Industrial Information and Control Centre

Aotearoa New Zealand’s ability to compete on a global scale increasingly relies on the ability to streamline our technologies. By effectively managing our industrial processes, resources and production, we can bring our country into the future by turning data into controlled industrial information.

The Industrial Information and Control Centre (I²C²) was established in 2008 in the Faculty of Engineering at The University of Auckland, and has become the premier national headquarters for industrial information and control in advanced process simulation and control. It’s objective is to provide a national focal point for research, postgraduate study, graduate training, continuing education and industry consultation in the field of industrial processes.

A significant percentage of New Zealand export industries, such as dairy, food, pulp and paper, and metals, depend on automation, but this is only part of the answer. New Zealand needs a greater range of opportunities for engineers working in its process and manufacturing industries.

By upskilling students and professionals in industrial information and control, companies can:

• Increase profits
• Improve sustainability with eco-efficiency and lower energy usage
• Improve quality (value add)
• Encourage the development of complex export products
• Open up new business opportunities
• Reduce dependence on imported software and expertise for modelling and control
  

The I²C² combines the strengths of three key institutions: the University of Auckland, Auckland University of Technology and the Institute of Measurement and Control, NZ (IMC). This relationship partners top research and academic institutions with instrument and control equipment vendors, and industrial clients. The I²C² also collaborates with the Light Metals Research Centre (LMRC) at the University of Auckland, which complements activities in the global light metals sector.

Process simulation

Process simulation is a model-based representation of chemical, physical, biological, and other technical processes and unit operations in software. Basic prerequisites are a thorough knowledge of chemical and physical properties of pure components and mixtures, of reactions, and of mathematical models which, in combination, allow the calculation of a process in computers.

Process simulation software describes processes in flow diagrams where unit operations are positioned and connected by product streams. The software has to solve the mass and energy balance to find a stable operating point. The goal of a process simulation is to find optimal conditions for an examined process.

Optimisation

Process optimisation is adjusting a process to optimise a specified set of parameters without violating a set of constraints.

Advanced Process Control (APC)

The I²C² is a strong proponent of Advanced Process Control (APC), and in particular a powerful controller known as Model Predictive Control (MPC) and optimisation techniques.

Model Predictive Control (MPC)

Model Predictive Control has widely been recognised as the most valuable advanced controller available to date. However, the implantation and maintenance of the controller requires significant expertise.

Process monitoring

Process monitoring is a statistical analysis tool that ensures processes are operating at their full potential to produce conforming products. Fault detection and diagnosis are two main parts of process monitoring.

Fault detection

Fault detection involves the recognition of a problem with an unknown root cause. Faults may be detected by a variety of quantitative or qualitative means. This includes many of the multivariable, model-based approaches. It also includes simple, traditional techniques for single variables, such as

• Alarms based on high, low, or deviation limits for process variables or rates of change
• Statistical process control (SPC) measures
• Summary alarms generated by package subsystems

Fault diagnosis

Fault diagnosis involves the pinpointing one or more root causes of problems, to a point where corrective action can be taken. This is also referred to as “fault isolation”, especially when emphasising the distinction from fault detection. In common, casual usage, "fault diagnosis" often includes fault detection, so “fault isolation” emphasises the distinction. Our methods include:

• Statistical Process Control (SPC)
• Principle Component Analysis (PCA)
• Partial Least Square (PLS)
• Linear Discriminant Analysis (LDA)

Control Performance Assessment (CPA)

Control loop performance directly affects the operability and profitability of industrial plants. Considering the importance of control loops, one would expect that they always perform at their peak, but this is not the case. In fact, many studies have shown that roughly one third of industrial control loops perform poorly.

Poorly performing control loops can make a plant difficult to operate and have several costly side effects, including:

• Reduced production rate
• Lower efficiency
• Poor product quality
• More off-spec product or rework
• Increased emissions
• Plant trips following process upsets
• Slower startup and transition times
• Premature equipment wear

For these reasons, control loop performance should always be kept at the highest possible level. Control Performance Assessment (CPA) is a recently developed tool to evaluate the control loop performance. CPA includes the use of statistical and signal processing techniques to help judge performance and effectiveness of control schemes for purposes of:

• Determination of performance benchmarks
• Diagnosis of underlying causes of poor performance
• Detection of poor performing loops
• Suggested improvement areas

At present most CPA techniques are restricted to linear systems, but most industrial processes are nonlinear to some degree. As a result, important phenomena such as a control valve stiction cannot be linearised, and hence adequately approximated by analysis based on linear systems. We have developed techniques to extend CPA into the general nonlinear systems including cases of valve stiction.

The I²C² team possesses a reliable academic and applied industrial track record in control performance assessment both in developing GUI software and developing new algorithms to detect and quantify valve stiction.

Software design

JSteam Excel Add-in (Steam utility modelling software)

JSteam is an Excel 2007 add-in to allow process and energy engineers to be able to model a range of industrial steam utility systems within the familiar Excel environment. JSteam utilises the latest code optimisation features to enable high speed thermodynamics for near instantaneous model convergence.

Utility models are built around a graphical Process Flow Diagram (PFD) using standard PFD symbols inserted similarly to standard Excel shapes. Unit operation results and thermodynamic properties are calculated in user defined cell locations, and are linked using standard cell reference techniques.

Automatically inserted and formatted tables speed up the model building process which results in an easy to use and learn modelling system. Utilising Excel enables the model to be expanded across multiple Excel sheets to best suit the plant, model, and engineers preference.

JSteam is written using the code libraries from Microsoft to enable enhanced functionality including an intuitive Excel ribbon interface and function wizard.

Controller auto-tuning

The tuning of PID loops is tedious, error prone and fraught with complications. However I2C2 has developed auto-tuning software based on relays. This provides a quick and robust way for operators to re-tune poor performing control loops.

jMPC Toolbox

Developed by an I²C²2 research engineer, the jMPC Toolbox is a mature MATLAB® Toolbox which enables research and development of MPC controllers within the powerful MATLAB® environment. It includes features such as:

• Full Simulink® integration for control of real processes via an external A/D & D/A interface
• Performance tuned Quadratic Programming (QP) solver
• Nonlinear simulations using Ordinary Differential Equation (ODE) models with automatic linearisation
• Advanced MPC features such as control move blocking, soft constraints and measured disturbance control
• Classroom-focused Graphical User Interface (GUI) for teaching MPC

• Professor Brent Young
• Dr Wei Yu
• Emeritus Professor John Chen
• Professor Mark Taylor