Refrigerant-grade ammonia contains only 33 parts per million (ppm) of water -- about one drop of water per gallon of ammonia. Put another way, a 48" diameter by 120" tall pump accumulator filled halfway (about 2,500 lb of ammonia) contains less than a shot glass full of water. In comparison, household ammonia solutions typically contain about 95 percent water. (Imagine about a cup of anhydrous ammonia mixed with a gallon of water.)
It might be intuitive to think that the ammonia needed to run in a refrigeration system would have to be purer than household ammonia, but how pure is pure enough? Isn't the ammonia refrigeration industry going a little overboard with its insistence on 33 ppm? Why such a fanatical obsession with water?
The reason is that water has a number of destructive effects on ammonia systems. For example, despite its caustic and toxic nature, pure ammonia does not affect many common metals. But when a little water is added, the aqueous ammonia will react rapidly with metals. This reaction occurs because ammonia mixed with water produces ammonium hydroxide (NH3+ H 2O → NH4++ OH-). And even though iron and steel do not react individually to aqueous ammonia, the ammonium and hydroxide ions can cause galvanic corrosion between two different metals near each other, especially if the metals lack a protective coating of oil.
Additionally, if water in an ammonia system gets into the compressor oil, it will cause a series of chemical reactions that create nitro compounds, which are commonly called sludge. Some of these compounds are soluble in ammonia and can escape with ammonia vapor through the oil separator and into the system. There they will clog strainers and filters and cause operational problems in valves. Refrigerant oil also contains no emulsifiers that would allow the water to dissolve in the oil. The water droplets cause the oil film inside to slide and roller bearings to break down, which dramatically reduces bearing life.
Water and EnergyAlthough the harmful effects of water in the oil are real, it can be difficult to trace them back to water in the ammonia as the root cause. The effect water has on energy consumption is easier to quantify. Water contamination changes the pressure-temperature relationship of ammonia in a definite, well-understood way: It raises the saturation temperature (boiling point) of the ammonia. To maintain a desired suction temperature after water is added, the pressure must be reduced, which requires the compressors to work harder and consume more energy.
Let's go back to the hypothetical 48" accumulator containing 2,500 lb of liquid ammonia. For refrigerant-grade ammonia, the pressure in the accumulator would need to be about 15.6 psig to maintain a temperature of 0°F (-18°C). Now let's add about 1.5 gal of water to make the ammonia more like agricultural-grade purity (0.5 percent). If we keep the pressure at 15.6 psig, the temperature in the accumulator will rise about 0.2°F (0.1°C) -- not because we've added warm water to cold ammonia, but because adding water raised the boiling point of the ammonia.
This temperature change probably is not enough to matter for most applications. Obviously, the effect of 1.5 gal of water is pretty small. It isn't until we've poured about 9 gal of water into the 48" accumulator, making the contamination about 3 percent, that we will see a 1°F (0.6°C) temperature rise. To maintain the original 0°F temperature, we will have to lower the suction pressure from 15.6 psig to 14.8 psig.
Of course, 9 gal seems like a lot of water. But there are a lot of ways that water can leak into a system over time. Ask yourself: How big is your system? Is the low side of the system operating below atmospheric pressure? How many years has it been operating? It might well be that your ammonia refrigeration system contains even more than 9 gal of water.
There's a rule of thumb that says each 1°F increase in suction temperature corresponds with a 2.5 to 3 percent loss in compressor capacity.1Translating this lost capacity into wasted energy dollars is a tricky business that depends on a number of assumptions. But some fairly straightforward calculations show that this 3 percent water can cost $1,000 or more per year for each 100 tons of refrigeration.
A Hidden ProblemThe problem of water contamination has been known for a long time. An excellent paper by J. Edmonds was presented on the topic at the RETA National Convention in 1996.2But if water is such a problem, why hasn't it been given much attention?
The main reason is that the problem generally is hidden. The 3 percent water mentioned earlier typically results in less than 1 psig change in the suction pressure of a 0°F coil. Unless there is a substantial leak in a heat exchanger, water will collect slowly over time in the low side of the system. The slow increase in water concentration, and the correspondingly slow change in system performance, can be ignored easily as measurement error or “normal aging.”
