Video-Based Motion Analysis

Reliable Verification of Design Parameters at Prototype and Serial Turbines

For reasons of cost reduction, modern large wind turbines and blades are designed to use less material and to make use of new materials. They are designed using new simulation tools and smaller safety factors, going beyond established knowledge. Therefore, turbine type certification requires simulation validation by real-life prototype measurements to prove that design dimensions (e.g. clearance between blade tip and tower), parameters and assumptions assure safe operation during the planned service life. For safe and reliable operation, every serial turbine also has to comply with certified design parameters. This keeps O&M costs and lifetime consumption low, despite unmanned 24/7 operation in remote areas. Hence, highly accurate but also cost efficient and safe measuring methods are needed to avoid excess fatigue loads, such as those from resonance issues related to the tower’s natural frequencies or intolerably high rotor imbalance and blade angle deviation. For these applications, video-based analysis is a suitable method to measure motion and vibration.

By Anke Grunwald, Christoph Heilmann and Michael Melsheimer, BerlinWind GmbH, Germany

For reliable, long-lasting and safe operation, as well as low O&M costs and avoidance of damage-related standstill, it is essential to validate design simulation by prototype measurements. In addition, serial turbines have to comply with design parameters relevant for fatigue and lifetime consumption. This requires accurate and effective field measurements.

Wind Turbine Safety During the Entire Service Life
Although in public many people talk about wind turbine costs, it has to be emphasized that for two decades safety and reliability have been the most important design criteria in the design standards for wind turbine certification and type approval. But often turbines go beyond established engineering and material science and this gave the challenges of dealing with:

• Unknown long-term fatigue loads from environmental and operational conditions
• Unknown long-term material properties of new composite materials
• Necessity of lightweight structures for cost reduction
• Very fast growth of serial turbine size
• High computational effort needed for design life simulation of the entire turbine system and a large operational range.

Therefore, field measurements are the key for design validation. In this context, video-based motion analysis has the advantage that the simultaneous motion and vibration can be analysed at numerous spots of the filmed structure without installing many single sensors.

Increased Fatigue at Many Serial Turbines
Statistics from field measurements at hundreds of serial turbines reveal that for a large proportion the design limit values of rotor imbalance (uneven mass distribution) and/or blade angle deviations are exceeded (Figure 1). The resulting higher vibration amplitudes of the tower–nacelle system cause problems for many components, and the entire turbine is liable to have increased fatigue loads, damage-related standstill and a shortened service life. These all increase O&M costs and reduce the return on investment. Detecting rotor imbalance by mobile expert vibration measurements in the nacelle using accelerometers is well established (see, for example, the annex on in-situ rotor balancing at wind turbines in the guideline VDI-3834 on vibration assessment at wind turbines, part 1 (ed. 2015)). However, this method is only reliable if other aspects like the potential falsifying impact of additional forces from aerodynamic imbalance (e.g. by blade angle deviation) are considered. The immediate availability of the result is a great advantage of the sensor-based method. Its main drawback is the time taken for climbing to the nacelle and sensor installation.

Video-Based Rotor Imbalance Checks from the Ground
Because of the large numbers of turbines affected by deviation from rotor design parameters, innovative measuring methods are needed to efficiently but reliably and accurately identify turbines where further action is needed to correct imbalance and blade angles. Through a two-year research project funded by the German Federal government an innovative video-based rotor imbalance check system has been successfully established. It is based on years of experience in sensor-based rotor imbalance and load measurements and optical blade angle determination at more than 1,000 turbines. If the nacelle provides objects suitable for continuous tracking (e.g. the screw holes in Figure 2, right) no climbing is necessary for the video-based vibration measurement, and this is very time-efficient. Working safety is also increased since people do not need to be in the nacelle during the measurement, but are sheltered in the tower base or at a remote position at the ground. The noise level in the amplitude spectrum (Figure 3) is also quite low because there is no impact from noise of the drive-train components.

