Spatio-Temporal Downscaling of Energy Demand Forecasts

Use of data from this repository must be cited as:

Cohen, Elliot, Vijay Modi, Henri Torbey, Michael Piccirelli and Yu-Tian (2015).  
Global Trends in Urban Energy Use. Working Paper of the Sustainable Engineering Lab,   
Columbia University, February 2015. Available online:  
http://ecohen4.github.io/Energy/Global_Trends_v4.html

Global energy forecasts such as the annual IEA World Energy Outlook inform energy planning at the highest levels of government and industry. These reports directly influence investment decisions today that will lock-in carbon emissions and environmental impacts for decades to come. However, such mid- to long-range demand forecasts are typically reported as annual totals and provide little insight to the temporal distribution throughout the year. This leaves a huge gap in investment decisions, particularly in the power sector where demand is non-uniform, growth is non-linear, and cost is driven by capital expenditures, which in turn are driven by peak demand, not integral energy consumption.

Peak electrical demand for a power control area may represent just a few days of the year, meaning that capital utilization is low and return on investment is long. Thus accurate long-range peak demand forecasts are essential for proper investment decisions today. This is particularly true for renewable energy, where utilization rates are already low due to non-dispatchability (wind) and limited availability throughout the day (solar).

To address this shortcoming, the Sustainable Engineering Lab at Columbia University is leading a study of diurnal and seasonal patterns of energy demand for a large number of global cities. This requires more and better data than is typically available to researchers, and thus we have embarked on an ambitious, multi-year project to cultivate a diverse network of electricity grid operator partners throughout the developing world. Data continues to pour in, strengthening our analyses. This will be a key area of research in 2015 and beyond.