Short-term CO2 emissions forecasting based on decomposition approaches and its impact on electricity market scheduling

Abstract

The world is facing major challenges related to global warming and emissions of greenhouse gases is a significant causing factor. In 2017, energy industries accounted for 46% of all CO2 emissions globally, which shows a large reduction potential. This paper proposes a novel short-term CO2 emissions intensity forecast to enable intelligent scheduling of flexible electricity consumption to minimize the resulting CO2 emissions. Two proposed time series decomposition methods are developed for short-term forecasting of the CO2 emissions of electricity. These are in turn benchmarked against a set of state-of-the-art models. The result is a new forecasting method with a 48-hour horizon targeted the day-ahead electricity market. Forecasting benchmarks for France show that the new method has a mean absolute percentage error that is 25% lower than the best performing state-of-the-art model. Further, the application of the forecast for scheduling flexible electricity consumption is studied for five European countries. Scheduling a flexible block of 4 hours of electricity consumption in a 24-hour interval can on average reduce the resulting CO2 emissions by 25% in France, 17% in Germany, 69% in Norway, 20% in Denmark, and just 3% in Poland when compared to consuming at random intervals during the day.