Recently Reinhard Schumacher "made some observations and questions"
about the CLAS gas distribution system which he shared with many of
us. Perhaps you read the message and were alarmed at the use of
"bad materials in the gas lines", the "chaotic routing of lines",
the "lack of control of gas in the hall", the "lack of consensus or
even planning" about how to monitor the flow of gas to individual
chambers, the apparent lack of "some fail-safe mechanism to prevent
over or underpressuring" of the chambers, the placement of pressure
transducers on the "`wrong' side" of bubblers or flow tubes.
Even we were alarmed. Sure, we spent years carefully designing the system,
and sure, the system had been fine-tuned by outside reviewers from
Cornell and Fermilab, but when a CMU professor with experience with
BNL safety people (kind of like getting a `purple heart') starts worrying
aloud you have to get a little nervous.
So, we rolled up our sleeves and took a long, hard look at these astounding
`facts'. Fortunately folks, not ONE of these claims is totally true. The
one valid suggestion - to have a way of REMOTELY monitoring the relative flow
to each of the six sectors individually - was discussed last week between Mac
and Reinhard and a consensus was reached about how to proceed (at least Mac
thought so).
So, here's our detailed response:
Reinhard says "1) Bad materials in the gas lines. On both Region Two and
Region Three I saw liberal use of Tigon tubing. This is one of the
worst materials for drift chambers because it is known to outgas and
poison a chamber. It has been shown that **even after removal** of
Tigon from the system, a chamber's performance will continue to
deteriorate {Blum and Rolandi "Particle Detection with Drift Chambers",
page 341]. In our case the Tigon might be on the exhaust lines, but
since we are recirculating the gas I think we should play it safe and
remove all of that stuff. I don't trust that the filtering we do will
remove all the poisons. Why risk the chambers with gas piping that is
known to be dangerous?"
We say: "The only Tigon used in the system is for the 1 inch exhaust lines
on the Region 3 chambers. This has been planned from the beginning to be
temporary only. Before recirculation starts, they will be replaced by
1 inch Nylon lines."
Reinhard says " 2) Chaotic routing of lines. This is a small point, since
perhaps all the junk hanging around now may be temporary. From
experience at BNL I predict that we will NOT get clearance to run
flammable gas in the hall until all the plastic gas lines are somehow
protected in trays or within conduits. At BNL our group argued that the
gas-sniffing system would warn us of breaks in the gas lines, but this
did not satisfy the safety mavens. We ended up routing the Poly-Flo
lines through 2.5" PVC plumbing pipe. This actually worked well, and
greatly reduced the danger of snagged or crushed lines."
We say: "By February all lines in the system will be polished stainless
steel except between chambers and manifold where we use Nylon in order
to electrically isolate the chambers. We are not using flammable gas in
the December run."
Reinhard says " 3) Lack of control of gas in the hall. We have 18 big drift
chambers in CLAS, but only three gas inlet/outlets in the hall. Each
Region has x6 manifolds which fan the gas to the six sectors. There
seems to be no consensus, or even any planning, for how to control or
monitor the gas flow to the six sectors separately. It would be wishful
thinking to believe that everything will go right if the just hook them
all up and start flowing gas. It seems to me that we need to
a) meter the gas flow to each sector,
b) meter the gas flow back from each sector, or in some equivalent way
c) have a way of telling when (not IF , but WHEN) one sector
develops a serious leak. Also
d) some fail-safe mechanism to prevent over or underpressuring
of the sectors. This mechanism should operate locally, at the chambers."
We say "First, there are only 12 BIG drift chambers in CLAS. Could size
envy be the actual cause of this outburst? Nah. There certainly was
planning on monitoring the gas flow to the chambers. Any indication of
leaks was to be followed by a manual search for the offending chamber
by selectively turning on and off flow to individual chambers. A good
suggestion which came out of a conversation with Reinhard is the addition
of low-impedance flow indicators on the output lines from individual
chambers to the exhaust manifolds which could be viewed by a remote camera.
We both liked the idea. I (Mac) thought that indicated consensus.
Maybe consensus has a different meaning in Pittsburgh."
Reinhard says "What complicates the issue (for me) is that we are reciculating
the gas, and we are doing some sort of differential pressure measurement to
keep the chambers from blowing up or collapsing like balloons. Yet there are
no pressure transducers in the chambers, so they must be in the gas
shed, on the 'wrong' side of any bubblers or flow tubes we install."
We say "We are doing "some sort" of differential pressure measurement.
The transducers are not on the chambers (very observant!) but neither are
they in the gas shed. They are on the gas panel on the second level of
the space platform. They follow about 16 feet of exhaust line from the
manifold to the panel which was sized large enough such that the pressure
drop is negligible even at purge flow rates (5 vol. exhanges per day).
This is the reason that impedances in the exhaust line must be minimized.
Any flow indicators on individual chambers must be low impedance (not
bubblers)."
Reinhard says "As a straw-man proposal I would say we need the following:
i) a wide-open rotameter flow tube on the input of each sector
so we know that each sector is getting its fair share of gas
ii) a mineral-oil bubbler on the output of each sector
to show that we don't have leaks
These two set of six devices can be mounted on separate panels which
will be mounted near the respective manifolds on Level 1. The pressures
measured in the shed will be 'off' by amounts determined by the drop
across these devices, but who cares?"
We say "We care. We do not want a pressure control point which depends
on flow rate. We need to control the pressure in the chamber within
very narrow limits or else we might well develop gas leaks."
Reinhard says "iii) on the input line there should be an overpressure relief
bubbler which vents to the hall to protect against the gas mixing system going
nuts and sending a high pressure blast of gas. Don't scoff, it can
happen."
We say "We already have an overpressure relief bubbler. We also have an
underpressure relief bubbler. Don't scoff, underpressure can happen also."
Reinhard says "iv) on the output line there should be a bubbler which vents to
the hall which protects against back pressure in the exhaust
(recirculation) line and in case a hurricane hits and drops the
hall pressure very fast (faster than the mixing system can respond).
Both of these devices need to be inserted in the in/out lines before the
manifolds, and therefore require some re-plumbing of what has already
been done. I would be satisfied of I were shown that these bubblers
already exist in the gas shed."
We say "to continue our previous point about over and under-pressure protection.
We have put ALL of the protection bubblers on the exhaust line ONLY because
it is connected by low impedance lines to the chambers themselves. Because
of the negligible flow-dependent pressure drop, any pressure change in the
chambers will be felt at the protection bubblers after a delay of 16 feet
divided by the speed of sound in the gas. We don't actually know what the
speed of sound in our gas is. Perhaps some university group could research
this for us."
Reinhard says" These are my concerns. I think we need to have some discussion
about these things, and then we probably need to take some action. For
myself, I especially want to understand the thinking that went into
designing the gas recirculation system. Are my suggestions above
reasonable and/or feasible? Any other comments?"
We say "We agree. About everything. Especially his desire to understand our
thinking. No more comments"
best regards,
Mac and Steve