Storage Futures Study: Four Phases Framework and Modeling

June 22, 2021

Speakers: Paul Denholm, Four Phases Lead AuthorWill Frazier, ReEDS Diurnal Storage Lead Author

NREL | 2

Storage Futures Study

NREL is analyzing the rapidly increasing role of energy storage in the electrical grid through 2050.

https://www.nrel.gov/analysis/storage-futures.html

• “Four Phases” - theoretical framework driving storage deployment• Techno-Economic Analysis of Storage Technologies• Deep dive on future costs of distributed and grid batteries• Various cost-driven grid scenarios to 2050• Distributed PV + storage adoption analysis• Grid operational modeling of high-levels of storage

One Key Conclusion: Under all scenarios, dramatic growth in grid energy storage is the least cost option.

SFS: Planned reports and discussed reports today

The Four Phases of Storage Deployment: This report examines the framework developed around energy storage deployment and value in the electrical grid.

Storage Technology Modeling Input Data Report : A report on a broad set of storage technologies along with current and future costs for all modeled storage technologies including batteries, CSP, and pumped hydropower storage.

Grid-Scale Diurnal Storage Scenarios : A report on the various future capacity expansion scenarios and results developed through this project. These scenarios are modeled in the ReEDS model.

Distributed Storage Adoption Scenarios (Technical Report): A report on the various future distributed storage capacity adoption scenarios and results and implications. These scenarios reflect significant model development and analysis in the dGen model.

Grid Operational Impacts of Storage (Technical Report): A report on the operational characteristics of energy storage, validation of ReEDS scenarioson capturing value streams for energy storage as well as impacts of seasonal storage on grid operations. Released late 2021

Key Learnings Summary: A final summary report that draws on the prior reports and related literature, generates key conclusions and summarizes the entire activity. Released late 2021

= discussed today

All reports are or will be linked to the SFS website: https://www.nrel.gov/analysis/storage-futures.html

Four Phases: A Visionary Framework

• The Four Phases report synthesizes a large body of work into a straightforward narrative framework we anticipate will be helpful to stakeholders.

• The important conclusions are the trends and the drivers of deployment rather than the specific quantities and timing.

The Four Phases of Storage Deployment

Phase Primary Service National Potential in Each Phase

Duration Response Speed

Deployment prior to 2010

Peaking capacity, energy time shifting and operating reserves

23 GW of pumped hydro storage

Mostly 8–12 hr

Varies

1 Operating reserves <30 GW <1 hr Milliseconds to seconds

2 Peaking capacity 30–100 GW, strongly linked to PV deployment

2–6 hr Minutes

3 Diurnal capacity and energy time shifting

100+ GW. Depends on both on Phase 2 and deployment of variable generation resources

4–12 hr Minutes

4 Multiday to seasonal capacity and energy time shifting

Zero to more than 250 GW

Days to months

Minutes

While the Phases are roughly sequential there is considerable overlap and uncertainty!

US Historical Deployment

Pumped storage continues to play an important role, and there are likely new opportunities for additional deployments

0

5,000

10,000

15,000

20,000

25,000

1950 1960 1970 1980 1990 2000 2010

Inst

alle

d Ca

pacit

y (M

W)

Year

Pumped Hydro Storage

All Other Technologies

NREL | 7

Framework: Power vs EnergyPower-related components are annotated in red and

energy components in yellow. Images are not to scale

Valuation Framework: Costs

Simplified relationship between capital cost of energy storage and duration using 2020 cost estimates

Valuation Framework: Marginal Cost

0

200

400

600

800

1000

1200

1400

1600

1 2 3 4 5 6

Mar

gina

l Sys

tem

Cap

ital C

ost (

$/kW

h of

st

orag

e ca

paci

ty)

Hours of Storage

H2 with Underground Storage

Pumped Storage

Li-Ion Battery

Li-Ion Battery (2024 estimate)

