Dynamic Mode Decomposition (DMD) is a useful tool to effectively extract the dominant dynamic flow structure from a unsteady flow field. However, DMD requires massive computational resources with respect to memory consumption and the usage of storage. In this paper, an alternative incremental algorithm of Total DMD (Incremental TDMD) is proposed which is based on Incremental Singular Value Decomposition (SVD). The advantage of Incremental TDMD compared to the existing on-the-fly algorithms of DMD is that Sparsity-Promoting DMD (SPDMD) can be performed after the incremental process without saving huge datasets on the disk space. SPDMD combined with Incremental TDMD enable the effective identification of dominant modes which are relevant to the results from conventional TDMD combined with SPDMD.
In this research, we conduct unsteady CFD to investigate the effect of engine bay flow on the steady and unsteady aerodynamics of the extended DrivAer model reproducing engine bay flow. Dynamic Mode Decomposition (DMD) is performed to analyze unsteady aerodynamics. As a result, it is revealed that different engine bay setups results in not only differences of steady aerodynamics but also the differences of unsteady aerodynamic characteristics. Furthermore, we perform an on-the-fly algorithm of DMD called Streaming Total DMD (STDMD) which can be conducted with much less memory than conventional DMD to investigate the relevancy and applicability of STDMD on the analysis of unsteady aerodynamics of a road vehicle.
Temporally resolved flow fields are commonly averaged in time, and mostly the time-averaged flow fields and forces are used for the aerodynamic optimization of road vehicles. Online DMD is found to be well suited for studying transient flow effects and leads to a deeper understanding of the complex flow around the vehicle. The investigated velocity field is computed by a Detached Eddy Simulation of the DrivAer reference body. The CFD setup and key considerations for the application of online DMD on large data sets are outlined, and the most dominant extracted coherent flow structures are analyzed independently.
It is believed that Dynamic Mode Decomposition (DMD) is a very useful method for the analysis of unsteady aerodynamics of road vehicles. However, it seems that the conventional DMD method is not practical regarding the application on the aerodynamic design of road vehicles, since DMD computation requires massive memory. Alternatively, online DMD methods seem to be useful in practice, as those require much less memory than the conventional method. In this paper, further validation of the online DMD on the aerodynamic forces on the DrivAer model is conducted, through the comparison with results from other enhanced DMD methods and FFT.
With the increase of available computer performance, unsteady Computational Fluid Dynamics (CFD) is now widely used for industrial applications. For the analysis of unsteady vehicle aerodynamics, massive data storage is required for saving time series of spatially highly resolved flow fields. The size of these transient datasets can be significantly reduced using the Incremental Proper Orthogonal Decomposition (POD) by computing POD modes in parallel to the CFD. In this paper, we present a successful approximation of the transient flow field using a reduced number of modes computed by Incremental POD.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.