Marc PerezClean Power Research

Inexpensive, Firm PV Without Conventional Backup:The Role of Supply Shaping Through Curtailment

This work is based upon work supported by:National Science Foundation GRF Grant No. DGE 1144155

DoE Sunshot Grant No. DE‐EE0007669Columbia University Center for Life Cycle Analysis

Clean Power Research

For achieving high‐penetration renewables!

Why do we think we need backup to achieve high penetration?

Jan Mar May Jul Sep Nov Jan

020

040

060

080

010

00

allMISO$datetime

min

ne_d

at.tp

i$G

loba

l

Hourly Interval

W/m

2Intraday Variability

Storage is considered to be the keystone to mitigate variability on all timescalesIt’s because wind + solar are intermittent and what we really need is firm power

Jan Mar May Jul Sep Nov Jan

020

040

060

080

010

00

allMISO$datetime

min

ne_d

at.tp

i$G

loba

l

Hourly Interval

W/m

2

Fri Sat Sun Mon Tue Wed

020

040

060

080

010

00

allMISO$datetime[1:(24 * 5)]

min

ne_d

at.tp

i$G

loba

l[1:(2

4 * 5

)]Diurnal Variability + clouds

W/m

2

Jan Mar May Jul Sep Nov Jan

24

68

dd$dates

dd$t

imes

erie

s

Daily Interval

kWh/m2

Interday variability: mesoscale weather

Jan Mar May Jul Sep Nov Jan

8010

012

014

016

018

020

0

allMISO$datetime

pred

ict(l

oess

(unl

ist(t

s) ~

unl

ist(d

d), d

ata.

fram

e(m

m),

span

= 0

.5),

s

eq(1

, 12,

leng

th.o

ut =

leng

th(a

llMIS

O$d

atet

ime)

))

Monthly Interval

kWh/m2

Seasonal variability: oribital

Jan Mar May Jul Sep Nov Jan

010

2030

4050

60

allMISO$datetime

prod

uctio

n.ta

rget

.TS/

1000

Say we want to produce baseload with PV

GW

Jan Mar May Jul Sep Nov Jan

010

2030

4050

60

allMISO$datetime

prod

uctio

n.ta

rget

.TS/

1000

Possible on an energy basis: GW

∫      =   ∫

Jan Mar May Jul Sep Nov Jan

010

2030

4050

60

allMISO$datetime

prod

uctio

n.ta

rget

.TS/

1000

Possible on an energy basis: GW

∫      =   ∫

Fri Sat Sun Mon Tue Wed

010

2030

4050

60

allMISO$datetime[1:(24 * 5)]

prod

uctio

n.ta

rget

.TS[

1:(2

4 * 5

)]/10

00But impossible due to diurnal intermittency…

GW

Night ☾

Jan Mar May Jul Sep Nov Jan

010

2030

4050

60

allMISO$datetime

prod

uctio

n.ta

rget

.TS/

1000

Possible on an energy basis: GW

∫      =   ∫

Fri Sat Sun Mon Tue Wed

010

2030

4050

60

allMISO$datetime[1:(24 * 5)]

prod

uctio

n.ta

rget

.TS[

1:(2

4 * 5

)]/10

00Possible with storage…

GW

Shortfall: discharge☾

Surplus: charge ☀

Jan Mar May Jul Sep Nov Jan

020

0040

0060

0080

0010

000

1200

0

Time

Ener

gyBut how much?  Storage State of charge for one year to produce baseload w/PV

Charge in the summer☀

❄ Discharge in the winter

GWh

1 year

Fri Sat Sun Mon Tue Wed

5200

5300

5400

5500

5600

TimeEn

ergy

1 week

Seasonal trend > diurnal trend on an energy basis

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

1 year of storage SoC       1 week of dispatch

GW GWh

Oversizing Effect on Aggregate LCOE

oversizing factor

LCO

E (c

/kW

h)

1 x 1.2 x 1.7 x 2.5 x 5 x 100 x

01

23

45

67

PV Load

Storage

PV>Load?

Call on Storage

N

Storage Full ?

Y

Curtail

Y

NCharge

A

Simplified PV/Storage Dispatch Schema

Optimally oversized PVOptimally oversized PV

A B

Physical Data High resolution (spatial/temporal) time‐synchronous SolarAnywhere® Resource Data: T, Irradiance Firm, Guaranteed Production Profile (time‐synchronous load)  NCLD Landcover Tech. specs for PV, storage 

Development Data High/Low Technological Development Residential/Community/Commercial/Utility Dev.