First hour of H2 storage is $3100/kWh

Simplified relationship between capital cost of energy storage and duration using 2020 cost estimates - incremental

Services

Service DescriptionCapacity Firm capacity Energy Energy shifting/dispatch

efficiency/avoided curtailment

Transmission Avoided capacity, congestion relief

Ancillary services Operating reserves, voltage support

Four Major Categories of Bulk Power System Storage Services

Phase 1: Short-Duration Storage for Operating

Reserves

Reserve Types

Regulating Reserves

Spinning Reserves

Non-spinning Reserves

Primary Frequency Response

mS S Min Hr Day

Inertial Response

Economic Dispatch

Replacement Reserves

Ramping Reserves

1. FrequencyResponsiveReserves

3. ContingencyReserves

5. Normal operationprovided by “energyand capacity”

2. RegulatingReserves

4. RampingReserves

Services currentlynot procured via markets

Proposed or early adoption market services

Currently procured via markets

Fast Frequency Response

Option 3mS S Min Hr Day

Timescale

Example Value

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60

Annu

al O

pera

ting

Rese

rves

Val

ue

($/k

W-y

r)

Storage Duration (Minutes)

Total Value

MarginalValue Example of the total and

marginal value of spinning reserves assuming tariff value of

$75/kW-yr

Example Benefit/Cost Ratio

0

0.5

1

1.5

2

2.5

3

3.5

0 10 20 30 40 50 60

Bene

fit -

Cost

Rat

io fo

r Pro

visio

n of

O

pera

ting

Rese

rves

Storage Duration (Minutes)

Total

Marginal

Example of total and marginal B/C ratio assuming 90% battery availability and 2020 Li-Ion cost estimates

Phase 1: Short-Duration Storage for Operating

Reserves

Region Deployment (MW)Alaska & Hawaii 27 California 139 Non-CAISO Western Interconnection 29Texas 108 PJM 182 New York & New England 66 Other Regions in the Eastern Interconnection 171 Total 721

Capacity includes 656 MW of Li-ion batteries, 47 MW of flywheels, and 18 MW of other battery types.

Phase 1 Utility-Scale (>0.5 MW) Storage Deployment with 1 hour or less capacity, 2011–2019

Limits to Phase 1

0

5,000

10,000

15,000

20,000

25,000

30,000

CAISO ERCOT ISO-NE MISO NYISO PJM SPP FRCC SERC WECC Total

Nat

iona

l Cap

acity

Pro

cure

d (M

W)

SpinningContingencyReserves

RegulatingReserves

FrequencyResponse

Market Regions Non-Market Regions

0

1,000

2,000

3,000

4,000

5,000

6,000

Regi

onal

Cap

acity

Req

uire

d (M

W)

0

1,000

2,000

3,000

4,000

5,000

6,000

Regi

onal

Cap

acity

Pro

cure

d (M

W)

Current U.S. grid requirements for high-value operating reserve products potentially served by energy

storage in Phase 1

Phase 2: The Rise of Battery Peaking Plants

Installation dates of U.S. peaking capacity (non-combined Heat and Power, Combustion Turbines, Internal Combustion, oil/gas steam) (EIA 860)

02468

1012141618202224

Peak

ing

Capa

city

Inst

alle

d (G

W)

Over the next 20 years, we would expect about 150 GW of peaking capacity to retire

Key Issue: Duration Required to Reliably Decrease Net Peak

NYISO = New York Independent System Operator FRCC= Florida Reliability Coordinating Council

Duration Requirements

MarketOperator

Duration Minimum(hours)

ISO-NE 2CAISO 4NYISO 4SPP 4MISO 4PJM 10

Regional Energy Storage Duration Requirements in response to FERC 841

[1] PJM is in the process of updating this value based on an effective load carrying capability calculation. (“PJM Interconnection L.L.C., Docket No. ER21-278-000 Effective Load Carrying Capability Construct,” PJM, October 30, 2020, https://www.pjm.com/directory/etariff/FercDockets/5832/20201030-er21-278-000.pdf.)FERC 841 https://www.ferc.gov/media/order-no-841

