Experts contacted:
- Jim Virdee (JV), CMS spokesman;
- Pierre Lazeyras (PL), formerly ALEPH technical coordinator, and long
experience with installing large experiments (UA2, BEBC);
- Alain Herve (AH), CMS technical coordinator. AH very kindly gave me
an extended, private tour of CMS, both on the surface and in the
cavern, followed by an long discussion.
While I did not talk to anybody in Atlas, I am in the cavern on a
regular basis, and also hear a lot of the pains and moans of my Atlas
muon colleagues involved in installation & commissioning.
The opinions relayed below will be quoted mostly without attribution,
simply for the sake of brevity - and also because the emerging picture
is highly consistent. Mistakes & misunderstandings (there must be some)
are my ownŠ
Witold
WK 18 Sep 06
Comments on surface vs. underground detector assembly
=====================================================
based on ATLAS & CMS experience
================================
1. Surface assembly is highly preferable to underground assembly, for
many reasons.
1.1 Cost: while you do need a larger surface building and a rented
gantry, it is felt that these items are largely offset by
> smaller access pit
> smaller underground cavern
> smaller, cheaper cranes.
Besides the fixed gantry (~ 1.5 MCHF incl. rental), the CMS
surface building has 1 (or 2?) 60 T cranes. The cavern has
one 20 T crane.
For cavern- or 'mixed' ass'y, the cost of two (1 surface, 1
underground) multi-100 ton cranes is comparable (per crane) to
the total cost of the one 2000 T gantry.
> more efficient, hence faster, assembly & commissioning
(salaries!). More on this below.
1.2 Simpler, more flexible, more efficient, and safer assembly
upstairs. Space in the surface building allows "easy"
manipulation of big objects, running several operations in // in
different parts of the building, decoupling conflicting schedule
requirements, easing access to components to be debugged, etc.
AH had been in charge of the L3 installation, where the magnet
had to be assembled in situ. It is that painful experience that
convinced him to push for CMS to be assembled on the surface.
I asked AH whether in hindsight, he saw ANY advantage to
underground assembly - financial or technical. He answered that a
few years ago, i.e. before seeing it play out, he wasn't yet
totally sure (characteristic caution of a highly competent
engineer). But given his experience since then, he stated that he
would not change a thing, and that he is strongly convinced this
has only advantages. As these words may sound categorical,
let me add that I have known AH for a long time. He is an
extremely solid engineer, and at the same time quite cautious.
He chooses his words carefully, and to me what he said above
carries a lot of weight.
PL pointed out to me that ALEPH followed a similar philosophy,
working on the surface whenever possible - even though
the ALEPH cavern was luxuriously large. For instance,
ALEPH assembled & fully tested their electronic hut on the
surface. When ready, connections were broken at patch panels, and
the entire electronic house lowered into the cavern as a single
unit. CMS has a similar system: each big piece (2000 tons max)
carries its patch panels. When surface testing is completed,
the disconnection is fast and painless. When the wheels arrive
underground, the final cable plant is already in place and
tested, allowing for fast reconnection and commissioning.
To me, with my Atlas experience, the simplicity & flexibility of
the surface ass'y, as I witnessed it at CMS, is perhaps the most
obvious advantage: there is no comparison whatsoever between
> the overcrowded Atlas cavern, the hours spent waiting for the
crane, the difficulties of accessing components buried deep
inside the toroid,.. and
> the 'relaxed' operation in the CMS surface building.
This is of course not because Atlas is less smart: the Atlas
detector concept simply makes it impossible to
assemble the biggest piece (toroid) on the surface, and that
forces underground ass'y. Cavern size restrictions (costs!) and
the coupling of many activities through shared-crane usage,
cramped underground floor space and safety concerns with hanging
loads, do the rest.
1.3 Decoupling the detector ass'y schedule from that of the cavern -
thereby minimizing the risk of serialized operations
(more on this in item 2. below).
2. If surface assembly saves so much time, why is it that CMS is not
far ahead of Atlas in terms of readiness?
Basically 3 reasons:
2.1 It was known that Atlas had no choice but to assemble
underground, so they needed their cavern first. Civil
engineering resources were limited - and went to Atlas first.
2.2 The geology at the CMS pit is much more hostile than at Atlas.
The tunnel and cavern digging was stuck for many months (or was
it > 1 year ?) because of surging underground water and other
major civil engineering challenges. This delayed beneficial
occupancy for a long time, and highlights the importance of
decoupling the CF schedule from that of the detector assembly.
2.3 Many other factors, that are technology- and detector specific,
can delay a project as technical difficulties arise.
3. Obvious perhaps, but absolutely essential: the detector concept
has to be thought through, taking into account in a very detailed
fashion the assembly & installation scenario.
I knew this "in my mind", but seeing it play out in CMS under AH's
guidance was educational - and sobering. Many aspects of the
detector modularity, assembly & testing sequence, in-situ maintenance
etc were driven by the deliberate choice of assembling/testing on
the surface, disassembling it and then putting it together again
underground. From what I could see (too long to write here), the
choices CMS adopted also made their underground access and
maintenance much easier than those of Atlas. Some of it clearly has
to do with the basic detector concept, but I do think that part of it
can be traced back to the modularity imposed by the assembly &
installation sequence.