Final Report Summary - AUTOWINSPEC (Automated mechanical property and fatigue life assessment of composite wind turbine blades in less than 4 hours)
Executive Summary:
Although wind power is the most promising renewable energy source and turbines are by far the most widespread means for harnessing wind energy, there are difficulties to overcome especially since their use is to expand. Despite the technological advances and use of composite materials, wind turbine rotor blades continue to fail due to the highly variable loads, significantly reducing the availability of the wind turbines and raising the downtime costs to over 600 million euro annually in EU. Current inspection methods fail to meet the end-user needs; reducing downtime as they focus on detecting defects. In fibre composite materials defects cannot be unambiguously matched with mechanical properties making traditional Non-Destructive Testing (NDT) techniques inadequate and many small distributed flaws cannot be reliably detected by any conventional methods. The AutoWinSpec system aims for a target reduction of 50% in maintenance costs, including the reduction of revenue loss through downtime, reducing them to 15% of overall turbine operating costs.
The AutoWinSpec projects has developed an Acousto-Ultrasonic (AU) device, integrated in an autonomous robotic mechanism, so as to provide fast and reliable inspection and mechanical integrity evaluation of wind turbine blades. Both FEM simulations and lab trials have been carried out to demonstrate the performance of AU technique. The main benefit of such a system is to replace the current inspection practice of wind turbine blade, whereby inspectors are hung from the rope and dropped along the blade while performing visual or conventional ultrasonic inspections.
The main selling points of the system is that it replaces using engineers who need to be hung from the nacelle via rope in order to access the blades in-situ. Instead, the robot can navigate on its own while deploying automatic inspection. Results can be accessed from ground level. In addition, compared to listening to the sound of a hammer test, the acoustic results are saved on record. It can be used for long term data comparison etc. The overall weight of the system is less than 40kg, also much lighter than a fully equipped inspector. Most importantly, the technique provides information that none of the conventional methods can at the moment.
Project Context and Objectives:
The wind turbine industry is one of the fastest growing markets. Current turbines are huge with turbine rotor diameters of over 100m becoming standard. EU energy policy calls for 20% of the EU's 3040Th/y electricity demand to come from renewable sources by 2020, which constitutes a market of some 140€ billion and wind energy is the clear front runner. BS Rotor, the initiator of the AutoWinSpec project, has identified the need for developing a novel and efficient NDE technique, integrated in an autonomous robotic mechanism, in order to provide fast and reliable mechanical integrity evaluation of wind turbine blades. BS Rotor has brought together a consortium of SMEs with complementary skills who also want to further improve their market penetration in the wind energy and NDT (Non Destructive Testing) sector. They have also chosen RTD performers with technical skills in NDT and robotics. The participating SMEs do not have internal R&D expertise in the different disciplines required to perform all the research in-house and thus have subcontracted the R&D activities.
The AutoWinSpec system provides a way to automatically estimate the condition of the blades and detect flaws that cannot be picked up by current inspection methods. It will also eliminate dangerous rope access and reduce downtime whilst increasing productivity. The product developed through this project will significantly enhance the Operation and Maintenance sector of Wind Power both onshore and offshore by providing an effective solution to the wind turbine blade failure problem. The core technique is Acousto-Ultrasonic inspection method, which is a full waveform analysis method, to enhance the inspection capabilities and enable accurate estimation of the mechanical properties of the blades.
The main objectives of the project are:
• Produce an NDT system that can detect flaws and structural degradations of wind turbine blades
• Produce a robotic system that can carry and deploy the system on the blades
• Perform tests and trials to demonstrate the capability of the integrated system
To achieve those objectives, a more detailed split of specific technical targets were also set and listed below:
• Decide on and secure the target wind turbine blade sample
• Perform finite element modelling of the AU propagation and interaction with various flaws in glass fibre composite, in order to simulate the AU signals obtained in measurements
• Perform advance signal analysis of the simulated signals in order to establish the technique’s capability in detecting the flaws
• Perform Eigen mode analysis to ensure the structural noise of an in-situ blade will not result in false alarm of flaw detection
• Based on the optimised modelling parameters, procure and assemble the NDT hardware
• Based on the optimised modelling parameters, select the appropriate probes and design the probe holding mechanism
• Based on the size and shape of the blade, application requirement of the AU techniques and dimensions of the NDT system, the robotic platform was designed to carry the NDT inspection system along the blade and deploy it for measurement
• The motion and control software of the robot is then integrated with the software interface of the NDT system
• The full AutoWinSpec system is then assembled and tested on the sampled blade secured in the beginning of the project
• The final test needed to be performed in a wind farm with an in-situ blade
Project Results:
RESULT 1: Crawling robotic system (Figures for Result 1)
The robotic system was designed as described in the relevant tasks in WP4. Following, its manufacturing and assembly was completed successfully. Next, several preliminary lab trials were designed and executed in order to assess its main components & subsystems performance under various typical scenarios. All necessary repairs, modifications and upgrades were determined and implemented. The robotic system performed highly satisfactorily according to initial planning. In more details:
o Design: The robotic system designed utilizes two wheels for its locomotion along a horizontally placed wind turbine blade (WTB) in cooperation with two safety arms, one at the front and one at the rear, each one incorporating a suction cup. This way, the system is always secured during operation.
