AKTUELLE FORSCHUNGSERGEBNISSE ZUR INSTANDHALTUNG UND WEITERBETRIEB VON WINDTURBINEN Lisa Ziegler 26. Windenergietage Warnemünde, 08.11.2017
AWESOME AWESOME = Advanced wind energy systems operation and maintenance expertise Funded by European Commission 11 PhD s O&M - Failure diagnostic and prognostic - Maintenance scheduling - Lifetime extension www.awesome-h2020.eu
AGENDA 1. Latest results from AWESOME 2. Lifetime extension of wind turbines State-of-art in Germany, Spain, Denmark, and the UK Decision-making: When to switch from lifetime extension to repowering? 3. Conclusion
AWESOME Fault detection based on SCADA Reliability models Stochastic maintenance planning Lifetime extension of wind turbines Other projects: Cost optimization Predictive maintenance Monitoring of induction generators Wind farm control Very short-term wind field forecast
SELECTED RESULTS Fault detection based on SCADA - SCADA data available without extra costs - Data-driven modelling of normal behaviour to indicate potential faults - Drive-train temperatures can predict mechanical problems Constant - average from training (Trivial) Linear modelling (LIN) Linear modelling with interactions (LIN-I) Artificial neural networks in feedforward configuration (ANN-FF) Artificial neural networks in layer recurrent configuration (ANN-LR) Gaussian process regression (GPR) Support vector machine (SVM) Adaptive neuro-fuzzy inference systems (ANFIS) Source: Tautz-Weinert J & Watson SJ (2016) Comparison of different modelling approaches of drive train temperature for the purposes of wind turbine failure detection, Journal of Physics: Conference Series, 753(072014).
SELECTED RESULTS Reliability models - What effect have weather conditions on component reliabilities? - Failure rate and downtime analysis for turbines >1MW - Increase of extreme weather events before failure Figure left: Reder M & Melero JJ (2017). Time series data mining for analysing the effects of wind speed on wind turbine reliability. Proceedings of the 27 th European Safety and Reliability Conference, 18-22 June 2017, Slovenia. Figure right: Reder MD et al (2016). Wind Turbine Failures - Tackling current Problems in Failure Data Analysis. J. Phys.: Conf. Ser. 753 072027.
SELECTED RESULTS Stochastic maintenance planning - What effect has uncertainty in maintenance planning? - Important for repair time and weather forecasts - and for more inputs? Seyr H & Muskulus M (2016). Value of information of repair times for offshore wind farm maintenance planning. J. Phys.: Conf. Ser. 753 092009.
AGENDA 1. Latest results from AWESOME 2. Lifetime extension of wind turbines State-of-art in Germany, Spain, Denmark, and the UK Decision-making: When to switch from lifetime extension to repowering? 3. Conclusion
WIND TURBINES AT END-OF-LIFE Large number of turbines approach the end of their design lifetime soon Today s turbines approaching end-of-life 20 are small size Decision between lifetime extension, repowering and decommissioning of site 2016 2020 2025 2030 <0.5 MW 0.5-1 MW 1-2 MW 2-3 MW >3 MW Source: Ziegler L et al. (2018). Lifetime extension of onshore wind turbines: A review covering Germany, Spain, Denmark, and the UK. Renewable and Sustainable Energy Reviews, 82(1), 1261-1271.
TECHNICAL ASPECTS Design lifetime min. 20 years according to IEC 61400-1 Analysis of fatigue limit state for lifetime extension Practical assessment Detailed inspection and review of maintenance history Analytical assessment Site-specific load simulations with (generic) aero-elastic model Data-driven assessment Use of operational and measurement data Fast and low costs Cannot predict RUL directly Estimation of RUL Requires site conditios and models Tracks load history Requires extrapolation
ECONOMIC AND LEGAL ASPECTS Germany: Fixed feed-in tariff until 2020 for old wind turbines UK: Renewable Obligation Scheme or Contract for Difference Spain and Denmark: Electricity spot market for old wind turbines Source: Ziegler L et al. (2018). Lifetime extension of onshore wind turbines: A review covering Germany, Spain, Denmark, and the UK. Renewable and Sustainable Energy Reviews, 82(1), 1261-1271.
