Abstract
In process industries there is an increasing trend in the -use of computers anchor distributed digital control. The invention of powerful 16 hit and 32 bit microprocessors is opening avenues for realising advanced control algorithms in practice and control of plant with distributed intelligence. Over the years, considerable work has been done towards evolving an appropriate configuration 1 of distributed digital control system for plant control. In many system configurations a computer has been chosen at the upper tier for carrying out plant optimization and management information report generation, leaving the regulatory control and monitoring functions to the processors in the bottom 2 , Optimization of an operating process unit performed in real time with an on line process control computer has a number of significant advantages over an optimization executed by an off-line computer. While the concept of Dynamic Matrix Control 3 to address the control of a plant at its economic optimum is new, the concept of process control computer assisting optimization by moving one set of operating conditions to another is well accepted in industry. A good account of this use of static optimization in computer control of process industries is available in reference 4. If. Two popular techniques e. g linear and nonlinear programming have found place in the on line optimization task 4 . Since the mathematical model of the plant and the constaint functions are nonlinear 5 , it is appropriate to use nonlinear programming technique to carry out the on line optimization. Various algorithms are available for solving nonlinear programming (NLP) problem 7 . An account of user experience on some of them is available in reference 8. Desirable features of NLP software has been discussed by Lasdon 9 . A brief description of the performance of M.J. Box's complex method for on line optimization is available in reference 10. This paper is conceited with making a brief survey of current industrial practice of real time plant optimization using nonlinear programming techniques. An account of desirable features of a good nonlinear programming software and relationship of software packages for on line execution from various data sources e.g, historical data base of the plant, operator data entry, input from Research and development activities will be included. The performance of various techniques e.g. successive linear programming (SLP) or Method of approximation programming (MAP), generalised reduced gradient (GRG) and complex method of M.J. Box will he evaluated in the context of on line application. Case studies from petroleum industries and thermal power plant will be chosen for this purpose.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have