Experiences of Modelling of Intermittent Renewable Energy Tom Kober - PowerPoint PPT Presentation

Experiences of Modelling of Intermittent Renewable Energy Tom Kober (ECN) JRC workshop on Addressing Flexibility in Energy System Models Petten, 4 Dec 2014 www.ecn.nl www.camecon.com

Rationale • Energy system models - strong tools for long-term energy analysis • Renewable energy (RE) assessment requires modelling innovation • No single model covers all facets of the integration of RE o How can energy system models be improved to better represent intermittent RE?  Linkage with power models  Adopt model structure & data  Sensitivity analysis

Wind production Germany: hourly profile vs. 12 time slices 100.0% 90.0% 80.0% 70.0% Autumn Winter Spring Winter 60.0% Summer 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 -10.0% DE onshore DE offshore times_de onshore times_de offshore

Power systems models  Detailed representation of the electricity system • What can energy system models learn? • How can they be linked? Two examples: – COMPETES (ECN) – E2M2s (IER)

COMPETES electricity market model (ECN) • Optimization-based model (e.g. LP/MIP) • Formulations for different goals: 1. OPF Static Economic dispatch model with perfect competition (LP) 2. OPF Static Unit Commitment model with perfect competition (MIP) 3. Dynamic model (LP): • Two-period under perfect competition • Investments in the first period (generation + transmission) • Dispatch in the second period

Modelling intermittent RE in COMPETES • Deterministic approach using hourly power factors or capacity factors per country or node • Capacity factors based on historic data: SODA Database, TradeWind Database, Websites European TSO’s • Future wind and solar profiles are similar to historic data • Future availability factors are scaled-up to reflect technological advancements (EWEA Pure Power report) • Curtailment allowed

COMPETES Unit Commitment Model Objective: Minimize Total variable generation cost+ Min-Load costs+ Startup costs + load-shedding costs Integer decisions subject to − Power balance constraints: These constraints ensure demand and supply is balanced at each node at any time. − Generation capacity constraints: These constraints limit the maximum available capacity of a generating unit. − Cross-border transmission constraints: These limit the power flows between the countries for given NTC values. − Ramping up and Down constraints : These limit the maximum increase/decrease in generation of a unit between two consecutive hours − Minimum Load Constraints: These set the min generation level of a unit when it is committed (Relaxed for neighboring countries with aggregated capacities) − Minimum up and down times (Only for NL)

Minimum load and corresponding costs for each unit in COMPETES Part-load LHV efficiency curves 100% 95% % of max efficiency 90% LHV efficiency as 85% Nuclear 80% PC 75% PC-CCS 70% IGCC 65% NGCC 60% NGCC-CCS 55% OCGT - Min Load Costs are incurred at Qmin 50% 0% 20% 40% 60% 80% 100% - Relaxation on minimum load for neighboring Production as % of max production countries

COMPETES’ flexibility assumptions Start-up cost a Minimum load Ramp rate Decade of Technology (% of max (% of max ( € /MWinstalled Min up time Min down time commissioning capacity) capacity/hour) per start) Nuclear <2010 50 20 46 ±14 8 4 2010 50 20 46 ±14 8 4 >2010 50 20 46 ±14 8 4 Lignite and PC <2010 40 40 46 ±14 8 4 2010 35 50 46 ±14 8 4 >2010 30 50 46 ±14 8 4 IGCC <2010 45 30 46 ±14 8 4 2010 40 40 46 ±14 8 4 >2010 35 40 46 ±14 8 4 NGCC <2010 40 50 39 ±20 1 3 2010 30 60 39 ±20 1 3 >2010 30 80 39 ±20 1 3 OCGT <2010 10 100 16 ±8 1 1 2010 10 100 16 ±8 1 1 >2010 10 100 16 ±8 1 1 CHP <2010 10 90 16 ±8 1 1 2010 10 90 16 ±8 1 1 >2010 10 90 16 ±8 1 1 Sources [1-9] [1-8, 10] [11] [11] [11] Sources: [1] (Jeschke et al., 2012); [2] (Dijkema et al., 2009); [3] (OECD/IEA, 2012b); [4] (IEAGHG, 2012a); [5] (Klobasa et al., 2009); [6] (Balling, 2010); [7] (Hundt et al., 2010); [8] (Isles, 2012); [9] (Stevens et al., 2011); [10] (NETL, 2012b); [11] (Lew et al., 2012). a) Warm start-up costs are assumed for all technologies but OCGT. For OCGT, a cold start is assumed.

