maintenance - norcowe 2016 presentation… · summary and highlights from norcowe maintenance. 2...
TRANSCRIPT
1
John Dalsgaard Sørensen Aalborg University, Denmark
• Introduction • AAU: Reliability and risk-based planning of O&M• UiS: Condition-based maintenance for offshore wind parks• UiA: Wind farm control including operation and
maintenance aspects• AAU: Wind farm control strategy• Concluding remarks
Summary and highlights from NORCOWE Maintenance
2
Introduction
Minimize the Total Expected Life-Cycle Costs
Minimize Levelized Cost Of Energy (LCOE)
140 (Anholt) 103 (HR 3) 64 (Near-Shore) [€/MWh]
• Dependent on Reliability Level Initial Costs
• Dependent on O&M Strategy, Availability and ReliabilityOperation &
Maintenance Costs
• Dependent on ReliabilityFailure Costs
6
• Corrective (unplanned): exchange / repair of failed components
• Preventive (planned):– Time-tabled: inspections, and evt. repair after predefined
sheme– Condition-based: monitor condition of system and decide
next on evt. repair based on degree of deterioration• Risk-based methods based on Bayesian decision theory
Operation and maintenance of wind turbines
7
Reliability modeling of wind turbines – exemplified by power converter systems as basis for O&M planning
PhD study: Erik Kostandyan, Aalborg University. 2009-2013
Objectives: • Develop reliability models for O&M planning
– Damage accumulation models for power electronics components – physics-of-failure models
– Reliability model(s) for selected electrical component: IBGT module
• to be integrated into O&M strategies for planning and development
8
Reliability modeling of wind turbines – exemplified by power converter systems as basis for O&M planning
10
Reliability modeling of wind turbines – exemplified by power converter systems as basis for O&M planning
• Reliability model of failure mode with crack growth in solderlayer developed, and using limit state equations using structuralreliability methods• Application for O&M strategy development• Application for DLC 2 reliability modeling considering faults
11
Risk-and reliability-based planning of inspections and maintenance
• Reliability-based cost-optimal planning of inspections andmaintenance for fatigue critical welded steel details in tower andsubstructures
• Maximum annual probability of failure = 5 10-4
• Used as background and basis for reduction of partial safetyfactors in CDV IEC 61400-1 ed. 4:2016
• Used in probabilistic design for wind turbines
12
Publications
• Kostandyan, E., & Sørensen, J. (2012). Structural reliability methods for wind power converter system component reliability assessment. 16th IFIP WG 7.5 Conference on Reliability and Optimization of Structural Systems, Yerevan, Armenia. pp. 135-142.
• Kostandyan, E., & Sørensen, J. (2012). Weibull parameters estimation based on physics of failure model. Industrial and Systems Engineering Research Conference (ISERC 2012), 62nd IIE Annual Conference & Expo2012, Orlando, Florida, USA. pp. 10.
• Kostandyan, E., & Sørensen, J. (2013). Reliability assessment of offshore wind turbines considering faults of electrical / mechanical components. 23rd International Offshore (Ocean) and Polar Engineering Conference (ISOPE 2013), Anchorage, Alaska, USA. pp. in press.
• Kostandyan, E. E., Lamberson, L. R., & Houshyar, A. (2010). Time to failure for a k parallel r-out-of-n system.International Journal of Modelling and Simulation, 30(4), 479-482.
• Kostandyan, E. E., & Ma, K. (2012). Reliability estimation with uncertainties consideration for high power IGBTs in 2.3 MW wind turbine converter system. Microelectronics Reliability, 52(9-10), 2403-2408.
• Kostandyan, E. E., & Sorensen, J. D. (2012). Physics of failure as a basis for solder elements reliability assessment in wind turbines. Reliability Engineering and System Safety, 108, 100-107.
• Kostandyan, E. E., & Sorensen, J. D. (2012). Reliability of wind turbine components — solder elements fatigue failure. Reliability and Maintainability Symposium (RAMS), 2012 Proceedings - Annual, pp. 1-7.
• Kostandyan, E. E., & Sørensen, J. D. (2011). Reliability assessment of solder joints in power electronic modules by crack damage model for wind turbine applications. Energies, 4(12), 2236-2248.
