Contact me


A website dedicated to advancing the art and science of ammonia refrigeration

About DTS

PSM audits

How does Ammonia Refrigeration Work?

How does Ammonia Refrigeration Differs?


Contact me


Ammonia Refrigeration How It Differs


Of the number of visitors to this website, the most frequently asked question is: "How does ammonia refrigeration work?"  Yes, it is different from refrigeration and air-conditioning using halocarbon refrigerants.  But there are a lot of similarities also.  So lets start with some basics a short review of vapor-compression refrigeration, native to all refrigerants.  Im going to leave out any discussion of ammonia/water absorption (heat-driven) cycles and limit this discussion to cycles involving vapor-compression.


The Bare Bones Basics

Every vapor-compression refrigeration system or unit ever built will have at least one each of these four components:

  • compressor
  • condenser
  • expansion device
  • evaporator

Figure 1 illustrates these components and their relative placement with one another.

Figure 1 Basic Vapor-Compression Refrigeration Cycle

The line numbers denote:

  1. Hot gas (high pressure, high temperature)
  2. Liquid (high pressure, warm temperature)
  3. Liquid + vapor (low pressure, cold temperature)
  4. Vapor (low pressure, cold temperature + ~10 F superheat)

Note the horizontal dashed line in Figure 1.  All portions of the system above this line are part of the system high side those components operating under a high(er) pressure than the pressure within the system low side.  As the absolute pressure of a gas increases, its temperature increases, therefore the system high side is usually hot or at least warm to the touch.  Everything residing below the dashed line operates under a low(er) pressure than the high side.  The pressure difference is a function of the temperatures involved in the process and the refrigerant selected.

Starting at the discharge connection of the compressor, line 1 conveys a high pressure superheated hot gas where it enters a heat exchanger (the condenser).  After entering, the gas is first desuperheated.  Upon reaching its saturation temperature, the vapor then begins to condense, changing from a vapor state back into a liquid state.  If additional heat is removed from this liquid stream, the process is known as "subcooling"[1].

Line 2 conveys the high pressure refrigerant liquid stream from the condenser into an expansion device.  There are many different types of expansion devices; in the following short list, Ive divided these into modulating devices and fixed devices.

  1. Capillary tubes (fixed)
  2. Orifices (fixed, short orifice, Accurator)
  3. Electronic expansion valves (modulating, senses refrigerant temperature rise across evaporator)
  4. Thermal expansion valves (modulating, senses refrigerant superheat generated within evaporator)
  5. Hand expansion valves (fixed but manually readjustable)
  6. High-side floats (modulating, senses liquid level)
  7. Liquid control valves (usually positioned by a remote high side float)

Expansion devices numbered 1 through 4 are commonly applied in halocarbon refrigeration systems.  Devices numbered 4 though 7 are used as throttling devices in industrial refrigeration systems and practices.

After leaving the expansion device, the refrigerant has now become a mixture of low pressure cold liquid and vapor as it travels down line 3.  In most cases, especially with halocarbon units, this line is very short maybe an inch or two long.  This mixture then enters an evaporator where the remaining liquid is boiled away while transferring heat energy across the evaporator tubing (and fins if they exist).  If the expansion device measures refrigerant superheat (or a temperature rise) occurring within the evaporator, the gaseous refrigerant is then superheated slightly before it leaves the evaporator and enters line 4.

Line 4 (also known as suction) conveys the now slightly superheated low pressure vapor back into the compressor where its pressure and temperature are simultaneously raised to a level where heat can be rejected from the condenser into a heat sink (air, water).

When looking at Figure 1, this energy balance becomes apparent:


As my learned colleagues at the University of Wisconsin remind me: "Denkmann, the system has to balance.  The sum of the gozoutas minus the sum of the gozintas has to be equal to zero".  All fine and well, Id say.  The bloody thing has to balance.

So far, everything Ive presented up until now is generic among all refrigeration systems and units.  This includes ammonia as well.  Where things become interesting are the differences between Commercial Refrigeration as commonly practiced (all DX) and Industrial Refrigeration and its practices (DX + liquid overfeed + thermosiphon).  These two fields of practice grew up separately, rarely if ever speaking with one another at ASHRAE meetings.  Consider the following Equations 2 and 3.  The system energy balance is simple and straightforward as weve already seen (Eq 1).  However, when it comes to mass flows, things get a little murky. 

All refrigeration systems (packaged units as well) can be grouped into either of two categories described by one of the following:


describes all direct expansion systems[2], and


describes all remaining liquid overfeed and gravity-flooded systems.

The three equations presented so far describe every vapor compression refrigeration system built, and regardless of the refrigerant charged into the system, where m is mass flow. 

Now lets look at the ways ammonia (industrial) refrigeration systems differ from their commercial halocarbon counterparts.


