Deliverables
Phase 1.
The Machine Condition and Structural Condition Monitoring System has been installed in the test wind turbine in Zoetemeer, Den Haag. The System has been integrated with the SCADA system and connected to the necessary process signals from the wind turbine control system. Unfortunately there has been some severe restrictions concerning installation and access to construction data which has made it very difficult to set-up an optimal system. Software for integration with the common SCADA system has been developed, see the document "Dataexchange 040219.pdf" and measurements has been recorded since medio February 2004. The results so far are presented in the document "Zoetemeer Test Results.pdf". Also software for doing improved re-sampling of measurements signals according to RPM (revolution per minute) has been improved. In March 2004 a bearing failed on the generator and where replaced. Unfortunately there is no monitoring of the generator due to the fact that we where not allowed to mount any sensors here. So here a great opportunity to demonstrate the possibilities and benefits of the system where lost.
Conclusion:
The measurement set-up has due to the lack of information regarding the construction, especially data for the gearbox, been very heuristic i.e. only a broadband FFT and Envelope and Cepstrum measurements have been set-up in a "standard" way. Conditional recording is based on an Active Power signal and the signal is resample according to RPM. The sensors are unfortunately situated on the gearbox at some very insensitive spots, due to restrictions regarding installation imposed by the owner of the wind turbine. Due to these circumstances the report do not present the necessary details to make an optimal MCM report. Due to these circumstances and due to the fact that no sensors where mounted on the generator, the system has not been able to pick up an evolving bearing fault on the generator, which had the bearing replaced medio March, i.e. no simple damage indicators where located (see the plots).
This is very sad since we lost a good opportunity to demonstrate the benefits of the CONMOW project. If the installation has been appropriate and we have had the information to compose an effective measurement set-up, this damage could have been detected in a much earlier state. Never a less the project should be carried on using other turbines (phase 2), so we will have a second change to demonstrate the benefits and opportunities of the CONMOW project.
Phase 2.
In the second phase of the CONMOW project two Nordex turbines where equipped with the Turbine Condition Monitoring System TCM. This phase has been used to develop methods for Automatic Fault Signature detection. Conclusion As only two turbines of same type, although over some time, has been producing data the material is too limited to finish the learning period for this specific type of machinery. No severe faults have been detected but beginning signs of wear can be seen.
The Signature Fault Detection technique can be improved and combined with the complementary Mask Fault Detection also used in the TCM system, an having sufficient experience with the specific machinery, i.e. enough wind turbines, to concluded the learning period, a good foundation to make predictive condition monitoring is available. With these techniques, implemented in an automated expert system as the TCM System, it is possible to monitor many wind turbines in an automated way. The outcome of the CONMOW project contributes to the possibility to implement renewable energy as wind power in large scale - and keep it up running.
The Machine Condition and Structural Condition Monitoring System has been installed in the test wind turbine in Zoetemeer, Den Haag. The System has been integrated with the SCADA system and connected to the necessary process signals from the wind turbine control system.
Unfortunately there has been some severe restrictions concerning installation and access to construction data which has made it very difficult to set-up an optimal system.
Software for integration with the common SCADA system has been developed, see the document "Dataexchange 040219.pdf" and measurements has been recorded since medio February 2004. The results so far are presented in the document "Zoetemeer Test Results.pdf". Also software for doing improved re-sampling of measurements signals according to RPM (revolution per minute) has been improved. In March 2004 a bearing failed on the generator and where replaced. Unfortunately there is no monitoring of the generator due to the fact that we where not allowed to mount any sensors here. So here a great opportunity to demonstrate the possibilities and benefits of the system where lost.
Conclusion
The measurement set-up has due to the lack of information regarding the construction, especially data for the gearbox, been very heuristic i.e. only a broadband FFT and Envelope and Cepstrum measurements have been set-up in a "standard" way. Conditional recording is based on an Active Power signal and the signal is resample according to RPM. The sensors are unfortunately situated on the gearbox at some very insensitive spots, due to restrictions regarding installation imposed by the owner of the wind turbine. Due to these circumstances the report do not present the necessary details to make an optimal MCM report. Due to these circumstances and due to the fact that no sensors where mounted on the generator, the system has not been able to pick up an evolving bearing fault on the generator, which had the bearing replaced medio March, i.e. no simple damage indicators where located (see the plots). This is very sad since we lost a good opportunity to demonstrate the benefits of the CONMOW project. If the installation has been appropriate and we have had the information to compose an effective measurement set-up, this damage could have been detected in a much earlier state. Never a less the project should be carried on using other turbines (phase 2), so we will have a second change to demonstrate the benefits and opportunities of the CONMOW project.
