25
Modeling EERE Deployment Modeling EERE Deployment Programs Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

Embed Size (px)

Citation preview

Page 1: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

Modeling EERE Deployment Modeling EERE Deployment ProgramsPrograms

Modeling EERE Deployment Modeling EERE Deployment ProgramsPrograms

Donna HostickDave Belzer

Pacific Northwest National LaboratoryNovember 29, 2007

PNNL-SA-58293

Page 2: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

2

ProblemProblemProblemProblem

EERE deployment programs contribute to overall program energy saving benefits, but are difficult to model in terms of the traditional cost and performance parametersWhen programs are modeled within GPRA (PDS) framework, the approaches vary widely – estimates may be inconsistent from one program to another

PNNL-SA-58293

Page 3: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

3

Purpose of StudyPurpose of StudyPurpose of StudyPurpose of Study

First phase of PAE effort to improve deployment modeling Identify and characterize modeling of EERE deployment

programs Address possible improvements to modeling process Note gaps in knowledge

PNNL-SA-58293

Page 4: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

4

Defining DeploymentDefining DeploymentDefining DeploymentDefining Deployment

Includes:

Addressing market barriers and consumer behavior

Currently available technologies

Preparing the market for future technologies

Demonstrations replicated as showcases

Does not include:

Research

Development

First-of-a-kind or scale-up demonstrations

“Activities that promote the adoption of advanced energy efficiency and renewable

energy technologies and practices.”EERE Deployment Inventory 2004

PNNL-SA-58293

Page 5: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

5

RD3 Activities over RD3 Activities over Development TimelineDevelopment Timeline

RD3 Activities over RD3 Activities over Development TimelineDevelopment Timeline

TIME

“R&D” Phase“Deployment”

Phase

“Demonstration”Phase

R&D

R&D

R&D

Demo

Demo

Demo

Deploy

Deploy

Deploy

TIME

“R&D” Phase“Deployment”

Phase

“Demonstration”Phase

R&D

R&D

R&D

Demo

Demo

Demo

Deploy

Deploy

Deploy

PNNL-SA-58293

Page 6: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

6

FY08 Request by Primary FocusFY08 Request by Primary FocusFY08 Request by Primary FocusFY08 Request by Primary Focus

$0.0

$100.0

$200.0

$300.0

$400.0

$500.0

$600.0

$700.0

$800.0

$900.0

R&D Deployment

FY

08

Re

qu

es

t ($

Mill

ion

) WIP

FEMP

Industrial Technologies

Building Technologies

Freedom Car/Vehicles

Wind

Solar Energy

Biomass

Hydrogen

PNNL-SA-58293

Page 7: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

7

FY08 EERE Deployment ActivitiesFY08 EERE Deployment ActivitiesFY08 EERE Deployment ActivitiesFY08 EERE Deployment Activities

EERE Program Deployment Activity

Hydrogen

Hydrogen Education and Outreach ($3.9 million)

Codes and Standards Program ($16 million, primarily code research)

Technology Validation ($30 million)

Distributed Energy Fuel Cell Systems ($7.7 million)

Biomass

Integration of BioRefinery Technologies (supports Biofuels Initiative) ($92.103 million)

Biomass Products Development ($10 million)

Bioconversion Platform R&D (R&D phase with marketing analysis) ($38.3 million total; only a small percentage is for deployment)

Thermochemical Platform R&D (R&D phase with marketing analysis) ($19.537 million total; only a small percentage is for deployment)

Regional Feedstock Partners Outreach, within Feedstock Infrastructure ($9.737 million total; only a small percentage is for deployment)

Education and Outreach (not called out separately within budget request)

Solar Energy

Solar America Initiative (includes Technology Pathway Partnerships, Technology Acceptance, and Technology Evaluation) ($16.34 million is primarily deployment-focused)

Solar Decathlon (not called out separately within budget request)

WindWind Resource Assessment (Systems Integration and Technology Acceptance) ($12.869 million is primarily deployment-focused)

Wind Powering America (not called out separately within budget request)

FreedomCAR/Vehicles

Clean Cities (now Vehicle Technologies Deployment, $9.593 million)

Graduate Automotive Technology Education (GATE) ($0.5 million)

Advanced Vehicle Competitions ($1.3 million)

Legislative and Rulemaking activities ($1.8 million)

PNNL-SA-58293

Page 8: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

8

FY08 EERE Deployment Activities, cont.FY08 EERE Deployment Activities, cont.FY08 EERE Deployment Activities, cont.FY08 EERE Deployment Activities, cont.

