Modelling of wind power fluctuations and forecast errors Prof. Poul Sørensen Wind Power Integration and Control Wind Energy Systems
Modelling of wind power fluctuations Simulation tool: CorWind Based on WRF mesoscale data Added stochastic variance Power curve for wind-to-power 24 22 20 Wind speed (m/s) 18 16 14 Turbine (A1) Wind Farm Weather model 12 10 00:00 02:00 04:00 06:00 08:00 10:00 12:00 Time 2 DTU Wind Energy, Technical University of Denmark
Consortium and budget Ireland UCD United Kingdom (2) ALSTOM GRID UNIVERSITY OF STRATHCLYDE France (2) RTE EDF 10 European Member States 1 Associated Country Norway SINTEF Denmark (3) DONG ENERGY ENERGINET DTU ENERGY Portugal INESC-PORTO The Netherlands TENNET Spain (5) RED ELECTRICA DE ESPAÑA IBERDROLA ITT COMILLAS GAMESA ABB S.A. DFFV conference, Herning Total budget: 56.8 M EU contribution: 31.8 M Germany (3) FRAUNHOFER IWES 50 HzT SIEMENS Wind Power Belgium (6) ELIA SYSTEM OPERATOR EWEA CORESO UNIVERSITY LIEGE UNIVERSITY LEUVEN UNIVERSITE LIBRE BRUXELLES Italy RSE 3
Demo 4 - The challenge Synchronous Area 2020 2030 MW MW Continental 21,421 57,685 Nordic 4,924 14,669 GB 13,711 33,601 Ireland 1,419 3,219 0 15 E 60 N DFFV conference, Herning 2030 map 4
The demonstration 91 Lead by Energinet.dk Horns Rev 2 wind farm owned by DONG Energy 91 x 2.3 MW Siemens wind turbines built with conventional storm control Siemens developed and installed High Wind Ride Through - (HWRT) DTU simulated and analysed impact on forecast errors SINTEF analysed coordination with HVDC and Norwegian hydro 7 1 84 DFFV conference, Herning 5
The performance of the two storm controls Conventional High Wind Shut Down (HVSD) wind turbine control Simplified representation of Siemens High Wind Ride Through - (HWRT) DFFV conference, Herning 6
Wind farm power generation during storm passage DFFV conference, Herning 7
Wind turbine forecast error February 7-8, 2011 January 30, 2013 Power (p.u.) 1.0 Meas 0.8 0.6 0.4 0.2 0.0 18:0019:0020:0021:0022:0023:0000:0001:0002:00 Power (p.u.) 1.0 0.8 0.6 0.4 0.2 Meas 0.0 18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 Error (p.u.) 1 0.5 0-0.5 comb Error real -1 18:0019:0020:0021:0022:0023:0000:0001:0002:00 Time (Feb 07, 2011) Error (p.u.) 1 0.5 0-0.5 comb Error real -1 18:00 20:00 22:00 00:00 02:00 04:00 06:00 08:00 Time (Jan 30, 2013) DFFV conference, Herning 8
Recorded storm events and max forecast errors Event nr Date Controller 1 11-nov-10 HWSD 2 12-nov-10 HWSD 3 07-feb-11 HWSD 4 24-sep-12 HWRT 5 14-dec-12 HWRT Legend: HWSD - High Wind Shut Down; HWRT - High Wind Ride Through Event Max forecast error [p.u.] 11-Nov-10 0.80 12-Nov-10 0.80 07-Feb-11 0.72 6 30-jan-13 HWRT 24-Sep-12 0.26 14-Dec-12 0.18 30-Jan-13 0.35 Average forecast error [p.u.] 0.77 0.26 Difference [p.u.] 0.51 DFFV conference, Herning 9
Upscaling of offshore wind power from Horns Rev 2 North Europe 2020/2030 0 15 E 91 84 60 N 7 1 Horns Rev 2 (0.21 GW) North Europe: 41 GW in 2020 109 GW in 2030 10 DTU Wind Energy, Technical University of Denmark
Upscaling approach Simulation tool: CorWind Addition of high wind shutdown / startup rules Model to aggregate at wind farm level Simulation of single wind farm Hysteresis model implemented in CorWind 8 full met years of simulation Assuming persistence forecasts (conservative assumption!) 11 DTU Wind Energy, Technical University of Denmark
Large scale challenge: Adequacy of primary reserves There must be sufficient primary reserves in the power system synchronous area to replace lost production corresponding to dimensioning fault This brings power system from normal state to alert state Frequency restoration (secondary / tertiary) reserves will return system to normal state in 15 minutes Larger faults (loss of generation) may bring system into disturbed (or emergency) state Therefore, maximum 15 minute wind power forecast errors are essential to esure adequacy of primary reserves Synchronous Area Dimensioning faultt MW Continental 3,000 Nordic 1,200 GB 1,800 Ireland 500 Nordic grid code 2007 12 DTU Wind Energy, Technical University of Denmark
