Medium-term reliability: System Balancing Debbie Lew WIEB/WIRAB Tutorial April 22, 2020 1
Medium-term reliability: System Balancing
Debbie Lew
WIEB/WIRAB Tutorial
April 22, 2020
1
Webinar Outline
• April 15 – Resource Adequacy – long-term reliability
• April 22 – System Balancing – medium-term reliability
• April 29 – System Stability part 1 – short-term reliability
• May 6 – System Stability part 2 – short-term reliability
• May 20 – 100% Clean Energy and Distributed Energy Resources
2
3
Short-term Medium-term Long-term
System Stability System Balancing Resource Adequacy
• Seasonal mismatch of supply and
demand
• Multiple days in a row of low
wind/solar/hydro
• 1 day in 10 years Loss of Load
Expectation
• Wind and solar variability and
uncertainty
• Reducing curtailment
• Managing imbalance
• High penetrations of inverter-based
resources (IBR)
• Essential reliability services
• Frequency, transient, and small signal
stability issues
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
5,500
0 1000 2000 3000 4000 5000 6000 7000 8000
Cu
rtai
led
En
ergy
(MW
h)
Hour of Year
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
5,500
0 1000 2000 3000 4000 5000 6000 7000 8000
Net
Lo
ad (
MW
h)
Hour of Year
Graphics from left to right: EU-MIGRATE 2016
CAISO, Fast Facts 2016
A. Bloom, Energy Systems Integration Group’s Planning WG Oct 2018
Timescales of grid reliability
4
Short-term Medium-term Long-term
System Stability System Balancing Resource Adequacy
• Seasonal mismatch of supply and
demand
• Multiple days in a row of low
wind/solar/hydro
• 1 day in 10 years Loss of Load
Expectation
• Wind and solar variability and
uncertainty
• Reducing curtailment
• Managing imbalance
• High penetrations of inverter-based
resources (IBR)
• Essential reliability services
• Frequency, transient, and small signal
stability issues
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
5,500
0 1000 2000 3000 4000 5000 6000 7000 8000
Cu
rtai
led
En
ergy
(MW
h)
Hour of Year
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
5,500
0 1000 2000 3000 4000 5000 6000 7000 8000
Net
Lo
ad (
MW
h)
Hour of Year
Graphics from left to right: EU-MIGRATE 2016
CAISO, Fast Facts 2016
A. Bloom, Energy Systems Integration Group’s Planning WG Oct 2018
Timescales of grid reliability
These are all fundamentally about the generation and
load balance
Wind and solar are uncertain and variable
5
Uncertainty
• Wind and solar generators are not always available when called upon
• Output is predicted by a forecast
• Actual power output differs from forecasts
Variability
• Wind and solar generation vary as the intensity of their energy sources
• Several timescales … minute (regulation), hour (ramping), daily, seasonal, inter-annual
0
5000
10000
15000
20000
25000
0 6 12 18 24 30 36 42 48
Hours
MW
Wind Actual
Forecasted Wind
0
500
1000
1500
2000
2500
0 6 12 18 24 30 36 42 48
Hours
MW
CSP Actual
CSP Forecasted
A perfect forecast eliminates uncertainty but there is still variability
Overview of system operations with conventional generation
6
• Forecast load
• Day-ahead – run security-constrained unit commitment and economic dispatch
• Determine hourly schedules for all generators
• Dispatch generators to set points at each interval (5-60 min)
• Regulation reserve adjusts generator output to manage variability within the interval
Graphic source: E. Ela, EPRI, ESIG Forecasting workshop, 2018
From coarse to fine control of generation/load balance
Security-constrained unit commitment and economic dispatch
7
• Optimize operations for lowest production cost (fuel and variable O&M)
• Which units need to be on (unit commitment)?
• How much output from each unit (economic dispatch)?
