Strategies for Nitric-Oxide Laser-Induced-Fluorescence in High-PressureJo
Combustion Systems

By
Tonghun Lee

Practical diagnostic strategies for detection of temperature and nitric oxide (NO) formation in high pressure (p<60bar) combustion systems using Laser-Induced-Fluorescence (LIF) of nitric oxide are investigated. NO-LIF, when applied to elevated pressures, suffers from decrease of signal due to pressure broadening and attenuation of the propagating laser beam/fluorescence signals. In addition, overlapping of neighboring excitation lines and interference from LIF of other species (mainly O2 and CO2) can significantly influence the overall signal. The main purpose of this study is to investigate NO-LIF strategies which minimize the impact of these complications or to allow for correction of their effects. A comprehensive study of NO-LIF in a laboratory high-pressure flame was carried out for various flame stoichiometries, pressures and excitation wavelengths to develop optimized excitation and detection strategies for high-pressure applications. Four main issues are addressed in this study. First, optimized excitation strategies are investigated for high-pressure applications in the A2Σ+-X2Π (0,0), (0,1) and (0,2) bands of NO. Second, CO2-LIF is identified as a major source of interference in the detection of NO-LIF in high-pressure combustion systems involving hydrocarbon chemistry. Third, an accurate multi-line thermometry technique for steady, high-pressure flames is proposed by fitting wavelength-scanned NO-LIF with computational simulations. Finally, measurements optimizing the detection strategies of 2-D NO-LIF imaging in high-pressure flames are reported. The discussion and demonstrations reported in this study provide a practical guideline for application of instantaneous 1-D or 2-D NO-LIF imaging in high-pressure combustion systems.

Self-organizing Systems:
Learning Algorithms and Multiple Robust Controllers

By
Jongeun Choi

The talk begins with a parametric diffeomorphism learning algorithm for one-dimensional topology preserving adaptive networks. The output weights of the network converge to a configuration that controls the usage probability of each agent, when the probability density function of the input signal is unknown and not necessarily uniform. In turn, the learning process provides an orientation-preserving diffeomorphic function from the known agent coordinate domain to the unknown input signal space, which maps a predefined agent coordinate probability density function into the unknown probability density function of the input signal. The convergence properties of the proposed learning algorithm are analyzed using the ODE approach (stochastic approximation) and verified by a simulation study. This new formulation can be readily utilized for various applications such as adaptive quantization, data clustering, learning control, intelligent systems, and coordination algorithms for multi-agent systems. The second part of the talk presents newly developed robust controller design techniques that utilize LMI convex optimization algorithms with an example of the robust track-following control of Hard Disk Drive (HDD) servo systems. Then we present how to design a convex polytopic partitioning of an uncertainty region, and its multiple robust controllers for LTI systems with a large convex polytopic parametric uncertainty set. Each controller takes charge of a local region and is designed to give a sub-optimal worst-case performance for that region. Minimization of the maximum of such worst-case performance of each local region is considered with respect to a partition and its multiple robust controllers. Future extensions to robust adaptive control systems for parametrically uncertain LTI systems and switching controllers for linear parameter-varying (LPV) systems will be discussed. Future research directions on learning algorithms and multiple robust controllers will be addressed. Surprisingly, these two apparently different topics, such as learning algorithms and multiple robust controllers, share a common idea: partitioning an uncertain space.

Two Problems in the Biomechanics of Soft Tissue:
Thermomechanics and Growth

By
Seungik Baek
Biomedical Engineering

Abstract - Supra-physiological temperatures are increasingly being used to treat many different soft tissue diseases and injuries. It has been long thought that the two main parameters affecting thermal damage are temperature level and duration of heating. Recent findings suggest, however, that increases in the mechanical loading on a tissue during heating decreases the thermal damage. Hence, we studied hear-induced changes in the thermomechanical behavior in a soft tissue membrane by heat treatments under biaxial mechanical constraints. Histological correlations of the extent of thermal damage reveal that the observed changes in biomechanical properties were due to changes in collagen organization.

Despite significant information on the biochemistry and cell biology of tissue growth there remains a need to develop a biomechanical framework for modeling. As a first step toward this end, we use a constrained mixture theory and, by specializing it to a 2D model, develop a computational framework to simulate enlargement of intracranial aneurysms. Specifically, we test multiple hypotheses for synthesis, degradation, and deposition of collagen fibers in the stability of these lesions. Computer simulations predict that stress-mediated enlargement proceeds via a competition between a local thickening and radial expansion, and the alignment of newly deposited collagen fibers has a significant influence on the potential stability of the enlargement. We are now expanding this model to incorporate more biochemical and cellular information on growth and remodeling of arteries with patient-specific geometries.

Investigation of Transport Phenomena using Magnetic Resonance Imaging

By
L. Guy Raguin

Abstract - The nuclear magnetic resonance phenomenon is widely used for spectroscopy in chemistry, biology and solid state physics (NMR), and for imaging for biomedical applications (MRI). Electrical engineers have accelerated MRI data acquisition through advances in hardware, as well as in processing algorithms. The development of the image reconstruction has been primarily qualitative, with focus on increasing the signal-to-noise ratio and enforcing smoothness constraints. As mechanical engineers, we can apply MRI to our research field and make significant contributions to MRI development as well. My talk will focus on my work in MRI velocimetry at the macro- and microscopic scales and in extracting quantitative information from diffusion-weighted MRI data.

The first part of my talk will deal with my efforts to make quantitative MRI velocimetry studies of flows in complex reactors and microsystems with limited or no optical access. First, a novel class of MRI velocimetry methods is proposed, drawing from prior knowledge about the flow. MRI velocity measurements are here projected onto a set of spatial functions that satisfy several fluid mechanical constraints, thus producing a realistic velocity field. The efficiency of this approach was demonstrated in the study of passive scalar transport in a mixing reactor, and led to the discovery of chaotic segregation, which is based on the effect certain "structuring" flows have on the transport of a scalar (species concentration or local temperature). Second, microscopic velocimetry protocols are developed, assessed and validated in microchannel networks.

The second part of my talk will focus on the theoretical, numerical, and experimental aspects of a quantitative analysis of q-space MRI data (QUAQ) that allows the extraction of the physical parameters of diffusion in fiber networks and the reconstruction of fiber crossings, e.g. for brain white matter fiber tracking. The advantages of QUAQ reside in its physical approach of the problem, the extraction of fiber orientations and diffusion characteristics, and the limited amount of data necessary for the reconstruction compared to existing methods. Experiments have been successfully implemented on a synthetic phantom and a section of human pons.