The two NDT transducers necessary are mounted on a probe holder. Each holder (one at the left and one at the right) is manipulated by a linear drive placed vertical to the axis of the platform, enabling scanning of the WTB, parallel to its chord. A specially designed pneumatic suspension ensures that the probe holder is always in contact with the WTB’s surface. Figure R1-01 shows a 3D CAD representation.
- The robotic system’s locomotion subsystem, is based on a differentially driven platform. It incorporates two driving wheels and a castor one. Each wheel is pneumatic and made of a high friction material.
- The chassis platform is a lightweight yet rigid design. A custom special compartment for the electronics and pneumatics was also foreseen.
- With regard to safety, two arms are incorporated on the main platform, one at the front and one at the rear side, each one carrying a moveable suction cup. This way the platform during its motion has both cups activated, increasing its traction as well as its stability in case of a gust or a severe umbilical disturbance.
- Each probe holder is manipulated by a linear drive, comprised by a linear guide and a timing belt transmission actuated by a motor A specially designed pneumatic suspension is employed to control the probe holder’s contact with the WTB’s surface.
- The control hardware incorporates the motor controllers, a set of digitally controlled pneumatic valves and a number of manually controlled air preparation control units (flow, pressure). Finally, for closed loop control needs a set of sensors are incorporated such as, rotary encoders, limit switches, vacuum sensors, etc.
o Manufacturing and assembly:
The robotic platform comprises a plethora of components and subsystems, some, off-the-shelf and others custom designed.
The pneumatics and electronics circuit was custom designed by Innora. During assembly all necessary adaptations and modifications were performed.
It proved that the assembly was made accurately, according to the drawings. Figure R1-02 shows the robotic system assembly by Innora.
o Preliminary good operation testing
A number of tests were performed to verify the correct operation of the various subsystems of the robotic platform as well as the overall platform’s performance under various scenarios. The operation and performance of each degree of freedom was assessed and optimized.
RESULT 2: Transducer Probe Configuration and Acousto-Ultrasound system (Figures for Result 2)
As part of WP3 (Development of novel application and probe configuration of AU), taking into account the project requirements and specifications defined within WP1 and the technical results obtained within WP2, InnoTecUK, with the collaboration of the rest of the partners, has designed, implemented, integrated and tested the Acousto-Ultrasound system used as part of the AutoWinSpec system.
Figure R2-01 shows the high level block diagram of the Acousto-Ultrasound system that has been implemented. As it can be seen, it is composed by two Acoustic emission sensors from Vallen used to transmit the broadband acoustic signals along the blade surface and receive them. The transmission and reception of the acoustic signal is fully synchronised and the signal processing is performed both in time and frequency domain. The main signal is generated by a high power pulser from ST microelectronic and digitized using an oscilloscope from PicoScope. Figure R2-02 shows the Acousto-Ultrasound system already assembled and used as part of the lab tests executed at InnoTecUK before its final integration on the robotic platform.
Also, as part of the Acousto-Ultrasound system, InnoTecUK developed, in labview, a set of signal post-processing algorithms used to validate the overall measurement system. Figure R2-03 shows its Man machine interface (MMI) and from where, the end user can control the status of the measurement system, see the signals that are being generated, acquired them and execute the signal post processing algorithms.
The measurement system was validated using a set of well-known defects samples provided by SpectrumLabs. Figure R2-04 shows some of examples of its detection capabilities (Teflon inserts).
In order to ensure a good coupling between the active surface of the acoustic emission sensors and the blade surface, a special elastomer from Olympus was used (Aqualene) as it is a dry and clean solution and its acoustic impedance is quite similar to the water one. Additionally, in order to ensure a good pressure force, a set of vacuum caps were included into the probe holder design (Figure R2-05) to attach the probe to the blade surface and a pneumatic piston was used to move down/up the acoustic emission sensors and apply on them an external force of approx. 10kilos. It is also important to highlight that the design process of the probe holders was an iterative process and several conceptual ideas were generated.
The probes holders were installed on the bottom part of the robotic arms as it can be also seen in Figure R2-05.
As part of the lab trials, the project partners used a wind turbine blade section placed at TWI (Cambridge, UK). As it can be seen in Figure R2-06, the NDT Acousto-Ultrasound system was already installed onboard (top part of the robot chassis) and a set of tests were executed. See also, within this report, the following section: Motion and control system.
Figure R2-07 shows a set of spot NDT Acousto-Ultrasound measurements using the AutowinSpec system placed on a real wind turbine in Greece - during the final testing process of the overall system. The shape of the blue and the red signals is strongly related to the position of the probe along the blade. The signal levels showed the capability of the system to transmit and receive the generated signals along a real wind turbine blade surface.
RESULT 3: Signal Analysis (Figures for Result 3)
To achieve the optimised signal analysis algorithm before constructing the NDT hardware, simulated signal responses were used. The glass fibre composite material modelled was based on the blade sample secured for the project. Each layer was homogenised individually based on Yang-Mal algorithm and then compiled into multi-layer composites in the finite element environment. The inspection set up modelled was optimised based on recommendations from relevant standard, and shown in Figure R3-01. The reference signal response was established based on the configuration shown in Figure R3-01.
