B31G remaining strength of corroded pipe

A blunt patch of external corrosion in a ductile line pipe does not fail by fracture — it fails when the remaining ligament reaches its flow stress and tears. B31G is a semi-empirical treatment of that: it replaces the real corrosion profile with a simple idealised shape, compares the metal lost to the metal that was there, and applies a bulging correction because a thin patch in a pressurised cylinder bulges outward and raises the local stress above the nominal hoop stress.

That bulging correction is the Folias factor M, and it is driven by the dimensionless group z = L²/(D·t). A short defect is restrained by the sound metal around it and M is close to 1. A long defect is unrestrained, M grows, and eventually the defect behaves like a uniformly thinned cylinder where only the ligament thickness matters. Both methods switch to a long-defect treatment past a limit on z.

The original B31G assumes a parabolic corrosion profile, so the area lost is two thirds of depth times length, and takes the flow stress as 1.1 × SMYS. It is deliberately conservative and was written when the assessment had to be done by hand or from a nomogram. The Modified B31G, sometimes called the 0.85dL method, uses a more realistic area of 0.85 × depth × length, raises the flow stress to SMYS + 68.95 MPa (10 ksi), and uses a better-fitted Folias expression. It typically returns 5 to 15 % more allowable pressure than the original for the same defect, and it is the method most operators now run first.

The scope limits matter as much as the arithmetic. B31G applies only to blunt, volumetric metal loss in ductile line pipe carrying internal pressure. It does not apply to cracks, crack-like flaws, laminations, gouges, grooving corrosion in a seam or girth weld, corrosion in a dent, or any defect where longitudinal or bending stress is significant. Corrosion deeper than 80 % of the wall is outside the method entirely and requires repair or replacement. And the answer depends on measuring the maximum depth and the axial length correctly — an under-measured depth from a single pit gauge reading is the most common way this calculation goes wrong in the field.

z = L² / (D·t)
Original B31G, z ≤ 20:  S_f = 1.1·SMYS · [1 − (2/3)(d/t)] / [1 − (2/3)(d/t)/M],  M = √(1 + 0.8z)
Original B31G, z > 20:  S_f = 1.1·SMYS · (1 − d/t)
Modified B31G:  S_f = (SMYS + 68.95 MPa) · [1 − 0.85(d/t)] / [1 − 0.85(d/t)/M]
  M = √(1 + 0.6275z − 0.003375z²)  for z ≤ 50 ;  M = 0.032z + 3.3  for z > 50
Failure pressure  P_f = S_f · 2t / D      Safe pressure  P_s = P_f / SF