EERE Program Deployment Activity

Buildings

Residential R&D: Building America (some deployment components) and the National Builders Challenge Initiative ($19.7 million total; only a small percentage is for deployment)

Commercial Buildings R&D (some deployment components) ($7 million total; only a small percentage is for deployment)

Equipment Standards and Analysis ($13.361 million)

Energy Star ® ($6.776 million)

Rebuild America: Building Application Centers ($2.834 million total for Rebuild America)

Rebuild America: EnergySmart Schools and EnergySmart Hospitals ($2.834 million total for Rebuild America)

Building Energy Codes and Advanced Energy Codes Initiative ($3.751 million, primarily code deployment)

Rebuild America: Commercial Lighting Challenge ($2.834 million total for Rebuild America)

IndustrialBest Practices/Save Energy Now Program ($8.833 million)

Industrial Assessment Center ($4.035 million)

FEMP

Technical Guidance and Assistance ($6.519 million)

Project Financing (ESPC Support, UESC Support) ($7.935 million)

DOE Specific Investments (new initiative, not included in FY08 request)

Federal Fleet (part of FEMP’s $2.337 million Planning, Reporting and Evaluation activity)

WIP

Weatherization Assistance Program ($144 million)

State Energy Program ($45.501 million)

Tribal Energy Program ($2.957 million)

Asia Pacific Partnership ($7.5 million)

PNNL-SA-58293

Page 9: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

9

EERE Deployment CategoriesEERE Deployment CategoriesEERE Deployment CategoriesEERE Deployment Categories

General information dissemination activities Targeted training and workshopsPartnerships with others to solve technical and administrative issuesRecognition for key products and awards for products, institutions and/or individualsSponsoring and promoting competitions to solve specific deployment issuesPurchasing enabling technologies and programsDeveloping and implementing standards and regulationsProviding technical assistance to “early adopters”Providing privileges and incentivesDemonstrations of key technologies, systems, and designs.

PNNL-SA-58293

Page 10: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

10

Target Audience and SectorTarget Audience and SectorTarget Audience and SectorTarget Audience and Sector

Knowledge community

Business Public Infrastructure

End-UserEnergy/ Fossil Fuels

Deployment activities target one or more groups. . .

Transportation Sector

PNNL-SA-58293

Page 11: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

11

Taxonomy of DeploymentTaxonomy of DeploymentTaxonomy of DeploymentTaxonomy of Deployment

Stage-Avenue Data Gathering/Market Research Advanced Market Preparation and Infrastructure

Development Identifying Promising Technologies Public Infrastructure and Policy, Regulation Manufacturing and Business Infrastructure Technology Adoption Supports Marketing and Outreach

Each activity has a target sector

Part of R&D Process

PNNL-SA-58293

Page 12: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

12

How do Deployment StrategiesHow do Deployment Strategies Save Energy? Save Energy?

How do Deployment StrategiesHow do Deployment Strategies Save Energy? Save Energy?

Reduce costs of energy-saving technologies and designs Explicit costs Implicit costs

Reduce risk associated with adopting new energy-saving or renewable technologies, designs, and strategiesReduce time to market entry of technologiesModify consumer behavior

PNNL-SA-58293

Page 13: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

13

GPRA (PDS) Modeling FrameworkGPRA (PDS) Modeling FrameworkGPRA (PDS) Modeling FrameworkGPRA (PDS) Modeling Framework

Modified versions of NEMS and MARKAL provide mid-term and long-term benefits estimates Detailed technology representations of electricity

markets, most residential and commercial end uses, and vehicle choice

Program cases represent adjustments to technology characterizations, cost, and performance parameters expected to result from program activities

Most deployment activities modeled “off-line”

PNNL-SA-58293

Page 14: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

14

Characterizing Current EERE Characterizing Current EERE Deployment ModelingDeployment Modeling

Characterizing Current EERE Characterizing Current EERE Deployment ModelingDeployment Modeling

Modeling R&D and Deployment Jointly Hydrogen, Fuel Cells, and Infrastructure Technologies Biomass Technologies FreedomCAR and Vehicle Technologies Solar Energy Technologies Wind Technologies

Modeling Deployment Activities within the NEMS Framework Building Technologies (Energy Star Appliances)

Off-Line (Non-Integrated) Modeling Approaches Building Technologies Industrial Technologies Federal Energy Management Program (FEMP) Weatherization and Intergovernmental Program

PNNL-SA-58293

Page 15: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

15

Modeling Deployment Activities Modeling Deployment Activities within the NEMS Frameworkwithin the NEMS Framework

