LHC finishes proton-proton collisions in 2018

[mfb]

New year, new thread! Here was 2017.

Yesterday the first beams this year circulated in the LHC. As every year, the machine operators start with a single low intensity bunch, checking that everything still works properly, and adjusting some parameters where the conditions changed over the winter shutdown. Meanwhile they slowly increase energy, protons per bunch and number of bunches, and finally the focusing of the beams. We will probably get a single full bunch at the full energy tomorrow, and then more and more bunches over the next weeks. The schedule is available online, first collisions are planned for end of April, and collision rates similar to last year will probably be reached by the end of May.
You can watch the current status online.

The total time assigned for collisions is a bit shorter than last year, but with improvements during the shutdown it is expected that more data can be collected this year than last year. In particular, the vacuum issue in 16L2 should have been resolved, allowing more bunches per ring.

Late least year the machine became able to deliver more collisions per bunch crossing than the two big experiments (ATLAS and CMS) could handle - they agreed to limit it to 60 (the design value is 25). This was maintained for about 1-2 hours until the beam had lost so many protons and the focusing decreased so much that the collision rate went down naturally. For this year the experiments adjusted their software to these conditions, they might be interested in even more collisions per bunch crossing. But even if they decide to stay at 60: The larger number of bunches will still lead to a higher luminosity. In addition, better focusing means the 60 collisions per bunch crossing can be maintained longer.
LHCb limits the number of collisions per bunch crossing to about 2, and ALICE limits it to much less than 1. These experiments look for processes where precision is more important than more collisions.

The long-term outlook: In November the LHC will collide lead ions with lead ions. Afterwards it enters the second long shutdown. In 2019 and 2020 LHCb and ALICE will upgrade their detectors significantly, CMS and ATLAS will do smaller upgrades. The accelerator will be prepared for higher luminosities, and potentially for going to the design value of 14 TeV collision energy (so far we have 13 TeV). Data-taking will resume in 2021.Related: SuperKEKB/Belle II are preparing for first collisions

 

[jedishrfu]

It seems the upgrades take so much longer than the experiments they run.

Is that a somewhat untenable situation that would affect future funding?

Basically bang for the buck.

 

[Vanadium 50]

It seems the upgrades take so much longer than the experiments they run.

Are you saying it take longer to install the upgrades than to run the experiment? That's not generally true.

Or are you saying that the experiment runs longer - or at least collects more data - in the upgraded configuration than the baseline configuration. That's often true - but isn't that a good thing?

 

[mfb]

Frequent upgrades are the usual and expected mode of operation. If you need two years of upgrades to double your data-taking rate, you have a "return of investment" time of two years after the upgrade. Unless you plan to shut down the whole accelerator earlier it is worth the time. After a few years of running after the upgrade the same calculation is done for the following upgrade.

Running without upgrades would mean you are stuck with the initial performance of the machine. In 2010 we had half the energy and 1% the collision rate of 2017. We would need 100 years of 2010-style running to match the number of collisions we had in 2017, and even then they would be at lower energy.

Since 2010 the LHC spent two full years (2013 and 2014) and a few months per year (the winter shutdown) on upgrades, machine maintenance and so on. In addition a few weeks per year are used for hardware work and machine development during the data-taking periods. These are necessary to keep the accelerator running and to improve the collision rate further.

 

[nikkkom]

Would it be possible to make a photo of the collider's pipe while it is in operation? It'd be cool to see an actual bremsstrahlung radiation.

 

[mfb]

The particle detectors are cameras, basically. Regular cameras are not suitable for x-ray images.

 

[Keith_McClary]

We would need 100 years of 2010-style running to match the number of collisions we had in 2017, and even then they would be at lower energy.

There's a computer analogy: If you have a calculation that will take 10 years on your computer, the trick is to wait 5 years until computers are 10 times faster. Then you can get your result in 5+1 years.

