LHC Experiments: Understanding "Turning On" the Collider

[DiracPool]

I understand basically what goes on at the LHC: They take protons, run them around the accelerator, they collide in the "experiment centers" such as Atlas and CMS, and then record the results of the collisions from detectors that span out radially from the collision site. That part is clear enough. What I haven't been able to find after much searching, though, is exactly what it means for the LHC to be "turned on" during an experiment.

I mean, I've seen documentaries where someone is being interviewed outside the LHC and they're talking about how they can't go in and show the interviewer the guts of the machine because it's "in operation." What I want to know is what does "in operation" mean, and how are these experiments run. They always say they generate an enormous amount of data out of each "run." How long is each "run." One hour, one minute, all day long, all week long?

Is this something where they just need to turn the collider on for a few minutes in order to create millions of collisions and generate tons of data? Or is it something where the collisions are relatively rare so they have to leave it running for hours or days at a time to get any meaningful data?

 

[kurros]

I understand basically what goes on at the LHC: They take protons, run them around the accelerator, they collide in the "experiment centers" such as Atlas and CMS, and then record the results of the collisions from detectors that span out radially from the collision site. That part is clear enough. What I haven't been able to find after much searching, though, is exactly what it means for the LHC to be "turned on" during an experiment.

I mean, I've seen documentaries where someone is being interviewed outside the LHC and they're talking about how they can't go in and show the interviewer the guts of the machine because it's "in operation." What I want to know is what does "in operation" mean, and how are these experiments run. They always say they generate an enormous amount of data out of each "run." How long is each "run." One hour, one minute, all day long, all week long?

Is this something where they just need to turn the collider on for a few minutes in order to create millions of collisions and generate tons of data? Or is it something where the collisions are relatively rare so they have to leave it running for hours or days at a time to get any meaningful data?

"run" is kind of a vague, non-technical term. It generally just refers to a period where the machine was running more or less constantly (i.e. with only short shutdown periods relative to the length of the "run"). So people are talking now about "run 2" of the LHC starting soon, with everything it has done so far being run 1. But it's not like the machine was on 24/7 for the last few years. There have been several periods of machine operation in different configurations throughout that time which people also refer to as "runs", i.e. the 2010-2011 runs with COM energy of 7 TeV, the heavy ion runs, the 2012 run at 8 TeV. During these periods the machine *is* running 24/7 as much as possible, and they only turn off the beams when they need to do some maintenance and other such things, for say a day or a week here and there.

Anyway the simple answer is that they run the machine as much as they possibly can, since yes the interesting processes are very rare and you need to do as many collisions as possible to create enough of these rare events that you can notice them above the background.

Oh and "turned on" just means that the beams are turned on. I.e. there are protons circulating in the ring and collisions occurring.

For some interesting plots showing the data collection over time (which tells you something about when the machine was running) you can check out some of the ATLAS public data (I'm sure CMS have something similar too) https://twiki.cern.ch/twiki/bin/vie...tyPublicResults#Multiple_Year_Collision_Plots

This one in particular shows you how much of each year they were running for. The integrated luminosity is a measure of how many collisions have occurred up to that point. Where the graph goes flat basically means that the LHC beams were turned off for maintenance. You see that each year the machine was running most of the time between early March and late October, but also kept going through to December in 2012.

 

[mfb]

There are multiple different meanings for "run", unfortunately.

One is on the timescale of years - 2010 to 2012 was "run 1", 2015 to ~2017 will be "run 2" and so on. In between those runs, the accelerator and experiments can be accessed, checked and upgraded on a large scale.

Then you have "runs" like "proton-proton collisions in 2011" or "lead-lead run in 2012", with many months for proton-proton collisions and typically one month for lead-lead and/or lead-proton collisions.

On a shorter timescale, there are "runs" of continuous collisions. Those typically last hours (ideally something like 10-15 hours, sometimes as long as a day, sometimes shorter due to technical issues). During such a run, collisions happen at a high rate - in 2012 up to something like 700 millions per second in ATLAS and CMS, with a lower rate in LHCb and ALICE. Those runs typically fill ~40% of the time, the rest is used for machine development, maintenance tasks, technical problems, ramping the magnets up and down, filling new protons in and various other things. 100% would be optimal to collect more data, but that is unrealistic.

Both ATLAS and CMS had about 2*10¹⁵ collisions so far. Most of those collisions are not very interesting because the physics of them has been studied in previous accelerators. In addition, there is no way to record or analyse all of them. Just ~10000 out of those 700 million collisions per second get studied in detail. The interesting processes are rare, so you still have to collect data for several months for most analyses.

 

[DiracPool]

Thanks to the posters, that's the information was looking for.

 

[nikkkom]

To piggyback on this thread: IIUC beams in colliders on the scale of LHC are sufficiently energetic to do all kinds of harm to life. The beam tunnel is closed during operation, but I don't know exactly why.

Are there significant levels of radiation there when beam is on? I imagine it can be bremsstrahlung not being fully absorbed by beam pipe?
Or it is just because there are many possible equipment failure scenarios which can kill a human (liquid helium leak in kilogram quantities is a bad thing)?
Or maybe magnetic fields are high enough to be dangerous?
Are there parts of accelerator which become significantly radioactive after a few years in use? I know of one: the beam dump. I take it beam dump gets quite radioactive (receiving an ultrarelativistic proton beam which heats graphite to ~1000 degrees is no fun), but how radioactive is it exactly? How it will be disposed of?

 

[mfb]

Are there significant levels of radiation there when beam is on?

Yes.
Some beam particles hit the walls, which leads to cascades of high-energetic particles.
RF cavities produce some radiation.
The collision areas are full of high-energetic particles and free neutrons produced from the interactions of those particles with the surrounding material.

The magnets are not dangerous on their own, but they make everything ferromagnetic more dangerous.

Failure scenarios are an issue as well - the beam has enough energy to burn a hole through nearly everything if it gets lost. The magnets, when ramped up, have even more stored energy (but distributed over the whole ring).

Are there parts of accelerator which become significantly radioactive after a few years in use?

The beam dump gets extremely hot (here: in the radioactive way), and the detectors become radioactive as well. This issue was not so important in the current long stop, but it will become more problematic in the next years. It is expected that ATLAS and CMS will need some rotation in the personnel operating at the innermost parts in order to stay within the allowed dose limits per person.

 

[nikkkom]

> Are there significant levels of radiation there when beam is on?

Yes.
Some beam particles hit the walls, which leads to cascades of high-energetic particles.

Can you give some figures? When beam is on, the radiation in the tunnel is how many mSv/h?

The beam dump gets extremely hot (here: in the radioactive way)

Approximately how hot it is expected to become by the time LHC is decommissioned?

 

[mfb]

Can you give some figures? When beam is on, the radiation in the tunnel is how many mSv/h?

That really depends on the position. As no humans are there, it seems that no one bothered to calculate mSv values humans would receive.
This talk discusses the radiation hardness of glass fibres from .1mGy/h to 625 mGy/h.
The inner parts of the big detectors will reach irradiation levels in excess of 1MGy with the HL-LHC, or >50Gy/h.

For radiation levels with beam off:
Here I found the quote "the induced activity in the dump caverns will only lead to dose-rates of the order of several tens of µSv/h" and "Iron shields of approximately 1 metre thickness will be required in some places to reduce dose rates from induced activty to tolerable levels in the passage-way alongside the machine elements" - but this was before the HL-LHC was planned...
And it really depends on the cooling time.

This short article has various numbers up to 1 mSv/h after waiting four months for the components to cool down.