An art or a science?

July 30th, 2012, Published in Articles: EngineerIT

by Dr. Kenny Uren, North-West University

Being a lecturer, teaching control courses to both mechanical and electrical/electronic engineering students, I increasingly become aware of students struggling to get their minds around these two concepts. This may be due to the fact that control systems is a kind of ethereal technology – since despite its wide spread application in industry, it is mostly invisible. Putting it differently, control systems may be viewed as a key driver in modern technological advances, only becoming visible if it failed in some way.

Devices such as a magnetic suspended metal ball or an inverted pendulum always intrigue students as they marvel at how then works. I also catch myself in awe looking at an Airbus A380 or a complex chemical plant. As practising control engineers we understand this kind of “magic”. We know the great effort going into simulations of such systems to make sure the actual systems are implemented with a minimum failure rate. Many of the mentioned systems are open-loop unstable, but with the correct sensors, actuators and computer control, the power of feedback control becomes evident. It stabilises an unstable system and provides smooth tracking control with disturbance rejection.

This illustrates both the art and science of systems modelling and control. There is an art to choosing appropriate mechanisms and combining them with the necessary sensors and actuators in a design that will allow controllability. There is also an art to selecting an appropriate mathematical model, with neither too little nor too much detail. Control theory then provides the “science” to allow control synthesis, analysis, simulation, and design. Therefore the art and science are both crucial for modern control system applications.

Throughout history we have been making models of the physical events we observe. From cave pictures to modern art, from models of the atom to models of the universe, all have sought to focus attention on particular ideas for analysis or for communication. As engineers, we must be particularly skilled in the art of making or choosing the mathematical models of our engineering concepts.

Today PID control is still the most popular workhorse in process control. However, models – in the sense of mathematic representations of systems – are critical to all modern control techniques. These models are not used solely for understanding the actual system anymore, but virtually all advanced control techniques rely on an explicit representation of the actual system. Models may be used in an on-line control algorithm; in optimisation loops; to allow fault tolerant behaviour; in adaptive control; the list goes on.

Present technological advances in computational power have enabled engineers to build increasingly complex models, which in turn have provided a means of performing increasingly difficult tasks in industry. This increased ability sometimes comes at the price of distancing ourselves from compact analytical models, typical of systems a few decades ago. As the systems to be controlled become increasingly more large-scale and complex, a single type of modelling approach will not be sufficient anymore. A future trend in system modelling may be to incorporate a variety of modelling approaches and physical domains. We may soon talk about the development of macro-models or multi-models referring to integrations of disparate models and domains in one framework. Today’s systems may become tomorrow’s subsystems. This only shows that in future control system design ventures, the comprehensive understanding of system modelling will continue to become increasingly paramount.

Contact Dr. Kenny Uren, Tel 018 299-1236,

Subscribe to our leading email newsletters


CLICK for other EE Publishers information products