Speaker
Description
Computational fluid dynamics (CFD) is widely employed in various disciplines to investigate flow fields, i.g., computing the drag coefficient of a road vehicle [1] or analyzing respiratory flows [2]. However, depending on the complexity of a problem or the desired accuracy, employing CFD simulations can be time consuming and costly. Recently, machine learning (ML) techniques have shown great potential in learning flow patterns [3, 4], predicting flow parameters [5] or fields [6], or controlling flow [7]. In this work, examples for combining CFD and ML for solving problems from environmental and energy
sciences are presented. These examples are based on collaborations between Korean and German universities and research institutes.
In the first example, an ML model is trained with flow data to predict the track and intensity of typhoons [8, 9]. It is explained how different combinations of meteorological and observational training data influence ML-based predictions, and how training improves when including a larger dataset with data from more cyclones from the past. Additionally, it is shown how those predictions can compete with numerical predictions from forecasting centers, while consuming only a fraction of the resources needed to compute the predictions. In the second example, it is shown how incorporating spatio-temporal data from potential wind farm sites can improve the precision of an ML-based wind forecasting model [10]. The wind forecasts can be used to control the yaw angle of a turbine and increase power generation [11]. Furthermore, spatio-temporal correlations of wind data are analyzed to evaluate locations for wind farms. In particular, locations of two existing off-shore wind farms in the USA and the UK are compared to a location for a potential off-shore wind farm in South Korea.
References
[1] M. Rüttgers, J. Park, D. You, Large-eddy simulation of turbulent flow over the
drivaer fastback vehicle model, Journal of Wind Engineering and Industrial Aero-
dynamics (2017).
[2] H. Aljawad, M. Rüttgers, A. Lintermann, W. Schröder, H. J. Lee, Effects of the
nasal cavity complexity on the pharyngeal airway fluid mechanics: A computational
study, Journal of Digital Imaging (2021).
[3] S. Lee, D. You, Data-driven prediction of unsteady flow over a circular cylinder
using deep learning, Journal of Fluid Mechanics 879 (2019) 217–254.
[4] S. Lee, D. You, Analysis of a convolutional neural network for predicting unsteady
volume wake flow fields, Physics of Fluids 33 (3) (03 2021).
[5] S. Lee, J. Yang, P. Forooghi, A. Stroh, S. Bagheri, Predicting drag on rough surfaces
by transfer learning of empirical correlations, Journal of Fluid Mechanics 933 (2022)
A18.
[6] M. Rüttgers, S.-R. Koh, J. Jitsev, W. Schr ̈oder, A. Lintermann, Prediction of
acoustic fields using a lattice-boltzmann method and deep learning, in: H. Jagode,
H. Anzt, G. Juckeland, H. Ltaief (Eds.), High Performance Computing, Springer
International Publishing, Cham, 2020, pp. 81–101.
[7] M. Rüttgers, M. Waldmann, W. Schr ̈oder, A. Lintermann, Machine-learning-based
control of perturbed and heated channel flows, in: H. Jagode, H. Anzt, H. Ltaief,
P. Luszczek (Eds.), High Performance Computing, Springer International Publish-
ing, Cham, 2021, pp. 7–22.
[8] M. Rüttgers, S. Lee, S. Jeon, D. You, Prediction of a typhoon track using a gener-
ative adversarial network and satellite images, Scientific reports 9 (1) (2019) 1–15.
[9] M. Rüttgers, S. Jeon, S. Lee, D. You, Prediction of typhoon track and intensity
using a generative adversarial network with observational and meteorological data,
IEEE Access 10 (2022) 48434–48446.
[10] H. Shin, M. Rüttgers, S. Lee, How regional wind characteristics affect cnn-based
wind predictions: Insights from spatiotemporal correlation analysis, Under revision
at Energy (2023).
[11] H. Shin, M. Rüttgers, S. Lee, Neural networks for improving wind power efficiency:
A review, Fluids 7 (12) (2022) 367.
| Keywords | Machine learning, typhoons, wind prediction, wind energy |
|---|
