Dear AMMRL's:
Thanks to the many (over 30) respondents to the email I
sent two weeks ago asking about room temperature control.
This clearly is a subject that many of you have thought
about and worked hard on. Much useful information was
contained in these emails, so I have listed below a copy
of all the initial responses I obtained. This email is
therefore quite long. I will briefly summarize the
responses prior to the pastes.
First, the main result is not a surprise: +/- 1C can be
done, as 22 of the respondents can verify in their labs.
One of the best examples is right here on our campus:
NMRFAM controls room temp in their labs to +/- 0.25C.
When I've used this information in the past, however, my
physical plant has been quick to mention the unique design
of that facility. There is no doubt that how easy/expensive
+/- 1C level of control would be depends on the specifics
of each individual lab. New buildings can/should have this
control for long-term, high-field instruments. Older
buildings, even under renovation, can present various
problems that make solutions more difficult/expensive. But
from the 22 respondents to my initial inquiry, all types of
buildings/environments seem to be included. The conclusion
is then that any lab can be corrected to give +/- 1C
control.
It is clear that heat-pumps--cooling in combination with
heating--are crucial to obtaining this level of control.
Various vendors were mentioned, starting with Johnson Controls
(used at NMRFAM), Liebert most often, Carrier, Landis & Gyr
Powers, and TRANE. Well designed air flow is another crucial
factor in obtaining the best environment for a spectrometer.
Some responses implied that +/- 1C over a limited period of
time (e.g., 1 h) would be sufficient. I have to disagree
with this assertion: the control is most important for
minimizing t1-noise during long (>12h) experiments. If the
temperature in the room changes slowly by 1C/h over 10 h,
and always in the same direction, t1-noise will be
significantly increased. The control should be maintained
over the time of the complete experiment, suggesting over
48 h or more.
I have stripped off the names and institutional information
from the emails below (except where specifically told it was
ok to post), as requested by some of the respondents. I can
forward more specific information to any of you about an
individual response, however.
This summary is a huge help for me at UW-Madison Chemistry in
our current efforts to get proper room temp control installed
during our remodeling project. So let me thank again all the
respondents that took the time to answer my email. You all
make AMMRL the hugely helpful group that it is!!
Regards,
Charlie
*****************initial email*************************
>Having a problem that I hope this group can help with:
>
>We are undergoing remodeling, and I have requested new
>air-conditioning with specified plus-or-minus 1 degree
>Celsius temperature control. I keep getting back from
>our physical plant designers something like:
>
> Client would like +/- 1 degree C. This is not
> feasible.
>
>I would appreciate hearing from those of you that have
>managed to make this temp control--clearly a critical
>aspect of any high-field facility--feasible, to send
>back as a response to my physical plant.
>
>Thanks,
>Charlie
1***************responses**************
i don't know all the terminologies for HVAC's, but
if you have a normal compressor and a condenser that
works like any old air conditioner on a window, then
it's true, you can't get that kind of fine control
in the room.
if you have a chilled water supply and heater and
the inlet air temperature itself is buffered, then
you can get pretty close to +/- 1C i think. this
type of setup is a lot more expensive, and also require
more space as well (usually the chilled water supply
is outside of the mechanical room altogether).
i can try and come up with proper terminologies for
the two different types, if you wish. we've got a
smaller building housing three 600's and three 500's,
and since 1996 we've never had a summer when temperatures
didn't rise no matter what we do. this building is
of the former type. then there's a much bigger building
that houses 800, 600's and a 900 soon to come, and
this building works like it should; probably +/- 1C.
also, with the more modern machines you don't actually
need +/- 1C, i believe. but the older AMX's we have,
every time we have an HVAC problem and the room warms
up, something goes wrong and sensitivity/noise params
are inevitably worse after each incident.
but i agree with you, temperature control in a magnet
room (especially if you have a bunch of machines in
one room) is still the single most important parameter
for getting good experiments and having stable hardware
performance.
2***************responses**************
We have one room with a very tightly controlled temperature which cost
lots and lots of money and we have 2 other spaces which are not so
tightly controlled. I think an easier requirement to ask for is for the
temperature to vary slowly. The temperature can vary a few degrees but
you want it to do so slowly, a couple of degrees per hour. We have 2
heat pumps in one room and if I had it to do over again they would be
place differently and we would have a means of keeping the air in the
room mixed better.