Modeling Deployment Activities Modeling Deployment Activities within the NEMS Frameworkwithin the NEMS Framework

General Approach Alter parameters related to consumer or business decision making Reduce “ancillary costs” associated with technology adoption

Consumer Decision Making Modify discount rates or “time preference premiums” (residential

and commercial modules) Modify parameters associated with “riskiness” of new technologies

(vehicle choice in response to FreedomCAR activities)Business Decision Making (Renewable Energy Suppliers and Investors) Modify risk premium component in cost of capital (Biomass, Wind) Risk premium sometimes represented as “beta” coefficient in

Capital Asset Pricing ModelAncillary Costs Interconnection costs Environmental studies and permitting

PNNL-SA-58293

Page 16: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

16

Example: Energy Star AppliancesExample: Energy Star AppliancesExample: Energy Star AppliancesExample: Energy Star Appliances

Logit choice algorithms in NEMS residential model Separate parameters on appliance cost and annual energy cost Ratio of parameters is roughly equal to the (average) discount rate

Modeling Energy Star for GPRA One of the logit parameters is adjusted to effectively lower discount

rate – reflects informational aspect of Energy Star labels Adjustment made to increase market penetration of Energy Star

products to meet program goals (currently, part of baseline)Key issue: NEMS framework is suitable for representing program activities, but not predicting outcomesEnergy Star has performed variety of assessment studies, but none indicate impact on parameters associated with consumer decision making

PNNL-SA-58293

Page 17: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

17

Example: Biomass Fuels Example: Biomass Fuels Example: Biomass Fuels Example: Biomass Fuels

Equity premium “beta” coefficients in NEMS Renewable Fuels Module can be used alter cost of capital for future cellulosic ethanol plants The risk of an average investment (i.e., broad portfolio of common stocks) is

multiplied by beta and then added to “risk-free” rate = cost of capital Corn-based ethanol plants (beta = 1.5), cellulosic ethanol (beta = 1.75) Unlikely the actual betas would be so similar

Recent work by NREL and On-Location for FY2009 GPRA Biomass Scenario Model used to characterize risk premium for different

classes of investors Blended Risk Premium used in NEMS model Risk premium declines on basis of increases in productive capacity –

presumably based upon Biomass Scenario Model

Issue: What empirical basis is there to establish appropriate risk premium and how much should it adjust as new plants are built?

PNNL-SA-58293

Page 18: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

18

Key Elements of Sebold-Fields Key Elements of Sebold-Fields Dynamic Adoption ModelDynamic Adoption Model

Key Elements of Sebold-Fields Key Elements of Sebold-Fields Dynamic Adoption ModelDynamic Adoption Model

A Framework for Planning and Assessing Publicly Funded Energy Efficiency Programs (California PGC, 2001)Adoption Process Model Market share = Awareness * Willingness * Availability

Awareness has dynamic elements: Awareness = (a0 + a1 INT ) x (1 – Awareness[t-1])

+ (a3 + a4 INT ) x Awareness [t-1]

Willingness has similar dynamic function; alternatively, Willingness can be described as function of PaybackPayback is function of intervention: Payback = c0 + c1 Payback [t-1] + c2 * INT

PNNL-SA-58293

Page 19: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

19

Newell-Anderson Study of IAC ProgramNewell-Anderson Study of IAC Program

Newell-Anderson Study of IAC ProgramNewell-Anderson Study of IAC Program

2002 study analyzed audit data from Industrial Assessment Center Program 9,000 assessments from period 1981-2000 Measure cost and estimated energy savings in database

Newell and Anderson hold out promise that study can quantify impact of information on discount ratesEmpirical results indicate very short payback periods required to undertake efficiency measures (typically 1.25 – 1.5 year payback)Conclusion is that program did not appreciably affect discount rates Discount rates generally in accordance with other studies Provides useful distribution of discount rates

Key points: Casts doubt on modeling approaches that significantly lower discount rates

as response to information programs Program success is measured by number of firms made aware of cost-

effective conservation options ($100 million in annual energy savings) Estimates of program influence on decision making would have required

careful program design with control group and pre- and post-participation interviews

PNNL-SA-58293

Page 20: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

20

Ordered Logit Techniques for Ordered Logit Techniques for Estimating MT InterventionsEstimating MT Interventions

Ordered Logit Techniques for Ordered Logit Techniques for Estimating MT InterventionsEstimating MT Interventions

Econometric technique to estimate market shares ACEEE paper in 2004 summer study – Skumatz, Weitzel Works with stated preferences, not revealed preferences

