WIND FARMS BASED MULTI-AGENT SYSTEM IRISE/CESI France Plan Context - PowerPoint PPT Presentation

CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey MODELING OF MAINTENANCE STRATEGY OF OFFSHORE WIND FARMS BASED MULTI-AGENT SYSTEM IRISE/CESI – France

Plan • Context • Renewable energy • Importance of wind energy ( especially offshore wind energy) • Energy cost • Maintenance cost and reduction • Failure rate of OWF • Most important part • Failure cause and failure mode • Relation between cost and down time in offshore wind farms • Multi-agent model of maintenance • Maintenance policies • Cost model • Simulator • Simulation and results • Simulations • Results • Conclusion and perspectives CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey 2

Context: Renewable energy • The renewable energy are the best alternative to replace the conventional energy ( Oil, coal, nuclear, etc ) • Solar and wind energies are the most reputed renewable energies • Offshore wind energy is a very interesting way to produce energy • Political strategies • Technological advances 3 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Development of OWF Energy (GW) 4 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Development of OWF Annual onshore and offshore installation EWEA ( EUROPEAN WIND ENERGY ASSOCIATION ) 5 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Development of OWF Onshore historical growth 1994 – 2004 compared to EWEA'S offshore projection 2010 – 2020 6 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Offshore Wind farms (OWF) • The OWF is expected to be the major source of energy • European countries are leader (117GW) • Characteristics : • higher wind speeds • smoother, less turbulent airflows; • larger amounts of open space; • the ability to build larger, more cost-effective Middelgrunden wind farm outside turbines (6 to 10 MW) of Copenhagen, Denmark. Image obtained with thanks from Kim • Cost of installation of offshore turbines is more Hansen on Wikipedia important than onshore • Cost of maintenance is very important in OWF 7 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Objective : Maintenance Cost reduction • Simulation of the behavior of all parts of an offshore wind farm during a to accomplish a maintenance task. • Evaluation of several maintenance policies • Maintenance optimisation 8 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Planning of maintenance tasks • Use of e-maintenanace (tele- maintenance, augmented/virtual reality, … ) • Management of transport of spar parts and personnel of maintenance (beats, helicopters, etc) • Management canes dimension and position • Storage centers management 9 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Multi-agents model • Each turbine is considered as an agent *..1 Impact Turbines Weather 5 agents type of maintenance: • • Preventive maintenance Depends > • Corrective Maintenance Supervise > • Condition Based Maintenance • Video-Assisted Maintenance • Proactive Maintenance Maintenance Select & Order > PM Monitoring VAM 1 agent representing the weather • PrM CM CBM 1 monitoring agent • S > Resources agents • *..1 Use Human Material • Human resources Resources Resources • Material resources 10 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Turbine agents Weather • Each Turbine is characterized by: Monitoring • Power rate (P r ), V cin , V rate and V cout Turbine • State indicator: On/Off, in_maint • Performance: EHF, MAR, inspection delay Energy • Component: Elec_sys, Yew_system, Gearbox, Hydraulic, Blade Maintenance • Production: energy, Peff = P * energy and energy depends of ehf • Behavior • Produce • Degrade ( time) • Interactions • Weather degrade the turbine and control the level of production • Maintenance repair the turbine and increase the E quipment H ealth F actor • Monitoring inspect the turbine 11 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Offshore Wind farms (OWF) “Example” • DOWEC wind farm • 80 turbines, 6MW each => 480MW • North sea at the location “NL7”, 50 Km offshore • Equipped with 50MT mobile crane • In each nacelle there is 1MT crane • A supplier with an Offshore Access System is used to transport personal and small components DOWEC 2003 12 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Failure mode and failure cause Resonances within Production Poor component Icing problem Frequent Improper resistor-capacitor defects in extreme installation (60%) quality and stoppage and (RC) circuits weather system abuse starting Turbulent High/Low Poor electrical wind Poor High vibration temperature Particle installation system level during contaminations design Corrosion overload Technical Out-of-control High loaded Lightning defects Vibration rotation operation conditions Electrical Blade Yaw Gearbox Hydraulic Control System Failures • Damages • Wearing, ● Generator windings, • Cracking of yaw drive shafts, Leakages • Cracks • Backlash, ● Short-circuit • Fracture of gear teeth, • Breakups • Tooth breakage ● Over voltage of • Pitting of the yaw bearing race • Bends electronics components • Failure of the bearing mounting ● Transformers Weather bolts ● Wiring damages Human Technical 13 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey EHF Degradation model 10 12 0 2 4 6 8 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121 127 133 139 145 151 157 163 169 Time (day) 175 181 187 193 199 205 211 217 223 229 235 241 247 253 259 265 271 277 283 289 295 301 307 313 319 325 331 337 343 349 355 361 Turbine 57 Turbine 33 14

Weather agent • It is characterized by : • Vs (wind speed) probabilistic variation regarding the season • Hs (high of waves) probabilistic variation regarding the season and the Vs • Lightning : appears randomly regarding the season • Visibility: appears randomly regarding the season Turbine • W1: Vs < 8 m/s and Hs < 1.5 m • W2: Vs < 12 m/s and Hs < 2 m Monitoring • Behavior Weather • Update (time) • Degrade M_ resources • Interactions • Weather degrade the turbine and control the level of production • Weather defines the window of intervention of maintenance team • Monitoring inspect the weather windows 15 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Resources agents • Material resources: • Characteristics • Number of big boats Weather • Number of small boats • Number of Cranes • Spares Behaviors • maintenance Monitoring • Degradation Resource • Update (maintenance) • Human resources: • Characteristics • Experience • Engineer • Technicians • Behavior • Get experience • Update 16 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Maintenance agents • Maintenance: • Characteristics Weather • It is executed at fixed dates • Needed engineers • Needed technicians Monitoring Resources • Needed cranes • Needed boats Maintenance • Needed weather window: • Weather window > W2 → No maintenance action • W1 < Weather window ≤ W2 → AVM telemaintenance • Weather window ≤ W1 → PM, CM, PrM, CBM • Time of execution • Behaviors CM CBM • Get resources SM • Repair • Release resources • Interactions • Monitoring maintenance order 17 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Monitoring agent Weather • Characteristics • Make order in the agents behaviors • Criterion : age, risk level, emergency Turbines Monitoring • Need actions • Concerned turbine Maintenance • Used maintenance Behaviors • Behaviors Resources • Monitor • Select • Order • Interactions • The monitoring agent inspects the characteristics of the other agents and select the turbine to maintain and the kind of maintenance to use 18 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Cost model Where: • NT : the number of turbine in the farm • N sm , N cbm and N cm : the number on systemic, condition-based and corrective maintenance respectively during the considered period (T unite of time) • X sm , X cbm and X cm are the decision variable where it is equal to • is an indicator of the state of the turbine • : measures the degradation level of the turbine tr at time i . • It is computed as follow: 19 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey

Simulation • Development on NetLogo • Possibility of defining: • The number of turbines in the farm • The size of maintenance teams (engineers and technician) • The number of material resources • Observations: • The generated energy • Weather variation • Turbines stats • Green : normal mode • Orange : degraded mode • Red : failed mode • Black : in maintenance • Maintenance agents • Simulation step = 1 day. 20 CIE44 & IMSS’14 Proceedings, 14-16 October 2014, Istanbul / Turkey