Strategies to avoid overshoot in PID Control

In the realm of automation and control systems, PID controllers stand as a cornerstone, widely adopted for their simplicity and efficacy in regulating various processes. However, a common challenge associated with PID control is overshoot, a phenomenon where the controlled variable exceeds its setpoint. This may or may not be an issue, depending on the process. A common process where overshoot can be an issue is in temperature control loops. This article explores some solutions to prevent overshoot that we’ve found to be used often in industry (quite often a ‘duct-tape strategy’), and solutions that are (in our eyes) more robust. Using a proper strategy ensures a smoother and more reliable control performance.

Understanding the Challenge: The PI Controller Case

Initially, consider a PID controller with its derivative term set to zero, effectively making it a PI (Proportional-Integral) controller. While PI controllers are simpler and less sensitive to noise, they often lead to significant overshoot in systems with substantial inertia or delay. This overshoot not only delays settling time but can also trigger undesirable system responses, making it crucial to address. In our example, we have a batch process where a product is heated from room temperature to 100[C]. This must be done sufficiently fast, while the temperature must not exceed 105[C].

Strategy 1: Setpoint Prefiltering

Setpoint prefiltering involves smoothing the setpoint changes before they are processed by the PID controller. This technique reduces the aggressiveness of the controller’s response to sudden setpoint changes, thus minimizing overshoot. By applying a filter, such as a first-order lag or a more complex algorithm, the setpoint variation becomes more gradual, allowing the system to respond more gently and reducing the likelihood of overshoot.

Disadvantages

Quite often it’s non-trivial how the setpoint should be filtered, and simulation tools are needed to optimize the setpoint trajectory and/or pre-filter type.

Strategy 2: Limiting the Manipulated Variable (MV)

Limiting the MV, the output of the PID controller, is a straightforward approach to prevent excessive overshoot. We’ve seen this method implemented quite often by our customers. More often then not, this leads to all kind of problems down the line, causing them to ask us for help and get a proper solution implemented.

Strategy 3: Disabling the Integrator on Large Errors

The integrator component of a PID controller accumulates error over time, leading to a stronger control action. However, during large errors, this can contribute to overshoot. A practical solution is to disable or freeze the integrator during these conditions, allowing the proportional and, if applicable, derivative actions to stabilize the system without the added aggressive push from the integrator.

Disadvantages

This can be very tricky, since turning the integrator off cause error to remain large, causing the integrator never to be turned on. In ‘integrating’ processes, for example temperature control loops, this may be a viable option though.

Strategy 4: Setpoint Weighting

A PID controller with setpoint weighting offers separate tuning parameters for the setpoint tracking and disturbance rejection paths. This decoupling allows for more refined control, where the aggressive response needed for quick disturbance recovery can be balanced with a smoother approach for setpoint changes, effectively reducing overshoot while maintaining robustness against disturbances.

The Setpoint weighting option in a PID controller is simple to implement, and most PID controllers have this implemented, for example the 'PID_Compact' block in Siemens it’s refered to as:

  • Proportional action weighting
  • Derivative action weighting

The setpoint weighting option is also implemented in our PID C-code implementation. Tuning is best done using the PID Tuner.

Strategy 5: Smith Predictor

For systems with significant time delays, the Smith Predictor can be powerful solution. This control strategy involves a predictive model of the process that estimates the future output based on current control actions. By compensating for the time delay, the Smith Predictor allows the PID controller to act on the predicted future state rather than the delayed measured output, significantly reducing overshoot caused by the delay. The PID Controller can be tuned ‘as-if’ there was no delay, while retaining stability.

Disadvantages

The predictive model must be ‘sufficiently’ accurate. If the model error is too large this can otherwise even lead to instabilities. It should thus be used only when the process can accurately be modelled for all process conditions – which is true for most processes.

Strategy 6: Model Predictive Control

Model Predictive Control can be a very powerfull control method, since it is able to ‘optimize’ towards a desired response using the process dynamics. It can predict the future, and as such, it can predict the overshoot before it’s actually there, and hence optimize the controllers output to make sure that no overshoot will occur.

Conclusion

Overshoot in PID control can undermine system performance and stability, but with the right strategies, it can be effectively mitigated. From simple tweaks like setpoint prefiltering and MV limiting to more advanced approaches like the 2-DOF controller, a Smith Predictor, or Model Predictive Control, there are numerous ways to enhance PID control for a variety of applications. By understanding the unique characteristics of the system in question and applying these strategies judiciously, engineers can achieve precise, stable, and efficient control.

More Information?

If you would like to have more information, or if you have a problem with PID Tuning yourself, please give us a call or send a mail to: info@dotxcontrol.com