by Jan Schuurmans, updated at 20-4-2019

Estimated reading time: 5 minutes

#### Can you make (more) money by tuning your PID loops? In this post we show how the mere fact of going through a systematic PID tuning process can increase profitability a lot.

There are actually two main arguments why going through the tuning process of PIDs, using the PID Tuner, can make your plant more profitable. The first is sometimes refered to as the 'classical' argument, and this one is actually just providing 'pocket money' compared to the second argument. So let's start with the first argument to get started

## PID tuning improvement allows to operate at more profitable setpoints

The key to increase profits, in this argument, is the setpoint value. The setpoints of PID controllers usually affect production speed, and/or production quality and/or costs. However, the setpoints have to be chosen at 'safe' distance away from high / low limits. Before we continue, let us define the control error as the difference between setpoint and the controlled variable, i.e. the input to the PID controller.

The control error varies with time. If you make a histogram of the control error, you usually see that the control error has a normal distribution, see the figure below. In this figure, we plotted the control error varitions with time in the left graph, and the histogram in the right plot. In the right hand plot we also show black dashed lines at +/- 3 times the standard deviation. For normal distributions we know that the control error will remain within those black lines in 99.7% of the time. If, for instance, the control error is not allowed to cross the lower limit more often than 0.15% of time, the setpoint value is ok now.

By improving the tuning, the standard deviation of the control error is reduced and this allows you to choose the setpoint at a more profitable level.

Let us consider an example to clarify this.

Estimated reading time: 5 minutes

The control error varies with time. If you make a histogram of the control error, you usually see that the control error has a normal distribution, see the figure below. In this figure, we plotted the control error varitions with time in the left graph, and the histogram in the right plot. In the right hand plot we also show black dashed lines at +/- 3 times the standard deviation. For normal distributions we know that the control error will remain within those black lines in 99.7% of the time. If, for instance, the control error is not allowed to cross the lower limit more often than 0.15% of time, the setpoint value is ok now.

By improving the tuning, the standard deviation of the control error is reduced and this allows you to choose the setpoint at a more profitable level.

Let us consider an example to clarify this.

This sounds like a nice way to increase profitability, if the loop is controlling quality, or adjusting some energy related input like heat or electricity. However, some control loops do not fall in this category, and even if they do, the payback on this is not always that much. So, you may wonder, what then is the main argument for going though the process of tuning loops in order to increase profitablity? That is discussed next!

- the controlled process variable (of the loop) does not respond to the manipulated variable (like a valve) at all!
- the controlled process variable (of the loop) responds much slower to the manipulated variable then previously!
- the controlled process variable responds in a non-smooth way , indicating sensor or software programming problems
- after PID tuning, the loop's closed loop response is much slower than predicted by the model, or
- after PID tuning, the loop's closed loop response is unstable (in contrast to what was expected of course)

Investigations indicate that, in practice, in most plants 30% of the control loops are in manual [1]! Another 25% (or so) of the loops shows oscillatory behavior. All in all, there is a huge potential to increase plant's profitability, but the results can be hard to calculate [2].

In the 'case studies' we show actual examples where the use of the PID Tuner created considerable increase of profits. Click here to read those stories

[2] N. Vatanskia, S-L. Jämsä-Jounelaa, A. Rantalac, T. Harjub, 'CONTROL LOOP PERFORMANCE MEASURES IN THE EVALUATION OF PROCESS ECONOMICS', IFAC 2005