• Kostandyan, E. E., & Sorensen, J. D. (2013). Reliability assessment of IGBT modules modeled as systems with correlated components. Reliability and Maintainability Symposium (RAMS), 2013 Proceedings - Annual, pp. 1-6.
13
Risk and Reliability based O&M Planning of Offshore Wind Farms
Industrial PhD study: Masoud Asgarpour, ECN / Vattenfall + Aalborg University. 2013-
Objectives: • Development of Bayesian and risk-based O&M techniques for
multiple component WT systems• Development of Reliability Matrix approach using CMS data• Updating the deterioration models used in the Reliability Matrix• Updating the model to include neighbor wind farms
14
Reliability Matrix for critical failure modes
Deterioration per Month/Park/Turbine/Component, updated using available information
Date Park Turbine Component Calculated "D" Reported "D" Reporting Method
2015 Jan WHV A01 MDA‐Rotor 0.2371002015 Jan WHV A01 MDC‐Pitch 0.6183302015 Jan WHV A01 MDK‐Drivetrain 0.1176592015 Jan WHV A01 MDL‐Yaw 0.7198482015 Jan WHV A01 MKA‐Generator 0.3835192015 Feb WHV A01 MDA‐Rotor 0.2375742015 Feb WHV A01 MDC‐Pitch 0.6208032015 Feb WHV A01 MDK‐Drivetrain 0.117777 0.3 Inspection2015 Feb WHV A01 MDL‐Yaw 0.7227272015 Feb WHV A01 MKA‐Generator 0.3846702015 Mar WHV A01 MDA‐Rotor 0.2380492015 Mar WHV A01 MDC‐Pitch 0.6232872015 Mar WHV A01 MDK‐Drivetrain 0.3051002015 Mar WHV A01 MDL‐Yaw 0.7256182015 Mar WHV A01 MKA‐Generator 0.385824 0.2 CMS2015 Apr WHV A01 MDA‐Rotor 0.2385252015 Apr WHV A01 MDC‐Pitch 0.6257802015 Apr WHV A01 MDK‐Drivetrain 0.3054052015 Apr WHV A01 MDL‐Yaw 0.728521 1.0 Failure2015 Apr WHV A01 MKA‐Generator 0.2006002015 May WHV A01 MDA‐Rotor 0.2390022015 May WHV A01 MDC‐Pitch 0.6282832015 May WHV A01 MDK‐Drivetrain 0.3057112015 May WHV A01 MDL‐Yaw 0.0082702015 May WHV A01 MKA‐Generator 0.201202
Baysian updating
Baysian updating
Rotor deteriorationmodel
Baysian updating
15
Maintenance Matrix based on Bayesian decision rules
Date Park Turbine Component Scheduled Service Condition‐based Inspection Condition‐based Repair Corrective Replacement
2015 Jan WHV A01 MDA‐Rotor2015 Jan WHV A01 MDC‐Pitch2015 Jan WHV A01 MDK‐Drivetrain2015 Jan WHV A01 MDL‐Yaw2015 Jan WHV A01 MKA‐Generator…2015 Jul WHV A01 MDA‐Rotor2015 Jul WHV A01 MDC‐Pitch2015 Jul WHV A01 MDK‐Drivetrain2015 Jul WHV A01 MDL‐Yaw2015 Jul WHV A01 MKA‐Generator…2016 Mar WHV A01 MDA‐Rotor2016 Mar WHV A01 MDC‐Pitch2016 Mar WHV A01 MDK‐Drivetrain2016 Mar WHV A01 MDL‐Yaw2016 Mar WHV A01 MKA‐Generator…2016 Jul WHV A01 MDA‐Rotor2016 Jul WHV A01 MDC‐Pitch2016 Jul WHV A01 MDK‐Drivetrain2016 Jul WHV A01 MDL‐Yaw2016 Jul WHV A01 MKA‐Generator…2017 Dec WHV A01 MDA‐Rotor2017 Jan WHV A01 MDC‐Pitch2017 Jan WHV A01 MDK‐Drivetrain2017 Jan WHV A01 MDL‐Yaw2017 Jan WHV A01 MKA‐Generator
16
Publications
• State of the art in O&M planning of offshore wind farms– Presented at European Safety & Reliability Conference (ESREL) - Poland– Published by Taylor & Francis Group, London, ISBN 978-1-138-02681-0 ©2015
• Framework of a risk and reliability based offshore wind O&M model– Annual Reliability and Maintainability Symposium (RAMS) - US– Published by IEEE 978-1-5090-0249-8/16/$31.00 ©2016
• Reliability matrix for O&M planning of offshore wind farms– Journal paper, expected submission 2016-Q4
• Maintenance matrix for O&M planning of offshore wind farms– Journal paper, expected submission 2017-Q1
• PhD thesis– Expected submission 2017-Q1
• Risk and reliability based O&M planning of offshore wind farms – A novel case study– Journal paper, expected submission 2017-Q2
17
Risk and Reliability based O&M Planning of Offshore Wind Farms
PhD study: Mihai Florian, Aalborg University. 2013-
Objectives: • Degradation modeling for decision support in maintenance
planning• Development of risk-based inspection planning framework for
wind farms• Cost optimal O&M planning using Bayesian decision tree
18
