Topic:"Forecasting Methodology" in M12249

1.4 KEY INPUTS The key input categories that informed the 2026 DSM Extension were as follows: - a) The prescribed statutory investment level of $63,750,000; - b) The 2023-2025 DSM Plan portfolio and corresponding programs; - c) To date 202...

AI summary The 2026 DSM Extension is informed by four key inputs: a statutory investment level of $63.75M, the 2023-2025 DSM Plan portfolio, implementation results to 2023, and the 2025 forecast. Statistics Canada's 2021 census data on Nova Scotia's Low-Income Measure After Tax (LIM-AT) is also referenced.

3. MODELLING FOR 2026 DSM EXTENSION

AI summary This section discusses the modelling for the 2026 extension of Demand-Side Management (DSM) programs in Nova Scotia. Key entities involved include the Nova Scotia Utility and Review Board (NSUARB) and the DSM Cost Recovery Rider (DCRR). The analysis involves regulatory considerations under the Public Utilities Act (PUA) and collaboration with the Independent Energy System Operator (IESO).

3.1 OVERVIEW In support of the 2026 DSM Extension Application, E1 has conducted a fulsome modelling process. An overarching objective of E1 in its modelling process for the 2026 DSM Extension was to adopt learnings drawn from the actual re...

AI summary E1's 2026 DSM Extension Application uses 2023-2025 data and 2025 forecasts to inform targets, noting no alternate scenarios were modelled. The NSUARB required alternative scenarios in past applications, with future submissions needing DSM budget scenarios and NSPI rate impact analysis.

8 3.3 MODEL INPUTS AND ASSUMPTIONS - 9 In collaboration with its consultant Guidehouse, E1 developed a set of inputs to use in the modelling - 10 process for the 2026 DSM Extension for both the Energy Efficiency Model and the Demand Respon...

AI summary E1, in collaboration with Guidehouse, developed model inputs for the 2026 DSM Extension, including line losses, avoided costs, discount rates, annual energy savings, peak demand savings, incremental costs, and incentives for both the Energy Efficiency and Demand Response Models.

those that are mainly electrically heated (i.e., they have a secondary heating system) to update unitary savings for heat pumps and wood/pellet burning equipment. The billing analysis resulted in significantly lower unitary savings for the...

AI summary The 2025 forecast highlights lower unitary savings for electrically heated homes due to billing analysis, increased unit costs in the Custom program driven by New Construction participation, and higher costs in Small Business Energy Solutions from new measures. E1 notes delayed availability of 2024/25 demand response capacity results until mid-2025, affecting the 2025 forecast.

2.5 MODELLING APPROACH - For the 2026 DSM Extension, E1 utilized the same modelling process and software tools as in the approved - 2023-2025 Plan. Modelling supports quantitative development by providing the following: - detailed cost eff...

AI summary E1 used the same modelling approach and software tools as in the approved 2023-2025 Plan for the 2026 DSM Extension. The modelling supports quantitative analysis through cost-effectiveness impacts, energy/demand impacts, DSM participation estimations, and investment projections.

2.5.1 EN ERGY EFFICIEN CY M OD EL - Following the same approach as in the approved 2023-2025 Plan, E1 engaged Guidehouse, to provide its - ProCESS™ short-term DSM planning tool for modelling the 2026 energy efficiency portfolio. - E1 and G...

AI summary E1 engaged Guidehouse to use its ProCESS™ tool for modeling the 2026 energy efficiency portfolio, building on the 2023-2025 Plan. Input data included line loss factors, avoided costs, and modified 2023 evaluation reports. Guidehouse developed and ran the model using technical measure data.

1 2.5.2 D EM AN D RESPON SE M OD EL - 2 As with the energy efficiency model, E1 worked with Guidehouse to complete demand response modelling - 3 using Guidehouse's DRSim™ model. This modelling approach was consistent with the approach used...

