Steel-Hull Ultrasonic Survey: Measurement Grids, Benchmarks, and How to Grade What You Find

A 1992 narrowboat changes hands. The pre-purchase survey runs to nineteen pages, and the hull paragraph reads "steelwork in good condition, sound throughout". Eighteen months later the new owner finds a weep at the waterline on the port quarter. Then a second one, a hand's width along. The boat goes into dock, and a thickness gauge tells the rest of the story: the topside plating has wasted to under 3mm across a band a metre long, just above the rubbing strake. It is enough for the insurer to call for remedial steelwork before the next renewal. The owner goes back to the survey to see what it said about that band. It says nothing. There are no readings in the document at all. Just the word "sound".

Hold that boat in mind, because it is the whole argument. Ultrasonic thickness testing exists to close exactly this gap, and the reason it so often fails to is that surveyors treat the gauge as a device that delivers a verdict. It does not. It delivers numbers. The skill is in where you take them, how you read them, and how you write them up so they cannot be argued with later. That skill is what separates a narrowboat surveyor from a yacht surveyor holding a UT gauge. This is the deep-dive sequel to our narrowboat survey primer: where that piece set out how inland steel work differs from coastal GRP, this one is about the single technique the entire hull finding rests on.

The standard that does not exist

Start with the uncomfortable part. There is no published minimum hull thickness for a privately owned narrowboat. None. The figure most often quoted in the yard, "4mm at survey", has no formal standing in any code or regulation. An IIMS discussion paper by Tom Keeling and Peter Brookes, both inland surveyors, found that surveyor opinion on the acceptable minimum "varies wildly", with some quoting 3mm, some a percentage loss, and some declining to name a figure at all.¹

The 25% wastage allowance you may have heard is real, but it is borrowed. It comes from commercial vessel survey practice, usually cited via MCA fishing-vessel guidance, not from any narrowboat code. Applying it to a private steel boat is a surveyor's choice, not a rule. The same paper records insurers running entirely different thresholds from each other: one declines a 5mm hull showing pitting, another applies a blanket number, a third reflects historic practice with no stated basis at all.

The practical consequence is precise. If there is no agreed benchmark, then a finding graded against a benchmark is a finding graded against your private opinion. The one thing nobody can dispute is what you actually measured. So that is what the survey must contain: the readings, by zone, in a table. The surveyor of our 1992 boat had no number to stand behind, because there were no numbers. A number table is defensible. "Sound" is not.

Know what "original" was before you call anything wastage

Narrowboats are commonly built to a shorthand such as 10/6/4 or 10/6/5/4: roughly 10mm baseplate, 6mm hull sides, then 5mm or 6mm cabin sides and 4mm roof. Treat that as a convention, not a guarantee. Builders vary, cabin sides are often 6mm, and older boats vary enormously. A 1980s Springer might have started life at 6mm or less on the bottom. The build certificate or the builder's specification governs, and without it you cannot calculate percentage loss honestly, because you do not know the starting point. State the nominal where you know it, and say so plainly when you do not.

There is a subtlety here that catches people out. Steel plate is rolled to a tolerance, and that tolerance is not symmetrical. Under EN 10029 a nominal 6mm plate can be supplied around 0.4mm under nominal under the common tolerance class, and it is the under-tolerance, the minus side of nominal, that matters to a surveyor.² So a reading of 5.6mm on a "6mm" side does not prove half a millimetre of corrosion. The plate may have left the mill at 5.6mm. Treat early, small shortfalls as the plate's own history, not as wastage, until the pattern across the whole zone tells you otherwise.

The instrument, and what defeats it

Two techniques sit inside the gauge in your hand. Single-echo measures from a bare, clean metal surface through couplant, so it means grinding the blacking back to bright steel at each test point. Multiple-echo, the method in a Cygnus and the reason it dominates narrowboat work, gates on consecutive steel back-wall echoes, so it reads a reliable metal thickness straight through normal coating systems: a few millimetres of paint, bitumen or epoxy, with the blacking left intact.³ That is what you want on a survey. Fast coverage, no damage, no argument with a broker about grinding patches on a freshly blacked hull.

Either way, the number is only as good as the metrology behind it, and that is the part a casual operator skips. Set the gauge to steel velocity (around 5920 m/s); a mis-set velocity quietly falsifies every reading you take. Verify the instrument against a reference of known thickness, a cal block or step wedge, at the start and again through the job, and record that you did. Use enough couplant; with multiple-echo a dry or starved contact will not lock at all rather than quietly read low, so a missing reading is the warning, not a wrong one. Where you do grind a patch for single-echo, that bare spot is itself a finding to note. A defensible UT survey is one where the reader can see the gauge was set right and checked, not just that a figure was written down.

