Condition Monitoring

for Wind Power Stations

Introduction of Condition Monitoring

In modern industrial enterprises the condition of manufacturing assets is a critical success factor to their business. An unforeseeable breakdown of one or more machines can cause logistical problems as well as losses of money, business partners or customers. Conditional monitoring is a

modern approach for maintenance of machines and allows through frequent or permanent recording and analyse of physical factors like vibrations or oil temperature to diagnose the actual condition of monitored machines. The main targets of conditional monitoring are e.g. early identification of

damages, safety of machines and people, efficiency of machines and avoidance of unforeseeable breakdowns [1]. The development of a modern Conditional Monitoring System (CMS) demands knowledge of mechatronic (mechanic, electrical and electronic engineering, measurement

technology and informatics) and advanced statistic (with growing importance).

Characteristics of Wind Power Stations

Wind Power Stations (WPS) are in most cases remote-controlled and often build in uninhabited areas. Inspections of plants are most only restricted possible. An unforeseeable breakdown of the power train can cause the operator the loose of earnings of several days or even weeks.

Conventional CMS are not suited for the use in WPS caused by the unsteady operating behaviour.

Research and Development

The research group of Measurement and Actuator Technology research and develop CMS for the special use in WPS. Our main targets are to develop and improve our CMS that offers reliable information about the condition of WPS and the reduction of the daily monitoring workload.

SPECTIVE

The developed System consists of hardware and Software for Powerful Envelope analysis and Calculation of Traces for Immediate Visualisation of Emergencies (SPECTIVE, analysing measurement data and viewing of results). The hardware primary consists of several acceleration sensors for

the measurement of the vibrations of the machine (standard configuration: eight), one sensor to measure the rotation speed, a programmable modular Controller (CompactRIO, National Instruments) for processing the data and an industrial PC for the storage of data and communication with the

server, see Figure 1. The software for data analysing is installed on a server. It offers state-of-the-art analysing features (e.g. Fast Fourier Transformation (FFT), Crest Factor, Kurtosis, RMS and Peak Level) and furthermore additional tools specially designed for WPS Condition Monitoring (e.g.

Frequency Lupe Spectrum, Cladding Curve Analyse, Cepstrum and Bearing Condition Indicator Value), see Figure 2. Figure 3 shows a spectrum of a WPS analysed with an industrial CMS. Figure 4 and 5 show the same data analysed with SPECTIVE. The pictures show that SPECTIVE is more

detailed and more precise than the state-of-the-art industrial one. The maximum amplitude at 602 Hz is more than four times higher than compared to the other one. The CMS SPECTIVE has been developed in cooperation with an industrial partner [2]. The hardware was installed in one of our

partners WPS and works reliable since two years without problems. The installed CMS software package is in daily use and helps to early identify potential defects on 46 WPS.

GetriebeStudio

Parallel to this analyse software the program GetriebeStudio was developed. It enables modelling of power trains and management of model data, see Figure 6. The program calculates important information e.g. meshing frequencies, which SPECTIVE use to assign frequency lines in a FFT to

specific components of the power train, see Figure 2.

SPECTIVExpert

The newest CMS-product is the program SPECTIVExpert, which is a Condition Monitoring Expert System (CMES). It uses multivariate statistics to determine the condition of each component of the power train based on the SPECTIVE output data. The condition of WPS is showed in

multiple views; see Figure 7 to 9.

Test stand

For future development and evaluation of our CMS a new test stand was build, see Figure 10. The key aspect of the test stand is a new developed gearbox, which enables switching between two possible configurations. The first one is a planet gear configuration, the second a spur gear one, see

Figure 11, 12 and 13. Both are commonly found in wind power trains. The gearbox design focus on the fast exchange of components (gear wheels and bearings see Figure 14). This enables to verify the efficiency of our CMS detecting on purpose damaged gearbox components.

SPECTIVE - Post analysis of known damage cases

1. Bearing damage: Figure 15 shows an endoscopy picture of a heavy damaged bearing. The used industrial CMS was not able to detect it. A post analysis of the stored data with SPECTIVE shows that the damage occurred three and a half years before the endoscopy (16. May 2013), see

Figure 16.

2. Broken tooth of a gear-wheel: Figure 17 shows an endoscopy picture of a broken tooth of a gear-wheel. The missing piece is one third of the length of the whole tooth. The used industrial CMS was again not able to detect the damage. A post analysis with SPECTIVE shows that the

damage occurred approximately five month before the endoscopy (15. May 2013), see Figure 18.

Publication

[1] Josef Kolerus, Johann Wassermann, German, „Zustandsüberwachung von Maschinen“, 4th edition, expertverlag

[2] Erwin Quintus, dissertation, German, 2014, „Entwicklung eines hochsensitiven Zustandsüberwachungssystems für Antriebsstrang von Windkraftanlagen“

Contact

Figure 1: Hardware

Figure 2: SPECTIVE - User Interface

Figure 6: GetriebeStudio - Model of a WPS power train

Figure 3: Spectrum - industrial provider

Figure 4: Spectrum - SPECTIVE

Figure 7: SPECTIVExpert - overview

Figure 8: SPECTIVExpert - detail view - table

Figure 5: Spectrum as shown in picture 3

and 4 with higher frequency

resolution – SPECTIVE

Figure 9: SPECTIVExpert - detail view - with geometry of

the analysed power train

Figure 10: Test stand

Figure 11: Cassettes used in spur gear configuration with

different bearings and with/without damaged components

Figure 12: Gear box - Spur gearbox configuration

Figure 13: Gear box - planet gearbox configuration

Figure 14: Gearbox components

Figure 15: Endoscopy -

Bearing damage

Figure 16: Post analysis with SPECTIVE - Trend of the amplitude of the overroll frequency

Figure 17: Endoscopy of a tooth

break on a gear-wheel

(recorded on 15.5.2013)

Figure 18: SPECTIVE post analysis - Trend of the amplitude of the third lower side band in the

frequency spectrum

Institute of Mechanics

and Mechatronics

Research group of Measurement

and Actuator Technology