Spectral analysis of OH-PLIF snapshots of a bluff body wake
Date
2021-12
Authors
Subramani, Harish
Major Professor
Advisor
Meyer, Terrence
Michael, James
Fox, Rodney
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Abstract
Advancements in laser based diagnostic measurement techniques have enabled the procurement of high-speed flame measurements such as chemiluminescence and planar laser-induced fluorescence. Availability of high-dimensional image sequences in turn leads to questions concerning mathematical tools that are capable of extracting fluid dynamic information. In this regard, spectral techniques such as the discrete Fourier transform, continuous wavelet transform, proper orthogonal decomposition, and dynamic mode decomposition are quite popular. In this work, these tools have been utilized on OH-PLIF snapshots to discern large-scaled patterns and their dynamics in the wake of a V-gutter flame holder. The fuel in these experiments is gaseous propane and forms a premixed mixture with air. Different conditions are studied by varying the equivalence ratio from 0.70 (fuel-lean) to 1.20 (fuel-rich), while maintaining the air mass flow rate at 0.4 lbm/s. It is found that the bluff body wake is prone to large-scaled, narrow-band, shedding patterns. This behavior agrees with findings of other scholars who have noted the excitation of thermoacoustic instabilities in this combustor rig. The 1-D discrete Fourier transform reveals a linear relationship between the frequency of the large-scaled structures and equivalence ratio. The 1-D continuous wavelet transform is able to show that the temporal prevalence of these large-scaled structures is diminished beyond the stoichiometric ratio. Both proper orthogonal decomposition and dynamic mode decomposition reveal the presence of symmetric patterns across the width of the wake. This indicates that the large-scaled shedding patterns are not associated with Von-Karman type of behavior. Wavelet coherence is a bivariate technique that quantifies the linear spectral relationship between two time series. By selecting pixels on either side of the bluff body wake, wavelet coherence has been used to quantify the large-scaled symmetry in the wake as a function of time. It is shown that the wake exhibits a predominantly symmetric pattern in the case of equivalence ratio equal to 0.70. During such cycles, values of coherence are high and close to 1. Other cycles exist with low values of coherence. During these cycles, the shedding pattern is not strongly symmetric. In fact, one cycle is detected wherein no large-scaled shedding pattern is visible. During this cycle, small-scaled Kelvin-Helmholtz type instabilities prevail. Overall, this work shows that various mathematical tools should be deployed on high-dimensional, scalar-valued, high-speed flame measurements so as to gain complementary and supplementary information.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
thesis