Multivariable Predictive Control

Applications in Industry

Sandip K. Lahiri

• Publisher: Wiley
• Year: 2017
• Print ISBN: 9781119243601
• eBook ISBN: 9781119243595
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Chemical & Biochemical Engineering
• Index: Technology & Engineering

A guide to all practical aspects of building, implementing, managing, and maintaining MPC applications in industrial plants

Multivariable Predictive Control: Applications in Industry provides engineers with a thorough understanding of all practical aspects of multivariate predictive control (MPC) applications, as well as expert guidance on how to derive maximum benefit from those systems. Short on theory and long on step-by-step information, it covers everything plant process engineers and control engineers need to know about building, deploying, and managing MPC applications in their companies.

MPC has more than proven itself to be one the most important tools for optimising plant operations on an ongoing basis. Companies, worldwide, across a range of industries are successfully using MPC systems to optimise materials and utility consumption, reduce waste, minimise pollution, and maximise production. Unfortunately, due in part to the lack of practical references, plant engineers are often at a loss as to how to manage and maintain MPC systems once the applications have been installed and the consultants and vendors' reps have left the plant. Written by a chemical engineer with two decades of experience in operations and technical services at petrochemical companies, this book fills that regrettable gap in the professional literature.

• Provides a cost-benefit analysis of typical MPC projects and reviews commercially available MPC software packages
• Details software implementation steps, as well as techniques for successfully evaluating and monitoring software performance once it has been installed
• Features case studies and real-world examples from industries, worldwide, illustrating the advantages and common pitfalls of MPC systems
• Describes MPC application failures in an array of companies, exposes the root causes of those failures, and offers proven safeguards and corrective measures for avoiding similar failures

Multivariable Predictive Control: Applications in Industry is an indispensable resource for plant process engineers and control engineers working in chemical plants, petrochemical companies, and oil refineries in which MPC systems already are operational, or where MPC implementations are being considering.

1
Introduction of Model Predictive Control

1.1 Purpose of Process Control in Chemical Process Industries (CPI)

Any industrial process, especially oil refineries and chemical plants, must satisfy several requirements imposed by its design and by technical, economic, and social conditions in the presence of ever‐changing external influences (disturbances). Among such requirements, the most important ones are as follows:

• Safety: This is the most important requirement for the well‐being of the people in and around the plant and for its continued contribution to economic development. Thus, the operating pressures, temperatures, concentrations of chemicals, and so on should always be within allowable limits.
• Product quality/quantity: A plant should produce the desired quantity and quality of the final products.
• Environmental regulations: Various international and state laws may limit the range of specifications of the effluents from a plant (e.g., for ecological reasons).
• Operational constraints: The various types of equipment used in a chemical plant have constraints (limits) inherent to their operation. Such constraints should be satisfied throughout the operation of the plant (e.g., tanks should not overflow or go dry).
• Economics: The operation of the plant should be as economical as possible in its utilization of raw materials, energy, and human labor.
• Reliability: The operation of the plant should be as reliable as possible to ensure that the plant is always available to make products.

These requirements dictate the need for continuous monitoring and control of the operation of a process plant to ensure that operational objectives are met. This is accomplished through an arrangement of instrumentation and control equipment (measuring devices, valves, controllers, computers) and human intervention (plant designers, plant operators), which together constitute the control system.

There are three general classes of requirement that a control system is called on to satisfy:

1. Suppression of disturbances
2. Ensuring the stability of the process
3. Optimizing the performance of the process

Traditionally, PID controllers are used in CPI to perform these tasks. PID regulatory controllers efficiently ensure stability of the process and suppression of disturbance. However, due to the multivariable nature of the process and complex interactions between process parameters, PID controllers cannot make a coordinated control move to optimize the process performance. Here lies the need of model predictive control.

1.2 Shortcomings of Simple Regulatory PID Control

PID control forms the backbone of control systems and is found in most CPI. PID control has acted very efficiently as a base‐layer control for many decades. But with increased global competitiveness, process industries have been forced to reduce production costs in order to maximize profit. They must continuously operate in the most efficient and economical matter possible.

Most modern chemical processes are multivariable (i.e., multiple inputs influence same output) and exhibit strong interaction among the variables. Let us consider an operation of a boiler whose main function is to produce and deliver steam to downstream units or steam header at a specified temperature and pressure. The boiler has a drum with inlet water flow and is heated by fuel gas to produce steam. Now consider a situation where demand of the steam in downstream units increases, and it starts drawing more steam from the header. As a result, water level in the boiler will drop, vapor space above the water will expand, and consequently pressure and temperature will drop. Note that the water level in boilers is not independent and can affect the steam pressure and temperature. As a corrective action, if inlet water flow increases to control level, this will drop the boiler temperature. It will call for more heating and more evaporation, which will again lead to level drop. This demonstrates that there are very strong multivariable interactions among steam pressure, temperature, boiler level, and inlet water flow. Everything affects everything.

Now consider a conventional basic regulatory control scheme in a boiler where multiple single‐input, single‐output PID controllers are used for controlling the plant (multiloop control). Say, boiler level is controlled by inlet water flow, temperature is controlled by fuel gas flow, and boiler pressure is controlled by outlet steam flow. One basic shortcoming of PID loop controls is that they act as a single‐input, single‐output (SISO) controller in an island mode. For example, level controller will see and maintain only level with no idea what is happening with pressure and temperature. The same is true for the temperature controller, which will adjust the fuel gas base on temperature feedback and will not care for level. Now consider the previous situation, where the level starts dropping due to more drawing of steam from header. Level controller will increase inlet water flow, which will reduce the temperature. Temperature controller will increase fuel gas flow, which will again lead to level drop. Again, the level controller allows more inlet water flow to maintain level. There is a lack of coordination among the controllers, and they all act as unconnected islands. Neighboring PID loops can cooperate with each other or end up opposing or disturbing each other. This is due to loop interactions and is a serious limitation of PID regulatory controller. It is very important to understand the multivariable interactions in the chemical process plant and then try to develop model predictive control.

MPC usually stands for model predictive control. Model predictive control is used in multivariable processes where multivariable interactions among the process parameters are significant. However, MPC also stands for multivariable predictive control. MPC is used for both model predictive control and multivariable predictive control throughout this book. In industry, sometimes it is also called advanced process control or APC. Reader should appreciate that MPC stands for all of them in this book and fundamentally they refer to same model predictive control in a multi variable process environment.

Unlike the PID controller, MPC is a multi‐input, multi‐output (MIMO) controller. MPC receives all the inputs (e.g., temperature, pressure, level, fuel gas flow, inlet eater f...

Table of contents

• Cover
• Title Page
• Table of Contents
• Figure List
• Table List
• Preface
• 1 Introduction of Model Predictive Control
• 2 Theoretical Base of MPC
• 3 Historical Development of Different MPC Technology
• 4 MPC Implementation Steps
• 5 Cost–Benefit Analysis of MPC before Implementation
• 6 Assessment of Regulatory Base Control Layer in Plants
• 7 Functional Design of MPC Controllers
• 8 Preliminary Process Test and Step Test
• 9 Model Building and System Identification
• 10 Soft Sensors
• 11 Offline Simulation
• 12 Online Deployment of MPC Application in Real Plants
• 13 Online Controller Tuning
• 14 Why Do Some MPC Applications Fail?
• 15 MPC Performance Monitoring
• 16 Commercial MPC Vendors and Applications
• Index
• End User License Agreement