Moving Marine Deficiency and Thickness Data Out of the Shared Workbook

The calculation nobody can reproduce

The moment that sends a marine team looking for a system is almost always the same. Two people compute diminution from what they believe is the same data and get different answers, and neither can demonstrate why. The cause is nearly always mundane. The original thicknesses were typed in from the midship section drawing as constants, so a transcription error is undetectable. Somebody pasted a corrected reading over a computed cell. A tab was duplicated for a new tank and kept a reference to the old tank's original-thickness column. None of these produce an error message; they produce a number.

Underneath the mechanical failure sits a definitional one that a workbook cannot express. Diminution needs an original thickness, and there are three defensible candidates: the as-built scantling from approved drawings, the rule-required minimum, and the previous survey's measurement. They are not interchangeable. Diminution against as-built tells you how much steel has gone; against rule minimum it tells you how much margin remains; against the previous survey it gives you a corrosion rate. A formula in a spreadsheet encodes one of these silently, and the person reading the output six months later has no way to know which.

A system replaces the formula with a stored, versioned calculation that names its inputs. The reading, the original-thickness basis used, the source of that basis, the calculation version and the person who accepted the result are all held together. Reproducibility then is not a matter of trusting the file; it is a matter of reading the record. That single change is usually worth the migration on its own.

Average or worst point — the arithmetic trap

Take five readings on a single strake plate: 12.1, 12.4, 11.9, 12.3 and 9.8 millimetres against a 14 millimetre as-built. The average is 11.7, which against a typical allowable margin looks entirely acceptable. The 9.8 point does not. Class assessment treats these differently on purpose — general diminution of plating is judged on the average of readings taken on that plate, while local wastage, grooving and pitting are judged on the individual measurement. One column of data, two legitimate verdicts, and the wrong one is the one most workbooks are built to produce.

This is not a subtlety that stays theoretical. Grooving corrosion along weld lines and pitting in ballast tank bottoms produce exactly this signature: a plate that is generally sound with localised loss that governs the outcome. A workbook that reports plate averages will pass it, and the deficiency surfaces when a surveyor with a gauge finds the same point in a tank with the vessel already in dock and the steel not ordered.

The module has to compute and present both, and it has to make the worst-point figure as prominent as the average rather than burying it in a detail view. It should also hold the reading pattern — how many points, where on the plate, which frames — because an average over five points and an average over twenty are not comparable quantities, and comparing them across surveys is how a false improvement gets reported.

Readings that were never comparable in the first place

Before any arithmetic, the readings themselves have to be sound, and marine gauging has two systematic error sources that a spreadsheet cannot see. The first is coating. A single-echo ultrasonic thickness gauge measuring through a coating includes the coating in the reading, overstating remaining steel — and ballast tanks, the very compartments where wastage matters most, are the most heavily coated. Echo-to-echo mode or subtraction of a measured coating thickness is the correct treatment, and whether it was used is a property of the reading that must be recorded alongside it.

The second is calibration and operator practice. Gauge calibration on a step wedge at the start and end of a session, the couplant used, the probe and frequency, surface preparation, and the temperature all affect a reading at the tenth-of-a-millimetre level that renewal decisions turn on. Under IACS UR Z17 the firm performing thickness measurement for class purposes must itself be approved, with qualified operators and controlled equipment. The evidence of that control belongs with the data, not in a separate quality file.

There is a governance point here that spreadsheets obscure. If a set of readings is later found to have been taken in single-echo mode through coating, you need to know exactly which readings, from which session, by whom, so you can quarantine and re-gauge that set. In a workbook the readings are anonymous once entered. In a structured record they carry their session, so the blast radius of a bad session is knowable in minutes rather than being a reason to distrust the whole file.

Substantial corrosion and the flag that dies quietly

Substantial corrosion — wastage that falls within roughly 75 to 100 percent of the allowable margin — is not an internal severity grade. It changes the survey regime. Areas identified as substantially corroded attract annual re-gauging and extended close-up survey at subsequent surveys, and they drive the steel renewal plan and therefore the docking duration and the material lead time. Getting the flag right is a commercial decision as much as a technical one.

In a workbook the flag is a conditional formatting rule or an IF formula, and it is exactly the kind of cell that gets overwritten. Once a pasted value sits where the formula was, the cell shows a number, the formatting no longer applies, and the area quietly leaves the flagged set. Nothing in the file records the change. The next person to open it sees a clean tank. The consequence surfaces during the survey, when the surveyor gauges the same area and finds it at renewal level, and the yard is holding a docking plan built on the wrong steel quantity.

A module handles this by treating the flag as derived output that cannot be edited, only recomputed. Any override becomes an explicit, attributed act with a reason, visible in the record and in the report. The test to put to a vendor is direct: show me what happens when someone tries to change a computed substantial-corrosion result, and show me where that attempt is recorded.

