Two hundred and forty ultrasonic readings, and the three of them that changed the quotation.
An ultrasonic thickness gauge does one thing: it tells you how much metal is left at the point you put the probe. It says nothing about the point ten centimetres away. That single limitation is the reason a thickness survey is either genuinely useful or completely misleading, and the difference is entirely a matter of how the readings are planned.
A number is not a survey
A yard can take twenty readings on a hull, find them all comfortably in tolerance, and report that the plating is sound. The report will be true about those twenty points and silent about everything else. Plating loss is not uniform — it concentrates where water stands, where a coating has failed at an edge, where a bilge does not drain, and where two dissimilar metals meet. Twenty randomly placed readings will usually miss all of it.
A grid changes the exercise. Readings taken by station and frame, at a spacing fixed in advance, produce a map: a field of values that can be compared with the as-built scantlings and, more importantly, with itself. Loss shows up as a pattern, and patterns point at causes.
One low reading is a data point. A cluster of low readings is a diagnosis.
The three readings that move the price
On a survey of any size, the great majority of readings confirm what was expected. They are not wasted — they are what makes the exceptions meaningful — but they do not change anything. What changes the quotation is a small number of readings that cluster, and clusters are only visible if the grid was dense enough to catch two of them.
A cluster tells you three things a single reading cannot. Where the loss is, which is a location on the vessel. How large the affected area is, which is a quantity of steel and a number of hours. And what caused it, which determines whether renewing the plate solves the problem or simply resets it. The third is the one that decides whether the owner is buying a repair or an annuity.
What we record, and why the format matters
- Every reading, not the summary. A mean thickness figure is close to useless: it averages away exactly the local minima the survey exists to find.
- Position by station and frame, so the map can be laid over the structural drawings and read against the original scantlings.
- Coating condition at the point of measurement, because a low reading under intact paint and a low reading under a blister are different findings with different remedies.
- The date and the instrument, so the next survey can be compared against this one rather than starting again.
That last point is worth more than it sounds. A single thickness survey establishes a condition. Two surveys, taken the same way, establish a rate — and a rate is what lets an owner plan rather than react. Knowing that a bilge area is losing measurable thickness is a maintenance decision. Knowing how fast is a budget.
What it means for the quotation
A dense, mapped survey usually makes the initial quotation larger and the final invoice smaller. It is uncomfortable to sell on that basis, because the competing quotation is lower on the day it is read. But an unmapped survey does not remove the steel that needs replacing; it only moves the discovery to a point in the programme where the yard is committed, the schedule is tight and the owner's negotiating position is gone.
Class attends the significant findings, and the readings that support them go into the file. What the owner ends up holding is not a reassurance. It is a map, a set of dates, and a defensible reason for every plate that was cut out — and for every plate that was not.
Written at the yard in Tuzla. Figures quoted are the ones we record on the job and hand over with the vessel.