Upscaling results and conclusion Result for 2020 indicates that there is sufficient primary reserves with current dimensioning fault to cover offshore wind power variability in the four main European synchronous areas Result for 2030 indicates that there is not sufficient primary reserves with current dimensioning fault to cover offshore wind power variability in Continental and GB synchronous areas Current requirements for primary reserves should be revised by 2030 to maintain secure operation Further studies will be done using more realistic forecast simulations than persistence Synchronous Area HWSD HWEP Dimensioning faultt MW MW MW Continental 1,661 1,548 3,000 Nordic 480 483 1,200 GB 1,212 1,222 1,800 Ireland 224 224 500 2020 Synchronous Area HWSD HWEP Dimensioning faultt MW MW MW Continental 4,729 3,933 3,000 Nordic 1096 1082 1,200 GB 4,418 4,440 1,800 Ireland 439 438 500 2030 13 DTU Wind Energy, Technical University of Denmark
Modelling of wind power forecast errors Simulation of consistent time series of wind power fluctuations and forecasts now also in CorWind 14 DTU Wind Energy, Technical University of Denmark
Simulation of balancing (Simba) Import Simulation of Balancing (Simba) Export Simba idea Simulation of intra hour balancing as supplement to day ahead Uses inputs from day-ahead market model Main imbalance included today is from wind Applications of Simba Planning of investment Assessment of new market designs (e.g. towards real time) Assessment of cost / value of reserves Assessment of needs for reserve capacities Economic optimisation of system services Assessment of flexible demand support to system balancing 15 DTU Wind Energy, Technical University of Denmark
Automatic Generation Control in a power system with high wind power penetration Danish case study CorWind Pwind(avail) Pwind (HA) Pwind (DA) SimBa Pwind(avail) Pplan (5 min) ΔPset Dynamic Power system model WILMAR Pplan (1 Hour) AGC Models data exchange Model overview fnominal + - factual PDCHP + PCHP + + Pwind B PGEN f/r - ACE - P + - PI controller -90 MW +90 MW Pset PCHP pfchp PDCHP pfdchp Pexchange PLOAD + + AGC model Result: simulated AGC performance 16 DTU Wind Energy, Technical University of Denmark
Wind Power integration into the Automatic Generation Control of power systems SimBa AGC Fmeasure Pmeasure Pavailable + ΔP_WF + Frequency droop Pref Pref (freq) Wind Power Plant Controller Pref_WT Pmeasure Aggregated Wind Turbine model Active power Controller Aggregated WPP model ip_cmd Static Generator Pset Curtailing Power P < 0 dpavailable P < curtailing yes no no P > dp_avail yes yes no P_WF = Pset P_CHP = 0 P_WF = -1* curtailing P_CHP = curtailing - Pset P_WF = dp_avail P_CHP = 0 P_WF = dp_avail P_CHP = Pset - Available Secondary (AGC) dispatch with wind 17 DTU Wind Energy, Technical University of Denmark
Modelling of wind power fluctuations and forecast errors. Conclusions regarding applications Users of Modelling of wind power fluctuations and forecast errors Transmission system operators Energy planning authorities Power producers with wind in portfolio Consultants R&D Application Tasks of Modelling of wind power fluctuations and forecast errors Planning of transmissions network development (e.g. TYNDP) Planning of power producer investments Assessment of wind power plant reliabillity Power system frequency stability assessment with massive wind Assessment of new market designs (e.g. towards real time) Assessment of cost / value of reserves Assessment of needs for reserve capacities Feasibility of ancillary service provision (power) from wind power Economic optimisation of system services Assessment of flexible demand support to integrate wind 18 DTU Wind Energy, Technical University of Denmark