Dynamic Models for Short-Term Wind Forecasting

Web Published:

Dynamic Models for Short-Term Wind Forecasting


Princeton Docket # 15-3145


Researchers in the Department of Civil and Environmental Engineering at Princeton University have developed a fundamentally new approach for short-term wind power forecasting. The new wind prediction model relies on a data-driven but physically-based approach that has successfully passed initial proof of concept tests. It is based on the extrapolation of the slowly-varying large scale forcing pressure gradients (rather than wind speed directly as is currently done) into the future, and then inferring the wind speed. This novel model can significantly reduce errors in short-term wind forecasts, which have been shown to be more critical to wind energy applications than day-ahead forecasts due to the limits of real-time response options. For this reason, it will have a transformative impact on the ability to handle variability and uncertainty of wind. Its incorporation into the energy market model will open new possibilities for optimizing grid integration.


The new wind forecasting model will benefit any application for which short term wind forecasting, within the next 1 to 5 hours, is needed. In the wind energy sector, wind power output varies as wind speed rises and falls, yielding uncertainties in the energy production. Hence, predicting the wind variability and forecast uncertainty is a critical component of managing the power systems to keep the electric supply and demand in proper and favorable balance. Wind farm operators obtain 24-hour forecasts from the National Weather Service or other sources using numerical models. These forecasts often have errors that reduce the efficiency of the wind farm since they either underestimate or overestimate the generated power in the next 2 hours, and thus lead to wasted power or a power shortage. To alleviate this problem, wind farm operators employ more accurate predictions such as statistical methods for forecasting the next 2 hours to plan their short-term operations, but these short-term predictions continue to also have errors. These statistical methods do not consider the underlying physical properties of the atmosphere and they have to be trained and calibrated for each site individually. The novel forecasting approach will significantly improve those short-term wind predictions, using a unique data-driven analytical model. Compared to previous works, this model is easier to generalize and more accurate than statistical approaches, and much faster than the fully 3D numerical weather prediction models with significant reductions in the prediction errors. It has been tested using some measurements in various conditions and the results indicate improved performance over the most widely used model at present in 70 to 90 % of the periods, with a reduction in errors averaging 66%. The adoption of the model can thus improve electricity market clearing, economic load dispatch planning, generation and load increment/decrement decisions, and the regulatory framework for wind energy.



Wind energy forecasting performed by:

•       Transmission system operators (TSOs)

•       Independent power producers (IPPs)




•       Significantly reduces errors in short term wind forecasting (about 66% reduction)

•       Will allow for more efficient wind power generation

•       More accurate and less computationally expensive (much faster) than current forecast models

•       Anchored to a robust physical model of wind variability that makes it more general and location-independent



Elie Bou-Zeid and Mostafa Momen, “Large-Eddy Simulations and Damped-Oscillator Modeling of Unsteady Ekman Boundary Layers” National Center for Atmospheric research, Boulder Colorado, November 2014.


Key Words

Wind energy, Short-term wind forecasting, Wind variability, Wind Farm, Energy Markets




Elie Bou-Zeid is an Associate Professor in the Department of Civil and Environmental Engineering and an Associated Faculty member in the Department of Mechanical and Aerospace Engineering at Princeton University. He obtained his Ph.D. in Environmental Engineering at Johns Hopkins University. Before coming to Princeton University in 2008, Bou-Zeid was a postdoctoral researcher at the Swiss Federal Institute of Technology. Bou-Zeid is a recipient of the E. Lawrence Keyes Jr./Emerson Electric Co. Faculty Advancement Award intended to recognize and assist promising junior faculty members at Princeton University, an award that comes with $40,000 unrestricted research funding.

Professor Bou-Zeid's current research focuses on combining numerical, experimental, and analytical tools to study the basic dynamics of flow and transport in environmental systems. The aim is to study how Environmental Fluid Mechanics relate to problems in climate change, air quality, hydrology, and sustainable development.


Mostafa Momen is a Ph.D. candidate at Princeton University in Civil and Environmental Engineering. During the academic year of 2014-2015, he was an exchange scholar in Mechanical Engineering at Massachusetts Institute of Technology. Momen earned his B.Sc. in Civil and Environmental Engineering in 2012 from Sharif University of Technology. Momen was awarded the Princeton University Gordon Y.S. Wu Fellowship in Engineering in 2012 and has two officially registered inventions in Iran’s Office for National Registration of Industrial Ownership. In 2006, he was awarded the best prize of the Iran’s national Kharazmi festival for his novel project among over 20000 inventions.


Intellectual Property Status

Patent applications are pending. Princeton is seeking industrial collaborators for further development and commercialization of this technology.



Michael Tyerech

Princeton University Office of Technology Licensing • (609) 258-7256• tyerech@princeton.edu


Anna Trugman

Princeton University Office of Technology Licensing • att@princeton.edu



Patent Information:
For Information, Contact:
Chris Wright
Licensing Associate
Princeton University
Elie Bou-Zeid
Mostafa Momen