BY KEM RUSSELL

Thinking about water LESSON and where to find it, I go back a few years ago to when one of my daughters and I were hiking the Pacific Crest Trail (PCT). We planned to do a north bound hike, which starts at the U.S./ Mexico border east of San Diego and heads to Canada. Hoping to avoid wet and cold weather (yes even in southern California), and to also avoid very dry conditions, we started hiking in midApril. Fortunately, and unfortunately, the year we hiked it was one of the driest on record.

The desert section of the PCT is a long 700 miles of almost constant search for water. The focus of each day was finding the next water source, which were at times many miles apart. Finding water became a valuable skill.

During these weird COVID-19 days, weeks, and months, I found myself with some extra time on my hands and I started thinking about finding water, specifically finding how much water there might be in an ammonia system. I would guess a similar thought has crossed the mind of many refrigeration operators and engineers, but I have also been somewhat surprised that a lot of people never look into this. So, I started looking into how finding the water might be done, and the effects on increasing quantities of water in a system. Then a refrigeration technician and I did some testing.

I have, for a long, long time, heard water in an ammonia system isn’t good. How come? I did some digging to find out the effect of excess water in an ammonia system. One of the helpful sources is IIAR Bulletin 108 “Guidelines for: Water Contamination in Ammonia Refrigeration Systems”. The first sentence in the preface of the bulletin was an eye opener “Water contamination of the ammonia refrigerant is common in many refrigerating systems.”

What is used in industrial refrigeration systems is not just ammonia but “anhydrous ammonia,” meaning free of water. What is used is 99.95% ammonia, with the other 0.05% being other contaminates. One of those contaminates is water. ANSI/ IIAR Standard 2, in chapter 5, Table 5.2.2 Purity Requirements, shows that water content can be a minimum of 50 ppm to a maximum of 5,000 ppm. Since ammonia and water have a great affinity to each other, it is possible to, over time, increase that percent of water over the 5,000 ppm maximum.

As the percent of water in a system increases, it dilutes the ammonia, which changes the pressure-temperature relationship. From Bulletin 108, “At a given pressure the saturated temperature for anhydrous ammonia will be lower than the saturated temperature for an aqueous solution. As the aqueous solution becomes more dilute (water content is increased), the saturate temperature becomes higher.”

The increasing percent of water in the system affects just about everything in the system. The bottom line is, as the water percent increases, the system has to work harder to maintain the desired refrigeration effect, which means more energy is used.

So, where am I going find the water in a system? Bulletin 108 gives some likely locations, such as: an LPR dropleg or pump discharge line; a Control Pressure Receiver (CPR) (transfer line between the transfer vessel and the CPR); a surge drum dropleg (oil drain valve). In a DX system it might be the suction accumulator. For additional insight read Bruce Nelson’s IIAR 2010 paper “Thermodynamic Effects of Water in Ammonia Evaporator Performance”.

Where is the water? Basically, the places in a system with the lowest pressure, which result in the largest difference between the vapor pressure of water and ammonia. So, now we know where to look, how do we do it?

You don’t just drain a sample into a pop can and see what’s left once most of the ammonia evaporates. You need to be more precise. Bulletin 108 gives a list for an apparatus. You can also start by purchasing a graduated flask, such as from H.A. Phillips. The lesson we learned: it is not as simple as it looks, or sounds.

I sketched up an assembly based on what I had learned from various sources, and purchased a few simple parts from a chemical supplier on the web. Then I gave my sketch to the shop to build. Several suggested modifications later I saw the end result, and it took me a few seconds before I realized, “Hmmmm, I guess that’s what I was thinking of.”

The new sampling apparatus had a means to hold the flask, plus the addition of a vertical pipe that could be filled with water. The thought was that the vent from the hose used to fill the flask could be bled into the water. Also, after the sample is taken, a person may want to vent the hose into water instead of releasing it to atmosphere, depending on where you are. There was also a couple of ¼” valves to direct flow either to the flask or vent.

You might be wondering . . . why the hose? Why not open a valve at the sample point and drain into the flask? We learned two things. One, there are many places where it is very challenging to drain directly into a flask. So, having a hose allows you to get into a better, and potentially safer location. Two, as we found out, trying to control the sample flow with a typical drain valve (1/2”, ¾”, whatever) was very challenging. We found it best to have the ¼” stainless steel braided hose connected to the sample point, and a ¼” valve to control flow.

Before heading out to take samples I wrote up my best guess for a sampling procedure. One reason was that neither I, nor anyone else, had ever done an ammonia/water sampling test, and two, I wanted to pre-plan how to safely do the entire process and the PPE precautions that should be taken.

should be taken. With apparatus in hand, well both hands, since the thing was almost tank proof except for the flask, we headed to a facility to do a test. The first machine room had a mechanical pumped recirculator, which should be a good sampling location. The recirculator had an oil pot, which we initially thought might be a good point to take a sample.

Another lesson learned: oil pots, droplegs, etc. are designed to retain oil. There should be very little or no oil in your sample, since the oil will take up volume in the sample, resulting in an inaccurate test. If you decide to take a sample from one of these locations, be prepared to spend time draining the oil so you can get a clean sample. After experimenting for a while we realized we needed to sample someplace else or this process was going to take us a lot longer

Because this was a pump recirculator, there was an access valve with a pressure gauge on the outlet of the pump. Since the pump(s) were connected to the lower section of the LPR dropleg, this should be a good sample location. We connected the hose to the pump discharge gauge valve. Then before we started to take a sample we verified the machine room exhaust fan was operating, and also opened the roll-up access door into the machine room to allow good air-flow.

With the sample hose connected, water in the apparatus vent pipe, and the flask securely held in place, we were ready. The sample was to be 100 ml. That was another lesson learned. It is tough to get just 100 ml. The lower the pressure at the sample point, the easier it is to control the flow rate. On the first sample test, we overshot the 100 ml line by a lot, because we underestimated the flow rate, the vaporization that occurs from everything being cooled by the ammonia, and how much residual ammonia would be in the sample hose. On the third sample, we got very close to 100 ml’s.

Now the slow part of the process begins. The sample needs to evaporate till all that is left is the ammonia/water mixture, and possibly (likely) some residual oil. Bulletin 108 suggested to have an 80°F to 90°F water bath to use as a heat source for the flask sample. The flask should not be heated quickly (accelerating vaporization) since this could result in some of the sample leaving the flask in the liquid state, which would offset the test results. We didn’t use a water bath since the outside temperature was in the low 70’s. However, if you have ever waited and watched a 100 ml sample evaporate, it is a very slow process. About as much fun as watching paint dry. It took over 45 minutes in our case, and we periodically heated the flask with our hands to add a small amount of heat and remove the insulating frost layer on the flask that kept building up.

When the sample was done evaporating I calculated the approximate water content based on the amount left from the 100 ml sample. In this first test, the water content was about 0.25%, which meant the system was pretty dry. We did a second test in another machine room, only this time from a Control Pressure Receiver. Learning more from this test too since the supply point pressure was 90 psig, which changed how much and how long the sample control valve is opened. However, after two tries we had a good sample, and the calculation showed this system was also pretty dry, about 0.34%.

Finding out how much water is in a system would be important information, and can help in deciding what might be done to get the water content percentage into the appropriate range. Doing the sampling process takes time, patience, and the use of appropriate PPE. The process is described in Bulletin 108 and other published literature. The key takeaway is that you will be learning some lessons on what works and what does not and realizing that just about every sample point is going to be a little different in what happens. Stay safe.