RESOLVE ModelingOverview

CPUC IRP WorkshopDecember 16, 2016

Nick Schlag, Sr. Managing ConsultantArne Olson, Partner

Jimmy Nelson, Consultant

2

Agenda

RESOLVE model background & overview

Key constraints impacting portfolio development

Summary of model inputs

Examples of model outputs

RESOLVE BACKGROUND

4

Defining the New PlanningProblem

Capacity planning paradigm hasshifted with the increasingquantities of variable renewableresources

• RPS targets

• GHG reduction goals

The new planning problemconsists of two relatedquestions:

1. How many MW of dispatchableresources are needed to(a) meet load, and (b) meet flexibilityrequirements on various time scales?

2. What is the optimal mix of newresources, given the characteristics ofthe existing fleet of conventional andrenewable resources?

5

The Renewable IntegrationChallenge

5

Many studies have highlighted centralrole of renewables in decarbonizingelectricity sector

Primary drivers of renewableintegration challenges at highpenetrations:

• Renewable oversupply during low load periods

• Inflexible conventional generation

• Must-run resources

• Technical constraints on ramping, minimumstable levels, minimum up and down times

• High costs associated with cycling

• Small balancing areas or constrainedinteractions with neighboring regions

Research has shifted to focus on gridintegration solutions

6

Optimal Solution Balances Non-Renewable Solutions with Overbuild

6

In reality, theintegrated

resource planningquestion hasmany more

dimensions thanshown here

7

RESOLVE Co-optimizes Investmentand Operational Decisions

RESOLVE is a linear programallows portfolio optimizationacross a long time horizon (10-20years)

Fixed costs capture capital,financing, and fixed O&Massociated with new physicalinfrastructure

Operational detail focuses onprimary drivers of renewableintegration challenges

RESOLVE may select portfoliofrom a variety of potential“solutions,” including:

• Renewable overbuild

• Energy storage

• Advanced demand response

• Conventional gas generation

• Gas retrofits

RESOLVE ObjectiveFunction

RESOLVE ObjectiveFunction

Fixed Costs of New Resources• Renewables• Energy storage• Demand response• Thermal

Fixed Costs of New Transmission

Total System Operating Costs• Variable O&M• Start costs• Fuel costs• Carbon

8

RESOLVE minimizes the NPV of total costs across a 20+year time horizon

• Additional weight applied to last year of analysis to account for end effects

• Because of computational complexity, RESOLVE is typically not used to modelall years in analysis horizon

Example Model Time Horizon

20202018 2022 2024 2026 2028 2030 2032 2034 2036 2038

In each modeled year, the portfolio is explicitlymodeled, and total cost is calculated as the sum offixed costs of investment and operating costs

In intermediate years, the total cost of the portfoliois calculated by linear interpolation between the twoadjacent modeled years

Decisions made withinone year carry forward to

subsequent years

Example of ‘modeledyears’ shown above is

shown to illustrateRESOLVE functionality but

does not necessarilyrepresent the final setthat will be used in IRP

9

RESOLVE & the IRP ScenarioDevelopment Process

Each candidate plan, future, and sensitivityrepresents a lever (or combination of levers) thatcan be adjusted in RESOLVE to capture theintention of the scenario

In each sensitivitycase, a single fixed

input to RESOLVEis varied to

evaluate its impacton the objectivefunction and the

least-cost portfolio

For each alternative candidate plan, RESOLVEassumes specific resources are added to theportfolio for evaluation against the ‘Base’ Plan

KEY CONSTRAINTS INPORTFOLIODEVELOPMENT

11

Key Constraints in RESOLVE

Renewables portfolio standard

GHG planning target

System planning reserve margin

Local capacity deficiencies

Resource technical potentials

System operating requirements

Generator operating characteristics

Required portfolioattributes &characteristics

Limits on availableresources

Constraints onsystem operations

12

Renewables Portfolio Standard

RESOLVE selects new resources to meet arenewable net short in each year

RPS constraint is based on delivered renewableenergy, so renewable portfolio is “overbuilt” tooffset for potential generation lost to curtailment

Renewable net shortselected by RESOLVE

13

Greenhouse Gas Planning Target

RESOLVE allows specification of a GHG planning target,which constrains the portfolio on an annual basis

No GHG credit is given for exports from CAISO(consistent with ARB inventory accounting)

GHG constraint, when binding, may result in renewableportfolios that exceed the statutory RPS target

≤GHG PlanningTarget[tons CO2]

GHG PlanningTarget[tons CO2]

PhysicalEmissions Rate

[tons/MMBtu]

PhysicalEmissions Rate

[tons/MMBtu]

CAISO Fuel Burn[MMBtu]

CAISO Fuel Burn[MMBtu]

GasCCGTGas

CCGT

Gas CTGas CT

Gas ICEGas ICE

GasCHPGasCHP

DeemedEmissions Rate

[tons/MWh]

DeemedEmissions Rate

[tons/MWh]

UnspecifiedImports

[MWh]

UnspecifiedImports

[MWh]