Throughout the life of an ammonia refrigeration system, water enters at various times and places as a result of the system's construction, operation and servicing, as well as through humidity in the air. Many of these sources are shown in figure 1. Moisture that leaks with air into the low side of the system will combine immediately with liquid ammonia. Meanwhile, the air that leaked in will move with the ammonia vapor to the high side of the system, where the air can be purged. If the air purger has been removing air from the high side, then you can be certain that there is water on the low side. However, water vapor can travel with gaseous ammonia only in extremely minute quantities. Any measurable water found on the high side of the system usually is the result of droplets carried over from an ineffective liquid separator. Water that does get to the high side of the system either mixes with oil in the compressor to form sludge, or it is carried at high temperature through the compressor with the ammonia vapor and into the condenser. There, the ammonia/water mixture liquefies, flows to the high pressure receiver and ultimately finds its way to the low side of the system.
Except in a few circumstances, the water simply cannot leave the low side. It will collect in evaporators, chillers and other places where ammonia is allowed to boil and the ammonia vapor is drawn off. In a mechanically pumped recirculation system, water will become concentrated in the accumulator. In a gas-pumped system, the water will be transferred through the dump trap (pumper drum) into the controlled pressure receiver. On a gravity-flooded system, the water will accumulate in the evaporator or heat exchanger. Water also will collect in “closed coil” (shell-and-tube or shell-and-coil) type economizers.
The water is trapped because it cannot evaporate with the ammonia; the temperature is simply too low. As the amount of accumulated water increases, either the temperature must increase, or the compressor must work harder to maintain the desired temperature.
There are some circumstances when water can escape from the low side of an ammonia system. For example in a direct expansion system, where water accumulates inside the evaporator, ammonia vapor exits at high velocity during normal operation. This vapor can carry water droplets out of the evaporator. If there is no suction accumulator, these droplets can move directly into the compressor and mix with oil. Also, as the water concentration inside the evaporator increases, the coil exit temperature will increase. This temperature increase might be misinterpreted as an improperly adjusted superheat. In attempting to reduce the superheat, the expansion valve might open too much, and the ammonia/water mix can slug the compressor.
Also, liquid ammonia remaining in a low-temperature evaporator at the start of a defrost cycle can contain a significant amount of water. If this liquid cannot be drained completely before hot gas is applied, the defrost condensate will contain water. This condensate can be routed to an intermediate pressure vessel and used for liquid-injection oil cooling.
Measuring Water ConcentrationSo how can you tell if there's too much water in the system? One way would be to measure the suction temperature and pressure accurately and then determine whether the measurements correspond precisely to the ammonia property tables. But the key words here are “accurate” and “precise.” Remember that a difference of less than 1 psi or 1°F is significant. For a meaningful answer, you'll need a sample of ammonia taken from the places where water can collect. As discussed earlier, evaporators, surge drums, pump accumulators and closed-coil economizers all are likely places where the water will concentrate.
But just as important as where to take the sample is the issue of when to take it. Don't forget that the liquid level inside pump accumulators can vary even though water will stay inside the vessels. This means that when the liquid level inside a given vessel is high, there is more ammonia present to dilute the water. When the liquid level drops, there is less ammonia, and the concentration of water rises. So if a 2 percent water concentration exists when the liquid level inside the accumulator is high, that concentration might increase to 6 or 7 percent while the system is operating with a low liquid level. The energy penalties, of course, vary accordingly.
So just how do you check the water concentration? In theory, it is simple: All you need to do is collect the sample from the system's low side in a container and let the volatile ammonia evaporate (in a safe place). What remains in the container is an aqueous-ammonia mix that is probably about 30 percent ammonia and 70 percent water. By performing some calculations based on the amount of this residual liquid, you can determine the concentration of water in your system.
That's the theory, anyway. In practice, you have to be sure that moisture from the surrounding air doesn't further contaminate the sample while the ammonia is boiling away. (Ammonia is extremely hygroscopic and will draw in moisture from the surrounding air.) You'll also need to measure precisely how much liquid remains after more than 90 percent of it boils away. To do this, you'll need a special container that is available from refrigeration system vendors. After you measure the amount of residual liquid, you'll need to convert the number to a percentage of water contamination. You can perform this conversion using a mathematical method given by some vendors of test containers, or you could use a chart like the one in figure 2, provided by other vendors, based on a 100-ml initial sample.
