LOADS, CUSTOMERS AND REVENUE

Size: px
Start display at page:

Download "LOADS, CUSTOMERS AND REVENUE"

Transcription

1 Page of LOADS, CUSTOMERS AND REVENUE The purpose of this evidence is to present the Company s load, customer and distribution revenue forecast for the test years. The detailed test year forecasts are shown in Exhibit K, Tab, Schedules -. The bridge year forecasts are in, Tab, Schedules -, and the historical loads, customers and revenues are shown in,, Schedules - and Tab, Schedules -. 0 Table below provides a summary of the loads, revenue, and customer forecasts. The revenue forecast is calculated based on proposed distribution rates, excluding commodity, and excluding rate riders. Table : Total Load, Revenues and Customers Year Total GWh Total kva Total Distribution Revenue Total Customers 00(A),,,,,, 00,,0,,,, 00,,,,,0, 00,,,0,,0,0 00,0,0,,,, Notes:. Streetlights and Unmetered loads measured as connections. Total kva are billed kva (eg: not including residential or GS<0 classes) net of CDM savings.. Total Customers are as of year-end and streetlight and unmetered loads measured as connections.

2 Page of On a rate class basis, forecast sales and revenues are shown in Table below. Table : Sales and Revenue by Rate Class Year Residential GS<0 GS 0- GS 0-000kW GS 000kW Large Unmetered Street Non Interval Users Load lighting Interval 000kW Load (GWh) 00(A),,,,0,, 0 00,,,,0,00, 0 00,,,,,0,0 0 00,,,00,0, 0 00,,,0,00,0 Distribution Revenue ($000s) 00(A),,,00,0,,,, 00, 0,,,,,,, 00 0,,,,, 0,,,0 00,0, 0,,,,,, 00,0,00,,,0,0,0,

3 Page of Table shows historical and forecast annual use per customer. Table : Annual Use Per Customer GS 0- GS 0- Year Residential GS<0 000kW Non- 000kW GS 000- Large Unmetered Street Interval Interval 000kW Users Load lighting Load (kwh per connection) 00(A),,0,0,, 0,0,,,00,0 00(A),,,00,, 0,,,,0, 00(A),,,,,,,,00,0, 00,0,,0,,,,,,, 00,0,,0,,,,,,0 00,0,,,,0,0,,, 00,0,, 0,0,,,0,0, Annual Growth % 00(A).%.%.% -0.% 0.% -.% -.% 0.% 00(A) -.0% -.% -.% -.% -.% -.0% -.% -.% 00 0.%.% -0.% -.% -.% -.% 0.%.% 00 -.% 0.%.% -0.% -.% 0.% 0.% 0.% 00 -.% 0.0% -.% -.% -.% 0.% 0.% 0.% 00 -.% 0.% -.% -.% -.% 0.% 0.% 0.% Note: Use per customer based on mid-year customer

4 Page of 0 0 LOAD FORECAST METHODOLOGY The Company s revenue and load forecast is developed in a three-step process. First, a total system purchased energy forecast is developed based on multifactor regression techniques that incorporate historical load, weather, and economic data. Second, a system demand forecast is developed based on historical and forecast load factor relationships. Finally, energy and demand by rate class are forecast based on historical billing statistics. A forecast of customers by rate class is also determined using timeseries econometric methodologies. Revenues are determined by applying the proposed distribution rates to the rate class billing determinants for the forecast period. KWh Load Forecast The forecast of total system purchased energy is developed using a multifactor regression model with the following independent variables: weather (heating and cooling degree days), economic output (GDP growth), load characteristics (peak hours per month) and calendar variables (days in month, seasonal). The regression model uses monthly kwh and monthly values of independent variables from through to the latest actual values (June 00) to determine the monthly regression coefficients. Data for Toronto Hydro s total system load is available only as far back as January. With the amalgamation of the six former municipal utilities in, consistent data is available only for this time period. The monthly model therefore has over 00 data points, which is a reasonable data set for use in a multiple regression analysis. Historically, load growth for Toronto Hydro has been fairly stable (on a normalized basis), averaging 0. percent annually over the -00 period. Figure and Table below show historical loads.

5 Page of,00 Figure Total Purchased GWh,000,00 GWh,000,00,000, Year Non-Normalized Normalized Table : Historical Annual Load Year Percent Percent Non-Normalized Growth Normalized Growth Change Change GWh GWh GWh GWh (%) (%),,,0., , 0.,. 00,., 0. 00,00.0, , () (.0), 0. 00, (00) (0.), 0. 00,., (0) (0.) Annual Average.0 0.

6 Page of 0 0 The main drivers of load growth over time are economic conditions, while the primary driver of year-over-year changes is weather. Both of these effects are captured within the multifactor regression model. Economic growth which encompasses both growth in the Company s customer base as well as general economic conditions within the Company s operating area is captured in the model using an index of economic output, Real Gross Domestic Product ( GDP ). Weather impacts on load are apparent in both the winter heating season, and in the summer cooling season. For that reason, both Heating Degree Days ( HDD a measure of coldness in winter) and Cooling Degree Days ( CDD measure of summer heat) are modelled. The third main factor determining energy use in the monthly model can be classified as calendar factors. For example, the number of days in a particular month will impact energy use. The modelling of purchased energy uses number of days in the month, hours of peak load in a month, and two dummy variables one to capture the typically lower usage in the spring and fall months, and the other to capture the impact of the 00 August blackout on energy use in that month. The process of developing a model of energy usage involves estimating multifactor models using different input variables to determine the best fit. Based on a priori assumptions about which input variables will impact energy use, different models were fit. Using stepwise regression techniques different explanatory variables were tested with the ultimate model being determined both by model statistics and by forecast accuracy. A dummy variable is a way of capturing a one-time, or recurring, impact in a time series model. It takes on the value of either 0 or.

