Hi everybody,
In responding to my query about compressed air systems, a number of
you used the phrase "you touched a nerve," or something similar.
Indeed, it appears that I touched the squid giant axon of NMR
facility manager nerves by probing this subject!
I'm not including the compendium of responses in this email because
it is so large - over 23 pages of 12-pt text. This is so much text I
fear it will appear as a result in all kinds of unrelated archive
searches. (Don't laugh just because this "summary" is so verbose! The
full set of unsummarized responses is even yet still larger.) Unless
there's demand to include it in the AMMRL archive, I'll just post the
full set of responses on the "Reports" page of our unofficial site
(homepage.mac.com/jkurutz):
http://homepage.mac.com/jkurutz/FileSharing18.html
In this summary, I'll describe the general lessons I've learned since
posting my query. then I'll describe my own situation, a little about
how I ended up needing to make this decision, and how we've decided
to proceed.
** THE BASICS:
Your standard compressed air system consists of, in order:
- One or more compressors that suck air from their surroundings and
feed it to...
- A big tank, also called a "ballast," "receiver," or "wet trap."
- A device to remove oil, if the compressor is lubricated
- Optional: A refrigeration-based dryer
- A desiccant-based dual-tower air dryer
- Optional: A nitrogen generator/separator
- Recommended: A second big tank, aka "ballast," "dry receiver," "dry
trap."
- Particle filters, maybe more oil condensers, maybe a color-
indicating desiccant
** AIR COMPRESSORS in this context come in three main designs: piston
(aka "reciprocating"), screw, and scroll (which spins one set of deep
spiral-shaped grooves against another to move gas into the center).
Scroll units rely on a teflon seal and require no oil in the air
line. Piston and screw compressors come in "oil-free," "splash
lubricated," and "pressure-lubricated" flavors.
Here are some key points to consider:
- Oil-free units eliminate the risk of oil contamination in the air
line.
- Oil-free units generally run hotter than their lubricated
counterparts because oil helps distribute heat away from moving parts.
- Oil-free units must be replaced much more frequently than
lubricated ones, probably because they run so hot.
- Scroll compressors are considered the premium units because they
are so quiet and contaminant-free. They are frequently the only
option for facilities that must place the compressor in the same room
as the spectrometer. However, they need to have their tip seals
changed every 3 to 4 months, and the whole device wears out after
just 2 to 5 years (even the salespeople admit to 5 years)!
- Piston compressors are generally reliable, but they're really
noisy. This doesn't seem to be a problem if there's sufficient
ballast air space in the line and the compressors are off in a
distant room. However, I didn't hear from anyone with a 600+ MHz
system or anyone with a cold probe saying they observed no vibration
from their piston system.
- Piston compressors are generally not designed to run a 100% duty
cycle. They need down time to cool off. I couldn't find any good data
to show whether the on/off cycling of these compressors yielded
problems in NMR spectra. The mechanical engineers say there should be
no problems from acoustic noise in the line or compressor switching
because you generate a supply of ~110 psi and regulate down to ~80,
so fluctuations shouldn't be propagated downstream. Several AMMRL
respondents told me that mechanical engineers didn't realize how
sensitive NMR spectrometers are to pressure fluctuations. I think it
all depends on the timescale on which the pressure regulators respond
to pressure changes, and I've found no specs or non-handwaving
arguments on this subject.
- Screw compressors are much quieter than ones with pistons, but not
quite as quiet as scroll units.
- Oil-lubricated screw compressors have a reputation among some
people as being even longer-lived than piston compressors
- There is much on the web that indicates screw compressors are more
"green" than pistons because they consume less energy. This may be
marketing hype, but there is no prevailing counterargument claiming
pistons are more green than screws.
- Heat given off by the units was a big subject, but hard to pin
down. Everyone with scroll compressors said they "run hot" compared
to pistons, probably because of the difference in lubrication
strategies. I got mixed messages when comparing screw compressors to
the other types. The rep selling piston compressors said there was no
spec available on how much heat his units gave off, but said they
were cooler than screws. The rep selling screw compressors said they
were about the same as pistons because you're compressing the same
amount of air in the same time, ergo they should produce about as
much heat; plus, he agreed to give me a spec on how much heat they
should give off (15,000 BTU/HR for a 19 SCFM system prior to air
dryer at 110 psi).