Ammonia Refrigeration Systems Ways They Differ

Probably the number one difference between a typical DX unit and an ammonia refrigeration system centers on this: oil.   In a DX halocarbon unit, oil is continuously returned to the compressor.  Oil is not returned to the compressor in ammonia refrigeration systems but is instead drained out of the system periodically.  Oils used in ammonia refrigeration systems are essentially insoluble in NH3 (some slight solubility exists at high pressures); oil is heavier than liquid ammonia, making it easy to drain out.  On the other hand, oil solubility is absolutely essential with the halocarbons in order to facilitate oil return.  This makes oil management in an ammonia system a relative piece of cake.  Its easy to manage just drain it out.  We dont have to maintain minimum gas velocities in order to bring it back through dry suction piping.  Paraffinic-based oils are commonly used with ammonia.  These oils do a good job of cleaning out old welding slag and dirt, hence another reason why it cant be returned to a compressor it is too dirty to reuse.  Oil, once drained from the system, is no longer usable.

The next difference lies in the choice of piping materials.  Copper, brass, bronze cannot be used with ammonia your metal choices are mild steels, stainless steels, nickel, but absolutely no copper.

The next difference lies in refrigerant management.  In nearly all halocarbon packaged units (air-conditioning, commercial refrigeration), the refrigerant charge is critical.  This means that the system has no alternate place(s) to store refrigerant not in use by the system at some particular point in time.  Stated differently, refrigeration units are built using the simplest designs no pressure vessels, no solenoid valves.  Any excess liquid becomes stored inside the condenser and any excess liquid decreases system refrigerating capacity.  Generally speaking, a critically-charged unit should be charged within 1-2% of the listed charge.  If it isnt, the unit will fail to produce its stated capacity.  Over-charge also increases the risk of liquid carryover to the compressor.  A few systems will have suction traps (a small vessel) to detain a liquid surge; some traps have an internal heat exchanger to facilitate liquid boil-off.

Have you ever noticed how a critically-charged system (or unit) behaves when its compressor initially starts?  A critically-charged system takes a few minutes to "get going" that is, start to produce some cooling effect.  During this interval, the compressor is evacuating refrigerant vapor from the low side and sending this now higher pressure vapor off to the condenser to be reliquified.  Then this liquid must start to back up in the high pressure liquid line so that it can seal off the opening into the liquid capillary (the expansion device).  Now your unit will start to produce cooling.  This time delay isnt acceptable with most ammonia systems, especially when process cooling or freezing is involved.  I know of only one ammonia refrigeration system designed for critical charging.  The owner of this system wasnt very happy with it either and has probably removed it by now.  And for a good reason the time delay associated with a critical charge just isnt acceptable.  Hence, another difference with ammonia the need for pressure vessels.


Pressure Vessels

Whenever someone familiar with halocarbon DX practices walks into an ammonia refrigeration engine room, their jaw usually drops open.  "You gotta be joking me, man!  What are those huge tanks for?  Are those filled up with ammonia?" 

Since the time delay associated with critically-charged systems cant be tolerated with industrial refrigeration, some means of preventing this delay, or at least shortening it becomes necessary.  That means takes the form of pressure vessels a device to store liquid ammonia.  Heres a partial list of the tasks that pressure vessels perform in ammonia refrigeration systems:

Repository of liquid not in use by the system at any point in time

Liquid operating reserve for mechanical drive refrigerant pumps (often 5 minutes or more)

Separate liquid from vapor (by gravitational forces)

Protect compressors from liquid slugging

Liquid transfer to other vessels within the system

Cool booster discharge vapor (two-stage systems with intercooler)

Collect oil for later removal

In the U.S., most ammonia pressure vessels are stamped for 250 psig service although youll occasionally see 300 psig vessels in the Desert Southwest for storing high pressure liquid.  The minimum allowable pressure (U.S.) is 150 psig for low-side vessels.  All vessels built in the U.S. must conform to ASME Section VIII requirements.


Water and Refrigeration Systems

Another way ammonia is different from its halocarbon counterparts has to do with a water inclusion.  Just about every field-erected halocarbon refrigeration system Ive ever seen or designed has at least one filter/dryer.  The desiccant used adsorbs water.  A high pressure drop across a dryer core is a sure sign that water has accumulated a liquid sight glass will probably read yellow (R22), indicating the system has taken on some water.  If this goes uncorrected, water will freeze inside distributor tubing if the evaporating temperature is
<32 F.

Ammonia is very different with its relationship to water.  They have this love affair going ammonia loves water!  As the water content in liquid ammonia increases, the evaporating temperature also increases with the evaporating pressure held constant.  But as ammonia becomes dissolved into water its freezing point temperature drops.  This makes freezing water in an ammonia mixture virtually impossible (at normal temperatures).  So distributor tubing doesnt suffer the same fate with an ammonia/water mixture as it does with the halocarbons.