• Generators have constraints – Pmin(minimum generation), minimum downtime, minimum runtime, ramp rates, startup times
• Hold headroom for operating reserves
• Power flows respect transmission limits. “Security-constrained” accounts for credible contingencies like a transmission outage and ensures resulting power flow will still respect limits
Graphic: Goggin, AWEA, 2017
Need to meet NERC standards
8
• Area Control Error (ACE) is essentially the Balancing Authority’s (BA’s) power balance error
• Real Power Balancing Control Performance BAL-001-2• Control interconnection frequency within defined limits• Control Performance Standard 1 (CPS1) is a measure of ACE
variability. Want CPS1≥100%• BA ACE Limit (BAAL) limits ACE as a function of frequency• Regulation reserves provide this control. BA’s decide how much
reg to hold
• Disturbance Control Standard BAL-002-3/BAL-002-WECC-2a• Contingency reserve for recovery from an event to balance
resources and restore ACE• WECC BA’s hold the greater of 1) most severe single
contingency or 2) sum of 3% of load plus 3% of generation• Capable of full response within 10 minutes• Half must be spinning (online, responds to frequency)
Graphic: NERC 2013,
https://www.nerc.com/pa/StandProject2010141Phase1of
BalancingAuthorityRe/BAL-001-
2_Background_Document_Clean-20130301.pdf
https://www.nerc.com/pa/StandProject2010141Phase1ofBalancingAuthorityRe/BAL-001-2_Background_Document_Clean-20130301.pdf
Ancillary services provide fine control and respond to events
9Source: Ela, et al, NREL, 2011 https://www.nrel.gov/docs/fy11osti/51978.pdf
https://www.nrel.gov/docs/fy11osti/51978.pdf
Add wind, PV, DER, more variable load
• Forecast wind/PV!• Forecasts are more accurate when they
are closer to real-time: make decisions as close to real-time as possible.
• Forecasts should line up with decision points (such as whether to commit a new unit)
• The grid will need for more flexibility for:• Variability and uncertainty of wind/PV• More variable load• DER
Graphic: T. Miller, SPP, ESIG Forecasting Workshop, June 2018; Goggin, AWEA, 2017
Mea
n a
bso
lute
% e
rro
r10
Wind and solar impact the grid differently
Source: Ela, 2008.
http://www.nrel.gov/
docs/fy08osti/43373.
Source:
CAISO
Projected net
load: spring
2020
Solar
• Ramp management
• Diurnal mismatch
• Turn off or turn down other power plants midday
Wind
• Ramp prediction
• With geographic diversity, relatively smaller ramps
• Hold more reserves
• Wind ramp forecasts
http://www.nrel.gov/docs/fy08osti/43373.pdf
Finer control and look-ahead
• Visibility and fine control for balancing and congestion
• Faster dispatch
• Far look-ahead to position system for ramps
• Situational awareness
• DER can complicate this
12
Finer control and look-ahead
Seconds (PrimaryFrequencyResponse)
Minutes (FrequencyRegulation)
Hourly (Dispatch) Day-ahead (Unitcommitment)
Traditional control for system balancing
13
Finer control and look-ahead
Seconds (PrimaryFrequencyResponse)
Minutes (FrequencyRegulation)
Hourly (Dispatch) Day-ahead (Unitcommitment)
Traditional control for system balancing
Four-hour-ahead(Unit commitment)
5-minute (dispatch)Fast Frequency
Response
Flexibility/ ramping reserves
Dispatch look-ahead 1-2 hours
Unit commitment look-ahead
14
New ancillary services
Graphic: C. Loutan, CAISO, “Operating Impacts of higher VG Penetrations – A Panel Discussion,” UVIG 2017 Fall Workshop, Oct 2017
• ERCOT recently revised ancillary services to include fast frequency response
• Ireland added seven system services including ramping products
• CAISO and MISO have ramping products
• Xcel/PSCO holds a 30-min flex reserve
15
It can be difficult for CAISO to balance their
system in the late afternoon
Reserve requirements based on risk
• ERCOT - Islanded grid: 73.5 GW peak; 22 GW wind and 2 GW PV
• Non-spin reserve service (NSRS) manages load/wind/solar forecast errors.