Challenges and Requirements for High-Quality Results
Despite hub heights above a hundred metres, wind fluctuation and yawing of the turbine, vibration amplitudes of only a few millimetres (Figure 3) have to be resolved for the nacelle coordinate system measuring axial and lateral directions (parallel and horizontally perpendicular to the rotor axis (Figure 4)). So there are several significant challenges and requirements to obtain the required high result accuracy:

• Measurement conditions defined similarly to the reference measurements for the individual turbine type to assure reproducibility despite changing winds
  No rain, fog or low light; no strong wind direction changes; careful camera adjustment; prevention of wind-induced camera vibration; defined turbine operating modus
• A high-quality camera system
• Low image distortion; high light sensitivity and frame rate (> 60 Hz); full HD resolution
• Expert evaluation software with advanced features
• Easy and semi-automatic image and result validation to reject outliers
• Precise image calibration to convert the length unit pixel to millimetres
• Sub-pixel evaluation to achieve the desired measurement accuracy
• Simultaneous multi-point tracking of objects at the nacelle bottom (Figure 2, right) and for re-appearing objects at the rotor
• Yaw correction to transform motion back to the initial nacelle coordinate system
• Further tools for validation
• Frequency analysis (Figure 3), and order analysis to give the imbalance-related amplitude
• Well-trained staff because of the high impact of user decisions
• Powerful computers and adequate evaluation time, with evaluation taking hours, compared to minutes for sensor-based acceleration measurements

Efficient Verification by Natural Frequency and Modal Analysis
For similar excitation by wind and/or rotor speed (i.e. similar operating conditions) the vibrational response and loads of the turbine structure are highly dependent on the natural frequencies of the blades and tower. Moreover, not only the natural frequency, but also the related shape of the structural movements (mode) is important (Eigen mode, see Figure 4). Here, video-based analysis with multi-point tracking of the structural movements allows a refined modal analysis, without the time-consuming ‘typical’ installation of many sensors. Linear motion in two directions as well as torsional vibration can be simultaneously evaluated. Moreover, later analysis of other structural locations that were not in the main focus of the original measurement is easily possible provided that objects suitable for tracking are available.

If the slender tower’s natural frequency lies within the excitation frequency range related to rotor speed and/or blade passage, turbine certification often requires repeated natural frequency measurements at the individual serial turbine during commissioning and periodical inspection to assure proper control parameterisation, which prevents excess fatigue loads caused from resonance issues. It has to be remarked that in the field independent experts regularly discover turbines with a wrong frequency parameterisation in the control or even natural frequencies outside the tolerable range from certification. For this issue, video-based motion analysis offers a quick and reliable tool for independent natural frequency measurements.

Video-Based Root Cause Analysis
If certain problems such as vibration-related shut-downs occur for particular operational or wind conditions (speed and directions) a measurement-based root cause analysis is helpful to prove, for example, wake effects or controller-induced vibration. Then, video-based analysis allows a very quick measurement set-up and start. Figure 5 shows nacelle displacements of up to 1.5 metres in the axial direction (red) during gusty full-load operation and a fault-related emergency shut-down for a multi-megawatt turbine with more than 100 metres hub height. These movements affect all components in the nacelle including power and control electronics. This measurement data serves as input for load simulation verification or investigation of the impact of site-specific control settings (e.g. wind sector management) on loads, thus avoiding costly and time-consuming strain gauge installation. Video-based nacelle displacement is also suitable for strain gauge calibration during load measuring campaigns. Low frequency nacelle housing vibration can be detected, too.

Tower Clearance Measurements at Prototypes
The tower clearance (TC) is the minimum distance between the tip and the tower during the blade passage (Figure 6). It is a very relevant design criterion for serviceability. Because of the strongly changing loading during operation, there are several metres’ variation of the vertical and horizontal displacement of the tip during the blade passage. Here, video-based motion analysis is very suitable because the motion analysis of two-dimensional video images is able to cope with these large displacements as well as with nacelle yawing, in contrast to, for example, a stationary laser-based distance measurement.

Further challenges in this kind of measurement are:

• Need for high motion resolution (< 0.01 metre) despite measuring distances of several hundred metres
• Tracking fast-moving blade tips (more than 250 km/h), temporarily disappearing from the images
• Calibration and yaw correction
• Strongly changing background and illumination
• Synchronisation with load measurement and operational data
• Development of filtering criteria and algorithms and their validation
• Extensive uncertainty analysis with elimination of outliers (e.g. flying objects like birds, insects, etc.)
• High computational effort for evaluation of many video files (typically 4 GB+ per 20 minute video)

However, both static and operational TC measurements have been carried out and evaluated successfully, showing a very good agreement between measured TC and loads.

Conclusion
The successful development and application of video-based motion analysis for several measuring tasks at wind turbines shows, that it is a powerful expert tool for assisting in turbine design and root cause analysis – decreasing O&M costs by avoiding excess loads at the serial wind turbines.

Acknowledgements
The two-year research project on video-based imbalance checks at wind turbines was kindly funded by the German Federal Ministry for Economic Affairs and Energy.