The API 653 record set a tank inspection contractor carries forward for decades

The tank register is the spine, not the report

Every API 653 calculation reads from construction data that does not change: diameter, height, number and thickness of shell courses, original construction standard, bottom design, whether a release prevention barrier is present, the product stored and its specific gravity, and the joint efficiency assumed. That data set is entered once and used for the next thirty years. Most contractors re-key it every inspection cycle from a scanned copy of the previous report.

Re-keying is where the errors enter. A course thickness transposed, a specific gravity carried over from a different product, a release prevention barrier assumed rather than verified. The last one is expensive, because the 30-year internal ceiling depends on the barrier existing, and claiming it on a tank without one produces an interval that cannot be defended when the owner or their insurer asks. The register must carry the source of each construction value.

A register also lets a contractor operate as a fleet manager rather than a job shop. Which of a client fleet of 60 tanks come due in the next 18 months, which have corrosion rates trending steeper than the last cycle assumed, which are approaching a minimum thickness threshold. That is the conversation that renews a multi-year tank programme. API 653 tank inspector services describes the delivery side of the same programme.

Interval arithmetic: RCA, N, and the divisor that changes

API 653 does not give one interval formula. It gives one pair of quantities used three ways. RCA is the remaining corrosion allowance in mils, measured thickness minus minimum required thickness. N is the corrosion rate in mils per year. External visual divides by 4N and caps at 5 years. External ultrasonic thickness measurement divides by 2N and caps at 15 years, or 5 years where the corrosion rate is not known. Internal divides by N.

The internal interval carries additional structure. Under the corrosion rate method the first interval does not exceed 10 years unless Table 6.1 safeguard credits apply, covering thicker bottom plate, thin-film or thick lining, cathodic protection, a release prevention barrier and alternative bottom materials. The credits are additive, so a lined tank with cathodic protection accumulates both. Section 6.4 caps the result at 20 years, or 30 years with a release prevention barrier.

Section 6.4.3 opens the risk-based route under API RP 580, setting the interval from likelihood and consequence rather than from corrosion rate alone. Whichever route is used, the software must store the inputs and the route, not just the answer. A due date without RCA, N, the divisor, the credits claimed and the minimum remaining thickness computed under Section 4.4.5 is an assertion the next inspector can neither verify nor update. See the API 653 tank inspection guide.

MFL bottom coverage is a claim you must defend plate by plate

Magnetic flux leakage is how the bottom gets scanned, and coverage is the number the owner cares about. API 653 requires the bottom to be examined; the method, the extent, the lane spacing and the percentage of each plate scanned come from the inspection plan agreed with the owner. Coverage is therefore a contractual commitment rather than a code number, which means your record has to evidence what was achieved, plate by plate.

A defensible MFL record carries a bottom plate map with each plate identified, the coverage achieved on that plate, the scan direction, the equipment and its calibration status on the day, the sensitivity verification against the reference plate, and every indication plotted to a position on the map. Indications carried forward then link to the prove-up ultrasonic reading that sized them, and from there to the disposition and any repair.

Field conditions make this harder than it sounds. Sludge, obstructions, annular ring proximity and appurtenances create areas the scanner cannot reach, and those are precisely the areas a leak later appears in. Recording exclusions explicitly, with reason and area, protects the contractor and tells the owner what was not covered. Offline field capture inside a tank, where there is no signal at all, is a hard requirement for capturing this at source.

Settlement survey data has to trend across cycles

API 653 Annex B evaluates tank bottom settlement: planar tilt, out-of-plane settlement of the shell, edge settlement, and localised bottom depressions. The out-of-plane evaluation fits a cosine curve to elevation readings taken at points around the shell circumference and measures each point deviation from that fitted curve against an allowable. The arithmetic is unforgiving of missing points, because the fit degrades with every gap in the series.

The problem contractors create for themselves is storing the conclusion and discarding the inputs. Next cycle, a different crew takes elevations at a different number of points, at different positions, against a different datum, and the two surveys cannot be compared at all. Settlement is a trend phenomenon. A single survey says the tank is within allowable today; only a series says whether it is moving, in which direction, and how fast.

