Estimation of Turbine Reliability figures within the DOWEC project DOWEC Nr. 10048 Issue 4 name: Signature: Date: Written by: DOWEC team 22-01-02 Checked by: Approved by: version date No of pages 0 19-12-01 19 1 21-01-02 20 Costs of repair and maintenance added 2 15-03-02 16 Repair time and maintenance costs deleted 3 17-06-02 19 6.5 Explanation reliability improvement added 4 22-10-03 Public version
Contents 1 General...2 2 Failure rate data...3 2.1 WindStats Newsletter...3 2.2 WMEP Germany...4 2.3 Landwirtschaftskammer Schleswig-Holstein...6 2.4 EPRI California...7 3 Comparison of the failure rates...8 4 Conclusions...11 5 References...12 6 Appendixes...13 6.1 WindStats failure data...13 6.2 WMEP Germany failure data...14 6.3 LWK Schleswig-Holstein failure data...14 6.4 EPRI California failure and repair time data...15 6.5 Explanation of reliability improvements offshore turbine...15 6.5.1 Shaft & Bearing...15 6.5.2 Brake...15 6.5.3 Generator...15 6.5.4 Parking brake...15 6.5.5 Electric...16 6.5.6 Blade...16 6.5.7 Yaw system...16 6.5.8 Blade tip...16 6.5.9 Pitch Mechanism...16 6.5.10 Gearbox...16 6.5.11 Inverter...16 6.5.12 Control...16 1 GENERAL This report gives a survey of the efforts to obtain the reliability figures of windturbines in general and of the DOWEC turbine in specific. This is done within the DOWEC concept study (task 17) in order to make an estimation of the overall costs per kwh. ECN and TUDelft perform these cost analysis and optimisation using their own theoretical models. The reliability figures mentioned in this report will be used as input for these models. In chapter four of the DOWEC report: Starting-point and Methodology of Cost Optimisation for the Conceptual Design of DOWEC [1], estimated failure rates are already mentioned. The estimation of this failure rate of the whole turbine was based on figures from wind energy data out of North-Germany [4]+[5]. The distribution over the turbine main components was made during a brain storm session. This report has to give a more realistic update of these figures. The same distribution over the main components is used. Remark: Failure is all events that ask for a service crew call-out; and can range from minor issues to larger repairs! 10048_004.doc Page 2 of 16 22-10-03
2 FAILURE RATE DATA To obtain reliability figures of windturbines and its components all kind of sources has been opened up. Extensive search on the Internet and consulting colleagues resulted in the following sources: WindStats Newsletter [2, WMEP [3], Landwirt-schafskammer Schleswig-Holstein [4] to [7] and EPRI [8]. 2.1 WindStats Newsletter Production and failure data of wind turbines out of Denmark and Germany are published every month in this quarterly. The failure data are even specified per component, but not per windturbine size or type. Every month failures of about 2000 turbines are reported from Denmark and about 2750 from Germany. Unfortunately the age of the turbines is unknown and the data from Denmark is collected in a different way comparing to the data form Germany. Every quarter this results in figures distributed over different components. If we look at only the mechanical failures of Germany, the average failure rate (from 4th quarter 1999 to 2nd quarter 2001) of the whole turbine is 3.0. For Denmark this is 0.7, which is 4 times less. See the next page for figure 1 and appendix 1 for the data. The reason for this difference is not known. Maybe in Denmark the turbines are smaller. Maybe in Germany the collected data are more profound due to the Deutsche Grundlichkeit. But the difference is striking. Figure 1. Mechanical failure rates according to WindStats. For the distribution of the failures over the different components see figure 2 on the next page. The difference between Denmark and Germany is again very large. This can be caused by other kinds of components. For instance for Denmark pitch 10048_004.doc Page 3 of 16 22-10-03
adjustment, instrumentation are not specified. And for Germany hub, coupling, tower, foundation and mechanical control are not specified. See the appendix 1 also for the distributing of the WindStats components to the report components. Figure 2. Mechanical failure rates distribution according to WindStats. 2.2 WMEP Germany In Germany at the ISET (Institut für Solare EnergieversorgungsTechnik) of the University of Kassel every year failures are collected per wind turbine power class. The data is available from 1998 up to 2000 of 1435 turbines. These data is even distributed over components and causes. Unfortunately the main part of the data (95%) is of the smaller turbines; under the 560 kw and the age of the turbines is not known. In the graphs a distinguishing has been made between the smaller and larger powers. The average mechanical failure rate of the whole turbine is 2.4 for all turbine powers and 5.2 for 560 to 1500 kw. See appendix 2 for the data. 10048_004.doc Page 4 of 16 22-10-03