• NSRS requires full output in 30 minutes.
• Previously procured fixed levels of NSRS, but now based on expected conditions and need.
• Cover 95% of uncertainty during high variability risk periods and 70% during low risk periods
Source: Lew, et al, Secrets of Successful Integration, IEEE PES Magazine, Nov/Dec 2019
Must ensure adequate reserves for these events
16
No wind/solar
High wind
High solar
High wind case
25% wind
8% solar
High solar case
8% wind
25% solar
No wind/solar
case
System dispatch
Lew et al, “Western Wind and Solar Integration
Study,” NREL, 2013
The grid needs flexibilityBefore you spend money on new infrastructure, consider:
18
• What do I need it to do? Can I do that with existing infrastructure?
• Can I recover my costs through revenue? Will it pay for itself?
• Is it the cheapest way to address my need?
• Storage competes with load flexibility from customer prices (eg, time-of-use rates) and demand response
• These options interact. The more constrained your system is, the more value one of these options will provide.
Source: US Dept of Energy Wind Program
There are many options for flexibility and a range of costs
Flexibility Options
1. Transmission, larger markets, big and fast trading
2. Forecasting
3. Extract flexibility from non-variable resources
4. Extract flexibility from wind and solar
5. Storage
6. Demand-side flexibility
Each tool can saturate – we need them all
Energy Strategies, “Western Flexibility Assessment,” WIEB, Dec 10, 201919
1) Geographic diversity provides flexibility
Left: Lew et al, Western Wind and Solar Integration Study Phase II, NREL, 2013; Right: GE, Western Wind and Solar Integration Study, NREL, 2010.
New Mexico (2006)
• Geographic diversity smooths load, wind and solar
• Geographic diversity also provides more dispatchable generators to help manage variability/uncertainty
• Transmission and markets enable diversity
20
Bigger and faster tradingReal-time energy imbalance market being transitioned to day-ahead market
21
EDAMEIM
Left: https://www.westerneim.com/Documents/ISO-EIMBenefitsReportQ2-2019.pdf;
right E. Ela, EPRI, ESIG Forecasting workshop, 2018
EIM allows trading partner to dispatch down in real-time, in order to buy lower cost generation from a neighbor
EDAM enables trading partner to not startup a generator, in order to buy
lower cost generation from a neighbor
https://www.westerneim.com/Documents/ISO-EIMBenefitsReportQ2-2019.pdf
22
• Denmark has 6.1 GW of peak demand, 5.7 GW wind and 1 GW PV. Half of their load is served by wind/PV annually.
• Instantaneous wind penetration reached 149% of demand
• Interconnections (6 GW) are well-exploited – day-ahead market coupling has expanded to intra-day markets which mitigates intra-day forecast errors and variability.
• Denmark sources flexibility from all resources: thermal plants, wind plants, demand-side resources can all provide down-regulation
Energinet has used many sources of flexibility
Graphic: P. Jorgensen, Energinet, 2019
2) Forecasting saves money
23Left: GE Energy, Western Wind and Solar Integration Study, 2010 https://www.nrel.gov/grid/wwsis.html ; Right: Kari Hassler, PSCO,
ESIG Forecasting Workshop, June 2018
Forecasting saves up to
14% in annual operating
costs. If forecasts were
perfect, an additional 1-2%
could be saved.
Savings over time from
optimized operations with
continually improvements
of wind and PV forecasts
https://www.nrel.gov/grid/wwsis.html
24
• SPP has 21.4 GW wind; 50.6 GW peak demand, 72% instantaneous
• Low pressure systems cause instabilities. Low-level jet movement significantly impacts wind output
• Largest wind down-ramp 16.1 GW in 21 hours – it was forecast. Largest one-hour ramp was 4 GW
• SPP has 5 GW that can start-up within 1 hour
• What if a forecast error occurred during:• In May 2018, all quick-starts were committed in day-
ahead market• The largest wind ramp had steep down-ramps during
morning and evening when load was increasing which exacerbated net load ramps.
