Tunnel and Aqueduct Inspection by ROV
By Ray Whitfield, ADCI-certified commercial dive supervisor. Reviewed by Dana Whitcombe, technical reviewer.
Underwater Inspection | Commercial Diving Contractors Alabama I remember the first time I put an ROV into a 2.4-meter-diameter raw water line at 40 meters depth. The boss had insisted on a video survey of a cracked liner, and I'd been pushing for a diver because the line was 'simple' — straight run, one gate valve, no bends. The ROV made a fool of me inside ten minutes. That little machine slipped through the valve opening, held itself in the current with a couple of bursts of its thrusters, and gave us a clear view of a rubble pile that had been building up for years. We never saw it from the gate chamber. I've been a fan of tunnel and aqueduct inspection by ROV ever since. In this post I'll walk you through what actually works for tunnel inspection ROV work — rigging, deploying, running the line, and making the data useful. It's not a marketing brochure. It's a field guide, plain and simple.
Start With the Water, Not the Robot
Before you size the ROV, you need the asset's basic numbers. What's the diameter, the typical flow velocity, the minimum depth the ROV has to handle? Is the tunnel gravity-fed or under pressure? Is it a municipal water tunnel, a power plant intake, or an irrigation aqueduct? Each one has its own set of access points and hazards. Call it a rule of thumb: if the tunnel is under pressure, you're likely working through a gate chamber or a riser. If it's free surface flow, you've got upstream or downstream manholes. You need a drawing of the tunnel, even if it's 40 years old. And you need to know if the water is clean or full of silt. A ROV in murky water is a different beast than one in gin-clear reservoir water. I've seen plenty of pilots show up with a shiny ROV and no idea what the flow does. In a crude sense, the ROV has to fight the current. Most small ROVs (call it the 10-20 kg class) can handle maybe 2 to 3 knots of flow in a straight line, but it'd rather not. If your tunnel runs at 4 knots, you're going to need a heavier unit with more thrusters, or you'll be crawling along the wall with a tether and hoping.
The ROV Rig for a Tunnel Job
For most tunnel inspections, I run a VideoRay Pro 4 or a Teledyne SeaBotix LBV, depending on the size and complexity. These are compact micro-ROVs, not the giant work-class units. They're rated for 300-400 meters, which is plenty. They've got a tether that's fiber optic or copper, and you need a tether management system if the run is long. For really long aqueducts, you might need a mid-sized unit like a BlueROV2 with custom thrusters, but that's a build-your-own route. Payload matters more than the vehicle itself. You want a high-definition camera with a good zoom and pan-tilt, plus a secondary wide-angle camera. Lighting is critical — LEDs, not halogens, unless you want a melted housing. In a tunnel, the walls are close and often dark. You'll have backscatter from silt and particles, so you need lights that mount away from the camera housing. Add a scanning sonar if you need to work in zero visibility. I like the Blueprint Subsea Oculus or the Teledyne BlueView. You also need a depth sensor, a compass, and a gyro. GPS won't work underground, so you're flying blind in a technical sense. The ROV pilot watches the compass, the depth readout, and the video. On top of that, you need an acoustic positioning system if you need to know where you are along the tunnel. Most of the time we get by with tether marks and a good chart, but if the tunnel's a few kilometers long, you'll want a transponder. Here's the digression. I remember a job at a power plant intake where the client insisted on no divers in the penstock due to safety protocols. So we launched an ROV at a downstream gate. The problem was the gate chamber had a 30-centimeter clearance around the ROV on the launch frame. We had to rig a special cage to guide it into the flow. If you're doing this work, always, always check the access hatch dimensions before you order the vehicle. An ROV that fits in the shop can easily be too fat for the site.