Example: generation flexibility in the Netherlands 3000,0 Supply of domestic flexibility per technology (GWh) Demand to ramp up(GWh) 2000,0 1000,0 Decentralized CHP Res-e Nuclear Gas Other 0,0 Gas GT Gas CHP Gas CCGT -1000,0 Demand to ramp down Coal -2000,0 (GWh) -3000,0 2012 2017 2023 2012 2017 2023 Source: ECN-E--14-039 (2014)

E2M2s (IER, Uni Stuttgart) • Electricity market model for Germany • All generation units • Inter-temporal optimisation • 144 time slices per year • Stochastic electricity production for wind and solar technology • Flexibility parameters for power plants – Ramp-up/down time & costs – Minimum load – Minimum down time

Link energy system model and power market model Electricity consumption CHP electricity generation Fuel prices European TIMES Power market model (PanEU) model (E2M2s) Power plant costs RE-generation (policy) Long-term Long-term 12 timeslices 144 timeslices LP Stochastics Capacity credit for wind and solar System reserve capacity Generation from flexible units

Example: wind capacity credit Germany (power market model) Capacity credit [%] ~150 TWh & ~60 GW in 2030 Source: IER Energieprognose 2009

Adopting the energy system structure in TIMES • Energy system and technology parameters of intermittent RE depend on the technology’s market diffusion • Unless RE deployment is exogenous to the model, introduce different model processes to control parameters Parameter set z Parameter set y Parameter set x

Improved data for TIMES energy system model • Capacity credit  NCAP_PKCNT • System reserve capacity  COM_PKRSV • Generation from flexible units  User constraints helps to model system flexibility that cannot be captured with low time resolution Negative balancing energy into 100% storages or flexible demand Positive reserve energy from 90% storages and flexible power plants 80% 70% 60% 50% 40% 30% 20% 10% 0% 4200 4300 4400 4500 4600 4700

User constraints for flexible generation • Determine energy production from flexible units as share ( p,n ) of production from intermittent RE (e.g. based on power model) • per time slice • per level of RE deployment (different technology processes) • User constraint for positive energy: ElcGen (storages, GT, IC) ≥ p % ElcGen(wind, pv) • User constraint for negative energy: ElcCons(storages, flex demand) ≥ n % ElcGen(wind, pv)

Storages in TIMES ● Pump storage ● Compressed air – Natural gas-CAES – Adiabate CAES ● Stationary battery systems – Natrium-Sulfid – Redox Flow ● Elektro mobility – E-vehicles loading from the grid only – E-vehicles to grid (V2G) ● Hydrogen storage ● Power-to-gas + storage

CAES storages • Base: natural gas caverns • Existing storages: 36 • Cavern storage projects: 38 • Major storage regions: Germany, UK, Poland, France, Portugal, Spain • Max CAES capacity estimated: 19 GW (of which 6 GW in Germany) source: Gillhaus 2007

Electricity infrastructure investments • Implemented via grid processes and Cost for transmission system extension 450 400 user constraints [Euro/kWnew capacity] 350 300 • Grid processes = solar and wind sector 250 Wind fuel processes (TIMES) 200 Solar PV 150 • 6 stages with costs up to 400 Euro/kW 100 50 refer to new installed capacity 0 0 100 200 300 400 Installed capacity [GW]  Good proxy but no trade-off between infrastructure investments and flexible generation / demand

The ‘extreme’ timeslice • Problem: hours of negative residual load level out when annual wind/solar power generation profiles are reduced to 12 time slices (no negative electricity prices in the model)  Introduce daynite timeslice per season that characterizes this condition (equivalent to peak time slice) and/or change distribution of annual profile to timeslices • Analysis of wind/solar peaks and the load during these hours .300 Annual availability .250 .200 .150 .100 .050 .000 RD RN RP SD SN SP FD FN FP WD WN WP

Conclusion • Model coupling is valuable • TIMES offers model framework to introduce flexibility mechanisms • Model link enables improved parameters for the energy system model (data and model structure to be adopted) • Challenge: incorporate trade-off between infrastructure investments and system flexibility

Thank you! Tom Kober Policy Studies | Global Sustainability T: +31 88 515 4105 | F: +31 224 56 83 38 Radarweg 60, 1043 NT Amsterdam, The Netherlands kober@ecn.nl

• Supplementary material