Science Stories

Discover More about XSEDE-enabled science, programs, trainings, and more.

Key Points
See how XSEDE supports science
« Back

Supercomputer Simulations Help Optimize Floating Wind Farms

 XSEDE-Allocated Supercomputer Models Provide Insight on Turbine Wakes

A large eddy simulation of the wake behind a floating offshore wind turbine. Credit: Hannah Johlas, University of Massachusetts Amherst

 

Over the past few years, offshore wind farms have emerged across the world as a viable source of energy.  While these powerful floating wind turbines are typically more than 800 feet tall and can weigh some 50 tons or more, they reduce land-use concerns, access better offshore wind resources, and can generate more power than land wind turbine farms. However, they present a complex engineering design problem: how can they be optimized to operate in the uniquely challenging offshore environment?

In a Journal of Physics: Conference Series paper published this summer called Large eddy simulations of floating offshore wind turbine wakes with coupled platform motion, researchers from the University of Massachusetts Amherst and the National Renewable Energy Laboratory (NREL) discussed their efforts to advance our knowledge of this issue. 

Specifically, the Comet supercomputer at the San Diego Supercomputer Center (SDSC) and the Stampede2 supercomputer at the Texas Advanced Computing Center (TACC) were used to perform simulations that showed how floating turbine wakes are very similar those of fixed-bottom turbines, except that floating turbine wakes are deflected upward and have slightly stronger turbulence at the edge of their wakes.

"We looked at how these wake effects can be accurately considered when designing floating offshore wind farms," said lead author and National Science Foundation (NSF) Graduate Research Fellow Hannah Johlas. "At about 20,000 computer processor hours (per run), these high-fidelity large eddy simulations are very computationally intensive and expensive, and as such, this research can only be performed using supercomputers."

The collaborative study aimed to better understand how the wake effects of large wind farm arrays decrease power output and reduce the lifespan of the turbines. Because of the growing prospect for floating wind farms, the researchers focused on the differences between floating turbine wakes and fixed turbine wakes.

The large eddy wind turbine simulations were completed with Comet and Stampede2 using the computational fluid dynamics software Simulator fOr Wind Farm Applications (SOWFA), coupled with the wind turbine modeling tool OpenFAST for the platform and turbine motion. The downstream wake characteristics of the floating platform were compared to equivalent fixed platform cases for different wind speeds, wave heights, wind-wave alignments, and turbine yaw angles.

Overall, the differences in wake shape between floating and fixed platforms were associated with mean platform displacements, while differences in turbulence were associated with time-varying platform motion. However, these observed wake differences between fixed and floating platforms were found to be quite small, especially for higher wind speeds and lower wave heights.

"With global-installed capacity of offshore wind increasing from 8.9 gigawatts in 2015 to 22.5 gigawatts in 2018, this research is becoming even more prevalent and now that we know more about how wakes behave for floating turbines, we will examine how those floating-turbine wakes affect downstream turbine power generation and structural loading," said Johlas, who has focused her Ph.D. in mechanical engineering at the University of Massachusetts at Amherst on this project.

"Comet and Stampede2's reliability and computing environment flexibility helped complete this research in a time-efficient manner," added Johlas. "As this research was funded by an NSF Graduate Research Fellowship, there were no funds available for purchasing supercomputing time, so access to XSEDE supercomputers really enabled this research to happen at all. Also, XSEDE's support team helped solve environment setup and file system usage issues for us."

Johlas' research group, led by PI David Schmidt, previously used XSEDE-allocated resources, including Comet and Stampede2, for several years.

"XSEDE makes available some of the most reliably useful high-performance computers available to university researchers in my field," she said. "Since the progress of my research directly depends upon reliable access to supercomputing time, resources such as XSEDE enable researchers like me to envision more accurate, more complex simulations than are possible when using lower-fidelity tools." 

The research was funded by an NSF Graduate Research Fellowship, grant #1451512. The project used the Extreme Science and Engineering Discovery Environment (XSEDE) program funded by NSF grant #ACI-1548562, as well as NREL computational resources sponsored by the Department of Energy's Office of Energy Efficiency and Renewable Energy.