# What do P, I and D do?

> A free online PID simulator for tuning practice: move P, I and D and see right away what a PID controller does with a temperature loop, from overshoot to settling time.

This is how an engineer tests a control loop: move the setpoint by a step and watch how the temperature follows. Here that happens again and again, between 50 and 60 °C. Move the controller sliders while it runs and watch what happens to the line: on the right you can already see the next step coming.

This is the text version of an interactive simulation. On the page itself the setpoint keeps jumping between 50 and 60 °C in a running trend, you move P, I and D while it runs, and for a step from 50 to 60 °C you see the overshoot, the time to setpoint and the remaining offset.

## The three actions

- **P (proportional)**: Reacts to how large the difference from the setpoint is right now.
- **I (integral)**: Reacts to how long the difference has existed.
- **D (derivative)**: Reacts to how fast the temperature is changing, and brakes.

## What you see

- **Overshoot**: how far the temperature goes above the setpoint.
- **At temperature after**: how long it takes until the temperature stays within half a degree of the setpoint.
- **Remaining offset**: the difference from the setpoint that remains at the end of the test.

## The model

The process is a first-order process with dead time (K = 1 °C per %, τ = 15 s, θ = 2 s, 55 °C at 50 % valve position). The controller has the ISA standard form, D action on the measurement and a scan time of 0.5 s. The slider for I runs from Ti = 60 s (weak) to Ti = 2 s (strong). Optimal here means: at the setpoint as fast as possible, with at most 0.3 °C overshoot and a calm valve. A different process needs different settings. The behaviour of P, I and D stays the same.

## P, I and D work exactly the same way in your plant.

This is a simulated process. In practice the setpoint usually stays put and the controller mainly absorbs disturbances; a step like this one shows how well it does that. In a real control loop we take the time constant, dead time and gain from your own trend data and calculate the settings from them. Would you like to know whether your control can be better? Take the quickscan, or read exactly how a PID controller works.

[Take the quickscan](/en/tools/pid-quickscan)

[PID controller explained](/en/knowledge/pid-controller-explained)