Project Director

Henry, Jim

Department Examiner

Rozema, Edward; Parten, Clifford; Wang LingJun; Jones, Frank


Dept. of Computer Science and Engineering


University of Tennessee at Chattanooga

Place of Publication

Chattanooga (Tenn.)


This paper researches the enhancement of accuracy for level control for an operational distillation column. This was achieved through an analysis of both controller and sensor components. The UTC distillation column reboiler level control, like any control system, has multiple sources of error resulting in increased imprecision and inaccuracy. Several fundamental inaccuracies in processing of data were identified and then minimized by the introduction of a non-linear mathematical model prior to any actual controller development. This provides more accurate level measurement and a base for developing accurate controllers.In exploring the effectiveness of alternative controllers, two non-classical controllers are considered in this project for the case of continuous operation. These two new, more effective controllers have been designed, implemented and evaluated through comparison to their classical counterparts. One controller uses a very simple binary proximity sensor as its basis leading to an almost simplistic concept that proves to be surprisingly effective in controlling the complex system. Fuzzy-logic is used in another controller configuration to minimize the complexity of mathematical calculations in decision-making by modeling a controller on abstract 'human' logic. Remote operation is achieved for this proximity controller to enable users to operate the distillation column via the internet. The performance of the proximity controller is found to be excellent. On the other hand, the fuzzy controller has not been completely debugged. The effort needed to develop the fuzzy controller was great and I recommend using this type of controller for more complex systems where the investment in time and effort is better justified. The binary proximity sensor based controller was especially effective considering its simple concept and even simpler development. Additional possibilities exist for enhancing the gathering and processing of level data using data fusion, alternate sensors such as a thermistor-based sensor, and alternate means of tuning the problematic fuzzy controller such as neural networking. Research into the development of parallel controllers based on different theory and sensors has been shown to be a way to improve and optimize the level control system.


B. S.; An honors thesis submitted to the faculty of the University of Tennessee at Chattanooga in partial fulfillment of the requirements of the degree of Bachelor of Science.




Distillation--Automatic control; Distillation apparatus; Systems engineering


Process Control and Systems

Document Type



v, 99 leaves





Call Number

LB2369.5 .R368 2002