Topic:"Forecasting Methodology" in M12665

III.A. Background The purpose of the Fuel Adjustment Mechanism—to ensure that power rates reflect the actual cost of the fuel used to produce the power and not simply a forecast of fuel need—highlights the importance of accurately forecast...

AI summary The Fuel Adjustment Mechanism (FAM) ensures power rates reflect actual fuel costs, emphasizing the need for accurate forecasting of energy and peak capacity requirements for NSPI's in-province customers and the least cost supply plan to meet those needs. This section outlines NSPI's forecasting and planning processes during the 2024-2025 Audit Period.

, which is filed with the NSEB each year. There were no significant changes to the forecasting methodology and no organizational changes impacting the load forecasting process during the Audit Period. The quantities of the various fuels re...

AI summary The document discusses the forecasting methodology used by NSPI, which involves using the PLEXOS model to estimate fuel requirements based on generating unit characteristics, fuel prices, GHG limitations, and transmission system capacity.

Energy Sales Forecasting NSPI forecasts long-term system load (energy) and system peak demand using separate models for its residential, commercial, and industrial sector customers. The forecasting methodology and organization involved in...

AI summary NSPI uses separate models for residential, commercial, and industrial sectors to forecast long-term system load and peak demand. The methodology and organization involved in load forecasting remain consistent with previous audit periods. The residential sales forecast combines average use and customer count forecasts, using historical data, economic indicators, appliance energy use data, and weather data.

III.B.1.d. Peak Demand Forecasting The long-term system peak forecast for the accrued classes is derived through a linear regression model that relates monthly peak demand (excluding large customer contribution) to heating, cooling, and ba...

AI summary The document discusses peak demand forecasting using a linear regression model that factors in weather, base load, and growth programs like heat pumps. It also references demand response savings from a third-party study and projects a 1.2% annual growth in peak load through 2035.

III.B.1.d.i. Accuracy of NSPI's Load and Peak Demand Forecasts The accuracy of the forecast of the Total System Requirement for the Audit Period years (2024 and 2025) was in line with results from the prior audit period (2022-2023), as sho...

AI summary The accuracy of NSPI's load and peak demand forecasts for 2024 and 2025 was consistent with the prior audit period (2022-2023) and comparable to other utilities, as illustrated in Figure III-1.

Figure III-1: Comparison of Total System Requirement Forecast vs. Actual (MWh, 2022 through 2025)100 Year Actual Forecast Variance 2022 11,133,919 11,279,702 -1.29% 2023 11,138,520 11,180,622 -0.38% 2024 11,326,320 11,240,054 0.77% 2025 11...

AI summary The text presents a comparison of actual versus forecasted total system requirements from 2022 to 2025, showing consistent underperformance of actual values relative to forecasts. NSPI's peak load was significantly lower than forecasted in 2024 and 2025, continuing a trend since 2020 where forecasts overestimated actual peak loads.

III.B.1.e. Short-Term Load Forecast The system operator relies on short-term load forecasts in its day-ahead unit commitment and scheduling processes. NSPI's operations rely on day-ahead, 2-day-ahead, 3-day-ahead, 4-day-ahead, and 5-day-ah...

AI summary The system operator uses short-term load forecasts provided by a third-party vendor for day-ahead scheduling. The vendor's model, based on historical load data and weather forecasts, is monitored for accuracy. After a cyber event in April 2025, forecast accuracy, measured by MAPE, significantly worsened, averaging 10.22% compared to 2.41% before the event.

Figure III-4: Day-Ahead Forecast Accuracy – All Hours (2024-2025) Monthly Mean Month Absolute Deviation Monthly Mean Absolute Percentage Error January 2024 38.3 2.37% February 2024 44.3 2.89% March 2024 50.1 3.63% April 2024 24.4 2.01% May...

AI summary Figure III-4 presents day-ahead forecast accuracy metrics for 2024-2025, showing monthly mean absolute deviation and percentage error. The analysis found no bias in the pre-cyber event period, but significant overestimation in the post-cyber event period, indicating substantial model bias.