AI summary E1 collaborated with Guidehouse to develop a demand response model using DRSim™, aligning with their 2023-2025 DSM Plan. The model uses bottom-up analysis with primary and secondary data, segmenting customers, defining DR options, and estimating cost-effectiveness.

ed in 2024 and the 128.7 GWh forecast for 2025. Consistent with the 2025 forecast, energy savings in 2026 will continue to be lower than what was achieved in 2024, as residential lighting is no longer

AI summary The 2026 DSM Extension Portfolio forecasts energy savings lower than 2024 levels, primarily due to residential lighting no longer contributing significantly. The 2025 forecast of 128.7 GWh aligns with this trend, indicating reduced savings from residential sectors.

A. Energy Efficiency ProCESS Model - E1 understands from NS Power that the avoided costs of carbon (electric utility compliance costs) are - embedded in the avoided costs of energy that NS Power calculated for the Evergreen IRP No Atlantic...

AI summary E1 used avoided energy costs (including embedded carbon costs) from NS Power for the 2026 DSM Extension, without modeling separate carbon costs. Avoided energy costs were not included in demand response cost-effectiveness testing. The same RBIA approach as energy efficiency programs was applied for the 2026 DSM Extension.

in annual avoided costs. - The annual impacts depicted in [Figure 3](#page-120-0) should not be interpreted to be the actual rate changes that will - occur in these years as experienced by customers. Figure 3: Annual Rate Impacts (2026-204...

AI summary The text discusses annual rate impacts of DSM activities, noting that NS Power's RBIA methodology explains varying benefits across rate classes based on fuel vs. fixed cost structures. It highlights that simplifying lifetime savings to a single weighted-average lifespan introduces discontinuous jumps in savings models, unlike real-world scenarios.

4.6 DEMAND RESPONSE ASSESSMENT In the 2022 Rate and Bill Impact proceeding, Synapse recommended that E1 monitor for models used in other jurisdictions that they may adopt to enhance the demand response assessment in the RBIA and E1 indicat...

AI summary In the 2022 Rate and Bill Impact proceeding, Synapse advised E1 to adopt models from other jurisdictions to improve demand response assessments. E1 committed to refining models with its consultant Elenchus but has not identified necessary changes yet. The NSUARB directed E1 to report on model developments in its next report.

2. RESOURCES AND SCENARIOS - The 2026 DSM Extension Analysis includes the NS Power rate model and the E1 RBIA model, filed - in Attachments 5 and 6 respectively. The analysis compares two scenarios: a DSM scenario and a - no-DSM scenario....

AI summary The 2026 DSM Extension Analysis compares DSM and no-DSM scenarios using NS Power's rate model and E1's RBIA model. It evaluates energy efficiency and demand response impacts, isolating 2026 DSM effects on rates and bills. Alternative scenarios include Energy Efficiency Only and Demand Response Only, with results summarized in Attachment 1.

2.2 DEMAND RESPONSE INPUTS - Demand response costs, savings, measure life, and customer incentives are calculated and - entered separately in the model from energy efficiency inputs. Demand response inputs are - determined separately from...

AI summary Demand response (DR) inputs are modeled separately from energy efficiency (EE) to enable scenario analysis, including DSM, EE-only, and DR-only cases. DR programs affect demand, not energy, with one-year measure life and continuous participant engagement. Data for the 2026 DSM Extension RBIA comes from Guidehouse's DRSim™ model and historical forecasts.

3. ENERGY AND DEMAND SALES - NS Power has provided historical and projected energy and demand sales within each rate class - for 2011 2040. Energy sales are provided at the customer's meter for both the with DSM and - without DSM scenario....

AI summary NS Power has submitted historical and projected energy and demand sales data from 2011 to 2040, including comparisons between scenarios with and without Demand-Side Management (DSM). The data includes energy sales at customer meters and average monthly demand forecasts under the DSM scenario.

7.2.1 ANN UAL TRACKED EN ERGY EFFICIEN CY PARTICIPA TION - For 2026, annual tracked participation was first estimated at the program component level. For - some program components this was done directly using inputs to Guidehouse's ProCESS...

AI summary The 2026 annual tracked participation for energy efficiency programs was estimated using Guidehouse's ProCESS model and scaled 2023 RBIA data with energy and unit factors. Results were allocated to rate classes based on historical 2023 participation patterns.

