Need for integrated simulations: Integration of electricity supply and demand Kenneth Bruninx with Erik Delarue and William D’haeseleer
Feb 23, 2016
Need for integrated simulations:
Integration of electricity supply and demand
Kenneth Bruninxwith Erik Delarue and William D’haeseleer
2
Basic principles of electricity generation
Electric power• Travels at speed of light• Is difficult to store
→ Supply must meet demand instantaneously • Network required for transport
• High voltage – Transmission• Low voltage – Distribution
Time [h]D
eman
d [M
W]
3
Basic principles of electricity generation
Different technologies used and scheduled to meet demand
4
TechnologiesElectricity generation system is mix of • Dispatchable units• Non-dispatchable units
o Often intermittent• Storage
5
TechnologiesDispatchable units: output can be actively controlled• Nuclear, coal, lignite
o Rankine steam cycleo Large units ~1000 MWo Continuous flat operation (?)
• Gas Combined Cycle o Brayton gas cycle + Rankine steam cycleo η ≈ 50%-55%o Typical size: ~400 MWo Flexible operation
• Gas turbineo Peak units, very flexible
• Renewable Energy Sources (RES) o Biomass, hydro (basin)
6
TechnologiesNon-dispatchable units: output cannot (or only limitedly) be controlled• Intermittent units
– Variable output– Output is predictable only to limited extent– Wind and solar photovoltaics (PV)
• Zero marginal cost
7
Technologies
Variable output & limited predictability
01/11/2012 00:00 02/11/2012 10:00 03/11/2012 20:00 05/11/2012 06:00 06/11/2012 16:000
100
200
300
400
500
600
700
800
900
Wind power production & forecasts (ELIA, BE)
ForecastProduction
Pow
er [M
W]
8
Technologies• Storage
o Pumped hydro storageo Compressed Air Energy Storage
(CAES)o Flywheelso Batterieso H2, CH4o …
9
Power plant schedulingGiven portfolio of power plants,
how to meet certain electricity demand?o At lowest variable costo Significant amount can be non-dispatchable generationo Taking into account technical constraints of power plantso Taking into account safety marginso Dealing with uncertaintieso Network restrictions, import/export
This optimization problem is known as the unit commitment problemo Difficult to solve because of on/off nature of decision variables
• Required to represent start-up costs and start-up behavior, minimum operating point, minimum up & down times
Power plant scheduling
Example of simplified unit commitment• Given set of power plants
o Capacity, minimum operating point, efficiency, fuel price, start-up cost, minimum up & down time
• Meet certain demand profile• Minimize fuel and start-up costs• Fuel cost dependent on load level• Technical constraints
o Respect minimum operating point if ono Respect minimum up & down time
Optimized with Mixed Integer Linear Programming (MILP)(Operations Research technique)
Power plant scheduling
Example of simplified unit commitment• Given set of power plants
o Capacity, minimum operating point, efficiency, fuel price, start-up cost, minimum up & down time
• Meet certain demand profile• Minimize fuel and start-up costs• Fuel cost dependent on load level• Technical constraints
o Respect minimum operating point if ono Respect minimum up & down time