Multiple-echo is not infallible, and its failures are informative. It needs two clean back-wall echoes to settle, and it will refuse or mislead where the back face is heavily pitted, where laminated rust or scale sits on the inside of the plate, or where spray-foam insulation breaks the acoustic path from within. Blistered blacking and thick, multi-layered old bitumen at the waterline, the everyday culprit, do the same. A broad, dished area of loss reads low and clean, but a sharp, narrow pit may refuse to lock at all, and that refusal is itself the finding: it usually means the back wall is no longer flat. Record which mode you used and flag any zone the instrument could not read, exactly as the condition survey an underwriter relies on expects you to log what could not be assessed rather than quietly pass over it.

The grid: hunt the minimum, not the mean

Here is the single habit that would have caught the boat at the top of this piece. Corrosion on steel is rarely uniform. You get general wastage, where a whole plate thins evenly, and you get pitting, where the average looks healthy but isolated craters drive deep. A gauge swept casually across a plate reports something near the mean. The mean is not what sinks a boat. The deepest pit is.

So the grid is not decoration. Take a structured pattern of readings across each plate, several per plate as a baseline and far more where you are suspicious, and within a worrying area close the spacing right down to chase the lowest figure. A thorough narrowboat hull commonly runs to dozens, sometimes low hundreds, of points. Report the minimum you find in each zone, not an average, and state how many readings the zone rests on. One low number among twenty is not noise to be smoothed away. It is the finding.

Prioritise the zones the first diagram flags, because corrosion has favourite places. The waterline band wets and dries and stays oxygenated, and it is a frequent perforation site on the topsides. It is exactly where the metre-long band on our 1992 boat ran. The baseplate corrodes from two sides at once: the canal bed below, and, more insidiously, trapped bilge water sitting under the ballast inside, where nobody looks and where the loss is often worse than on the visible outer face. The bow rake takes grounding damage and loses its blacking first. Add the swim and uxter plate at the stern, the chine welds where base and side wastage compound, the steel around skin fittings and the weed hatch, and any plate near a tired anode. Test alongside welds rather than on them: weld metal's coarse grain scatters the beam, and readings taken each side of a weld also catch the heat-affected zone, where corrosion tends to concentrate. A clean sweep of the easy midships sides while the bilge baseplate and the chines go unread is precisely how a boat passes a survey and then fails one.

Two things must not be missed while you gauge. If the hull has already been overplated, your gauge reads the combined sandwich or only the new doubler, not the wasted original plate trapped beneath, where moisture caught between the layers keeps eating. Identify existing overplating and say what it does and does not tell you. And report the anodes in their own right: material, percentage wasted, and whether each is soundly welded on, electrically continuous, and not painted over. Magnesium is the correct choice in the low-conductivity fresh water of the canals, where zinc and aluminium cannot drive enough current to protect. A boat that lives in salt or brackish water, the tideway, an estuarial or coastal mooring, needs zinc or aluminium instead, because magnesium over-protects and wastes fast in conductive water. Note which the boat carries against where it is kept. The anodes are central to how the hull has survived this far.

Grading what you find

You now have a table of minima by zone. Classify, and show your working. Separate general wastage (a whole zone down a millimetre, gradual, monitorable) from isolated deep pitting (a 2.9mm crater in otherwise sound plate, far more urgent), because the two carry different consequences at the same reading. And weigh whether the loss is live or historic: two plates measuring 4mm grade differently if one sits under sound blacking and the other is actively weeping at the waterline, as ours was.

Use the standard Category A, B and C framework and tie each grade to the measured number, not to a private rule of thumb. Plate at or near perforation, or so wasted the insurer is likely to require remedial work, is a Category A finding the buyer must act on. The band on our 1992 boat belonged there, and the original survey never graded it because it never measured it. Significant wastage that warrants closer monitoring, a defined re-survey interval, or an overplating quote is Category B, a band wide enough that you should say which end of it you mean. Light, early loss against the original specification, with no structural concern, is Category C, recorded as a baseline for the next survey to measure against. Where you apply a threshold, name it and name its source, so the reader can see it is a judgement and not a law.

One decision sits behind all of it: drydock or afloat. A proper UT grid wants the boat out of the water so the external shell is reachable. Afloat, you are limited to internal readings through bilge access and floor plates, the external below-waterline plating is out of reach, and you must say so. It is also worth one line for the buyer's benefit: a valid Boat Safety Scheme certificate says nothing about hull thickness, so it is no substitute for this. The same discipline that makes a keel attachment finding hold up under challenge applies here, and the hull condition you record drives the boat's valuation and insurability directly. The evidence chain is the deliverable.

A grid of recorded thickness readings by hull zone, each tied to the instrument that took it and the access that allowed it, is a finding an underwriter can act on and a buyer can trust. "Hull sound" is neither, as the owner of our 1992 boat found out eighteen months too late. Marine Inspect's narrowboat checklist gives ultrasonic thickness readings their own structured item, recorded by named zone with the gauge make, model and probe logged alongside, and prompts you to flag any zone the foam or the waterline put out of reach. See how it reads in a finished report in our sample condition survey. The numbers go where the report can use them.