7.2.2 ACTIVE TRACKED ENERGY EFFICIEN C Y PARTICIPATION - For 2026, in the forward-looking RBIA, all annual participants are considered to be active - participants, as the forward-looking RBIA does not account for any impacts prior to 2026....

AI summary The forward-looking RBIA assumes all annual participants are active in 2026 and remains flat until their energy savings expire, after which participation drops to zero. This approach does not account for pre-2026 impacts.

7.3 UNTRACKED (POINT-OF-SALE PROGRAM) PARTICIPATION - E1 operates two program components that offer rebates at the point-of-sale: residential Instant - Savings and the Instant Rebates portion of Business Energy Rebates (BER-IR). These prog...

AI summary E1's Untracked Point-of-Sale Program includes residential and business rebate components (BER-IR) with participation estimated via transaction records and assumptions about rate class participation. For 2026, annual and active participants are estimated using forward-looking RBIA methods, with assumptions about flat participation until energy savings expire. Residential Behaviour and Demand Response participation methods are also detailed, including cross-participation rates and DRSim™ model inputs.

Cost of Service Studies COSS provides the most insight into class cost causation as based on changes in its energy and demand usage. It shows in a transparent way how rate class usage of demand and energy services within each functional ar...

AI summary COSS (Cost-of-Service Study) is critical for analyzing class cost causation by leveraging NS Power's Load Forecast Report and E1's long-term usage forecasts. This approach simplifies pricing adjustments by utilizing existing data rather than future investment details, ensuring transparency in rate class changes due to DSM.

3.2.4 Generic COSS Results The actual results from the above cost allocation process under the "With DSM" and "No DSM" scenarios are presented in the "COSS Outputs" tab within NS Power's rate model, where the long-term trends in annual rel...

AI summary The COSS Results compare 'With DSM' and 'No DSM' scenarios, showing long-term unit cost trends by rate class. Historic periods show higher DSM cost impacts, while out-years show reduced differentials. Fuel-cost-heavy classes (e.g., Large Industrial) benefit more from DSM savings, whereas fixed-cost-heavy classes (e.g., Domestic) see less benefit. Differences arise from DSM spend, usage changes, and cost allocation methods.

"COSS Data Inputs" tab This tab includes all annual test year class usage and embedded costs from the COSS and BCF COSS filed in GRA and BCF proceedings as well as a forecast of annual usage by class per the most recent ten-year Load Forec...

AI summary The 'COSS Data Inputs' tab contains annual test year data from COSS and BCF COSS filings, load forecasts, and DSM expenditures, used to determine class unit costs and revenues. It includes data from regulatory proceedings and forecasts for usage by rate class.

III. Resource Acquisition and Other Evaluation Frameworks Efficiency Nova Scotia programs are almost entirely resource acquisition programs that treat saved energy as equivalent to generated energy. This is the original framework for the e...

AI summary Efficiency Nova Scotia's energy efficiency programs are evaluated under a resource acquisition framework, equating saved energy to generated energy. Econoler's approach is highlighted, with mentions of evolving evaluation frameworks and market transformation. DSM evaluation types (impact, process, market) are discussed.

VIII. General Findings • The method followed in each program impact evaluation follows a recognized analytic approach appropriate for each program type. The structure and format of each impact evaluation follows a consistent template (exce...

AI summary The document highlights that program impact evaluations follow recognized methods and consistent templates, with Econoler using Efficiency Nova Scotia's CIRx Screening Tool. Evaluations include executive summaries, methodological diagrams, and appendices. The Evaluator conducted four process and two market evaluations, with process evaluations being desirable for future work. Carbon emissions offsets are appropriately developed.

since MHEEP began in 2019. - A unitary savings review, - An effective useful life (EUL) update study - Calculations using evaluation results, NTGR ratio of 1, and avoided GHG reduction calculations. As part of the 2024 evaluation the Evalu...