Optimized with Mixed Integer Linear Programming (MILP)(Operations Research technique)
12
Power plant scheduling
• Consider power plants i
• Consider time periods jo E.g., one week, hourly time steps: j = 1 … 168
Unit Capacity [MW] Full load efficiency [%]
Nuclear 1000 33%
Coal 800 40%
Gas CCGT 400 50%
Gas GT 60 35%
Oil turbojet 20 30%
13
Power plant scheduling• How to meet demand at lowest cost?• Minimize
• Supply = Demand
With gi,j electricity generation of plant i in period j [MW] and dj electricity demand in period j [MW]
𝑐𝑜𝑠𝑡=∑𝑖 , 𝑗
𝑓𝑢𝑒𝑙𝑐𝑜𝑠𝑡 𝑖 , 𝑗+∑𝑖 , 𝑗𝑠𝑡𝑎𝑟𝑡𝑢𝑝𝑐𝑜𝑠𝑡 𝑖 , 𝑗
∑𝑖𝑔𝑖 , 𝑗=𝑑 𝑗
14
Power plant scheduling
Unit Capacity (Pmax) [MW]
Full load efficiency
[%]
Minimum operating
point (Pmin)[MW]
Efficiency at Pmin,
relative to full load
efficiency [%]
Startup cost S [€]
Fuel price [€/MWhp]
Nuclear 1000 33% 600 90% 60000 3.3
Coal 800 40% 240 90% 30000 12
Gas CCGT
400 50% 150 80% 20000 30
Gas GT 60 35% 15 60% 1000 30
Oil turbojet
20 30% 5 60% 0 40
Power plant scheduling
Fuel costs• Assume linear cost behavior between Pmin and Pmax• Other function possible, e.g., quadratic
𝑓𝑢𝑒𝑙𝑐𝑜𝑠𝑡 𝑖 , 𝑗=𝑐𝑖 ∙ 𝑧𝑖 , 𝑗+𝑏𝑖 ∙ δ𝑖 , 𝑗
Fuel
cos
t [€/
h]
Output, g [MW]
~ η
c
b
Linear :
Pmin Pmax
δ 𝑖 , 𝑗≤(𝑃𝑚𝑎𝑥 ¿¿ 𝑖−𝑃𝑚𝑖𝑛𝑖) ∙𝑧 𝑖 , 𝑗¿𝑔𝑖 , 𝑗=𝑃𝑚𝑖𝑛𝑖 ∙𝑧 𝑖 , 𝑗+δ𝑖 , 𝑗
δ 𝑖 , 𝑗≥0𝑧𝑖 , 𝑗=0𝑜𝑟 1δ
16
Power plant scheduling
• Deriving c and b from data previous table yields
Unit Capacity [MW]
Minimum operating
point [MW]
c [€/h] b [€/MWh]
Nuclear 1000 600 6667 8.3Coal 800 240 8000 28.6Gas CCGT 400 150 11250 51.0Gas GT 60 15 2143 66.7Oil turbojet 20 5 111 103.7
17
Power plant scheduling
Startup costs• Bringing power online (from zero to 1) incurs a cost• E.g., amount of heat required to bring steam to appropriate
temperature and pressure
𝑠𝑡𝑎𝑟𝑡𝑢𝑝𝑐𝑜𝑠𝑡𝑖 , 𝑗≥ 0
𝑠𝑡𝑎𝑟𝑡𝑢𝑝𝑐𝑜𝑠𝑡𝑖 , 𝑗≥𝑆𝑖 ∙[𝑧𝑖 , 𝑗−𝑧 𝑖 , 𝑗−1]Unit Capacity
[MW]Startup cost
S [€]
Nuclear 1000 60000Coal 800 30000Gas CCGT 400 20000Gas GT 60 1000Oil turbojet 20 0
18
Power plant scheduling: an example
Total installed capacity: 6520 MW
One week period (168 h)o Peak demand = 5244 MW (95% of dispatchable capacity)o Given certain intermittent profile
Unit Capacity [MW] # units [-]
Nuclear 1000 2
Coal 800 2
Gas CCGT 400 4
Gas GT 60 4
Oil GT 20 4
RES intermittent 2000
Power plant scheduling
20
Power plant scheduling
Results
Unit Total generation
[MWh]
Load factor
[%]
Operational cost [k€]
Average cost
[€/MWh]
CO2 emissions
[ton]
Average CO2 emissions [kg/MWh]
Nuclear 333 963 99% 3 343 10.0 0 0
Coal 226 141 84% 6 858 30.3 194 599 861
Gas CCGT 79 558 30% 5 156 64.8 33 490 421
Gas GT 2 311 6% 222 96.2 1 398 605
Oil GT 215 2% 31 145.1 210 980
RES intermittent 95 011 28% 0 0 0 0
Challenges and issues:
Challenges and issues: a paradigm shiftResidual load in Germany in 2050
Challenges and issues: a paradigm shiftYearly load duration curves of supluses due to fluctuating electricity supply
Challenges and issues
Flexibility will be required• Generation side• Storage• Interconnections• Curtailment• Demand side activation
o Through smart gridso Demand side response
Challenges and issues
Activation of demand side• Modeling of demand response
4 8 12 16 20 240
500
1000
1500
2000
2500
3000