Initial cracking
Development Failure
• One dimensional fracture mechanics model• Loading based on turbulence mean wind speed and
turbulence intensity
Monte Carlo simulations
Deterioration modelling for WT blades
19
Dynamic Bayesian Networks
• inspection at time t• update of uncertain parameters in damage model
Distribution of crack size before/after inspection
Inspection plan
20
CTV HLVNumber 4 1Wave limit [m] 1.5 2
Wind limit [m/s] - 20Mobilisation time [days] - 30
Mobilisation cost [€] - 250000
Speed [knots] 20 11Day rate [€] 1000 100000
Activity Cost [€] Duration [h]Inspection 1000 6Repair 10000 24
Replacement 400000 80
Vessels
Cost model
Preventive strategy• Time/condition based• Risk/reliability based
Case study – NORCOWE Reference Wind Farm
21
Time/condition based model• Time interval of inspection• Repair threshold
Optimal decision• 2 year interval• 0.4 [m] crack sizeTotal cost [€] Downtime [%]5.25 106 0.37
Optimal decision• 1% failure probability
Total cost [€] Downtime [%]4.55 106 0.27
11 inspection & 2.3 repairs/turbine 7.2 inspection & 1.9 repairs/turbine14% cost reduction
Case study – NORCOWE Reference Wind FarmRisk/reliability based
22
Publications
• Wind turbine blade life-time assessment model for preventive planning of operation and maintenance - Journal of Marine Science and Engineering 2015
• Planning of operation & maintenance using risk and reliability based methods - Energy Procedia Journal 80 ( 2015 ) 357 – 364, 2015
• Risk-based planning of O&M for wind turbines using physics of failure models - European Conference of the Prognostics and Health Management Society 2016
• Case study for impact of D-strings on levelised cost of energy for offshore wind turbine blades – under revision for publication to the International Journal of Offshore and Polar Engineering
• Cost optimal risk-based inspection planning for offshore wind farms – to be submitted to DeepWind conference, Trondheim, 2017
• Risk-based inspection planning for offshore wind farms, a comparison with traditional maintenance – in progress
23
Condition-based maintenance for offshore wind parks
PhD study: Ole-Erik Endrerud, University in Stavanger. 2013-
Objectives / Scientific Contributions:• Methodological contribution to hybrid simulation for modelling
of large scale industrial assets.• Increased understanding of the marine logistics system used for
O&M of offshore wind energy assets. (Work process optimization)
• A framework for simulation modelling of operation and maintenance. (Decision support)
24
Publications
Conference articles• Decision support for operations and maintenance of
offshore wind farmIn Engineering Asset Management - Systems, Professional Practices and Certification. Springer 2015 ISBN 978-3-319-09506-6. s. 1125-1139
• Marine logistics decision support for operation and maintenance of offshore wind parks with a multi method simulation modelIn Proceedings of the 2014 Winter Simulation Conference. IEEE Press 2014 ISBN 978-1-4799-7486-3. s. 1712-1722
• New Vessel Concepts for Operations and Maintenance of Offshore Wind FarmsIn Proceedings of the Twenty-fifth (2015) International Ocean and Polar Engineering Conference Kona, Big Island, Hawaii, USA, June 21-26, 2015
Journal articles• Reference Cases for Verification of Operation and
Maintenance Simulation Models for Offshore Wind FarmsIn Wind Engineering, Volume 39, No. 1, 2015
• In progress: Efficiency of condition based maintenance in operation and maintenance of offshore wind farms
• In progress: Assessment of maintenance strategies using an agent-based and discrete event simulation model
25
SR Bank Innovation Award 2015
Endrerud fikk innovasjonsprisSøk i rogalandsavis.no
• Ole-Erik Endrerud got the SR Bank Innovation Award for the establishment of Shoreline AS and commercialisation of O&M simulation technology for the offshore wind industry.
