/μb/s with 912 bunches. Also the cryogenics and vacuum have been pretty stressed at this intensity. Perhaps that will ease with scrubbing. If they can keep good long runs it is a good sign. Not sure if my understanding is accurate though.

  7. Kevin says:

    They just have filled with 1.27 E11 protons per bunch, they’re gonna reach 1000/μb/s…

    • Philip Gibbs says:

      They are more than 10% over nominal intensity but the luminosity has come out worse. There must be some problems. I’m guessing it’s blow-up of some sort. Some of the bunches may become unstable and spread out reducing their luminosity. Hopefully they can find a tweak to fix it.

      • carla says:

        Now they’re filling at 750/pb/s rather than 850/pb/s. I wonder what’s the point of this. Maybe they’re carrying out studies to sort out this blow up problem.

      • algernon says:

        Fill #1802 seems a total triumph, with ATLAS reaching as high as 1100/ub/s…

        Great! 🙂

      • felix says:

        CMS too!

      • Philip Gibbs says:

        That’s more like it!

      • carla says:

        On Tuesday they’re doing a TCLIB adjustment with physics, so perhaps this will mean running at 1100/ub/s for longer.

      • Philip Gibbs says:

        TCLIB is a collimator used at injection. Perhaps they need to adjust it to reduce injection losses. Hopefully it wont take too long and they will be back for more physics runs quickly with better luminosity from 912 bunches.

      • carla says:

        Finally they’re refilling after the cryo problem. It’s very noticeable that the beam size was very stable during the ramp and is still stable after an hour. This bodes well for when they decide to increase the number of bunches.

      • Philip Gibbs says:

        They are now big factors beyond anything a collider has been asked to do before so they will move more slowly.

        I think they fixed the problem with losses at injection so the luminosities are better now. Perhaps there are still other problems.

        I think the e-cloud is also preventing them from taking the next step too soon. The cryogenics and vacuum systems have to work harder to remove the extra heat and gas. They may need to wait for the scrubbing effect to reduce the overload to reasonable levels before they can increase bunch numbers again.

      • Philip Gibbs says:

        The morning report says they are ready for next step up to 1056 bunches. Could be later today or tomorrow.

      • carla says:

        Machine coordination planning says possibly 1092b. Wow, fingers crossed 🙂 just another 2 increases after that. When will they have to change the 108b inj?

      • Philip Gibbs says:

        They were going to use 144 bunch injection for 1056 so maybe they found a new pattern for 1092 that can be done with 108. Good stuff.

        I hope they let this fill run to over 30/pb though, it would be nice to see a new record total for one fill.

  8. tpqr13 says:

    anyone has any idea why so many of the ATLAS events shown at http://atlas-live.cern.ch/ seem to be with just one or no particle tracks at all?

    Also the background grey lines seem to be a lot more intense.

    (I’m not a physicist, I’m just asking out of curiosity)

    • Philip Gibbs says:

      Most of the collision events are very dull with particles just scattering off each other and not producing very much. The experiments use triggers to select the few that seem more interesting and record them for latest analysis. The ones you see on the live display have already passed these basic selection criteria.

      Even then, only a small number of those are likely to contain signals of new physics, and those signals can only be extracted after analysing many collisions and averaging over all the known effects. ATLAS has so far produced 25 million million collision events and is just getting started.

      The collisions which produce the most particles are not necessarily the most interesting. When a Higgs particle is produced it may decay into two high energy photons or two muons with a few lower energy charged particles produced from the noise. Other interesting events produce jets of hadrons when quark pairs or other more unstable particles are produced. All these different types of events are analysed to figure out what is going on.

%d bloggers like this: