One thickness dataset across every terminal, or none at all

A roll-up problem is always an identity problem

Terminals fail to consolidate thickness data for a reason that sounds trivial until it is quantified: the same physical location is named differently at every site. Shell readings are referenced by course, by elevation and by circumferential position, and each of those three has competing local conventions. Course numbering runs upward from the bottom angle at most sites and downward at some. Elevation appears as height above the bottom plate, above the chime, or as a nominal band such as mid-course. Circumferential position turns up as degrees from true north, degrees from the north manway, a clock position painted on the shell a decade ago, or the name of the nearest stairway.

The consequence is that a corporate report showing average course 2 thickness by terminal is comparing the wettest and thickest bottom course at nine sites with an upper course at three. Nobody catches it, because every number is plausible. What is actually lost is the corrosion rate. A rate requires the same point measured twice, so if point identity is not stable across contractors and years, the only defensible statistic left is the tank minimum — which discards exactly the location information that would make the minimum actionable.

Standardising does not mean renumbering every tank at every terminal on day one, which is how these programmes stall. It means defining one corporate convention, storing each tank's original site labels as aliases against it, and mapping at import, so historical campaigns stay readable in their own terms while the trend runs on a stable key.

One population, several standards, and a minimum thickness that is not comparable

A terminal rarely holds a single class of tank. Field-erected API 650 tanks stand next to shop-fabricated tanks, alongside older tanks built to superseded editions and tanks whose service has changed since construction. API 653 governs in-service inspection, repair, alteration and reconstruction of the field-erected population, with shell minimum thickness derived from the tank's own geometry, product specific gravity, fill height and joint efficiency. STI SP001 addresses shop-fabricated tanks through a different framework driven by tank category and the presence of a release prevention barrier.

A dashboard showing the percentage of tanks above minimum thickness across twelve terminals, without recording which standard, which edition and which inputs produced each minimum, is not a compliance view. It is an average of incompatible calculations. Two tanks with identical readings can sit on opposite sides of the line because one site used the API 653 shell calculation with actual product density and the other applied a conservative corporate floor that was set for a different service twenty years ago.

The remedy is to store the minimum thickness basis as data on the tank: the governing standard and edition, the fill height and specific gravity used, the joint efficiency assumed, who performed the calculation and when. The roll-up can then be filtered to like-for-like, and a change of product service triggers an explicit recalculation instead of quietly invalidating every remaining-life figure attached to the tank.

Bottom data: four methods, four different kinds of number

The floor is where the standardisation problem becomes physical. Magnetic flux leakage scanning produces an estimated percentage of wall loss per indication, not a thickness, and its output is intended to be verified by ultrasonic prove-up at selected indications. Manual ultrasonic grids produce thicknesses at discrete points. Vacuum box testing produces a pass or fail on weld integrity and no dimension at all. Robotic in-service inspection produces thicknesses over limited coverage while the tank stays in product.

Sites mix these freely and the numbers land in the same column. Loading MFL percentage estimates into a thickness table converts an estimate into a measurement, and that measurement then feeds a bottom minimum remaining thickness calculation, which in turn sets the internal inspection interval. It is a chain in which nobody at any step did anything visibly wrong, and at the end of which an interval measured in years rests on a figure nobody ever measured.

Every value therefore needs a method flag. MFL indications need a link to the ultrasonic prove-up that verified them, or an explicit record that none was performed. Coverage needs storing as coverage rather than being implied by the number of rows returned: a floor scanned at forty percent coverage and a floor scanned at full coverage produce datasets that look identical in a spreadsheet and mean entirely different things when an inspector is deciding whether a bottom can run another decade.

The assumptions have to travel with the reading

Bottom remaining life at a terminal is not a straightforward extrapolation of measured loss, because most of the loss is on the soil side and nobody measures the underside of a floor in service. The projection subtracts an assumed underside corrosion rate as well as the measured product-side loss across the proposed interval. That assumed rate comes from somewhere — a previous bottom removed at the same site, a sister tank, published experience, or a corporate default — and its provenance quietly determines the interval.

