How Do I Tune a PID Controller?

A Simple Recipe for Tuning Controllers Manually

Techniques for modelling a process’ dynamics and calculating tuning parameters are widely available.  These approaches enable practitioners to tune controllers consistently and with reasonable results.  At Control Station we know a little about this topic as we’ve been sharing these techniques for decades and innovating software-based solutions that deliver even better results.

Techniques for manually tuning controllers is often characterized as a recipe that involves the following steps:

1.  Specify the Design Level of Operation

Every control loop has a specific function or purpose – a specific range of operation for which the control loop was designed.  Before doing anything else you need to know the design level of operation (DLO) as that dictates how you address each subsequent step.

2.  Step the Process Within the DLO

PID controllers are model-based.  The model allows the PID to react appropriately to dynamics within the process.  In order to understand how your process reacts to changes, you must step the process within the DLO.  It’s essential for your step to be large enough that it clearly dominates any noise in the process.  As a rule of thumb a step of 3x-5x the noise band is recommended.  Just as important, manual tuning requires that you begin the step when the process is “quiet” – when it is steady and not impacted by disturbances.

3.  Calculate FOPDT Model Parameters

This step involves math.  Using results from the step test you can now calculate a First Order Plus Dead-Time (FOPDT) model with specific values for Gain, Time Constant, and Dead-Time: 

Gain indicates how far the process moves in response to changes

Time Constant is how fast the process reacts to those changes

Dead-Time reflects the delay in the process’ response 

 4.  Apply Tuning Correlations

The previous step provided model parameters with which your tuning parameters are calculated.  Those tuning parameters are based on correlations which specify the controller’s responsiveness.  Control Station recommends the IMC tuning correlations as a controller’s responsiveness can be easily adjusted for the following range of outcomes: 

Aggressive?     Moderate?     Conservative 

These values are then applied to your controller’s algorithm.  Note that the algorithms themselves can be complex so be sure to reference documentation supplied by the manufacturer.

Manual tuning is a reasonable approach when a process is steady and free of noise.  Unfortunately, most industrial processes are highly oscillatory and noisy.  As a result, manual tuning oftentimes requires numerous bump tests which can be costly in terms of time and lost production.  Only one company offers a software solution that eliminates the steady-state requirement across the full range of industrial applications.  If you can’t tune your loops manually, give us a call and we’ll tell you how!