thank you… I will check the article when I am back to office – I do not have a personal subscription to Nature, my University does though.

Thanks again for notifying the piece.

Cheers,
T.

Your excellent April Fools joke has now got an honourable mention in Nature (although your blog is not explicitly named).

http://www.nature.com/nature/journal/v446/n7138/full/446836a.html

Congratulations.

I am writing from the mountains, unarmed. I think you are highlighting critical but plain-to-spot engineering issues, which I am quite sure have been addressed thoroughly during design. What happened to the G-11 support structures is instead a much subtler failure, which has much more potential for creating problems.

I remain interested in the issue, but until I come back to my office I am unable to discuss it meaningfully… I really hope we will get more information on the subject in the meantime.

Cheers,
T.

http://www.iop.org/EJ/article/1367-2630/8/11/290/njp6_11_290.html#nj230233s8.3

So the key is they have a fast kicker magnet which could quickly switch from zero to norminal magnetic field in just 3 microseconds, and when it’s fully turned on it deflect the beam 280 micro radian angle. The trick is they allow a 3 us “particle free gap” within the beam and the kicker started the 3 us switch right when the particle free gap arrives. The kicker finishes the 3 us switch-on while there is no particle within that 3 us window.

Sounds good but how do you make a fast kicker that switches on that fast? The main magnet is 14.3 meters long with a magnetic field of 8.4 tesla, and bend the beam 5.1 mili radian. For the fask kicker to bend the beam for 280 micro radian, the field strength times the length of the path within the magnetic field then must be 6.6 meter*tesla. That’s quite a big magnet to be turned on within 3 us.

Second problem is 3 us gap corresponds to 900 meters of gap within the beam. The beam is composed of 2808 bunches uniformly distributed on a 27 km circonference so the particle free gap between bunches is only 9 meters or less.

Main components of the LHC accelerator:

http://lhc-machine-outreach.web.cern.ch/lhc-machine-outreach/lhc-components.htm

Notice that everything has a valid URL link, except for the “beam dump” where there is NO url page link at all. Guess they have nothing to say about beam dump?

In general physics you know magnetic fields could not change suddenly. It could only change gradually, changing magnetic field induces electric current.and that’s how transformers work. It means you really can not instantaneously change the bending angle of a magnet from 17.5 arcminutes to 7.5 instantaneously. It will be a gradual transition, depending how fast you can allow the magnetic field energy be safely released. Such gradual change of the bending angle will mean part of the particle beam, actally 100% of them, if the change takes longer time than the round trip time of the beam. This part of the particle beam will hit the metal part of the vacuum enclosure before they could be diverted enough to enter the safety dump path. If the beam touches any metal part, it’s a huge release of energy and a huge explosion, destroying anything. I just don’t see how you can avoid that? It’s as if you are trying to exit a highway, and you are not turning the steeling wheel enough and you hit the midway curb between the highway lanes and the exit lanes. And it is a catastrophy.

Another factor to consider is if one of the magnet is quenched, and it is at least half circumference away from the beam dumping site. So this magnet fails to bend the beam for 17.5 arcminute that it is responsible to bend, to keep the beam safely within the vacuum tube the next magnet must compensate and bend the beam 35 arcminute to make it up. How could the next magnet instantaneously double its magnetic field strength, and instantaneously, if it is already working near its maximum capacity (8T) by design. And very likely, one failed magnet and the beam will be diverted enough to hit the vacuum tube, which is barely an inch or so in diameter, before it even enters the next magnet for corrective action!!! How do you deal with that?

my studies of Accelerator Physics date back to when I was a graduate student – yes we had a 20 hour course on that, and there was an exam at the end! Anyway, that was more than 10 years ago and I admit I am quite rusty on the details.

That said, you ignore the fact that to deflect a beam you can just _decrease_ the field of a bending dipole, rather than turning one on. I have no time to dig the data out for the LHC, and I actually do not think the use of the normal bending dipoles is the way the safety system of that machine work for dumping the beam, but consider that much less than a degree is necessary to deflect the particles: for instance, 10 arcminutes of deflection off a curved path in the LHC ring provide a separation of the beam of more than one inch per 10 meters.

About catastrophic power failures, they are never instantaneous -again, 0.3 milliseconds are a long time. Along 27km of power lines, I am sure there are devices that provide the needed protection against what you envision.

About the beam being lost, that is indeed possible, but it is not as destructive as a global quench of all the LHC magnets… If you really insist I will try to get some more information on how the accelerator is designed to deal with those situations, but again, one quench is not nearly enough to cause the destruction you describe: quenches are normal, to some extent, in a system of that complexity.

If you find the above too vague, I apologize… I tend to try and avoid running for information until my arm gets twisted.

Cheers,
T.

It only takes one magnet falling into an unstable state, and the beam will be ever so slightly deflected from where it should be, and it is just a few milimeter off and would hit the metal part of the next magnet head on, totally destroying the superconductive state of that magnet, and then BOOM all the beam hit the same spot on the metal shell of LHC, like a chain reaction traffic accident on snowy highways. I don’t see how that can be avoided.

“A small deflection from the ordinary path” is easier said than done. Remember those are extremely high energy particle beams. Even with thousands of super strong magnets you barely make one small deflection at a time that the whole beam barely makes a U turn in 25 miles in diameter. That’s the whole point why the accelerator is built in such a huge circular tunnel. Just do some simple calculation, even if you want to deflect just one degree, how long a distance does the beam have to travel within a 8T magnetic field.

