-
© 2015 Electric Power Research Institute, Inc. All rights reserved.
Aidan Tuohy, PhD
Project Manager/Technical Leader, EPRI
Grid Operations and Planning
University of Illinois Dept of Electrical
and Computing Engineering
Feb 22, 2015
Assessing
Operational Flexibility
in Systems with
Increased Penetration
of Variable Generation
-
2© 2015 Electric Power Research Institute, Inc. All rights reserved.
Grid Operations, Planning & Integration Area
Grid Operations &
Planning
Bulk Integration Variable
Generation
Integration of Distributed
Renewables
Information & Comm.
Technologies
Transmission & Subs
P162 HVDC
PS-D HVDC Planning PS-A Modeling/Simulation
-
3© 2015 Electric Power Research Institute, Inc. All rights reserved.
Control Center
Bulk Renewable Integration R&D Focus
Schedule, Dispatch
& ReservesVoltage &
Frequency Control
Monitoring
Analysis
Decision Support
Control
New Methods/Tools
Reliable & EfficientOperation
Modeling &
Protection
0 20 40 60 80 100 1201.9
2
2.1
Time (seconds)
Vfd
(pu
)
0 20 40 60 80 100 1200.95
0.96
Vt
(pu)
0 20 40 60 80 100 1201.95
2
Ifd
(pu)
Measured
Fitted
Variability & System
Flexibility
Conventional Gen
Emerging Flexible Resources
VG Power Management
-
4© 2015 Electric Power Research Institute, Inc. All rights reserved.
US Installed Solar PV
Source: NREL Open PV Project; Bloomberg
Total US: 16 GW
2014 Install Est.: 6.5 GW
CAISOWG Capacity = 5.8 GW
PV Capacity = 8+ GW
Peak Load = 48 GW
HECOWG Capacity = 100 MW
PV Capacity = 254 MW
Peak Load = 1200 MW
ERCOTWG Capacity = 11.2 GW
PV Capacity = 250 MW
Peak Load = 56 GW
-
5© 2015 Electric Power Research Institute, Inc. All rights reserved.
How much PV and Wind Is Possible?
DOE SunShot Initiative Scenarios…
Source: NREL, “Sensitivity of Rooftop PV Projections in the SunShot Vision Study to Market Assumptions”
Optimistic, but policy objective PV assumptions, leads to
Rooftop PV of 120 GW in 2030 & 240 GW in 2050.
-
6© 2015 Electric Power Research Institute, Inc. All rights reserved.
Wind & PV Variability/Uncertainty Increases the Need for
System Flexibility
• It’s the Wind Ramp,
Not the Ripple!
• Forecasting Is Key
• System must have
ramping & cycling
capabilities
Source: Constructed from EIRGRID online data (www.eirgrid.com).
-
7© 2015 Electric Power Research Institute, Inc. All rights reserved.
0
10.000
20.000
30.000
40.000
50.000
60.000
70.000
80.000
EE D
Installiert EE D
Jahres-Mittel EE
MW
PDE max. RE*: 37.642MW
wind: 23.574MW
PV: 14.069MW
approx. 53% (14.04.2014)
min. RE*: 148MW
wind: 148MW
PV: 0MW
approx. 0,2% (17.02.2013)
h-Values
Renewable Energy
Installed Capacity
yearly average
Sourc
e: A
mprion G
mbH
Germany - Installed Renewable Capacity versus
Real Infeed Capacity since 2011
Source: Amprion GmbH - CIGRE Session 2014 - Opening Panel | Klaus
Kleinekorte | 25th August 2014 | © Amprion
-
8© 2015 Electric Power Research Institute, Inc. All rights reserved.
VER Integration Impacts
Integration Issue Bulk Renewables
(Wind and Solar)
Distributed Resources
(incl. rooftop PV)
Scheduling & Dispatch
Reserves & Frequency
Regulation
Resource Adequacy &
System Flexibility
Generation Cycling &
Retirement
System Voltage and
Frequency Impacts
Distribution Feeder
Impacts
Bulk Impacts of DER
Utility Revenue &
Business Models
-
9© 2015 Electric Power Research Institute, Inc. All rights reserved.
Ensuring Sufficient System Flexibility
-
10© 2015 Electric Power Research Institute, Inc. All rights reserved.
