Why hydraulic fittings leak

1. Two different cone angles

A 37 degree male in a 45 degree female, or the reverse, touches on a line instead of on a face. The joint holds on the bench, weeps under pressure cycling, and every extra quarter turn cold-flows the softer of the two cones rather than closing the gap.

It happens because the threads are the same. At six of the eight shared sizes an SAE 45 degree flare and a JIC 37 degree flare run the same thread - 5/16-24, 3/8-24, 7/16-20, 1/2-20, 3/4-16 and 7/8-14. One identification guide states this outright and warns that the sealing surface angles are not the same.

The 30 degree seats add a third angle to the same problem. A JIS 30 degree female nut will run onto a plain BSPP male and seal against nothing at the correct angle.

2. A parallel thread with nothing at the face

A BSPP thread physically cannot seal on itself. It is parallel, so the flanks never wedge; the seal is a bonded washer, an O-ring or a cone at the face, and the thread only carries the load.

Assemble a BSPP male into a BSPP port with the washer left in the box and it will leak at any torque. Somebody then tightens it, strips the port, and concludes the fitting was faulty.

3. A tapered thread in a parallel port

This is the imported-machinery classic. An NPT male will start in a BSPP or Rc female at several sizes because the diameters are close, and at 1/2 and 3/4 in trade size the thread counts even match at 14 per inch.

The taper jams, so it feels tight. It still leaks, because a 60 degree thread in a 55 degree port bears on its crests rather than on its flanks and the wedge that makes the seal never forms. Worse, the taper is levering a parallel port open from the inside, and a cast or aluminium port bells out and is then scrap.

Both identification guides this site cites describe NPT and BSP as not interchangeable and note that the thread form and diameters are close but not identical.

4. A missing, pinched or nicked O-ring

On an ORFS or an SAE O-ring boss connection the O-ring is the entire seal and the thread carries load only. A missing ring, a ring pinched on assembly, or a nick across the groove will weep no matter how hard the nut is pulled up.

The two connections fail slightly differently. On ORFS the ring lives in a groove in the flat male face, so damage to that face is the usual cause. On an O-ring boss the ring sits on a plain shoulder behind the thread and is squashed against a machined spotface in the port, so a rough or damaged spotface is the usual cause - and once the spotface is damaged, the port is the part that needs replacing.

On a used fitting the O-ring is frequently already gone. The groove or the shoulder is still there, which is how you know what the fitting was.

5. Tightening past the point where it helps

A tapered thread seals by wedging, so there is a point past which extra torque only deforms the port. A cone seal is metal to metal, so extra torque past contact cold-flows the cone. An O-ring joint is already sealed at the moment the ring is compressed, and extra torque past that yields the thread and loses the clamp load that holds the ring where it belongs.

In every one of the three cases, the extra turn that felt like the fix is the thing that made the leak permanent. If a joint weeps at the correct assembly torque, take it apart and find out which of the four failures above you have.

A note on pressure ratings

A rating is not a property of a thread. It belongs to a particular part, in a particular material, with a particular design factor, and both pressure catalogues this site cites state their conditions explicitly: steel, at a 4 to 1 design factor against minimum burst.

They also both state a rule that is easy to lose: a fitting with two different ends takes the LOWER of the two ratings, and on most adapters that carry a pipe thread, the pipe end is the lower one. Half of the standards on this site carry no published working pressure at all here, because no two independent sources gave the same figure under conditions they both stated.