• Uncertainty Response Team assesses risk in 1, 4 and 8 hour look-ahead horizons over next 7 days
Advanced forecasting in Southwest Power Pool (SPP)
Source: B. Rew, SPP, 2019; Lew, et al, Secrets of Successful Integration, IEEE PES Magazine, Nov/Dec 2019
7 GW wind forecast error
3) Flexibility of non-variable generatorsCalifornia Pumped Hydro Storage
25
Left, adapted from: A. Park, Stanford University, “Data-
driven unit generating parameters,” WIEB, May 2019; Right:
ORNL, DOE 2017 Hydropower Market Report, April 2018
Pmin from operating data in WECC
Pm
in(%
)
Some WECC plant operators are already
achieving very low Pmin
Impacts of wind/PV on cycling costs and emissions
26Lew et al, “Western Wind and Solar Integration Study Phase 2,” NREL, 2013
$0.035 – $0.157 Billion
26
4) Wind and PV can provide flexibility
27
• Dispatch wind and PV (eg, MISO DIR)• Move away from ‘must-take’ resources and manual curtailment
• Use real-time forecast as maximum output
• Rooftop PV is not easily dispatched
• Regulation reserves (eg, Xcel)• Reduced curtailment and improved control
• Contingency reserves (eg, Xcel)
• Wind and PV can provide other essential reliability services (see next week’s webinar)
• If wind and PV provide essential reliability services, you may avoid reliability-must-run thermal generation, which has economic and emissions consequences
Think of curtailed wind/solar as a resource
The grid needs essential reliability services to operate stably
Graphic: Milligan, et al, “Alternatives No More,” IEEE PES Magazine, Nov/Dec 2015.
MW
ERCOT reduces regulation while increasing windDeclining regulation reserve requirements Improved CPS1 reliability metric
Source: Lew, et al, Secrets of Successful Integration, IEEE PES Magazine, Nov/Dec 2019
Reg-up
requirement
decreasing
Wind
increasing
28
5) Storage
29
• Energy storage is the most obvious answer to the duck curve.
• Difficult to make a case for stand-alone batteries based on price arbitrage alone. But battery storage is increasingly providing capacity to meet daily peak, and other services.
• A portfolio of storage resources will likely be in our future
• Electricity storage technologies: pumped hydro storage, compressed air energy storage, liquid air energy storage, flow batteries, hydrogen, vehicle-to-grid
• Thermal energy storage may provide low-cost solutions
Top: CAISO; bottom: Valmoro, Viking Cold Solutions, “Intelligent Thermal Energy Storage,” ESIG Spring 2020
• Hybrid plants proliferating, largely because of the 30% investment tax credit:• PV/battery
• Wind/battery
• PV/wind/battery
• Gas turbine/battery
• Hybridizing plants can improve their attributes• Diurnal mismatch between load and generation
• To firm variable renewables
• Meet ramp-down rates (eg Hawaii)
• Faster startup times
• Faster ramp times
• Reduce Pmin of conventional units to zero (eg SCE)
Hybrid plants increase flexibility of generators
Pictures: THEenergy, 2017; V. Kaushik, SCE, “Reality Check on Storage,” IEEE PES T&D 201830
Kauai manages high penetrations of PVRan at 100% renewables for 8 hours
31Source: Rockwell, KIUC, personal communication, 2020
• Operating reserves:• Ample quick-start diesel
reciprocating engines but can’t start-up fast enough (seconds not minutes) to meet cloud events
• Pioneered hybrid PV/battery plants. Battery storage provides spinning contingency reserves for those PV plants.
