Ganesh, MahadevanDworzak, Bradley2015-10-212022-02-032015-10-212022-02-032015https://hdl.handle.net/11124/203242015 Fall.Includes illustrations (some color).Includes bibliographical references.This research is centered around improving and developing forward and inverse models to characterize a class of wave scattering configurations. The improvement is based on a project suggested by a group (P-23) at the Los Alamos National Laboratory (LANL). The scattering configuration is determined by a set of parameters. The main aim of the project is to develop data oriented forward and associated efficient inverse wave configuration parameter estimation algorithms. A key constraint associated with the project is that the input data is from a single incident wave impinging on the configuration. Based on a set of LANL experiments, the LANL group developed a preliminary configuration parameter estimation model by making strong assumptions about the experimentally determined configurations. The initial LANL model is based on the assumption that all particles in a dynamic blast configuration have the same simple (spherical) shape and also that the wave interaction between particles is weak. Such assumptions are not practical for the experimental data, leading to the research project investigated in this report. The main components of this research comprise the following: (i) Develop a forward computer model for multiple particle wave propagation that allow various shapes and material properties to facilitate collection of synthetic data from general configurations. (ii) Using the synthetic data, quantify the validity of the preliminary LANL inversion model. (iii) Develop an efficient inversion model to characterize an associated class of configurations. We achieve the objectives in our research by first implementing a high-order algorithm and developing a high performance computing (HPC) code for generating synthetic data for various configurations. The forward algorithm is based on an efficient non-polynomial finite element method that facilitates data generation from configurations with smooth and non-smooth particles. Then we generalize a multiscale inverse algorithm for a class of multiple particle configurations. Consequently, we develop an efficient inverse multiple parameter wave propagation model for our application.born digitalmasters thesesengCopyright of the original work is retained by the author.computer modelinversewave propagationfar-fieldacousticslocationText