How easy is it to switch between 25ns and 50ns operation?

[jmc2000]

Round of applause for the guys on the LHC - 6.6/fb delivered! unbelievable! ICHEP will rock in July!

They have to do some 25ns operation this year to get some useful results for the long shutdown, so I'm curious how they're going to do this, as things progress. If things progress at the recent rate, then a conservative estimate for the additional integrated luminosity over the next 6 week slot will be 5.5/fb giving 12/fb. They then have a final 7 week slot to accumulate the remaining 3/fb for a 2012 total of 15/fb requested and delivered

*switch over to 25ns over the technical stop in preparation for a 25ns start
*wait until 15/fb total delivered at 50ns, then switch to 25ns

The stratedy chosen will be influenced by how easy it is to switch between 25ns and 50ns so - what's involved? days of preparation? a simple download of new firmware?

Cheers in advance,

Jmc

[chriwi]

In my opinion biggest problem of 25ns will be the handling, reduction and the interpretation of the collected rawdata. They will have more collisions at least over short periods of time so it will be harder to tell them apart.
Also the problem with the overheating of the injectionkickers will sure not become less with even closer and maybe more bunches.
I wonder if 25ns spacing would really bring much more usefull data per day than the 50ns setup.

[DCWhitworth]

JMC,

I think they will run flat out for the full time of the machine running if they can. The 13.5 fb-1 figure is merely an estimate of what they need to find (or not) the Higgs, it's a best guess. If thay can provide more data than that they almost certainly will.

They had a go at 25 ns last year but hit enough issues for them to prefer to stick with 50 ns. I think 50 ns allows them to go with larger bunch sizes/higher intensity and thus the luminosity gain for switching to 25 ns wouldn't be as big as you might think.

The upside of switching to 25 ns spacing is the pile up issue. With more bunches, but less intensity they will get less pile up and this clearer data.

[DCWhitworth]

In my opinion biggest problem of 25ns will be the handling, reduction and the interpretation of the collected rawdata. They will have more collisions at least over short periods of time so it will be harder to tell them apart.
Also the problem with the overheating of the injectionkickers will sure not become less with even closer and maybe more bunches.
I wonder if 25ns spacing would really bring much more usefull data per day than the 50ns setup.

Not sure I agree with you here chriwi. The experiments and their associated equipment were designed and built with 25 ns spacing in mind so they will probably be better able to handle that than the current situation.

In order to run at 50 ns (i.e. 1440 bunches instead of 2880 at 25 ns) and still get as much luminosity as possible they have pushed the LHC beyond it's original design in terms of bunch intensity. A move to 25 ns will probably involve a reduction in bunch size/intensity and hence drop things back into the design criteria.

They've addressed some of the data issues by inserting storage between the level 1 and level 2 triggers, so that they can buffer the data and not lose any if pile up becomes too bad. Beyond the level 2 trigger it's just a matter of buying more CPUs !

I don't know what is going on with the injection kickers heating up. It's not an issue I've come across before and I don't know if it was expected or not. Presumably they will look at ways of addressing it in LS1. If it's related to bunch intensity then presumably it will be eased by 25 ns too.

[chelle]

I wonder if 25ns spacing would really bring much more usefull data per day than the 50ns setup.

In order to run at 50 ns (i.e. 1440 bunches instead of 2880 at 25 ns) and still get as much luminosity as possible they have pushed the LHC beyond it's original design in terms of bunch intensity. A move to 25 ns will probably involve a reduction in bunch size/intensity and hence drop things back into the design criteria.

@DC, There is little logic in your comment, first you say that they boosted up the LHC to get as much luminosity, and now they'll double the frequency you suggested that they will lower the intensity, leveling the effect out. It seems that the only thing that they want to do, is to produce the highest luminosity possible.

@Chirwi, do you mean that doubling the results would keep the percentages the same, or does the number of results works differently when getting the five-sigma approval stamp?

