Analysis of numerically solved large-scale periodic fluid flows

Date of Completion

January 2004


Engineering, Mechanical




Periodic flows are those unsteady flows in which the flow variables have a repeated behavior over time. Flow periodicity may occur naturally as in the case of flow over bluff bodies, or artificially as in the case of flows forced at specific frequencies. Many engineering flow applications exhibit some sort of periodicity such as in turbomachinery, jet combustors, internal combustion engines, etc. The periodic flow may be associated with laminar or turbulent fluid flows. ^ Unsteady numerical simulations have been generated for several relevant physical problems. Two convergence criteria for numerical solutions of periodic fluid flows have been developed and applied, one based on the Fast Fourier Transform and the other on the Proper Orthogonal Decomposition. This study is also concerned with analyzing the unsteady numerical simulations of periodic flow and considers a variety of parameters to examine these simulations, e.g. the frequency content, mixing performance, and entrainment. The simulations are validated against available data. Extracting the coherent structures, data reduction, and relationships between frequency spectra and the modes of reconstruction of the flow variables are also addressed. Scalability, which extracts the solution of a new case from previously solved cases, is studied using three different methods. ^ Although most of these techniques are general and can be applied to any numerically solved periodic fluid flow problem, three specific problems are investigated. The first problem is jet in crossflow where many cases have been numerically solved ranging from a steady jet in a steady crossflow to forcing of the jet and/or the mainstream with sinusoidal waveforms. The second problem is flow over a triangular bluff body. The third problem is flow over an open cavity. ^ The new two convergence criteria based on FFT and POD are proved to be easily applicable to assess convergence of unsteady numerical solutions of periodic fluid flow problems and can identify directly some of the physical characteristics of the fluid flows. Application to the analysis of the jets in cross flow, both steady and forced cases, demonstrated that forcing enhances mixing and penetration. Furthermore, the primary flow can be predominantly driven by super-harmonics of the input forcing frequencies. The POD and FFT analyses can be used to reduce the large unsteady data sets to more measurable sizes. A scalability analysis was developed and successfully used to predict the flow field of new cases from the numerical solutions of other cases of the same geometry. ^