1

ASAC REFRIGERATION DIVISION

LIQUID OVERFEED OR RECIRCULATION SYSTEMS
Colin Hewetson
23/6/05
Reference Cornell Pumps/Hanson/Witt/Hermetique

This type of system was developed to overcome the shortcomings of the DX system whereby the refrigerant
quality is only approximately 60% or in simple terms only approximately 60% of the evaporator surface is fully
wetted. In such cases the heat transfer or heat flux will be low as only latent heat transfer will arise in the wetted
portion and sensible heat in the vapour section or useful superheat.
A liquid overfeed evaporator is essentially the same as in the DX system except that the means of distributing
the liquid among the circuits of the coil may be somewhat different. Overfeed implies that more liquid is fed to
the coils than will be evaporated (or boiled off), and in fact, the overfeed rate is typically about 2 to3:1.which is
the ratio of units of liquid that exit the evaporator to units of vapor that exit the evaporator. This is compared
with the recirculation rate which is the ratio of units of liquid that enter the evaporator to units of vapor that exit.
The recirculation rate is most commonly used to specify the amount of liquid to be fed. The various references
and tables are generally set up in terms of recirculation rate. By overfeeding the coils it is assured that the inner
surfaces will be thoroughly wetted and will thus have the optimum heat transfer. Overfeeding also assures that
the vapor that does exit the coil will be close to saturation, not superheated, and this will ensure lower
compressor temperatures and more efficient operation of the condenser. There is little point in overfeeding by
say 4:1 as this just increases the pressure drops & pump power.
By feeding considerably more than will be evaporated, it becomes unimportant that the feed be precisely
controlled - there is great latitude without any significant effect on the temperature control of the refrigerated
space. So, since there is liquid exiting the coil, how does the compressor operate without damage? Protecting the
compressor is the primary function of the LP surge drum or more commonly referred to as the LP receiver.
The low pressure receiver is simply a horizontal or vertical cylindrical vessel placed in the compressor suction
line between the outlet of the evaporator and inlet of the compressor. Its function is to separate the excess liquid
from the vapor, thus it is sometimes also known as a separator. The diameter and volume of the receiver must be
carefully designed to provide a) adequate reduction in gas velocity to ensure disengagement of the liquid
droplets from the vapor so that it is a true vapor that enters the compressor rather than an aerosol b) sufficient
surge volume for liquid storage is provided to store refrigerant during low loadings or coil defrost c) sufficient
static head to avoid cavitation at the vessel outlet/pump inlet.
A secondary function of the receiver is that it helps to ensure that the vapor is saturated and not superheated any superheat in the vapor tends to transfer into the adjacent liquid and boil it off , thus maintaining saturation.
There is an outlet at the top of the receiver that takes the vapor to the compressor and an outlet at the bottom of
the receiver that takes the liquid to the pump. There is an inlet, known as a wet return or wet suction which
brings the liquid and vapor from the evaporator into the receiver. There is also a line from the high pressure side,
the condenser, through which the high pressure, high temperature liquid is flashed back to low pressure, low
temperature liquid and vapor in the low pressure receiver. Ideally however the high pressure liquid should be
subcooled to near as possible to the evaporator temperature via the intercooler on a two stage system or shell &
coil flooded economizers or subcoolers on either the high and or low stage compressors. Open flash economizers
should not be used (unless the pressure head available is in excess of liquid feed device or LP receiver pressure
requirements) as the pressure (screw compressor systems) will correlate to the side load port pressure which
will be at some mid point between compressor suction & discharge pressure. Subcooling reduces the volume of
flash gas & hence compressor displacement in comparison to systems where the liquid is expanded directly from
the HP receiver at condensing pressure to the LP receiver via an expansion device. The liquid feed or make up