AI summary The 2024 evaluation of the Mi'kmaw Home Energy Efficiency Program (MHEEP) reports 0.401 GWh in net electrical energy savings, below the 0.618 GWh target. Methodology changes using a new billing analysis and adjustment ratios (AR) improved accuracy, reflecting higher heat pump adoption. The average effective useful life (EUL) was 20.3 years, with 8.141 GWh in lifetime savings.

not make sense. Similarly, unless NSP demonstrates otherwise, a 6.27 GWh effect at the system level means that a savings claim for the program at the system level does not make sense.[55](#page-60-1) Currently, behavioural RCTs, of which t...

AI summary The document critiques the energy savings claims of a program, arguing that without NSP's demonstration, a 6.27 GWh effect at the system level lacks validity. It highlights that behavioral RCTs are 'black boxes' lacking coherent mechanisms and warrants, referencing Cartwright and Hardie's methodology for establishing program efficacy through causal analysis.

Recommendations SVR2024-Demand Response – 13 . In the next evaluation, include an analysis of the relative importance or lack of importance to the possible capacity shortfall problem to Nova Scotia Power, the roles of the load research sho...

AI summary The document recommends evaluating Demand Response (DR) programs' impact on Nova Scotia Power's capacity shortfall, clarifying their practical benefits beyond learning experiences, and justifying their business case. It critiques DR programs for minimal kW demand reduction and calls for analysis of whole-home vs. device-level approaches in residential DR. A citation to Econoler's report is included.

2. BNI Efficient Products Rebates SVR2024-BNI Efficient Products – 8. Change baselines for BER and IR rebates to reflect current market practices that have indoor DLC-Standard products as the new baseline with incentives offered for compar...

AI summary The document proposes adjusting BER and IR rebate baselines to DLC-Standard products, offering incentives for DLC-Premium alternatives. It also recommends reviewing in-situ meter studies for LED lighting products or commissioning a study if necessary.

XII. References American Statistical Association, Statement on Statistical Significance and P-Values, Provides Principles to Improve the Conduct and Interpretation of Quantitative Science, March 7, 2016 [(www.amstat.org/asa/files/pdfs/p-va...

AI summary The references include academic and industry sources on statistical methods, energy efficiency programs, and policy evaluation. Key entities are organizations like the American Statistical Association, the Consortium for Energy Efficiency, and reports on thermostat programs and energy sufficiency.

3. ANNUAL NEW WIND MODELLING CONSTRAINT E1 reviewed the annual amount of new wind resources built in each year by scenario, as shown in Figure 1. Apart from planned wind projects expected to be online by 2026, E1 understands an annual new...

AI summary E1 reviewed annual new wind build rates under different scenarios, noting a 200 MW/yr cap in most scenarios except No DSM, which allows 400 MW/yr. This difference likely caused a dip in avoided costs from 2029-2035. E1 emphasizes applying consistent constraints across all scenarios to ensure meaningful avoided cost results.

4. GHG EMISSIONS BY SCENARIO NS Power's August 23, 2024 avoided costs deliverable provided annual GHG emissions for the scenarios CE1-E1-R2, Base DSM, No DSM ("No EE, No DR"), and No DR. E1 has plotted the annual CO 2 emissions intensity b...

AI summary NS Power's analysis of GHG emissions under various scenarios (CE1-E1-R2, Base DSM, No DSM, No DR) shows the No DSM scenario exceeds the 50g/kWh post-2035 carbon cap. E1 argues that the same constraints should apply to all scenarios for avoided costs to be meaningful, citing the 2022 IRP's 50g/kWh limit. Figures 2 and 5 illustrate emission trends and discrepancies.

Discussion of 2021 and 2024 avoided cost results: Energy and carbon combined ( Figure 3 ) : - When assessing the 2026-2045 period, average avoided costs have decreased 11% compared to 2021 and decreased 15% compared to the 2021 CPI adjuste...

AI summary The document discusses avoided costs for energy and carbon from 2021 and 2024 analyses, noting an 11% decrease in average avoided costs by 2026-2045 compared to 2021 and a 15% decrease compared to 2021 CPI-adjusted costs. Energy avoided costs show fluctuations, peaking in 2028 and dipping in 2031. The text requests NS Power to provide estimates of the embedded avoided cost of carbon and explain calculations.