In the next step, various flaws were introduced into the sample. The flaws of interest were distributed micro-porosity, aging and large delamination. With each type of flaw, a number of signal processing methods were performed in order to optimise detection. For distributed micro-porosity, each pore was set to be 50µm. A range of porosity densities were modelled as shown in Figure R3-02.
The following signal responses were obtained for increasing porosities, shown in Figure R3-03. Spectrum analysis, homomorphic processing, and stress wave factor analysis with and without partition were applied to the raw signals. The full wave energy without partition was shown to be the most robust method. The result is shown in Figure R3-04. A clear trend of decreasing energy with increasing porosity was observed. A max of 67% drop in energy was observed. Even at 0.6% porosity, the energy content decreased by 50%. That means the wave attenuation increased with the porosity. This is consistent with the fact distributed micro-flaws disperse and attenuate the propagating waves.
For aging, the stiffness matrix was reduced to 40%, 50%, 60%, 70% and 80%. The full waveforms of all signals received are shown in Figure R3-05.
Similarly spectrum analysis, homomorphic processing, and stress wave factor analysis with and without partition were applied to the raw signals. The full wave energy without partition was shown to be the most robust method. The result is shown in Figure R3-06. Figure R3-06 displayed almost a linear relationship between the energy content and percentage of stiffness. Overall the energy decreased significantly as the material ages with decreasing stiffness. The energy content dropped by 84% as the stiffness was reduced to half. Even for a 20% drop in the stiffness, the energy has decreased by 40%. Such drops in energies are rather significant and would serve as a good indicator during the inspection.
Large delamination is not a particularly difficult flaw for conventional UT to detect. However, it is still important to demonstrate that AU technique can also detect such flaws. In the model, delamination was inserted at different depths, as shown in Figure R3-07. The signal responses are shown in Figure R3-08
In this case, spectral energy analysis gave the most robust result as shown in Figure R3-09.
As the delamination gets shallower, more energy is directed towards the sensor and the measured energy content. The degree of changes in the indicator is also observed to be much larger than the other flaws. So when multiple flaws are present, large delamination will produce the dominant effect. Because large delamination is the one of the most serious types of flaws, it is promising to find out that in AU measurement, its signals will not be drown by the signals produced by other types of flaws simulated.
Vibration analysis was also carried out to ensure the natural resonance of the blade will not produce false alarm. The analysis of up to 500 modes was carried out and the highest frequency was around 500Hz, which is way below the AU frequency range. So such vibrations will not even be picked up by the sensor.
The last part of the signal analysis is the presentation of data. A C-scan type of presentation was chosen. At each scan position, the signal processing method will produce one characteristic parameter, for example energy content or signal arrival time. When all the positions are scanned, a plot of the chosen characteristic was obtained. An example is shown in Figure R3-10 (arbitrary number) and Figure R2-04 (experimental data obtained at a later stage).
RESULT 4: Motion and Control System
In line with the mechanical development process and according to the working procedure defined by Innora, InnoTecUK has developed, in labview, the monitoring and control software used to control and monitor all the robotic platform parameters and components. That means: monitor the status of all sensors and control the status of the pneumatic valves and electrical motors.
Due to safety issues and the environmental conditions where the final tests were executed, the monitoring and control loop was developed using a supervised philosophy. In other words, all the steps of the working procedure are autonomously executed but before releasing the vacuum cups from the wind turbine blade surface, the end user has to validate the process. Additionally, the main movement of the robotic platform is always performed in a straight line, except, if the end user is specifically commanded to turn right or left in order to correct the movement direction.
The control software was tested at InnoTecUk's lab to validate the overall software status. Afterwards, the software was tested and improved using the wind turbine blade section placed at TWI.
The final control tests were executed on a real wind turbine blade in a wind farm in Greece, and at CRES, Greece. The results of these control tests can be seen in the project videos.
RESULT 5: The Integrated AutoWinSpec System (Figures for Result 5)
The trials with the complete integrated system were made in two types of venues: (a) Full WTB placed horizontally on the ground, (b) WTB of an operating WT, stalled horizontally
In case (a) the system was tested in spring 2016 in Greece, on two different WTBs: a1) On a small WTB (LM make) belonging on a WT with max. output power of ~0,5MW at CRES (Figures R5-01); a2) On a WTB (Gamesa make) belonging on a WT with max. output power of 1.2MW (Figures R5-02).
In case (b) the system was tested on a WTB of an operating WT (Vestas V47-0,67MW) in CRES in Attica-Greece.
Regarding the three trials in total, the organizing and preparation was performed by Innora in close collaboration with CRES, InnoTecUK, TWI, SpectrumLabs and WLB.