Two editors, one file, and the drydock window

The trigger for most marine deficiency projects is concurrency. One workbook was fine when one superintendent owned it. It stops being fine when a surveyor is gauging in a tank with no connectivity, a second is working the deck areas, the class surveyor wants a status, and the planner is pulling numbers for the steel order — all on the same day, from files named final, final revised and final revised 2.

The structural fix is to stop passing files and start writing records. Each reading, finding and closure belongs to a location and an identified person at a recorded time. Field capture has to work fully offline, because tanks and double bottoms have no signal, and synchronise afterwards without creating duplicates. When two people record at the same location, the result should be two attributed readings that a chief surveyor resolves deliberately — not a last-write-wins overwrite that removes the earlier one without trace.

Concurrency also changes what management can see. In a workbook world, the status of a survey is whatever the last person to send a file said it was. With records, the open deficiency list, the areas still to gauge and the items awaiting evidence are live, which matters intensely inside a docking window where every day has a cost and decisions about scope extension are made on incomplete information. That visibility, more than the calculation itself, is usually what convinces the operations side to fund the change.

Recommendations, conditions and dates you do not control

Marine work has a category of deficiency that fixed-plant systems rarely model: the finding imposed from outside. A class recommendation or condition of class carries a due date set by the society, is tied to a specific survey, and has consequences for the vessel's class status if it is not discharged. Statutory findings from flag or port state have their own regimes. These are not internal findings with negotiable dates; they are constraints.

Merged into one undifferentiated list, the imposed items compete for attention with items the yard could reschedule freely, and the list gets prioritised by whoever shouts loudest. The module should hold the issuing body, the survey the item arose from, the imposed due date and the evidence required for discharge as distinct attributes, and it should be able to produce the imposed-date list on its own. The single most useful report in a docking window is the set of items with dates that cannot move, ordered by date.

There is a second reason to separate them. Discharge evidence for a class item is not the same as internal closure. A society will want the specific evidence its recommendation named — a re-gauging report, a repair record with welding procedure and NDT results, sometimes attendance. Structuring the item so that its discharge requirement is recorded when it is raised, rather than reconstructed when the surveyor returns, removes the most common cause of an item being carried forward to the next survey for want of paperwork.

What to move out of the workbook first

Do not attempt to migrate the whole workbook estate. Move the thickness measurement data first, because it is the part with the arithmetic, the part with the regulatory consequence and the part where reproducibility failure is doing active damage. Take the current survey cycle plus the previous one, so trend and corrosion rate are computable from day one, and leave older cycles in a read-only archive with a pointer.

When you extract, expect the original-thickness columns to be the worst data in the file. They were typed once, corrected inconsistently, and copied between tabs. Rebuild them from the approved drawings and the class-agreed scantling data rather than trusting the workbook, and treat any location where the drawing figure and the workbook figure disagree as a finding in its own right — because until it is resolved, every diminution figure computed at that location is unreliable in both directions.

Sequence the rest by pain. Coating condition ratings under the PSPC performance standard are usually straightforward and immediately useful because they drive re-gauging scope. Weld NDT findings from newbuild or repair work follow, since they carry their own acceptance criteria under AWS D1.1 or the applicable class rules. Deficiency items imported from the last survey report come last, because they need the class-item structure to be right first, and rushing them produces a list of undifferentiated text that recreates the original problem in a new place.

How a yard or TM firm should evaluate the module

Bring one real tank. Take the readings from a ballast tank on your most recent survey, with the original-thickness basis and the coating condition, and ask the vendor to load them and produce both the general diminution assessment and the local wastage assessment, with the substantial-corrosion set identified. Then ask them to show you the calculation provenance for one specific number, and to show what happens when someone tries to change it. That exercise separates systems that model marine assessment from systems that hold marine data in generic fields.

Test the field reality next. Gauging happens in a tank, in gloves, with no signal, sometimes on a raft. If the capture flow requires connectivity, or more than a few taps per reading, technicians will keep a notebook and key it in later, and you will have bought a slower workbook. Ask to see offline capture, a sync that produces attributed duplicates rather than silent overwrites, and photograph attachment at the reading level.

Finally, test the report. The attending surveyor will want the thickness measurement report in the format their society expects, tied to frames, strakes and structural members — not a spreadsheet export. A system that produces the assessment correctly but cannot render the required report has moved the rework rather than removed it. Atlantis builds inspection management and reporting software on Odoo, configured around stored calculation provenance, offline field capture and society-specific report formats — affordable, accessible and fully customisable. Bring a real tank's data to a working session; contact info@atlantisndt.com to arrange one.

Request a consultation