Skip to main content
Open Access Publications from the University of California


UCLA Electronic Theses and Dissertations bannerUCLA

Tomographic Laser Absorption Imaging of Combustion Gases in the Mid-wave Infrared


This dissertation describes advancements in mid-infrared laser absorption tomography for spatio-temporal measurements of thermochemistry in reacting flows relevant to combustion systems. Tunable laser absorption spectroscopy is combined with tomographic reconstruction techniques to resolve small diameter ( < 1 cm) non-uniform flow fields with steep spatial gradients, leveraging emerging mid-wave infrared photonics. Multiple novel measurement methods, hardware configurations, and image processing techniques were investigated. Initially, a mid-infrared laser absorption tomography sensing method was developed for quantitative measurement of CO and CO2 concentrations and temperature distributions in turbulent premixed jet flames using a translation-stage-mounted optical system. This sensing approach was used to examine effects of varying fuel structure on carbon oxidation over a range of Reynolds number regimes. It was found that spatial and temporal resolution is limited in this method due to the finite laser beam size (~ 1 mm) and the slow mechanical translation of the optical system. To address these limitations, a novel laser absorption imaging (LAI) technique, that expands a single laser beam and replaces the detector with a high-speed infrared camera, was introduced to achieve enhanced spatial and temporal resolution for thermo-chemical imaging. As a demonstration of this new technique, distributions of combustion species were imaged in both axisymmetric and non-axisymmetric flow fields using linear tomography algorithms. For non-axisymetric flows, the limited view tomography problem often results in a blurring effect and artifacts in the reconstructed flow-field. In an effort to address these issues, state-of-the-art deep learning neural networks were developed and applied to solve the limited angle inversion. Initial results suggest that deep neural networks have potential to more accurately predict flame structures with fewer projection angles than linear tomography. This work provides a foundation for a new approach to quantitative time-resolved 3D thermo-chemical imaging in high-temperature reacting flows.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View