2
line to the LP receiver simply needs to be 2 service stop valves + solenoid valve + hand expansion valve. The LP
receiver liquid level float controller or float switch signals or energises or deenergises the liquid line solenoid
valve according to the liquid level in the LP drum. There is no need for complicated metering valve or expensive
metering valves to be used. In pumped systems the hand expansion valve is used mainly as a flow control or
regulating device it is not an expansion device as such, assuming the HP liquid has been subcooled to within
10C of the evaporating temperature. Obviously there is some expansion and flash gas generated in cooling the
HP liquid to the final evaporating temperature, but not as much as there would be if HP liquid at condensing
temperature were used.
The receiver introduces a second "loop" into the refrigeration system. One loop takes the vapor from the
evaporator to the receiver to the compressor, to the condenser, to the high pressure receiver and then back to the
receiver. The other loop takes the liquid from the evaporator to the receiver, through the pump and then back to
the evaporator.
Liquid overfeed systems require a circulating pump to transfer a continuous flow of refrigerant to the
evaporators. In the DX system (and others) the motive force for distributing refrigerant through the system is
high side pressure or pressure differential. The liquid pump removes that dependence on maintaining an
artificially high compressor discharge - the high side and low side are now "decoupled." The refrigerant pump is
sized according to the total refrigeration load that will be served by that pump and the pressure required to get
the resultant liquid flow to all the coils being served. To determine the required refrigerant flow rate it is
necessary to know the total tons of refrigeration to be served by the pump. Every heat exchange device
connected to the liquid supply will be designed to absorb a specified quantity of heat and evaporate refrigerant at
a particular temperature. Ideally, the pump will be supplying liquid to devices (evaporators, plate freezers, etc.)
all evaporating refrigerant at the same refrigerant temperature and pressure, such as a number of evaporators in a
large cold storage space. Then the flow rate will be determined from:
lbs or kgs Per Minute = Total tons (or kw) x lbs/m/ton (or g/m/kw) x recirculation rate
The lbs/m/ton can be looked up in a refrigeration reference such as the ASHRAE Refrigeration Handbook, and it
is a function of temperature. If such tables are unavailable then the flow rate can be determined by calculating
the refrigeration effect or more simply taking the change in enthalpy of the liquid & vapour at the liquid supply
temperature & dividing the cooling load in Btu/m by the change in enthalpy to give you a lb/m/ton flow rate. For
example the change in enthalpy for R717 assuming the evaporators are fed from LP liquid @ -20C & the
enthalpy change between vapour & liquid @ -20C is 613btu/lb 48.3btu/lb = 564.7Btu/lb. If the load is 100TR
or 1200000btu/hr the flow rate is 1200000/60[/564.7] = 35.4lb/m or 0.35lb/m/TR. There is no significant heat
flux increase in overfeeding by more than 3:1 in fact 2:1 is more than enough for most applications and even
then this provides only a slight improvement over flooded operation. So to select the pump flow multiply the
theoretical 35.4lb/m x 2 or 3 to give you the total flow. Additionally you will need to determine the pump head,
this calculation is based on the total equivalent length of piping from the pump discharge back to the LP receiver
which normally would be based on 1psi/100' pressure drop + evaporator pressure drop; on average this would
work out at 15-20 psi. Additionally you have to add the static lift which is the difference between the pump and
the highest evaporator in the circuit. As 1psi = 4' for R717 if we assume an average static height of 10mtr or 33'
this equates to 8.25psi so total head would be say 20 + 8.25= 28.25psi.
If pipe runs are longer than about 300 feet then the pipe friction should be calculated. The pipe friction can be
determined from tables issued to you. Sizing & selecting a refrigerant pump is no different from selecting a
water pump the same principles apply.
The most significant exception to the process outlined above for determining pressure requirement is
brought about by the presence in the system of one or more back pressure regulators. These are control
valves, placed at the outlet of an evaporator, whose purpose is to hold the evaporator at a higher pressure
than the saturation pressure corresponding to the temperature of the liquid refrigerant supply. Holding

