Graduation Semester and Year
Spring 2026
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Industrial Engineering
Department
Industrial and Manufacturing Systems Engineering
First Advisor
Dr. Victoria C.P Chen
Second Advisor
Dr Jay M. Rosenberger
Third Advisor
Dr. Chen Kan
Fourth Advisor
Dr. Kevin Schug
Abstract
Supercritical fluid extraction (SFE) and supercritical fluid chromatography (SFC) are powerful analytical techniques widely adopted in pharmaceutical, food, and environmental analysis. Their performance depends critically on equipment parameters, such as pressure, temperature, and flow rate, etc. This dissertation investigates surrogate optimization techniques to identify optimal parameter settings for SFE-SFC systems in the extraction and separation of various analytes. A novel composite objective function is developed to guide the surrogate model, enabling efficient navigation of the complex parameter space and rapid identification of the global optimum.
The results demonstrate that the composite objective function along with the sequential optimization approach, which aims to improve the experimental outcomes over a sequence of experiments yields improved parameter settings, resulting in higher extraction yields, better separation, and more robust equipment operation with reduced variance and fewer failed experiments. Additionally, the sequential optimization approach facilitates the rapid determination of optimal settings for other analytes in the column.
The surrogate model utilized in this study is a Quintic Multivariate Adaptive Regression Splines (QMARS) model, which introduces bilinear terms to capture two-way parameter interactions during optimization. These bilinear terms pose significant challenges in nonlinear optimization due to their non-convexity. While traditional McCormick envelopes provide convex relaxations, they often produce loose bounds, particularly when variable limits form elongated rectangular regions. This study proposes a novel geometric approach to enhance these relaxations. By analyzing the feasible region’s geometry, a 45-degree coordinate system rotation is applied when one edge of the bounding rectangle significantly exceeds the other. This rotation enables the derivation of directional bounds that consistently outperform the upper bounds of conventional McCormick envelopes. Unlike standard axis-parallel cuts, the proposed 45-degree cuts tighten the relaxation more effectively. This method delivers superior upper bounds in all cases, enhancing global optimization solvers, particularly within spatial branch-and-bound frameworks, and accelerating convergence to the global optimum.
Keywords
Surrogate Optimization, Operations Research, Black-box optimization, global optimization, bilinear terms, non-convex optimization
Disciplines
Industrial Engineering | Operational Research
License
This work is licensed under a Creative Commons Attribution-No Derivative Works 4.0 International License.
Recommended Citation
Sood, Jaivardhan Vinodkumar, "Surrogate Optimization for Analytical Chemistry" (2026). Industrial, Manufacturing, and Systems Engineering Dissertations. 1.
https://mavmatrix.uta.edu/industrialmanusys_dissertations2/1
Comments
I would like to acknowledge my advisors, Dr. Victoria C.P Chen and Dr. Jay M. Rosenberger.