1. Evaluating flexibility and adequacy in future EU power systems: model coupling and long-term forecasting
Sylvain Quoilin, Wouter Nijs and Andreas Zucker
Joint Research Center (JRC)

2. Flexibility assessment in long-term planning models
Main challenge: integrate flexibility contraints into low time-resolution models
Should be expressed linear constraints between model variables (no integers!)
Information about the chronology of timesteps is lost
Focus on high shares of renewables
Dispa-SET Model ?
JRC EU-TIMES
JRC models:

3. JRC-EU-TIMES in a nutshell
Model horizon: (2075)
Technology rich (300+) bottom-up energy system optimisation (partial equilibrium) model based on the TIMES model generator of the IEA
Designed for analysing the role of energy technologies and their innovation for meeting Europe's energy and climate related policy objectives
Electricity multi-grid model (high, medium and low voltage grid), tracking demand- supply via 12 time slices (4 seasons, 3 diurnal periods), and gas across 4 seasons
70 exogenous demands for energy services

4. The JRC-EU-TIMES model optimises the energy system over long time periods
Aligned to latest EU Energy Reference Scenarios
Energy services demands (pkm, tkm, PJ, Mt)
Supply and demand technologies
Objective
Minimise total energy system costs
Constraints
Demand and supply balances: Transport, industry, buildings, agriculture - primary energy (RES, fossils), refineries and electricity
Impacts of high variable RES-e
Flexible use possible excess RES-e: curtailment, Power2gas and storage
Reduced operation dispatch. power
Material and energy flows
Resources and costs
Technology costs (O&M, investments)
Technologies
ETRI
Policies (GHG and energy target, subs.)
Emissions (t CO2) and prices

5. JRC EU-TIMES - illustrative results

6. Flexibility constraints in JRC-EU-TIMES
Water Heating
Space Heating
DSM
Space Cooling
Excess (20 GWh)
Batteries
PtX
Diesel/Kero
Demand (100 GWh)
Methanol
Curtailment
Methane
Deal with the surplus
Power (GW)
VRE (60 GWh)
Other (40 GWh)

7. Dispa-SET in a nutshell
Unit commitment and dispatch model of the European power system
Optimises short-term scheduling of power stations in large-scale power systems
Assess system adequacy and flexibility needs of power systems, with growing share of renewable energy generation
Assess feasibility of power sector solutions generated by the JRC-EU- TIMES model

8. Dispa-SET 2.1: unit commitment and dispatch model
Wind, PV Generation (MWh/h)
Objective
Minimise variable system costs
Constraints
Hourly demand balances (power and reserve)
Ramping constraints, minimum up and down times
Storage balances (PHS,CAES)
NTC based market coupling
Curtailment of wind, PV and load shedding (optional)
Plant output (MWh/h)
Power
Demand (MWh/h)
Plant on/off status (binary)
Commodity
Prices (EUR/t)
Variable costs/prices (EUR/MWh)
Plant data (MW, eff,…)
Emissions (t CO2)
Formulated as a tight and compact mixed integer program (MILP)
Implemented in Python and GAMS, solved with CPLEX

9. Dispa-SET 2.1 Current coverage

10. Dispa-SET 2.1: typical outputs

11. Example Dispa-SET simulation: the Belgian power system
Nuclear
plants:
Doel 1
Doel 2
Doel 3
Doel 4
Tihange 1
Tihange 2
Tihange 3
CCGT
plants:
+ Pumped hydro, wind, solar

12. Increasing VRE capacity
Peak load: 11 GW
VRE: 2.6 GW
7% of demand
VRE: 23 GW
49% of demand

13. Increasing power plant flexibility
6 GW
Nuclear
5 GW
CCGT

14. Increasing power plant flexibility
0 GW
Nuclear
11 GW
CCGT

15. Effect of storage power and capacity
Power: 1.3 GW
Capacity: 4.5h

16. Effect of storage power and capacity
Power: 4 GW
Capacity: 4.5h

17. Effect of storage power and capacity
Power: 6.3 GW
Capacity: 4.5h

18. Effect of Wind/Solar share0%
100%
Total: 20GW

19. Effect of Wind/Solar share
25%
75%
Total: 20GW
20 September September 2018

20. Effect of Wind/Solar share
50%
50%
Total: 20GW
20 September September 2018

21. Effect of Wind/Solar share
75%
25%
Total: 20GW
20 September September 2018

22. Effect of Wind/Solar share
100% 0%
Total: 20GW
20 September September 2018

23. Simplified evaluation of curtailment and storage
Focus on high shares of renewables!
Approach:
Simulation of the power system
Identification of the curtailed energy
Express the curtailed amount as a function of the power system characteristics

24. Methodology Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration

25. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

26. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

27. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

28. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

29. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

30. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

31. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

32. Simulation in Dispa-SET
Methodology
Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET

33. Methodology Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET
Cleaning up: discarding simulations with lost load
=> 219 results

34. Methodology Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET
Cleaning up: discarding simulations with lost load
=> 219 results
Gaussian Processes regression of the results to build the most likely response surface

35. Methodology Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET
Cleaning up: discarding simulations with lost load
=> 219 results
Linearization of the results
e.g.: Excess power vs thermal capacity, flexible share, storage, wind, PV
Gaussian Processes interpolation of the results to build the most likely response surface

36. Methodology Input parameters of the simulations:
Thermal capacity margin
Share of flexible capacity (CCGT vs Nuclear)
Storage Power [MW]
Storage Capacity [MWh]
Wind penetration
PV penetration
Latin hypercube:
Definition of 245 Runs
Simulation in Dispa-SET
Cleaning up: discarding simulations with lost load
=> 219 results
Implementation in TIMES
Linearization of the results
e.g.: Excess power vs thermal capacity, flexible share, storage, wind, PV
Gaussian Processes interpolation of the results to build the most likely response surface

37. Regression results (examples)

38. JRC-EU-TIMES Variable Renewable Energy (VRE)
Parametrization based on detailed analysis outside JRC-EU-TIMES
Applied to all timeslices

39. Conclusions
A methodology has been defined to extract simplified flexibility constraints from a large number of runs from a power system model
Demonstrated for the excess power, but also used to limit the dispatch of baseload generation
Could be an alternative to soft or hard-linking, or to the increase of the simulation time-resolution
Work in progress! Future work will focus on:
Implementation and extensive testing in the TIMES model
Validating the current approach in other conditions
Integration of interconnections
All methods and models are released as open-source (Dispa-SET side):

40. Evaluating flexibility and adequacy in future EU power systems: model coupling and long-term forecasting
Sylvain Quoilin, Wouter Nijs and Andreas Zucker
Joint Research Center (JRC)

42. Dispa-SET 2.1 Inputs Input database: RES generation profiles
Power plants
Demand curves
Outages
Fuel prices
Lines capacities
Minimum reservoir levels
From the same database different levels of model complexity are available:
MILP
LP with all power plants
LP one cluster per technology
LP presolve + MILP