Date of Award

Fall 2007

Project Type

Thesis

Program or Major

Chemical Engineering

Degree Name

Master of Science

Abstract

Process control is required in most of the industries to regulate the output of a specific chemical process. Any improvement in control design results in process optimization, consistent production, hence less waste. In this thesis, process dynamics of first and second order systems was studied in terms of a response to a step input change. A software called LabVIEW from National Instruments was used to simulate a number of processes to observe time response and frequency response to calculate gain, phase margin and Bode plots. This was followed by system stability analysis. Programs in LabVIEW were created to calculate Ziegler Nichols settings for controller tuning and dynamic Relative Gain Array for Multiple-Input Multiple-Output (MIMO) systems.

Studies were carried out on the dynamic performance criteria for controller tuning by minimizing Integral Absolute Error (IAE). This was possible since data could be collected and analyzed real time. LabVIEW programs were created to fine tune the controller starting with P, I, D values obtained using Cohen and Coon method. Process parameters required for the calculations were determined from Process Reaction Curve (PRC). Open loop circuit was used to measure the temperature/level to obtain a Process Reaction curve. Control of temperature in a heater was achieved by means of closed loop in which power supplied to the heater by a solid state relay was regulated according to the feedback obtained from the thermocouple. Results showed that PRC method was unsuitable for this process. Temperature controller was tuned using trial and error method and best settings were obtained as P = 2, I = 0.02, D = 0.5. It was desired to use Compact FieldPoint by National Instruments (NI) for liquid level controller tuning. After configuration and testing, it was found, however that output signal from the FieldPoint was in the range of 4-12mA which resulted in opening the control valve to only half of its full capacity. The problem was solved by using traditional Data acquisition device from NI to acquire data. PI controller was tuned from the starting values obtained by Cohen and Coon method using error-integral criteria. The best controller settings were obtained as P = 24, I = 0.35.

Download

Share

COinS