Underwater NDT: Ultrasonic Thickness Testing Explained

By Ray Whitfield, ADCI-certified commercial dive supervisor. Reviewed by Amanda McCallister, editor.

Underwater Inspection | Commercial Diving Contractors Alabama I'll never forget a bridge pile inspection in the Ohio River. The thickness gauge read 0.412 inch on what was supposed to be 0.500-inch steel. I re-zeroed the probe, checked my calibration block, and shot it from three angles. Still 0.412. That number saved someone from a lot of trouble, because we found the member was losing about nine mils a decade. Underwater ultrasonic thickness testing is the only way I know to get that number without tearing the structure apart. It's not glamorous. It's a diver with a probe, a cable, and a topside unit reading a screen. But it tells you whether that dam gate will last another season or whether the hull of a chemical barge needs a patch before it leaks. Call it UTT, or UT thickness, or just "pulse-echo" if you're talking to the old-timers. The principle doesn't change. You send a sound wave through the metal, it bounces off the far side, and the round-trip time tells you how much material is actually there. A lot of divers can run the tool, but understanding what it's telling you is a different skill. That's what this article is about. I've been doing this for 28 years on inland and industrial jobs — dams, bridges, water tanks, power plant intakes. I'll walk you through how it works, when to use it, and how to make sure you're not fooled by the readings.

What Underwater Ultrasonic Thickness Testing Actually Is

Underwater ultrasonic thickness testing is nondestructive testing (NDT) that measures remaining metal thickness from one side. We use a handheld probe with a transducer that sends high-frequency sound waves — typically 2.25 MHz for thick steel, 5 MHz for thinner pipe, sometimes 10 MHz for thin wall. The sound travels through a coupling medium, hits the far wall, and reflects back. The instrument measures the transit time and converts it to thickness using the acoustic velocity of the material. For steel, the sound speed is roughly 0.232 inches per microsecond, give or take. The gauge does the math. You just need to get a good, repeatable echo. In a dry shop, that's easy. Underwater, it takes a bit more patience. We use direct contact probes with a water-resistant gel, or sometimes a "stand-off" probe that uses the water itself as the couplant. Depends on the plant. The critical difference underwater is that you're not just fighting corrosion. You're fighting marine growth, barnacles, rust, paint, and often poor visibility. The surface has to be cleaned down to bare metal. Not bright steel, just clean enough that the sound gets in without bouncing off a loose scale or an air pocket.

When UTT is the Right Tool

I've spent a lot of time doing visual inspections and underwater photographs. Those have their place. But where UTT shines is when you need numbers for a structural assessment. Coastal bridges, lock-and-dam structures, bulkheads, pilings, ship hulls, mooring chains, and intake screens. If it's steel and it's wet, I'll likely be asked to take thickness readings. Sometimes it's a baseline survey. Sometimes it's following up on a visual indication like scabbing or a leak. Rarely it's in response to an incident — a tank overfilled, a vessel struck something, a piling has a bend. The engineer wants to know what the remaining wall is. That's a question for UTT. And there are times when UTT is not the answer. If you're looking for surface cracks, you'd use magnetic particle testing or eddy current. If you're looking at welds, ultrasonic shear wave or radiography. But thickness? That's our tool. I've had contractors ask me to "UT the weld" when they meant "check the weld for cracks" — that's a different procedure. We need to be clear.

