Off-campus WSU users: To download campus access dissertations, please use the following link to log into our proxy server with your WSU access ID and password, then click the "Off-campus Download" button below.

Non-WSU users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Access Type

WSU Access

Date of Award

January 2023

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Chemical Engineering and Materials Science

First Advisor

Helen Durand

Abstract

Next-generation manufacturing technologies offer increasing production autonomy and process profitability for the chemical process industries. However, they also demand unique requirements for real-time process monitoring and safe design. In particular, the design of control strategies to prevent process losses while at the same time being resilient against external attacks is critical to ensuring safe and profitable operations. Motivated by this, one of our research foci is to develop novel control designs and theories for cyberattack-handling strategies for nonlinear process systems under model-based control designs that use nonlinear control theory and optimization to compute optimal control actions that would cause hard process requirements to be satisfied. In particular, we developed systematic strategies for cyberattack detection in tandem with optimization-based controllers for industrial systems with rigorous proofs of the mathematical conditions under which the closed-loop process trajectories are guaranteed to be maintained inside a safe region of operation even when sensor/actuator attacks occur. In particular, the control/detection procedure and the mathematical conditions for detection and safe process operation were based on: a) randomized online modifications to a model-based control formulation with safety constraints to potentially detect cyberattacks; b) whether a threshold on the difference between process state measurements and process state predictions is exceeded; c) the use of multiple process state estimators to flag deviations from “normal” process behavior. Furthermore, these control schemes have been extended to cyberattack-handling for nonlinear processes whose dynamics change with time. We have demonstrated how certain combinations of the control/detection strategies (a-c) can detect and handle multiple attack scenarios simultaneously (while other combinations cannot achieve the same effect), and revealed the mathematical conditions under which process safety is guaranteed when multiple attack events happen. This research work has contributed to the process systems engineering field by opening new directions in the theory and design of controllers for next-generation manufacturing. On the process modeling front, a digital twin that captures all the relevant phenomena of an industrial system can be used to understand and predict the physical counterpart’s performance and impacts on process economics. Motivated by this, our research work has also been focused on the enhancement of digital twin development for dynamically operating systems, including a preliminary investigation into simultaneous design and control of processes under economics-based control formulations (i.e., optimization-based controllers that maximize process profitability). Specifically, we have implemented a chemical process example consisting of a continuous-stirred tank reactor followed by a heat exchanger to investigate the interactions between economics-based control formulations and processes and how they can impact the computational complexity of the controller and the design procedure. With regard to probing and exploring industrial uses of a digital twin, we have developed a tunable digital twin using the computational fluid dynamics method via ANSYS Fluent software that can serve as a testbed for exploratory studies to evaluate different operating conditions, materials, and design/operating costs for next-generation etching processes. Finally, motivated by the growing interest in incorporating machine learning techniques, including image processing analysis, for digital twins and next-generation manufacturing, we have been working toward exploring a virtual environment for image-based control strategies using the 3D graphics software Blender. In particular, to provide preliminary insights into how image-based process systems might be evaluated via process simulations, we have developed a testbed for a closed-loop process system using Blender's Python programming interface with camera images provided to the controllers as measurements for a tank level control problem.

Off-campus Download

Share

COinS