3
the pressure at this higher value causes the refrigerant to evaporate at a higher temperature, and this
enables one pump recirculation package to handle more than one temperature zone. For example a pump
and receiver package may supply -40F liquid to a freezer room to maintain that room at -30F and also
supply -40F liquid to a loading dock whose evaporators are back pressure controlled to cause that liquid
to boil off at +25F. In such a system you still apply the same process to determine the required pressure
except that now you must add an additional pressure equal to the difference between the liquids
saturation pressure and the pressure to which the coil is being held. As an example, a system is employing
10F liquid which has a saturation pressure of 24 psi, but one coil is being back pressure regulated so that
the liquid will boil off at 30F instead. The saturation pressure at 30F is 45 psi, so the pump must be sized
to provide an additional 21 psi (45 - 24) above what the system calculations indicate.
With the introduction into the system of the LPR (low pressure receiver) the coils can now be operated flooded
for maximum efficiency. The addition of the pump removes the dependence on high side pressure as the means
of circulating the refrigerant, so now head pressure can be allowed to "float" down with the ambient temperature
to take advantage of cooler weather.
Pump Leg Line Velocities and Characteristics
The pump leg should be sized for an optimum line velocity of 3 Ft. Per Second. A 3 Ft. Per Second line velocity
represents the most effective compromise between the heat capacity of the line and excess friction loss. Pump
leg over sizing should be limited to one pipe size, however the pump leg should never be smaller in diameter
than the pump suction size. In addition, the pump legs should be kept as short as possible while allowing 2 to 3
pipe diameters between the suction stop valve and the pump suction flange.
Bypass Line Characteristics
Recirculation is a phenomenon common to all centrifugal pumps when operated at low capacity. At one half to
one third of the best efficiency point, a secondary flow begins within the impeller whereby the fluid actually
reverses direction and exits the eye and/or enters the discharge. This results in turbulence and small vortices. The
high velocity at the core of the vortices results in low pressure, often below the vapor pressure of the fluid, and
cavitation may ensue. The bypass line is a means to safeguard against insufficient flow. The bypass line is
connected between the pump discharge flange and the stop valve to a wet return back to the vessel. The bypass
line should be at least 3/4 in. and should include a needle or globe valve. A relief valve should not be utilized
within a bypass line. Due to the relative flatness of a centrifugal pump curve, the pressure difference between
sufficient and insufficient flow may not be enough to assure that a relief valve will open or close properly.
Remember that the bypass flow should be added to the system flow and the total used to specify the total pump
flow.The minimum flow requirement for a given refrigeration pump can be established by the following
procedure; (1) Fully open the bypass line valve, (2) Close the discharge stop valve, (3) Slowly close the bypass
line valve until the pump discharge pressure starts to become unstable as indicated by "bouncing" of the pressure
gauge needle, and (4) Open the bypass line valve until the gauge stabilizes. Once the bypass line has been
adjusted for minimum flow requirements, the line should be left alone.
Refrigerant Pump Capacity Characteristics
The following matrix highlights the minimum and maximum flow characteristics for refrigerant pumps
Model Mini Flow Max Flow
1.5CLB 12 USGPM 55 USGPM
1.5CB 13 USGPM 88 USGPM
1.5CBH 10 USGPM 116 USGPM
2CBS 12 USGPM 54 USGPM

4
2CB
3CB

50 USGPM 156 USGPM

50 USGPM 400 USGPM

*Maximum flow limits are established at 3 Ft. NPSHR

Operation below the minimum flow threshold may create insufficient flow through the pump which may result
in heating and boiling of the liquid or in cavitation due to "recirculation". Recirculation is a phenomenon
common to all centrifugal pumps when operated at low capacity. At one half to one third of the best efficiency
point, a secondary flow begins within the impeller whereby the fluid actually reverses direction and exits the eye
and/or enters the discharge. This results in turbulence and small vortices. The high velocity at the core of the
vortices results in low pressure, often below the vapor pressure of the fluid, and cavitation may ensue.
The bypass line is an effective means to safeguard against insufficient flow. Consequently, bypass capacity
should be include within the capacity requirement to enable the end user to achieve the minimum flow
requirement for a given refrigeration pump.
Operation above the maximum flow threshold may initiate cavitation. Net Positive Suction Head Available
(NPSHA), is a measure of the total energy available at the suction of the pump, above the vapor pressure head,
to suppress boiling. Whereas, Net Positive Suction Head Required (NPSHR) is a function of pump design. The
liquid velocity increases and the pressure decreases as the liquid passes over the leading edges of the impeller
vanes.
The NPSHR is the positive head in feet absolute required at the pump suction to overcome these pressure drops
in the pump and maintain the liquid above it's vapor pressure. The NPSHR varies with the pump design,
capacity, and operating speed. These NPSHR values are stated in terms of feet of head on the manufacturer's
performance curves.
When the NPSHA in the system has been determined, it is compared to the NPSHR as published on the liquid
overfeed pump performance curve. The Net Positive Head Available (NPSHA) must be greater than the NPSHR,
otherwise cavitation will occur. It is recommended that NPSHA is equal to or greater than NSPHR + 2 feet.
Drop Leg Sizing and Characteristics
As an ideal target the drop leg diameter should be sized for a velocity of approximately 1/4 Ft. per second at the
system design flow rate. Using Stoke's settling laws it can be calculated that about 1/4 foot per second is about
the optimum velocity for the drop leg flow in order to allow entrained bubbles to rise at a higher rate than the
downward liquid velocity, preventing the bubbles from entering the pump suction.
The drop leg mouth should contain a crossed vortex eliminator and/or horizontal baffle plate. In addition, the
absolute minimum submergence of the drop leg opening below the vessel minimum liquid level should be 12
inches.
Finally, the drop leg should extend below the pump leg entrance. This assumes the conventional horizontal pump
leg connected to a vertical drop leg. Where the drop leg is omitted in favor of a pump leg with a long radius
elbow connected directly to the vessel, the pump leg should be connected off the vessel vertical centerline so oil
can be collected from a tap off the bottom vertical centerline of the vessel.
Bypass Line Characteristics
Recirculation is a phenomenon common to all centrifugal pumps when operated at low capacity. At one half to
one third of the best efficiency point, a secondary flow begins within the impeller whereby the fluid actually

