Start from the failure mode, not the sensor
Blade failures cluster in a few places: the root and its inserts, the spar caps, the shear web and trailing edge bond lines, and the leading edge where erosion or lightning has opened the laminate. Each of those failure modes changes something measurable, but not the same thing, and not at the same stage.
- A root that is loosening changes how one blade carries load relative to the other two, long before it cracks visibly.
- A spar cap wrinkle changes nothing measurable until it begins to delaminate, and then it changes local stiffness.
- A trailing edge bond line that is opening changes the blade's edgewise behaviour and, when it opens under load, its sound.
- Erosion changes the power curve, slowly, and nothing else a blade sensor can hear.
A system that measures root bending will see the first and not the fourth. A microphone will hear the third and not the second. The right question for any vendor is therefore which of your fleet's likely failure modes their sensor is physically able to register, and at what stage.
Root load sensors and blade root monitoring
Root load sensors measure flapwise and edgewise bending at the blade root, usually with strain gauges or fibre Bragg gratings bonded to the inner laminate, sometimes derived from pitch drive torque or current. They are the most common retrofit because the root is the one place reachable from the hub.
What they detect: changes in load asymmetry between the three blades, which is the early signature of root insert movement, root laminate cracking and lost bolt tension. They also register rotor imbalance from ice or mass loss, and they are used by the turbine controller for individual pitch control on some platforms.
What they miss: the whole outer blade. A trailing edge crack at maximum chord, a tip lightning strike or leading edge erosion produce no root signal until the damage is large. In our judgement blade root monitoring is a root and hub system, and it should be bought as one.
Distributed strain and fibre-optic sensing
Fibre-optic strain sensing runs a single fibre along the blade with many measurement points, so strain can be read at the root, at ply drops and in the outer blade from one installation. It is fitted in manufacture rather than retrofitted, because the fibre has to be laminated in.
What it detects: local stiffness changes wherever a measurement point sits, so a growing delamination or a spar cap crack near a point shows as a strain redistribution. It gives real load data that can also be used to validate the design assumptions after installation.
What it misses: anything between measurement points, and anything that does not change strain, which includes erosion, receptor damage and a bond line that has not yet started to open. The value is proportional to how many points were fitted and where, which was decided at the factory.
Accelerometers and natural frequency tracking
Accelerometers on or in the blade track its natural frequencies. Damage that reduces stiffness lowers those frequencies, and the change is in principle detectable.
What they detect: large stiffness loss, in practice near the root where the modes are most sensitive.
What they miss: early damage. In our judgement the frequency shift from a developing defect is small compared with the shifts caused by temperature, ice, and normal variation between blades, so the method flags damage late unless the baseline is unusually well characterised. It is better at confirming that something has changed than at saying what.
Acoustic emission and cavity microphones
Two different things are sold as blade acoustic monitoring, and they should be separated.
Acoustic emission sensors are contact transducers that listen for the ultrasonic bursts released when fibres break or an adhesive crack advances. They detect damage as it grows, which no camera can, but the sensor must be close to the source, so a long blade is only partly covered.
Cavity microphones sit inside the blade, usually at the root, and listen to airborne sound: the change in a blade's own noise when a crack opens or a trailing edge splits, and the impact of loose material on the shell. In our experience this is the most common commercial form of a blade condition monitoring system. One microphone at the root hears a long way along the cavity, but locates the source less precisely than a contact sensor, and its alarm needs an inspection to interpret.
What both miss: slow, quiet degradation. Erosion, coating loss, a wrinkle that has not begun to delaminate, and a bond line void that is not yet growing produce no sound.
When a blade condition monitoring system is worth installing
Sensors are a capital item per turbine plus a data service for the life of the fleet. In our judgement they pay in four situations:
- The blade type has a known structural weakness at the root, a bond line or a web, and the system measures the quantity that weakness changes. This is the strongest case.
- Access is expensive or slow: offshore, remote, or a site where a rope access team takes weeks to arrange, so the value of early warning is high.
- An insurer, lender or warranty position asks for continuous evidence, and the system produces a record that can be relied on.
- A serial defect is suspected and the owner wants to keep operating a monitored sample while the population is assessed.
They do not pay as a substitute for inspection. A fleet with a sensor system still needs periodic external imaging, and still needs an internal survey at the events that call for one, because the sensors do not see the surface and do not see the outer blade.
What to ask a vendor before signing
- Which physical quantity does the sensor measure, and which of our blade type's failure modes change that quantity?
- How much of the blade length is covered, and where is the nearest sensor to maximum chord and to the tip?
- What is the false alarm rate on a comparable fleet, and what inspection response does an alarm require from us?
- How long is the baseline period before the system can distinguish a change from normal variation, and what happens to alarms in icing or extreme temperature?
- Which real blade failures has the system been validated against, on which blade types, and can we speak to that owner?
- Who owns the data, in what format is it exported, and what happens to it if the contract ends?
A vendor who can answer the fifth question with specifics is selling a monitoring system. One who cannot is selling a sensor.
Frequently asked questions
What does a blade condition monitoring system actually detect?
It depends on the sensor. Root load sensors detect load asymmetry and the early signs of root failure. Acoustic sensors detect cracks that are growing or opening. Fibre-optic strain detects stiffness changes near its measurement points. None of them detects erosion, coating loss or a defect that has not yet started to move.
Is real-time blade monitoring better than periodic inspection?
It is different. Real-time systems catch events and growth; periodic inspection catches what is visible on the day, including the surface damage sensors cannot see. A fleet at risk needs both, in proportion to its failure history.
Can blade sensors be retrofitted?
Root sensors and cavity microphones can, because the root is reachable from the hub. Distributed fibre-optic sensing is laminated in at manufacture and cannot be added later.
How do I know if a monitoring system is worth it for my fleet?
Start from the blade type's failure history and known weaknesses, then ask which sensor measures the quantity those failure modes change. If the answer is none, the system will not see the failure that matters to you.
How Apex Wind can help
We are an independent blade engineering consultancy. Apex Wind sells no sensors, software, inspection services or monitoring products and takes no referral fees; for disclosure, our founder separately co-founded a company that builds crawler robots for internal blade inspection. In a transaction, our blade technical due diligence reads what the monitoring record could and could not have seen. For the teams who specify these systems and act on their alarms, our blade training covers how blades fail and which signals precede which failures. If you are deciding whether to fit a system to a specific fleet, contact us.