Minimum thickness and MAWP: the numbers auditors reread

Two numbers decide whether a pressure vessel keeps running: the minimum thickness the wall can corrode to, and the maximum pressure the wall as measured can carry. Everything downstream depends on them. Corrosion rate becomes remaining life only against a tMin. The inspection interval is capped by that remaining life. A repair or rerate decision hangs on MAWP. And when an auditor, an insurer, or an incident investigator opens the file, these are the two numbers they reread first, because they are the ones that were calculated rather than measured. This paper covers how the calculations are built, which inputs drive them, and what the record must carry so the number can be reproduced years after the person who ran it has left.

What is minimum required thickness and why does the method matter?

Minimum required thickness (tMin, or t-required) is the smallest wall that still satisfies the construction code at the design pressure and design temperature. It is not the nominal thickness, which includes corrosion allowance and mill tolerance. It is not the retirement thickness set by an owner-user, which may be more conservative. It is the code floor.

The method matters because tMin is not a single formula. For a cylindrical shell under internal pressure, ASME VIII Division 1 UG-27 gives two expressions, one for circumferential stress and one for longitudinal stress, and the larger required thickness governs. Heads have their own formulas by geometry: ellipsoidal, torispherical, hemispherical. Nozzles bring reinforcement rules. External pressure follows a chart-based procedure rather than a closed formula. A tMin that names only a number, with no method, cannot be checked.

Owner-user practice adds another layer. Many programs set a structural minimum for small-bore piping and thin components because the pressure formula returns a value too thin to survive handling loads. API 574 gives structural minimums for piping by diameter for that reason. Some programs add a further margin on top of the code value. All of these are legitimate, but each one is a decision, and the record must state which was applied.

The practical test is simple. Pick a CML at random. Ask what governs its tMin: pressure, structural minimum, or owner margin. If the answer requires opening a spreadsheet on someone's laptop, the method is not in the record.

How is MAWP calculated from a measured thickness?

Maximum allowable working pressure is the tMin formula solved the other way. For the same cylindrical shell under UG-27, MAWP = SEt / (R + 0.6t), with t the thickness available for pressure. At construction, t is the nominal thickness less corrosion allowance, and the result is the nameplate MAWP.

In service, API 510 changes what t means. The section on MAWP determination requires the calculation to use the actual thickness found by inspection, minus twice the estimated corrosion loss expected before the next inspection. The doubling is deliberate: it protects against the case where the corrosion rate is underestimated and the next inspection is pushed to the interval limit. The code edition used should be the current edition of the construction code, and the result is compared against the nameplate. The lower of the two governs unless a rerate is formally engineered and documented.

This has three consequences for the record:

• The corrosion rate feeds the MAWP. A change to the rate, or to the planned interval, changes the pressure the vessel is allowed to run at. The MAWP record must reference the rate and interval it assumed.
• MAWP is a moving value. It is recalculated after every thickness survey. Every historical value must remain retrievable so the trend can be seen.
• A rerate is a separate act. Recalculating MAWP does not change the nameplate. Raising or lowering the rating requires the rerate procedure in API 510, including engineering review and, for some cases, pressure testing. The record must distinguish an assessment MAWP from a rerated MAWP.

Which inputs does a tMin or MAWP calculation depend on?

Every input below has caused a disputed number somewhere. The list is the audit checklist.

• Design pressure and design temperature. From the U-1 data report or the equivalent design document. Temperature matters because allowable stress falls with it.
• Allowable stress (S). Read from ASME Section II Part D for the material specification, grade, and design temperature. The code edition changes the value, and so does using the wrong product form.
• Joint efficiency (E). Depends on weld type and the extent of radiography recorded at construction. Assuming E = 1.0 when the radiography record shows spot RT understates tMin.
• Inside radius or outside diameter. UG-27 is written in terms of inside radius. Field measurements are usually outside diameter, so the conversion must be shown.
• Corrosion allowance. Included in nominal thickness, excluded from tMin. Confusing the two is the most common error in spreadsheet calculations.
• Static head and external loads. For tall vessels, liquid head adds to the design pressure at the bottom courses. Wind and seismic loads add to longitudinal stress.
• Material specification. Drives S and, for some materials, the temperature limits of the formula.

A calculation record that carries the result and not these inputs cannot be verified. A record that carries the inputs but not their source (which data report, which code edition, which radiography record) can be verified only by someone who already knows the vessel.

How does the equipment type change the formula?

The same measured thickness means something different on a pipe, a vessel, and a tank, because three different codes govern.

Piping (ASME B31.3). Straight pipe under internal pressure uses t = PD / (2(SEW + PY)), where D is outside diameter, W is the weld joint strength reduction factor at temperature, and Y is a coefficient that depends on material and temperature. B31.3 works in outside diameter, unlike UG-27. Mill undertolerance (typically 12.5 percent) is subtracted from nominal when setting the corrosion allowance available.