With respect to the ground trials all in prior preparations were performed absolutely successfully as planned, that is:
• Human operators’ safety measures
• WTBs proper and secured placement
• Power supplies (electrical, compressed air)
• Robot preparation
• Robot and all equipment & tools transport and storage
• Multimedia capturing equipment
The setup was quite straightforward; the system was placed manually on a horizontally mounted WTB and controlled remotely via a control laptop. The ground trials proved to be extremely useful as the system’s performance was evaluated under highly realistic conditions on two WTB sizes. The system’s performance was in general successful under all operations tested. The following were tested and evaluated:
o Robot navigation and stability
- Automatic straight motion
- Under operator’s supervision cornering
- Stability vacuum cups system
o NDT operation
- NDT data capturing, mapping and post processing
- NDT probe holders’ manipulation
o Robot’s status evaluation
- Sensors’ feedback
- Control algorithm interlocks
In all cases, various improvements were performed aiming to increase the readiness level of the developed system for the follow-on air trials. This mainly refers to safety and secondarily on performance optimization. For this reason, improvements and upgrades were performed on the following areas:
o Electromechanical area
- Cables management
- Pneumatic cylinders speed and pressure fine tuning
- Wheels air pressure fine tuning
- Umbilical improved lightweight design
o Motion control
- NDT probes movement procedure
- Platform motion sequence optimization
o NDT system
- Optimized probes mounting
- Coaxial cables management
- Data acquisition optimization
- Coupling method improved
- Data processing
With respect to the field trials (Figures R5-03), the system was tested on a WTB of an installed WT, a Vestas V47-0,67MW. The preparation works were more complicated as they involved the usage of a high range external truck-crane for maximum safety. The scenario designed and pursued was as follows:
i. Technicians on top of the nacelle would help to:
- Assist with the robot’s preparation for operation
- Assist with the placement of the system on the WTB
- Evaluate system’s status and communicate with the ground control station operator
ii. The WT’s onboard nacelle crane is used to transport in the nacelle all required assistive equipment, that is:
- The air compressor
- The umbilical
- A camera for video capturing
iii. One technician mounts an Ethernet cable on the WT’s internal ladder in order to establish communication with the ground control station.
iv. The crane lifts the complete robotic system and places it on top of the nacelle for preliminary good operation check.
v. The crane places the system on the WTB and keeps it secured throughout the whole process.
vi. When trials are completed the crane lowers down the system for inspection and storage.
After having secured the communication with the robot:
o The robot was tested on the nacelle’s roof for a preliminary check.
o The system was easily placed on the WTB with fine guidance by the operators, through the robot’s umbilical.
o Its operation was satisfactory (locomotion, NDT operation).
- No disturbance was witnessed due to winds, umbilical weight, WTB’s oscillations
- No slippage was witnessed
- The NDT system performed satisfactorily
Any issues encountered were solved on site so that the trials could proceed unhindered.
Finally, a video on youtube, https://www.youtube.com/watch?v=54H5TLeIsjM showcases in the best way the highlights from the field trials performed.
Potential Impact:
Wind Power both globally and in Europe is the clear front runner when it comes to renewable forms of energy. At the end of 2011 worldwide nameplate capacity of wind-powered generators was 238 GW growing by 41 over the preceding year. Between 2005 and 2010 the average annual growth in new installations was 27.6%. As of 2011, 83 countries around the world were using wind power on a commercial basis; with China making a dynamic entry in Wind Power industry accounting for nearly half new installations in 2010 (16.5 GW) and aiming at 100 GW by 2015. In terms of economic value, the wind energy sector has become one of the important players in the energy markets, with the total value of new generating equipment installed in 2007 reaching €25 billion.
Although the wind power industry was affected by the European and global financial crisis in 2009 - 2012, Global Wind Energy Council (GWEC) expects a substantial annual installed growth rate beyond 2013 and up to 2016. In the forecast to 2016 the expected average annual growth rate is 9 percent. Between 2012 and 2016, more than 255 GW of new wind power capacity could come on line before. Wind power market growth rate is expected to reach 11.9 percent by 2015 and 7.6 percent by 2018. In 2011, installed Wind power capacity in the EU totaled 93,957 MW while achieving an average annual growth of 15.6% over the last 17 years (1995-2011). EWEA estimates that 230 GW of wind capacity will be installed in Europe by 2020 as three times greater than Europe’s expected electricity demand rising to a factor of seven by 2030. Corresponding maintenance market is set to reach 42 billion by 2020. These figures and market prediction are directly correlated with the number of installed WT and are indicative of the growing number of WTs installed every year.
The specific blade failure problem exceeds €600 million in Europe annually. Lost revenue due to downtime sums up to €512 million; considering that approximately 94 GW equals to 50,000 WT and utilizing the information for blade failure (Annual failure frequency=0.2 Downtime per failure=4.1 days; Source: ISET). Additional cost savings will result from expanding the blade lifespan up to 30% by accurately estimating the areas to be repaired and eliminating the need for trained operators to perform the inspection.
Currently a trained aerial platform operator costs ~€400 per day and a team of three operators per WT needs a day to inspect its blades (400 x 3 = €1200). The integrated AutoWinSpec will require only two non-IRATA certified technicians (€200 per day). The inspection time will reach 50 min per WTB. A commonly used WT facilitates 3 blades, therefore 3 hours plus 1 hour for the required equipment mobilization and setup are required to fully inspect a WTB. Consequently, the integrated AutoWinSpec system will be able to fully inspect 2 WTs per day, increasing the inspection speed by 100% with regards to the currently applied methods, while the outage duration of each WT is reduced by 50%.