3500
4000
hour [h]
pow
er [M
W]
unit commitment, without demand shift
windoilnatural gascoal
4 8 12 16 20 240
500
1000
1500
2000
2500
3000
3500
4000
hour [h]
pow
er [M
W]
unit commitment, with demand shift
windnatural gascoalnuclear
26
Challenges and issuesModeling of demand responseo Cost based models
• Centrally planned• Incentive payment to consumers (function of amount shifted)• Explicit modeling of flexibility
• E.g., heating/cooling systems, transport (including storage)
o Price based models• Time-of-use pricing (e.g., two tariff system), critical peak pricing,
real time pricing • Demand elasticities
• Own and cross price elasticities
• Maximizing overall social welfare• Quadratically constrained programming or iterative piecewise linear
optimization
Challenges and issues
Source: De Jonghe, Delarue, D’haeseleer, Belmans, 2011, PSCC
Fixed demand
-0.2 own-price elasticity
Challenges and issues
Source: De Jonghe, Delarue, D’haeseleer, Belmans, 2011, PSCC
Electricity price
-0.2 own-price elasticity
Challenges and issues
Source: EEX Spot prices
Challenges and issues
• Old situation:o Load drives generation
• New paradigm:o Generation drives load
o Consume when & where there is electricity generation
Integrated modelling: a case study‘Impact of intelligent thermal systems on electricity generation – a systems approach’• Research questions:
o Effect of DS flexibility on electricity generation • Cost• Carbon intensity• RES utilization
o Quantify the usefulness of DS flexibility• Motivation: in literature
o OR focus on DS: technology-based analysiso OR focus on SS: economics-based analysis (see example above)Here: integrated approach, where the (economic) flexibility stems from a technology-based model.
• Ongoing research
32
Integrated modelling: a case studyModelling approach: cost-based• Minimize
• Supply = Demand
With gi,j electricity generation of plant i in period j [MW] and dj electricity demand in period j [MW]
𝑐𝑜𝑠𝑡=∑𝑖 , 𝑗
𝑓𝑢𝑒𝑙𝑐𝑜𝑠𝑡 𝑖 , 𝑗+∑𝑖 , 𝑗𝑠𝑡𝑎𝑟𝑡𝑢𝑝𝑐𝑜𝑠𝑡 𝑖 , 𝑗
∑𝑖𝑔𝑖 , 𝑗=𝑑 𝑗 , 𝑓𝑖𝑥𝑒𝑑+𝑑 𝑗 ,𝑣𝑎𝑟
Optimization variable
Integrated modelling: a case studyModelling approach: cost-based• Variable demand is sum of demand of all flexible devices
of all consumer groups• Constraints
o Technology: peak power, stored heato Comfort level
Integrated modelling: a case studyNo flexibility 20% flexible demand
Integrated modelling: a case studyChallenges in modelling• Consumer behaviour
o Attendance pattern determines optimal solutiono Limited number of consumer groups currently
considered (25)• Correct representation of the building stock and its energy
storage potential• Problem size
o computational effort rises as more flexible device
Conclusions
• Forms of RES are non-dispatchableo Intermittent: variability & limited predictability
• Impact on dispatchable generation systemo Increases flexibility requirements
• Can be absorbed by flexible generation if limited• At higher penetration additional flexibility required
• Can lead to negative instantaneous residual demand• Storage, interconnections, curtailment, flexible demand• New paradigm: generation drives load
• Also other technical, economic and regulatory challenges