Presentations

Abstract

Fast simulations of wind turbine wakes are crucial during the design phase of optimal wind farm layouts as well as for the active wake control. Physics-informed neural networks (PINNs), a deep learning approach to simulate dynamical systems governed by partial differential equations, are gaining traction in computational fluid dynamics due to their fast inference capability. This study presents a physics-informed neural network implementation for simulating wake behind multiple wind turbines in a windfarm. Our research focuses on simulating turbulent flows in a mesh-free environment using physics guided neural network methodology. We developed a PINN model using the 2-equation 𝑘 - 𝜖 model and the actuator disc method to simulate the wakes behind the wind turbines. The novelty of the research is that the developed model is trained from scratch without relying on experiment data or simulation data from traditional numerical solvers. The developed PINN framework can give quick (near real-time) inference using a trained model in contrast to traditional high-fidelity numerical solvers. The study shows that the developed model can simulate various turbulent flows. Two different wind farm scenarios are simulated. Our experiments reveal that the results obtained from the PINN models are in good agreement with the experimental data and numerical simulation. This current work serves as a technology demonstrator towards real-world engineering applications like turbulent flow and wake modelling in a windfarm.