Suppose, for example, you start with a 100-ml sample of liquid from the system and let the ammonia boil away until you are left with about 6 ml of residual liquid. Comparing this quantity to the chart in figure 2 shows that 6 ml of liquid corresponds to about 5 percent water present. In other words, the sample you started with was significantly contaminated.
Incidentally, both the initial ammonia sample and the liquid residue should be clear and colorless. If the liquid is amber or brown, it indicates sludge formation due to the compressor oil having been contaminated by water. In that case, a sample of the oil should be sent to a laboratory for a chemical analysis to determine if its quality is still acceptable.
Removing the WaterIf your system contains a significant amount of water, won't your air purger take it out?
No. As noted earlier, water stays on the low side of the system; only dry air moves to the high side, where the purger is connected. Even if water vapor did move with the ammonia, air purgers simply cannot remove water from ammonia vapor. This limitation occurs because conventional airpurgers* work by cooling and condensing the ammonia vapor to separate it from non-condensable gases. Water vapor is not a non-condensable gas and therefore cannot be separated from ammonia by cooling it in a purger. However, an air purger connected to the high side of the system can be a qualitative indicator of the presence of water. As noted previously, if you have a purger that has been removing air from the high side, then there's some water on the low side.
To remove the water, instead of cooling the warm ammonia vapor, you must warm and vaporize the cold contaminated liquid ammonia through distillation to separate it from the water, oil and any other contaminants. This is the same process that allows water to become concentrated on the low side of the system, only now you are capturing and concentrating the water mixture in a separate vessel. Once the concentration of water is high enough, you can drain it from the system.
On smaller systems, it might be acceptable to shut down the system, remove the complete charge of ammonia, then recharge and restart the system. The new charge could be either new ammonia or the old ammonia that has been cleaned in a distillation unit. Unfortunately, the combination of downtime and the quantity of ammonia that must be handled often prohibits this approach.
Larger systems sometimes are cleaned of water in batches by sending a fraction of the contaminated liquid ammonia into a distillation unit. There, the ammonia is evaporated to separate it from the water, and the pure ammonia is collected and returned to the system. The separated water and other impurities are drained and disposed of, and then another fraction of the contaminated system charge is admitted and purified. The process continues until the entire system charge reaches the desired level of purity. It can take weeks to clean out a large system by the batch method. Often these distillation units need to be adjusted carefully for particular operating conditions, monitored, and emptied at unpredictable intervals. For some of these batch-type water removers, a change in the system's operating conditions can result in sudden and violent boiling inside the distillation vessel. This, in turn, can send some of the impurities back into the system.
Other types of ammonia-cleaning units, sometimes called stills or “regenerators,” can be integrated directly into the refrigeration system. These machines typically use hot discharge gas or warm condensed liquid to provide an inexpensive heat source for boiling the cold contaminated liquid ammonia. Some units use hot water in a jacket surrounding the main vessel for the heat source. These units can vary in complexity and effectiveness but generally are capable of operating continuously. As with the batch-type units, if the integrated unit is not designed properly, changes in operating conditions can send water droplets back into the system. With a well-designed unit though, intervention is needed only periodically to drain the collected impurities.
Some water removal units use electric heaters to warm the ammonia-water mixture to temperatures above 150°F (66°C). Warming the mixture allows as much ammonia as possible to be removed from the water and other contaminants before the system is drained. This approach can indeed lower the ammonia remaining in the effluent to as little as 5 percent (about the same as household cleaner). Unfortunately this approach adds the complexity of heaters and thermostats. In addition, heating the mixture generates steam -- just like a pot of water warming on a stove. If the temperature is high enough and the pressure in the vessel is low enough, the liquid might even boil. The resulting steam and water droplets are then sent with the ammonia vapor back into the system.
Water contamination in an ammonia refrigeration system is a serious problem. Understanding how it got there, how to measure it, and some methods for removing it can help you choose the best course of action.
*One relatively new purger takes liquid from the system's low side, as well as vapor from the system's high side, to separate both air and water inside the same unit. However, the water is being removed from the liquid ammonia, not the gaseous ammonia.
References1. Bulletin No. 108, “Water Contamination in Ammonia Refrigeration Systems,” International Institute of Ammonia Refrigeration, Arlington, Va., www.iiar.org.
2. Edmonds, J.M., “Ammonia System Water Contamination,” presented at the Refrigerating Engineers and Technicians Association National Convention, Valley Forge, Pa., Oct. 1996.