Figure 3. Mechanical failure rates according to WMEP. Because the larger turbines are younger the failure rate is much worse (factor 2.2). This effect can also be seen in the WMEP data of the large turbines. During the measurement period of three years of the same 70 large turbines the failure rate decreases, from 6.4 to 4.7 and 4.6. The distribution over the components is pointed out in figure 4. It is peculiar that there is hardly any difference between both power classes. One would expect other type of problems with younger or larger turbines. See the appendix 2 also for the distributing of the WMEP components to the report components. Figure 4. Mechanical failure rates distribution according to WMEP 1998-2000. 10048_004.doc Page 5 of 16 22-10-03
2.3 Landwirtschaftskammer Schleswig-Holstein In the northern part of Germany the province Schleswig-Holstein is located. The output data and failures of all turbines in his province are collected and presented every year [4] - [7]. For the years 1997 to 1998 the average failure rate of the turbines, larger than 500 kw, is 2.2 per year [1]. In this report only the results of year 1999 and 2000 are presented, because I do not have the details of the previous years. For each type and size of the 510 monitored turbines the data is mentioned. The greater part of these turbines (57%) is larger than 500 kw. Unfortunately the age of each turbine is unknown. The average mechanical failure rate of the whole turbine is 1.7 for all turbine powers and 1.9 for 500 to 1500 kw. The latter value is slightly better than the 2.2 of 1997 and 1998. See appendix 3 for the data. At LWK there is again hardly any difference between both power classes for the distribution over the components. But comparing to WMEP the distribution differs. See the appendix 3 also for the distributing of the LWK components to the report components. Figure 5. Mechanical failure rates according to LWK Schleswig-Holstein. 10048_004.doc Page 6 of 16 22-10-03
Figure 6. Mechanical failure rates distribution according to LWK 1999-2000. 2.4 EPRI California In the south-eastern part of United States of America the state California is located. During 1986 and 1987 output data and failures of 290 turbines are collected and presented by the Electric Power Research Institute (EPRI) [8]. For these years the average mechanical failure rate of the turbines was 10.0 per year. This is very high, probably due to the old standard of technique. The power range of these turbines is 40 to 600 kw. Unfortunately this is rather small and the age of each turbine is also unknown. See appendix 4 for the data, also for the distribution of the EPRI components over the report components. 10048_004.doc Page 7 of 16 22-10-03
Figure 7. Mechanical failure rates according to EPRI California. Figure 8. Mechanical failure rates distribution according to EPRI 1986-1987. 3 COMPARISON OF THE FAILURE RATES There is a huge difference in average failure rates of wind turbines. It depends highly upon the source of the data, the size and age of the turbines and the method of data collecting. The EPRI data is considered as too old. The figures are very bad due to the old standard of technique. When we compare the other sources we get the next graph. 10048_004.doc Page 8 of 16 22-10-03
Figure 9. Failure rate comparison over the years. When we look at all power classes, the failure rate differs from 0.6 to 5.4. When we look at the large powers, the failure rate differs from 1.9 to 6.4. It is hardly possible to extract an average value due to the large scatter of values. When we are looking at the failures per component we see also a large scatter. See figure 14. To obtain the right figures for all components the three latter components (other, instrumentation and hydraulics) need to be distributed over the others. That is not possible with the present data. Average values can hardly be extracted due to the wide variations of values. 10048_004.doc Page 9 of 16 22-10-03
Figure 10. Failure rate ratio comparison between the different data sources. During a brainstorm session estimations have been made of the state-of-the-art onshore values and new failure rates of an offshore turbine with a price increase. The result is shown below. Figure 11. Failure rate onshore and offshore. 10048_004.doc Page 10 of 16 22-10-03