• Holds 50% of real-time PV output for those PV plants that are not backed by battery storage
Batteries enable 100% renewables in the Kauai island system
6) Dispatchable demand
We used to control a portfolio of
generators to balance a fixed demand
In the future, we’ll control both demand
and generation to balance the system
Demand
Instead of using generation to balance demand,
can we use demand to balance generation? 32
April 22, 2020 33
We know how to clean this part up
Electrification can help clean this part up
We need to integrate other energy sectors
We need to integrate other energy sectors
April 22, 2020 34
High renewables
penetrations in the
electricity sector will require
significant flexibility that
may be cost-effectively
sourced from the existing
fuel-based heating and
transportation sectors
Programs and pricing for demand-side flexibility
35Hines, Tierra Resource Consultants, PLMA, Apr 2020; Thompson, APS, “What if Your Neighboring Utility Goes to 100% Renewables,” CREPC April 2019;
Lew, GE, “DER Compensation: Colorado Springs Utilities Solar Program Design Study,” ESIG Oct 2017
APS sells cheap midday power
Rates and programs are critical enablers
of demand-side flexibility
DSM exacerbating duck curveOrchestrating TOU rates and DR programs
Conclusions
36
• System operators already have a number of tools to balance the system. Additional tools may be needed to provide finer control and better look-ahead.
• There are various measures that can provide flexibility to integrate wind/PV. At some point, the effect of one mitigation option saturates so a suite of mitigation options is necessary.
• Wind and PV can provide essential reliability services and this can help reduce reserves and balance the system.
• Electrification is a key because other energy sectors may have natural flexibility in their demand, but it can help or hurt the grid depending on how it done. The ‘ultimate lever’ that we have to address system balancing is controllable and/or price-sensitive demand. Orchestrating demand with prices and programs will be critical.
36
References
37
• Integration studies: • Western Flexibility Assessment https://westernenergyboard.org/wp-content/uploads/2019/12/12-10-19-ES-WIEB-Western-
Flexibility-Assessment-Final-Report.pdf
• Western Wind and Solar Integration Study Phases 1&2 https://www.nrel.gov/grid/wwsis.html
• Electrification futures study https://www.nrel.gov/analysis/electrification-futures.html and https://www.nrel.gov/docs/fy18osti/71500.pdf
• Best practices https://www.esig.energy/wp-content/uploads/2020/01/MPE2930855-final-Lew.pdf
• Demand response https://brattlefiles.blob.core.windows.net/files/16639_national_potential_for_load_flexibility_-_final.pdf
• Essential reliability services from wind/PV https://www.caiso.com/Documents/UsingRenewablesToOperateLow-CarbonGrid.pdf
• Operating reserves https://www.nrel.gov/docs/fy11osti/51978.pdf
• NERC Standards full set 2020 https://www.nerc.com/pa/Stand/Reliability%20Standards%20Complete%20Set/RSCompleteSet.pdf
37
https://westernenergyboard.org/wp-content/uploads/2019/12/12-10-19-ES-WIEB-Western-Flexibility-Assessment-Final-Report.pdfhttps://www.nrel.gov/grid/wwsis.htmlhttps://www.nrel.gov/analysis/electrification-futures.htmlhttps://www.nrel.gov/docs/fy18osti/71500.pdfhttps://www.esig.energy/wp-content/uploads/2020/01/MPE2930855-final-Lew.pdfhttps://brattlefiles.blob.core.windows.net/files/16639_national_potential_for_load_flexibility_-_final.pdfhttps://www.caiso.com/Documents/UsingRenewablesToOperateLow-CarbonGrid.pdfhttps://www.nrel.gov/docs/fy11osti/51978.pdfhttps://www.nerc.com/pa/Stand/Reliability%20Standards%20Complete%20Set/RSCompleteSet.pdf
38
Debbie [email protected]
(303) 819-3470
mailto:[email protected]