Almost all industrial ammonia systems have water inside —typically 2 to 6% by weight— and almost nobody sees it as a problem until evaporators perform below catalog specs. Here is what water does to an evaporator, why the damage spikes when you reduce overfeed or switch to dry expansion, and what can be done about it.
There is an uncomfortable fact behind many ammonia plants that "can't quite cool as they should": a small amount of water dissolved in the refrigerant. That water always ends up concentrating on the low-pressure side of the system, right where the evaporators are. For years it was ignored because in pump-recirculated systems its effect is small. But the industry is moving toward lower overfeed rates and toward dry expansion (DX), and there the water stops being a detail.
At Thermomac we are IIAR members and closely follow this type of technical work. This post is based on the paper "Thermodynamic Effects of Water in Ammonia on Evaporator Performance" (Bruce I. Nelson, P.E., Colmac Coil — IIAR Technical Paper #8, 2010).
Why DX and Low Overfeed Are So Attractive
First, the context of why this matters more and more. Operating evaporators at reduced overfeed rates (n less than 2), up to dry expansion, brings concrete advantages: liquid pumping power is reduced or eliminated, suction lines stay dry or nearly dry, liquid and suction diameters shrink, initial cost drops, and — above all — the ammonia charge of the evaporators is drastically reduced (up to 30 times less in DX vs. pumped ammonia). In times when reducing the ammonia inventory is a priority, the temptation is enormous. The problem is that this same change makes water weigh much more.
What Water Does to Ammonia: A Zeotropic Mixture
The key is to understand that ammonia with some water behaves like a binary zeotropic refrigerant mixture, just like the blends of the R4XX family (R410A, R404A, etc.). These mixtures have "glide": the boiling temperature changes as they evaporate.
Why? Because ammonia is far more volatile than water. Throughout the evaporator circuit, ammonia evaporates first and leaves behind a liquid that becomes increasingly water-rich. That remaining liquid has a higher bubble point, which keeps rising until the evaporator outlet. It is, literally, a distillation process occurring inside your coil.
In pumped systems with high overfeed, that bubble-point shift is small (barely 1 to 2 °F), which is why evaporator manufacturers and designers usually neglect it. But when you reduce the overfeed, the liquid fraction remaining at the end of the circuit becomes increasingly water-concentrated, and the shift becomes large.
In Dry Expansion, the Valve Is "Fooled"
Here is the heart of the problem. In a DX evaporator, the expansion valve controls flow by measuring superheat at the outlet: when it sees a temperature above the saturation temperature, it assumes the refrigerant has fully evaporated.
But when water is present, that rise in the bubble point looks like superheat without being superheat. The valve interprets this as superheated gas and lets a liquid ammonia-water solution escape to the suction line. In other words: when water is present, it becomes impossible to truly superheat the refrigerant at the outlet of a DX circuit. There will always be some liquid coming out.
The paper gives an example that is alarming in its specificity: a DX evaporator operating at 18.3 psia (1.3 bar), with ammonia containing 5% water and a superheat setting of 10 °F. When the bubble point shifted those 10 °F, the vapor quality is 0.80. That is, 20% of the mass exiting the evaporator is liquid, and that liquid has a water concentration of around 25%. Sufficient suction accumulator capacity must be ensured to capture and handle that liquid before it reaches the compressor and damages it.
The Performance Cost: Worse Than You Think
The paper's calculation model compares two typical cases: constant refrigerant flow (ammonia pumped with a fixed-displacement pump) and constant outlet temperature (DX with superheat expansion valve). The conclusions are clear and worth quoting:
- The performance penalty due to the bubble point shift is significantly greater than what one would estimate with a simple calculation of the initial water concentration (the "ideal capacity ratio").
- The penalty grows as overfeed is reduced (n less than 2), and is "masked" when overfeed is high (n greater than 2).
- It is always worse than predicted by the ideal calculation when n is less than 1.5, or when the initial water concentration exceeds 1%. In the author’s own words: «it is worse than you thought.»
- The penalty also increases with operating pressure (higher suction temperature).
Practical rule: water is a problem that grows exactly when you want to modernize. While you overfeed heavily, it forgives you; when you drop to DX or low overfeed to reduce ammonia charge, it charges you with interest.
What to Do: Capture and Remove the Water
The central recommendation of the paper is direct: always plan for means to capture and remove water from the ammonia system, especially in DX operation and with low overfeed (n less than 2). Commercial water removal devices are available that can be installed in new systems or adapted to existing ones.
Now, removing water has its physical limits, and it is worth knowing them. These devices concentrate the water by heating the solution, but the ammonia-water phase equilibrium diagram shows that by simple heating, only a maximum practical water fraction of approximately 0.75 to 0.85. At typical freezer pressures (10.4 to 18.3 psia), heating the captured solution much beyond 120 °F (49 °C) starts to generate water vapor that, if vented back to the suction side, recombines with the ammonia: you gained nothing.
And there remains a point that is not about process but about disposal: the concentrated ammonia-water solution has a high pH (very little ammonia is needed to exceed 10.0). Diluting it to an acceptable pH for drain disposal may require an impractical volume of water, so it often must be chemically neutralized before disposal, in compliance with local environmental regulations.
In Summary
Water in ammonia is not a new problem, but it becomes critical exactly in the direction the industry is heading: less refrigerant charge, low overfeed, dry expansion. Ignoring it means evaporators that perform below design, expansion valves that pass liquid, and compressors exposed to liquid slugging. The good news is that it can be measured, removed, and managed. The condition is to stop treating it as a negligible detail and to design the system — from day one or in the retrofit — to remove it.
From Thermomac
As IIAR members, at Thermomac we apply these criteria when designing, evaluating, and retrofitting ammonia systems. If you are thinking about reducing your ammonia charge by switching to low overfeed or DX, the water content of your system stops being a secondary data point: it becomes a design variable. We can measure the water concentration, evaluate the impact on your evaporators, and define the appropriate removal scheme.
Want to know how much water your system has and how much performance it's costing you? Contact us and we'll analyze it.
Technical source: Bruce I. Nelson, P.E. (Colmac Coil Manufacturing, Inc.), "Thermodynamic Effects of Water in Ammonia on Evaporator Performance", IIAR Technical Paper #8, 2010 IIAR Industrial Refrigeration Conference & Exhibition, San Diego, California.
