With higher coverage of instrumentation comes a higher probability of false positives, i.e. indication of failure due to sensor or communication failure, according to Anders Valland (pictured), Research Manager, Marintek Energy Systems and Technical...

By Ritesh Gupta

Condition monitoring is a vital initiative. If handled efficiently, it can pave way for fruitful results. The idea is: even if things are shaping up well, it’s always better to avoid complacency. For instance, one may put a lot of instrumentation into the machine in order to come to grips with probable known failure modes. But this still might not be enough as there will always be a significant portion of failures that are so-called hidden failures.

These categories of malfunction or breakdown are the ones that abruptly surface, without pre-warning. Some components either do or do not work and they do not demonstrate any in-between deteriorated state behaviour. A high coverage of instrumentation is also very expensive, and all sensors may eventually fail.

“With higher coverage of instrumentation comes a higher probability of false positives, i.e. indication of failure due to sensor or communication failure,” says Anders Valland, Research Manager, Marintek Energy Systems and Technical Operations.

Valland continues: “There are several approaches in use to alleviate this, such as flight leaders where some turbines are fully dressed sensor-wise and are expected to provide coverage for the remainder, less instrumented machines.”

He says certain approaches try to figure out a “cost-effective” suite of sensors, and some try to keep it to a minimum. “In any case there is a profound need for a methodological approach which enables overview of an entire wind park coupled with the ability to drill down into details on each turbine,” says Valland.

Expanding condition monitoring

The demand for sustaining higher operational availability and the growing value of offshore wind turbines, as well as their restricted accessibility, call for evaluating options pertaining to expanding condition monitoring to other wind turbine components. Valland says the area where this will be most evident is with offshore turbines, as the cost of accessibility is significant.

He recommends that an operator must always start with the basic question: “What decisions will I be making with regard to operation and maintenance of my wind park?”

“The answers given to this question provides the foundation for determining what information you really need, and the required quality of measurements and sensors. It really does no good having a high resolution, high sampling rate static pressure sensor if what you need is a rough estimate of differential pressure. It is necessary to be pragmatic related to what the latest and greatest in sensors can do compared to what you need to know in order to support your decisions,” says Valland.

The turbine blades are among the most sensitive parts of the turbine, and they are a challenge to instrument properly.

“Important failure modes of the blade are surface damage due to bird strikes, icing, drop and particle erosion, and the most dreaded – delamination. Being composite structures they are difficult to inspect using traditional methods (e.g. ultrasonic) and you cannot have remote instrumented coverage of this sort. Using accelerometers and embedded fiber sensors will provide some clues to structural failures but there is still a lot of work to be done on how well this can be covered,” says Valland.

He says that the use of drones and robots to inspect both inside and outside the turbine components is an emerging technology which holds a lot of promise.

“One line of thought should also be to challenge the need for high efficiency blades that are all the more sensitive to surface damage and abrupt loads – it might be more cost-effective in the long run to have less efficient and more robust blades,” he comments.

Direct-drive and other low speed applications are a challenge with regard to monitoring main shaft fatigue and cyclic loading of other components.

Optimisation

Regarding how to aptly work on extending condition monitoring to other wind turbine components, Valland says one should ask the question: “What decisions will I be making with regard to operation and maintenance of my wind park?” By looking into this question and the potential options, it opens up how a wind park's O&M is handled.

He explains: “There is no absolute answer to this, as it relates both to the physical setup/ layout of your wind park as well as your organisational culture and behaviour. I believe the wind industry has a good opportunity to learn from the efforts within integrated operations (IO) in the offshore oil & gas industry – one may check www.iocenter.no. I also believe that the current level of maturity of the wind power industry is a major challenge to the ability to incorporate the IO way of working. But it is worthwhile to go that way.”

In this sector, the SKF condition monitoring system, SKF WindCon, is capable of monitoring vibration acceleration on all mechanic components on a wind turbine, way beyond the monitoring of just the bearing, says Stefan Karlsson, Head of Marketing & Strategic Development, SKF Renewable Energy Business Unit.

In fact, wind farm operators are using the system for monitoring the entire drive train including gears of the gearboxes and generators.

“Results from SKF WindCon is not limited to failures in the bearing itself, but through enveloping of vibration acceleration and data sampling the SKF WindCon provides detailed information on initiating failures on all mechanical parts in the drive train, enabling the system to detect emerging failures at early stage also in other areas of the drivetrain of the wind turbine,” explains Karlsson.

Measurements

Offshore wind turbines often involve a lot of standard measurements. Some of these instruments could also be utilised for condition monitoring through suitable trending and analysis methods.

It is worth considering which information could any measurement provide compared to the information you need to support your decisions.

“When using “standard measurements” it is imperative to understand that they were designed for a purpose which may be different from what you want in a condition monitoring perspective. This relates, but is not limited to stability, resolution, sampling frequency and accuracy. You will most probably need to use such measurements in a calculation or model, and it is important to understand then what assumptions the model is based on with regard to these properties of the measurements. How do you handle data spikes, instability in measurements, loss of data in certain time frames etc.? Does your model require raw vibration data, or is it based on processed values? How are such measurements stored, how is data quality assessed and is such metadata actively used in models?” Valland says.

The utilisation of existing measurements is important because it alleviates, to some extent, the need for specialised sensor systems. In some cases it can give a very useful assessment of technical condition, and it should always be used as a valuable supplement to any specialised sensor system.