4 CONCLUSIONS - Before collecting data it is necessary to think how to collect them in order to make the right conclusions. In this report it is tried to compare all kind of data which has been collected differently. This was not really possible because there was always another parameter not measured. In the following table a cross illustrates the missing parameters: Source Age Size Turbine type Cause Windstats Germ. X X X Windstats Denm. X X X X WMEP Germany X X LWK Schlesw.-H. X EPRI California X X Figure 12. Missing parameters during data collection. Remark: Cause means: has a distinction been made between: grid, weather, maintenance, mechanical failure, etc. - The EPRI data have to be omitted because they are too old. The standard of technique in 1986 and 1987 is not comparable with what it is today. - The age of the turbine (type) is very important by determine the failure rate, new turbines have three times as much failures as turbines of at least 4 years old. - New types of turbines have not significant higher failures than older types of the same age. The complexity of the components increases but also the quality. - The failure rate of an onshore turbine without commissioning is considered to be 2.3 failures per year. 10048_004.doc Page 11 of 16 22-10-03
5 REFERENCES [1] M.B. Zaaijer: Starting-point and Methodology of Cost Optimisation for the Conceptual Design of DOWEC, Report T9-MZ-99-R01-D, March 23th 2000, TUDelft, The Netherlands [2] WindStats Newsletter, Volume 12 No 4 (Autumn 1999) to Volume 14 No 3 (Summer 2001), Denmark [3] Wissenschaftliches Meß- und EvaluierungsProgramm Jahresauswertung 1998-2000, Institut für Solare EnergieversorgungsTechnik, Universität Gesamthochschule Kassel, Germany [4] W. Eggersglüß: Windenergie X Praxisergebnisse 1997, Landwirtschaftskammer Schleswig-Holstein, Germany [5] W. Eggersglüß: Windenergie XI Praxisergebnisse 1998, Landwirtschaftskammer Schleswig-Holstein, Germany [6] W. Eggersglüß: Windenergie XII Praxisergebnisse 1999, Landwirtschaftskammer Schleswig-Holstein, Germany [7] W. Eggersglüß: Windenergie XII Praxisergebnisse 2000, Landwirtschaftskammer Schleswig-Holstein, Germany [8] EPRI 1986-1987, California, USA 10048_004.doc Page 12 of 16 22-10-03
6 APPENDIXES 6.1 WindStats failure data 10048_004.doc Page 13 of 16 22-10-03
6.2 WMEP Germany failure data 6.3 LWK Schleswig-Holstein failure data 10048_004.doc Page 14 of 16 22-10-03
6.4 EPRI California failure and repair time data 6.5 Explanation of reliability improvements offshore turbine In figure 12 the final reliability figures are given of an onshore turbine. Also estimation is made of these figures for an offshore version. Because offshore turbines are more difficult to reach the reliability is more important than for onshore turbines. So for offshore turbines more attention shall be paid on components with high failure rates. During a brainstorm session estimations have been made of the new failure rates of an offshore turbine with a price increase. For each component an explanation is given of the estimated reliability compared to the onshore version. 6.5.1 Shaft & Bearing Beacause the shaft is a simple part, reliability improvements are not expected. Altough the size increases the manufacturing techniques stays the same, so also the failure rate stays the same. The principle of the main bearing stays the same so new improvements are not expected. 6.5.2 Brake We expect that the principle of the brake remains the same. So the same number of parts, so the same reliability. 6.5.3 Generator We expect an improvement with a factor 2. A larger cooling and greasing system cause the extra costs. 6.5.4 Parking brake The principle of the (rotor) parking brake will stay the same; a manually acted pin. But we hope to make it an add-on tool, because it have to be used rarely. In that case we could expect a lower failure rate. But 0.01 is already very low, so we keep it that way. 10048_004.doc Page 15 of 16 22-10-03
6.5.5 Electric The overall conclusion was that we stayed to the same failure rate. 6.5.6 Blade The construction of the three blades will not change. The chance for a stroke of lightning will increase offshore but with the lightning protection experience we will stay at the same failure rate. 6.5.7 Yaw system A number of improvements was defined. Because of this we expect an improvement with a factor 1.5. Heavier motors mainly cause the price increase. 6.5.8 Blade tip The pitch-regulated blades do not have blade tips anymore. So the new failure rate is zero. Still there is a little price increase because now the blades have to be longer (in one piece) which causes higher mould, storage, transportation costs. 6.5.9 Pitch Mechanism The failure rate of the pitch mechanism is based on data out of report T9-MZ-99 of M.B. Zaaijer [1]. We expect an improvement with a factor 1.5. The price increase is expected on more sophisticated pitch system and higher reliable (electrical) components. 6.5.10 Gearbox We expect a 40% improvement. The price increase is caused by larger gearbox and cooling system. 6.5.11 Inverter This failure rate is based on data out of report T9-MZ-99 of M.B. Zaaijer [1]. At the time the 6MW turbine is being build we expect that the reliability of the inverter is improved with 25%. The price increase is expected on higher reliable (electrical) components. 6.5.12 Control We expect to improve the reliability of the control system with 45%. The cost increase is mainly caused by more expensive electrical components. 10048_004.doc Page 16 of 16 22-10-03