However I have also seen evaporators that stopped performing because so much water had entered (through a failed tube in a water-cooled condenser) and the oil turned to a sludge.  This sludge then coated the inside of each evaporator tube on nearly all evaporators in the plant so that all that remained was a " hole in a 1" tube certainly not sufficient for a gravity evaporator.  All of them had to be scrapped.  The tell-tale sign was clearly visible only a few U" bends had any frost on them most were warm to the touch.

So the take home on NH3 + H2O comes down to this: some can be tolerated but it shouldnt be more than 0.4% by weight[3].  A refrigerant remediator (a batch-fed still) can be used to remove water from ammonia.  Dont try using a halocarbon desiccant the bond between the desiccant and water is weaker than the bond between ammonia and water.  Besides, if you try doing this, you wont like the outcome because the liquid ammonia will dissolve the desiccant, sending little bitty pieces of it down the liquid line which plugs up everything.  Been there.


Ammonia Remediators & Foul Gas Purgers

One way to remove water from ammonia is to use a batch remediator a tall, skinny vessel with a couple of float switches and usually one or more belly-band resistance heaters.  "Weak aqua" is then manually drained out.  If desired (or required), neutralize the aqua with a little citric acid before sending it down a sewer.  Ok, that takes care of getting water out.  Now how about air?

This is the purpose of a "foul gas" (non-condensable gas) purger remove collected air, trapped during servicing.  These two devices - a remediator and a purger make operating an ammonia refrigeration system low side below atmospheric pressure feasible.  It is not uncommon to see a blast freezing line run at -60 F suction, which is 18.6 inches Hg vacuum.  This is not possible with any of the HFC halocarbons even minute quantities of moisture + air + oil forms acids.


Evaporator Liquid Feed Methods

The three methods of supplying refrigerant to an evaporator are:

  1. Direct-expansion (DX)
  2. Liquid overfeed (via mechanical drive pump or controlled pressure receiver)
  3. Gravity flooded recirculation (one surge drum required for each evaporator, gravity handles all refrigerant circulation also known as "thermosiphons")

All three are commonly applied in industrial refrigeration using ammonia.  No one method is "better" than the other.  Each has its best applications.  Most systems have a mix of evaporators "A dog from every town[4]".



The major difference between the halocarbon series and ammonia with respect to compressors has to do with the motor open drive versus hermetic.  With only one exception (Japan), all ammonia compressors are open-drive design due to the incompatibility of copper and NH3.

The most commonly applied compressor design is twin rotary screw in todays industrial marketplace.   Reciprocating compressors are still applied and have an inherent advantage of better part-load efficiency over a screw at off-load (slide valve positioned).

With only one exception that I am aware of, centrifugal compressors are not used with ammonia.  The low mol weight of the NH3 molecule (17) makes this particular refrigerant a poor candidate for turbo compression which relies on a high density gas.

Another factor that distinguishes ammonia systems from their halocarbon counterparts has to do with compressor capacity unloading.  One of the most energy-wasteful practices ever to come about in refrigeration is using hot gas bypass as a means of unloading a compressor.  When the hot gas bypass solenoid opens, hot gas is sent directly back into the compressor suction.  While effective at reducing system suction cfm, it does little or nothing for a corresponding reduction in compressor energy use.  Fortunately, hot gas bypass cannot be applied to ammonia systems, unless it is injected into a side-inlet distributor and then into a DX evaporator.  Any attempt at sending compressor discharge vapor directly back to suction will either shut the machine down on high oil temperature or high discharge temperature (screw compressors) or blacken the discharge heads of a reciprocating compressor.  You may also coke up your compressor oil as well. 



Ive seen all three methods of heat rejection (to ambient) applied:

  1. Air-cooled
  2. Water-cooled
  3. Evaporative-cooled

An air-cooled condenser is seldom used with ammonia.  One system Im aware of, operating at 0 psig suction (Tsat = -28 F), requires two stages of compression in order to reject heat at 110 F condensing temperature.  However, this system seldom sees hot days and the regional water quality is low and difficult to obtain.  Other than this system, I know of no others using air-cooled heat rejection.

Evaporative cooled condensers are by far the most common, followed by water-cooled condensers (with a cooling tower).  Id say the percent ratio is ~1% with water-cooled condensers (shell & tube design).  All others (the vast majority) use evaporative condensers selected for a 95 F maximum condensing temperature.

[1] The removal of heat from a liquid at its boiling (saturation) temperature.  This is no different than a pan of water on your stove after youve turned the burner off.  The only variations are the temperatures involved.  For example, that mug of coffee youre drinking lets say its measured temperature is 120 F.  To find the amount of subcooling your coffee has, subtract its measured temperature from 212 F (boiling temperature at one standard atmosphere).  Answer: 92 F of liquid subcooling.  This subcooling temperature can then be used to find other data about handling a fluid near its boiling point, usually involving static lift or pump NPSHA.

[2] The < character in Eq 2 is necessitated by hot gas bypass as a means of compressor capacity unloading.

[3] IIAR

[4] Source: Doug Reindl

website design and promotion: Web Design Chicago