- There are a couple of good websites that discuss various aspects of
different compressor types, but they don't touch on the heat issue:
http://en.wikipedia.org/wiki/Gas_compressor
http://process-equipment.globalspec.com/Industrial-Directory/
piston_compressor
http://www.engineeringtoolbox.com/air-compressor-types-d_441.html
- The NMR facility managers who appeared most satisfied with their
systems (aside from those who use reliable house air or, best of all,
dry N2 boiloff from a big outdoor liquid N2 tank) use dual pressure-
lubricated screw compressors. They're very quiet, though not as
supremely quiet as scroll units. Oil lubrication helps extend their
lives to 20+ years of potential use. They need regular filter
changes, but this is uncomplicated enough that university maintenance
staff can be trained how to keep them going.
** DUAL/MULTIPLE COMPRESSORS
Whatever the compressor type, it is important that the initial
receiver tank be equipped with two or more compressors. It is very
common for systems to be configured with "dual" compressors sitting
atop one receiver tank, but this is normally done to reduce the duty
cycle on each compressor. In our context, we need two compressors so
we'll always have one working in case the other gives out. This must
be explained explicitly to the sales reps because its not "normal"
for them. You can also configure things so each compressor has its
own tank, but the sales reps all said this was unnecessary.
** DRYERS
- Virtually everyone who said something about their air dryers is
using a "dual-tower" desiccant drying system. Here, compressed air
passes over a bed of desiccant to achieve a low dew point while the
other tower "purges," i.e. some compressed air is sent in reverse
direction over the "wet" desiccant to dry it out. After awhile, a
control panel will switch roles of the towers.
- Several people have refrigerated dryers on their systems to take
our most of the moisture before the air gets to the desiccant dryer.
There doesn't seem to be a big advantage to this unless it'll help
reduce the load on the desiccant dryer. However, it's standard
equipment for many compressors, and the technology is quite reliable,
so there's no big incentive to exclude a refrigerated dryer unit.
- Many expressed concern that switching between towers induced a
momentary pressure fluctuation that would affect data collection, so
they put a big (30 to 200 gal) tank after the dryer to dampen them.
This sounds like a very sensible idea.
- Some units have control panels that monitor the dew point of the
air produced to determine when to switch between towers. This is
supposed to reduce its power need or something like that.
** NITROGEN GENERATORS
We won't be getting a nitrogen generator with this system because we
don't need to go that low in temp, but I'll include a discussion on
them for the record. Many of you with Bruker systems described having
a nitrogen "separator" that enriches N2 to about 98% of the
compressed "air." This seems to work well for just about everybody,
and few have reported problems with their units. Some said they had
molecular seive-based generators, but these seem geared more for
systems demanding less gas flow than most spectrometers
** SOLIDS NMR NEEDS
The Varian Inova Installation Guide says that the dew point for a CP/
MAS system needs to be an incredibly miniscule 80K (-193 deg C, -315 °
F)! Yet many of you responded that you're doing lots of solids NMR
with dew points of approximately -80 deg. C by using a nitrogen
generator/membrane accessory. I'd like to hear more from people with
solids setups and what your dew points really are on operation
spectrometers.
** SEPARATION OF SUPPLIES FOR DIFFERENT LOADS
A few labs reported that they split their loads:
A) High-flow/low-pressure/very dry needs such as VT are hooked up to
more costly N2
B) Low-flow/high-pressure loads such as antivibration legs are hooked
up to compressed air.
This seems to make the most sense if you're running solids, which
seems to require a very high flow rate of enriched N2 gas.
**!! SPECIAL NOTE FOR COLD PROBES
- I received one or two mentions from people hooking up their cold
probe pneumatics to their high-pressure supply. This sounds OK, but
it still leaves your cold probe vulnerable to failure or shutdown of
the air system. If you're configuring a new cold probe to run with
the common air system, I RECOMMEND YOU DON'T DO IT! We had big
problems with our cold probe when our house air started occasionally
dipping down to 55 psi.