 

[jedishrfu]

Are you saying it take longer to install the upgrades than to run the experiment? That's not generally true.

Or are you saying that the experiment runs longer - or at least collects more data - in the upgraded configuration than the baseline configuration. That's often true - but isn't that a good thing?

I guess what I’m trying to say is it seems the LHC is offline more than it is online and we hear from news reports that it’s starting up again. I would feel from those reports that the bang is less than the buck.

The wiki article tells the history better:

https://en.m.wikipedia.org/wiki/Large_Hadron_Collider#Timeline_of_operations

Big science is expensive but scientifically the payout is greater so let’s keep it going.

 

[mfb]

I guess what I’m trying to say is it seems the LHC is offline more than it is online and we hear from news reports that it’s starting up again.

"Offline" is not right. "No collisions while it is getting improved" is a better description. Which takes up about half of the time of a typical big accelerator. This maximizes the "bang". What do you suggest? Keep running at the low initial energy and collision rates just to avoid a few news articles that the accelerator gets shut down for a while for upgrades? That would be a huge waste of money.

 

[jedishrfu]

I'm not attacking the LHC, I'm just trying to understand why it seems to be offline so much relative other big science projects like the Hubble or LIGO.

Your answers about the kinds of upgrades needed helps me to see that its impossible to continue running for certain types of changes.

There was once a plan to build a collider in Waxahatchie Texas but it got killed due to its large budget:

https://en.wikipedia.org/wiki/Superconducting_Super_Collider

So I imagine the CERN folks faced these same kinds of budget problems but still made it work.

 

[fresh_42]

So I imagine the CERN folks faced these same kinds of budget problems but still made it work.

Others dream to win in the lottery, mine is to once cash in the yearly power bill of CERN.

 

[nikkkom]

Yes, I know what detectors do. The visible light photos of the insides of the beamline while it is operating would not be for scientific purposes.

 

[mfb]

Hubble is very difficult to upgrade, so it was done rarely.

LIGO follows the same pattern as the LHC, just with more rapid iterations. I didn't find details for the initial runs but from 2002 to 2004 it had four science "runs", with upgrade phases in between. From end of 2004 to end of 2005 there was another upgrade phase. Afterwards LIGO took data until 2007, again followed by upgrades. A sixth science run was done from July 2009 to October 2010. Afterwards LIGO was stopped for major upgrades to Advanced LIGO. It then took data from September 2015 to mid of 2016, followed by a few months of upgrades, until it resumed data-taking from late 2016 to August 2017. Since then it is in another upgrade phase. Probably from mid of 2018 on it will resume data-taking for an expected 9 months, followed by the next upgrade phase. Virgo follows the same schedule to have the three detectors running at the same time as much as possible.

Compared to the initial LIGO, it now has more than 10 times the sensitivity, which leads to more than 1000 times the rate of events. Same as for the LHC: These upgrades are absolutely crucial to reach the full potential of the devices.

Your answers about the kinds of upgrades needed helps me to see that its impossible to continue running for certain types of changes.

It would be possible to keep running, but you wouldn't achieve the full potential of these machines.

The SSC was mainly killed for political reasons, but that is a different story.

Others dream to win in the lottery, mine is to once cash in the yearly power bill of CERN.

Tens of millions of Euros.

Yes, I know what detectors do. The visible light photos of the insides of the beamline while it is operating would not be for scientific purposes.

I don't think they would see anything apart from random charged particles crossing the sensors. They would quickly die from the accumulated radiation dose.

 

[jedishrfu]

Thanks @mfb !

 

[Vanadium 50]

The visible light photos of the insides of the beamline

Won't show you anything anyway. The synchrotron radiation is in the UV; about 200 nm. Getting it to a camera requires some non-trivial optics, and it's far from clear that a direct image is what you want. It is used for beam monitoring, and it's unclear that position is the most useful quantity to monitor.

 

[Vanadium 50]

why it seems to be offline so much relative other big science projects like the Hubble or LIGO.