I don't know what issues you have humidity you have out there but we
often have to add dehumidifiers in the summer and humidifiers in the
winter because this turns out to be a big problem for us. You don't
want someone breaking a sample in the probe because they jerked when
they caught static from the magnet. Our normal HVAC system is not
setup for controlling humidity. You also need to find out about the
between season issues. You may need to cool when everyone else is
heating and vice versa. HVAC engineers usually do not consider this
issue at all. This is another reason we like heat pumps.
3***************responses**************
I don't wish to disappoint you, but I would have to agree with your Physical Plant comment to a point. A more reasonable specification would be +/- 2 degrees Celsius for the ambient temperature near the spectrometer.
I have tried over the years to do better, but without success in a real world building. I got the 2 degree level of ambient temperature stability only by installing a prefabricated environmental chamber in a very large lab. The chamber was big enough that two spectrometers could be put inside of it.
My operating experience suggested that the rate of temperature change is as important as the stability of the set point. This means that the heating and cooling systems must be capable of PID control. The capacity of the system must be carefully sized such at it runs at least 80% of the time in either mode. Estimating the heat loading from spectrometers and people is something of an art.
Another important factor is to keep the probe gases well regulated. This requires a ballast heat exchanger placed in a fairly stable temperature environment and some other tricks.
If you are interested in any of the above in more detail, contact me via e-mail. It is possible to achieve something near the value that you would like, but it will cost additional money over a conventional set-point system.
4***************responses**************
We do... when it works. On a good day we maintain +/- 1.5 F or better.
Only thing is that we have had many bad days due to break-downs of
either the building heat, chiller, compressed air (which drives the
stats), or the whole lot. By now, much has been repaired or replaced,
so we are hoping for better reliability
5***************responses**************
That could be a problem. There are a couple of things to consider.
Most Chem buildings are OSHA required to turn over the air every
so often via a exhaust system. Here at [univ] we have quite a vacuum
on the building. I can't remember the specs but it is several
times a day. To regulate room air temperature along with this
requirement could be a problem. When they did our building 10
years ago I also looked into this for our labs. I was told it
was a tough requirement that would cost a pile of money.
Currently, with standard air control, I see +/- 2 C at a max.
6***************responses**************
Right now, it is 80+ in the lab and 85+ in my office. At best, I can
sometimes get down to around 70. So I have +/- 5 degrees around 75.
The university chilled water is non-functional more that it is
functional. There is _nothing_ I can do.
So, no I don't worry about temperature swings at all. I worry about
things getting too hot to run.
[Similar to what we've been experiencing at UW-Chemistry all spring!!
The not at all amusing part of this is how many problems, many
intermittent, show up as a result of the lab being too warm.]
7***************responses**************
Rest assured that one can achieve +/- 1 deg C. I have managed to get
our air-conditioning people to do it even in rather large rooms. I would
be happy to send you plots from a digital temperature monitor that I bought
from Omega that demonstrate that we do it. (By the way, one of the best
purchases I ever made was these temperature monitors from Omega. They end
all arguments about room temperature stability.)
8***************responses**************
It is, of course, possible to do this. All it takes is an HVAC engineer with the appropriate can-do attitude and a (possibly) unlimited amount of money.
When I upgraded my high-field lab, we faced the same problems. The HVAC (facility) people said that +- 1C was just not going to be possible. I also demanded reasonable humidity control ([here], we worry about too little humidity!).
In the end, the answer they came up with was to take our needs off-line. That is, the high field lab has its own heating and cooling system, complete with private evaporator outside. This runs 24/7/365. The room is about 300 square feet, three exterior sides (including the roof), about 100 square feet of outside window (solar gain, winter loss), one emergency exit to the exterior, and double wide doors with vents to the main lab area. An exhaust fan runs continuously and pulls air from the main lab, through the high field lab, and then exhaust to the outside (don't know the flow specs). The HVAC system is a two-ton design with room to spare (we don't use it to capacity). There is one Inova-500 in the lab now with a second system planned for the future. There are six Sun, SGI, and PCs, and two laser printers. On the hottest day of the year last year the room was 68 degrees. On the coldest day of the winter this year the room was 68 degrees. The temperature varies by less t
han 1C in 24 hours (my spec.). The humidity...well, that is another story (25 to 45 percent, sometimes drops below 20% in the winter).
