Intelligent Light will be in San Jose, CA for the AIAA Joint Propulsion conference. Dr. Earl Duque, IL's manager of applied research is a co-author on a paper "Applications of Dynamic Mode Decomposition and Snapshot POD to Time-Accurate Turbomachinery CFD" which will be presented by Trevor Blanc of Brigham young University.
- Meet with Intelligent Light at JPC! - inquiry/meeting request
- 49th AIAA/ASME/SAE/ASEE Joint Propulsion Conference
Applications of Dynamic Mode Decomposition and Snapshot POD to Time-Accurate Turbomachinery CFD
Trevor J. Blanc and Steven E. Gorrell and Matthew R. Jones
Brigham Young University, Provo, UT, USA
Earl P.N. Duque
Intelligent Light, Rutherford, NJ, USA
Post-processing of both experimental and Computational Fluid Dynamics (CFD) analyses has always been an integral element to understanding the results of a case study. However, with the increase in computing power and development of high frequency experimental measurements, the amount of data to process has grown signicantly. Working with large data sets often poses a problem for users because there is so much information available that sifting through the data requires a lot of time, and any extensive inquiry into flow characteristics can be cumbersome. Thus, the goal of post-processing is to recover the most information using the least amount of time or effort.
Two data analysis techniques, Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD), present themselves as viable techniques that work to filter through large data sets and gather the most relevant flow information. The POD often goes by a variety of names depending on the application, such as Principal Component Analysis, the Karhunen-Love Decomposition, and Singular Value Decomposition. Using the POD method of snapshots as presented by Sirovich,. one can isolate the main modes from which a flow is constructed. Within these modes contain statistical information that define the coherent flow structures that relate to the physical phenomena present. The DMD is a relatively new technique developed by Schmid that serves to isolate the dynamic modes present within a data set. It is a way to observe the modes characterizing the evolution of the flow through time as well as isolating the coherent structures present in the flow field.
These analysis techniques have been applied to PIV cases and both simple and turbulent CFD flows but have yet to be applied to large, unsteady, and time-accurate turbomachinery simulations. Therefore, it is advantageous to observe the relevance of the POD and DMD within the realm of these highly complex flows.