Well, Hubble's original plan was to be brought back down for servicing every 5 years for 2.5 years. And roughly half the time its target is blocked by the earth, so it really is not that different from the LHC.

The LHC has downtime at every scale:

• A small number of year or multiyear upgrades - without which the energy and luminosity would be right where it was when the LHC started. We have already surpassed where we would be had we not had that upgrade, in some cases weeks after the 13 TeV run began.
• A technical stop of a few months at the end of each calendar year. Major repairs and minor upgrades fit in here. This is how the LHC went from 7 to 8 TeV and this is how bad magnets are repaired or replaced.
• A technical stop of a week or so every 5-8 weeks. Minor repairs and preventative maintenance fit in here.
• Refilling the machine with protons takes a few hours per day. In principle, one could wait until the beam falls out on its own, but the luminosity drops over time, and after a day or so if you want to optimize luminosity better to dump and refill than to drag it on.
• There are gaps of a few hundred nanoseconds in every beam per turn. This is to give the abort kickers time to fire and direct the beam safely to the aborts.
• There are gaps of one or a few bunches (25 ns) because of how the LHC is filled from the injectors.

 

[Jonathan Scott]

Note that the annual downtime for maintenance is normally deliberately scheduled for the winter months because that's when the electricity needed to keep it running would cost the most.

 

[mfb]

Note that the annual downtime for maintenance is normally deliberately scheduled for the winter months because that's when the electricity needed to keep it running would cost the most.

In the US it is the opposite, annual downtime in the Summer due to higher electricity prices (more cooling, less heating in the US).Over easter some protons were accelerated to 6.5 TeV and the focusing was tested as well. Now we "just" need that with 30,000 times as many protons, and then with colliding beams.

 

[nikkkom]

Won't show you anything anyway. The synchrotron radiation is in the UV; about 200 nm.

I imagine it does not disappear without anything visible. There are some absorbtion and re-radiation by walls, at lower wavelengths. I doubt the beam pipe is dark in visible. So, how exactly does it look?

Getting it to a camera requires some non-trivial optics, and it's far from clear that a direct image is what you want. It is used for beam monitoring, and it's unclear that position is the most useful quantity to monitor.

Guys, it's not about monitoring. It's about "cool science pics". PR. You know, that thing.

 

[fresh_42]

CERN has a few better PR programs. E.g.they have a "photo of the week" contest on FB where readers can guess what is seen on a photograph of usually a technical apparatus, a summer school program for students and of course their website, where such "photos" can be found, too, although not real photos of collisions.

 

[mfb]

The machine operators continue their test program to make sure the beam behaves as expected, and they are making good progress - they are a few days ahead of the schedule. First collisions might happen next week. Even if the detectors can use these collisions at all the collision rate will be small, but it is an important milestone on the way back to large datasets.

 

[dukwon]

First collisions (not stable beams) could be tomorrow mid-morning if loss maps and transfer-line collimator validation are performed during the night. First stable beams could be expected on Monday.

 

[dukwon]

We have collisions:

 

[mfb]

At 0.06% the design rate ;). Still not bad considering that only a single bunch is colliding in each experiment. ~40-50 collisions per bunch crossing for ATLAS and CMS. Scaling that up to 2500 bunches we get 1.5 times the design luminosity.

For tomorrow it is planned to inject many bunches at the same time.

 

[mfb]

Stable beams! Data-taking in 2018 begins.
Just 2 colliding bunches, this number will increase to ~2500 later.

Bunch trains were tested before, but not accelerated to high energy yet. Increasing the number of bunches is the next step.