The whole thing including duct work, unit, blah and blah was over $30K. By the way, be sure you specify Aluminum duct work and banding and, if you can afford it, a stainless HVAC unit in the lab. (Since you are remodeling, you might also consider incandescent lights rather than fluorescent. There *may* be some noise reduction gained from this switch although you will probably really get the Texas-Oil award if you do go that way!
My main lab temperature varies by five degrees over twenty-four hours. It is on house air and the doors are open since the lab is open-access.
9***************responses**************
Well, using refrigeration only, the physical plant
statement is quite correct. However, that's only
true if one only uses refrigeration.
To get a +/- 1 degree C variation, one must reheat
the air after cooling it just like we do in VT probes.
For a room, one uses an oversized refrigeration unit
that runs constantly, and the air then flows through
a digitally controlled heater of at least equal power
to the refrigeration. Of course, for a room, because
the refrigeration unit is removing humidity, a humidistat
controlled water mist or steam based rehumidification
system {fed with deionized water to avoid deposits
build up} will also be required with this arrangement.
This is the only way I've ever seen to get this amount
of precision in room temperature control.
I wish you lots of luck getting anyone to agree to
such an arrangement. In my experience, physical
plant designers do not like these more complicated
designs and do not like the idea of constantly
running refrigeration units.
10***************responses**************
Temperature control within +/- 1 degree C, or even +/- 1 degree F is certainly technically achievable and is commonly used in, for example, mainframe computer rooms. In fact, the best people to contact for this are contractors who provide environmental controls for computer rooms. The major manufacturer of this type of climate control hardware is Liebert, Inc. Although I don't have any of their equipment, I did look into buying some a few years ago and have a lot of their product literature. Generally, they regulate the temperature to within +/- 1 degree C by cooling below the set point with an air conditioner that is on continuously and then reheating to the desired temperature with an electrical coil connected to a proportional heater controller. The critical idea here is to avoid any switching of air conditioning equipment on and off, which could cause t1 noise in multidimensional NMR experiments due either to the cyclical change in temperature or to a cyclical change i
n barometric pressure. You can look at their products and find distributors at Liebert's web site:
http://www.liebert.com
11***************responses**************
Our facilities people built me a lab to those specs but it required double
sheet rock insulated walls, special alarmed temp controls and huge HVAC
ducts. It was nice but expensive and not practical for refitting an existing
lab. When we built the new R&D building with good but standard HVAC in all
the labs I didn't notice any problems. One key may be the huge diffusers
that spread the incoming air widely to avoid temp gradients and wind
currents. While its no longer guaranteed, I think we maintain the most
important spec which is +- 1 deg C /hour. As long as the temperature change
is slow and within reason, the instrument VT control should suffice for long
experiments.
12***************responses**************
We have a Liebert system (special model) for this purpose on our 600 MHz
instrument. I was told that there are only three this special model
system in [our state], one is in air force base lab, we have one, and one
in somewhere else.
13***************responses**************
From my experience with conventional air conditioners in our NMR lab, I am
of the opinion that these units cannot control air temperature to better
than +/- 2 degrees. There are many reasons for this. First is the
control system that is used to decide when to turn the compressor on or
off. It is either a bimetallic element or a bourdon-tube that has
significant hysteresis (this is generally for the sake of the compressor
as this should not be cycled on and off too rapidly). Then, the fan speed
that is used to blow air is not generally dependent on room temperature,
but is set by the user. Finally, the temperature sensor is positioned at
the output of the air duct and not in the center of the room where it
would be best positioned to sense the actual room temperature.
Also, unless air circulation is very good, there will be hot and cold
spots inside the room. People entering the room would cause a fluctuation
in temperature that takes time to restore.
