Limits, Unreachable Targets, and Windup

The Core Idea

A controller's requested output is limited by what the equipment can deliver. Integral windup happens when the integral contribution continues accumulating while that limit prevents the requested correction.

What you will learn

• Recognize an output that has reached its physical limit.

• Separate an unreachable target from a tuning problem.

• Compare integral windup with anti-windup protection under the same conditions.

Why it matters

A loop at an output limit needs a different diagnosis from one that still has room to respond. Recognizing the difference helps you choose a useful correction.

The problem

A heater starts at 20 °C. Its commanded output, controller output (CV), can range from 0% to 100%. You can change controller settings, but you cannot ask the physical heater for more than its full power.

First, the requested temperature is 140 °C. Predict what you would need to see before deciding that a different gain would solve the problem. Run the full 120 simulated seconds and compare process variable (PV), setpoint (SP), and CV near the end.

Read the limit before changing a gain

CV stays at 100%, while PV levels out near 120 °C below the 140 °C target. This is saturation: the requested correction has reached an actuator limit. Asking the controller to react harder cannot create additional heater capacity.

Now use a reachable 80 °C target on the same heater. Run the starting settings below with Integral anti-windup off. The purple reference is the same unprotected trial. Watch how long CV stays at its limit, then inspect the temperature peak and the recorded integral contribution after the run.

Change one protection setting

Enable Integral anti-windup, press Restart, and repeat the full 120 s. Keep P, I, D, the target, and the observation time unchanged for this first comparison. Compare the peak, the period at maximum output, and the I contribution at 20 s.

What happened and why

While PV is below SP, integral action builds a positive output contribution. With protection off, that contribution can keep growing even while CV is already clamped at 100%. After the temperature catches up, the stored correction can keep asking for heat and carry PV past the target. That is integral windup.

Here, anti-windup pauses accumulation while the output is saturated and the error would push it farther into the limit. It leaves less stored correction to unwind. The recorded I contribution is a contribution before the output clamp; it can exceed the heater's available output when protection is off.

Protection improves the reachable-target response. It does not make the 140 °C target reachable. Diagnose those two situations separately.

Check your response

Complete 120 s at the fixed 80 °C target. The measured response must reach the target's ±1.2 °C band, keep overshoot within 6 °C, and have average absolute error over the final 24 s no greater than 1.8 °C. These limits are based on the 60 °C start-to-target change.

The run exceeds the heater's 83 s minimum steady-state observation window: 3 s dead time plus four 20 s time constants. After comparing protection off and on, you may adjust I as well. The check accepts different settings that meet the response limits.

Hint 1 — Which problem do you have?

Is CV already at its maximum while PV stays below SP, or does PV reach SP and then go too far? Those observations point to different causes.

Hint 2 — Inspect stored correction

Compare the recorded I contribution with protection off and on. Restart both trials from 20 °C so the comparison includes the same warm-up.

Hint 3 — A useful starting repair

Enable Integral anti-windup and restart with I at 0.2% output per (°C·s). If you want another trial, try a smaller I such as 0.15% output per (°C·s) and compare the full response.

Compare the prepared trials

With P at 2% output per °C and I at 0.2% output per (°C·s), the unprotected 80 °C trial peaks near 97.7 °C. Protection on reduces the peak to about 83.1 °C, within the stated limit. D stays at zero.

On the 140 °C trial, the heater approaches 120 °C while commanding full power. The remaining gap calls for a capacity or operating-target decision, rather than another increase in gain.

Common mistakes

Enable JavaScript to run the embedded simulator and complete the interactive exercises.