A hydraulic manifold flange leaked at 70 bar. The flange was rated for 250 bar. The gasket was a spiral-wound, correct size. The bolts were M12 8.8, torqued to 85 N·m. The customer re-torqued them. It leaked again. They replaced the gasket. It leaked again. They called us.
We checked the torque. The bolts were dry (no lubrication). The torque-tension relationship for dry bolts is uncertain — the friction coefficient can be 0.15 to 0.25. At 85 N·m on a dry M12, the actual clamp force might be 30 kN or 50 kN. The gasket needs 45 kN to seal at 70 bar. If the actual clamp was 30 kN, the gasket wasn’t compressed enough. It leaked. The customer re-torqued to 85 N·m, but the dry friction was still unpredictable. The clamp force varied bolt to bolt.
We put anti-seize on the bolt threads and under the bolt heads. The friction coefficient dropped to 0.10. We re-torqued to 65 N·m (the correct torque for lubricated M12 8.8). The clamp force was now 45 kN, predictable and even across all bolts. The leak stopped. We also re-torqued after 24 hours (the gasket and surfaces settle). It didn’t leak for two years.
The gasket was never the problem. The bolts were dry. The torque was a guess. The customer had been replacing gaskets because that was the visible part. The real issue was the bolt friction. Next time a flange leaks, check the bolt lubrication before replacing the gasket. A dry bolt at 85 N·m might be providing less clamp than a lubricated bolt at 50 N·m.
