[IP] Re: Bad electrical connection cited as cause of LHC shutdown.

[David Farber <dave@farber.net>]

Begin forwarded message:

From: Vern Paxson <vern@icir.org>
Date: October 30, 2008 1:12:57 AM EDT
To: dave@farber.net
Subject: Re: [IP] Re: Bad electrical connection cited as cause of LHC  
shutdown.

In case this now weeks-old thread is worth further commentary: "bad
electrical connection" is a layman's explanation of what accelerator
physicists refer to as a "quench".  This is in fact a routine event when
working with superconducting magnets: somewhere something in the winding
is imperfect, such that, when under the influence of extremely high  
magnetic
fields, there is physical motion in the magnet's wire, leading to  
friction.
The friction generates local heat which transitions a small amount of  
wire
from superconducting to "normal".  The magnets carry such immense  
currents
that the small resistance of the normal wire turns into a large amount  
of
local heat; this in turn *transitions more wires* from superconducting  
to
normal; and the process spreads.

It all happens fast, with the risk of explosion.

To counter this threat, superconducting magnets have high-speed  
circuitry
designed to detect the onset of quench events.  When detected, they very
rapidly (and counter-intuitively) heat up the entire magnet, causing it
*all* to go normal.  The goal of doing so is that while individual wires
can't possibly bear (when normal) the heat of the amperage going through
them, the entire magnet has enough bulk volume to do so.

When a superconducting magnet is first commissioned, it undergoes  
"training",
in which it is repeatedly (1) powered up to the point of quenching, (2)
the quench-protection circuitry kicks in, (3) the whole thing goes  
normal
and thus doesn't explode ... and (4) the imperfect winding that moved  
due
to the high magnetic field is now in a new position where it likely  
won't
move again (since the original motion was driving it towards such a
potential-energy minima).  Therefore, that particular quench won't  
recur.
However, others might.  So during training, you do this again and again
until no more quenches occur.

So, Yes, such failures are expected, but in this case something went  
wrong
with the quench-protection mechanism leading to the explosion.  A bit  
more
than just "a poor solder connection" (though clearly still not good, and
I believe magnets are generally trained *before* installing them in the
beam tunnel).

		Vern


(Explanation/disclaimer: this is all from memory rooted in my former  
life
doing software for the Superconducting Super Collider back in the  
1980s.)




-------------------------------------------