A bolted flange on a hydraulic line. It leaked at 70 bar. The flange was rated for 250 bar. The customer used a standard paper gasket. The system pressure was well within the rating. But the bolts weren’t torqued correctly, and the gasket was wrong for the fluid. This is how I size bolted flanges for pressure retention.

The flange load

The pressure inside the flange creates an axial force that tries to separate the flange:

F_p = P × A_inner

Where P is the pressure (bar), and A_inner is the internal area (π × d²/4). For a DN50 flange at 250 bar: F_p = 250 × 10⁵ × π × 0.05²/4 = 250 × 10⁵ × 0.00196 = 49,087 N. That’s 5 tons trying to push the flange apart.

The bolts must hold the flange together. The total bolt force must exceed F_p with a margin. For a hydraulic system, I use a design factor of 2: F_bolt_total ≥ 2 × F_p = 98,174 N. With 4 bolts (standard for DN50), each bolt must provide 24,544 N. An M12 bolt at property class 8.8 has a yield of 640 MPa. The stress area is 84.3 mm². At 70% yield (preload limit): F_preload = 0.7 × 640 × 84.3 = 37,766 N per bolt. That’s enough. So why did it leak?

The gasket seating stress

The bolts must not only hold the pressure — they must compress the gasket enough to seal. Different gaskets need different seating stresses. A soft paper gasket needs about 20 MPa. A spiral-wound metal gasket needs 60 MPa. The seating stress is the compressive stress on the gasket from the bolt preload:

σ_gasket = F_bolt_total / A_gasket

For a DN50 flange with a spiral-wound gasket (ID 52 mm, OD 72 mm): A_gasket = π × (72² – 52²)/4 = π × (5184 – 2704)/4 = 19,478 mm². The required seating stress is 60 MPa. F_required = 60 × 19,478 = 1,168,680 N. But our bolts only provide 4 × 37,766 = 151,064 N. That’s 8x less than needed. The gasket never seats. It leaks.

What I changed

1. Changed the gasket. For a hydraulic line at 70 bar, I use a nitrile rubber O-ring (or a rubber-clad gasket). The seating stress for rubber is about 5 MPa. F_required = 5 × 19,478 = 97,390 N. Our bolt preload (151,064 N) is enough. The gasket seats properly. No leak.

2. Increased bolt size for high-pressure. At 250 bar, I use M16 bolts instead of M12. The stress area jumps from 84.3 to 157 mm². Each bolt provides 0.7 × 640 × 157 = 70,336 N. Four bolts provide 281,344 N. That covers the pressure force (49,087 N) and seats the spiral-wound gasket at 60 MPa (1,168,680 N… still not enough). At 250 bar with a spiral-wound gasket, I go to 8 bolts (M16) or use a ring joint (RTJ) flange.

3. Specified torque values. I write the exact torque on the drawing: “M12 bolt, 8.8 grade, torque = 85 N·m.” The mechanic doesn’t guess. Undertorqued bolts leak. Overtorqued bolts stretch and lose preload.

The relaxation issue

Over time, bolts relax. Gaskets creep. The preload drops. For a flange at 250°C, I add a 10% preload margin. For a flange at -40°C, the bolts contract — I use spring washers or Belleville washers to maintain preload. The flange that leaked at 70 bar wasn’t over-pressurized — it was undertorqued with the wrong gasket.

The two numbers I check: bolt preload must exceed 2x the pressure force, AND the gasket seating stress must be achieved. For hydraulic lines, use rubber gaskets (low seating stress). For steam/oil at high pressure, use spiral-wound and spec more/bigger bolts. The leak wasn’t the flange rating — it was the wrong gasket and wrong torque.