Storage Tank Inspection (API 653): A Practical Field Guide

API 653 vs. API 650: Knowing Which Code Governs

Every aboveground storage tank (AST) inspection scope starts with one question that gets skipped more often than it should: is this tank still under construction, or is it in service? API 650, Welded Tanks for Oil Storage, governs the design and fabrication of a new tank — plate thickness by design (not just corrosion allowance), weld procedures, foundation design, and the hydrotest and acceptance criteria applied before the tank ever sees product. Once that tank is hydrotested, placed into service, and reaches its first scheduled inspection, API 653, Tank Inspection, Repair, Alteration, and Reconstruction, takes over. It governs everything from that point forward: when to inspect, what to measure, how to calculate minimum required thickness, and what separates a routine repair from an alteration that requires design review. A tank built to API 650 in 1987 that has never been reconstructed is still, today, an API 653 tank for inspection purposes — the design code doesn't reset just because the operator replaced a roof drain or re-leveled a course.

Confusing the two shows up constantly in bid packages. A client issues an RFQ asking for “API 650 inspection” when what they actually need is a 653 external or internal, and that mismatch drives scope disagreements, incorrect acceptance criteria on repair welds, and, occasionally, a repaired shell course that never should have passed on new-construction thickness tables instead of in-service minimums.

The Inspection Interval Decision: External, Internal, and What Actually Drives the Clock

API 653 splits condition monitoring into two independent inspection types with two independent interval logics, and mixing them up is the single most common scheduling error in a tank farm integrity program.

External Inspection Intervals

The formal external inspection — performed by an individual meeting API 653's Authorized Inspector qualifications — is required at least every 5 years, or at the interval set by RCA (risk-based interval calculation) or one-quarter of the remaining corrosion-allowance life of the shell, whichever is less. Layered on top of that formal interval, the code expects routine in-service visual monitoring by operations or facility personnel far more often — typically monthly, sometimes as part of continuous operator rounds — watching for shell bulging, foundation settlement, coating breakdown, nozzle leaks, or vegetation encroaching on the shell. That routine check isn't a substitute for the formal external; it's the bridge between formal externals that catches a developing problem before it sits unlogged for up to five years.

Internal Inspection Intervals

Internal interval is where the corrosion-rate math actually lives. The governing calculation takes the bottom plate's measured minimum remaining thickness, divides by the calculated corrosion rate (from comparing current UT readings against original construction thickness or the prior inspection's readings), and produces a remaining bottom life. The next internal inspection interval is the smaller of half that remaining life, or a code-set ceiling — historically up to 10 years when there isn't a defensible corrosion-rate history behind the tank, extending to up to 20 years when the operator has enough inspection history and a proper RBI (risk-based inspection) justification to support it. In practice, a tank with a well-documented, low, stable corrosion rate and a strong RBI case can run twenty years between internal entries; a tank with an unknown or aggressive rate — first inspection, no history, or a bottom that's clearly pitting fast — gets pulled back to that 10-year ceiling or less, sometimes far less once the half-life math forces it.

The decision sequence an inspector actually works through once a set of thickness readings is in hand:

• Confirm which inspection type is due — has 5 years (or the RCA-adjusted interval) elapsed since the last formal external?
• Pull the corrosion rate from the last two internal thickness data sets; on a first inspection, use a conservative assumed rate until real field data exists.
• Calculate remaining bottom life: (measured minimum thickness minus minimum required thickness) divided by corrosion rate.
• Take half of that remaining life as the candidate internal interval.
• Compare against the code ceiling — 10 years by default, up to 20 years with adequate RBI justification — and select whichever number is smaller.
• Re-run the same logic per shell course using 1-foot method minimum thickness; a thin upper course can force an earlier UT re-gauge even when the bottom looks fine.
• Document the basis in the inspection report — an interval that can't be traced back to a corrosion rate and a calculation is an audit finding waiting to happen.