The ideal would be an A/C unit with an electronic control, utilizing a
sensor in the middle of the room (or multiple sensors throughout the
room), coupled to fan speed and an elaborate control of the compressor (or
better yet, two compressors for faster cycling) would be what you are
looking for, but I'm afraid this doesn't exist (yet). But good luck
anyway. If you find such a system, I would be glad to hear of it.
14***************responses**************
Don't let them get away with it ! - +/- 1 deg C is definitely achievable.
For us temp. control was always a big problem in our old lab - leading to
horrible spectral artifacts. So when the new lab was built, to get this right
was one (of many !) priorities.
I went through a similar situation with our architects/planners/contractors, we
finally got a system installed which although it wasn't entirely satisfactory
initially, with a little bit of tweaking from our maintenance guys it now works
fine.
In the early days I monitored the room temp religiously but now I hardly
bother. Most temperature changes are due to doors opening, bodies in the room
and some diurnal variations.
I've attached an Excel spreadsheet with some randomly chosen data sets for your
inspection. These are not necessarily the best but I think all show +/- 1 deg.
to be achievable.
15***************responses**************
The best system that I encountered, less than +/-1 0C drift was at [a]
University [in Europe]. Here they used chilled water in aluminium panels to cool
the rooms and as the flow was constant there was no cycling. The temperature
regulation was PID based with a very long time constant.
16***************responses**************
yes it is feasible ---- you use electric heat against the cruder air-conditioning system to avoid the swings. Run the chilled water continuously through a variable throttle valve into a fan coil unit (i.e. chilling is always on though the rate can be controlled - fan speed constant so no drafts) and heat the output for fine control and you can do 1 degree farenheit --- we've done that for 40 years in nmr and mass spec labs I've dealt with.
jack
--
Prof. Jack M. Miller,
Associate Vice-President, Research and
Dean of Graduate Studies,
Professor of Chemistry,
Brock University,
St. Catharines, Ont.,
Canada, L2S 3A1.
Phone (905) 688 5550, ext 3789
FAX (905) 684 2277
e-mail jmiller_at_brocku.ca
http://chemiris.labs.brocku.ca/~chemweb/faculty/miller/
17***************responses**************
Charlie,
Get the best HVAC unit you can afford. Your p-plant people just don't want
to guarantee +-1 degree C in case it doesn't work out. The single most
important thing is to NOT have ducts blowing directly onto the magnets.
Of course there are standard ways of figuring out what you need based on
room volume and estimated heat output.
My lab has a 75 kW air handler that is mucho overkill, especially in light
of the fact that consoles are getting smaller and thus produce fewer BTUs.
Unless you are planning on stacking shielded magnets on top of each other
like cordwood, you can probably get an air handler that is reasonable based
on the power consumption of your present instrumentation and your lab
dimensions.
BTW, my air handler is a TRANE and it has been working very well for the
last three years since I moved into my lab (ca. 4000 sq ft with ~13'
ceiling, with 6 NMRs, an EPR, FT-MS, GC-MS, FT-IR, polarimeter, seven
workstations).
Feel free to post this information.
Jeff
Dr. Jeffrey H. Simpson, Instrumentation Facility Director
Department of Chemistry, Massachusetts Institute of Technology
77 Massachusetts Avenue, Cambridge, MA 02139
617-253-1812 ofc, 617-253-1806 lab, 617-253-0873 fax
e-mail: jsimpson_at_mit.edu
18***************responses**************
This is not easy but here are some keys to doing it.
1. Have a huge exchanged air volume by circulating air between this and
adjacent labs.
2. Multiple independent thermostats controlling independent zones,
carefully adjusted so that no one thermostat becomes the "master"
3. Lots of air circulation between the zones
The secret is to have enough noise in the system so that the temperature
oscillations caused by the thermostats on/off cycles are incoherent and
averaged out.
On a good week our lab does +/- 1 F so don't let them tell you it cannot be
done. It might cost a bit, but it can be done.