26
Health Assessment of pitch and yaw systems in offshore wind turbines
PhD study: S.T. Kandukuri (Surya), University in Agder. 2014-
Objectives: • Develop health assessment techniques for electrical pitch & yaw
systems that are suitable for farm-level implementation– Diagnostics for pitch and yaw systems – Prognostics for pitch and yaw systems – Capturing effects of corrosion and fatigue
27
Overview
Wind Turbine Systems
knowledge
• Wind & wave simulations
• FAST analysis – loads on pitch system
• Pitch system components, health assessment requirements
Laboratory Demonstration
• Efficacy of diagnostics algorithms in variable load/speed conditions
• Effects of ’exogenous’ faults on component level diagnostics
Real‐world O&M scenario
• Sate‐of‐the‐art
• CM exists but as ‘end products’ on specific components
• Offshore wind farms in nascent stages
• Pitch & yaw systems fail often
Health Assessment of Pitch & Yaw Systems
• Failure modes and effects
• Most common failure modes
• Diagnostics & prognostics algorithms
• Fault classification
28
Induction motor
4-stage planetary gearbox
Pinion gear
Focus: Wind Turbine Pitch/Yaw systems
Motor GearboxTypically, 3 phase Induction Motors, servo motors
Common faults
• Stator windings (30%)• Bearings (40%)• Imbalance, Shaft Eccentricity• Broken rotor bars (IM)• Loss of rotor magnetization (SM)
Diagnostic Methods: • Motor current signature analysis
(MCSA), Vibration, Temperature…
2-3 stage planetary gearbox
Common faults
• Scuffing of gear tooth • Teeth crack• Carrier plate cracks• Ring gear damage• Bent shafts• Corrosion effects
Diagnostic Methods• Vibration signature, Acoustic
emission, oil debris monitoring…
Motivation:About 45% of all installations have electrical pitch and almost all turbines have electrical yaw systems.
29
Algorithms - Motor current signature analysis (MCSA)
Mechanical faults
Electrical faults
Variations in electric circuit/ airgap magnetic field
Manifest as periodic disturbances in supply current
Accomplishments: • Detailed modeling of induction motor faults based on modified winding
function theory (MWFTh) • Diagnostics based on Fourier spectrum analysis of motor currents
Next Step: • Time – frequency analysis of current signature for variable speeds and
torque analysis
30
Laboratory DemonstrationDemonstrate feasibility of techniques for incipient fault detection in pitch/yaw drives using current and vibration signature, suitable for farm-level implementation.
