Ultrasonic Testing for Condition Monitoring

Ultrasonic testing monitors machine condition by detecting the high-frequency sound that faults produce. A bearing running dry, gas escaping a valve, or electricity arcing across a gap all emit ultrasound above what your ears can hear. The instrument shifts that sound down into an audible range and puts a number on it, so one technician with one tool can check bearings, valves, steam traps, and electrical cabinets on a single route.

This matters because ultrasound is frequently the earliest of the condition-monitoring signals. Friction, turbulence, and electrical discharge all start quietly and high in frequency long before they generate the heat a thermal camera needs or the amplitude a vibration alarm needs. A bearing an hour into grease starvation is already screaming in ultrasound while its temperature and its 1x vibration are still normal. This guide covers the two modes of ultrasound, what each one inspects, and where it fits among the other tools on the truck.

What is ultrasonic testing in maintenance?

In maintenance, ultrasonic testing means listening to the ultrasound a running machine or system produces to judge its condition, a form of acoustic condition monitoring, not the flaw-detection ultrasound used to inspect welds and castings. The same physical fact underlies every application: friction, turbulence, and electrical discharge all radiate energy well above the roughly 20 kHz ceiling of human hearing. Instruments capture that band and heterodyne it down so a technician can hear a bearing tick, a valve hiss, or an arc crackle, and can log a decibel level to trend over time.

Ultrasound is short-wavelength and directional, which is what makes it useful on a busy floor. Low-frequency machine noise fills the plant, but ultrasonic energy drops off fast with distance and does not travel around corners, so a fault reads loud only when the sensor is near and aimed at it. That directionality lets you isolate one bad component in a crowded skid, and the same short wavelength is why a decibel reading taken the same way each week trends cleanly as a component degrades.

Airborne vs structure-borne, what is the difference?

The difference is whether the sensor touches the machine. Airborne ultrasound uses a non-contact sensor to pick up ultrasound traveling through the air; structure-borne ultrasound uses a contact probe, or stinger, pressed against the asset to pick up ultrasound traveling through the metal. Each mode suits a different family of faults.

Airborne ultrasound is the leak-and-electrical tool. It finds compressed air, gas, and vacuum leaks, tests steam traps for pass-through, and, held near, never touching, energized gear, detects the arcing, tracking, and corona discharge that precede electrical failures. Structure-borne ultrasound is the mechanical tool. Pressed on a bearing cap it hears friction and impacting long before heat or amplitude rise, so it catches lubrication problems and early bearing wear, and it hears cavitation and internal leakage in pumps and valves. The leak-hunting side of airborne ultrasound is covered in depth in ultrasonic leak detection; this guide focuses on the mechanical and electrical condition-monitoring side.

What can you inspect with ultrasound?

You can inspect anything that leaks, rubs, impacts, or arcs. That covers a surprising share of a plant's failure modes, which is why one ultrasound instrument earns its keep across so many routes.

Two applications carry outsized savings. Steam traps fail silently and expensively: the Department of Energy reports that in systems left unmaintained for three to five years, 15 to 30 percent of traps may have failed, many stuck open and blowing live steam. Ultrasound tests each trap live, telling a healthy cyclic discharge from a failed-open blow. On the electrical side, listening for arcing and corona in switchgear finds the discharge that leads to insulation breakdown and flashover, a safety and reliability win that a de-energized inspection would miss entirely.

Why does ultrasound catch faults so early?

Ultrasound catches faults early because the physics of an incipient fault is high-frequency and low-energy, exactly ultrasound's band. A bearing losing its lubricant film starts with metal-to-metal micro-contact that radiates high-frequency friction energy. That energy is real and detectable long before it accumulates into the heat a thermal camera reads or the low-frequency amplitude a standard vibration alarm trends. The same is true of an electrical fault: a partial discharge crackles in ultrasound well before it carbonizes enough insulation to make heat.

This places ultrasound at the leading edge of the failure-development, or P-F, interval, the window between the point a fault becomes detectable and the point it becomes a functional failure. Detecting a problem earlier in that interval buys more time to plan the repair. Ultrasound and vibration are complementary rather than competing: ultrasound often raises the first flag and confirms lubrication state, while vibration analysis and its spectrum diagnose the specific mechanical defect and how far it has progressed. Together they cover more of the interval than either alone.

How do you set up an ultrasound route?

An ultrasound route is a repeatable list of points, each read the same way every time so the decibel trend means something. Absolute readings vary with machine and sensor; the value is in the trend and in comparing like assets. Build the route to keep readings comparable.

1. Pick the assets and points. Choose bearings, steam traps, valves, and electrical cabinets where early warning pays off. Mark a fixed contact point on each bearing and a fixed standoff distance and angle for each airborne target.
2. Baseline every point. Record a decibel level and a short sound sample on a known-good machine. The baseline is what every future reading is judged against.
3. Standardize technique. Same probe, same contact pressure and location, same frequency setting, same distance and aim for airborne points. Inconsistent technique creates noise that looks like a fault.
4. Read on a schedule. Walk the route at a fixed interval, logging level and sound at each point. Listen as well as read: the character of the sound tells failure mode where the number tells severity.
5. Set alarms off the baseline. Alarm on a decibel rise above baseline, not on an absolute number. A common rule of thumb flags a several-decibel increase for review and a larger jump for action.
6. Confirm before acting. When a point alarms, confirm with the sound signature and a second tool where it helps, vibration for a mechanical defect, infrared for an electrical hot spot, before writing the work order.
7. Trend and close the loop. Store readings so each point shows a history, and tie alarms to work orders so a rising trend turns into a planned repair, not a surprise failure.

How does ultrasound fit with vibration and thermal?

It fits as the wide, early, low-cost first pass, with vibration and thermal as the deeper diagnostics behind it. No single technology sees every fault, and the strongest programs layer them by what each does best. Ultrasound covers the most application types with one inexpensive instrument and often gives the earliest warning; it is the natural front line of a route.

Where a point alarms, the follow-up tool depends on the fault. For a suspected bearing or gear defect, vibration analysis and its bearing defect frequencies pinpoint the component and stage. For an electrical hot spot, infrared electrical inspection confirms and grades it. For the stress-wave energy of advanced crack growth, acoustic emission monitoring extends into a related band. Ultrasound rarely replaces these; it points them at the right machine, which is where its return comes from. For how these tools combine into a strategy, see condition-based maintenance and predictive maintenance.

Where ultrasonic testing fits your reliability program

Ultrasonic testing is one of the highest-payback entries into condition monitoring because a single affordable instrument spans leaks, bearings, lubrication, valves, steam traps, and electrical faults, and it usually gives the earliest warning of any of them. It is often the first predictive tool a plant buys and the one that touches the most routes. The catch is the same as every other technique: readings only matter if they are trended and acted on.

That is a data problem. A decibel reading in a technician's notebook, disconnected from the machine's vibration trend, work-order history, and run hours, cannot show whether a bearing is degrading or a repair worked. Pulling ultrasound readings together with the plant's other signals into one operational view is where machine monitoring platforms like Harmony come in, so a rising ultrasound level on a bearing lands next to its vibration trend and its open work orders instead of in a silo. It layers onto the systems you already run, with no rip-and-replace. See how the platform works or read the CLS case study.