GHG planning targetreflects the collectiverequired emissions

targets for the LSEs ofCAISO

GHG planning targetreflects the collectiverequired emissions

targets for the LSEs ofCAISO

14

System Planning Reserve MarginConstraint

In each year modeled, RESOLVE imposes a planning reservemargin constraint on the total CAISO generation fleet

Contribution of each resource to PRM requirement depends onits attributes

≤PRM Requirement1-in-2 peak x 115%

PRM Requirement1-in-2 peak x 115%

Based on NQC list

Calculated in RESOLVEvia ELCC surface

Planning assumption

Based on forecast 1-in-2 peak load impact

Function of capacity &duration

Available CapacityAvailable Capacity

Thermal NQCThermal NQC

Hydro NQCHydro NQC

Renewable ELCCRenewable ELCC

ImportsImports

Demand ResponseDemand Response

StorageStorage

PRM constraint designedto ensure that sufficientgeneration capability isavailable to meet load

during system peakconditions; constraint is

unlikely to be bindingexcept in cases thatassume substantial

retirements of existingfleet

PRM constraint designedto ensure that sufficientgeneration capability isavailable to meet load

during system peakconditions; constraint is

unlikely to be bindingexcept in cases thatassume substantial

retirements of existingfleet

15

Local Capacity DeficiencyConstraint

The constraint on local capacity requires the totalamount of new capacity associated with local areasto exceed the sum of all local deficiencies

The addition of local RA constraints providesadditional location-specific value for candidateresources that can be used to meet local needs

≤Local Deficiency[MW]

Local Deficiency[MW]

Assumed NQC[%]

Assumed NQC[%]

New Resources[MW]

New Resources[MW]

Gas CCGT/CTGas CCGT/CT

StorageStorage

Distributed PVDistributed PV

Demand ResponseDemand Response

Local RA constraintdesigned to ensure that

sufficient generationcapability is available tomeet load in local areas

during peak periods

Local RA constraintdesigned to ensure that

sufficient generationcapability is available tomeet load in local areas

during peak periods

16

Renewable Resource TechnicalPotential

Northern CaliforniaLassen North, Round Mountain,Sacramento River

SolanoCentral Valley North & Los Banos

WestlandsGreater CarrizoCarrizo North, Carrizo South,Cuyama, Santa Barbara

Greater ImperialImperial East, Imperial North, Imperial South,San Diego South, San Diego North Central

Mountain Pass & El Dorado

Riverside East & Palm Springs

Southern California DesertIron Mountain, Pisgah, TwentyninePalms, San Bernandino - BakerTehachapi

Kramer & InyokernBarstrow, Kramer, San Bernandino – Lucerne,Victorville, Inyokern

Example renewable resource andtransmission development zones inRESOLVE model built for CAISO• Renewable-driven transmission build

solved for within each zone• Transmission costs factor into optimal

resource selection

17

System Operating Requirements

System operating costsincluded in objective functionusing a linear (LP) productioncost model

• Zonal representation of WECCregion with transmissionconstraints

Additional operationalrequirements are imposed toreflect CAISO operations:

• Spinning reserves

• Load following reserves

• Regulation reserves

• Frequency response

Captures operational impacts ofrenewable integration challenges

Renewables

Gross Load

Net Load

18

Plant Operational Constraints

Hourly operations is constrained by many factors:

Hydro constraints:• Daily energy budget• Daily Pmax• Daily Pmin• Hourly ramping

Hydro constraints:• Daily energy budget• Daily Pmax• Daily Pmin• Hourly ramping

Gas generator constraints:• Pmax• Pmin• Min up/down time• Max hourly ramp

Gas generator constraints:• Pmax• Pmin• Min up/down time• Max hourly ramp

Transmission constraints:• Minimum/maximum flowTransmission constraints:• Minimum/maximum flow

Renewable constraints:• Hourly availabilityRenewable constraints:• Hourly availability

Storage constraints:• Pmax• Pmin• Energy neutrality

Storage constraints:• Pmax• Pmin• Energy neutrality

SUMMARY OF KEY INPUTS

20

Key Model Inputs

RESOLVE requiresdetailed inputs on boththe demand side (loadforecasts, loadmodifiers, candidateresources) and supplyside (existingresources, candidateresources)

21

(5) Time-of-use rates (+/-) (6) Final retail load

(4) Electric Vehicles (+)

(2) Energy Efficiency (-)

(3) Behind-the-Meter PV (-)

(1) Consumption

Load Forecast by Component

Load forecastincorporates multipledemand-sideadjustments:

• Energy efficiency

• Behind-the-meter PV

• Electric vehicles

• Time-of-use rates

Each adjustment ismodeled with anindependent profile,allowing RESOLVE tocapture changes inthe load shapethrough time

Primary data source:CEC IEPR DemandForecast

Shape from (1)

Resultingshape

Shape from (2)

Resultingshape

Shape from (3)

Resulting shape

Shape from (4)

Resultingshape

Shape from (1)

Final shape

22

Load shapes for CAISO and other WECC BAs based on2007-2009 historical period

Renewable shapes derived from NREL’s latest wind andsolar data sets:

NREL Wind Prospector (link)