Capacity Value as a Function of Duration

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5 6 7 8

Annu

alize

d Ca

paci

ty V

alue

($/k

W-y

r)

Storage Duration (Hours)

Total Value

Marginal Value

Value of storage providing capacity, assuming a 4-hour duration requirement and a $90/kW-yr capacity payment

Energy Shifting Value

0

10

20

30

40

50

60

70

80

90

0 1 2 3 4 5 6 7 8 9 10 11 12

Tota

l Ann

ual V

alue

of E

nerg

y Ti

me

Shift

ing

($/k

W-y

r)

Storage Duration (Hours)

CAISO

PJM

0

5

10

15

20

25

0 1 2 3 4 5 6 7 8 9 10 11 12

Mar

gina

l Ann

ual V

alue

of E

nerg

y Ti

me

Shift

ing

($/k

W-y

r)

Storage Duration (Hours)

CAISO

PJM

(a) Total value (b) Marginal value Example of the total and marginal value of energy time-shifting using

2019 energy market values

Total Value

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0

20

40

60

80

100

120

140

160

0 1 2 3 4 5 6 7 8 9 10

Frac

tion

of T

otal

Val

ue fr

om C

apac

ity

Annu

al T

otal

(Ene

rgy

and

Capa

city

) Va

lue

($/k

W-y

r)

Storage Duration (Hours)

Total Value

Marginal Value

Fraction of total value from capacity

Example of the total and marginal value and of a battery storage system providing peaking capacity

Example Benefit/Cost Ratio

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6 7 8 9 10

Bene

fit -

Cost

Rat

io fo

r Pro

visio

n of

En

ergy

and

Cap

acity

Storage Duration (Hours)

2024 Cost Estimates

2020 Cost Estimates

Example of the Benefit/Cost ratio of a battery storage system providing peaking capacity

Limits – Widening Peaks

20,000

25,000

30,000

35,000

40,000

45,000

50,000

55,000

0 6 12 18 24

Net

Dem

and

(MW

)

Hour of Day

No Storage

With Storage

With storage peak demand period is

now > 4 hours

Simulated impact of increased 4-hour storage

deployment on net load shape

But Peak Narrows with PV Deployment

0

10,000

20,000

30,000

40,000

50,000

60,000

0 6 12 18 24

Net

Dem

and

(MW

)

Hour of Day

0% PV

5% PV

10% PV

15% PV

20% PV

PV increases opportunities for storage

as peaking capacity –California Example

Potential for Phase 2

0

20

40

60

80

100

120

140

160

0% 5% 10% 15% 20% 25% 30% 35% 40% 45%6-Ho

ur o

r Sho

rter

Bat

tery

Dur

atio

ns w

ith

High

Cap

acity

Cre

dit

Annual Percentage of Electricity from PV (Conterminous U.S.)

Trend of 20,000 scenarios

The potential opportunity of Phase 2: National potential of 4–6 hour

batteries with high capacity credit

Phase 3: The Era of Ubiquitous Storage?

0

10,000

20,000

30,000

40,000

50,000

8/29 0:00 8/29 12:00 8/30 0:00 8/30 12:00 8/31 0:00

Dem

and

(MW

)

Load Net Load with 4- and 6-Hr Storage (Phase 2)

Net Load w/VG Net Load with 8-Hr Storage

Net load peak is about 7 hours wide after Phase 2

Net load peak is about 10 hours wide with example Phase 3 deployments

Example of Phase 2: 9,000 MW with 4.5-hour average duration

Example of Phase 3: 7,000 MW with 8-hour average duration

Longer peaks with greater storage deployment

NREL | 28

Storage Costs Are Projected to Decline Beyond 2020

Cost declines (from 2018) range from 21% to 67% by 2030

Source: Augustine, Chad, and Nate Blair. Energy Storage Futures Study: Storage Technology Modeling Input Data Report. Golden, CO: National Renewable Energy Laboratory. NREL/TP-5700-78694. https://www.nrel.gov/docs/fy21osti/78694.pdf.

