Applications of convex optimization in plant-wide control of a solar Powered MDBR water recycling system

• Main Author: Avinash Vijay
• Other Authors: Ling Keck Voon
• Format: Theses and Dissertations
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering
• Online Access: https://hdl.handle.net/10356/48681
• Institution: Nanyang Technological University

The objective is to design a system which determines optimal operating points that will enable the solar powered Membrane Distillation Bio-Reactor (MDBR) water recycling plant to maximize the output produced while being self-sufficient. The system must optimize production rate during periods of adequate solar radiation and concentrate on survival of the essential micro-organisms in the MDBR during periods of inadequate solar radiation. Hence, uninterrupted plant operation during periods of unfavorable weather entails the management of a back up reserve. A balance has to be struck between minimization of power consumed and maximization of output produced to facilitate smooth operation. The challenge faced by this thesis is to translate these engineering objectives into convex optimization problems. This requires the modeling of plant dynamics to describe the relations between the variables of interest. Control is implemented through a hierarchical framework consisting of the scheduling, supervisory and regulatory layers. The main focus of this thesis is on the scheduling and supervisory layers which require optimization. The scheduling layer decides how to cope with changes in weather conditions several days in advance. It manages system reserves based on information available from the solar radiation prediction system to ensure continuous operation. The supervisory layer determines how the system will handle variation in solar radiation from dawn to dusk. Different control strategies and optimization problems required for this purpose have been considered. Since some relations in these problems are non-convex, the modifications required to solve them using convex optimization algorithms have also been discussed. Subsequently, simulations have been conducted to compare the performances of different strategies and to select the most desirable alternative. Illustrative examples have been provided for both layers to demonstrate the operation of the proposed system when subjected to different situations the plant is likely to encounter.