19***************responses**************
This is certainly feasible! We have such a facility here for our
Inova500. In fact, we do it with diffusive air flow. I designed the
system such that the cold air is dumped into the ceiling plenum via ducts
and it diffuses through several properly located (determined empirically)
ceiling panels with ~0.25 inch holes. There are properly located (at each
end of the room) return air vents which are ducted back to the AC unit. It
maintains the RT at 22 š +-0.5 F. The room is ~ 20 ft X 20 ft and 15 ft
high. It has one spectrometer in it. I am sure that it can be made to
work with 3 times that area and two or three more spectrometers. It has
worked flawlessly for ~4 years. The system has a computer controller by
Landis &Gyr Powers, Inc., 1000 Deerfield Parkway Buffalo Grove, IL
60089-4513, 708-215-1000.
20***************responses**************
I think much of this depends on hour your HVAC system works in your
particular space (I've learned a lot about this in the two years that I've
served as liaison to our building-overhaul project). In my labs, I have
supplemental A/C units in each room; however, they (the A/C units) are often
fighting (rather than assisting) the dysfunctional environment control
systems in the building. If the building is properly controlled and
balanced, and if the general environmental controls are working, then
supplemental air should get you close to +/- 1-deg.C. (at least over a
reasonable period). You can't expect this precision over several months,
but you should be able to get it over a 48-hour period. Some labs hover
around 30 Deg.C. in the hottest months of summer, and I drop to around 20-22
Deg.C. (sometimes) in the winter. However, these are gradual changes, and
don't affect instrument performance during the time scale of a set of NMR
experiments.
Maybe you could re-phrase your specification, giving a time-constraint on
the +/- 1-degree C. control. They should be able to assure that the room
temperature doesn't fluctuate more than that (assuming that you don't have
people leaving doors open or something) over 2-3 days.
21***************responses**************
Way back when... We had a WH 90 that used a Haskris cooling
unit for the magnet. The cooling compressor in this unit was a
standard A/C unit (can't remember who made it) but instead
of cycling on and off like a regular A/C unit it had something called
a "hot gas bypass valve" to maintain a very constant temperature.
Perhaps your physical plant people or HVAC contractor would
consider installing one of these?
Another option might be a close differential thermostat.
The objection to the hot gas bypass valve might be energy
consumption, as the compressor runs continuously. The
close differential thermostat causes frequent compressor
cycling which wears it out faster.
I hope that you find a solution. +/- 1C is a two degree range
which shouldn't be too hard (about 4 deg F?)
22***************responses**************
Sounds like it would cost a lot of money to insure +- 1 degree
through-out the lab area. Especially if the doors are opened a lot.
Definitely try to reduce the large excursions and when you pick the unit
, double the capacity. Try to keep the a/c draft from blowing directly
on the equipment.
23***************responses**************
They have a point. It's not trivial ... But is definitely cheaper to do
during a remodel than from the ground up. You may only need some
variable motors, servo valves and a good multi-channel PID controller.
+/-1 degree in the room implies hiring (or borrowing) a HVAC
designer/consultant who knows classical control theory ... you may have
a local expert willing and able. It also requires special fans and
motors to control mass flows. PP can do this for you if you talk to the
right people. This is the stuff the mech engineers learn in school but
don't usually practice in the field in HVAC. If they haven't done it
before, they would need to learn.
Most ac systems are either on or off and the temp oscillates. You can
control (reduce) the amplitude of the oscillation by having the desired
target temperature closer to the cooling temperature of the ac outlet.
Think of a heater that heats a water bath (load). If the heating
element is full on every time it turns on then the oscillations will be
high. If it is only partially on then the temp. will be moderated ...
but takes longer to heat up.
Consider;
Omega (high end temp control experts)
What are the constant loads?
What are the changing loads?
Can changing loads be mitigated? ... Are there people coming and going
... etc.?
Perhaps you should enclose the magnet and control it's environment?
Enjoy the problem ... Sound like a lot of fun:)
24***************responses**************
After my old 5 ton AC unit died last June, I started look into a replacement
unit. My dept chairman decided that I did not need one over the winter but
promised to produce the money for a replacement by spring. The old unit was
about 3 ton overkill except in the hot days of summer. We have a Bruker
DPX400 and DPX300 and a west facing brick wall (no insulation just concrete
and brick). The replacement unit was install two weeks ago so I do not have
a lot of experience yet but I am so impressed with the unit I will install a
similar unit in our new house we are building.