The “Duck” Curve
0
500
1.000
1.500
2.000
2.500
00
.00
01
.00
02
.00
03
.00
04
.00
05
.00
06
.00
07
.00
08
.00
09
.00
10
.00
11
.00
12
.00
13
.00
14
.00
15
.00
16
.00
17
.00
18
.00
19
.00
20
.00
21
.00
22
.00
23
.00
MW - Lunedì, 30 Agosto 2010
MW - Lunedì, 29 Agosto 2011
MW - Lunedì, 27 Agosto 2012
Source: ENEL – Measured Data from Southern Italy
Increased requirement for
downward ramping capability in
the morning
More upward ramping
capability is required when
sun goes down
Need lower minimum generation
levels to avoid over-generation
Not Just Resource Adequacy but the Adequacy of Resource of the Right Type
-
11© 2015 Electric Power Research Institute, Inc. All rights reserved.
Flexibility Considerations & Metrics
Many Regions (Regulators + ISO+ Utilities) Considering Future Flexibility Needs Now – Planning and Operations time frame
Other systems experiencing similar needs (Renewables and/or Retirements)– Germany, Spain, New York, Hawaii etc.
New flexible resources now becoming deployable in the bulk system
California
•Flexible Resource Adequacy
•Flexi-Ramp Market Product
•Long Term Procurement Plan
Ireland
• Long Term Flexibility Incentives
Oregon
• Integrated Resource Planning
Process
MISO
• Market Rule Changes to
Incentivize Flexibility
-
12© 2015 Electric Power Research Institute, Inc. All rights reserved.
EPRI Flexibility Metrics for system planning
Multi-Level Approach– Levels 1 and 2 screening metrics– Levels 3 and 4 detailed metrics
Four detailed metrics for planning time frame:– Periods of Flexibility Deficit – Expected Unserved Ramping – Well-being analysis – Insufficient Ramping Resource
Expectation
Post-processed metrics based on production cost study or historical data
White paper available on epri.com
Level 1
• Variability Analysis & Flexibility Requirement
Level2
• Resource Flexibility Calculation
Level 3
• System Flexibility Metrics
Level 4
• Transmission and Fuel Constrained Flexibility
-
13© 2015 Electric Power Research Institute, Inc. All rights reserved.
Level 2: What flexibility is available?
Determine Ramping Available in Each Hour of the YearDifferent time scales, need to make assumptions
about intertie and energy limited resources
-
14© 2015 Electric Power Research Institute, Inc. All rights reserved.
Need to consider operational aspects…
How you operate the system – including reserves – impacts on the availability of flexible capacity
-
15© 2015 Electric Power Research Institute, Inc. All rights reserved.
Level 3: What is the net flexibility after ramps?
Examine either net available flexibility or against extreme ramps
-
16© 2015 Electric Power Research Institute, Inc. All rights reserved.
Level 3: Metric 1: Periods of Flexibility Deficit
30%
40%
50%
60%
70%
80%
90%
100%
0 2 4 6 8 10 12 14 16
Pe
rio
ds
of
Fle
xib
ilit
y D
efi
cit
(%
)
Time Horizon (Hours)
Example for extreme ramping requirements (97th percentile)
-
17© 2015 Electric Power Research Institute, Inc. All rights reserved.
Level 3: Metric 2. Expected Ramping Deficit
Expected value of ramping deficit values observed in each direction and time horizon
8760
1
/,0||/,8760
1 i
i
DCit iDeficitERD
-
18© 2015 Electric Power Research Institute, Inc. All rights reserved.
Example Conclusions from Studies
“Flexibility shortages seen over 60 minute interval”
– Higher resolution data would be beneficial
“Peak at 540 minutes indicates that the system may need to add additional capacity”
– Unlikely that system leaves long start units offline when available
– In that case the system had interties assumed inflexible
Assumptions on intertie flexibility may alleviate issue
Frequency of shortages for
-
19© 2015 Electric Power Research Institute, Inc. All rights reserved.
Time of Day and Year
Example Ramp Rates
1 2 3 4 5 6 7 8 9 101112131415161718192021222324
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
0
200
400
600
800
1000
1200
1400
1600
1800
Flexibility resources will need to be available at different times of day and year depending on system requirements
Max 1 hour up ramps for each hourly interval in a given month
Based on example data from Northwest US
Calculate for different time intervals, up and down ramping
-
20© 2015 Electric Power Research Institute, Inc. All rights reserved.
Demand Response as Flexible Resource
Types of Loads:– What will be available and when?
– How long will it be available for?
– How is it controlled?
