jump to navigation

Some insight on the LHC triplets April 6, 2007

Posted by dorigo in news, physics, science.

In an attempt to find out more about the recent problem experienced with a triplet of low-beta quadrupoles at the Large Hadron Collider, I dug a bit into the technical publications related to the design and test of those structures.

There is not a lot online on the issue. Many of the papers are not accessible freely… But I found a paper that describes the choice of material for the support structures (the so-called G11, a glass cloth-epoxy laminate which was the part which broke in the triplet at point P5 in the LHC ring. The paper also discusses several of the tests performed to study the mechanical stresses under quench conditions. Here are the conclusions of the paper:

The HGQ [short fully instrumented model magnets – TD] mechanical design has been refined and improved through a combination of analytical, computational, and empirical studies. Quench performance has improved substantially, and is well reproduced among magnet assembiles. The essential design details leading to achievement of the performance goals are being finalized. Two additional model magnets will be fabricated in order to refine production techniques and test minor modifications to cable desing, in preparation for full-scale prototype fabrication.

Not very illuminating. Let’s see another paper titled “Quench performance of Fermilab high-gradient quadrupole short models for the LHC interaction regions“. It is a 1999 publication, so it is an earlier one than the one quoted above. In it, one reads that the supports were still designed to be made of G-10. It said:

Model HGQ05 included a set of changes which addressed the issues raised by previous models. The most important of the changes from the base-line design included n HGQ05 are:

  • Use of G10 as end part material
  • Re-cure of inner coil at higher pressure, resulting in a higher inner layer elasticity modulus and more uniform inner/outer coil mechanical properties
  • A continuous body/end transition, including elimination of key extension

In the conclusions, one learns that

…Significant improvement in magnet training at 4.5K and 1.9K was achieved in last model (HGQ05) as a result of the optimization of design details of magnet support structure. Test of HGQ05 in a second termal cycle continues. Although the results obtained demonstrate that the magnet desing can provide the required quench performance some further optimization is desirable…

Interesting. It looks like these support structures were understood as critical points early on, as demonstrated by the attention they received and the changes – from an earlier design to G-10, and then G-11. And one also infers that the quality of the support is instrumental in preventing the quenches.

Is any expert willing to elaborate on these issues ? My comments column is always open. 


1. non-expert - April 6, 2007


what is a “triplet of low-beta quadrupoles”? A sequence of quadrupoles is used to focus beams, triplet maybe means that you have 3 quadrupoles, and low-beta?

And why the maximum intensity of magnetic fields possible today is not much better than what possible 50 years ago? Simply because too strong magnetic fields generated by too strong currents break materials?

2. Christian - April 6, 2007

Ciao “scenziato” ! Stavo quasi pensando di commentare questo tuo post – che come puoi intuire saprei senza problemi di sorta argomentare ;-P ma per questa volta ti risparmio il mio discorso sulla fisica nucleare ! Ho scritto tanti di quei saggi che potrei riempirne una libreria intera. Strano tu non ne abbia qualcuno.
Stop bullshit , come dicono i tuoi lettori anglofoni. Passa una buona Pasqua ! That’s all !

3. Quantoken - April 6, 2007


When the magnetic field is too strong it breaks superconductivity of the material. So 8T magnetic field is about the state of the art of current technology. The whole reason why LHC needs to have a circumference of 27 km, because you need a long path to bend the orbit of the high energy particle beam!

Whether LHC can be operated safely is a very critical question to ask at this point. If it can not be operated safely then it is a piece of junk. Quench of magnets and loss of beam will occur during normal operation and the safety of both must be guaranteed.

I believe the test failure is just tip of an iceberg. From an engineering point of view you normally have at least a 5 times or 10 times safety factor. That means if the structure fails at 20 ATM pressure, you increase the strength just enough so it doesn’t break at 20 ATM. And then, on top of that you further increase the strength by 5 times to make it safe. That means the new design needs to withstand at least 100 ATM without breaking. That means that piece have to be re-designed to about 10 TIMES bigger, thicker, stronger. The increased size means the whole magnet will need to be re-designed to allow room for the extra size of the supporting structure. And then the cryogenic system needs a new design because it now needs to cool down several times more mass. The list goes on and on. It’s going to be nightware.

Remember we are not talking about a few percentage of change. We need a safety factor of multiply by 5 or 10!!!

I am also very skeptical whether they can actually safely quench a magnet from normal working condition. Remember we are talking
about a magnet with 11700 amperes of current flowing. The coil suddenly becomes conductive and a tremendous amount of heat will be generated when that huge current flows through that normal resistive coil. It’s like a transformer, you simply can NOT instantly cut off the current. The current will continue and the heat generated equals to resistance times current squared times time. Sure the whole thing is surrounded by 27 tons of cold mass. But it doesn’t help you at all, heat does not dissipate well in liquid helium. You can pull liquid helium directly onto your finger and you won’t feel cold at all, because the layer of helium next to your finger evaporate into gas, protecting you from making direct contact with the super cold liquid. Likewise the helium next to the coil will be evaporated, preventing heat to further dissipate. So all the tremendous heat will melt the superconductive coil completely, destroy the whoe magnet.

Even after the coil is melted broken, the huge current will continue. The inductive electric potential induced by the magnetic field will be so high that an electric arc will be generated continuing the current. The electric arc will destroy anything in the path, melting things down. The whole process continues until the full 9 MJ of energy stored in the magnetic field is totally dissipated in the form of concentrated heat. I don’t see how that can be avoided. The experts probably assumed that 27 tons of cold mass is enough protection. But the problem is heat simply won’t dissipate that fast into the 27 ton cold mass.

4. Tony Smith - April 7, 2007

According to a Nature online article dated 3 April 2007 by Jim Giles:

“… Fermilab staff say that the responsibility for the failure lies with them. Judy Jackson, the lab’s director of communications, says that one of the structures that holds the magnet in place – made from a woven glass cloth named G11 – was not designed to the specifications required by the LHC.
Jackson adds that the documentation for the parts was reviewed four times without this error being spotted. …”.

Does anyone blogging here know the exact “specifications required by the LHC” ?

How many levels of bureacracy were between the people who wrote the LHC specifications
the hardware lab people who actually built the thing ?

For example, were there layers at LHC, layers at Fermilab, layers at contractors, and various liason layers in between ?

Did the people who actually built the thing (I am NOT talking about their bosses) ever see the LHC specifications ?

Tony Smith

5. dorigo - April 7, 2007

Hey Christian,

well… Thank you and happy easter to you too. This forum is open for you to dump some of your knowledge here, anyway.


6. Bee - April 11, 2007

Hi Tom,

I was wondering if anybody currently has any plan whether the schedule for the first run is completely messed up or one can still expect it this year?



Sorry comments are closed for this entry

%d bloggers like this: