Date of Award

5-30-2014

Degree Type

Dissertation

Embargo Date

5-30-2015

Degree Name

Doctor of Philosophy (PhD)

Department

Mechanical and Aerospace Engineering

Advisor(s)

John F. Dannenhoffer

Second Advisor

Mark N. Glauser

Keywords

Airfoil, Data Fusion, Fluid Dynamics, Jet Flow

Subject Categories

Engineering

Abstract

In recent years, there have been improvements in the methods of obtaining fluid dynamic data, which has led to the generation of vast amounts of data. Extracting the useful information from large data sets can be a challenging task when investigating data from a single source. However, most experiments use data from multiple sources, such as particle image velocimetry (PIV), pressure sensors, acoustic measurements, and computational fluid dynamics (CFD), to name a few. Knowing the strengths and weaknesses of each measurement technique, one can fuse the data together to improve the understanding of the problem being studied. Concepts from the data fusion community are used to combine fluid dynamic data from the different data sources. The data is fused using techniques commonly used by the fluid dynamics community, such as proper orthogonal decomposition (POD), linear stochastic estimation (LSE), and wavelet analysis. This process can generate large quantities of data and a method of handling all of the data and the techniques in an efficient manner is required. To accomplish this, a framework was developed that is capable of tracking, storing, and, manipulating data.

With the framework and techniques, data fusion can be applied. Data fusion is first applied to a synthetic data set to determine the best methods of fusing data. Data fusion was then applied to airfoil data that was obtained from PIV, CFD, and pressure to test the ideas from the synthetic data. With the knowledge gained from applying fusion to the synthetic data and airfoil data, these techniques are ultimately applied to data for a Mach 0.6 jet obtained from large-window PIV (LWPIV), time-resolved PIV (TRPIV), and pressure.

Through the fusion of the different data sets, occlusion in the jet data were estimated within 6% error using a new POD based technique called Fused POD. In addition, a technique called Dynamic Gappy POD was created to fuse TRPIV and LWPIV to generate a large-window time-resolved data set. This technique had less error than other standard techniques for accomplishing this such as pressure-based stochastic estimation.

The work presented in this document lays the groundwork for future applications of data fusion to fluid dynamic data. With the success of the work in this document, one can begin to apply the ideas from data fusion to other types of fluid dynamic problems, such as bluff bodies, unsteady aerodynamics, and other. These ideas could be used to help improve understanding in the field of fluid dynamics due to the current limitations of obtaining data and the need to better understand flow phenomena.

Access

Open Access

Included in

Engineering Commons

Share

COinS