Centralised vs. Decentralised Control Systems

Centralised control is usually carried out via computer software, having as inputs all the available sensors and producing signals for all the available actuators in the system. This control strategy is the most powerful, at least in theory, capable of extracting “optimal” performance. However, in practice, it requires non-standard apparatus (industrial computer data acquisition cards, communications), its tuning is normally non-intuitive (involves matrix computations) and in case of a fault the whole system may break down. That is the reason why it is not widely employed in manufacturing systems. Nonetheless, for complex strongly coupled plants, centralised control may be the only solution with a limited set of actuators and sensors. Centralised control can be implemented either in open or closed loop.

In most industrial plants, the basic extension of classical PID controller design, implementation and tuning is the decentralised control approach, where structural concepts are used to decouple the interaction between variables. The use of standard equipment and the ease of hand-tuning or understanding by non-specialist technicians are the main advantages of decentralised control technique. Nonetheless, the control effort is decomposed into two stages: first to decouple the different subsystems and then to control them. The extra effort rewards consist in simpler subsequent design, implementation and tuning.

Decentralised control tries to divide the plant and design the independent controllers for each of the subsystems as a way to handle the control of the overall plant.  The information flow when implementing decentralised control in a complex plant may easily become complex and difficult to maintain and understand, with lots of nested loops, block-diagrams, many individual regulators to tune, etc. Nevertheless, a fundamental reason to use decentralised control in most practical applications is because it requires less modelling effort.

Other advantages of decentralised control are:

• Standard equipment (such as PIDs) is used, thus, it often results in cost-effective solutions.
• Its behaviour can be understood by technicians and plant operators, hence aiding in maintenance and repairs.
• Their localised and decoupled behaviour enables easier tuning, often on-line, with very few parameters to be tuned (e.g., a gain and an integral action), with model-free strategies such as PID tuning charts.
• Decentralised implementation tends to be more fault-tolerant, as individual loops will try to keep their set-points even in the case where some other components have failed (if coupling does not destabilise the overall combination). Fault tolerance, also increases when implementing override selectors. On the other hand, if a centralised controller fails, this may result in a catastrophic fault and significant downtime.

Undoubtedly, in present-day industrial processes, the most popular choice is decentralized PID implementation, with industrial PIDs, sensors and actuators connected to a communication network based on an industry-wide standard. At times PLCs act as middleware between PIDs and the network, and also provide the necessary set-point scheduling. Note that, some PLCs do incorporate PID calculations, sparing the need for local regulator hardware.  

It is advisable to use centralised control as an intermediate-hierarchy part of a larger cascade and decoupled structure for subsystems where:

• A stronger degree of coupling and the lack of additional sensors and actuators hinder the use of non-centralised structures.
• An accurate sufficient model is available.

But, note that, a centralized approach using the extra instrumentation can also be designed for enhanced robustness, but in many cases, with a significant modelling cost.

Also read:

• Basic Steps to Consider in Designing a Control System
• Cascade Control
• Ratio Control
• Proportional-Integral-Derivative (PID) Control Systems
• The Performance Limits for PID Controllers