Aircon refrigerant distributor: why a coil cools unevenly

A refrigerant distributor is a small brass fitting with one inlet and several outlets. The component maker Sporlan defines it as a device connected to the outlet of a thermostatic expansion valve. Its outlet end is machined to accept tubing, and that tubing runs to each circuit of the evaporator coil.

Its stated function fits in one line. The distributor equally distributes refrigerant flow from the expansion valve into each circuit of a multi-circuit evaporator coil.

Not every indoor coil carries one. A coil with a single refrigerant path has nothing to divide. Daikin Applied ties the two together in its evaporator coil manual, where single or multiple distributors are supplied depending on the number of circuits required. Its interlaced coil types use two distributors, each feeding every other tube in the first row.

The parts are few. There is a body, a nozzle inside it, a retainer ring, and the tubes. On most models the nozzle lifts out. Daikin's installation steps have the technician inspect the distributor and confirm the nozzle is in place, held by a retaining ring or formed into the body itself.

Where it sits is worth fixing in mind. A multi-split system can also carry a branch box, and the two are different machines doing different jobs. The branch box sits in the pipe run, upstream of any indoor unit. The distributor sits at the coil, directly downstream of the expansion valve that meters it.

Downstream of the valve is a precise position. The valve decides how much refrigerant enters. The distributor decides where it goes once inside. Confusing the two sends a technician back to the metering device while the split goes unexamined.

The refrigerant arriving at the distributor is already part vapour. Some of the liquid flashes as it passes the expansion valve. What reaches the fitting is a two-phase mixture of liquid and vapour, and the two do not behave alike.

Lopsided proportions decide the whole problem. Sporlan works a case for R-22 with liquid entering the valve at 110°F and a 45°F evaporating temperature. Liquid is 77% of the flow by weight but only 7% by volume. Vapour is the rest, at 23% by weight and 93% by volume.

Nearly all the cooling is carried by the liquid. Nearly all the volume is vapour. A split that divides the volume evenly can still hand one circuit most of the liquid.

Speed makes it worse. Liquid and vapour move at different velocities, an effect the bulletin calls slip, and it gives gravity as the reason. Gravity has a greater influence on the liquid portion of the flow.

A plain header therefore fails, and it fails predictably. Sporlan's account is that the lower circuits of the evaporator invariably receive the most liquid. That can set the valve hunting and send liquid back towards the compressor. The upper circuits are starved at the same moment, reducing the effective evaporator surface.

So the design problem is not a better-shaped header. It is stopping the two phases separating at all. The bulletin sets out two requirements. Mix the liquid and vapour portions of the flow, then hold that mixture homogeneous until equal portions have been divided into each circuit.

The nozzle does both jobs with speed. It increases the velocity of the two-phase flow, which mixes the liquid and vapour, and it aims that flow at a dispersion cone. The cone divides the mixture into passageways spaced evenly around it. High velocity, Sporlan writes, is the key to the distributor's success.

Pressure drop is the price of that velocity, and it is deliberate. Drop across the nozzle focuses the flow and provides the mixing. Drop across the tubes helps balance flow entering each passageway. The bulletin adds that this drop does not reduce system capacity.

Tube length is specified, and the reason is resistance. For each circuit to offer equal resistance, Sporlan says, distributor tube lengths must be the same. The tubes also have to be bent with care, because sharp bends and kinks reduce the cross-sectional area and raise resistance to flow.

The second condition sits outside the fitting entirely. Since the distributor disperses equal amounts to each circuit, the bulletin calls it essential that the heat load on all circuits of the coil be the same. Where it is not, the effect is the same as poor distribution. Circuits with the greatest load are starved and circuits with the least are overfed.

That clause matters for buildings more than for equipment. A filter blocked across half its face, or a duct that dumps air unevenly onto the coil, breaks the equal-load assumption while every pipe stays perfect. The fitting does its job and the coil still runs uneven.

Position is specified as well. Optimum performance comes from mounting the distributor directly on the valve outlet. Where that cannot be done, Sporlan allows a straight length of tube or pipe under two feet, sized to keep velocity up. Elbows between valve and distributor are ruled out as a hindrance to distribution.

Orientation is a preference in that document, not a prohibition. The distributor can be positioned in any direction. Best performance is usually obtained when it feeds vertically upward or downward, and vertical feed is what Sporlan recommends where the fitting is not mounted straight onto the valve.

Sizing happens on a drawing board, long before anyone owns the machine. Tube size is chosen by dividing the total load by the number of circuits and reading a rating table. The nozzle is chosen against the total load. Ratings assume a 30 inch tube length and a 10 psi drop across the tubes.

