CuWind – CUDA (GPU)-Based High-Fidelity Wind Farm Simulator
Date:
In this workshop, I have introduced how to use CUDA to complete the Wind Farm Simulation to our collaborators with GE.
Why wind farm simulation is important? The wake of upstream wind turbine can greatly affect the wind on the rotation plane of downstream turbines, decreasing the power generation of wind farms. To solve it, we need to develop wind farm control techniques to mitigate the wake effect. Meanwhile, we also expect to reduce the fatigue load on blades, ensuring the safety operation of wind turbines. For offshore wind turbines, we may also need to control the floater at the bottom, reducing its vibration. What’ more, the power system also asks the wind farm to offer the auxiliary service. For example, when there is a frequency drop in the grid, wind farm should increase the active power output the help recovery the grid frequency. To achieve these applications, it is necessary to develop a high-performance wind farm simulator.
I introduced the wind field model (N-S equations) and wind turbine model (actuator line model, ALM) to them. By adding the aerodynamical force calculated via ALM to the N-S equations, it can simulate the operation of the entire wind farm.
Results show that the time taken by the GPU and CPU to complete the calculation. Overall, the calculation on a parallel CPU is three times faster than on a single-process CPU. Additionally, the speed on the GPU is more than 17 times faster than on the CPU.