A Comprehensive / Integrated DR Value Framework Presented at: Demand Response Research Center TAG Technical Advisory Group Meeting Pacific Energy Center

A Comprehensive / IntegratedDR Value Framework

Presented at:Demand Response Research Center TAG Technical Advisory Group Meeting

Pacific Energy CenterSan Francisco, California

January 31, 2006

By:Daniel M. Violette, Ph.D.Summit Blue ConsultingBoulder, [email protected]

Summit Blue Consulting 2

Objectives

1. Develop a “more comprehensive DR conceptual valuation framework.”

2. Develop a method capable of addressing different stakeholder perspectives.

3. Other considerations:

– Linking energy efficiency, frequent time-of-use, and less frequent DR response.

– Appropriately defining DR.

– Purpose of DR evaluation method(s).

– Customer bill management opportunities.

– Risk and opportunity costs of investments in DR.

– What methodologies best support DR valuation and integration into current resource plans.


Definition of DR

1. Conventional View – A resource for extreme events.

– DRRC original definition:

“Actions taken to reduce load when contingencies (emergencies and congestion) occur that threaten the supply-demand balance, and/or market conditions occur that raise supply costs.”

DR typically involves peak-load reductions and strategies that differ from energy efficiency in that they represent transient versus permanent changes in peak demands.

A customer DR response is typically associated with a customer notification.


Definition of DR (cont.)

2. Evolving Definition -- a working definition for DR (U.S. DOE):

Reductions in electric usage by end-use customers from their normal consumption patterns at times of high wholesale market prices or when system reliability is jeopardized in response to changes in the price paid for electricity or to incentives designed specifically to induce reduction.

– Two components:

1) reductions in energy use at “critical times,” and

2) the method by which DR is elicited from customers.

– Reconciling these two concepts is important for characterizing the available DR as well as valuing DR – the unique attributes of DR give it its value and also its limitations.

– Issue -- Change in “normal consumption” could rule out certain types of pricing programs, i.e., they always respond to high prices in a certain way.


Definition of DR (cont.)

3. More expansive definition:

The definition of DR can move from one focused on critical system and market events to alternatives that influence electricity demands for almost all hours, with impacts on market efficiency and resource allocation.

– This broader definition is consistent with the two prior views – it simply extends the definition to hours that may not meet the definition of a “critical event.”

– A comprehensive conceptual definition valuation framework should consider this more expansive definition.

– The view of demand response as a substitute for supply should shift to also emphasize its role as a customer cost management resource. (From EPRI, EP-6035).

– Integrate DR into the fabric of the electricity market using price signals (ISO-NE, 2005)


Definition of DR: Implications for the Framework

• Framework should fit both the conventional event-based view and the more expansive view of customer-driven usage decisions.

• Each DR type will have benefits and costs – a comprehensive approach should provide insights into the merits of each:

1) a event-based reliability approach to DR (conventional view); and,

2) a broader view focused on market efficiency and decentralized price-based customer response.

♦ Consumers who make individual decisions to shift or reduce demand without direct communication with the system operator.

♦ A non-event based real-time pricing program would fall into this category.

♦ Customers would respond to real time prices (day-ahead or real-time market prices) every day, not just on days deemed as “event days.”


Several Perspectives of DR Value

1. Event-Based Value of DR:– Value of DR in mitigating a single extreme reliability event (often uses a

value of reliability approach)

– Restoration of power after a blackout (August blackout in NY)

2. Benefit-Cost Screening assessment of a DR program:– Often focuses on one-at-a-time program assessments using “standard

practice tests” similar to those used by EE programs for cost-effectiveness.

– The demand-side and supply-side are linked by avoided supply costs, but the analyses are generally static in nature.

3. Evaluation – Retrospective assessment of DR over a past time period -- often used to assess pilot programs or in regulatory filings.

4. DR in a Resource Portfolio – Examine portfolios of DR as part of a resource plan.


Analysis Flow – DR Assessment

1. Resource Planning and Portfolio ModelingIncremental System Costs, Reliability and Risk Metrics(Long term assessment that provides inputs into other

analyses)

2. EvaluationEvidence that DR is Producing the Benefits Expected?

3. Benefit-Cost ScreeningUse of Benefit-Cost Tests For Program Assessment

Augmented for Peak Events Conditions

4. Event Based DR ValuationFor Major Select Events, e.g., A Substantial Outage.

These Analyses Can Confirm the Risk Mitigation Value of DR


Practical ChallengesC1. Many different types of DR with each producing different types of

benefits – All the DR possibilities can not be assessed.

C2. Characterizing the cost and performance of DR programs/options.