[DCWhitworth]

@DC, There is little logic in your comment, first you say that they boosted up the LHC to get as much luminosity, and now they'll double the frequency you suggested that they will lower the intensity, leveling the effect out. It seems that the only thing that they want to do, is to produce the highest luminosity possible.

You missed one of my points from my previous post - "They had a go at 25 ns last year but hit enough issues for them to prefer to stick with 50 ns. I think 50 ns allows them to go with larger bunch sizes/higher intensity and thus the luminosity gain for switching to 25 ns wouldn't be as big as you might think."

I cannot recall or remember why, but if I'm remembering correctly they would not be able to have such a big bunch size at 25 ns.

If the switch to 25 ns would double luminosity (as you seem to think) then they would already have done it, it would be worth giving up a lot of machine time to get it to work.

But the debate was - move to 25 ns with a lot of development time or stick to 50 ns and slowly edge the bunch intensity up.

Currently the bunch intensity is at 1.5 e11 (design limit - 1.1e11) and luminosity is at 6.7 e33 (design limit 1 e34). So they're two thirds of the way to top designed performance in luminosity. No doubt if they move to 25 ns they'll find ways to push it over 100% of design limit, but I doubt they'll double the 50 ns luminosity.

[Kasuha]

As far as I remember, important factors with (not) going to 25 ns were that it would require long scrubbing runs and long beam commisioning period. While it was relatively certain they can meet the target with 50 ns beams and already prepared parameters, they'd run into much less certain land with 25 ns beams and shorter remaining time.

They are planning some 25ns MD runs (likely with few and/or low intensity bunches) this year to be ready to start serious 25ns runs right after the LS1.

[terryburton]

I don't know what is going on with the injection kickers heating up. It's not an issue I've come across before and I don't know if it was expected or not. Presumably they will look at ways of addressing it in LS1. If it's related to bunch intensity then presumably it will be eased by 25 ns too.

From the Monday weekly summary report (slide 21), they appear to be planning to remove the bake-out jackets during next week's technical stop to provide only a small improvement, but a more adequate solution appears to rely on replacing hardware during the following technical stop.

[cypherpunks]

Old thread, but just to answer... not very hard, but currently 25 ns bunch spacing causes electron cloud effects that cause excessive beam losses.

The experiments would very much prefer 25 ns. They want as much luminosity (which causes collisions) as possible, but spread out enough that it's not hard to figure out what's going on inside the detector.

With fewer, larger bunches, it's easy to cause a lot of collisions (the number of collisions is proportional to the product of the two bunch intensities, so 1.414x more protons in each doubles the collision rate), but it also causes pileup, which is the number of different collisions per bunch crossing.

High pileup makes it hard for the detectors to figure out what the heck happened. Detectors want high luminosity with low pileup, which means collisions every 25 ns (or even more often, if possible), every 25 ns, 24/7.

However, bunches very close together suffer from electron cloud effects. The protons in the bunch attract stray electrons, which gives them enough energy to hit the walls of the beam pipe and release more electrons. If the next bunch comes too soon, it will accelerate these new electrons and the number of loose electrons in the beam pipe rapidly becomes enough to interfere with the beam focus. Bad bad bad.

Anyway, although the machine can technically operate at 25 ns bunch spacing, the e-cloud effects are so bad that the beam is quickly degraded, you haven't made a neat gun, and science doesn't get done.

There is a process known as "scrubbing" that can be done where you deliberately cause this effect and it's known to eventually "clean" the beam pipe to knock fewer electrons loose as part of this process. However, in the last couple of years it was decided that the time taken away from physics to do the scrubbing wasn't worth the gain from 25 ns spacing.

This is because there have been improvements in the detectors' software which lets them deal with bigger bunches and more pileup. That's the hidden part of increasing luminosity you don't see: there's no point producing more than the detectors can use. Already, they are talking about "luminosoty leveling" where a slight deliberate misalignment of the beams reduces the collision rate at the beginning of a fill.