– Will be examined in this project and quantified for case studies
– Can contribute to system operators assessment of DR as a resource for providing operational flexibility
DR operations:– Ramp limits
– Call rate limits
– Energy limits
– Duration limits
– Time of day/week/year availability
– Efficiency of pre-loading & make up energy
System Operators will need to be able to characterize if and how DR can provide operational flexibility
-
21© 2015 Electric Power Research Institute, Inc. All rights reserved.
What about storage?
Storage is a very flexible resource
But also very expensive and inefficient
Need to identify key places where storage will play a role in aiding integration of VG
Still in progress – as more VG comes online, storage becomes more attractive
-
22© 2015 Electric Power Research Institute, Inc. All rights reserved.
Today
Energy
Capacity
Ancillary Services
Future ?
Energy
Capacity
Ancillary Services
Central Station Energy StorageDemand Response
Value of Capacity and Services
Variable Generation
-
23© 2015 Electric Power Research Institute, Inc. All rights reserved.
Increased variation in system conditions
7,2
50
8,0
00
8,7
50
9,5
00
10,2
50
11,0
00
11,7
50
12,5
00
13,2
50
14,0
00
14,7
50
15,5
00
16,2
50
17,0
00
17,7
50
18,5
00
19,2
50
20,0
00
20,7
50
21,5
00
1
10
100
1000
0
800
1,600
2,400
3,200
4,000
4,800
5,600
6,400
Region Load (MW)
Frequency of Occurence
Renewable Output (MW)
100-1000
10-100
1-10
Low probability,
high impact events
Low probability,
high impact events
Average system conditions
-
24© 2015 Electric Power Research Institute, Inc. All rights reserved.
Ongoing and future work
Impact of transmission constraints on system flexibility adequacy
– Working with Ecco International to demonstrate how metrics can consider flexibility and how transmission can be a flexibility resource
Flexibility from Energy Limited Resources
– How do demand response, storage, hydro etc provide flexibility?
Case Studies to demonstrate framework and ‘baseline’ flexibility
Potential future directions:
– Gas network and interaction with gas markets
– Transmission planning interaction with resource adequacy
– Operational versus planning issues – making sense of why there is insufficient flexibility
– Markets for incentivizing flexibility (see next section)
-
25© 2015 Electric Power Research Institute, Inc. All rights reserved.
Managing Uncertainty in Operations
Forecasting Solar and Operational Tools for Reserve Procurement
-
26© 2015 Electric Power Research Institute, Inc. All rights reserved.
Responding to Variability
Frequency
Droop
AGCCapacity
Reserves
Operating
Reserve
Spinning
Reserve
Va
riabili
ty &
Uncert
ain
ty (
MW
)
1 day 4 hr 30 min 5 min 10 s 0 s
Energy Storage
Intelligent Distribution Devices
Demand Response
Inertia
Governor Response
Regulating Units
Hydro
Simple-Cycle GT
Combined Cycle
Warm ST
Cold ST
Source: Russ Philbrick, PES General Meeting, Detroit, July 2011
-
27© 2015 Electric Power Research Institute, Inc. All rights reserved.
Solar Forecasting – Texas solar plants
-
28© 2015 Electric Power Research Institute, Inc. All rights reserved.
Daily Mean Absolute Percentage Error
Many other metrics could be calculated – research needed
the best ways to assess forecast accuracy and value
Weekly MAE shows range of performance of individual and combined forecasts
Difficult to forecast distributed PV, but may have some geographic diversity?
-
29© 2015 Electric Power Research Institute, Inc. All rights reserved.
Operating With Increased Uncertainty
Source: Pierre Pinson, DTU, Denmark
Average day ahead error: 8%-10% for wind farm, 4% for system
Ramp error: Over 50% for large ramps
-
30© 2015 Electric Power Research Institute, Inc. All rights reserved.
Ancillary Services/ Reserves – Industry Rethink
FlexiRamp
– Reserving flexible capacity for use in real time
– New methods to quantify requirements
Ramp Product & Look Ahead Dispatch
– Similar requirement to California ISO
Ancillary Service Review
– Wide scale reorganization of ancillary services
Cooperative balancing in Europe, CAISO /Pacificorp
Other methods being developed in BPA, HECO, etc
-
31© 2015 Electric Power Research Institute, Inc. All rights reserved.
Flexible ramping reserves
Requirements
based on short-
time variability
and uncertainty
– Confidence
intervals inform
requirements
– Demand
curves for
flexibility
31
Based on offline analysis will be included in markets soon
-
32© 2015 Electric Power Research Institute, Inc. All rights reserved.