You literally have to have a giant magnet and turn the magnetic field from 0T to 8T instantly to make even that small deflection. That is a huge injection of energy to build the magnetic field up. I don’t know ho you can do it within 0.3ms time?

0.3 milliseconds seem like a blink, but it is a long time interval in terms of acceleration operations. Remember the principle behind stochastic cooling of a beam ? A sensor “picks up” the divergence of a bunch of protons on one side of the ring, and sends a signal through the diameter to a magnet which is able to give a small kick to the incoming packet…
And you do not need a magnet to turn on from scratch to get the beam down a dump line, but just a small deflection from the ordinary path.

As for the total energy of the magnets: it is indeed enormous, but they are decoupled systems to a large extent: to think they all release their energy at once is like thinking all the rain falls from a thunderstorm cloud in the same instant…. Both things would cause a lot of damage, but it just does not happen.

Cheers,
T.

The accident exposed structural weakness in dealing with the catastrophic quench events, which WILL occur during normal operation. Quench may be caused by power failure to any part of LHC, or just power unstability, or any of other failure conditions.

Just think about how much energy is stored in the high energy particle beam, in the energy stored in the magnetic field of the super strong supercondutive magnets. All the energy will have to released instantly upon any quench incident. And that is a very destructive energy.

Now think about how much that energy is, it is equivalent to roughly 1000 kilogram of TNT dynamites. Now try to devise a way, to detonate 1000 kilograms of dynamite, within the tunnel of LHC, safely, without causing any damage to any equipment or personel. This is not a design issue. This is not an engineering issue. This is an issue of an obstacle that is virtually impossible to overcome in today’s technology.

Energy equivalent to 1000 kilogram of dynamite. How do you quench that energy safely? You can’t!

Here is the FAQ, scroll dow to the bottom “What Happens If the Beam Becomes Unstable”:

http://public.web.cern.ch/Public/Content/Chapters/AskAnExpert/LHC-en.html

Any magnet that can deflect such high energy beam must necessarily be big and strong and store extremely high energy just in the magnetic field itself. I do not think they have such a deflecting magnet that can be swiftly turned on in just 0.3 ms and instantly deflect the beam. It’s a flat out lie. I do not think it exists.

Oh yes I love it!

PS Tommaso you are a troublemaker, certainly had me fooled ha!

From memory (too lazy right now) the triplets weren’t tested for the quench stresses when mounted. Presumably, less complex mounts doesn’t have that problem.

“… Lattice quadrupoles MQ 392 twin

… Quadrupole in the insertions (3.4 m) MQM 46 twin

Quadrupole in the insertions (4.8 m) MQML 36 twin

Wide aperture quadrupole in the insertions,
twin aperture MQY 24 twin

Quadrupole in the insertions (2.4 m) MQMC 12 twin

Twin aperture warm quadrupole in IR3 and IR7.
Asymmetrical FD or DF MQWA 40 twin

Twin aperture warm quadrupole in IR3 and IR7.
Symmetrical FF or DD MQWB 8 twin

Inner triplet quadrupole, single aperture (Q1, Q3)
MQXA 16 single

Inner triplet quadrupole, single aperture (Q2)
MQXB 16 single

… Skew quadrupole (a2) in MQSXA MQSX 8 single …”.

So, as you say, there are lots of quadrupoles,
but as for triplets, there are only 16 MQXA and 16 MQXB.

Have all the other types of magnets been fully tested,
or is it possible that other unpleasant surprises might happen in the future ?

Tony Smith

Thanks, with the update we have to come clean and note that we *did* see the AF context. But Tony had a large concern.

Tony:

That it was the only “triplets” affected was mentioned in Dorigo’s links, which is why I didn’t motivate it closer. The exact number of quadrupoles involved is confusing OTOH.

Don’t say that I never did anything for the LHC project…

Seriously, this is unbelievably bad news.

the number of structures affected is, to my understanding, 16 (two pairs for each of four points along the beam where low-beta squeeze happens). 16 or 32 makes little difference, though. What makes a difference is whether an easy fix is found, that is if the problem is found to be virtual (say, if the stresses of the test are above what can happen, or if the stress can be reduced in some other easy way than modifying the magnet support design), or if a real overhaul of the pieces is needed.

Removing the structures should not be a big issue. Replacing them is, since no replacement exists that I am aware of with different characteristics.

Cheers,
T.

The Physics Web news article ( NOT dated April 1 ) by Jon Cartwright said, in addition to the excerpt that I quoted in a comment above:
“… The protons will be guided around the ring by some 6000 superconducting magnets of various types. These include 392 “quadrupole” superconducting magnets that are designed to focus the proton beams before they collide at four interaction points around the accelerator. …
scientists at CERN performing preliminary tests on three of these quadrupoles witnessed a serious failure when structures supporting one of the magnets broke at a pressure of 20 atmospheres in response to “asymmetric forces” that were applied during the test. It is essential that the magnets can survive such unwanted pressures, which will occasionally be generated when the LHC is up and running. …”.

Since I don’t know a lot of details of how the LHC is put together, I must rely on information from people like Torbjorn and publications like Physics Web.

So, I personally don’t know for sure whether the number of questionable quadrupoles is 32 or 392, and I hope that some day CERN will issue an official statement in detail.

Whatever is the number, if the quadrupoles are already placed in the collider, would it be difficult (i.e., expensive in time and money) to remove them so that they could be replaced ?

Tony Smith