Bottom Inspection: MFL Screening, UT Verification, and Vacuum Box Testing

The internal inspection lives or dies on how well the bottom gets covered, and 100% manual UT grid coverage of a 150-foot-diameter tank floor is neither practical nor how it's actually done anymore. The standard approach runs MFL (Magnetic Flux Leakage) floor scanning first: a robotic crawler pulls a strong magnetic field through the plate and picks up flux leakage from pits, general corrosion, and laminations on both the top and underside surfaces, covering large floor areas quickly compared to point-by-point manual gauging. MFL is a screening tool, not a sizing tool — every indication it flags gets a follow-up UT thickness reading to confirm actual remaining wall and rule out false calls from mill scale, weld caps, or coating buildup. That MFL-then-UT-verify sequence is what makes practical full-floor coverage achievable within a normal outage window instead of turning a tank entry into a month-long manual grid exercise.

Bottom weld seams get a separate test entirely: vacuum box testing with a soap solution. A clear-topped box with a gasketed edge is set over a wetted weld seam and vacuum is drawn; any leak path through the weld shows up as bubbles forming under the box. It's slow per linear foot, but it's the standard method for confirming leak-tightness of the bottom-to-bottom and bottom-to-shell welds, and it catches through-wall defects a thickness reading alone won't — a weld can measure full nominal thickness on a UT scan and still leak through a crack or porosity path invisible to a single point gauge.

Shell thickness gauging follows a grid pattern per course — enough points to establish a representative minimum, concentrated near the lower courses and any known corrosion cells such as the vapor-space line or external coating failures. Comparing this cycle's grid against original construction thickness and the prior inspection's readings is what produces the corrosion rate driving the interval math above. That's a longitudinal data set that's genuinely hard to manage on paper across a tank farm with dozens of ASTs and years of inspection cycles — overlaying this cycle's floor-scan corrosion map against the last one, course by course, is one of the more useful applications built into our digital twin platform, since a 3D corrosion trend is a lot easier for a reliability engineer to act on than a stack of PDF grid reports pulled from five separate inspection cycles.

Shell Evaluation: The 1-Foot Method

API 653's 1-foot method calculates the minimum required thickness for each shell course based on the hydrostatic stress at a point one foot above the bottom of that course, using the specific gravity of the stored product, the tank diameter, and the course's height above the bottom of the tank (not above the course itself). Because hydrostatic head decreases going up the shell, lower courses generally need more thickness than upper ones, and the calculation is run course by course rather than treating the shell as uniform. The code also permits variable-point and elastic-analysis methods for tanks where the 1-foot method's simplifying assumptions don't fit well — very tall tanks, or unusual diameter-to-height ratios — but the 1-foot method is what most field inspectors reach for first because it's conservative and straightforward to apply against a UT grid.

The output of that calculation is the remaining corrosion allowance per course — current measured thickness minus minimum required thickness — and that number, run through the same corrosion-rate math used for the bottom, sets the next shell UT interval. Where original construction records exist and weld quality needs verification (new construction, or a shell alteration performed under 653), radiographic testing of shell welds follows API 650's RT acceptance criteria. On a routine in-service inspection, RT is used more selectively — typically where a repair weld or a suspect indication needs volumetric confirmation rather than as a blanket resurvey of the shell.

Settlement Survey: API 653 Annex B

Differential settlement doesn't show up on a thickness gauge, which is exactly why Annex B exists as its own discipline within the code. A settlement survey measures elevation at a series of points around the tank shell perimeter — typically taken during the internal inspection when the tank is out of service and accessible, though external surveys on a filled tank are also performed — and the results are run through Annex B's acceptance criteria, which check both the maximum out-of-plane deviation and the rate of change between adjacent survey points, not just the absolute low spot. A tank can show several inches of uniform settlement and be perfectly fine; what Annex B is actually hunting for is localized or edge settlement that concentrates bending stress at the shell-to-bottom weld or distorts the shell enough to bind a floating roof against its guides.

The consequences of skipping a settlement survey compound with everything else covered here: a shell that's gone slightly out-of-round from foundation settlement changes the local stress the 1-foot method assumes is uniform around the circumference, it stresses the shell-to-bottom weld in a way plain UT gauging won't detect, and on a floating roof tank it can bind the roof rim against the shell wall hard enough to tear a primary seal or crack a rim vent. A settlement survey skipped because “the shell looks fine” from the outside is one of the more expensive shortcuts taken during a tank turnaround.