5
reverses direction and exits the eye and/or enters the discharge. This results in turbulence and small vortices. The
high velocity at the core of the vortices results in low pressure, often below the vapour pressure of the fluid, and
cavitation may ensue. The bypass line is a means to safeguard against insufficient flow. The bypass line is
connected between the pump discharge flange and the stop valve to a wet return back to the vessel. The bypass
line should be at least 3/4 in. and should include a needle or globe valve. A relief valve should not be utilized
within a bypass line. Due to the relative flatness of a centrifugal pump curve, the pressure difference between
sufficient and insufficient flow may not be enough to assure that a relief valve will open or close properly.
Remember that the bypass flow should be added to the system flow and the total used to specify the total pump
flow.
The minimum flow requirement for a given refrigeration pump can be established by the following procedure;
(1) Fully open the bypass line valve, (2) Close the discharge stop valve, (3) Slowly close the bypass line valve
until the pump discharge pressure starts to become unstable as indicated by "bouncing" of the pressure gauge
needle, and (4) Open the bypass line valve until the gauge stabilizes. Once the bypass line has been adjusted for
minimum flow requirements, the line should be left alone.
Vapor Entrainment
Centrifugal pumps for refrigerant recirculation in liquid overfeed systems are very simple devices, they employ
essentially one moving part. Consequently, problems from the pump's perspective are also simple. When too
much flow is demanded of the pump by the system, cavitation will result. Conversely, insufficient flow through
the pump may result in heating and boiling of the pumpage or in cavitation due to "recirculation". These are
indicative of an improperly sized pump or an improperly adjusted system. A third simple characteristic of
centrifugal pumps is that while they are very efficient liquid movers, they do a very poor job of pumping gasses
or multi-phase fluids. Vapor entrained in the liquid stream to the pump suction in large enough volume does not
simply pass through the pump. A popular analogy is that of transporting helium filled balloons in your car. When
you accelerate away from a stop, the balloons actually move forward toward the windshield. This is because the
air surrounding the balloons is heavier than the helium; when you accelerate, the heavier air is forced toward the
back of the car and displaces the lighter helium forward. In a similar fashion, the heavier liquid component of
the pumpage is thrown outward by the centripetal action of the rotating impeller, and the lighter vapor is
"centrifuged" toward the center of rotation. This creates a vapor blockage at the eye of the impeller which
reduces or prevents the passage of the liquid - the pump is "vapor bound." The phrase "vapor entrainment"
implies that the vapor is carried into the pump from an external source. This is as distinct from cavitation where
the vapor is actually generated within the pump. However, where the vapor content is not sufficient to stop the
flow through the pump, it may actually result in noise which sounds exactly like cavitation. Nevertheless, it is
not directly related to or affected by the pump's Net Positive Suction Head Required (NPSHR) or the system's
Net Positive Suction Head Available (NPSHA). How do you tell the difference? Diagnosis begins with an
evaluation of how likely vapor is to be the problem.
A horizontal vessel is far more likely to contribute to vapor problems than a vertical. The distance between the
operating level and outlet of the vessel (the mouth of the drop leg) is always smaller with a horizontal. This lack
of submergence means that the pump suction line is closer to the vapor phase in the vessel so that vortexing is
more likely. More important, the pump suction is closer to any surface boiling that occurs, and this increases the
likelihood that vapor will be drawn into the suction line.
Where process loads predominate vs. cold storage loads, there are going to be more load variations and resulting
pressure transients in the vessel. The "instabilities" will result in more boiling within the vessel and simply
increase the likelihood that the pump will ingest vapor.
The preceding two system design/application characteristics individually or in unison make vapor entrainment a
very likely cause of problems, and there are simple methods available to verify this. At the onset of a vapor