!!- To get around the problem, we hooked up our cold probe pneumatics
to a cylinder of dry compressed air last year, and we've had ZERO
problems despite having to shut down our compressed air system a
number of times for various reasons. We've been configured like this
for over half a year, and our cylinder is still over 2/3 full.
** SIZING THE SYSTEM
- Getting a properly-sized system is very important, and it seems to
be trickier to figure out than one might think. The spectrometer
manufacturers provide specifications for how much air is required for
one spectrometer in a typical setup. In the U.S., the standard unit
is SCFM, standard cubic feet per minute. What these specs don't tell
you about is the extra load imposed by the air dryer and, if you have
one, the nitrogen separator/generator. These can be huge! One
facility reported their 10 SCFM compressor system lost 1-2 SCFM to
the dryer, and another 5-6 to their N2 separator, and they ended up
with insufficient air for their spectrometer, and they had to upsize
their compressors!
- One other tricky thing is that the specs you have for air use, in
SCFM, may be given for different pressures. For instance, the Varian
Inova Installation Guide says normal operations will demand 1.6 SCFM
at 45 psi during sample eject. But if you have a CP/MAS unit, it will
require 2.8 SCFM at 90 psi. The Kaeser compressor product literature
has a table o' multipliers you can apply to your SCFM requirements at
different pressures to gauge what the normalized flow should be at
100 psi.
** BRAND CONSIDERATIONS
- Because this message is intended for peer discussion, not
publication, per se, I'll frankly describe the following
recommendations I garnered unscientifically from my review:
- The happiest managers seem to have Kaeser-brand screw compressors.
The parts are a bit more expensive in the U.S. because of the German
origin, but they're apparently worth it.
- One person said they've seen or worked with both Kaeser (pronounced
"Kay-zer") and Ingersoll-Rand screw compressors and find that the I-
R's aren't made quite as well as the Kaeser's.
- Some people found I-R compressors to be fine for their purposes.
- One said the Ing-Rand control panel was very touchy. It's very
fancy and does great things when it's operational, but it's prone to
total failure if exposed to power fluctuations. I haven't read the
document, but this person reported the I-R warranty requires the
control panel be on a power conditioner, else the warranty is void.
- One person also said their I-R compressor had a problem allowing
oil vapor into their lines, even with filters in the line.
- Quincy was a compressor brand that came up when talking with our
vendor. Though no one owning this brand responded to my AMMRL query,
I found two instrumentation facilities happily using Quincy piston
compressors. In each case, however, their compressors were on
different floors than their magnet rooms. This looks like a great
brand for this type of compressor.
- Atlas-Copco got mixed reviews for their oil-free scroll
compressors, but this may be due to the intrinsically low life of
this type of compression mechanism.
- One person reported they had Sullair compressors and had problems
with them.
- Two people reported having Jun-Air systems, and they like them OK.
- Vast numbers of respondents love their Balston or Parker/Baltson
air dryers. This appears to be the best default choice in dryers.
Kaeser systems come with kaeser dryers, and they seem to work fine, too.
- Most everyone loves their Hankinson air dryers. I've had some
trouble w/ ours, but it's 12+ years old, and it may have been fouled
by our bad house air. I wouldn't mind having another one.
- The name Wilkerson came up a couple of times with respect to dryers
and filters, and it seemed their products were OK.
- People seemed happy with Parker N2 generators
** OUR SYTEM & FUTURE PLANNING
We have three Varian Inovas (2x600, 500), one 600 with a cold probe
and carousel sample changer. One 600 needs 1.5 LPM air to cool its
shims. All are equipped with pneumatic antivibration legs and FTS VT
systems that heat and refrigerate the VT air. Our cold probe's
pneumatics are on a cylinder of high-grade compressed air. We do
biomolecular work, but a fair amount of it is in the -5 to 25 degree
C range, so we need to use the FTS chillers frequently. We had been
in the habit of leaving the FTS units on all the time to achieve
stable 25 °C sample temperatures, but we're reluctant to do that now.
The spec dew point for these chillers is -60F. There is faculty
interest in adding solids capability to the 500, which would
dramatically increase our compressed air needs. Getting an extra-
capacity system now won't cost much more than one that just meets our
current needs, so we're trying to pretend that we've got a solids
system.
