Operational wind & solar power forecasting The Perfect Wind Power Prediction Dr. Hans-Peter (Igor) Waldl, Overspeed, Germany Overspeed GmbH & Co. KG, Germany Large-scale Grid Integration of Renewables in India Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 1
Overspeed: 25plus Years of Experience Core: Consulting for investors, banks, project developers R&D as background Dr. Hans-Peter Waldl System development Main areas: Thomas Pahlke Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 2
Wind Energy energy Consulting Assessments System and Software Development Wind and Solar Power Predictions overspeed.de Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 3
Anemos predictions: What is Anemos? Leading edge research and development Prediction models and modules Wind and Solar Power Prediction System Commercial wind & solar predictions Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 4
Anemos Wind/Solar Power Predictions: Partners Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 5 5
Anemos predictions: What is Anemos? Leading edge research and development Prediction models and modules AU.SA Wind and Solar Power Prediction System Commercial wind & solar predictions World-wide around 100 GW Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 6
Wind power predictions: Principles The Perfect Prediction: Models The Perfect Prediction: Power data Summary Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 7
Wind power predictions: Principles NOW Production [MW] 6 5 4 3 The Perfect Prediction? 2 1 0 +1h +6h +12h +18h +24h +30h +36h. +48h DAY +1 DAY +2 Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 8
The Perfect Prediction: Ingredients Historical Power Prediction Model(s) NWP Wind Prediction System Online Power Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 9
The Perfect Ingredients?: Chili Peppers India The Perfect Curry Meal Me Tamil Nadu Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 10
Perfect for? -- Applications Grid and system operations Power plant dispatching Reserve planning Congestion management Extreme event handling Other applications Energy trading Load predictions O&M optimisation Rules and regulations! Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 11
The Perfect Prediction: Models Historical Power Prediction Model(s) NWP Wind Prediction System Online Power Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 12
The Perfect Prediction: Ingredients Prediction Model(s) NWP Wind Predicted Power Online Power Historical Power Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 13
Modelling approaches Commitment of physical or statistical models strongly depends on purpose From experience: Combination of physical and statistical modelling leads to best results Use all knowledge you have about wind and wind power Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 14
A note on Machine Learning Machine learning methods based on power production Could be a great asset for some applications Needs big data volumes Issue: Not available in most applications Risk: Modelling of extremes may be completely wrong Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 15
Prediction Principles: Time domains Model types Statistical models NWP + physical/statistical modelling time Input Data 5 min 15 min 3 h 6 h intraday day ahead Online SCADA NWP domain Historical SCADA 10 d Model Tuning Auto-adaptive model tuning Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 16
Auto-adaptive model improvement Model error improvement over 6 month, different NWPs Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 17
The Perfect Prediction: Power data Historical Power Prediction Model(s) NWP Wind Prediction System Online Power Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 18
Power data should be as ideal as possible, but Turbine failure Turbine availability is limited Missing SCADA data from wind farms Missing data connection to wind farms Substation availability Grid availability Grid congestions System services Issues with access to wind farm SCADA Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 19
Data with turbine availability and curtailment Power [%] 50 Unconstrained wind farm production 45 40 35 Limited turbine availability 30 25 20 Grid curtailment 15 4:30 5:30 6:30 7:30 8:30 9:30 10:30 11:30 12:30 13:30 14:30 15:30 16:30 17:30 18:30 19:30 20:30 21:30 Time [hours] Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 20
Upscaling for curtailed wind farms Upscaling for curtailed wind farms Based on local wind speed measurements Automatic detection of curtailment Plus excellent data Quality Management Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 21
Data channels: Power, and? Current power production per farm if possible. Current turbine availability Set-points for curtailment, Data quality! Wind speed measurements for upscaling Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 22
Prediction Principles: Time domains Model types Statistical models NWP + physical/statistical modelling time Input Data 5 min 15 min 3 h 6 h intraday day ahead Online SCADA NWP domain Historical SCADA 10 d Model Tuning Auto-adaptive model tuning Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 23
Summary Historical Power Prediction Model(s) NWP Wind Prediction System Online Power Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 24
Typical prediction system: Anemos Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 25
Summary 1: Resilience Prediction systems must be robust and flexible Communication and IT infrastructure is important! Models must be robust and flexible Data quality management is essential Well designed regulations and rules for wind/solar farms help the prediction business! Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 26
Perfect predictions? In order to get an optimal prediction: Use more than one NWP; consider local models Get the best measurement data you can get Regulations: Make good power data quality a must Excellent predictions need good models and excellent data sources Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 27
Perfect predictions? Activities in India REMC advice Wind and solar power predictions Training for NIWE team (giz): Indigenous solar power prediction system for India igor@overspeed.de Workshop Large-scale Grid Integration of Renewables in India, Delhi Sep. 2017 28