API 579 Level 1 general metal loss

General metal loss is corrosion spread over an area large enough that the component behaves as if it were uniformly thinner, rather than as a shell with a local dimple in it. API 579-1 Part 4 handles it, and the central idea is thickness averaging. A single low ultrasonic reading does not by itself condemn a vessel, because a pressure boundary carries load by membrane action over a finite length. What matters is the average wall over the length that actually shares the load, together with a floor on how thin any single point may be.

The length over which averaging is permitted is L = Q·√(D·t_min). The √(D·t) group is the characteristic decay length of a shell — the distance over which a local disturbance in a cylinder dies away — and Q scales it according to how thin the worst point is relative to what is required. Q comes from Table 4.4 and follows Q = 1.123·√{[(1−R_t)/(1−R_t/RSF_a)]² − 1}, where R_t = (t_mm − FCA)/t_min is the remaining thickness ratio and RSF_a is the allowable remaining strength factor, 0.90 for most pressure equipment. As R_t approaches RSF_a the permitted averaging length grows without limit, because the wall is barely below requirement anywhere. As R_t falls, Q shrinks and the averaging window closes down onto the defect.

Level 1 then applies two acceptance criteria, and both must pass. First, the average measured thickness less the future corrosion allowance must be at least the required minimum: t_am − FCA ≥ t_min. Second, the single minimum measured thickness less FCA must be at least the greater of half the required thickness and 2.5 mm: t_mm − FCA ≥ max(0.5·t_min, 2.5 mm). That second criterion is the guard against local perforation and against a very thin spot being hidden by a favourable average.

Two things sit outside this calculation and must be checked separately. Level 1 in Part 4 assumes the metal loss is remote from major structural discontinuities — nozzles, head-to-shell junctions, stiffeners — and does not apply where supplemental loads such as wind, weight or thermal expansion are significant. Where the loss is clearly local rather than general, Part 5 for a local thin area is the correct route and will usually be less conservative. And the future corrosion allowance is an engineering judgement about remaining life, not a measurement: it is the corrosion rate multiplied by the intended interval to the next inspection.

R_t = (t_mm − FCA) / t_min
Q = 1.123·√{ [(1 − R_t)/(1 − R_t/RSF_a)]² − 1 }   for R_t < RSF_a ;  Q = 50 otherwise
L = Q · √(D · t_min)      (length for thickness averaging)
Check 1:  t_am − FCA ≥ t_min
Check 2:  t_mm − FCA ≥ max(0.5·t_min, 2.5 mm)
MAWP = S·E·(t_am − FCA) / (R + 0.6·(t_am − FCA))