LHC ended 2016 proton collisions - exceeded all records; now: proton-lead collisions

  • I
  • Thread starter mfb
  • Start date
  • #1
35,248
11,502
Data collection can begin! This night the luminosity ("collision rate") was negligible (0.05% of the design value), but it should go up quickly as more and more bunches are filled in for the runs.

By August we might know if the diphoton excess is something real or just an extremely weird statistical fluctuation.

Edit: Another run, now with 0.2% the design luminosity.
Edit2: Another run, 0.4% of the design luminosity.
 
Last edited:
  • Like
Likes quantumgeography, Amrator, Andreas C and 13 others

Answers and Replies

  • #2
ProfuselyQuarky
Gold Member
817
527
0.05% of the design value
Edit: Another run, now with 0.2% the design luminosity.
I daresay that that is a big jump.
 
  • Like
Likes makemoneyu
  • #3
35,248
11,502
Well, still irrelevant. We had about 3/fb integrated luminosity last year, collected at about half the design luminosity for several weeks. Adding 0.0006/fb luminosity from this morning does not really help. Time for the machine operators to verify that nothing is in the way for a higher luminosity, and time for the experiments to check that everything works as expected. Ramping up the luminosity should be much easier this year, due to the experience gained last year.

We'll get another run with something like 0.2% of the design luminosity this night, then LHC will do "scrubbing". The name is more fitting than it might look like: the ring is filled with a lot of particles that are kept in the machine for as long as possible. Some of them will hit the beam pipe and remove imperfections there. You don't want this happening too much during regular operation at the full energy (would be bad for the magnets), so it is done at a lower energy. Scrubbing will probably take 3-4 days. Afterwards the LHC returns to deliver stable beams, with a quickly increasing luminosity.
 
  • Like
Likes ProfuselyQuarky
  • #4
ProfuselyQuarky
Gold Member
817
527
We'll get another run with something like 0.2% of the design luminosity this night, then LHC will do "scrubbing". The name is more fitting than it might look like: the ring is filled with a lot of particles that are kept in the machine for as long as possible. Some of them will hit the beam pipe and remove imperfections there. You don't want this happening too much during regular operation at the full energy (would be bad for the magnets), so it is done at a lower energy. Scrubbing will probably take 3-4 days. Afterwards the LHC returns to deliver stable beams, with a quickly increasing luminosity.
Never heard of that before! So, this “scrubbing” is pretty much just a way to “clean” the vacuum?
 
  • #5
35,248
11,502
Mainly the beam pipe, but yes.

Protons hitting the beam pipe release electrons. Those electrons can have a high energy, impacting the beam pipe again, releasing more electrons... if electrons come close to the proton bunches, they can get accelerated, hit the beam pipe again, release more electrons... this effect is called "electron cloud". It is not a runaway effect, but it leads to significant heat load in the beam pipe and the magnets around it. The magnets are superconducting, if they get too much heat they quench (stop being superconducting). The number of electrons released goes down over time, scrubbing is trying to accelerate this process as much as possible. At the low energy (=low magnetic field), the magnets tolerate more heat than at the full energy.

Edit 28th: Now 50 bunches, 2% design luminosity.
 
Last edited:
  • Like
Likes atyy, dlgoff and Lord Crc
  • #6
35,248
11,502
After the weasel incident (see e.g. BBC), that damaged a transformer and lead to a CERN-wide power cut, everything is back running now. Collisions resumed, and we are back to about 2% design luminosity (with 50 colliding bunches), with plans to increase that quickly. Tomorrow night we probably get 300 bunches, for more than 10% the design luminosity. That will become notable for physics analyses.

Afterwards we'll see how fast the luminosity and the amount of collected data can go up. One of the preaccelerators has an issue with its vacuum, which limits the number of bunches that can get delivered to the LHC. It is unclear how fast that gets fixed, and how many bunches they manage to inject while that issue is still there. Certainly more than ~300, but certainly not as much as planned (up to ~2700) until that vacuum issue gets fixed.

For reference: last year the LHC reached up to ~50% the design luminosity. All the values are for ATLAS and CMS, LHCb has a lower luminosity and ALICE has a much lower luminosity.


Edit: Sunday afternoon: Stable beam with 300 bunches, ~10% design luminosity for ATLAS/CMS (LHCb: 1/4 of their design luminosity).
 
Last edited:
  • Like
Likes nrqed, 1oldman2, Imager and 4 others
  • #7
314
162
Interesting stuff man, keep us posted.
 