Vessels (ASME VIII Division 1). UG-27 for shells, UG-32 for formed heads, UG-28 for external pressure. Inside radius, joint efficiency, and the 0.6P term are the identifying features.

Storage tanks (API 653). Shell courses are assessed using the hydrostatic formula in API 653, which relates required thickness to tank diameter, liquid height above the course, product specific gravity, allowable stress, and joint efficiency. Bottoms follow a separate minimum remaining thickness rule. The inputs are entirely different from a vessel: no design pressure, and liquid level is the driver.

A calculation that picks the wrong code for the equipment type gives a confident wrong answer. The record must state the code that was applied, and the classification of the equipment that led to it. A program that stores CMLs on a vessel and on the attached piping in one table must route each to its own formula.

What does an auditor look for when rereading a thickness calculation?

Auditors do not check arithmetic first. They check reproducibility. The sequence is predictable:

1. Can the code and edition be identified? If the record says "ASME" with no section, division, or year, the audit stops there.
2. Are the inputs present and sourced? S traced to Section II Part D for a stated temperature. E traced to a radiography record. Design pressure traced to the data report.
3. Does the arithmetic reproduce? The auditor recomputes. A small difference usually means a rounding or unit issue; a large one means a different input.
4. Who performed and who verified? Two identities, two dates. A calculation verified by its author has not been verified.
5. Has the value changed, and why? If tMin for a CML differs from the previous survey, there must be a reason on file: new code edition, corrected input, or reclassification.
6. Is the result actually used? The auditor follows tMin into remaining life and into the interval. A correct tMin that the interval calculation ignored is still a nonconformance.

A calculation that cannot be reproduced from the record is not a calculation. It is a number someone once believed.

Programs that pass this review share one trait: the record was built to be reread, not to be filed. That is a data design decision made long before the audit.

Where do thickness calculations go wrong in practice?

The failures are rarely in the formula. They are in the handling of inputs across time and across people.

• Spreadsheet drift. A tMin sheet is copied for a new vessel, one cell is edited, and the others silently carry the previous vessel's E or S. Years later nobody knows which cells were checked.
• Edition mismatch. Allowable stress read from a 1998 table used with a 2023 formula, or the reverse. The values move, sometimes materially, between editions.
• Nominal versus measured. MAWP calculated on nominal thickness after twenty years of service. This overstates the rating and hides the corrosion already consumed.
• Unit and diameter confusion. Inside radius in one calculation, outside diameter in the next, without the conversion shown.
• Lost provenance after a system migration. The result survives the migration; the inputs and the method do not. The new system shows a tMin with no way to say how it was derived.
• Overwriting instead of versioning. A corrected tMin replaces the previous value in place. The trend, and the fact that a correction was made, disappear.

Every one of these is a record-keeping failure rather than an engineering one. The fix is structural: store the inputs with the result, name the method, version the record, and require a second identity to verify.

How intelliSPEC handles it

intelliINTEGRITY treats tMin and MAWP as calculations with provenance, not as fields to type into.

• Routing by equipment type. The minimum thickness calculation routes by equipment classification to ASME B31.3 for piping, ASME VIII Division 1 for vessels, or API 653 for tanks. The output returns the design code and the formula alongside the number, so the record answers the auditor's first question before it is asked.
• Inputs on the CML. Each CML point carries 64 fields, including nominal thickness, tMin and the method that produced it, design pressure, material, outside diameter and schedule, allowable stress, joint efficiency, corrosion allowance, and the latest thickness with three corrosion rates. The inputs live with the point, not in a side spreadsheet.
• MAWP as a registered calculator. MAWP is one of 199 registered calculators. Calculators are versioned, and a segregation-of-duties rule prevents a calculator author from approving their own work, so the method itself has a second identity behind it.
• Records that render as executed. A completed record pins the field configuration, locale, and reference options it was executed with. When the form changes next year, the historical calculation still renders exactly as it was performed.
• A hash-chained audit trail. Every create, update, and export carries the actor, tenant, request id, and a SHA-256 hash chain that a verify routine can recompute for an auditor. Audit events are never auto-deleted. The question "who changed this tMin and when" has a permanent answer.
• Honesty about scope. The platform implements API 579-1 Part 4 Level 1 screening but deliberately does not implement Part 5 Level 1 pending licensed-engineer verification, and remaining strength results stay provisional until an engineer verifies them. A number the engine is not confident in is labelled that way.

For the definitions used here, see the glossary. For how these calculations sit inside a refinery or chemical integrity program, see oil and gas and chemical.