What causes O-ring failure in a machined seal joint, and how do you trace it back to the groove?

How do you recognize which O-ring failure mode caused the leak?

Look at the ring before you clean it or throw it away: its shape, the side the damage is on, and its surface tell you the failure mode. Flat sides mean compression set, a chewed edge on one side means extrusion, and a clean ring in a leaking joint usually means the groove or sealing surface is at fault.

Remove the ring carefully, mark which side faced the pressure, and photograph it next to the groove it came from. The common modes, in the order they are usually found on machined hardware, are:

• Compression set. The cross-section is no longer round; it has taken the shape of the gland, with flat faces where it was squeezed. The ring has lost its rebound and no longer pushes back against the sealing surfaces. ASTM D395 is the standard test for this property, and its scope names seals among the applications it covers.
• Extrusion and nibbling. Small bites or a frayed lip along one edge, always on the low-pressure side, where rubber was forced into the clearance gap and then pinched off.
• Installation damage. Clean cuts, skived slivers or a peeled strip, often in one place on the circumference, where the ring was pushed over a thread, a cross-drilled port or a sharp groove edge.
• Abrasion. A flat, polished or scored band on the face that rubbed against a moving surface, found on rods, pistons and rotary shafts.
• Spiral (twist) failure. Cuts or marks that wind around the ring in a corkscrew, from a reciprocating ring that rolled in its groove.
• Chemical attack and swell. A ring that is soft, oversized, sticky or blistered, or the opposite, shrunken and hard. ASTM D471 is the test method used to compare how rubber withstands a given liquid.
• Heat and ozone damage. A hard, cracked surface from heat ageing, or fine cracks at right angles to the stretch direction from ozone, the effect tested under strain by ASTM D1149.
• Decompression damage. Blisters, pits or internal splits on rings in high-pressure gas service, caused by dissolved gas expanding when pressure drops quickly.

If the ring looks undamaged and still leaks, stop looking at the rubber. A healthy ring in a leaking joint is the strongest sign that it was never squeezed enough, that the sealing face has a leak path across it, or that the mating parts move apart under load.

How do you diagnose a failed O-ring and measure the groove it came from?

Treat the ring, the groove and the mating part as one set of evidence: record the ring, measure the actual gland, recalculate the squeeze and gap at the measured sizes, and inspect the sealing surface under magnification. Only then decide whether the fix is the ring or the part.

Use this checklist on the failed joint and, if possible, on a few parts that have not failed:

1. Keep and label the ring. Note its position, the pressure side and the service conditions: fluid, temperature range, pressure and whether the joint is static or moving.
2. Confirm the ring is what was specified. Check the size against the SAE AS568 dash number or the metric size on the drawing, and the compound against the bill of materials. Wrong rings from a mixed bin are a common and cheap finding. Physical properties of finished rings, including changes after ageing, are measured by the methods in ASTM D1414.
3. Measure the gland, not just the groove. Record the groove diameter, groove width and the mating bore or shaft diameter on the actual parts, then calculate the radial gland depth and the diametral clearance. For a face seal, measure groove depth and the flatness of both faces.
4. Recalculate squeeze, fill and gap at worst case. Use the measured sizes and the ring tolerance, with the shaft pushed to one side. Squeeze (%) is the ring cross-section minus the gland depth, divided by the ring cross-section, times 100: take the smallest ring section with the deepest gland for the minimum, and the largest section with the shallowest gland for the maximum. Fill (%) is the ring’s cross-section area divided by the groove’s cross-section area, times 100, worked out at its own worst case and again with any swell from the fluid. Compare the results with the gland standard or the seal maker’s data, not with the nominal drawing.
5. Inspect every edge the ring passes. Groove corners, the bore entry, ports and thread starts: look for burrs, sharp corners and missing chamfers.
6. Inspect the sealing surface. Measure roughness across the seal path and look at the lay direction under a loupe or a vision system. One radial scratch matters more than an average Ra value. As a reference point, one seal maker’s face-seal gland chart lists 0.8 µm Ra (32 µin) on the groove bottom and sealing face for liquids, 0.4 µm Ra (16 µin) for vacuum and gases, and 1.6 µm Ra (63 µin) on the groove sides (ERIKS O-ring technical handbook, static axial glands); use the values your seal maker publishes for your compound. The surface roughness guide covers how to specify it.
7. Check the joint under load. If the leak appears only at pressure, check whether the bore expands, the flange lifts or the bolts stretch enough to open the gap.
8. Separate part from process. If failed and good joints have the same measured gland, look at assembly (lubrication, twisting, tools) and at the ring lot instead.

A leak test with the right sensitivity tells you whether the fix worked; the leak testing guide explains how to choose one. Write the findings up as a root-cause record so the corrective action is traceable.

What should the drawing and RFQ say so the seal joint does not fail again?

Put every condition the seal depends on in writing: the gland standard and ring size, the gland dimensions with tolerances that hold squeeze at both extremes, edge and chamfer notes, the roughness and lay of the sealing surface, and the compound, fluid and temperature range. A note that says only “O-ring groove” leaves each of those to the machinist.

• Ring and gland reference. The ring size by AS568 dash number or metric size, and the gland standard the groove follows, for example “Gland per SAE AS4716, static”.
• Dimensions that control squeeze. Tolerance the groove diameter and the mating bore or shaft together, and add a concentricity or position control between the groove and the sealing diameter where eccentricity matters.
• Edges. “Lead-in chamfer on bore entry”, “groove edges broken, no sharp edges or burrs” and “ports and threads the ring passes deburred”.
• Sealing surface. A roughness value on the sealing face with a lay symbol, and a note such as “no radial tool marks or scratches across the seal path”.
• Flatness and clamping. A flatness tolerance on face-seal flanges, and the bolt torque and sequence on the assembly drawing.
• Material and service. The compound by name or by an ASTM D2000 line call-out, plus the fluid, pressure and temperature range. Where compression set matters, state the requirement and the test method.
• Inspection. Which gland dimensions are critical and how they are to be checked, the leak test and its acceptance limit, and whether a first article report is required.

On the RFQ, add whether the supplier fits the ring or ships the part bare, and whether the joint has failed before. A short note such as “previous lot leaked; extrusion on low-pressure side” tells the supplier where to look before quoting.