Project Details
Projekt
Data analysis and flow modeling
Applicant
Dr. Richard Semaan
Subject Area
Fluid Mechanics
Term
from 2020 to 2024
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 406435057
Buffet is a byproduct of shock-boundary layer interaction in the transonic regime. Coupled with three-dimensional geometry (swept wing, nacelle, pylon), the buffet phenomenon yields a complex non-linear flow which is challenging to analyze and to model. Analyzing the flow results, particularly the large-scale LES and DES simulations that are planned in this research unit, requires dedicated tools to handle the large data as well as to extract the relevant information. Large data create both storage and an analysis problem. Existing approaches to reduce the data size typically rely on selective sampling and on filtering. In this project, the so-called Sparse Spatial Sampling (S3) method is to be extended and applied in order to reduce the large numerical grid sizes and to compress the simulation results. This, in turn, enables better data management and simplified evaluation of the results. Data analysis methods are numerous. They include proper orthogonal decomposition (POD) and dynamic mode decomposition (DMD), which are currently the two most powerful and popular tools in fluid dynamics. Both spectral methods shall be further developed and adapted in this project as part of the xROM toolbox, which shall be shared with the entire research group. In addition to the adaptation and sharing of the xROM, TP2 will closely work with the FOR partners to interpret the results. Modeling fluid flows provide physical insights as well as the ability to perform predictions. In this project, two modeling strategies will be considered and extended: The dynamic mode decomposition (DMD) and the cluster-based reduced models (CROM). The DMD identifies and quantifies the relevant modes and their stability. Two DMD variants (multi-resolution DMD and randomized DMD) shall be combined to better identify the range of spatiotemporal modes and to enable calculations on very large datasets. The results of the DMD stability analysis shall be compared with those of the cooperation partner Dr. Sebastian Timme, who uses the global stability analysis. In addition, two cluster-based approaches shall be used for analysis and modeling: the Metric Attractor Overlap (MAO) and the Cluster-based Reduced Models (CROM). These data-driven methods are well suited to characterize and to quantify the differences between different flow conditions (or different types of simulations) and to model nonlinear flows.
DFG Programme
Research Units
International Connection
France, United Kingdom, USA
Cooperation Partners
Professor Steven L. Brunton, Ph.D.; Professor Dr. Bernd Rainer Noack; Dr. Sebastian Timme