The old unit was a 1960's vintage floor unit that was noisy (the fan bearing
were shot) and when it kicked in the unit would over cool the room (5-6
degree F) swing. The room originally had two electro-magnets with a much
larger heat load. If I needed to run any NOE's I normal ran them in the
morning when the AC unit was not cycling.
Here is what we have done. The building has a closed loop glycol-water
system connected to a cooling tower on the roof. We also have a duct system
installed in the ceiling of the NMR room to distribute the AC air. We
settled on a liquid exchange heat pump too both cool the NMR room during
normal days and heat the room if needed during the winter. (the physical
plant has decided not to heat the building during the weekends in winter to
save money.) I purchased a Carrier two stage heat pump 50YBV060R. The unit
is quiet. It has a very low speed fan setting to continuously move air. When
the low speed cooling unit kicks in, the compressor is a very quiet hum.
Most important is the temperature drop is 1 F. I was told that the low speed
cooling is approx 1.5 ton. If I change the thermostat several degrees I can
get the unit to jump up to the high speed cooling. This increases the fan
speed and the unit will bring the room temp down. So I expect that I will
have no problem with the summer heat load with the sun on the west wall. And
right now I am enjoying a nice comfortable room with a 1 F temp control.
If your physical plant people are like those that work here at [univ], they are
vintage (old style) AC people with experience on big noisy units and not the
new high tech units.
25***************responses**************
I tried for 0.1 degrees. Didn't work either, but they did give
us a lot of large vents. [co-worker] went to those meetings,
and said that they thought we were crazy for asking. I think
that having more capacity does even out the temperature swings.
Probably the best way would be to super insulate the facility.
26***************responses**************
We installed one of the Liebert units used in computer rooms. It can also control humidity. It cost about $30,000. It was a good investment because all of the electronics are much more stable. The university heating/cooling tends to shut down nights and weekends. We recycle most of the air in the room through the unit.
27***************responses**************
A bit late, but here are my two bits on the subject. It depends
what kind of a system you have. An air conditioner is an on-off
device that needs a 2 degree hysteresis so it will not cycle all
the time. However, if you are in a fortunate position (like we are)
that you have a central chilled water supply, a fan coil unit (heat
exchanger) can be equipped with a proportional control. With
such a system, the only variation comes from the environment,
i.e. doors opening etc.
28***************responses**************
It can be done if you start out with the right type of HVAC system. Our
building is designed so that hot air and cold air are brought to the rooms.
The hot air and cold air are mixed using mixing boxes located above the
false ceiling to give the ambient room temperature via air ducts located
throughout the room. Air flow dampers in the mixing boxes are controlled
by a pneumatic Johnson control regulator located on the wall. Once the
temperature is adjusted the dampers on the mixing boxes will adjust to mix
the appropriate amount of hot air with cold air to give the desired
temperature. Once this is done dampers pretty much stay in that position
so one does not get temperature cycling caused by the compressor cycling as
with an air conditioner. Hot and cold water is fed to our building by our
power plant and very large heat exchangers and blowers are used to generate
the hot and cold air for the mixing boxes.
For added stability we (campus engineering) have installed a process
chiller to work along with the building's HVAC system. The process
chiller's compressor run continuously using a shorting valve between the
high pressure side to the low pressure side of the compressor. This avoids
compressor cycling. It circulates cold water to a heat exchanger designed
to work with the hot and cold air deck mixing boxes. We keep the
temperature of the room set to 68F and it does not vary more than +/- 1F
per day when everything is working.
That is a basic idea of how the system works, there is a complicated set of
pneumatic controls to integrate the process chiller into the mixing box
system's operation. There is a certain amount of air that is recirculated
in the room and a number of nuances designed by engineering. The design is
not trivial and the installation will require competent HVAC people.
The cost was I think ~ $17000 not including the chiller. If you want more
info let me know.
***************2 others via phone****************
both controlling better than +/- 1C
************************************************
----------------------------------------------------------
Charles G. Fry, Ph.D. Tel: (608)262-3182
Director, MR Facility Fax: (608)262-0381
Chem. Dept., Univ. Wisconsin
Madison, WI 53706 USA email: fry_at_chem.wisc.edu
----------------------------------------------------------
Received on Tue May 22 2001 - 18:39:47 MST