Reserve Determination for VG - Survey
• Many areas already considering wind and/or PV
• Multiple manners to consider:
• Static versus dynamic
• Forecasted versus historical
• Scheduling & dispatch practices also matter
Broad Spectrum of Approaches Being Used
-
33© 2015 Electric Power Research Institute, Inc. All rights reserved.
Methods for managing uncertainty
Static
Reserve
Dynamic
Reserve
No
Reserve
Stochastic
UC
Re
liab
ility
/effic
ien
cy im
pro
vem
en
t
Computation time
-
34© 2015 Electric Power Research Institute, Inc. All rights reserved.
EXISTING
NEW
EPRI project - Stochastic Reserve Procurement Process
System
Data
Hour-Ahead (HA)
Deterministic
Commitment
Intra-Day
(ID)Determinist
ic Commitment
Forecast
Probabilistic
Data
Real Time
(RT) Dispatch
(ID_ST)
Reserve
Determination
Integrate probabilistic information into existing deterministic processes
Day-Ahead (DA)
Deterministic
Commitment
-
35© 2015 Electric Power Research Institute, Inc. All rights reserved.
Overview of multi cycle modeling
Updated information in each cycle data requirement
Updated unit schedules as time progresses
Fewer options available to meet errors
Models consumption of load following type reserve
Source: Russ Philbrick, Utility Variable Generation Working Group, April 16, 2012, San Diego
Cycle 1: ~4 HA
Cycle 2: 90 MA
Cycle 3: Real time
-
36© 2015 Electric Power Research Institute, Inc. All rights reserved.
Stochastic Scenario Development
Step 1: Probabilistic
Site Forecasts
Step 2: Probabilistic
Aggregate Forecast
Step 3: Forecast
Trajectories
MW
MW
MW
MW
Time
Prob.
Prob.
Prob.
Coherent scenarios with
trajectory for each site
-
37© 2015 Electric Power Research Institute, Inc. All rights reserved.
Stochastic Multi Cycle
LFU and LFD Procurement
Load Following procurement seen to follow solar forecast uncertainty.
-
38© 2015 Electric Power Research Institute, Inc. All rights reserved.
Findings from Stochastic and Multi-Cycle Modeling
1. Important to replicate the time constraints associated with each
type of resource on multiple time horizons
– As uncertainty is realized, dispatch and commitments may need to
change.
2. Choose reserve procurement policies which cover the
uncertainty at each time
– E.g. load following reserve should be held by online units as
commitment is not possible when the reserve is to be released
– Choices relating to when and how to reserve are released can also
impact on energy prices.
3. Stochastic Modeling is achievable and realistic
– May want to use stochastic methods to inform deterministic process
at first to improve operators understanding and comfort
-
39© 2015 Electric Power Research Institute, Inc. All rights reserved.
Overview of the Integrated
Grid Benefit/Cost
Framework
An
Integrated
Grid
-
40© 2015 Electric Power Research Institute, Inc. All rights reserved.
Integrated Approach to Deploying DER
Consistent, transparent framework for
assessing benefits/costs of
transitioning to an Integrated Grid.
An
Integrated
Grid
-
41© 2015 Electric Power Research Institute, Inc. All rights reserved.
Strategic Planning with DER
Value of
Solar?
Storage
Deployment?
Community
Solar?
Smart Inverter
Cost/Benefit?
Proactively
Upgrade?
Research Questions Core Assumptions
Study
Timeframe
Regulatory
Framework
Resource Mix
Expected DER
Growth
Environmental
Impact
Analytical process must be consistent, repeatable, and transparent
-
42© 2015 Electric Power Research Institute, Inc. All rights reserved.
EPRI’s Integrated Grid Benefit-Cost Framework
Hosting
CapacityEnergy
Thermal
Capacity
Distribution System
Bulk System
Customer or
Owner
Cost/Benefits
Societal
Costs/Benefits
Benefit/Cost
1 4
5
3
2
6
Core Assumptions
Adoption/
Deployment
Scenarios
Market
Conditions
Resource
Adequacy
Flexibility
Operational Practices &
Simulation
Transmission
Performance
Transmission
Expansion
System
Net Costs
System
Benefits
Reliability
-
43© 2015 Electric Power Research Institute, Inc. All rights reserved.