• 126,000 sites• 5-min temporal resolution• 2007-2013 historical period

Load and Renewable Profiles

NREL Solar Prospector (link)

• 120,000 sites• 1-min temporal resolution• 2007-2013 historical period

23

Existing & Planned CAISO Renewable Portfolio

Primary sources:

• CAISO conventional generators: CPUC NQC list

• Non-CAISO generators: TEPPC 2026 Common Case

• CAISO existing renewables: CPUC IOU Contract Database

Existing & Planned CAISO Conventional Fleet

Existing Generation Resources

24

Resource Cost & Potential

For each candidate resource, RESOLVE requires inputassumptions to specify:

• Technical potential (MW): total available resource that mat be selected

• Fixed costs ($/kW-yr): annualized cost of investment + ongoingmaintenance

• Operating characteristics: e.g. hourly profiles for variable resources;operational constraints & variable costs for thermal & storage resources

Primary sources:• Renewables: RPS Calculator Cost & Potential Assessment (Black & Veatch)

• Gas generation: California Cost of Generation (CEC)

• Advanced DR: 2015 California Demand Response Potential Study (LBNL)

• Storage: market research (E3)

SUMMARY OF RESOLVEOUTPUTS

26

Summary of RESOLVE Outputs

RESOLVE produces a variety of outputs that mayinform planning decisions

• Incremental resource portfolio (MW)

• Fixed costs of new investments ($)

• System-wide operational cost ($)

• Renewable curtailment (GWh)

• CAISO GHG emissions (MMTCO2e)

• CAISO fuel consumption (MMBtu)

• System-wide GHG emissions (MMTCO2e)

• System-wide fuel consumption (MMBtu)

• Shadow prices of key constraints ($ per unit)

27

- Scenario 1a Scenario 2 Scenario 3CAISO simultaneous export limit 2,000 8,000 8,000Procurement Current practice Current practice WECC-wideOperations CAISO WECC-wide WECC-widePortfolio Composition (MW)California Solar 7,601 7,804 3,440California Wind 3,000 1,900 1,900California Geothermal 500 500 500Northwest Wind, Existing Transmission 1,447 562 318Northwest Wind RECs 1,000 1,000 0Utah Wind, Existing Transmission 604 604 420Wyoming Wind, Existing Transmission 500 500 500Wyoming Wind, New Transmission 0 0 1,995Southwest Solar, Existing Transmission 0 500 500Southwest Solar RECs 1,000 1,000 1,000New Mexico Wind, Existing Transmission 1,000 1,000 1,000New Mexico Wind, New Transmission 0 0 1,962Total CA Resources 11,101 10,204 5,840Total Out-of-State Resources 5,551 5,166 7,694Total Renewable Resources 16,652 15,370 13,534

Energy Storage (MW) 972 500 500

Example Outputs: Resource BuildCAISO SB350 Regionalization Study

Scenarios shown hereanalogous to “candidate

plans” in IRP

Scenarios shown hereanalogous to “candidate

plans” in IRP

Optimalresource mix

adjusts toreflect changesin economicsof operations

(S1a – S2) andincreased

availability ofhigh qualityrenewables(S2 – S3)

28

Example Outputs: Shadow PricesCAISO SB350 Regionalization Study

RESOLVE can produceshadow prices on keymodel constraints toinform planningdecisions

Marginal cost of RPScompliance varies byscenario

• Reflects the marginal costof procuring an additionalMWh of renewablegeneration

• Also represents themarginal cost ofrenewable curtailment toratepayers

-

5

10

15

20

25

30

35

40

45

2016 2020 2025 2030

$/M

Wh

Year

Marginal RPS Compliance Cost

Scenario 1a

Scenario 2

Scenario 3

$40

$20

$5

29

Example Outputs: SensitivitiesCAISO SB350 Regionalization Study

Fixed Costs ($MM) Scenario 1a Scenario 2 Scenario 3Base assumptions $3,292 $2,612 $2,492A. High coordination under bilateral markets $3,003 $2,612 $2,492B. High energy efficiency $2,790 $2,214 $2,098C. High flexible loads $3,138 $2,643 $2,522D. Low portfolio diversity $3,196 $2,301 $2,192E. High rooftop PV $3,256 $2,418 $2,312F. High out-of-state resource availability $3,104 $2,526 $2,443G. Low cost solar $3,137 $2,627 $2,490H. 55% RPS $4,385 $3,221 $3,044

Annual savings from regional integration range from$391 million to $1 billion per year under 50% RPS

• High flexible loads and high energy efficiency reduce savings

• Low Portfolio diversity, high rooftop PV, and higher RPS increase savings

• High out-of-state availability has limited effect on savings

Sensitivities shownhere analogous to“sensitivities” (or

futures) in IRP

Sensitivities shownhere analogous to“sensitivities” (or

futures) in IRP

Scenarios shownhere analogous to

“candidateplans” in IRP

Scenarios shownhere analogous to

“candidateplans” in IRP

Thank You!

Energy and Environmental Economics, Inc. (E3)101 Montgomery Street, Suite 1600San Francisco, CA 94104Tel 415-391-5100http://www.ethree.com