[labnetwork] Toxic Gas Monitoring Systems

[John Shott]

Jerry et al:

As we have a newly commissioned toxic gas monitoring system here at the 
Stanford Nanofabrication Facility, let me tell you what we ended up with 
to complement all of the other good input you are receiving.

In February of this year, we replaced an aging system that had about 80 
monitored points.  Our new system monitors nearly 150 points with gas 
detectors.  Ny guess is that about 125 of those points are monitoring 
our main shared facility and the remaining 25 are monitoring points in 
private labs that share our building.  In our case, we went with "draw 
type" sensors.  We have 16 channels of paper tape monitoring of hydrides 
using a DOD Technologies CL-96 unit.  The remainder of our sensors are 
DOD Technologies PS-7 electrochemical sensors that are also draw type 
sensors.  Note: we would have likely only used the PS-7 electrochemical 
sensors were it not for the fact that we have an epitaxial reactor that 
uses arsine.  To my knowledge, none of the electochemical sensors can 
reliably detect the 50 ppb PEL level required for arsine gas.
We have also added digital input fire eyes in all of our silane 
cylinders and in our silane VMBs.  Note:  all of our sensors put out 
old-fashioned 4-20 mA signals that are monitored by analog input modules 
in the PLC.  I know that there are a number of detectors that have web 
interfaces, etc., but we chose not to go that route.

We do like draw-type sensors:

They don't tend to dry out if they are sensing a point in either a 
high-velocity or high-temperature area.

You can look at a non-zero reading without being nearly so close to an 
actual leak.

Being able to cluster a bunch of sensor bodies in one area makes annual 
live-gas testing easier and allows you to get a better idea if more than 
one sensor is "smelling" something.

Our system is PLC based with inputs going to a GE 3Xi PLC.  HMI in our 
case is provided through a WonderWare interface.  Our TGO computer is in 
the FACP control room which is, we believe, out of harms way in the 
event of a gas leak.  Our fire department always goes there first and is 
sufficiently well trained that they can get a pretty good idea of what 
is going on even if none of us is there yet.  We also have VPN remote 
access to the HMI as well as synthesized voice phone dialer and email 
messaging.  The voice and email options provide a listing of the sensor 
involved and whether it is reporting a fault, warning level, or alarm.  
All of our sensors are set up to provide a warning signal at about 50% 
of the alarm level.  The warning level does not evacuate the building or 
call the fire department.

In general, even though legal requirements only require alarms in 
exhausted spaces at 1/2 IDLH, we have all of our alarms set at PEL for 
several reasons:

1. Dilution in most gas cabinets and exhausted enclosures is high.  In 
many cases it would be quite a large leak to reach 1/2 IDLH in those spaces.

2. If our detectors are set at PEL in exhausted enclosures, we can also 
use that sensor to monitor the nearby breathing air as any gas outside 
the enclosure will soon get drawn in to the enclosed space.

3. We'd rather find and deal with a problem as soon as possible.  Little 
leaks seem to have an annoying habit of becoming larger leaks if left 
unattended.

Note: in the same vein, we have also installed 0-1000 ppm hydrogen 
sensors rather than 0-100% LEL sensors.

I recommend early discussions with your Authority Having Jurisdiction.  
Their interpretation and preferences can have a large bearing on what 
you may end up with.  For example, in our case, Santa Clara county would 
categorically be opposed to any approach that was not fully designed and 
stamped by a Registered PE including comparatively minor 
changes/upgrades to an existing system.  While I know that there are a 
number of examples of very nice in-house designed and built systems ... 
and, at some level, I envy those that have full control ... that is 
something that simply wouldn't fly with our AHJ.

Good luck, there are a lot of choices and options.  You are welcome to 
contact me should you care for any additional detail.

Thanks,

John