C3. Uncertainty must be dimensioned if the hedge benefits, insurance aspects, and risk metrics of DR portfolios are to be developed.

C4. The planning horizon must be long enough to address low-probability, high-consequence events, i.e., 15 to 20 years.

C5. Addressing the locational value of DR – particularly for T&D.

C6. Estimating the customer-side benefits of DR.

C7. Many values associated with DR are difficult to quantify.

C8. Different levels of detail are needed for different types of DR assessments.


Table 3‑1: Potential DR Benefits

1. Direct Financial

DF1. Incentive payments to participating customer.

DF2. Bill reductions from customer load usage reductions or shifts in use.

DF3. Incentive payments to DR implementers (e.g., load aggregators or distribution companies).

2. Pricing

P1. Wholesale market price reduction – short term spot and long term as supply adjusts.

P2. Reduced price volatility & hedging costs.

P3. Reduced market interventions (price caps).

P4. Deterred market power (as compared to “reduced market power” shown below).


Table 3‑1: Potential DR Benefits (cont.)

3. Risk management and Reliability

RM1. Physical hedge against extreme events – system or market.

RM2. Lower "insurance costs" for market participants against extreme events.

RM3. "Real Options" due to the increased resource diversity and a larger set of options for meeting loads both ongoing and in emergency situations.

RM4. Lower cost ancillary services to meet reliability criteria.

RM5. Ability of market participants to manage their ongoing financial risks.

4. Market Efficiency Impacts

E1. Equitable pricing.

E2. Incentive for innovative competitive retail markets.

E3. Incentive for development of efficient controls and end-use technologies.

E4. Reduced market power.

E5. Overall productivity gains by better utilizing industry investment.


Table 3‑1: Potential DR Benefits (cont.)

5. Lower Cost Electric System & Service

ES1. Reduced peaking capacity requirements.

ES2. Lowered transmission capital & operating expense.

ES3. Lowered distribution capital & operating expense.

ES4. Decreased or shifted generating costs (operating costs).

ES5. Reduction in LSE commodity costs.

ES6. Reduction in long-term resource adequacy requirements.

6. Customer Services

CS1. Increase in customer choice.

CS2. Possible increase in services.

7. Environmental

EN1. Potential avoided land-use, water, and air impacts.


Costs Associated with DR

1. Initial Implementation Costs (Fixed Costs)

IC1. Program Design Costs

IC2. Marketing

IC3. Metering/Communication Equipment

IC4. Business Integration

2. Ongoing Operating Costs (Variable Costs)

OC1. Incentive Payments (event-based, monthly, seasonal, etc.)

OC2. Ongoing Administrative and Maintenance

OC3. Customer Opportunity Costs


Stakeholder Views

1) Participating Customers

2) Non-Participating Customers

3) Load Aggregators

4) Distribution Companies (DISCO)

5) Transmission Companies (TRANSCO)

6) Load Serving Entities (LSEs)

7) Independent System Operator (CA ISO)

8) Generation Companies (GENCOs)

9) Market-Wide Views (System Level Perspective)


Needs Assessment

• Traditional supply-side planning recognizes the value of different resources to produce a least-cost resource plan, i.e., a portfolio of resources:

– High capital, low variable cost baseload generation

– Low capital, high variable cost peaking plants.

• An assessment of DR (and DSM) designed to deliver the system benefits likely will produce a similar portfolio:

– Energy efficiency is comparable to baseload generation.

– A decentralized price-response option to address demands in all hours (similar to a mid-merit power plant).

– Event-based system operator controlled dispatchable DR for critical events.

• Need to assess portfolios!


Needs Assessment:Questions to be Addressed

1) Baseline Question – What is the value of existing DR and existing resources?– Base case demand forecast.– Existing generation resources.– Existing transmission and distribution resources.– Existing levels of demand response or DSM resources.

2) What types of DR products/options should be assessed as part of a DR portfolio?

3) What size of DR products/options is appropriate?

4) Timing of DR deployment, expansion and/or maintenance?

5) Do different DR products have positive or negative synergies?


Needs Assessment:Questions to be Addressed

6) What are the “insurance” and “portfolio” benefits of DR:

– Diversity in resources, e.g., mix of fuels.

– Locational diversity, e.g., located near load centers.

7) How to assess the overall impacts on the electricity market now that incentives exist to shift loads?

– Technology innovation.

– Customer innovation in use of energy.

– Deterred market power.

– Appropriate use of supply-side capital investment.


DR Valuation Framework

Task Work Area 1 – Generation Expansion and Production Costing with Transmission Constraints and Load Flow Modeling.

Task Work Area 2 – Transmission investment avoided/ deferred costs based on project engineering and modular cost estimation (10 year project plans).