  • #8
35,248
11,502
They just went to 600 bunches, 21% design luminosity for ATLAS/CMS (for LHCb: 50% of their lower design value). As with the previous steps, they want 20 hours of stable beams at that intensity before they move on, probably to ~900 bunches (~1/3 design luminosity). Each step takes about 2 to 3 days and typically adds ~300 bunches.

At higher beam intensities, things get a bit more problematic. Electrons in the beam pipe get more problematic (see post #3, scrubbing), which probably slows down the intensity increase beyond 1500 bunches, and might need additional time for scrubbing. Another potential issue appeared in one of the preaccelerators (SPS): It has a small vacuum leak. Gas in the beam pipe leads to losses of protons, which heats all the elements around it - not good. It currently limits the number of bunches the SPS can have at the same time, which will then limit the number of bunches it can inject into the LHC. It is unclear when exactly this limit will be hit, and if the leak can be repaired before that.


For the experiments, it is a race against the clock. The most important conference this summer is ICHEP (3rd to 10th of August). All the experiments want to present new results, and improve the precision compared to 2015. Take as much data as possible? Well, you still have to analyze it, the more data you include (=the more relevant your result might become) the less time you have for the analysis (=less time for all the necessary cross-checks, especially for important results).
Last year both ATLAS and CMS presented first results 6 weeks after data-taking ended, that would point to June 22nd. That is soon, if you take into account that there will be machine development and a technical stop in between (~2 weeks), and the LHC is still running at low collision rates. For possible impact on the diphoton excess, see here.


Collected luminosity so far for ATLAS+CMS, excluding the current run: 84/pb = 0.084/fb. (LHCb: 4.4/pb)
As comparison: last year we had 4/fb.
What is this weird unit?
 
  • Like
Likes nrqed, atyy and Garlic
  • #9
334
43
How long would it take to repair the leak in the SPS? I'm guessing it needs to be heated up and cooled down in a similar to the main ring itself, and finding the culprit may take some time?

What I'm trying to get an idea of is how few bunches they can live with before it's better to take the repair downtime.
 
  • #10
35,248
11,502
How long would it take to repair the leak in the SPS?
That is unclear.
The leak is at the beam dump, and the SPS does not use superconducting magnets, so heating/cooling times are not an issue.

Two good runs with 600 bunches per beam increased the collected integrated luminosity to 195/pb. The step to 900 bunches is planned for the weekend.
 
  • #11
ChrisVer
Gold Member
3,373
459
is the IBL turned on? :biggrin:
 
  • #12
35,248
11,502
The ATLAS IBL? I don't know, if you work for ATLAS ask your coworkers. Why should it be off?
 
  • #13
334
43
That is unclear.
The leak is at the beam dump, and the SPS does not use superconducting magnets, so heating/cooling times are not an issue.
Thanks, for some reason I thought the Super bit had something to do with superconducting... No idea where that came from, I'm blaming a lack of caffeine.

IIRC they upgraded the SPS beam dump last year, maybe just coincidence?
 
  • #14
35,248
11,502
There is a smaller and older Proton Synchrotron, they just named the next bigger machine Super Proton Synchrotron.
IIRC they upgraded the SPS beam dump last year, maybe just coincidence?
Most things get upgraded frequently. Found this meeting from last year about upgrading the SPS beam dump.
 
  • #15
334
43
There is a smaller and older Proton Synchrotron, they just named the next bigger machine Super Proton Synchrotron.
Yea I know about PS, I just really need to stop posting too early in the morning I think :)

In any case I find it impressive that they don't have more issues, given the complexity of the whole thing. That said it must be really frustrating for everyone involved to have this string of issues given the earlier hints of something new.

Anywsy, thanks again.
 
  • #16
35,248
11,502
900 bunches in now, initial luminosity for ATLAS/CMS was 30% the design luminosity. Total integrated luminosity as of now: 290/pb.
LHCb values are about 5% of the ATLAS/CMS values.


For ATLAS and CMS, the two experiments with the highest luminosity, the bunches are made to collide head-on. As the bunches lose some protons over time (from collisions in the experiments but also from losses in the machine) and the focussing of the bunches gets worse over time, they start with a high luminosity which then goes down over the lifetime of a fill (typically a few hours).
LHCb cannot handle the high collision rate the LHC could deliver. There, the beams are shifted a bit so they don't collide head-on. While the intensity and focus quality goes down, the shift is reduced, so the collision rate stays constant all the time, and LHCb is always operating at the optimal collision rate.
That said it must be really frustrating for everyone involved to have this string of issues given the earlier hints of something new.
In 2008, we hoped to have 14 TeV collisions in 2009 or even 2008. One or two weeks delay don't really matter in the long run.
 