List of Websites:
WEBSITE address: www.autowinspec.eu
Contact Point: Ms. Silvia Sotiropoulou, WLB LIMITED
Although wind power is the most promising renewable energy source and turbines are by far the most widespread means for harnessing wind energy, there are difficulties to overcome especially since their use is to expand. Despite the technological advances and use of composite materials, wind turbine rotor blades continue to fail due to the highly variable loads, significantly reducing the availability of the wind turbines and raising the downtime costs to over 600 million euro annually in EU. Current inspection methods fail to meet the end-user needs; reducing downtime as they focus on detecting defects. In fibre composite materials defects cannot be unambiguously matched with mechanical properties making traditional Non-Destructive Testing (NDT) techniques inadequate and many small distributed flaws cannot be reliably detected by any conventional methods. The AutoWinSpec system aims for a target reduction of 50% in maintenance costs, including the reduction of revenue loss through downtime, reducing them to 15% of overall turbine operating costs.
The AutoWinSpec projects has developed an Acousto-Ultrasonic (AU) device, integrated in an autonomous robotic mechanism, so as to provide fast and reliable inspection and mechanical integrity evaluation of wind turbine blades. Both FEM simulations and lab trials have been carried out to demonstrate the performance of AU technique. The main benefit of such a system is to replace the current inspection practice of wind turbine blade, whereby inspectors are hung from the rope and dropped along the blade while performing visual or conventional ultrasonic inspections.
The main selling points of the system is that it replaces using engineers who need to be hung from the nacelle via rope in order to access the blades in-situ. Instead, the robot can navigate on its own while deploying automatic inspection. Results can be accessed from ground level. In addition, compared to listening to the sound of a hammer test, the acoustic results are saved on record. It can be used for long term data comparison etc. The overall weight of the system is less than 40kg, also much lighter than a fully equipped inspector. Most importantly, the technique provides information that none of the conventional methods can at the moment.
Project Context and Objectives:
The wind turbine industry is one of the fastest growing markets. Current turbines are huge with turbine rotor diameters of over 100m becoming standard. EU energy policy calls for 20% of the EU's 3040Th/y electricity demand to come from renewable sources by 2020, which constitutes a market of some 140€ billion and wind energy is the clear front runner. BS Rotor, the initiator of the AutoWinSpec project, has identified the need for developing a novel and efficient NDE technique, integrated in an autonomous robotic mechanism, in order to provide fast and reliable mechanical integrity evaluation of wind turbine blades. BS Rotor has brought together a consortium of SMEs with complementary skills who also want to further improve their market penetration in the wind energy and NDT (Non Destructive Testing) sector. They have also chosen RTD performers with technical skills in NDT and robotics. The participating SMEs do not have internal R&D expertise in the different disciplines required to perform all the research in-house and thus have subcontracted the R&D activities.
The AutoWinSpec system provides a way to automatically estimate the condition of the blades and detect flaws that cannot be picked up by current inspection methods. It will also eliminate dangerous rope access and reduce downtime whilst increasing productivity. The product developed through this project will significantly enhance the Operation and Maintenance sector of Wind Power both onshore and offshore by providing an effective solution to the wind turbine blade failure problem. The core technique is Acousto-Ultrasonic inspection method, which is a full waveform analysis method, to enhance the inspection capabilities and enable accurate estimation of the mechanical properties of the blades.
The main objectives of the project are:
• Produce an NDT system that can detect flaws and structural degradations of wind turbine blades
• Produce a robotic system that can carry and deploy the system on the blades
• Perform tests and trials to demonstrate the capability of the integrated system
To achieve those objectives, a more detailed split of specific technical targets were also set and listed below:
• Decide on and secure the target wind turbine blade sample
• Perform finite element modelling of the AU propagation and interaction with various flaws in glass fibre composite, in order to simulate the AU signals obtained in measurements
• Perform advance signal analysis of the simulated signals in order to establish the technique’s capability in detecting the flaws
• Perform Eigen mode analysis to ensure the structural noise of an in-situ blade will not result in false alarm of flaw detection
• Based on the optimised modelling parameters, procure and assemble the NDT hardware
• Based on the optimised modelling parameters, select the appropriate probes and design the probe holding mechanism
• Based on the size and shape of the blade, application requirement of the AU techniques and dimensions of the NDT system, the robotic platform was designed to carry the NDT inspection system along the blade and deploy it for measurement
• The motion and control software of the robot is then integrated with the software interface of the NDT system
• The full AutoWinSpec system is then assembled and tested on the sampled blade secured in the beginning of the project
• The final test needed to be performed in a wind farm with an in-situ blade
Project Results:
RESULT 1: Crawling robotic system (Figures for Result 1)
The robotic system was designed as described in the relevant tasks in WP4. Following, its manufacturing and assembly was completed successfully. Next, several preliminary lab trials were designed and executed in order to assess its main components & subsystems performance under various typical scenarios. All necessary repairs, modifications and upgrades were determined and implemented. The robotic system performed highly satisfactorily according to initial planning. In more details:
o Design: The robotic system designed utilizes two wheels for its locomotion along a horizontally placed wind turbine blade (WTB) in cooperation with two safety arms, one at the front and one at the rear, each one incorporating a suction cup. This way, the system is always secured during operation.