So the record has to fix the measurement geometry: point count, point identifiers, angular positions, datum, and instrument, alongside the elevations themselves. The next crew then measures the same points and the software produces the trend without anyone reconciling two spreadsheets by hand. This is the same discipline as condition monitoring locations applied to a different measurement, and it is what makes a twenty-year tank file useful rather than merely large.

Thickness history per course, per measurement location

Shell thickness is not one number per tank. It is a reading per condition monitoring location per course, and the value of the dataset lies entirely in its history. Short-term corrosion rate comes from the last two surveys; long-term rate comes from original thickness to current. API 653 uses both, and the governing rate drives the interval and the remaining life calculation the owner relies on for budgeting.

That arithmetic requires every historical reading to survive at the location level. Contractors storing only the current survey, because the previous one arrived as a PDF from a different contractor, cannot compute a long-term rate at all and default to conservative assumptions that shorten intervals and cost the owner money. Preserving prior readings is therefore a commercial argument as much as a technical one, and it wins bids.

Each reading also needs its provenance: date, instrument and serial number, calibration status, technician and their certification, procedure revision, and surface condition. That is what lets a reading be defended when it disagrees with the previous cycle by an implausible amount. Equipment calibration tracking supplies half of that automatically and certification tracking supplies the other half, without anyone typing a serial number twice.

What API 653 Section 6.9 requires the report to carry

Section 6.8 covers records and Section 6.9 covers reports, and the distinction matters. Records are the permanent file held for the life of the tank. The report is the document issued to the owner, and it must contain enough for a third party to reproduce the inspection: what was examined, by which method and procedure revision, with which equipment, by whom, and what was found and where.

Beyond findings, the report carries the analysis: thickness data with short-term and long-term corrosion rates, minimum required thickness calculations, remaining life, settlement evaluation against Annex B, the next inspection due dates with the arithmetic behind them, and recommended repairs or alterations. The API 653 inspector identity and certification number appear on it, because the certification is what gives the determination its standing with the owner and the regulator.

Generating that from a system rather than a template has a second benefit: the owner can be given the underlying data, not only the PDF. Owners increasingly want thickness readings and settlement points loaded into their own integrity systems. A client portal for inspection reports exposing both the report and the structured data behind it removes a recurring source of post-delivery email traffic and speeds up payment.

Breakout tanks are held to a different API 653 edition

A US tank inspection contractor works two populations that look identical and are regulated differently. Non-jurisdictional tanks are inspected to whichever edition of API 653 the owner specifies, typically the current one. In-service atmospheric and low-pressure steel above-ground breakout tanks on hazardous liquid pipeline systems fall under 49 CFR 195.432, which incorporates API Std 653 by reference at a fixed edition that changes only by rulemaking.

That fixed edition is currently the 3rd edition, issued December 2001. PHMSA published a notice of proposed rulemaking on 2 June 2026 proposing to move the incorporation to the 5th edition of November 2014, to authorise risk-based inspection procedures for setting breakout tank intervals, and to allow a maximum of 20 years for an initial internal inspection and 25 years for subsequent inspections supported by RBI analysis. Comments closed 3 August 2026.

The software consequence is immediate and easy to miss: the governing standard edition belongs on the inspection record as a field. A contractor whose report template hard-codes one edition will misstate the basis on half its work, and an interval computed under one edition rules cannot be silently carried into the other. Stamp the edition, the jurisdiction and the interval route onto every tank inspection when it is created.

Selecting inspection software for API 653 work

Score candidates on whether they store calculations or answers. Ask to enter a tank, a set of thickness readings and a corrosion rate, then see whether the system produces a due date with its inputs exposed and re-computable when a new reading arrives. Ask what happens when a reading is corrected two years later: does the interval recompute, and does the change leave a trail naming who made it and when.

Then score on the fleet view. A contractor holding registers for several hundred tanks across a dozen owners has a renewal pipeline sitting inside its own data: due dates, trending corrosion rates, tanks approaching thresholds. Software that surfaces that turns a completed inspection into the next contract. Software that files it as a PDF leaves the pipeline invisible until the owner happens to remember to call somebody.

Atlantis builds this on Odoo, cloud or on-premise, with offline field capture, certification and calibration tracking, and a configurable record model, which is necessary because owner specifications differ on coverage, reporting format and data handover. Affordable, accessible, fully customizable. Request a demo or a quote and bring one tank file and one owner specification, because that pair exposes more about a system than any feature list.

Request a consultation