Lynna Deng

Lynna Deng

Biography

Lynna Deng is a third-year student at Northwestern University pursuing a B.Sc. inMechanical Engineering and Environmental Sciences. She is driven to preserve the environment and address global energy challenges through research on the fluid dynamics of renewable energy technology. She pursues research in the Aristilde lab at Northwestern University and gained additional research experience in environmental fluid dynamics and computer vision in the Luhar lab as a University of Southern California SURE fellow. At MIT, she is working with ProfessorMichael Howland to model wind farm wakes and create reliable, low-computational-cost tools for wind farm developers to optimize power production. She strongly believes that research insights and, more broadly, information about energy should be accessible and beneficial to all. Thus, Lynna serves as the executive director of the Chicago Energy Conference, a new collaboration between Northwestern University and the University of Chicago bringing free conversations across energy technology, policy, and finance to Chicago, as well as the co-president of Engineers for a Sustainable World, a barrier-free club for engineering students to gain their first hands-on experience with sustainability-centered projects. Beyond her academic pursuits, she enjoys testing chocolate chip cookie recipes, running along Lake Michigan, and attempting crossword puzzles.

Abstract

Extending Wind Turbine Wake Models for Wind Farm Wake Modeling

As wind turbines extract energy from the wind, they generate low-energy, low-wind-speed regions downwind known as wind turbine wakes. A wind turbine inside the wake of another turbine experiences slower, less energetic wind and thus produces less power. Analytical engineering wake models such as the Turbulence Optimized Park (TurbOPark) model are commonly used to estimate wake losses due to their low computational cost, but they rely on flow simplifications and empirically fit parameters that may not generalize beyond the turbine or farm scenarios to which they were fit. In this project, we compare the TurbOPark model to large-eddy simulations (LES), more computationally expensive models that capture complex flow phenomena. By running LES of the same wind farm in different atmospheric conditions, we demonstrate that the empirically fit TurbOPark parameters do not generalize across operating conditions. Furthermore, refitting the parameters to minimize error in wind farm power prediction results in underpredicting the impact on neighboring wind farms. Using the flow properties predicted by LES, we evaluate the flow property relationships that underlie TurbOPark. In our ongoing work, we modify the TurbOPark model form based on the flow property relationships observed in LES, creating an updated tool for wake loss modeling.