The two NDT transducers necessary are mounted on a probe holder. Each holder (one at the left and one at the right) is manipulated by a linear drive placed vertical to the axis of the platform, enabling scanning of the WTB, parallel to its chord. A specially designed pneumatic suspension ensures that the probe holder is always in contact with the WTB’s surface. Figure R1-01 shows a 3D CAD representation.
- The robotic system’s locomotion subsystem, is based on a differentially driven platform. It incorporates two driving wheels and a castor one. Each wheel is pneumatic and made of a high friction material.
- The chassis platform is a lightweight yet rigid design. A custom special compartment for the electronics and pneumatics was also foreseen.
- With regard to safety, two arms are incorporated on the main platform, one at the front and one at the rear side, each one carrying a moveable suction cup. This way the platform during its motion has both cups activated, increasing its traction as well as its stability in case of a gust or a severe umbilical disturbance.
- Each probe holder is manipulated by a linear drive, comprised by a linear guide and a timing belt transmission actuated by a motor A specially designed pneumatic suspension is employed to control the probe holder’s contact with the WTB’s surface.
- The control hardware incorporates the motor controllers, a set of digitally controlled pneumatic valves and a number of manually controlled air preparation control units (flow, pressure). Finally, for closed loop control needs a set of sensors are incorporated such as, rotary encoders, limit switches, vacuum sensors, etc.
o Manufacturing and assembly:
The robotic platform comprises a plethora of components and subsystems, some, off-the-shelf and others custom designed.
The pneumatics and electronics circuit was custom designed by Innora. During assembly all necessary adaptations and modifications were performed.
It proved that the assembly was made accurately, according to the drawings. Figure R1-02 shows the robotic system assembly by Innora.
o Preliminary good operation testing
A number of tests were performed to verify the correct operation of the various subsystems of the robotic platform as well as the overall platform’s performance under various scenarios. The operation and performance of each degree of freedom was assessed and optimized.
RESULT 2: Transducer Probe Configuration and Acousto-Ultrasound system (Figures for Result 2)
As part of WP3 (Development of novel application and probe configuration of AU), taking into account the project requirements and specifications defined within WP1 and the technical results obtained within WP2, InnoTecUK, with the collaboration of the rest of the partners, has designed, implemented, integrated and tested the Acousto-Ultrasound system used as part of the AutoWinSpec system.
Figure R2-01 shows the high level block diagram of the Acousto-Ultrasound system that has been implemented. As it can be seen, it is composed by two Acoustic emission sensors from Vallen used to transmit the broadband acoustic signals along the blade surface and receive them. The transmission and reception of the acoustic signal is fully synchronised and the signal processing is performed both in time and frequency domain. The main signal is generated by a high power pulser from ST microelectronic and digitized using an oscilloscope from PicoScope. Figure R2-02 shows the Acousto-Ultrasound system already assembled and used as part of the lab tests executed at InnoTecUK before its final integration on the robotic platform.
Also, as part of the Acousto-Ultrasound system, InnoTecUK developed, in labview, a set of signal post-processing algorithms used to validate the overall measurement system. Figure R2-03 shows its Man machine interface (MMI) and from where, the end user can control the status of the measurement system, see the signals that are being generated, acquired them and execute the signal post processing algorithms.
The measurement system was validated using a set of well-known defects samples provided by SpectrumLabs. Figure R2-04 shows some of examples of its detection capabilities (Teflon inserts).
In order to ensure a good coupling between the active surface of the acoustic emission sensors and the blade surface, a special elastomer from Olympus was used (Aqualene) as it is a dry and clean solution and its acoustic impedance is quite similar to the water one. Additionally, in order to ensure a good pressure force, a set of vacuum caps were included into the probe holder design (Figure R2-05) to attach the probe to the blade surface and a pneumatic piston was used to move down/up the acoustic emission sensors and apply on them an external force of approx. 10kilos. It is also important to highlight that the design process of the probe holders was an iterative process and several conceptual ideas were generated.
The probes holders were installed on the bottom part of the robotic arms as it can be also seen in Figure R2-05.
As part of the lab trials, the project partners used a wind turbine blade section placed at TWI (Cambridge, UK). As it can be seen in Figure R2-06, the NDT Acousto-Ultrasound system was already installed onboard (top part of the robot chassis) and a set of tests were executed. See also, within this report, the following section: Motion and control system.
Figure R2-07 shows a set of spot NDT Acousto-Ultrasound measurements using the AutowinSpec system placed on a real wind turbine in Greece - during the final testing process of the overall system. The shape of the blue and the red signals is strongly related to the position of the probe along the blade. The signal levels showed the capability of the system to transmit and receive the generated signals along a real wind turbine blade surface.
RESULT 3: Signal Analysis (Figures for Result 3)
To achieve the optimised signal analysis algorithm before constructing the NDT hardware, simulated signal responses were used. The glass fibre composite material modelled was based on the blade sample secured for the project. Each layer was homogenised individually based on Yang-Mal algorithm and then compiled into multi-layer composites in the finite element environment. The inspection set up modelled was optimised based on recommendations from relevant standard, and shown in Figure R3-01. The reference signal response was established based on the configuration shown in Figure R3-01.