• CapEx for PV, Storage ($/kW | $/kWh) • OpEx for PV, Storage ($/kWh | $/kWh)• WACC/discount rate• Spatial Allocation

Modeling Inputs

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

p: hourly w: 0 g: 0 s: mid.term_high_utility.led dr: 3

Curtailment %

PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, Low Technological Development in 2050, WACC of 3% single point.

PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, Low Technological Development in 2050, WACC of 3% single point.

MN LoadNormalized PVMN LoadNormalized PV

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

18.7 c/kWh18.7 c/kWh

Non‐official results

PV Spread more evenly (yet stillAvoiding sensitive landcover)PV Allocated just in Minneapolis

Geographic Dispersion Has an Effect on Cost

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

p: hourly w: 0 g: 0 s: mid.term_high_utility.led dr: 3

Curtailment %

PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, Low Technological Development in 2050, WACC of 3% distributed.

PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, Low Technological Development in 2050, WACC of 3% distributed.

12.6 c/kWh12.6 c/kWh

MN LoadNormalized PVMN LoadNormalized PV

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

Non‐official results

PV $/WBattery $/kWh

…As the degree of technological development 

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

p: hourly w: 0 g: 0 s: long.term_high_utility.led dr: 3

Curtailment %

PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

7.04 c/kWh7.04 c/kWh

MN LoadNormalized PVMN LoadNormalized PV

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

Non‐official results

PV Load

Storage

PV,wind>Load

Call on Storage

N

Storage Full

Y

Curtail

Y

NCharge

A

Simplified PV/Storage Dispatch Schema

Wind

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

: hourly w: 100 g: 0 s: long.term_high_utility.led dr: 3

Curtailment %

Wind + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

Wind + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

7.02 c/kWh7.02 c/kWh

MN LoadNormalized WindMN LoadNormalized Wind

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

Non‐official results

Costs can be reduced further by blending the wind and solar resources

…where anticorrelated

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

p: hourly w: 62 g: 0 s: long.term_high_utility.led dr: 3

Curtailment %

Optimal Wind/PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

Optimal Wind/PV + Storage Meeting 100% of Hourly Load in MN, Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

5.3 c/kWh5.3 

c/kWh

MN LoadNormalized RenewablesMN LoadNormalized Renewables

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

Non‐official results

Adding a small amount of conventional gas e.g. from stranded assets

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

p: hourly w: 62 g: 5 s: long.term_high_utility.led dr: 3

Curtailment %

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

Optimal Wind/PV + Storage Meeting 95% Hourly Load in MN, 5% met by gas Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

Optimal Wind/PV + Storage Meeting 95% Hourly Load in MN, 5% met by gas Utility‐scale‐led, High Technological Development in 2050, WACC of 3% 

3.6 c/kWh3.6 

c/kWh

MN LoadNormalized RenewablesMN LoadNormalized Renewables

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

Non‐official results

0

4

8

10

12

Current MN Load+ 5 million EVsw/load shifting

One Day 

EV Modeling in MN

DSM Can shift load to surplus hours, but unless V2G is only intraday

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5

p: hourly w: 62 g: 5 s: long.term_high_utility.led dr: 3

Curtailment %

Jun 30 Jul 02 Jul 04 Jul 06

050

0010

000

1500

0Hourly Production Target (zoom)

date

MN LoadProduction TargetNormalized Hybrid RE production

Optimal Wind/PV + Storage Meeting 95% Hourly Load in MN, 5% met by gas Utility‐scale‐led, High Technological Development in 2050, WACC of 3% +EV electrification & load‐shifting

Optimal Wind/PV + Storage Meeting 95% Hourly Load in MN, 5% met by gas Utility‐scale‐led, High Technological Development in 2050, WACC of 3% +EV electrification & load‐shifting

3.4 c/kWh3.4 

c/kWh

MN LoadNormalized RenewablesMN LoadNormalized Renewables

MWh

MWh

LCOE ($/kWh)

LCOE ($/kWh)

Oversizing factorOversizing factor

1x       1.2x       1.7x     2.5x       5x        100x1x       1.2x       1.7x     2.5x       5x        100x

Non‐official results

Key Takeaways

55% PV, 40% wind, 5% gas

3.6 c/kWh

• Supply shaping via oversizing + curtailment has significant value for minimizing cost at high penetration

• Major cost reduction by optimizing relative penetration of PV + wind• Major cost reduction by including minor amounts of gas backup at key 

times to minimize need for storage

Thanks!

Marc PerezClean Power Research

Inexpensive, Firm PV Without Conventional Backup:The Role of Supply Shaping Through Curtailment

This work is based upon work supported by:National Science Foundation GRF Grant No. DGE 1144155

DoE Sunshot Grant No. DE‐EE0007669Columbia University Center for Life Cycle Analysis

Clean Power Research

For achieving high‐penetration renewables!