  • Like
Likes nrqed
  • #17
135
48
As a layman I'm trying really hard to understand what you folks are talking about. But don't dumb it down too much cuz then I won't learn much. However, I have a question. It may sound dumb but here goes. What happens if a magnet fails while the protons are circulating? Will the protons "hit the wall" or another mag and do a lot of damage? Or is there some kind of back-up in place to keep them on track? Or does that only seem like a lot of energy because of the density but is actually not a big deal? I was under the impression that a freight train was flying through that thing.
Those things must be synchronized pretty tightly in order for that energetic mass to stay 'on track'.
 
  • #18
Vanadium 50
Staff Emeritus
Science Advisor
Education Advisor
26,351
9,836
If a magnet begins to fail, the beam is steered to the dump. This takes about 3 microseconds.
 
  • #20
35,248
11,502
The beam has a lot of energy, and could burn a hole through the machine if it would not be contained within the beam pipe. The bending magnets store even more energy, however - and that energy does not disappear at once. The magnets are superconducting coils in a closed circuit, during operation they do not need additional power - as long as they stay cold, they work. If they get too warm, the coils get a resistance and current starts to drop - but slow enough to dump the beam before the magnetic field gets too far away from its design value.
 
  • Like
Likes hsdrop, Uranic_Wabbit and dlgoff
  • #21
dukwon
Science Advisor
Gold Member
80
62
If a magnet begins to fail, the beam is steered to the dump. This takes about 3 microseconds.
The kicker rise times are 3 microseconds. Machine protection dumps happen within 3 orbits (~300 microseconds) of something going wrong.
 
  • Like
Likes Greg Bernhardt and mfb
  • #22
Vanadium 50
Staff Emeritus
Science Advisor
Education Advisor
26,351
9,836
That's right -the steering takes 3 microseconds, and then the travel time to the dump is whatever it is.
 
  • #23
35,248
11,502
Well, the travel time is up to one orbit. The kicker magnet can only ramp up in the abort gap (a region without bunches), so in the worst case we have to wait nearly one orbit until the kicker magnet can ramp up, and then another orbit until all bunches are out. That leaves one orbit time (~90µs) for the accelerator system to decide that the beams have to get dumped. Not much time, given that the signals do not travel faster than the particles (but have a shorter path).


Edit: 470/pb, the 900 bunch step is done, next will be 1177 bunches tonight. Some issues in the preaccelerators degraded the beam quality in the last runs, so the luminosity might be a bit lower than the 40% you could expect from a linear extrapolation. Maybe 30% to 35%.
Edit2: Stable beams with 1177 in the night to Wednesday. Initial luminosity was somewhere between 30% and 35%. Just 4 hours of stable beams, unfortunately.
 
Last edited:
  • Like
Likes nrqed and ProfuselyQuarky
  • #24
35,248
11,502
One of the preaccelerators (the PS) had a fire or something similar Friday morning, and won't work before Wednesday. It was decided to keep the current fill in as long as possible. Data-taking started at around 5:00, now we are at 21 hours. Not the longest fill ever (yet? I think the record is a bit below 30 hours), but still a lot of data: 190/pb for ATLAS and CMS so far, 14/pb for LHCb. For ATLAS and CMS, luminosity dropped from 35% design value to 19%, while LHCb is running constantly at its chosen luminosity (~90% of its design value).

Collected integrated luminosities as of now:
ATLAS/CMS: 690/pb
LHCb: 46/pb
~30% of that from the last 24 hours!
 
  • Like
Likes Lord Crc and Drakkith
  • #25
dukwon
Science Advisor
Gold Member
80
62
Not the longest fill ever (yet? I think the record is a bit below 30 hours)!
The longest fill was #2006 at 25:59:08
 

Related Threads on LHC ended 2016 proton collisions - exceeded all records; now: proton-lead collisions

  • Last Post
Replies
12
Views
4K
Replies
57
Views
11K
Replies
7
Views
3K
  • Last Post
Replies
6
Views
3K
  • Last Post
Replies
1
Views
2K
  • Last Post
Replies
2
Views
3K
Replies
8
Views
2K
  • Last Post
Replies
2
Views
29K
Replies
1
Views
3K
  • Last Post
Replies
7
Views
3K
Top