Roof Inspection: Fixed vs. Floating, and the IFR Wildcard

Fixed-roof tank inspection is comparatively simple — roof plate thickness and corrosion pattern (often concentrated where condensation drips consistently from structural members), roof-to-shell weld condition, and venting are the main line items. Floating roofs add an entirely separate inspection discipline: primary and secondary seal condition and compression, roof drain integrity (a flexible or articulated drain that fails shut can sink a roof in a heavy rain event), rim vent condition, pontoon or full-contact deck integrity, and leg or support condition when the roof is landed for entry. None of that shows up in a UT thickness grid — it's a hands-on mechanical inspection of components that fail from fatigue, UV degradation, and mechanical wear rather than from corrosion alone.

Internal floating roof (IFR) tanks — a fixed roof over a floating pan, common where vapor emission control is the driver rather than weather protection — introduce a corrosion mode that neither a straight fixed-roof tank nor a straight external-floating-roof tank sees to the same degree: vapor-space corrosion on the underside of the fixed roof and the upper shell, driven by condensation and product vapor trapped in that annular space above the floating pan. That vapor-space band is a standard place to find accelerated shell thinning on an IFR tank that wouldn't be predicted by the liquid-line corrosion rate alone, which is a good reason not to assume one uniform corrosion rate applies from the bottom of the shell to the top on these tanks.

STI SP001: The Standard Next Door

Not every AST in the field falls under an API 653 program, and it's worth being precise about why. STI SP001, the Standard for the Inspection of Aboveground Storage Tanks, is commonly applied to smaller shop-built and field-erected tanks — the tanks found at petroleum marketer terminals, bulk plants, and many smaller industrial and commercial sites — rather than the large field-erected refinery and terminal tanks API 653 was written around. SP001 shares the same underlying goal, structured periodic inspection to catch corrosion and mechanical degradation before it becomes a release, but its inspection categories, intervals, and inspector qualification requirements are scaled to a different population of tanks. A program manager overseeing a mixed fleet — a refinery's crude and product tanks alongside a smaller satellite terminal's shop-built tanks — needs to know which standard governs which asset before setting up the schedule, since applying API 653's full formal-external-and-internal cycle to a small SP001-scope tank, or the reverse, either overspends the inspection budget or leaves a gap in the program an auditor will find.

Running the Program Across a Tank Farm

Everything above is manageable on one tank with a clipboard and a spreadsheet. It stops being manageable somewhere between the tenth and thirtieth tank in a farm, when external and internal intervals for dozens of tanks are all running on independent clocks set by independent corrosion rates, and a missed or late formal external on even one tank is a compliance finding. Scheduling that reliably — tracking each tank's last inspection date, calculated interval, and next-due date against a fleet-wide calendar, with enough lead time to plan the outage, book the MFL crew, and coordinate vacuum box work — is what we built the tank-inspection scheduling module in Atlantis NDT ERP around, rather than leaving that math to live in a spreadsheet that doesn't survive a staff turnover.

The other place programs quietly fail is report continuity: the corrosion rate driving every interval calculation in this article only means anything if this cycle's thickness readings get compared against the right prior-cycle readings, at the same grid locations, without transcription error. That's the case for standardizing report generation through NDT reporting software rather than free-form PDFs — the interval calculation is only as trustworthy as the data pipeline feeding it, and a field crew re-keying numbers from a paper form into a spreadsheet is exactly where corrosion rates quietly get miscalculated. If your program's underlying calculations, RBI justification for extended internal intervals, or overall API 653 compliance structure need an outside technical review, that's the kind of program audit that fits an engagement with ASNT Level III consulting — not to replace the Authorized Inspector's sign-off, but to pressure-test the corrosion-rate math and inspection data quality that sign-off depends on.

None of the code sophistication above substitutes for a UT technician who reads a genuine 0.180-inch reading correctly instead of second-guessing it against what the report “should” say, or an MFL operator who can tell a real pit indication from a coating artifact. The standards define what to calculate and when; field data quality determines whether that calculation means anything at all.

Atlantis NDT Products & Services