Task Work Area 3 – Distribution investment deferred costs based on based on project driven budget estimates (10 year project plans).

Task Work Area 4 – Market effects related to overall productivity and customer choice (secondary research)


Robust Planning Requirements

• Need to dimension uncertainty – rational decision making is best served when risk-reward tradeoffs are explicitly evaluated.

• Assess Risk Management and Value at Risk from different options.

• Fully address the portfolio of demand-side and supply-side options.

• Need to work with distributions of outcomes:– Closed form solutions and analytics.– Monte Carlo methods.– Decision-tree variants.

• Need to incorporate “time steps” and value of information to assess the value of flexibility contained in different plans.

• Models such as @Risk and Crystal Ball allow for representations of market uncertainties as a pre-processor step for planning models.

• Some resource planning models also incorporate Monte Carlo solutions within their input and solution algorithms. Adaptations of these models may be useful.


Robust Planning Needs (cont.)

• The tools exist to assess a resource portfolio of supply-side and demand-side options.

• This requires:

1. Appropriate resource characterization.

2. Representations of the uncertainty around key factors in the analysis.

• The challenge is to change perspectives and develop better (i.e., more accurate) representations of uncertainty:

– Correlations in distributions across different factors.

– Correlations over time for the distributions for the same factor (e.g., demand)

• Representing uncertainty and the value of information over time is the key challenge as both contribute to the value of options and hedges; and therefore to the value of DRR.


Overall Approach – Nine Steps

STEP 1 -- Base Case: Develop the base case set of resources that represent the without-DR scenario.

STEP 2 -- Pivot Factors: Determine pivot factors influencing the market costs of electricity.

STEP 3 -- Distributions: Assess uncertainty around these factors and express that uncertainty via probability distributions.

STEP 4 -- Create Joint Probability Surface: Combine the probability distributions to create a joint probability surface and make the Monte Carlo draws to represent alternative futures (approximately 100 sets of inputs to the model).

STEP 5 -- Base Case Planning Model Runs/Analyses: Run the model for each draw. In some cases, this may require adaptations to the model to reduce runtimes. OUTPUT: A distribution of system costs that incorporates uncertainty into the base case.


Overall Approach – Nine Steps (cont.)

STEP 6 – Benchmark DR WTP Valuations: Run willingness-to-pay scenarios for DR, i.e., reduce loads at specific locations that represent viable future DR scenarios (essentially this is a zero cost DR analyses)

STEP 7 – Develop DR Options: A representative set of DR programs/ options developed with costs of initiation and ongoing operation; along with realistic load reductions.

1. Interruptible Product – An amount of load reduction based on a 2 hour call period.

2. Direct Load Control Product – A known amount of load reduction with 5 to 10 minutes of notification.

3. Dispatchable Purchase Transaction – A call option where the model looks at the “marginal system cost” and “takes” the DR offered when it is less that marginal costs of production.

4. TOU/CPP Pricing Product – Modeled as a resource using elasticities to calculate demand reductions in TOU periods, and a critical peak price on event days.

5. Real-Time Pricing Product – This DR pricing option is modeled as a reduction in demand based on estimates from the literature.


Overall Approach – Nine Steps (cont.)

Step 8 – Estimate Value of DR Options: The base case model is run with various sets of DR options. The specification of different portfolios will allow this framework to address the relative value of different types of DR.

Step 9 – Analysis of DR Value Results: This final step will take the results from Step 8 and develop: 1) hedge values for the reduction in risk resulting from DR portfolios; and 2) values for the change in reliability from DR.


Model Concepts:A Model Neutral Approach

• Task Work Area 1: Select Generation and Transmission Models

– PROMOD IV Suite of Tools – NewEnergy Associates, a Siemens Company.

– ProSym and the Capacity Expansion Module from Global Energy Decisions (formerly Henwood Energy).

– AURORA model from EPIS, Inc.

– Candidate load flow models:♦ General Electric’s Positive Sequence Load Flow (GE PSLF) model.

♦ Power System Simulator for Engineering (PSS/E) offered by Siemens Power Technologies International.

• PJM suggested approach to assessing the impacts of DR on nodal prices – assume 3% load reduction during 100 peak hours.

– Calculate savings by reduction of dispatch of highest cost generation.

– Requires model capable of complete re-dispatch of the grid.


Two-Stage Modeling Approach

Set up/runMarket Model /

Production Cost Model with Transmission

Risk &ReturnMetrics

Calculate probability metrics

Set up/run Portfolio

Expansion Model

EfficientFrontier

PortfolioAssessment

PortfolioOptimization

Create distributions for risk analyses

Produce the efficient frontier Review & adjust portfolio

Market Modeling Data Crunching Final Portfolio Modeling Outputs


Three Implementation Approaches

1) Implementation Scenario 1: Collaboration with one of the California IOUs.

