A Repeatable P/PI Tuning Workflow

The Core Idea

Tuning is a controlled comparison of settings against operating requirements. A repeatable workflow changes one thing at a time, observes the whole response, and records enough evidence to explain the result.

What you will learn

• Make gain comparisons from the same starting conditions.

• Change one setting for a stated reason and keep a useful baseline.

• Choose a response that meets explicit limits rather than chasing one gain value.

Why it matters

Without a consistent trial, you cannot tell whether a setting improved the response or whether you simply compared different conditions.

The job

Use the familiar heater: 20 °C start, 50 °C target, and 0–100% output. D stays at zero, measurement noise is off, and anti-windup is on. Every trial lasts 60 simulated seconds.

The operating limits are a peak no higher than 53 °C, a settled response inside 49.4–50.6 °C by 35 s, and average absolute error over the final 12 s no greater than 0.9 °C. Meet all three with a response you can explain.

Use the same five steps each time

1. Record a baseline. Run the starting P-only settings and save the finished trace. The supplied purple reference is that same P-only baseline.

2. Name one problem. Is the response too far below setpoint (SP), too slow, or going past the target? Point to a part of the trace.

3. Change one gain. Keep P fixed while you try adding I. On another trial, keep I fixed while you adjust P. Write down what you expect the change to improve.

4. Restart and observe. Start again from 20 °C and run the full 60 s. A gain edit halfway through a hot run is a different experiment.

5. Compare and keep evidence. Compare the peak, settling, final gap, and output trace. Save an improved trial as your new reference, or retain the previous baseline if it remains more useful.

Read the result before another edit

An improved ending error can come with a larger peak. A quick first rise can still be followed by a long settling period. The check evaluates each response limit separately, so read the failed condition rather than changing both gains at once.

If a trial passes, record its P and I settings and a sentence about the compromise. Try a second passing combination. The lesson grades the measured response; it does not require a particular pair of gains.

Check your method

Compare two trials from identical starts and describe one deliberate gain change. Keep a short record: settings, predicted effect, measured peak, whether process variable (PV) stayed in the target band by 35 s, and whether the trial met the final-error limit.

Hint 1 — Which effect are you changing?

P changes the response to present error. I builds a correction while error persists. Decide which observed problem your next edit is intended to address.

Hint 2 — Keep the comparison fair

After changing a gain, use Restart before Run. Keep the target, D setting, noise setting, and observation time fixed. A saved trace makes the comparison visible.

Hint 3 — A useful trial

Keep P at 2% output per °C and try I at 0.2% output per (°C·s). If you want to explore another response, keep I fixed for a P trial rather than changing both together.

Two responses that meet the brief

P at 2% output per °C with I at 0.2% output per (°C·s) passes. P at 3% output per °C with I at 0.3% output per (°C·s) also passes. D remains zero in both.

These are two acceptable results on this heater. Your explanation should refer to the operating limits and the finished traces, not to the idea that one of these numbers is universally correct.

Common mistakes

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