Dynamic reserve requirement in real time operation Daniel Huertas Hernando SINTEF Energy Research SINTEF Energy Research 1
Wind power forecast uncertainty Uncertainty is higher for 1 wind farm than for many wind farms over a large area 1 wind farm, Germany 2 hours ahead Actual production Day-ahead Relative error 40 wind farms, Germany Region size (km) 4 hours ahead Source: dena Grid study I Uncertainty increases with time Large prediction errors day-ahead Intraday-trading is an important instrument for reducing the need for regulating power SINTEF Energy Research 2
Influence of forecast uncertainty Traditionally, point prediction has been the trend: Single value for each look-ahead time, Expectation or most-likely outcome only Focus on assimilating more and more observations in the models or refining the resolution of physical models to decrease of the level of error SINTEF Energy Research 3
Influence of forecast uncertainty There will always be an inherent and irreducible uncertainty in every prediction Instead, uncertainty is expressed in the form of probabilistic forecasts with confidence intervals provided along with the traditional point predictions These probabilistic forecast methods are known as: Optimal dynamic quantification of reserve requirements Optimal operation of combined systems including wind SINTEF Energy Research 4
Integration of renewable generation. The case of Spain. Abril 2007 EES-UETP - Madrid, November 2010 Miguel de la Torre Javier Paradinas 1
Generation Mix in Spain Today (I) Installed capacity (30/09/2010) Technology SEP MW % IBG % Solar CSP 0,5% Solar PV 3,6% Mini-hydro 2,1% Wind 20,6% Biomass 0,7% Cogeneration 6,3% Waste Treat. 1,3% Hydro 17,6% Coal 11,4% Nuclear 7,9% Hydro-power 16 657 17.6 8 847 53.1 Nuclear 7 455 7.9 3 409 45.7 Coal 10 789 11.4 1 253 11.6 Fuel-Gas 1 849 2.0 157 8.5 Combined cycles 24 633 26.1 5 893 23.9 Total (Conventional) 61 383 65.0 19 559 31.9 Wind power generation 19 442 20.6 5 199 26.7 Solar power 3 853 4.1 50 1.3 Combined cycle 26,1% Fuel-Gas 2,0% Biomass 684 0.7 2 0.3 Small hydro 1 965 2.1 131 6.7 Cogeneration 5 946 6.3 410 6.9 Waste treatment 1 204 1.3 0 0.0 Total (RES and CHP) 33 114 35.0 5 792 17.5 Total 94 497 MW 25 351 MW 4
Generation Mix in Spain Today (II) Demand supply 2010 S1 132 942 GWh = 139 039 Net Generation - 2 793 Hydro-pump storage - 3 304 International exchanges Solar PV 2.2% Small Hydro 2.8% Technology GWh % IB-G % Hydro-power 28 702 8.9 13653 47.6 Wind 15.9% CHP and other RES 12.4% Combined cycle 20.5% Hydro-power 20.6% Coal 4.9% Nuclear 20.0% Nuclear 27 791 19.1 12651 45.5 Coal 6 815 12.1 262 3.8 Fuel-Gas 800 0.7 39 4.9 Combined cycles 28 556 29.0 4309 15.1 Total (Conventional) 92 664 69.8 30 914 33.4 Wind power generation 22 089 13.7 4942 22.4 Solar PV 3 091 2.0 22 0.7 Small hydro 3 960 2.0 158 4.0 Fuel-Gas 0.6% CHP and other RES 17 235 12.4 991 5.7 Total (RES and CHP) 46 375 30.2 6 113 13.2 Total 139 039 GWh 37027 GWh 5
Secondary and tertiary control (I) Scheduled program Real demand TSO demand forecast The scheduled program must be modified in order to meet the expected demand: Generation-demand unbalance market Tertiary regulation 9
Wind production variability 16,000 14,000 12,000 Max. Coverage 08/11/2009 3.29 h W P Record 12 916 MW 24/02/2010 11.21 h Wind production record 14 962 MW 09/11/2010 14.46 h 10,000 8,000 6,000 4,000 2,000 0 01/11/2009 01/12/2009 01/01/2010 01/02/2010 01/03/2010 01/04/2010 01/05/2010 01/06/2010 01/07/2010 01/08/2010 01/09/2010 01/10/2010 01/11/2010 Int. Int -4.67% Minimum coverage 03/09/2010 12.33 h Wind; 1.26% Nuclear; 16.89% Hydro-Power; 13.10% Combined cycles; 35.09% Cog + Other RE; 20.31% Coal; 16.62% Maximum coverage 09/11/2010 3.35 h Hydro-Power; 5.42% Wind; 53.09% Int. Int -4.67% Manageable generation must compensate this variability to maintain the equilibrium between generation and demand. Nuclear; 31.84% Cog + Other RE; 17.65% Pumping; - 12.45% Coal; 2.99% Combined cycles; 6.10% 13