Develop alternative option sets Technical and cost (size, efficiency, system cost) Factors influenced by deployment activities (i.e., rebates) Reliability (warranty, experience in the field)

Construct sample of potential adopters, (HVAC installers, 200 50 in sample)Respondents order option sets (on cards) Option sets described by characteristics – not by name In particular study, name of efficient technology was disclosed at

end to reveal bias

PNNL-SA-58293

Page 21: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

21

Modeling Choices of SteamModeling Choices of Steam Generation Technologies in CIMS Generation Technologies in CIMS

Modeling Choices of SteamModeling Choices of Steam Generation Technologies in CIMS Generation Technologies in CIMS

CIMS Model of Canadian economy – Marc Jacquard (UBC) Energy Journal – January 2005Survey of nearly 600 industrial firms (260 in final sample)Three types of steam generation: Conventional boiler High efficiency boiler Cogeneration system

Stated preference approach yields flexibility for policy analysisMultinomial logit model estimated from responses Intangible cost (constant term) is a key output l Intangible cost is highest for cogeneration, lowest for high efficiency Interpretation is that cogeneration brings safety and reliability issues

Modeling information by segmenting market (“well-informed” or not)

PNNL-SA-58293

Page 22: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

22

Key Modeling Aspects –Key Modeling Aspects – Demand Sectors Demand Sectors

Key Modeling Aspects –Key Modeling Aspects – Demand Sectors Demand Sectors

Basic question: what particular market barrier is being addressed1) Lack of awareness (i.e., not “well informed” ) (new technology)

Market segmentation in NEMS or MARKAL (Can use existing choice framework) Ongoing surveys to track awareness – link to EERE activity

2) Consumers not familiar with trade-off between first cost and operating costs General information programs would affect average discount rate (or distribution of

discount rates Ongoing surveys to track consumer sensitivity – issue: specific link to EERE activity

3) Risk perceptions, high costs of gathering information for specific technologies, ancillary implementation costs

Use terms in logit (or similar specifications) to adjust implicit or intangible cost (as is now done for vehicle choice)

Tracking impact of EERE deployment activities would require periodic studies to quantify these factors

Stated preference studies could disentangle effects from tangible and intangible costs

PNNL-SA-58293

Page 23: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

23

Retrospective Analysis: Retrospective Analysis: CFL Sales in the Northwest CFL Sales in the Northwest

Retrospective Analysis: Retrospective Analysis: CFL Sales in the Northwest CFL Sales in the Northwest

Addressing uncertainty in evaluation of MT (ACEEE 2004) Stratus Consulting, Summit Blue Consulting Energy Star program Activities of NW Energy Efficiency Alliance

2001 CFL sales increased by 7.7 million units in NW 3.6 million from rebates, giveaways Of remaining 4.1 million, how many due to Alliance?

Interviewed retailers, utility program managers (31) Alliance totally responsible – set up infrastructure… OR Alliance had minor influence – CFL sales up sharply elsewhere

Alternative scenarios “High influence” (4.1 million), and “low influence” (2 million) Used @Risk to characterize other uncertainty

PNNL-SA-58293

Page 24: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

24

Modeling Risks of RenewableModeling Risks of Renewable Energy Investments Energy Investments

Modeling Risks of RenewableModeling Risks of Renewable Energy Investments Energy Investments

A very few publicly available studies have characterized the risk premium As with discount rates, no study has yet been found to directly link to governmental intervention jor European study in 2004 suggests initial approach EC-funded study surveyed 650 stakeholders – representatives from

Utilities Project developers Investors Banks Manufacturers Government

Survey augmented by in-depth interviewsEC study developed ranges of risk premiums for different types of renewable projectsOther aspect is to assess deployment impact on ancillary (implementation) cost

PNNL-SA-58293

Page 25: Modeling EERE Deployment Programs Donna Hostick Dave Belzer Pacific Northwest National Laboratory November 29, 2007 PNNL-SA-58293

25

Key Issues/Questions Re: Modeling Key Issues/Questions Re: Modeling Key Issues/Questions Re: Modeling Key Issues/Questions Re: Modeling

Energy Models include technology cost and choice behavioral parameters Few (no?) empirical studies as how interventions might change

behavioral parameters – typically used to represent deployment Program evaluation studies typically focus on market outcomes,

rather than characterizing behavioral parameters

This study suggests some approaches for gathering empirical data to estimate deployment impacts To what degree does a better understanding of past deployment

activities serve to inform probable effects from future activity? How should effort should be prioritized for analysis of activities that

do not fit into NEMS framework? Representation or Prediction?

PNNL-SA-58293