6
entrainment event, the pump discharge pressure will begin to fall off very noticeably. When this begins to occur,
it will be necessary to quickly start closing the pump discharge stop valve and observe the behavior of the pump.
If the pump is suffering from cavitation due to too high a flow rate, the closing down of the stop valve will bring
about almost immediate recovery. However, if the problem is due to vapor entrainment, the pump will stumble
even more quickly and will likely lose prime. This is because the liquid velocities inside the pump impeller
become inadequate to sweep vapor through, and more of the vapor becomes "stuck" in the impeller eye. Pump
performance that is faltering due to low flow recirculation will also worsen as the stop valve is closed down.
However, if the pump is connected to an adequately sized and adjusted bypass line and/or there are system loads
on line (not all liquid solenoids closed), then recirculation is so unlikely as to be relegated to the end of the check
list. This simple test using the discharge stop valve will most likely reveal whether or not there is a vapor
problem but not what its cause is. Unfortunately, the causes are numerous and often interrelated.
Vessel and piping design is always involved no matter what is deemed to be the greatest contributing factor. As
previously mentioned, horizontal vessels always have less submergence available to separate the surface boiling
from the mouth of the drop leg. Simple vortexing is not often a problem anymore as most vessel manufacturers
have learned to include crossed plate vortex eliminators or similar devices in the mouth of the drop leg.
However, vapor entrainment due simply to the proximity of the boiling layer to the mouth will still occur if the
submergence is less than about 18 inches. If any pressure drop in the vessel is quicker than about 1 psi/min.,
boiling will occur about 3 feet below the liquid surface, and then obviously 18 inches of submergence will not
prevent vapor from entering the drop leg. In a vertical vessel it would be unusual if there were not more than 3
feet of submergence. However, if the drop leg is properly sized, then the full liquid height from the operating
level to the pump level can be used to protect against vapor entrainment rather than just the submergence above
the mouth of the drop leg. If the drop leg is sized for a liquid velocity of about 25 ft./min., the bubbles will rise at
a rate faster than the downward liquid velocity. In many cases a drop leg of larger diameter than the pump
suction size is not used; rather, the pump suction pipe is simply connected directly to the vessel, and the
horizontal to vertical transition is made with a long radius elbow. In such designs, the likelihood of vapor
entrainment is greatly increased by the higher velocity in the smaller diameter line. Often the drop or pump leg is
made reentrant to the vessel projecting through the vessel wall rather than flush with it. This is done to prevent
oil entrance to the pump. However, if the projection is made too long, some of the submergence is lost.
Along with the vessel design, all but a few vapor entrainment problems are also directly related to the rate of
pressure drop during any transient in the vessel. This leads to the discussion of false loads. The key to avoiding
vapor entrainment in a vessel is to keep the rate of pressure drop as low as possible during any pressure
reduction. Pressure reductions occur when a system is started up from ambient - the temperature and pressure
must be brought down to their design values before they stabilize. Pressure reductions also occur when
something upsets the system, such as an increased refrigeration demand caused by a new batch of warm product
being brought to a freezer or increased shipping and receiving activity which allows more warm air into the
refrigerated space.
A false load is not directly related to an actual increase in refrigeration required but is due to the way the system
functions. For example, when the liquid level in the LPR drops and the float switch causes the makeup solenoid
to open, high pressure, high temperature refrigerant flashes into the LPR. This make-up refrigerant is as much as
30% vapor by weight, and this vapor is simply flash gas, it is not the result of any refrigeration load. In order to
maintain the pressure in the LPR at saturation pressure the compressor(s) will load up to a higher percentage of
capacity in order to remove the excess gas from the receiver. When the LPR liquid level is restored the float
switch will close the solenoid, and the flash gas supply will be instantly removed. However, the compressor
cannot unload instantly, so it will continue to operate for a time, fully loaded but drawing on a reduced gas
supply. As a result the LPR pressure will drop until the compressor can unload. If this pressure drop is greater
than about 1 psi/minute the liquid in the LPR will boil about 3 feet below the liquid surface, the pump will draw
in the vapor and will lose prime until the vapor blockage "burps" through.

7
Another source of false load occurs as a result of hot gas defrosting practices. During hot gas defrost of an
evaporator the liquid supply solenoid to that evaporator closes, the defrost regulator on the evaporator outlet
closes, and a hot gas supply solenoid opens. Hot gas from the high side enters the evaporator and warms the
coils. After a while the pressure inside the evaporator builds high enough that the defrost regulator opens, and
now hot gas starts blowing down the wet return line to the LPR. This, again, is a false load, and the compressors
respond just as with the make-up cycle. Sometimes there are ways to control false loads more closely, and
sometimes compressors can be loaded more slowly, but false loading is a common source of pump grief. This
grief is more from the standpoint of the operator than from the pump, because our pump will run vapor locked
without a problem. If the vapor lock lasts too long, though, then the plant begins to have trouble maintaining
temperature.