Hosting
CapacityEnergy
Thermal
Capacity
Distribution System
Bulk System
Customer or
Owner
Cost/Benefits
Societal
Costs/Benefit
s
Benefit/Cost
1 4
5
3
2
6
Core Assumptions
Adoption/
Deployment
Scenarios
Market
Conditions
Resource
Adequacy
Flexibility
Operational Practices & Simulation
Transmission
Performance
Transmission
Expansion
System
Net Costs
System
Benefits
Reliability
Integrated Grid Benefit Cost Framework
B
Feeder Performance Characterization
Feeder Hosting Capacity Analysis
Feeder Clustering Based on Hosting
Feeder Clustering Based on Energy
Impacts
Capacity Analysis
Energy Analysis on Select Feeders
Losses/Consumption Results Extrapolated
to System
Mitigation Evaluation
Benefit/Cost Analysis
Bulk System Analysis
Feeder Models
Load Data
Hosting Capacity Analysis
Energy Analysis
Capacity Analysis
Asset Deferral
DER Performance Characterization
DER Data
Optimal Hosting Capacity Location
B
C
D
Reliability Analysis
Reliability Analysis
E
Substation – Level Hosting Capacity for
DER
Feeder-Specific Hosting Capacity for
DER
A
Transmission System
Performance Studies
DER Scenarios
Resource Adequacy
Existing SystemModel(s)
Load Forecasts
Variability Profiles
Existing Generation
Existing Network Model
Resource Epxansion
LOLE/Reserve Margin & Capacity
Credit
New Resources/Expansion
Plan
Thermal / Voltage Impacts
Operational Simulations
Resource Dispatches
Transmission System
Upgrades
Technology options
Transmission Expansion
Losses
Reliability Impacts
Reserve & Operational
Changes
LOLE/Reserve Margin & Capacity
Credit
New Reserve & Operational
Modes
Integrated Grid
Bulk System
Analysis
Framework
Costs of new resources
Production Costs & Marginal
Costs
Costs of mitigation/upgrades
Cost of Losses
Cost of Base Case
Cost of Scenario
System Flexibility
Assessment
Flexibility Metrics
Line Type Legend
Data Input
Final Result
Feed-Forward Result
Feed Back Result
Frequency Impacts
Hosting Capacity PV & Demand
Profiles (See Fig. 5.3)
PQ & Protection Impacts
-
44© 2015 Electric Power Research Institute, Inc. All rights reserved.
Transmission System
Performance Studies
DER Scenarios
Resource Adequacy
Existing SystemModel(s)
Load Forecasts
Variability Profiles
Existing Generation
Existing Network Model
Resource Epxansion
LOLE/Reserve Margin & Capacity
Credit
New Resources/Expansion
Plan
Thermal / Voltage Impacts
Operational Simulations
Resource Dispatches
Transmission System
Upgrades
Technology options
Transmission Expansion
Losses
Reliability Impacts
Reserve & Operational
Changes
LOLE/Reserve Margin & Capacity
Credit
New Reserve & Operational
Modes
Integrated Grid
Bulk System
Analysis
Framework
Costs of new resources
Production Costs & Marginal
Costs
Costs of mitigation/upgrades
Cost of Losses
Cost of Base Case
Cost of Scenario
System Flexibility
Assessment
Flexibility Metrics
Line Type Legend
Data Input
Final Result
Feed-Forward Result
Feed Back Result
Frequency Impacts
Hosting Capacity PV & Demand
Profiles (See Fig. 5.3)
PQ & Protection Impacts
Integrated Grid: Bulk System Analysis
1 RESOURCE ADEQUACY
2 FLEXIBILITY3 OPERATIONAL
SIMULATION
4 TRANSMISSION PERFORMANCE
5 TRANSMISSION EXPANSION
-
45© 2015 Electric Power Research Institute, Inc. All rights reserved.
Summary/Conclusions
Large penetrations of variable generations already being seeing and will continue to grow– Much of it is distributed which adds particular challenges
Planning time frames will need to ensure sufficient operational flexibility is available– Need methods and metrics to assess flexibility adequacy
– Consider demand response, energy storage, VG itself and transmission as well as conventional generation
Operational planning and market operations will see increased uncertainty– Requires additional operating reserves to manage wind/solar in appropriate time
scales
– New methods for scheduling and dispatch stochastic or other
Frameworks to assess the impacts and benefits of new technologies– Consider distribution, transmission and economics
– Fully integrate rather than just interconnect new resources
-
46© 2015 Electric Power Research Institute, Inc. All rights reserved.
Together…Shaping the Future of Electricity