Subsystem Component Faulty type
Pitch Drive
MotorBroken rotor bars
Stator winding faultWorn bearings
Gearbox
Planet gear faultSun gear fault
Carrier plate cracks
Seeded fault tests
Test motorABB brake motor 1.1kW 4-pole 3phase IM
Test gearbox 2 stage planetary gearbox (1:48)
Load motorABB motor 3kW 8-pole 3phase IM
Load gearbox BPH gearbox (1:27)
31
Publications• S.T. Kandukuri, A. Klausen, H.R. Karimi, K.G. Robbersmyr, A review of diagnostics and prognostics of low-speed machinery
towards wind turbine farm-level health management (2016). Renewable & Sustainable Energy Reviews• S.T. Kandukuri, K.G. Robbersmyr, H.R. Karimi, Towards farm-level health management of offshore wind farms for
maintenance improvements (2015). The International Journal of Advanced Manufacturing Technology• S.T. Kandukuri, V.K. Huynh, H.R. Karimi, K.G. Robbersmyr, Fault diagnostics for electrically operated pitch systems in
offshore wind turbines (2016). Torque 2016, IOP Conf. Series• A. Chougule, S.T. Kandukuri, H.G. Beyer, Assessment of synthetic winds through spectral modelling and validation using
FAST (2016), Torque 2016, IOP Conf. Series
32
Wind-farm Control strategy
S. Christiansen, T. Jensen, T. Knudsen and T. Bak, Aalborg University:
Control problem: minimize WT fatigueCentral assumptions:
1. Fatigue is positively correlated with turbulence; 2. Sum of individual fatigues is a relevant fatigue measure.
Minimisation problem: – Sum (over WTs) of added turbulence as objective function;– Max/min WT power and WF power reference as constraints.
Two approaches have been examined:1. Static optimisation. (24h updates of WT references);2. Dynamic optimisation. (5 sec updates of WT references).
T N Jensen, T Knudsen and T Bak; Fatigue minimising power reference control of a de-rated wind-farm; Torque 2016
33
Comparison of strategies(Relative to a fixed distribution strategy)
5,6 5,6 5,7
7,98,5 8,2
3D 5D 10D
%
Spacing in Rotor diameters
Reduction in sum of added turbulence
Static Dynamic
Simulation: 3 turbines in a row. The sum of the added turbulence is reduced in both cases
34
Comparison of strategies(Relative to a fixed distribution strategy)
Reduction in damage equivalent loads (DEL)
5,8
3,6
1,4
4,6
2,3
-3,5
3D 5D 10D
%
Spacing in Rotor diameters
Tower bending moment DEL
Static Dynamic
Reducing added turbulence translates to reduced DEL on the tower
35
Floating wind turbines
• Estimating wave and wind disturbances• Optimal control and estimation – use pitch system, as if turbine was
bottom fixed• Allows loads to be lowered, with acceptable loads on pitch system.
11
11
22
11
22
36
18
-121
15
1
54
1
20
-105
17
0
51
3
7
-95
13
-4
62
1
-150 -100 -50 0 50 100
ELEC. POWER (STD)
BLADE PITCH RATE (ABS)
TOWER FORE-AFT (STD)
TOWER SIDE-SIDE (STD)
PLATFORM PITCH (STD)
PLATFORM ROLL (STD)
% compared to baseline controller (NREL)
Perpendicular wind-wave forces: statisticalanalysis of relative controller performance (standard deviations
and abs values)
10 sec 5 sec 2 sec
Floating wind turbines - results
Peak wave period:
Baselinebetter
Optimal controlbetter
Less loads on platform and tower
More pitch
But this is more in line with typical
bottom fixed turbine operation
Christiansen, S., Bak, T., & Knudsen, T. (2013). Damping Wind and Wave Loads on a Floating Wind Turbine. Energies, 4097-4116.
37
International cooperation …
• LEX - Torsional Stiffening of Wind Turbine Blades – Mitigating leading edge damages (Danish EUDP)
• RATZ (Root Area and Transition Zone) - Reduction O&M cost of WT blades (Danish EUDP)
• LEANWIND - Logistic Efficiencies And Naval architecture for Wind Installations with Novel Developments (EC – FP7)
• MANTIS – Cyber Physical System based Proactive Collaborative Maintenance (EC - H2020)
• IEA-WIND Task 33 - Reliability data – for O&M optimization of wind turbines
• IEC 61400-1 ed. 4 (CDV 2016) Wind turbines – Design requirements
38
Concluding remarks
• Next steps:– Implementation of risk-based OM strategies– Better understanding of degradation mechanisms – and reliable estimates
of RUL (Remaining Useful Life) – Improved use of data (big-data) for OM planning– More application of
• integrated WT and WF control & OM planning• information from condition monitoring systems & structural health
monitoring (SHM) – accounting for uncertainties / reliability