Phase 3: Continued Deployment of <8 Hr

storage?

-10,000

-5,000

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

4/13 0:00 4/13 12:00 4/14 0:00 4/14 12:00 4/15 0:00

Pow

er (M

W)

Load Net Load w/VG VG Supply

Supply of VG exceeds electricity demand

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

4/9/24 0:001

4/9/24 6:007

4/9/24 12:0013

4/9/24 18:0019

4/10/24 0:0025

4/10/24 6:0031

4/10/24 12:0037

4/10/24 18:0043

PV G

ener

atio

n (M

W)

PV Stored PV Used Directly PV Curtailed

Curtailment lasts 6 hours

10,0004/13 0:00 4/13 12:00 4/14 0:00 4/14 12:00 4/15 0:00

Stored with Phase 2 deployments

VRE supply and net load during two spring days

Availability of curtailed energy during a spring period showing length of curtailment events

Residual curtailment after Phase 2 storage deployments

Phase 3: Ultimate Limits Due to Seasonal

Mismatch

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

9/2 0:00 9/2 12:00 9/3 0:00 9/3 12:00 9/4 0:00

Dem

and

(MW

)Load Net Load w/80% RE Net Load with 80% RE and Dirunal Storage 80%

Charging raises net load

Discharging reduces net peak load

Decline in capacity value due to a flattened net load

Simulated flattened loads in ERCOT at 80% RE

Phase 3: Ultimate Limits Due to Seasonal

Mismatch

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0

10

20

30

40

50

60

70

3/11 3/12 3/13 3/14 3/15 3/16

Stor

age

Stat

e of

Cha

rge

(%)

Gen

erat

ion

(GW

)

RE Curtailed RE Stored Discharge from Storage

RE Used Directly Normal Load Total VG Potential

Storage SOC

Residual load is not shown and is zero or negativeduring this entire period

3/17

Simulated flattened loads in ERCOT at 80% RE

Decline in time-shifting value due to zero net load.

Phase 3 Opportunities

0

20

40

60

80

100

120

140

160

180

200

0% 5% 10% 15% 20% 25% 30% 35% 40% 45%

Stor

age

with

Hig

h Ca

paci

ty C

redi

t (GW

)

Annual Percentage of Electricity from PV (Conterminous U.S.)

Up to 12 Hours

Up to 6 Hours

National opportunities for long-duration (up to 12-hour) storage providing capacity services in Phase 3

Phase 4: The Need for Residual Capacity

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

0 2000 4000 6000 8000

Net

Loa

d (M

W)

Hours at Load

Base

VG Only

VG with diurnal storage

Firm capacity needed to meet net peak demand and serve remaining 10%

0

10,000

20,000

30,000

40,000

50,000

60,000

0 2000 4000 6000 8000

Net

Loa

d (M

W)

Hours at Load

Base

VG Only

VG with diurnal storage

ERCOT

Residual load duration curves at 90% RE showing the need for significant

firm capacity California

Seasonal Storage Operation at 98% RE

Phase 4 Opportunities

0

50

100

150

200

250

300

350

80 82 84 86 88 90

Rem

aini

ng C

apac

ity R

equi

red

Pote

ntia

lly

Met

by

Seas

onal

Sto

rage

(GW

)

% of Demand Met By RE and Diurnal Storage

Low RE Capacity Credit Scenario

High RE Capacity Credit Scenario

Bounding the size of Phase 4 by estimating the national residual capacity requirements under 80%–90% RE scenarios