Running the Line: Navigation and Data
Once the ROV is in the water, the job is all about maintaining a good, steady run along the tunnel axis. You want to keep the vehicle off the walls and off the invert. In a concrete tunnel, the wall will have joints every few meters. You want to record those joints in sequence with a running timecode, so the engineer can map defects to footage distances later. We run a series of longitudinal passes — low, mid, and high if possible. That's the simplest way to document conditions. On the way in, the ROV goes upstream, fighting the current, so you get a slower, more controlled pass. On the way back, with the current pushing you, you can cover the other side. We log video continuously, but we also take stills every second or so via the software. The camera's timecode is synced to the ROV's navigation, so every image has a depth and heading stamp. If visibility is bad, that's where the sonar earns its keep. A scanning sonar can give you a pretty good picture of the tunnel wall even in zero visibility, and it's how you spot large debris, rock fall, or delaminated lining. But sonar is not video. You can't see fine cracking or corrosion. So if the client needs a detailed condition assessment, we might have to do it the hard way — slow runs, extra lights, and maybe even a diver if the tunnel is safe enough and the ROV can't get close enough. There's a real art to not getting stuck. Tunnels are full of old pipes, cables, and debris. You can snag the tether on a protruding bolt, and then you've got a problem. The trick is to keep the tether as tight and straight as possible. The pilot and the tender have to communicate constantly through the tether tension. And if the current is strong, the tether sags, creating drag. That drag can pull the ROV sideways into the wall. We use a tether float system to reduce drag in larger tunnels.
Aqueducts Are a Different Animal
Aqueducts are open-channel or low-pressure tunnels, and they have their own set of quirks. The water can be flowing fast and shallow, sometimes only a few feet deep. An ROV that's at the surface may not have enough clearance above it. Or, if the tunnel is only half full, the ROV can get wedged under an air pocket or slam into a roof of rock by a hydraulic jump. In a free-surface aqueduct, we often run the ROV with a floatation collar so it stays near the surface, and we launch it from a drop pipe or access shaft. You need to keep the ROV in the main flow and avoid getting sucked into a low-flow area where it might settle into silt. The current can be deceptive. A gentle surface might hide a 3-knot undercurrent around a bend. Also, aqueducts often carry raw water for municipal supply. That means you have to deal with chlorination or other water treatment conditions. Avoid any oil leaks from the ROV. Check all the O-rings. Keep a clean tether. If you're working in a drinking water supply, you'll likely need a washdown station for the ROV and approval from the operator's water quality team. That can add days to the schedule — depends on the plant, really. We've also seen aqueducts with fish screens or grates that can catch the ROV. Those are places where a live feed to the control room pays off. The client's engineer can direct you to avoid a critical structure while you wiggle around it.
Delivering a Report People Can Use
The ROV footage is only part of the deliverable. The report is what gets the repair budget approved. We put together a spreadsheet with the timestamp, distance along tunnel, depth, heading, and a description of each defect — crack, spall, joint offset, debris, sediment build-up. We include a short video clip for each defect, with the ROV speed and direction marked. We also make a simple log file that can be imported into the client's asset management system. Don't try to invent a fancy new format. Clients can use a CSV or a KML if they have GIS. We always include a preface that spells out the limitations — what we could and couldn't see, what the sonar resolution allows, and what we'd recommend for further inspection. One of the biggest lessons I've learned is to give the client an honest confidence level. If the water was too silty for good video, say so. If the sonar shows a big void but we can't confirm the structure, say that. The whole point of a tunnel inspection ROV operation is to give the owner reliable information. That means not sugarcoating the data, and not promising something the ROV didn't actually deliver. And when the job is done, the ROV gets a fresh set of sacrificial anodes and a thorough rinse in fresh water. The gear has to live for the next job. That's not glamorous, but it's part of the business.
Frequently Asked Questions
How long can an ROV work inside a tunnel? Battery life varies, but most small ROVs run 2 to 4 hours on a charge. Tunnel inspections often take longer than that, so you need a spare battery or a generator to recharge. Some ROVs can be powered from the surface through the tether, which gives you unlimited endurance. What visibility do you need for a successful tunnel inspection? Ideally, at least a meter or two of clear water. If visibility is less than 50 cm, video becomes nearly useless. That's when you rely on sonar and touch-based methods — but if you're using sonar, you'll need to explain that the inspection is image-based, not visual. How much does a tunnel inspection ROV cost to run? Costs vary widely. A small inspection ROV package with operator can run between $1,500 and $5,000 per day, depending on the site, equipment, and whether you're including reporting. That's simpler than diving, which has more complex logistics and higher safety requirements. Can an ROV replace divers entirely? No. ROVs are great for routine inspection and data collection, but they can't do hands-on repair work. If you need to replace a bolt or patch a crack, you either need a diver or a work-class ROV with manipulators, which is much more expensive. The key is to use the ROV for reconnaissance, so you know exactly when and where you need human hands.