A parallel computational code for the Proper Orthogonal Decomposition of turbulent flows

A parallel computational code for the execution of the Proper Orthogonal Decomposition (POD) of turbulent flow fields is developed. The POD is an analytically founded statistical technique that allows the extraction of appropriately defined modes of the flow from the background field, allowing the determination of the coherent structures of turbulence. The technique operates through the selection of a basis for a modal decomposition of a set of signals, to be processed with the use of the spectral theory of compact self-adjoint operators. Details are given on computational aspects with regard to the different phases of the numerical method and the development of the related parallel code is described. Computational tests corresponding to different domains and number of processors are executed on a HP 9000-V2500 parallel computer and the results are shown in terms of parallel performance of the different phases of the calculations separately considered, and of the computational code as a whole.