Background
Many renewable energy sources produce electricity intermittently and unpredictably throughout the day. In contrast, the demand for electricity varies hourly in a predictable fashion. Ensuring that renewable energy production matches energy needs requires a model of hourly demand. Knowing hourly demand is important because it allows predicting gaps between energy use and energy production. However, electricity demand that is available publicly only exists at an annual resolution. There are hourly residential data, originating from smart meters, but these data are property of utility companies. The Canada Energy Regulator (CER) has a long history of publishing energy supply and demand projections for Canada at an annual resolution, in a report released yearly (Energy Futures), and is interested in addressing this research question.
Data Description
The CER provides the provincial-level hourly demand data for all sectors aggregated (residential + industrial + commercial/institutional + agriculture + transportation). This data is publicly available for Ontario from the Independent Electricity System Operator (IESO).1 The CER also provides annual demand data for each sector. Data is from Natural Resources Canada (NRCan).2 Hourly air temperature and weather data (precipitation, snowfall, snow mass, air density, ground-level solar irradiation, top of atmosphere solar irradiation, cloud cover fraction) are available from ETH Zurich and Imperial College London.3
The goal of this case study is to develop a model of hourly electricity demand in the residential sector, in Ontario. To that end, participants are given a set of annual and hourly variables. Which analytical approach is taken is to the discretion of participants.
We focus on residential demand because this sector is the most well-understood of all sectors—other sectors being industrial, commercial/institutional, agriculture, and transportation. In the residential sector it is known what components, termed end uses, drive annual electricity demand (for example, cooling, heating, and others). We are also interested in methods how to derive a maximal utilization of existing information or overcome the lack of knowledge when developing a model.
The model performance can be evaluated by leaving one year out (leave one out cross-validation). The sector-specific annual data can be estimated by summing the predicted hourly data for a year based on the built model. The accuracy of the estimated sector-specific annual data will be calculated using mean absolute error (MAE):
Where n is the number of years and y is the sector-specific annual data.
Hourly Demand 2003-2016 (SSC2020_hourly_demand.xslx)1
Variable |
Unit |
Description |
Total Energy Use from Electricity |
megawatt (MW) |
Hourly electricity usage for all sectors combined (residential, industrial, commercial, agriculture, transportation) |
Annual Demand 2003-2016 (SSC2020_annual_demand.xlsx)2
Variable |
Unit |
Description |
Total Energy Use from Electricity (residential) |
petajoule (PJ) |
Annual electricity usage for the residential sector |
Energy Use from Electricity by End-Use (residential): |
||
Space Heating |
petajoule (PJ) |
Annual electricity usage for the residential sector for heating |
Water Heating |
petajoule (PJ) |
Annual electricity usage for the residential sector for heating water |
Appliances |
petajoule (PJ) |
Annual electricity usage for the residential sector for appliances |
Lighting |
petajoule (PJ) |
Annual electricity usage for the residential sector for lighting |
Space Cooling |
petajoule (PJ) |
Annual electricity usage for the residential sector for cooling |
Total Energy Use from Electricity (industrial) |
petajoule (PJ) |
Annual electricity usage for the industrial sector |
Total Energy Use from Electricity (commercial) |
petajoule (PJ) |
Annual electricity usage for the commercial sector |
Total Energy Use from Electricity (agriculture) |
petajoule (PJ) |
Annual electricity usage for the agriculture sector |
Total Energy Use from Electricity (transportation) |
petajoule (PJ) |
Annual electricity usage for the transportation sector |
Hourly Weather 2003-2016 (SSC2020_hourly_weather)3
Variable |
Unit |
Description |
precipitation |
millimeter per hour (mm/hour) |
Amount of drizzle, rain, sleet, snow, graupel, or hail per time |
temperature |
degree Celsius (°C) |
|
irradiance surface |
watt per square meter (W/m^2) |
Amount of power per area from the Sun on the Earth's surface |
irradiance toa |
watt per square meter (W/m^2) |
Amount of power per area from the Sun at the top of the atmosphere |
snowfall |
millimeter per hour (mm/hour) |
Amount of fallen snow per time |
snow depth |
centimeter (cm) |
Total amount of snow on exposed ground |
cloud cover |
fraction [0,1] |
Portion of sky that is covered in clouds |
air density |
kilogram per cubic meter (kg/m^3) |
Mass per unit volume of Earth's atmosphere |
The data set can be downloaded from here:
Please email Dr. José Ribas Fernandes jose.ribasfernandes@cer-rec.gc.ca if you have any questions about the data set.
Organizers
Dr. José Ribas Fernandes and Dr. Ryan Hum (Data and Information Management, Canada Energy Regulator)
Mr. Mantaj Hundal, Mr. Lukas Hansen, Mr. Michael Nadew, and Mr. Matthew Hansen (Energy Outlooks, Canada Energy Regulator)
Dr. Chel Hee Lee (Department of Mathematics and Statistics, University of Calgary)
Acknowledgment
We thank Dr. Pingzhao Hu (Department of Biochemistry and Medical Genetics, University of Manitoba) and Dr. Ehsan Karim (School of Population and Public Health, University of British Columbia) for their feedback in preparing this case study.
1 Independent Electricity System Operator. (2019). Hourly demand report. Retrieved from http://reports.ieso.ca/public/Demand/
2 Natural Resources Canada. (n.d.). Comprehensive energy use database. Retrieved from http://oee.nrcan.gc.ca/corporate/statistics/neud/dpa/menus/trends/comprehensive_tables/list.cfm
3 Pfenninger, S., & Staffell, I. (2019). Renewables ninja. Retrieved from https://www.renewables.ninja/