Anyway, they've got the acceptable protons/bunch up to the point that having that many protons with 25 ns spacing would exceeed a different limit on the total number of protons in the ring, so to go to 25 ns spacing actually requires reducing the number of protons per bunch, so you don't get the full 2x luminosity gain.

Now, if they can tighten the focus to increase a different luminosity factor without having to increase the proton numbers, then they will be able to reach the detector limit with 25 ns bunches, and the 2x gain will start to look very tempting again.

[chelle]

Thanks 'cypherpunks' for scratching this thread open again, it made me remember an article from about a month ago:

http://www.symmetrymagazine.org/article ... _issue=205

Smallest lab-made drop of liquid might cause strange particle behavior

Scientists think a drop of hot plasma spurts away from the high-pressure epicenter of collision to lower-pressure areas, and newly created particles are flushed out horizontally around the line of the incoming beam. The fluid cools as it expands, but the droplet remains liquid long enough to sweep particles before it like debris on wave.

... and yes I'm going to BS again about safety, but aren't these droplets kinda like a vapor that hangs around the collision center? ... it makes me think of a vaporizer of fuel in a combustion engine, so when there are enough droplets than only a couple of extra *sparks* are needed to set of the ignition of matter.

[CharmQuark]

Good to see you posting Chelle

[chelle]

Thanks Charms ; )

[chriwi]

I am also happy to see that you are still around chelle.

But asuming for any kind of vapor that it must be combustable is a bit far fetched, I guess there are more noncombustable vapors around in the world than the epecially desind combustable vapors in engines.
We should first consider if this possible vapor as you call it could be combvustable at all or rather not.

[chelle]

I am also happy to see that you are still around chelle.

Yes, me too, but I guess that's because the LHC is in maintenance right now

But asuming for any kind of vapor that it must be combustable is a bit far fetched, I guess there are more noncombustable vapors around in the world than the epecially desind combustable vapors in engines.
We should first consider if this possible vapor as you call it could be combvustable at all or rather not.

Yes that's a fair argument.

But if I may bring up a first point regardless if the 'vapor' is dangerous or not, we haven't got a reference frame for this situation. For the safety report we look at Cosmic-rays but they happen all over the place and those that are in the reach (and beyond) the energies of the LHC are very rare, and thus there could be in the LHC a unique vapor-cloud build up, at one specific place. Check out the density and frequency to compare the situations:

In nature there are about a thousand Cosmic ray collisions of a few GeV’s (1 GeV= 10^9 electron Volt) per second per m^2. In LHC it are about one 1 billion per second per cm^2. That’s 1.000.000 times more for an area which is 10.000 smaller, it is a density & frequency difference of 10 billion.

At the end of last year we humans have even generated collisions on this planet, who were an other 1000 times more intense, with energies of 8 TeV (1 TeV= 10^12 eV). These collisions are in nature of course even less frequent per m^2 while the density & frequency at the LHC of 10 billion per cm^2 was maintained.

--

Well to be or not to be, is the 'vapor' dangerous or not; to answer that question we have to look at the conversion factor of energy into mass and visa versa, could this vapor be regarded as stored unstable energy, just like kerosine is less stable than tab water, and thus more combustable.

Another question is, can you only break protons apart, by smashing them into each other, or could you also make them decompose and break up, when an extra Quark or Meson enters the Proton's inner mechanism and shakes it up; if so let's say that you have a large amount (a cloud) of these drops hanging around, and domino-wise they suddenly all start to break open, spreading out, just like how a single *spark* ignites the fuel in your car's engine; and by doing so disrupting the inner processes of a large percentage of atoms at one area, who in turn burst open (decompose); then you have a scenario where combusting protons generate in turn a fresh amount of these drops who spread out ... and the beginning of an unstoppable chain reaction at a sub-atomic level, releasing enormous amounts of energy ... So what happens when protons come in contact with this vapor ... and the tricky part to find this out, is that you have to first create a mass amount of these droplets.