Discussion: • Some of the observations noted above may in part be driven by inconsistent scenario constraints, as discussed elsewhere in this document. If, as recommended above, new wind build out and GHG emissions constraints are applied...

AI summary The discussion highlights concerns about inconsistent scenario constraints affecting embedded carbon cost estimates and recommends consistent GHG emissions constraints. E1 requests NS Power to reassess carbon cost impacts and provide detailed breakdowns of carbon compliance and non-carbon related costs in future avoided cost analyses.

E1 Responses to Nova Scotia Energy Board (NSEB) Information Requests NON-CONFIDENTIAL 1 Avoided costs were updated for the 2026 DSM Extension to reflect the most o 2 up-to-date avoided costs of capacity, transmission, and distribution – as...

AI summary E1 updated avoided costs for the 2026 DSM Extension, leading to lower TRC due to BTM Battery Control changes. Factors include reduced battery adoption and higher variable costs from DRMS. Insights from 2023-2025 DSM Plan improved 2026 modeling accuracy.

E1 Responses to Synapse Energy Economics (Synapse) Information Requests NON-CONFIDENTIAL 1 (c) The avoided costs used to calculate the Lifetime Benefits (TRC and PAC), TRC ratios and PAC 2 ratios for 2026 in the 2026 DSM Extension were dev...

AI summary EfficiencyOne (E1) outlines that Nova Scotia Power (NSP) provided avoided cost data for TRC and PAC calculations to the DSMAG in 2024 and 2021, using the 2022 and 2020 IRP updates respectively. Updated transmission/distribution avoided costs were shared in 2024, developed outside the 2022 IRP modelling. References to matter numbers M12249 and M10473 are included.

9 Response IR-12-e: - 10 In terms of this discussion the total claimed savings of 10% or greater for a well-maintained - 11 system as the upper bound.

AI summary The response discusses a claim of 10% or greater energy savings for a well-maintained system as an upper bound, emphasizing the potential of demand-side management in regulatory proceedings.

Response IR-14: - The verification team specifically requested an on-site verification effort be conducted at the - site, at the original request for site visits this explicit request was bolded to - highlight it as the highest priority. T...

AI summary The verification team requested an on-site visit, which was denied by the Customer as they had already accommodated a prior site visit. E1 used tracking sheet data and evaluator reports to calculate total system savings.

DSM Potential Studies in New York, New Jersey, and Pennsylvania Optimal Energy, Inc. - Vermont (December 2018 – December 2019) - Assisted Optimal Energy, Inc. with the development of measure assumptions and characterizations for statewide,...

AI summary Optimal Energy, Inc. assisted with developing measure assumptions and characterizations for electric and gas demand-side management (DSM) potential studies across New York, New Jersey, and Pennsylvania from December 2018 to December 2019.

Comments on EmPower Maryland Programs Sierra Club, Maryland (September 2011 – October 2011) - Research for and development of comments on EmPower Maryland's energy efficiency programs, including the development of alternative energy effici...

AI summary Sierra Club conducted research and developed comments on EmPower Maryland's energy efficiency programs during September–October 2011, focusing on alternative energy efficiency potential projections. The analysis aimed to evaluate and propose modifications to the programs' methodologies.

Vermont's 20-year Forecast of Electricity Savings from Sustained Investment Efficiency Vermont – Burlington, Vermont (December 2008 – October 2009) - Provided components of final report relating to long-term trends for the environment (cli...

AI summary Efficiency Vermont's 2008–2009 report outlines a 20-year forecast of electricity savings, focusing on environmental trends (climate change, land-use, water-use), population growth, regulatory impacts, and technical analysis of electric demand-side savings potential.

nt, the treatment group would have been larger, but no savings would have probably been detected. The Peach Report states as follows in relation to sampling and independence of cases:[12](#page-24-1) […] standard significance tests assume...

AI summary The Peach Report highlights concerns about statistical dependencies in energy use data, noting that weather patterns and climate trends may influence utility data, leading to non-independent observations. This challenges the reliability of standard significance tests. A reference is made to NREL's residential behavior evaluation protocol and a specific exhibit from matter M12249.

