Figure 1. The ammonia refrigeration system was laden with the mystery powder. A chemical analysis identified the material as ammonium fluoride.
One of the critical steps in the standard procedures for starting up a refrigerating system is to evacuate the system prior to introducing refrigerant to it, regardless of the refrigerant being used. This step applies to first-time startups and to service- or maintenance-related shutdowns where the whole system (or part of the system) has been opened up.

This column discusses a case where the failure to evacuate the system caused contamination so severe that the system was rendered totally inoperative.

The case involved a large Midwestern university with a food store and commissary using a two-stage ammonia refrigerating system. The system served a 20,000 ft2 storage freezer and a small sharp-freeze on the low-temperature side and a number of storage and workrooms on the high-temperature side. There were 11 such high-temperature rooms for meat storage and cutting, produce and egg storage, food preparation, and bakery and flower storage. The refrigerating system cooled the storage freezer and the sharp-freeze with flooded-type air-cooling units fed with surge drums. The refrigerating system also cooled all of the high-temperature rooms, which used air-cooling units fed with thermostatic expansion valves (TXVs).

Early one Saturday morning, it was noticed that the temperatures in all of the refrigerated spaces were rising. An inspection of the system and the major system components showed nothing unusual. Almost an entire day was spent trying to track down the source of the problem, focusing mainly on the machinery room equipment, after which a consultant was called in.

After reviewing the symptoms, the performance of several of the cooling units in two of the high-temperature rooms was checked. With the room thermostat calling for cooling and the liquid solenoid valve open, temperature measurements of the air into and out of the units indicated that no cooling was taking place. The piping downstream of the TXVs was neither frosted nor wet and, in fact, was close to room temperature, as were the liquid distributors on the air coolers.

One of the units was selected for further study. The liquid shutoff valve was closed, the suction pressure setpoint for the compressors' controls was lowered, and the unit was allowed to pump out as much as possible. After a time, the liquid solenoid valve and the suction shutoff valve were closed. The system internals were then inspected. The first step was to open the liquid piping and examine the valves and strainer. An unidentified powdery substance filled the strainer basket (figure 1). The system was so laden with the substance that on a number of the units, the accumulation completely filled the strainer body as well as the upstream piping for several feet.

A chemical analysis of the powder identified it as ammonium fluoride. The analysis also revealed that the substance was present in two forms, crystal and filament, one being more refined than the other. The source was not immediately clear. The piping on all of the cooling units was disassembled, and the pipes, valves and strainers were cleaned. The system was then reassembled.

Further investigation revealed the source. The surge drum on the sharp-freeze unit had been leaking ammonia. The maintenance crew undertook to find and repair the leak. The surge drum was pumped down and isolated. It was then charged with nitrogen to a moderately high pressure, and R-22 was injected so a halide leak detector could be used. The leak was discovered in a flange in the liquid piping. The flange was disassembled, new gaskets were installed, and the flange bolts were retightened.

Figure 2. To capture the ammonium fluoride that was distributed throughout the refrigeration system, an elaborate filtration system was designed and installed in the liquid main at the outlet of the liquid receiver.
Although standard protocol and procedure were to evacuate the surge drum and piping before recommissioning and reintroducing refrigerant, for whatever reason, it wasn't done. The pressure in the surge drum was released to atmosphere, valves were opened, ammonia was introduced into the surge drum, and the compressors were allowed to pump out the remaining test charge. (Note that this event occurred prior to when Clean Air Act amendments prohibited venting chlorinated refrigerants to atmosphere.) The ammonium fluoride was produced by the chemical reaction between the R-22 and the ammonia mixing at the elevated temperatures and pressures encountered in the compressors.

Having determined the cause of the contamination and having cleaned out the major accumulations, it was necessary to recommission the system. First, however, it was imperative to consider how to prevent further shutdowns and to try to capture any contaminant that was potentially distributed throughout the system. An elaborate filtration system was designed and installed in the liquid main at the outlet of the liquid receiver (figure 2). The chemists who had conducted the analysis estimated the particle size of the contaminant substance to be about 60 micron. A liquid-line filter for ammonia with a particle size-capture capability of 15 micron was applied in the design. The dual parallel filter arrangement was intended to capture the contaminant, monitor pressure to indicate filter plugging, and permit switchover so that filters could be changed without shutting the system down. It was believed that this design would allow for immediate resumption of operation and the ability to clean up the system over time.

The system shut down again in about three days. Pumping out one unit and inspecting the strainer revealed that the same contamination had reoccurred. The dual filter arrangement had not worked. Further discussion with the chemists indicated that they had erred in their original estimate of particle size. Instead of the particles being 60 micron, they were actually about 6 micron.

In the ensuing investigation, it was learned that the only substance in which ammonium fluoride is soluble is water. The good news, of course, was that water is readily available, low cost and easy to work with. The bad news was that the water had to be introduced into the system in large quantities and then completely removed.

The cleanup protocol involved removal and disposal of the entire ammonia charge (approximately 6,000 lb), mechanical cleaning of valves and strainers at each of the cooling units, flushing the entire system with warm water, drying the system internally by blowing it out with heated nitrogen, evacuating it, and charging it with new refrigerant-grade ammonia. The system worked flawlessly after recommissioning.

The phenomenon of ammonium fluoride contamination became better known after the CFC phase-out resulting from of the Montreal Protocol agreements. A number of large R-22 systems were converted to ammonia in anticipation of the R-22 phase-out, coming in the next 15 years in the United States but already in effect in Europe. Failure to adequately clean the system and remove all remnants of R-22 from it can result in the same experience as the university when ammonia is introduced to the system. PCE