– Work with one of the California IOUs to collaboratively implement the resource planning models.

– Cost-effective implementation, but would the IOUs have the resources.

– Need utility collaboration on T&D regardless.

2) Implementation Scenario 2: A Demonstration Grade Analysis.

– Use data bases that exist.

– Limit number of scenarios to demonstrate the tools and methods.

3) Implementation Scenario 3: A Production Grade Analysis.

– Performed at the level of a utility planning project, with data cross checked and a broad set of scenarios.


Discussion of Select Issues


Amount of Uncertainty in System Costs

(From IEA Study, 2006)


Ranges of System Costs by Year

• Generally, a 100% increase from low to high in system costs over time.

Ranges of System Costs for Select Years Range of Total System Costs for Selected Years - Base Case($ Billions)

Year 2010 2012 2015 2018 2020 2023Maximum 7.7 8.2 10.2 10.3 12.4 15.0Minimum 3.5 3.8 5.1 5.6 6.5 7.5Range 4.2 4.5 5.1 4.6 5.9 7.5 Ratio 118.5% 118.8% 101.7% 82.2% 89.9% 99.3%


Reduced System Marginal Costs on Peak Days

• The model output hourly system marginal costs for a week in July (Note: actual market prices could be 3 – 4 times higher than these).

• The worst day of the week was typically Monday and the worst time was 4 – 6 in the afternoon.

• Peak hour(s) marginal costs were significantly reduced for the scenarios with DRR in both stress and non-stress cases.

• With DRR, in a stress case, savings were $24.5M on the peak day, about $45.0M for the week; and approximately $190.0M in savings for the month of July.


System Marginal Cost – Stress Case

System Marginal Costs - Monday in July 2015Case 4

0

100

200

300

400

500

0 2 4 6 8 10 12 14 16 18 20 22 24 26

Hour

$/MWh

Base Case With DR & RTP


Overall Risk Profile

• There was a change in the risk profile associated with the planning scenarios with the addition of DRR.

• There were significant savings when looking at value at risk (VAR) at the 90th percentile (VAR90) and at the 95th percentile (VAR95). Results for the three scenarios are shown below.

Risk Metrics – Reduced System Costs at Risk ($M)

VAR 90 VAR 95

DR No Pricing 238 213

DR with Peak -Period Pricing 966 924

DR with aggressive RTP 2,766 2,673


The Efficiency Frontier

• “Highest expected return at any level of risk”

“Efficient”portfolio Higher risk/

Higher returnExp

ecte

dR

etu

rn ()

Standard Deviation of Return ()

Lower risk/lower return

Higher risk/lower return

EfficientFrontier

The area under the curve representsevery possible portfolio combination


RTP Effect on Other DRR Programs

• The addition of more RTP meant that slightly less of the other three programs was used.

– BUT, not as big an impact as might be expected.

Average DR Capacity Usage

0%

5%

10%

15%

20%

25%

30%

35%

> 10% > 30% > 50% > 70%

Percentage of Capacity Used

Percentage of Years

Without RTP

With RTP


Reliability -- Loss of Load

• Loss of load was high in the base case in only a few cases of the 100 examined:– Due to the large system with many options for the model to call on in

an emergency situation (imports, other generation, etc.).

– The highest number of loss of load hours over the 20-year period was 32 hours, and the average was 7.6 hours.

• Importantly -- On a case-by-case basis, the addition of DRR reduced LOL hours by a large percentage – a maximum of 99% and a minimum of 51%.

• While high values for LOL were found, i.e., around 30, the average LOL hours was 7.6 hours and this was reduced to .5 hours with DRR.

• Next step is to value this increase in reliability?


NPV Savings by DRR and Pricing

• DRR and aggressive RTP produced at the low end approximately $1.5 to $2 billion in incremental system costs going forward.

Savings in Incremental NPV Over 19 Years

-1000

-500

0

500

1000

1500

2000

2500

3000

3500

0 20 40 60 80 100

Case

Savings ($M)

With DR and RTP With DR and CPP DR Only


A One Year Look

• Shows greater variability and price spikes not associated with the highest net system cost future.

Savings in Incremental NPV for 2023

-100

0

100

200

300

400

500

600

0 20 40 60 80 100

Case

Savings ($M)

With DR With DR and RTP With DR and CPP


Production Cost Data versus Price Formation

Documents

A Comprehensive / Integrated DR Value Framework Presented at: Demand Response Research Center TAG Technical Advisory Group Meeting Pacific Energy Center