Can data science help resolve the power grid's operational challenges? Evidence from Ireland, Great Britain, and continental Europe.

The ESRI organises a public seminar series, inviting researchers from both the ESRI and other institutions to present new research on a variety of public policy issues. The seminar series provides access to specialised knowledge and new research methodologies, with the objective of promoting research excellence and facilitating productive dialogue across the policy and research fields.

Join the conversation #ESRIevents

Guest speaker: Kevin Forbes, Associate Professor of Economics in the Department of Economics, The Catholic University of America, Washington DC

Seminar topic: The integration of wind and solar energy resources into the power grid has brought about environmental benefits but has also made it more challenging for system operators to balance electricity supply with demand. For example, the increasing utilization of distributed energy resources (DERs) such as rooftop solar has reduced the accuracy of the load forecasts that system operators use to optimize the resources of the power grid. Moreover, solar and wind energies are difficult to predict using conventional methods accurately.   The presentation begins by documenting this general point using data from Ireland, Great Britain, and Continental Europe. Specifically, evidence is presented that the energy imbalances associated with wind energy in Great Britain are significantly larger than the imbalances corresponding with electricity generation that uses coal or natural gas as fuels.

The presentation proceeds by presenting a modeling framework to improve the accuracy of the load,  wind energy, and solar energy forecasts. The modeling framework is also applied to high voltage transmission flows, an area of increasing importance that nevertheless can give rise to operational challenges when the actual electricity flows into an importing control area are less than the scheduled imports.  

A central feature of the modeling framework is that the errors in the standard approach to forecasting do not have the property of “white noise.” In other words, the errors are systematic. By incorporating this systematic component in a model, more accurate short-run predictions can be generated.

Speaker bioKevin Forbes is a Visiting Researcher at ESRI and an Associate Professor of Economics in the Department of Economics at The Catholic University of America in Washington DC. He contributed to the design and development of the United States Government’s National Energy Modeling System (NEMS).  His efforts primarily focused on the application of econometrics to enhance the modeling system’s forecasting capabilities with respect to oil and natural gas extraction.  His current research interests include electricity load forecasting, solar energy forecasting, wind energy forecasting, electricity loop flow issues, electricity balancing market issues,  oil market issues, space weather power grid issues,  and the issue of causality between atmospheric concentrations of greenhouse gases and meteorological outcomes. Most of his current research makes use of time-series econometric methods. Publications include papers in The American Economic Review, The Energy Journal, Energy Policy, The Space Weather Journal, The Journal of Law, Economics, and Organization, and The Electricity Journal.