Econoler Response: Econoler disagrees with this evidence. In the evaluation, savings were calculated with a panel regression analysis, therefore individual household consumption values were not used in the savings calculation, but instead...

AI summary Econoler disputes the evidence, arguing that savings were calculated using group averages via panel regression analysis rather than individual household data, leading to aggregated results (referenced in section 4. Causal Effects).

uced energy use across the treatment group compared to the control group, the causal effect is demonstrated through observed outcomes, even if the precise internal mechanism is not fully disentangled. The Peach Report states as follows in...

AI summary The text discusses a study showing reduced energy use in the treatment group compared to the control group, demonstrating a causal effect. The Peach Report highlights the need for systematic analysis of adopted practices and their measurable impacts to optimize program design, though current data is insufficient.

Residential Demand Response description In 2024, Residential DR was composed only of Eco Shift – a residential "bring your own device" offering generating available DR capacity through three pathways: 1) Smart thermostats for electric spac...

AI summary In 2024, Nova Scotia's Residential Demand Response (DR) program, Eco Shift, focused on smart thermostats due to limited participation in other pathways (EVs, batteries). Econoler evaluated DR capacity using regression models on AMI data, aiming to aggregate participant impacts to reduce NS Power's need for new capacity or expensive peak-period electricity purchases.

5.3 Savings and Verification Report Recommended Disallowances [50] Dr. Gil Peach, Board Counsel's consultant, recommended that savings from the residential behavioural program, the residential and BNI demand response programs, and the comp...

AI summary Dr. Gil Peach recommends disallowing savings from residential behavioral, demand response, and compressed air programs due to insufficient independent evaluation. Econoler defends its methodology, arguing it balances accuracy and cost, and notes no other jurisdictions require the disputed test. Disagreement centers on evaluation protocols and reliability of reported savings.

b. Please explain why 2023 actuals are not included in the scaling factors for the Residential Behavior program component. c. For program components in which the actuals are not relatively consistent from 2023 to 2024, please discuss why E...

AI summary The NSUARB requests clarification on scaling factors for the Residential Behavior program, excluding 2023 actuals and using 2023-2024 averages. It also questions methodology for attributing low-income savings in DSM Reporting, focusing on Business Energy Rebates, Custom, and Small Business Energy Solutions programs.

1 Request IR-1: 2 3 Reference: EfficiencyOne's Evidence, p. 13 4 5 E1 refers to the potential impact of "current economic and geopolitical uncertainty," but further 6 states that "there has been no specific adjustment made in the 2026 targ...

AI summary The document contains a request (IR-1) directed at EfficiencyOne (E1), questioning their 2026 DSM extension modelling. It asks whether E1's model uses historical data and forecasts, acknowledges market uncertainty, assesses economic/geopolitical impacts on DSM costs, and explains consequences if no assessment was conducted.

Request IR-5: Refer to M12249, Exhibit E-2, 2024 Peach Report, Section X, Individual Program Component Review, subsection J. BNI Efficient Products Rebates (BER), pages 58-60, which states, at pages 58-59: Point-of-sale rebate evaluated sa...

AI summary The 2024 Peach Report notes a 7.7% decline in BER program savings, with LED Linear Lamps dropping 41%, but a slight increase in NTGR from 81% to 84% mitigated the decline. The report recommends updating baselines for BER and IR rebates to use DesignLights Consortium-Standard products as the new baseline. Questions are raised about the statistical significance of BER results compared to demand response programs and how the baseline change would address savings decline.

Request IR-11: - Reference: page 54 of the 2024 Verification Report (Section I. Residential Behavioral Program - (Efficiency Insights): Currently, behavioural RCTs, of which the current program is an example, are black boxes. There is no c...

AI summary The text critiques current behavioral RCTs in energy programs as 'black boxes' due to lack of coherent mechanisms or energy-saving specifications. It references Cartwright and Hardie's more robust approach and requests examples of jurisdictions using this method. The proceeding relates to E1's 2026 DSM extension application (M12249).