Aircon flare nut torque: too loose and too tight leak

A flare joint seals with nothing added to it. There is no gasket in there, no sealant, and no filler metal. The seal is a contact patch between two pieces of shaped metal, and that patch is produced on site in a few minutes.

Two separate things have to be right for the contact to hold. The cone on the end of the pipe has to be the correct shape and finish. The nut behind it has to apply the correct load. Get either one wrong and the joint can still pass on handover day, then give up later.

Order matters here, and it cannot be corrected afterwards. Mitsubishi Electric's installation manual for the MXZ-2B20NA has the installer take the flare nuts off the units and slide them onto the pipe before the cone is formed. The manual notes that fitting them afterwards is not possible. A cone formed with the nut still in the box means cutting the end off and starting again.

Only half of this joint is made on site. The port it closes against arrives on the equipment, machined at a factory to a fixed taper. The cone that meets that port is formed by hand, on the day, in whatever conditions the location offers. Every variable in the joint therefore sits on the half made at the property.

The tool does one job. It forces the end of a soft copper tube into a cone of fixed angle. The pipe is clamped in a die with a set length standing above it, and a cone-shaped head is screwed down onto the projecting end. How much pipe stands above the die decides the size of the finished cone.

That projection is published, and it changes with the tool. Daikin's R32 Split Series manual gives three settings. A flare tool made for R410A and R32 takes 0 to 0.5 mm of projection. A conventional clutch-type tool takes 1.0 to 1.5 mm. A wing-nut type takes 1.5 to 2.0 mm. Toshiba's service manual publishes the same three figures.

Manufacturers describe an acceptable result in plain visual terms, which is unusual for a spec sheet. Daikin asks for a pipe end evenly flared in a perfect circle, with an inner surface that is flaw-free. Mitsubishi Electric wants the flare smooth all around and of even length all around, with the inside shining and free of scratches. Panasonic's manual says a correct flare shines evenly and has even thickness.

The rejected shapes get named as well. Mitsubishi Electric illustrates tilted, uneven and burred flares as defective. Panasonic lists inclined, surface-damaged, cracked and uneven-thickness flares. Toshiba names obliquity, roughness and warp. No manual suggests fixing any of these by tightening harder. Mitsubishi Electric's instruction is to cut the flared section off and flare the pipe again.

Burrs earn their own line because of where they end up. Three of these four manuals require the cut edge to be reamed clean, with the pipe pointed downward while it is done. Panasonic states plainly that gas leakage may be caused if burrs are not removed. A burr on the sealing edge leaves a path straight through the finished joint, and a chip that drops inside the pipe travels into the system instead.

Tightening past the published figure makes the seal worse, on a delay. Toshiba's service manual states both failure directions in one sentence: when the torque is weak, the gas leakage may occur. When it is strong, the flare nut may crack and may be made non-removable.

The delay is what makes this so hard to catch. Daikin warns that an over-tightened flare nut may crack after prolonged use, causing refrigerant leakage. Mitsubishi Electric uses the same shape of words, warning that a nut fastened too tight may break after a long period. Read either warning closely and the timeline is the alarming part. Nothing looks wrong on the day, and nothing improves while the system runs.

Two separate parts can be ruined by the same excess. The nut is the one the manuals name most often, and a cracked nut leaks while still looking sound from outside. The flare is the second part at risk, and Panasonic lists a cracked flare among the results it will not accept. Neither shows up on a photograph taken at handover.

Over-tightening also costs the one property this joint was chosen for. Toshiba warns that a nut tightened too hard may be made non-removable. A flare exists so the connection can be opened again, for a unit swap, a relocation, or an indoor coil going out for an overhaul. A seized nut quietly converts a reversible connection into a cut.

Falling short of the figure fails more quietly and probably more often. A joint left loose seats well enough to survive a pressure test and well enough to pass a handover. It gives up slowly afterwards, and the room registers it long before anyone suspects a connection. Those refrigerant leak signs arrive years detached from their cause, which is why the joint is rarely where the search starts.

A figure gets published because feel is not accurate enough at this joint. Daikin, Mitsubishi Electric, Panasonic and Toshiba all instruct the installer to use a torque wrench, and none of them offers a hand-tight alternative. Daikin's wording is the bluntest of the four: use torque wrenches when tightening the flare nuts to prevent damage to the flare nuts and gas leakage.

The figures are not shared between brands. Four manuals give four answers for the same size of copper. The columns below come from Daikin document 3P601788-4K, Mitsubishi Electric document JG79A253H04, Panasonic order number PAPAMY1503095CE, and Toshiba service manual SVM-17057.

The spread is narrow at the thin end and decisive at the thick end. On 15.88 mm pipe, Panasonic specifies 65 N·m. Mitsubishi Electric's range for the same copper starts at 73.5 N·m. A joint tightened correctly for one of those units would be under-tightened on the other, by the second manufacturer's own numbers.

The 6.35 mm row runs the other way. Panasonic specifies 18 N·m, which sits above the top of Daikin's range of 14.2 to 17.2 N·m. Neither figure is wrong, because they apply to different hardware. The disagreement is the finding here, not a mistake in one of the manuals.

Printing a number is an admission about the joint. A manufacturer that publishes a range to one decimal place is saying the acceptable window is narrow enough that guessing will miss it. Toshiba puts the governance in a sentence: when choosing the tightening torque, comply with values designated by manufacturers. The manual for the unit on the wall governs, and the table below is evidence that a figure exists, not a setting for any particular system.

The flare nut is not the only fastener with a number. Daikin's R32 Split Series manual publishes a tightening torque for the stop valve caps and a separate, lower one for the service port cap. Those caps are seals in their own right, which is the argument the schrader valve core makes on its own page.

Whether a torque wrench came out of the van is a rare thing on an installation: a quality question with a clean yes or no. The people who did the job know the answer already. There is no technical vocabulary to get wrong in asking for it.

The answer carries more than the tool. A crew that torques flares had to look a figure up for that exact model, which means a manual was open on the job. A crew tightening by feel may still be first-rate, and no record exists of what they applied. Neither reply settles the matter by itself. The second one costs the option of clearing these joints later without testing them.

One reply deserves pushback. If someone says a flare cannot be over-tightened, or that tighter is safer, four manufacturers contradict that in print, in the same direction, using almost the same words. Excess force is a documented leak path, not a cautious margin.

Timing decides whether any of this is checkable. Insulation and casing go on within the hour, and after that the flare face cannot be reached without undoing finished work. A nitrogen pressure test recorded at handover comes nearest to proving the joints held on the day they were made.

A complete answer to the torque question contains five things.

• •The figure that was set, stated as a number, not as an assurance that everything is tight
• •The manual that figure came from, and whether it covers this exact model
• •Whether the flare nuts supplied with the units were fitted, or replacements from stock
• •Whether the flaring tool was one built for R32, or a conventional tool adjusted for it
• •The pressure test reading taken after the joints were made, and how long it was held