In the next step, various flaws were introduced into the sample. The flaws of interest were distributed micro-porosity, aging and large delamination. With each type of flaw, a number of signal processing methods were performed in order to optimise detection. For distributed micro-porosity, each pore was set to be 50µm. A range of porosity densities were modelled as shown in Figure R3-02.
The following signal responses were obtained for increasing porosities, shown in Figure R3-03. Spectrum analysis, homomorphic processing, and stress wave factor analysis with and without partition were applied to the raw signals. The full wave energy without partition was shown to be the most robust method. The result is shown in Figure R3-04. A clear trend of decreasing energy with increasing porosity was observed. A max of 67% drop in energy was observed. Even at 0.6% porosity, the energy content decreased by 50%. That means the wave attenuation increased with the porosity. This is consistent with the fact distributed micro-flaws disperse and attenuate the propagating waves.
For aging, the stiffness matrix was reduced to 40%, 50%, 60%, 70% and 80%. The full waveforms of all signals received are shown in Figure R3-05.
Similarly spectrum analysis, homomorphic processing, and stress wave factor analysis with and without partition were applied to the raw signals. The full wave energy without partition was shown to be the most robust method. The result is shown in Figure R3-06. Figure R3-06 displayed almost a linear relationship between the energy content and percentage of stiffness. Overall the energy decreased significantly as the material ages with decreasing stiffness. The energy content dropped by 84% as the stiffness was reduced to half. Even for a 20% drop in the stiffness, the energy has decreased by 40%. Such drops in energies are rather significant and would serve as a good indicator during the inspection.
Large delamination is not a particularly difficult flaw for conventional UT to detect. However, it is still important to demonstrate that AU technique can also detect such flaws. In the model, delamination was inserted at different depths, as shown in Figure R3-07. The signal responses are shown in Figure R3-08
In this case, spectral energy analysis gave the most robust result as shown in Figure R3-09.
As the delamination gets shallower, more energy is directed towards the sensor and the measured energy content. The degree of changes in the indicator is also observed to be much larger than the other flaws. So when multiple flaws are present, large delamination will produce the dominant effect. Because large delamination is the one of the most serious types of flaws, it is promising to find out that in AU measurement, its signals will not be drown by the signals produced by other types of flaws simulated.
Vibration analysis was also carried out to ensure the natural resonance of the blade will not produce false alarm. The analysis of up to 500 modes was carried out and the highest frequency was around 500Hz, which is way below the AU frequency range. So such vibrations will not even be picked up by the sensor.
The last part of the signal analysis is the presentation of data. A C-scan type of presentation was chosen. At each scan position, the signal processing method will produce one characteristic parameter, for example energy content or signal arrival time. When all the positions are scanned, a plot of the chosen characteristic was obtained. An example is shown in Figure R3-10 (arbitrary number) and Figure R2-04 (experimental data obtained at a later stage).
RESULT 4: Motion and Control System
In line with the mechanical development process and according to the working procedure defined by Innora, InnoTecUK has developed, in labview, the monitoring and control software used to control and monitor all the robotic platform parameters and components. That means: monitor the status of all sensors and control the status of the pneumatic valves and electrical motors.
Due to safety issues and the environmental conditions where the final tests were executed, the monitoring and control loop was developed using a supervised philosophy. In other words, all the steps of the working procedure are autonomously executed but before releasing the vacuum cups from the wind turbine blade surface, the end user has to validate the process. Additionally, the main movement of the robotic platform is always performed in a straight line, except, if the end user is specifically commanded to turn right or left in order to correct the movement direction.
The control software was tested at InnoTecUk's lab to validate the overall software status. Afterwards, the software was tested and improved using the wind turbine blade section placed at TWI.
The final control tests were executed on a real wind turbine blade in a wind farm in Greece, and at CRES, Greece. The results of these control tests can be seen in the project videos.
RESULT 5: The Integrated AutoWinSpec System (Figures for Result 5)
The trials with the complete integrated system were made in two types of venues: (a) Full WTB placed horizontally on the ground, (b) WTB of an operating WT, stalled horizontally
In case (a) the system was tested in spring 2016 in Greece, on two different WTBs: a1) On a small WTB (LM make) belonging on a WT with max. output power of ~0,5MW at CRES (Figures R5-01); a2) On a WTB (Gamesa make) belonging on a WT with max. output power of 1.2MW (Figures R5-02).
In case (b) the system was tested on a WTB of an operating WT (Vestas V47-0,67MW) in CRES in Attica-Greece.
Regarding the three trials in total, the organizing and preparation was performed by Innora in close collaboration with CRES, InnoTecUK, TWI, SpectrumLabs and WLB.