7 Page of 0 Appendix A contains the historical and forecast load and input variable details. Also shown is the in-sample model forecast of load. The results of the model testing are shown in Appendix B. From the regression model, the forecast of energy usage is determined by applying the model coefficients to forecasts of the input variables. The forecast for heating and cooling degree-day inputs is based on a ten-year historical average of HDD and CDD. A ten-year average was chosen over the 0-year average based on analysis of the annual HDD and CDD data that shows a definite trend (see Figure below). The forecast of Ontario GDP is based on forecasts of 00 GDP growth from the six Canadian chartered banks for 00, and extended over the test years using time-series forecasting methodology. The forecasts of the calendar variables are based on the calendars. 000 Figure Heating and Cooling Degree Days 00 HDD CDD HDD CDD The forecast determined by the regression model is evaluated on a monthly basis against historical monthly load to determine any outlier months. For those months where the

8 Page of 0 0 model predicts load which is beyond historical maximum and minimum bands, a manual adjustment to the model forecast may be applied based on professional judgement. In the case of the current forecast, an adjustment was applied to a number of points throughout the forecast (see Appendix C for details). In total, the manual adjustments adjust the forecast downwards in each of by less than 0. percent of total load, and in the Company s view improve the forecast. The preceding describes the process to develop the kwh load forecast that is used to determine revenues. The company has also provided, in Appendix D, sensitivity analysis around the base load forecast. This analysis provides the impacts on the total load for both a one-percent and a five-percent variance in the degree-day, and a one-percent and a two-percent variance in economic growth parameters of the model. Peak Demand Forecast The forecast of peak demand by customer class, which is used to determine revenue for those customers billed on a demand basis, is established using historical relationships between energy and demand. Appendix E shows the historical and forecast load factors and peak demand forecast. CDM impact on kwh and kw Forecast The load forecast as described above does not explicitly take into account any load impacts arising from CDM programs undertaken by the Company. Savings derived from the planned CDM programs are developed based on the programs, program designs, anticipated take-up and free-rider estimates. Details on THESL s CDM programs are contained in Exhibit G,, Schedules -.

9 Page of CDM savings, by rate class, are shown in Table below. The demand (and revenue) forecast shown in Table above is net of these savings. Table : CDM Savings by Customer Class Year Residential GS<0 GS 0- GS 0-000kW 000kW GS 000- Large Unmetered Street Non- Interval 000kW Users Load lighting Interval Energy Savings (kwh) 00,,,,0 0,,0,,,,000,, 0 00,,0,,0 0,,,,,,000,, 0 00,,,,0 0,,0,,,,000,, ,,,,0 0,,,,,,000,, 0 Demand Savings (kva) ,,, ,0,00, ,,, ,,, 0 0 Customer Forecast The forecast of customer numbers by rate class is shown in Table.

10 Page 0 of Table : Total Customers Year Residential GS<0 GS 0- GS 0- Total 000kW GS 000kW Large Unmetered Street (w/o Non Interval Users Load lighting street Interval 000kW lighting) Total Customers,0,, 0,,,, 000,0,,,,0 0,, 00,, 0,,,,, 00,,,,,,0,0 00 0,0,0,,,0,,,0 00,,0,,,0,,, 00,,,,,0, 0,0, 00,00,0,0,,,,, 00 0,,,,00,, 0,, 00,,0,00 0,00,,,, 00,,0 0,0,,,, ,0, 0,,,,0, Annual Increase 000 -, -,00 0,0, 00,0-0 -,0, 00, ,0, 00, ,0,0 00, - 0 0, 0,0,0 00, ,0 0,,0 00, , 0,, 00, - -,00, 0 0 0,, 00, - -,00, ,, 00 0,0 - -,00, ,00 0, , , 0, Note: Customers numbers as of year-end

11 Page of 0 Customer additions in the company s operating area have been fairly flat over recent history, with about,00 to,00 new customers (excluding Unmetered loads and streetlighting) added annually. On top of natural customer growth forecast based on historical trends, over the forecast period however, the mix of customers is expected to be altered significantly as a result of the Smart Meter program, and specifically the Government of Ontario s Installation of Smart Meters and Smart Sub-Metering Systems in Condominiums legislation which calls for individually-metered multi-residential units. The forecast of customers for the residential sector in 00 through 00 anticipates that new residential condominium buildings in Toronto will incorporate individually-metered units, and thus where previously the load associated with these customers would have been billed as a single General Service customer, it will now be billed as individual residential units. Additionally, it is anticipated that most existing multi-unit buildings will also be converted to individual meters by the end of 00, and the associated load will convert from being bulk-metered to individually-metered.