Date of Award

Spring 2015

Degree Type

Thesis

Degree Name

Master of Science in Electrical and Computer Engineering (MSECE)

Department

Electrical and Computer Engineering

First Advisor

Raymond A De Carlo

Committee Chair

Raymond A De Carlo

Committee Member 1

Jianghai Hu

Committee Member 2

Zoltan K Nagy

Abstract

The purpose of this research is to investigate the feasibility and advantageous outcomes of modeling a complicated non-linear hybrid dynamical process as an interconnected dynamical system for the purpose of solving a hybrid optimal control problem under the framework of nonlinear model predictive control. This work considers a hybrid model of the startup process of an evaporation system. In this evaporation system a liquid containing mixture of a non-volatile component A and volatile solvents B (water) and C (alcohol) is heated to evaporate the solvents and obtain component A at a higher concentration using a column that is temperature controlled by valves that control the flow of steam through a heat exchanger; valves also control the input feed inflow, vapor outflow and the drain of the concentrated product. The hybrid nature of this process was established in the work of C. Sonntag et. al. In this thesis we reformulate the mathematical model as an interconnected dynamical model with two autonomous modes. The reformulated model is then used as a constraint set for the optimization of a performance metric characterizing the startup of the evaporation process and its evolution into steady-state operation. The algorithm used to solve the optimization problem is a new version of the existing algorithms in which the model constraints and cost function computation are performed outside of the sequential quadratic (optimization) program inside fmincon in MATLAB. Extensive comparisons to the work of C. Sonntag et. al. are made.

Share

COinS