Adding up all of our needs yields a total of between 7 and 8 SCFM
with the solids system and two samples being ejected. We're shooting
for a compressor system capable of 18-21 SCFM that can deliver ~15
SCFM with a dryer, but no N2 generator. We'd like to hit that dew
point of -100F, but won't complain if we get -70.
** OUR SOLUTION
We've decided to go with a Kaeser system of oil-lubricated dual-screw
compressors (SX-6, SCB-5) feeding a 60 gallon wet receiver, which
then feeds a Kaeser desiccant dryer with a 60 gallon dry receiver.
The pair of 5 HP compressors should deliver 21 CFM _at_ 110 psi, but
that should get cut to ~13 CFM of -100 deg F dew point air after the
Kaeser dryer (KADW-20). If you detect flaws with this setup,
especially if you have experience comparing the heat output of screw
and piston compressors, please let me know now before we take delivery.
** COST
Price is always a touchy subject, but I thought I'd give you the
ballpark figures (+/- 10-20%) so you know what order of magnitude
we're dealing with. Of course, your system's costs would depend on
your sizing, dew point needs, and your vendor's whim, so these
figures cannot be used for the precise computations on which sales
people depend. Each of these compressors is approximately $5K, the
dryer is approximately $1.5K, and with all the tanks, filters,
gauges, etc., the whole thing will run approximately $15K without
installation. We received a quote on a Quincy system that also looked
very good and was approximately $1500-$2K cheaper, but we ended up
being turned off by the noise and potential vibration issues with the
pistons.
** OUR SAD STORY
Last year, our cold probe started having intermittent "low cooling
power" faults. It turned out - after spending long periods of time
watching pressure gauges - the pressures of the compressed air, which
feeds the cryogenic pneumatic valves, was fluctuating. Sometimes it
dropped much lower than the regulated pressure (down to ~55 psi),
shutting some of these valves for a few seconds, thus leading to
faults observed on the cryobay window. After more gauge-staring, we
determined the difficulty lay in our aged house air compressor. We
got around the problem by hooking up the cryo system to a gas
cylinder, whose gas we filter and regulate down to 80 psi. This
arrangement has been great.
Early this year, I walked in to the spectrometer room and found all
three spectrometers' temperatures were high but unregulated and
fluctuating a little. Near the magnet with the cold probe, I smelled
charred plastic. Turned out that all three FTS chillers had frozen
shut, depriving VT air to all the spectrometers. Those of you
familiar with Inovas will know that they're configured in a way that
VT air loss leads to catastrophic overloading of the heater in the
probe. On a cold probe, this melts the expen$ive plastic VT/tuning
knob assembly.
So our house air was failing us both by giving inconsistent pressure,
but also by sometimes being so wet that water would condense and ice
up in the VT air line. After talking with the administration, we were
given approval to get our own air system. That's when we started
shopping and I began talking to so many of you.
**!! THANKS !!**
Thanks so much to all of you who responded, including Andrew Fowler,
Andy Staley, Andy King, Andy Soper, (alphabetical order, you see),
Carlos Amezcua, Craig Butts, Dave Gindelberger, David Vander Velde,
Deane McIntyre, Elwood Brooks, Eric Paulson, Dr. Sukenick, Herve
Bizot, Jeff Ellena, Jerry Hirshinger, John Witte, J. Kotesh Kumar,
Ken Fishbein, Ken Osbore, Klaas Hellinga, Mark Edgar, Martha Morton,
Rich Shoemaker, Robert Honeychuck, Robin Kinnel, Sara Kunz, Tom
Stringfellow, Walt Niemczura, Bill Kearney, Wing-kong Kwan, and
Xavier Lemercinier.
It's been most educational. I hope this compiled result helps those
of you making similar decisions now, and those who are reading this
as an archive entry in the future.
- Josh
Josh Kurutz, Ph.D.
Technical Director, Biomolecular NMR Facility
University of Chicago
Gordon Center for Integrative Science, room W123C
929 E. 57th St.
Chicago, IL 60637
Office: (773) 834-9805
Spectrometer Room: (773) 702-4052
Cell: (773) 315-5732
Fax: (208) 978-2599
nmr.bsd.uchicago.edu
homepage.mac.com/jkurutz
Received on Thu Nov 02 2006 - 13:56:06 MST