Figure III-5: Percentage of hours where actual load exceeded day-ahead forecast (2024-2025) Month Hours Where Actual Load Exceeded Forecast Hours in Month Percentage of Hours Where Actual Load Exceeded Forecast January 2024 429 744 57.7% F...

AI summary NSPI's short-term load forecasts showed bias and high MAPEs after a cyber event. The issue was raised with NSPI to determine if intentional bias was used for reliability, but NSPI denied this and stated its calculations aligned with historical results. Discrepancies in calculations were noted despite using NSPI's data.

There were contributing factors to these deviations that were outside of NSPI's control. For example, Surplus Energy from Muskrat Falls was lower than forecasted. This can be seen in the "Imports" row of the following two figures. In 2024,...

AI summary The text discusses deviations in energy supply forecasts due to factors outside NSPI's control, including lower-than-expected surplus energy from Muskrat Falls and underperformance of renewable resources. These deviations required replacement energy from other sources, and the text references specific reports and figures for further details.

III.C. Conclusions Conclusion III-1: NSPI's load forecasting approach remains largely unchanged from the prior audit period. NSPI continues to file an annual 10-year load forecast with the Board and allows for stakeholder involvement in th...

AI summary NSPI's load forecasting approach has remained largely unchanged, with annual 10-year forecasts filed with the Board. The 2025 Load Forecast Report predicts a slight decrease in net system requirement due to factors like RTR load migration and increased energy efficiency. NSPI's actual peak loads have been lower than forecasted in recent years, suggesting a trend of over-forecasting.

5 Rebuttal Evidence, page 9 lines 1-3. 328 NSPI, Exhibit N-3(C), M11533, September 13, 2024, section 2.10.1. 329 Bates White's February 21, 2025 Rebuttal Evidence, page 6 lines 19-20, page 7 line 5. 330 NSPI, Exhibit N-3(C), M11533, Septem...

AI summary Bates White's 2022-2023 recommendations were addressed by NSPI, which confirmed it periodically reviews forecasts and assumptions. NSPI's February 2026 update affirmed its ongoing practice of reviewing forecasting assumptions and outputs.

443 2020-2021 Audit Report, pages 204 to 205. 444 NERC GADS data, https://view.officeapps.live.com/op/view.aspx?src=https%3A%2F%2Fwww.nerc.com%2Fglobalassets%2Fprograms%2Frap a%2Fgads%2Freports%2Fgenerating-unit-statistical-brochure-2-2024...

AI summary The 2020-2021 Audit Report indicates that actual output during the Audit Period was 35,367 MWh, which is similar to the prior Audit Period of 33,855 MWh. The data is compared to forecasts and visualized in Figure X-21, with NERC GADS data referenced for generating unit statistics.

Quarter Slope Coefficient R-Squared Q1 2024 - 0.98 Q2 2024 - 0.89 Q3 2024 - 0.81 Q4 2024 - 0.87 Q1 2025 Q2 2025 - 0.96 Q3 2025 - 0.95 Q4 2025 - 0.87 Figure XIII-7: Portfolio-Level Regression Analysis Output 789

AI summary Figure XIII-7 presents a portfolio-level regression analysis output with slope coefficients and R-squared values for various quarters from 2024 to 2025, indicating the statistical relationship between variables over time.

XIII.C. Conclusions Conclusion XII-1: The hedging program as executed during the Audit Period generally conformed with the objectives of the Fuel Hedging Plan. Conclusion XIII-2: Our review shows that NSPI conducted quarterly rebalancing a...

AI summary The hedging program executed by NSPI during the Audit Period generally aligned with the Fuel Hedging Plan. Quarterly rebalancing was consistent with plan requirements, though impacted by liquidity issues and SO2 compliance verification. NSPI's success in shielding FAM ratepayers depends on the accuracy of fuel consumption forecasts, and changes in energy flows from NLH on the Maritime Link also influenced hedging activities.