The valve in front has to be chosen knowing the fitting is there. Because the distributor takes pressure drop, Sporlan calls for an externally equalised expansion valve to compensate, and the drop has to be estimated before the valve is picked. Adding or changing a distributor therefore changes what the valve ahead of it has to be.

One more rule catches out shared systems. Each distributor must be fed by its own expansion valve. If one valve feeds two distributors, Sporlan states, the refrigerant flow will not be divided evenly between them.

Blanking off a circuit is discouraged for the same reason the tubes must match. Plugging one or more outlets ordinarily results in poor distribution. Where circuits genuinely have to be plugged, the bulletin says it should be done symmetrically.

A distributor fault can hide behind readings that look ordinary, and the reason is arithmetic. Researchers at NIST describe the standard arrangement plainly. Most evaporators use an inlet expansion valve with a flow distributor to control the overall superheat at the evaporator exit manifold.

Exit manifold is the operative phrase. The circuits recombine before anything measures them. Whatever the instrument reads at the suction line describes the mixture, and a mixture carries no record of what each stream brought to it.

The coil only works as designed when every circuit agrees with that single figure. The same paper sets out the condition. A coil performs optimally when the superheat at individual circuit exits matches the desired overall superheat in the exit manifold. Nothing at the suction line reports whether they match.

An experiment puts a number on that blind spot. Testing a three-circuit finned-tube evaporator, the NIST team found capacity degradation from refrigerant maldistribution reaching as much as 30%, with overall superheat held at its 5.6°C target throughout. The target was met and close to a third of the capacity was gone.

One of their test cases shows the mechanism in miniature. With the middle and bottom circuits running 16.7°C of superheat, holding the overall figure at 5.6°C meant overfeeding the top circuit, which then flooded. Part of the coil starved, part flooded, and the composite reading sat on target.

There is a second reason the number holds still while the coil does not. Circuits that finish boiling early spend the rest of their run carrying superheated vapour, which moves heat far less effectively than a two-phase mixture. That metal is still in the airstream and still costing fan power, while contributing much less than its area suggests.

The authors close with a caution that makes the field picture worse. Their rig held evaporator exit pressure constant. A real system answers lost capacity with a drop in evaporator pressure, so the disparity between circuit superheats in service will be greater than what the experiment recorded.

None of this makes superheat and subcooling useless. It makes a single averaged figure insufficient on a multi-circuit coil. The reading rules out a whole class of faults. It stays silent about which circuits were carrying the load when it was taken.

Distributor faults are built in, not worn in. Nothing inside the fitting has a service life or a cleaning interval. The NIST paper names the causes as bends or other blockages in the distributor tubes, or a non-optimal design. Each of those is something a person did.

Brazing is the usual field origin. Sporlan instructs that tubes soldered into the coil be pushed in far enough to prevent coil plugging from excess solder. It then delivers the line worth remembering. A leak check of the coil will not reveal a plugged circuit. The coil passes its pressure test and one circuit still carries nothing.

Heat damages the fitting in a quieter way too. Overheating a brass body drives the zinc towards the surface, the bulletin warns, leaving passageways in the brass for refrigerant leakage. Large bodies are prone to stress cracking, and quenching a hot distributor can crack it outright. A leak found at a distributor is often a record of how it was fitted.

The nozzle is the other made fault. It is interchangeable by design, which is exactly what allows a wrong one to be fitted. Sporlan is blunt that the nozzle orifice number itself does not imply a specific capacity rating, and asks for loading between 50% and 200% of rating across the operating range. Below 50%, it says, distribution can suffer.

Absence is worth checking before anything subtler. Daikin's coil instructions have the installer confirm the nozzle is present, and note that a hot gas bypass kit requires it to be moved into that assembly.

What a household can see is a pattern, and the pattern is the part worth recording. Frost may form as a band across one part of the coil while the rest stays clear. The face may be properly cold across part of its width and only mild across the rest. A photograph showing where the cold stops carries more diagnostic weight than a description of how cold it felt.

The question that moves this forward is about method. Sporlan's own instruction is that each circuit be checked individually with a probe wire or an air jet, and that a well-placed flow meter can detect a restricted circuit. Asking which circuits were tested separately, and what each one gave, separates a coil that was measured from a coil that was looked at.

One further check costs nothing to request. On a distributor with a removable nozzle, taking the nozzle out lets restrictions in the tubing or the body be seen directly, or found with a probe wire. Where a coil has been opened for a repair and the cooling came back uneven, that inspection belongs in the sequence before more refrigerant is discussed.