With respect to the ground trials all in prior preparations were performed absolutely successfully as planned, that is:
• Human operators’ safety measures
• WTBs proper and secured placement
• Power supplies (electrical, compressed air)
• Robot preparation
• Robot and all equipment & tools transport and storage
• Multimedia capturing equipment
The setup was quite straightforward; the system was placed manually on a horizontally mounted WTB and controlled remotely via a control laptop. The ground trials proved to be extremely useful as the system’s performance was evaluated under highly realistic conditions on two WTB sizes. The system’s performance was in general successful under all operations tested. The following were tested and evaluated:
o Robot navigation and stability
- Automatic straight motion
- Under operator’s supervision cornering
- Stability vacuum cups system
o NDT operation
- NDT data capturing, mapping and post processing
- NDT probe holders’ manipulation
o Robot’s status evaluation
- Sensors’ feedback
- Control algorithm interlocks
In all cases, various improvements were performed aiming to increase the readiness level of the developed system for the follow-on air trials. This mainly refers to safety and secondarily on performance optimization. For this reason, improvements and upgrades were performed on the following areas:
o Electromechanical area
- Cables management
- Pneumatic cylinders speed and pressure fine tuning
- Wheels air pressure fine tuning
- Umbilical improved lightweight design
o Motion control
- NDT probes movement procedure
- Platform motion sequence optimization
o NDT system
- Optimized probes mounting
- Coaxial cables management
- Data acquisition optimization
- Coupling method improved
- Data processing
With respect to the field trials (Figures R5-03), the system was tested on a WTB of an installed WT, a Vestas V47-0,67MW. The preparation works were more complicated as they involved the usage of a high range external truck-crane for maximum safety. The scenario designed and pursued was as follows:
i. Technicians on top of the nacelle would help to:
- Assist with the robot’s preparation for operation
- Assist with the placement of the system on the WTB
- Evaluate system’s status and communicate with the ground control station operator
ii. The WT’s onboard nacelle crane is used to transport in the nacelle all required assistive equipment, that is:
- The air compressor
- The umbilical
- A camera for video capturing
iii. One technician mounts an Ethernet cable on the WT’s internal ladder in order to establish communication with the ground control station.
iv. The crane lifts the complete robotic system and places it on top of the nacelle for preliminary good operation check.
v. The crane places the system on the WTB and keeps it secured throughout the whole process.
vi. When trials are completed the crane lowers down the system for inspection and storage.
After having secured the communication with the robot:
o The robot was tested on the nacelle’s roof for a preliminary check.
o The system was easily placed on the WTB with fine guidance by the operators, through the robot’s umbilical.
o Its operation was satisfactory (locomotion, NDT operation).
- No disturbance was witnessed due to winds, umbilical weight, WTB’s oscillations
- No slippage was witnessed
- The NDT system performed satisfactorily
Any issues encountered were solved on site so that the trials could proceed unhindered.
Finally, a video on youtube, https://www.youtube.com/watch?v=54H5TLeIsjM showcases in the best way the highlights from the field trials performed.
Potential Impact:
Wind Power both globally and in Europe is the clear front runner when it comes to renewable forms of energy. At the end of 2011 worldwide nameplate capacity of wind-powered generators was 238 GW growing by 41 over the preceding year. Between 2005 and 2010 the average annual growth in new installations was 27.6%. As of 2011, 83 countries around the world were using wind power on a commercial basis; with China making a dynamic entry in Wind Power industry accounting for nearly half new installations in 2010 (16.5 GW) and aiming at 100 GW by 2015. In terms of economic value, the wind energy sector has become one of the important players in the energy markets, with the total value of new generating equipment installed in 2007 reaching €25 billion.
Although the wind power industry was affected by the European and global financial crisis in 2009 - 2012, Global Wind Energy Council (GWEC) expects a substantial annual installed growth rate beyond 2013 and up to 2016. In the forecast to 2016 the expected average annual growth rate is 9 percent. Between 2012 and 2016, more than 255 GW of new wind power capacity could come on line before. Wind power market growth rate is expected to reach 11.9 percent by 2015 and 7.6 percent by 2018. In 2011, installed Wind power capacity in the EU totaled 93,957 MW while achieving an average annual growth of 15.6% over the last 17 years (1995-2011). EWEA estimates that 230 GW of wind capacity will be installed in Europe by 2020 as three times greater than Europe’s expected electricity demand rising to a factor of seven by 2030. Corresponding maintenance market is set to reach 42 billion by 2020. These figures and market prediction are directly correlated with the number of installed WT and are indicative of the growing number of WTs installed every year.
The specific blade failure problem exceeds €600 million in Europe annually. Lost revenue due to downtime sums up to €512 million; considering that approximately 94 GW equals to 50,000 WT and utilizing the information for blade failure (Annual failure frequency=0.2 Downtime per failure=4.1 days; Source: ISET). Additional cost savings will result from expanding the blade lifespan up to 30% by accurately estimating the areas to be repaired and eliminating the need for trained operators to perform the inspection.
Currently a trained aerial platform operator costs ~€400 per day and a team of three operators per WT needs a day to inspect its blades (400 x 3 = €1200). The integrated AutoWinSpec will require only two non-IRATA certified technicians (€200 per day). The inspection time will reach 50 min per WTB. A commonly used WT facilitates 3 blades, therefore 3 hours plus 1 hour for the required equipment mobilization and setup are required to fully inspect a WTB. Consequently, the integrated AutoWinSpec system will be able to fully inspect 2 WTs per day, increasing the inspection speed by 100% with regards to the currently applied methods, while the outage duration of each WT is reduced by 50%.
List of Websites:
WEBSITE address: www.autowinspec.eu
Contact Point: Ms. Silvia Sotiropoulou, WLB LIMITED