 

[dukwon]

You beat me to it, so I will post some event displays:

 

[pinball1970]

You beat me to it, so I will post some event displays:
]

I googled what's next at the LHC and got the following link
http://www.iflscience.com/physics/whats-next-for-cern-and-the-large-hadron-collider/

There is no mention of super symmetry/particles or anything on dark matter/dark matter particles

Are those areas pie in the sky? I am sure they have been mentioned but could have been pop science/news paper talk

 

[mfb]

It does mention invisible Higgs decays briefly (relevant both for supersymmetry and dark matter), but apart from that there is just a general discussion of searching for anything new without going into details. The article focuses more on the machine and detector operation.

The searches for new particles continue, but if you don't find anything with the 2016 dataset (where most of these searches have been published already) it is unlikely that the full run 2 dataset (2015-2018) finds a clear signal of new particles. With the largely increased statistics in the following runs this might change.
Apart from that there is always precision physics: Look for deviations from the expected properties of known particles. This is a big topic with the Higgs boson now as it is quite new, but the other particles can be studied in more detail as well.

 

[dukwon]

I posted this in another forum, and maybe some people here will be interested. I tried to reduce the amount of jargon.

Good news: the LHC seems to be about 7–10 days ahead of schedule, although some of that extra time may be spent on extra "machine development" tasks rather than more physics.

Bad news: during "scrubbing" (reducing electron clouds in the beam pipe with diffuse beams) yesterday, there were five beam-dumps with the signature pattern of the 16L2 problem that plagued the LHC last year.

16L2 stands for cell 16, left of point 2. Throughout 2017 there were a number of beam dumps triggered by an initial spike in beam losses in the same location. It is believed that these were caused by air that had accidentally entered the pipe and become frozen. Warming up this area in August made the problem worse, and the eventual solution was to operate with "8b4e" beams, referring to gaps of 100ns (4 empty) after every 8 bunches in a train. This was successful but unfortunately reduced the number of bunches from 2556 to 1920.

During the year-end technical stop, 16L2 was warmed to 90 K and approximately 8.5 g of gas was extracted, which, when analysed, was found to be consistent with air. However, as we saw yesterday, this doesn't seem to have eliminated the problem.

Here are some LHC reports from last year which cover the 16L2 issue and 8b4e operation:
https://home.cern/cern-people/updates/2017/08/lhc-report-something-nothing
https://home.cern/cern-people/updates/2017/09/lhc-report-operation-holes

Here are some publicly-available slides:
https://indico.cern.ch/event/705545/
https://indico.cern.ch/event/664048/contributions/2718791/

 

[Vanadium 50]

LHC is at over 600 bunches now.

 

[dukwon]

We've had three fills with 600b. There needs to be another using 96b trains.

The timetable for the week is filled in (see below). A lot of the commissioning steps have been completed already, but there are still a few more to be done in the next couple of weeks. It's going to be solid physics from Friday afternoon and over the weekend. Optimistically, if a step in https://lpc.web.cern.ch/RampupPolicies.htm takes 4 or 5 shifts (20 hours stable beams in 3 fills), the 1200b step could start on Sunday morning. Compare this with the schedule for the year, which says 1200b on 11th May.

 

[mfb]

Stable beams with 987 bunches (96 bunches per injection). The initial luminosity was 80% the design value, and 60 collisions per bunch crossing for ATLAS/CMS have been achieved.

Luminosity statistics
ATLAS/CMS are at 0.5/fb, about 1% of the target for this year.

 

[mfb]

2300 bunches, with an initial luminosity of more than 150% the design value (~175%?). This exceeds the luminosity record for regular data-taking from last year.
0.5/fb collected in a single 14 hour long fill from midnight to 2 pm.

 

[dukwon]

The next steps are 2460b and then 2556b. I think they've abandoned the policy of 3 fills per step. I'm happy that LHCb is getting at least 400 Hz/μb, and may this last the rest of the proton run.

A combination of controlling the corrector-magnet current and adding "bumps" to both beams has dramatically reduced the spikes in losses around 16L2.

 

[dukwon]

Stable beams with 2556 bunches per beam. The intensity ramp is finished, and the machine is in full swing for physics production.