Are you need IT Support Engineer? Free Consultant

What Are My Options for Tuning a PID Controller?

  • By Control Guru
  • April 10, 2014

The Strengths and Weaknesses of Common Tuning Approaches

Not all approaches to PID controller tuning are equal. Regardless of your approach a key to successfully tuning a controller is modeling the process dynamics accurately.

Without an accurate model of the process the controller can’t do its job effectively. Each of the three most common approaches to tuning has pros and cons especially with regard to their ability to model process dynamics. Whether you choose to tune manually or with the aid of software, a common series of steps should be applied. It’s essential that the recipe include the following steps: 1) Clearly establish the Design Level of Operation, 2) Step the process sufficiently to reveal the process dynamics, 3) Calculate model parameters, including Gain, Time Constant and Dead-Time, and 4) Apply appropriate tuning correlations.

Some approaches to tuning are straight-forward whereas others leave the outcome to chance. We highly recommend that you choose the former and avoid the later. Here’s our assessment of those approaches:

  • Manual Manually tuning a PID controller is a common practice and it’s generally fine for processes that are quiet. By quiet we mean that the process is not oscillating and noise is limited. Unfortunately, most control loops have some degree of variability and they are inherently noisy, making it difficult to calculate the model parameters.  A typical result is that several attempts are required to achieve the desired level of control.  Sometimes a process is simply too erratic for this approach.

Pros  No additional cost.
Cons  Difficult to accurately model noisy/oscillatory data; Involves “Trial and Error” as multiple attempts are typically required; Time consuming.

 

  • Auto-Tune Many PLCs, microcontrollers, and DCSs are equipped with auto-tune functionality. Like manual tuning, this approach can prove effective when applied to a quiet process, and its easier just push the button! When initiated the auto-tuner performs a step test and utilizes the controller’s response to automatically calculate a process model and implement new tuning parameters.  That’s easy enough. The downside is that the model may or it may not be accurate and the tuning parameters may or they may not improve the controller’s performance. Since auto-tuners don’t allow you to either see the data that was used or to verify the model-fit, they keep you in the dark which is rarely a good thing.

Pros  No additional cost (usually included with the controller); Quick and easy.
Cons Difficulty with modeling noisy/oscillatory data accurately; Acts like a “Black Box” and restricts the user from adjusting the data and verifying the model fit.

 

  • 3rd Party Software Many companies have developed software specifically for the purpose of tuning PID controllers. In general 3rd party software follows the recommended recipe, and these products allow users to both adjust the model of the process and customize the tuning parameters. However most packages use traditional modeling techniques and struggle when noisy and/or oscillatory data is involved. Unfortunately noisy and/or oscillatory conditions are typical of industrial production processes, meaning the software is only useful in situations when manual tuning could achieve similar results.

Pros Quick and easy; Involves less attempts when the process is quiet; Allows the user to adjust the data, the model, and the tuning parameters.
Cons Additional cost; Cannot calculate an accurate model when the process data is noisy/oscillatory.

 

The purpose of tuning a PID controller is to improve control and enhance productivity. However, without an accurate model of the process the controller cannot respond appropriately to a process’ typical dynamics. In spite of their benefits, each of these approaches has clear limitations which should considered in advance.