How We Do It Underwater: Equipment and Technique

The basic setup is a probe, a long cable, and a data logger. I've used the Olympus 38DL PLUS for years. It's rugged, has a real-time display, and lets me store readings with comments. The probe I use most is a 5 MHz straight beam, 0.25-inch element. For rough and pitted surfaces, I'll use a 2.25 MHz with a bigger footprint because it averages over more area. Preparation is half the job. The diver scrapes and brushes the surface until it's clean, then grinds if needed. In heavy marine growth, that's a chore. You don't need a mirror finish, but you need enough removal that the probe sits flat. I've seen guys try to save time and shoot through a patch of barnacles. You get a thickness reading, but it's wrong. It's often reading the barnacle plus the steel, or worse, just air. That waste of time costs more in the long run. For reading, we gate the first back-wall echo. Some gauges let you set a range so you don't get fooled by echoes from paint or scale. We calibrate on a known thickness block — I keep a 1.000-inch and a 0.250-inch block on the boat. Each shift, re-check zero. The water temperature, steel temperature, even the pressure at depth can change the velocity slightly. That's why we calibrate on site. If the reading is between 0.400 and 0.410 on a nominal 0.500 plate, we say it's about 0.400, with a tolerance of maybe ±0.005.

Reading the Numbers: Accuracy, Pitfalls, and Reporting

So you've got a bunch of numbers. Now what? First, watch for laminations. A lamination is a plane of separation in the steel, usually left over from the rolling process. It'll give you a false thin reading because the sound reflects off the inclusion, not the back wall. On the initial bridge job, that cross-member was actually 0.500 at the lamination-free area, but the area we tested was thinned by corrosion. We flagged it as "minimum thickness" and recommended an expanded grid. Pitfalls come in a few flavors. Paint: if the paint is intact and well bonded, you might measure paint plus steel. The standard trick is to set the gate on the second back-wall echo. But more common is the false echo from a loose or corroded surface. And a rookie mistake: not realizing the probe angle. The transducer isn't meant to be tilted. The contact surface must be parallel with the back wall. If the surface is curved, you use a different shoe. Reporting is where the value shows. We don't just write "0.412 minimum thickness". We plot the readings on a sketch or a CAD photo, mark the low spots, and give the engineer a table with the nominal thickness, the measured minimum, and the estimated corrosion rate. On a recent penstock job, we took 140 readings across a 60-foot diameter section and found a patch where the wall had gone from 0.750 to 0.620 in ten years. That's 13 mils per year. The owner used that number to prioritize a replacement.

Getting Qualified: What It Takes to Do This Work

If you're a commercial diver, you already know your certification: ADCI, DCBC, or IMCA depending on your region. But for UTT, you need nondestructive testing certification too. Most commonly, that's ASNT Level II in Ultrasonic Testing, with specific training in thickness gauging. It's not as deep as a full UT weld-testing course, but it covers calibration, application, and interpretation. In the U.S., you might also need to meet the requirements of SNT-TC-1A or CP-189. Some clients will want an operator qualified to a specific code, like AWS or API 510 for vessels. And on top of that, many jobs require a degree of inspection experience beyond the diving. For instance, a bridge inspection might require you to work under a structural engineer who has been trained in fracture critical inspection. The diving is the easy part. I've had dive school graduates think they can grab a thickness gauge and be a tech. The gauge is a tool, not a brain. You have to know how to identify when the reading's lying to you, and when to say "I need more prep" or "I need a different probe". That comes with time. Call it five hundred hours of supervised readings, give or take. And even then, you'll still see something new that'll make you question your numbers.

Frequently Asked Questions

How accurate is underwater ultrasonic thickness testing? In good conditions, with clean steel and proper calibration, we're accurate to ±0.005 inch (0.13 mm) on mild steel. On painted or scaled surfaces, less so, sometimes ±0.02 inch. The accuracy also depends on the instrument and operator experience. Can you test through a layer of marine growth or paint? No. Well, sometimes you can test through a thin, well-bonded paint layer, but we don't trust it. We clean to bare metal whenever possible. Marine growth is a no-go. You'll get ghost echoes and false readings. How long does it take to do a survey? Depends on the size and the surface condition. A single reading can take five minutes if the area's dirty. A hundred-point survey on a bridge pier might take a full dive shift of two or three hours. The prep is the time consumer. What's the difference between ultrasonic thickness testing and acoustic emission testing? Thickness testing uses a probe to send and receive echoes, giving you a wall thickness reading at a specific point. Acoustic emission listens for sounds from cracks or other sources while the structure is under load. They answer different questions.