Graduation Semester and Year
2022
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Industrial Engineering
Department
Industrial and Manufacturing Systems Engineering
First Advisor
Victoria Chen
Second Advisor
Jay M. Rosenberger
Abstract
Surrogate Optimization and Global Optimization approaches to optimize underlying functions have been studied and used extensively in the field of Operations Research. However, there are very few instances where these approaches have been applied and tested in applications with uncertainty. Additionally, extensive focus and effort have been put into developing highly complex metamodels rather than globally optimizing these metamodels. In this study, we propose a Mixed Integer Quadratically Constrained Program (MIQCP) based approach that globally optimizes a Quintic Multivariate Adaptive Regression Splines (QMARS) metamodel. The QMARS-MIQCP based optimization is applied to a global optimization framework called QMARS-MIQCP-OPT to optimize several test functions both with lower and higher dimensions. The QMARS-MIQCP based optimization is also applied to a surrogate optimization framework called QMARS-MIQCP-SUROPT. In any surrogate optimization algorithm, the process of exploring the space to find new points to evaluate plays a key role in the performance of the algorithm. Even though, traditional Exploration and Exploitation Pareto Approach (EEPA) works well with surrogate optimization algorithms that use EEPA to select candidates points for evaluation, because our proposed approach optimizes the space around the selected EEPA candidates there is inherent exploration in the algorithm itself. Using traditional EEPA caused too much exploration, slowing down the algorithm's capability to find optimal solutions in fewer function evaluations. So, a modified Sorted EEPA approach is developed in the study. Additionally two different candidate selection strategies are also studied. The QMARS-MIQCP-SUROPT algorithm is also applied to test the performance when optimizing standard test functions that have been used as benchmarks in both surrogate optimization and global optimization applications. Optimal parameter settings play an important role in the efficiency of analytical chemistry instrumentation. The application of the developed QMARS-MIQCP-SUROPT algorithm to analytical chemistry instrumentation will provide an abundance of knowledge about the instrument, eliminate trial-and-error runs, as well as help in reducing the sample preparation time and cost of materials used. The QMARS-MIQCP-SUROPT algorithm is applied to ChromSim, a simulation software developed by Dr. Dwight Stoll and team to optimize the parameter settings of a 2D - Liquid Chromatography (LC) system for efficient separation of different analytes. The developed QMARS-MIQCP-SUROPT algorithm is also applied to guide a series of real-world laboratory experiments to optimize the parameter settings of the Shimadzu Liquid Chromatography Mass Spectrometry (LCMS) 2020 instrument for efficient flow injection analysis of Acetaminophen.
Keywords
Surrogate optimization, Global optimization, Mixed integer quadratically constrained programming, Quintic multivariate adaptive regression splines, Black-box functions, Limited data, Uncertainty, Analytic chemistry, Mass spectrometry, Electrospray ionization, Ionization efficiency
Disciplines
Engineering | Operations Research, Systems Engineering and Industrial Engineering
License
This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License.
Recommended Citation
Sekar, Srividya, "Optimizing the Performance of Analytical Chemistry Instrumentation" (2022). Industrial, Manufacturing, and Systems Engineering Dissertations. 132.
https://mavmatrix.uta.edu/industrialmanusys_dissertations/132
Comments
Degree granted by The University of Texas at Arlington