PID Temperature Controller Tuning

As a general rule there are 2 types of tuning you can implement with a PID controller at the setup stage. The tuning process involves evaluating the best values for the PID parameters and applying them to the control output to achieve best response. Here we will discuss tuning methods for temperature control but many of these themes are common to generic process control and can be applied to PID control of pressure, flow and speed.

Automatic tuning - the controller performs an algorithm to calculate best settings for PID

Manual tuning - implemented by the user, parameters are adjusted to achieve desired response

Looking first at Automatic tuning, most modern temperature controllers incorporate this feature to speed up the commissioning process. There are a number of different categories of Autotune. With a Novus controller these are Fast, Full and Adaptive. With some variation on each depending on the controller model.

Fast Autotune is sufficient for the majority of basic temperature control applications. One of the key points to remember is to allow adequate span between process value and setpoint when performing the Autotune. We normally recommend a 100C differential. This ensures there is time for the algorithm to accurately run through the calculations required. With a Novus device, at the end of the Autotune the AT or Tune light will go out and the parameters are stored to the controller. Generally it is not necessary to repeat this process unless there is a change in the process configuration. For example, adding or removing heat/cool capacity, changing physical dimensions of area under control, replacing or moving sensor.

Controller types: N1020, N1030, N1040, N1050, N1200, N20K48

Full Autotune will take longer to perform the necessary calculations. This can be beneficial where the sensor feedback can take some time to settle out. Again at the end of the Autotuning process the controller will store the best values to the controller and there is no need to perform further tuning unless the response deteriorates. In theory a Full Autotune is more precise than Fast Autotune, although often in practise the difference in PID parameters is very small with no noticeable loss in performance. You should always avoid switching the controller off until any tuning process is complete.

Controller types: N1020, N1030, N1040, N1050, N1200, N20K48

Adaptive Tuning is advised for processes with a potential for degrading response. If the controller detects a change in performance it will initiate a cycle of Autotuning to restablish the correct PID values. Consider a complex heating operation with a number of elements, it is possible that over time some elements may fail. Adaptive tuning will adjust the PID values to provide best performance.  One disadvantage of Adaptive tuning is that is not available on some models of controllers and generally comes with high end devices which are more costly. For industrial processes a controller with Adaptive tuning is often the best choice. It may not be required in all cases but is beneficial to future proof the control process.

Controller types: N1020, N1200, N20K48

Manual Tuning has the significant disadvantage of being subject to operator intervention and requires a degree of trial and error to establish the best PID values. We recommend following an established pattern of Manual tuning where one parameter is adjusted at a time before noting the response and refining as necessary.

1. The first parameter to adjust should always be Pb, looking for the optimum performance, but understanding that there will always be some steady state error until the other terms are set.

2. Once happy with Pb introduce the Ir term and check to see if the steady state error is minimised. At this point the controller should hold temperature relatively well

3. Finally slowly increase the Dt such that the controller achieves any change in setpoint quickly with little to no overshoot

If all else fails revert the controller to default settings or perform a Fast Autotune to restore PID values. It is sometimes the case that a small amount of manual tuning is required following Automatic tuning. For example, to increase the rate of response or dampen small oscillations.

PARAMETER	BY INCREASING, THE PROCESS …	BY DECREASING, THE PROCESS …
Pb	It gets slower.   It generally gets more stable or less oscillating. It has less overshoot.	It gets faster.   It gets more unstable or more oscillating. It has more overshoot.
Ir	It gets faster, reaching the Setpoint quickly.   It gets more unstable or more oscillating. It has more overshoot.	It gets slower, taking longer to reach the Setpoint.   It gets more stable or more oscillating. It has less overshoot.
Dt	It gets slower.   It has less overshoot.	It gets faster.   It has more overshoot.

 

IF THE PROCESS PERFORMANCE …	TRY THE OPTIONS ONE BY ONE:
It is almost good but the overshoot is a little high.	Increase Pb by 20 %.   Decrease Ir by 20 %. Increase Dt by 50 %.
It is almost good but there is no overshoot, and it takes a while to reach the Setpoint.	Decrease Pb by 20 %.   Increase Ir by 20 %. Decrease Dt by 50 %.
It is good, but MV is always varying between 0 % and 100 % or is varying too much.	Decrease Dt by 50 %.   Increase Pb by 20 %.
It is bad. After startup, the transient has several periods of oscillation, which reduces very slowly or not at all.	Increase Pb by 50 %.
It is bad. After startup, it moves slowly towards the Setpoint, without overshoot. It is still far from the Setpoint and MV is already less than 100 %.	Decrease Pb by 50 %.   Increase Ir by 50 %. Decrease Dt by 70 %.