The Four Phases of Storage Deployment

Phase Primary Service National Potential in Each Phase

Duration Response Speed

Deployment prior to 2010

Peaking capacity, energy time shifting and operating reserves

23 GW of pumped hydro storage

Mostly 8–12 hr

Varies

1 Operating reserves <30 GW <1 hr Milliseconds to seconds

2 Peaking capacity 30–100 GW, strongly linked to PV deployment

2–6 hr Minutes

3 Diurnal capacity and energy time shifting

100+ GW. Depends on both on Phase 2 and deployment of variable generation resources

4–12 hr Minutes

4 Multiday to seasonal capacity and energy time shifting

Zero to more than 250 GW

Days to months

Minutes

While the Phases are roughly sequential there is considerable overlap and uncertainty!

Grid-Scale Diurnal Storage Scenarios

Variable Renewable Energy

CostStorage Cost Natural Gas Price

• Reference Cost• Low Wind Cost• High Wind Cost• Low PV Cost• High PV Cost

• Reference Cost• Low Battery

Cost• High Battery

Cost

• Reference NG price

• High NG price• Low NG price

Transmission cost

• Reference transmission cost

• High transmission cost

Combinations of these sensitivities are used to create a total of 19 scenarios

Improvements to Storage Representation in ReEDS 7 years of weather and

load data for analyzing system reliability

Storage resources are split out by duration to capture the relationship between duration, capacity provision, energy arbitrage, and cost

Dispatch of generation, storage, and transmission simulated for each hour of year to inform investment decisions

Modelled storage deployment in ReEDS

Interaction of Storage and Net Load (2050 in California)

Net load profiles inform capacity value of storageEnergy price profiles inform energy time-shifting value of storage

Storage Correlates with PV More than Wind

Peaking capacity potential (GW) (determined by net load shape)

Energy time-shifting potential (TWh) (determined by energy price profiles)

Amount of Generation that Goes through Storage

Transmission Correlates More with Wind

Relative Value of Storage Services

Economic Deployment versus Peaking Capacity Potential

29% PV 43%

PV 22% PV

31% PV

41% PV

Storage is optimized based on the relationship between:

- capacity value- energy value- storage duration- storage cost & performance

Grid-Scale Diurnal Storage Scenarios Key Takeaways

Questions and Discussion

Encouraging everyone to share/forward NREL outreach on social media• https://www.nrel.gov/news/program/2021/nrel-

launches-storage-futures-study-with-visionary-framework-for-dramatic-increase-in-deployment.html

• NREL also posted items on Linkedin, Twitter, etc.

https://www.nrel.gov/analysis/storage-futures.html

NREL/PR-5C00-80366

NREL | 48

Future Battery Costs by Cost Scenario - Moderate

• Use same cost projections for 4-hour BESS as in Cole & Frazier 2020*– Projections based on

literature review of 16 projections

• Adjusted cost projections for other durations to account for reductions at component– BNEF data for component-

level reductions– LIB pack costs reduce faster

than rest of component costs

– Long-duration BESS costs reduce faster (LIB make up greater share of costs

• Compared here to EPRI, BNEF and Schmidt

*https://atb.nrel.gov/

NREL | 49

Current Battery Costs – Residential

• Stand-Alone Battery Energy Storage System (BESS)– Based on methodology

used in NREL PV and BESS cost benchmarking study*

– Assumes $176/kWh battery pack (BNEF 2020)

– Installation, overhead and profit margin assumptions aligned with NREL Residential PV model

– Costs converted to linear equation based on battery power and energy capacity for use in dGen *Feldman et al. 2020 (forthcoming) U.S. Solar Photovoltaic System and Energy

Storage Cost Benchmark: Q1 2019, NREL/TP-6A20-75161

NREL | 50

Future Battery Costs by Component

• Battery pack costs decline much faster than other cost components

• Must account for this or will skew BESS costs as a function of duration

• Determined current year cost breakout by component for all durations and then applied cost reductions by component