Cathodic protection status, the presence and type of release prevention barrier, whether the tank sits on sand, oiled sand or a concrete ringwall, the product history and any period spent on water bottoms all bear on that assumption. If the assumed rate lives in an engineer's spreadsheet rather than beside the reading it acts on, two terminals with identical measured data will produce different internal inspection intervals and neither will be able to explain why to an auditor, an insurer or a buyer.

Storing the assumption as a field on the record, with its justification and its author, has a second and larger effect: it makes the assumption reviewable. When a bottom does come out, the actual underside condition can be compared against what was assumed, and the corporate default can be corrected with evidence instead of defended by seniority. Very few terminal operators can do this today, and it is the highest-value thing a standardised thickness history unlocks.

Exclusions at a terminal are structural, not incidental

A great deal of a tank cannot be measured, and the reasons repeat at every site. The annular ring beneath the shell is inaccessible without excavation. The critical zone of bottom plate within three inches of the shell-to-bottom weld is exactly where corrosion concentrates and exactly where scanning equipment cannot reach. Areas under floating roof support legs, beneath the pontoon, around sumps and under permanent internal structures are excluded by geometry. Shell plate behind stairways, under insulation support rings and behind attached piping is excluded by access. And a tank in service cannot have its floor inspected at all.

Because these exclusions are predictable, they can be recorded once against the tank and reported against every campaign, rather than rediscovered independently by each contractor. The value becomes obvious when someone asks why a bottom minimum was reported from a scan that could not reach the critical zone, or when an API 653 inspector's report contains a recommendation that turns on an area no method covered.

The record should hold the location, the intended method, the reason it was not taken from a controlled list, and what would recover it: excavation, landing the roof, an out-of-service window, insulation removal. At a terminal that list is a maintenance plan for the next out-of-service opportunity, and those opportunities are years apart and scheduled around commercial commitments rather than inspection convenience. Losing the list means losing the window.

Where the record gets audited, and by whom

Terminal thickness records serve more masters than most inspection data. The API 653 inspector relies on them for interval determination and for the judgements behind a repair or alteration recommendation. The SPCC plan under 40 CFR 112 requires integrity testing of bulk storage containers in accordance with industry standards, with inspection records retained, and a professional engineer's certification of that plan rests in part on the same evidence. State programmes add requirements at many terminals. Insurers ask. Buyers in an asset transaction ask hardest of all, and they ask with a deadline.

Each audience asks a slightly different question, and all of them are answered badly by a folder of contractor PDFs. What they want is the ability to take one number — a reported shell minimum, a bottom remaining thickness, an interval — and follow it down to the readings behind it, the method used, the technician's qualification on the day, the instrument's calibration, and the assumptions applied. A system that can do that turns an audit into a query. A system that cannot turns it into a fortnight of somebody's working life and a set of answers hedged with caveats.

Risk-based inspection raises the stakes further. An assessment under API 580 and API 581 is only as sound as its inputs, and feeding it thickness data whose location identity, method and assumptions differ by site produces confident-looking intervals with nothing underneath them. Standardising the record is a prerequisite for RBI, not a refinement to be attempted afterwards.

What to ask before standardising twelve sites

The questions that matter are about migration rather than features, because the features all demonstrate well. Can the system take twelve years of contractor reports in twelve formats and land them against one location model, keeping the original labels as aliases so the historical report remains reproducible? Can it record method, coverage and prove-up links on every value? Can it hold the minimum thickness basis and the underside corrosion rate assumption as structured data rather than as notes? Can it store an exclusion with a controlled reason and report exclusions across the whole estate? For any tank, can it show every reading ever taken at a given location, with the campaign each came from?

Then ask the operational question, which is the one that justifies the project. Can a corporate integrity manager see the ten tanks across all terminals with the shortest projected remaining life, and immediately see which of those ten rest on assumed rather than measured underside rates? That view changes capital planning and turnaround sequencing, and it is unreachable without every field described above being captured at the moment of the reading.

Atlantis configures this on an Odoo-based inspection management platform built around the tank, its courses and its inspection history rather than a generic equipment record, and it is affordable, accessible and fully customisable to a terminal group's own conventions. To see it run against historical reports from two of your own sites, request a consultation at info@atlantisndt.com.

Request a consultation