The Building Blocks of Proportional Integral Derivative Controllers (PID)

Proportional Integral Derivative Controllers (PID for short) are relatively simple yet irreplaceable components that are commonplace within the control panel industry, and despite PIDs being heavily relied on, they’re definitely not without their flaws.

Whilst the basic function of the PID is relatively simple – or at least when they were first introduced to process control – they work within a more complicated loop system, known as Ladder Logic. Ladder Logic gets confusing quite quickly, which only gets worse with each subsequent fault that occurs within a panel, then further compounded by the essential-but-damaging troubleshooting of said faults.

Proportional-only controllers

When they were first introduced, controllers weren’t known as PIDs as their function was ‘proportional-only’. Meaning they were far simpler than their three-charactered progeny that we’re used to today, however, they quickly became unable to handle errors occurring within the panel. Proportional controllers had a tendency to give-out whenever an error between the process variable and set-point needed remedying. As such, it was evident that the function for integral action within the controller was needed, and the proportional controller was updated to suit.

Putting the I in PID: Integral Action

To overcome the latency period created from dwindling proportional action (PA) between the process variable (A) and the set-point (B), electrical engineers soon figured that they could manually bias the control effort. Or put in simpler terms; they could manually reset the loop between points A and B.

It wasn’t long before the manual piece to this puzzle became automated, which further minimised the inherent delay manifested by manual intervention. It is this automatic reset which became what we now know as integral action (IA) or ‘reset rate’. However, integral action is not without its own flaws.

For example, if the process under control is already slow, then the error will take a while to clear whilst the operator employs aggressive integral action to remove it. The issue with IA is that it continues to grow whilst the error is still present, so if the reset rate is set too high, the error will be overcompensated by the controller. This sends an even larger error back in the opposite direction, producing ‘hunting’ cycles back and forth between points A & B – at least until the error is eliminated.

Integral action is appurtenant to processes whose actuator isn’t large enough to produce the required control effort to solve the error, such as an insufficient burner that can’t provide enough heat, or a valve with an insufficient flow rate. This situation is known as ‘actuator saturation’. As the errors between points A and B begin to climb as a result of aggressive IA, there is a limiting value where the actuator bottlenecks, stalling any option of fixing the error.

This bottleneck jams the actuator at 100%, allowing the errors to build to a huge value and renders the controller unresponsive, this in turn prevents the operator from fixing the error – the remedy to which would usually be to reduce the set-point level (B) to a more attainable range.

This type of malfunction is known as ‘reset windup’, and systems are now designed and built with the ability to shut down the integrator in order to protect the controller.

Proportional, Integral and D?: Derivative Action

Derivative action (DA) reduces the control effort in relation to the error’s rate of change, this way the speed at which the process variable (A) is descending to the set-point level (B) can be controlled and reduced. Meaning A doesn’t drop too quickly to meet point B – reducing the chance of hunting cycles as in Integral Action.

That said, this only works as long as the DA is being applied correctly, if it comes on too aggressively as in IA – it will cause hunting of its own. More often seen in the control of robotic equipment and motorised apparatus as they are quick to respond to the controllers command.

Because DA responds to the error’s rate of change, it can sometimes cause the control effort to spike if an error occurs or changes suddenly. Under these circumstances, the derivative action would immediately kick into action, long before the proportional or integral elements of the PID will. Each element of a PID controller has its place, which is why two-term PI controllers or just straight-up proportional controllers are not often in use.

The only situation where this level of predictive control is detrimental, is in a situation where the end result of a spike from derivative action could be dangerous. For example, the control of heat in an industrial blast furnace. To mitigate this risk, modern PIDs have been developed to include:

• reset windup protection
• a derivative action calculator – to ensure accurate operation
• noise filtering – to reduce spikes and surges
• loop tuning – allowing appropriate value selection for proportional, integral and derivative action, narrowing the gap in response time to changes in the process variable

Need more info? Contact the DualTEC team today

If you would like to discuss the application of PIDs into the design of your process control setup get in touch with our team of electrical engineers for more info. With over fifteen years experience as a company, and a collective knowledge surpassing the majority of control panel builders today – we’re well placed to help. Just call 01535 609314 or email info@dualtec.co.uk.