Murdoch University Research Repository

Welcome to the Murdoch University Research Repository

The Murdoch University Research Repository is an open access digital collection of research
created by Murdoch University staff, researchers and postgraduate students.

Learn more

Multivariable model predictive control of a pilot plant using Honeywell profit suite

Wheat, Paul (2018) Multivariable model predictive control of a pilot plant using Honeywell profit suite. Honours thesis, Murdoch University.

[img]
Preview
PDF - Whole Thesis
Download (7MB) | Preview

Abstract

This thesis documents the first implementation of Profit Suite into Murdoch University’s Pilot Plant. This Pilot Plant is a small scale model of the Bayer Alumina Process. Profit Suite is a Honeywell application that uses Model Predictive Control (MPC) for Multivariable Control (MVC). The major project objective was to connect Profit Suite to the exiting Experion PKS control system then compare multivariable model predictive control to the existing PI control scheme. The project objectives were achieved. Multivariable controllers were built that controlled temperatures and levels in both halves of the plant.

The OPC connections between Profit Suite and Experion were completed and documented, as well as the procedures used to build and commission Profit Controllers in the Pilot Plant. Multivariable level controllers were designed using accurate models that performed well. These MVCs performed better than PI control in that they managed all tank levels and recycle streams throughout the plant. Linear objective functions were used to optimize flows and levels with success.

Baseline testing of the PI Controllers showed they were better than the MVCs for temperature control. The steam pressure disturbance had no effect on temperatures controlled by fast executing Experion PI controllers. Models found for steam pressure caused MVCs to overcompensate for this temperature disturbance. An MVC built that could manipulate steam valve positions to control temperature performed poorly compared to PI control. Multivariable temperature control was significantly improved when all pumps and steam valves were used as Manipulated Variables by the MVC. Models between water flow rates and temperatures enabled the MVC to use additional pump MVs to counteract the steam pressure disturbance.

There was no existing instrumentation to measure steam flowrates from each valve. This required Profit Suite to connect to the OP point of the PI Controllers to directly manipulate valve position for temperature control. Temperature control by cascaded PI steam flow control is recommended to improve the performance of multivariable temperature control. The installation of steam flow transmitters will enable the set point of a PI flow controller to be used as an MV by Profit Control. Fundamental models between steam flowrates and tank temperatures could then be acquired for multivariable control.

Item Type: Thesis (Honours)
Murdoch Affiliation: School of Engineering and Information Technology
Supervisor(s): Bahri, Parisa
URI: http://researchrepository.murdoch.edu.au/id/eprint/41285
Item Control Page Item Control Page

Downloads

Downloads per month over past year