Specifying a recirculating chiller in Bangladesh: capacity, stability and condensation
How to work out the heat load, why catalogue capacity misleads at Bangladeshi ambient temperatures, the difference between setpoint accuracy and stability, and the condensation problem that damages instruments every monsoon.
The laser ran without complaint through January. From the middle of April it began shutting down in the afternoon, always between two and four o'clock, always on an over-temperature fault, and always fine again the next morning. Nothing had changed except the outside air temperature. The chiller had been chosen from a catalogue figure quoted at an ambient nobody in Dhaka sees between March and October.
Work out the heat before you look at a chiller
For most laboratory equipment the heat to be removed is close to the electrical input, because nearly all the electrical energy ends up as heat somewhere. Start from the instrument's electrical input, then ask the instrument manufacturer directly for the figure they specify. Any serious instrument manual states a required cooling capacity, a flow rate and a coolant pressure range. Those three numbers are your specification, and they come from the instrument, not from the chiller supplier.
Rated capacity is quoted at a stated coolant temperature and a stated ambient. Both matter, and the second one is the one that catches people. A chiller rated at a temperate ambient loses capacity in a Dhaka plant room in May, and the loss is not marginal.
Ask for the capacity curve rather than the headline number, then read it at your worst case ambient and your required coolant temperature. If the supplier cannot produce a curve, that is information too.
Add headroom deliberately. An instrument that trips on temperature during the hottest fortnight of the year is unavailable during the hottest fortnight of the year, which is also when the building services are least able to help.
Count everything that will end up on the loop. Chillers get shared: a second instrument arrives two years later, gets connected to the loop that was already there, and nobody resizes anything.
Air cooled, water cooled, or facility water
Air cooled is simplest and needs nothing but power, and it rejects the entire heat load into the room it stands in. If that room is air conditioned you have now paid twice, once to move heat out of the instrument and again to move it out of the room. Above a modest load, put the chiller outside the laboratory or duct its discharge out.
Water cooled needs a facility loop or a cooling tower and gives up much less to the room. It suits higher loads and hot plant rooms, at the cost of making your instrument's availability depend on a system that somebody else maintains.
Direct facility water on a once through arrangement wastes a large volume of water and its temperature swings with the season and with the time of day. It will not hold a tight temperature and it is not acceptable for any instrument whose output depends on thermal stability.
Stability is the specification, not the setpoint
Two chillers can both be set to 20 degrees Celsius and behave completely differently. Setpoint accuracy describes how close the average sits to the number on the display. Stability describes how much the temperature moves while it sits there, and stability is what appears in your data.
Where it shows up: a laser's output power and wavelength move with the temperature of the head; an optical assembly drifts as its mounts expand and contract; a detector's dark current changes; and any instrument holding a focus or a stage position produces a slow drift that looks exactly like a real sample effect until somebody plots it against the chiller cycle. If the instrument manufacturer states a stability requirement, treat it as a hard specification and confirm the chiller meets it at your actual load rather than at full load or no load.
Ask about the control method as well. A chiller that holds temperature by switching its compressor on and off maintains a wider band than one that modulates its capacity, and that cycling appears as a periodic ripple in sensitive measurements. For a teaching laboratory it does not matter. For interferometry, Raman work or an electron microscope column, it does.
Fluid, plumbing and the condensation problem
Use the fluid the instrument manufacturer specifies. Laser heads commonly require deionised water held within a stated conductivity range, because ordinary water corrodes the internal passages and can bridge electrically. Tap water in a laser head is a repair, not a shortcut.
Glycol only where it is specified. It raises viscosity, reduces heat transfer and lowers the effective capacity of the system. Never use automotive coolant: its additives are formulated for engines and attack the seals and platings used in laboratory equipment.
Fit a strainer and a particulate filter, and change them on a schedule. Micro-channel cold plates block, and once blocked they usually cannot be cleared on site.
Use opaque tubing. Clear tubing in a lit laboratory grows algae, the algae ends up in the filter, and then in the cold plate.
Wire the flow switch into the instrument's interlock so that loss of coolant shuts the instrument down, rather than merely displaying a message on the chiller in an empty room.
Condensation is the failure nobody designs for. If the coolant runs below the dew point of the room air, water condenses on the lines and inside the instrument. Through the Bangladeshi monsoon, the dew point in a room that is not tightly humidity controlled can sit high enough that a modest chiller setpoint drips steadily onto electronics. The options are to control room humidity, raise the setpoint above the dew point where the instrument permits it, insulate every cold surface including the fittings and valves, or specify a chiller that tracks dew point. Choosing between them after the instrument has been dripped on is the expensive route.
Installation, servicing and what actually fails
Give it clearance in and out, with the discharge not blowing at a wall, at a person, or into another instrument's air intake.
Put condenser coil cleaning on a schedule with a named person and a date. In a laboratory near a construction site or an unpaved road the coil blocks within months, and a blocked condenser is the most common reason a chiller quietly loses capacity long before it fails outright.
Change the fluid and flush the system at the specified interval, and keep the record with the instrument's maintenance file.
Keep the chiller off the optical table and off the instrument bench, and out of the room where you can manage it. A compressor is a vibration source, and putting one on the same structure as a Raman stage or a microscope column undoes the isolation you paid for.
Check the refrigerant type and confirm that a technician in Bangladesh can obtain it and is competent to handle it. A chiller charged with something unobtainable locally has one life.
Decide whether the chiller sits on the standby supply. An instrument on a UPS with its chiller on the raw mains will keep running and overheat during an outage, which is worse than a clean shutdown. Either interlock them or power them together.