Power balance feasibility A favorable landscape to renewables Reserve margins Keys of the technical success 22 GW CCGT (2001-2009) 75% of coal fit scrubbers SO 2 Peak load: 44 GW Installed Wind: 19 GW <1 GW Grid Reinforcements in last decade Dispersed wind Flexibility, storage ~52 CCGTs Units in Spain 17 GW hydro (~5 GW of pumping hydro under operation or construction) X Interconnections (France) Export capacity < 1 GW Source: REE thanks to investments committed before the 20/20/20 political decision 14
Wind forecast uncertainty in operation (I) MW W 12000 10000 8000 6000 4000 Real wind Forecast production uncertainty vs Forecast Uncertainties on how much manageable generation is required during the peak load. Higher levels of reserves needed. 2000 0-2000 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 25/10/2010 Real production Forecast Error More thermal generation needs to be connected to provide them which may take up space for RES during off-peak hours. 35
P-85% : 85% probability that the wind forecast is below the actual wind production for each hour. This constitutes the lower-bound of the wind forecast with a 15 % confidence margin. P-15%: 15% probability that the wind forecast is below the actual wind production at the given hour, i.e. 85% probability that the wind forecast is above the actual wind production at the given hour This constitutes the higher-bound of the wind forecast with a 15 % confidence margin P-50%: 50% probability that the wind forecast is below the actual wind production at the given hour. This is the forecast with the lowest mean square error. Point prediction forecast SINTEF Energy Research 5
Wind forecast uncertainty in operation (III) Critical time horizons are 24 or 32 hours in advance for D-1 reserve evaluation and 5 hours for real-time evaluation. Positive evolution in forecast error in the last years has resulted in fewer need for reserves to cover wind forecast errors, specially in D-1. 37
Control of wind generation (I) The SO may issue wind generation curtailments to wind producers to restore downward reserves: System runs out of downward reserves. Thermal groups needed to supply peak demand. Conventional generation for LFC reserves. Last resort to recover downward reserves: Curtailment of wind generation. Done within 15 minutes thanks to the control center structure. 12000 11000 10000 9000 8000 7000 6000 5000 4000 3000 MW Wind Production Wind Program Wind Forecast 24
Control of wind generation (II) The SO may issue wind generation curtailments to wind producers to restore downward reserves: System runs out or downward reserves. Thermal Wind groups curtailments needed due to to supply power peak balance demand. feasibility (GWh) 250 Conventional generation for LFC reserves. Last resort to recover downward reserves: Curtailment of wind generation. 200 175.26 Done within 15 minutes thanks to the control center structure. 150 100 50 0 0 3.7 12000 11000 10000 9000 8000 7000 6000 5000 4000 3000 MW March 1 st 2010 2007 2008 2009 2010 (30/09/2010) 0.019% of total 0.58% of total production production 6.9 Wind Production Wind Program Wind Forecast 25
2010 Experience Main Figures The System has been able to integrate a high share of wind with minimum spillages Maximum ratio Wind/Demand Wind curtailments Hydro reservoirs (End Q1) 50-55% 0,15 TWh 85% but thermal generators have undergone a remarkable cost Thermal capacity load-factor Switched off units (lack of downwards reserve) 23% 88 Source: REE, Iberdrola 26