Conclusion
As only two turbines of same type, although over some time, has been producing data the material is too limited to finish the learning period for this specific type of machinery. No severe faults have been detected but beginning signs of wear can be seen. The Signature Fault Detection technique can be improved and combined with the complementary Mask Fault Detection also used in the TCM system, an having sufficient experience with the specific machinery, i.e. enough wind turbines, to concluded the learning period, a good foundation to make predictive condition monitoring is available. With these techniques, implemented in an automated expert system as the TCM System, it is possible to monitor many wind turbines in an automated way. The outcome of the CONMOW project contributes to the possibility to implement renewable energy as wind power in large scale - and keep it up running.
Throughout the CONMOW project, turbine(s) will be modified with the installation of inline/online gearbox lubricating fluid and filtration diagnostic monitoring equipment. This will provide both fluid cleanliness [per ISO4406 or AS4059D standards] and fluid water contents [%Saturation and temperature] with data output for proactive condition monitoring programmes.
This data will allow for early detection of abnormal operating conditions, further investigations, scheduled maintenance and/or corrective actions can then be implemented in a timely manner. System components will be operating within specified limits with respect to cleanliness and this will help ensure continued operation, uptime and productivity.
Filtration and online/inline Fluid diagnostic condition monitoring products yet to be installed.
Further details will be provided in the final version of the TIP.
Three condition monitoring system were installed at Nordex N80 turbines. The one VIBROWEB® XP, a certified system according to GL-Regulations, was working properly during the entire test period. Unfortunately the system was supposed to be added with one oil- monitoring device provided by Pall but the project time expired before the installation could be realized.
The measurement results have shown that machine failures, like misalignment, as well as bearing and electrical failures were detected. Specific measurement set-ups including the wide- and narrow- band trending were working properly to analyse the typical machine behaviour of the drive train. In addition two VIBNODE® devices were installed which have been developed for smaller turbines (stall system). These systems were equipped with additional displacement sensors next to its standard use of ICP-type transducers. The measurements have impressively shown the movement of the input shaft during typical operating condition up to high wind periods.
Out of several experiences one VIBNODE® was equipped with a newer controller version to use specific developed set-ups for monitoring especially low frequency ranges up to 25Hz. Besides to the installed hardware, further developments of the reporting tool OMNITREND® Web were realized and implemented. Therefore a remote monitoring of all installed condition monitoring systems was established via Internet for all registered users. The project has provided a solid platform for constant developments on the hard- and software to improve the efficiency of the condition monitoring systems and to secure a stable operation particularly with regard to further offshore wind applications.
A monitoring system in a small wind farm of 5 turbines has been applied consisting of drive-train vibration monitoring systems, each with different sensor configurations and analysis capabilities. This was combined with SCADA logging and traditional measurement systems. Several online oil-monitoring systems were selected and, although not implemented, the integration with the existing CM-systems was prepared.
A data communication network was implemented for online system configuration and automated data upload. Web-based reporting of CM-systems analysis results was implemented such that reports of user-selected time-periods and subsystems are generated automatically. Also automated reporting of SCADA statistical data was implemented.
Algorithms for analysis of high-frequency data have been developed focused on the electrical power output.
Mathematical processing techniques have been developed to help detect potential problems/future failure under two main headings:
1) Medium frequency (30-32Hz) data - analysis of the power signal by use of wavelets and FFTs to extract details in the signal indicative of changes to the generator flux field by looking at the slip frequency and associated side bands. This has been used to detect generator shaft misalignment
2) Low frequency (10-min) SCADA data - trending of temperature signals over time to detect potential bearing faults, pitch faults or yaw/anemometer/controller faults.
3) State of the art review of condition monitoring techniques and recommendations for future use in wind turbines.
The techniques are being used in a UK consortium project and being further developed in conjunction with wind farm operators using existing SCADA/error log data. The aim is to improve overall availability of wind turbines by pro-active and predictive maintenance.
Searching for OpenAIRE data...
There was an error trying to search data from OpenAIRE
No results available