Laser Sheet Imaging and Image Analysis for Combustion Research

Abstract: This Thesis presents techniques that aim at exploiting the potential of image analysis and processing in order to solve problems of data reduction, interpolation, quantification, and interpretation within the field of experimental laser imaging of combustion processes.

Combustion is the most important source of energy for power generation, heating, and transportation in the world today and its strong dominance is projected to continue in the foreseeable future. There are, however, many concerns regarding health effects and risks on humans, environmental pollution, climate changes, as well as availability of fuel resources, fuel cost, and competing markets. Therefore, a great interest in studying and better understanding the combustion processes emerged within the academic and industrial communities. Laser based optical diagnostics has been proven to be a valuable tool for characterizing combustion processes in great detail. These methods are appreciated for their ability to combine non-intrusiveness with sensitivity and selectivity for specific chemical species.

The first part of the Thesis deals with the analysis of spray images obtained through the application of tunable excimer lasers to spray diagnostics. The aim is to form a better understanding of the spray behavior, which may in turn lead to performance improvements in many applications of sprays in aerosols and combustion systems. The images of fuel sprays are experimentally produced by planar laser imaging where Mie scattered light from a cross section of the spray is imaged onto a CCD detector. Spray characterization then involves analyzing the resulting images by segmenting the sprays and investigating a number of their characteristics such as the cross sectional area, perimeter, and penetration length. Also, since the studied sprays are optically dense, a method for compensating laser attenuation based on the inversion of Beer Lambert's law is developed.

The second part of the Thesis deals with the analysis of flame images obtained through the application of time resolved laser imaging to turbulent combustion diagnostics. The data is produced by planar laser induced fluorescence (PLIF) imaging, where a laser diagnostic system for high speed spectroscopic imaging is used to record image sequences with very high frame rates (several kHz). Images reflecting the OH radical concentrations in flames are used to investigate the flame front structure in both non-premixed jet flames and spark ignited premixed flames. The aim is to study the influence of fluid motion and reaction chemistry on flame structure, velocity, topology etc. Image analysis methods for edge preserving smoothing, segmentation, tracking, frequency domain interpolation, and velocity estimation are developed for these purposes. Curve matching based on the computation of geodesic paths is used to track (interpolate) the flame motion. Implicit representations incorporating level set methods are deployed to allow proper handling of complex flame front curves with arbitrary topology present in high turbulence scenarios. Finally, a scheme which combines high speed PLIF imaging of OH with particle image velocimetry (PIV), is used to facilitate the separation of the effects of flow and chemistry on local flame front velocities and structures.

In conclusion, the work presented in this Thesis, which applies image analysis techniques to laser sheet imaging data, represents novel approaches for analyzing time resolved combustion processes both qualitatively and quantitatively. This will provide insight into fundamental mechanisms of turbulent combustion and a better understanding of such processes.