INTERVIEW RESULTS 24 guideline-based interviews with key market players > Operators, manufacturers, independent experts, certifiers,... Motivations for lifetime extension Increase return on investment Site impossible or uneconomic to repower Uncertainty for new projects Better public acceptance Large availability of new sites Lifetime extension Repowering Source: Ziegler L et al. (2018). Lifetime extension of onshore wind turbines: A review covering Germany, Spain, Denmark, and the UK. Renewable and Sustainable Energy Reviews, 82(1), 1261-1271.
INTERVIEW RESULTS Parameter Germany & Spain Denmark UK Analytical assessment Generic aero-elastic models to reassess site-specific loads None Wind history; occasionally loads Practical assessment Monitoring Extended inspection; O&M history SCADA; no load measurements or monitoring (few exceptions) Frequency Once (AA); maybe periodical (PA) Annually Every 3-5 years Costs 5000-15000 single turbine (DE) 1500 per turbine Not specified Concerns Uncertainty in electricity market price Spare parts unavailble Access to design information Unclear authority requirements
DECISION MAKING LIFETIME EXTENSION IF Market price electricity ~35 /MWh > Operational costs?? /MWh Simple question - but how to do it?
LIFETIME EXTENSION IF Availability Maintenance Energy production Market price electricity ~35 /MWh > Operational costs?? /MWh
LIFETIME EXTENSION IF Wind resource power curve Availability Maintenance failure rates repair costs technicians weather Energy production Market price electricity ~35 /MWh > Operational costs?? /MWh The better the data basis the better the decision!
DECISION MODEL Deterministic model based on expected values (averages) Optimisation of length of lifetime extension Objective function = maximise capital value Applicable when you are certain about your input for large asset numbers Case study Offshore wind farm Old: 100 turbines 3MW New: 60 turbines 6MW
MAINTENANCE MODEL - SIMPLIFIED Failure rates Major replacement, major repair, minor repair, no cost Wear out components Constant failure rate components Wear out and remaining useful lifetime 1. Data driven - Own data? Extrapolation? - Reference data? 2. Physics driven - Physical models - Monitoring and inspections Wear out Pitch/ hydraulic Other components Generator Gearbox Blades Grease oil/ liquid Pumps/ motors Hub Heaters/ coolers Yaw system Tower/ foundation Transformer Constant Electrical Circuit breaker Controls Safety Sensory Power/ converter Service item
Turbine [failure/year] Generator [failure/ year] MAINTENANCE MODEL - SIMPLIFIED Bathtub curve Early failures 2 1.5 1 Early failure Constant failure rate replacement major repair minor repair Wear out Constant failure rate Wear out 0.5 Fit to 8-year of data Weibull distribution failure rate = λ β (λ t) β 1 λ: shape parameter (data fit) β: scale parameter (data fit, 1, 4) 0 0 5 10 15 20 25 30 Lifetime [years] 12 10 8 6 4 replacement major repair minor repair 2 0 0 5 10 15 Lifetime [years] AW E S O M E
Availability [%] Net present value [%] CASE STUDY RESULTS 106 104 102 100 OPEX-based 5 years optimum 98 96 Direct repowering, no lifetime extension 94 0 2 4 6 8 10 Lifetime extension [years] Drop afterwards due to Increase failure rates Increase maintenance costs Decrease availability Decrease revenues 90 85 80 75 Lifetime extension 70 0 2 4 6 8 10 Lifetime extension [years]
CONCLUSION Research in O&M helps to reduce cost of energy. Target Safety Level Large market for lifetime extension; strongly influenced by country-specific aspects. Germany and Denmark leading in consistent technical assessment. Today s procedure suits small turbines with limited monitoring data. Decision for lifetime extension driven by uncertain electricity spot market price. Risk OPEX CAPEX Strategy 1 Strategy 2
THANKS! Lisa Ziegler PhD researcher M +49 151 44 006 445 lisa.ziegler@ramboll.com Ramboll Wind Hamburg, Germany www.ramboll.com/wind
HOW TO DETERMINE THE REMAINING LIFE? Wind turbines are designed to withstand a design lifetime of min. 20 years according to IEC 61400-1 with an appropriate target safety level for several limit states and failure modes Analysis of fatigue limit state Aerodynamic loading and rotor excitation 10 8-10 9 load cycles over lifetime Damage calculation with SN-curves D i n N i i n i : number of occurred stress cycles N i : number of stress cycles until failure
HOW TO DETERMINE THE REMAINING LIFE? Structural reserves exist, if loading is lower, or Assessment Environmental conditions more benign than IEC class Advances in numerical modelling material capacity is larger than in design practical analytical datadriven