What is PID?

What is PID control

What Is PID Control? A Practical Introduction to Closed-Loop Control

PID control is a feedback method used to keep a measured process variable close to a desired value. It appears throughout industrial automation: maintaining pressure in a pumping system, temperature in an oven, flow through a process line, tank level, motor speed and even position in motion-control systems.

The letters PID stand for Proportional, Integral and Derivative. Each term reacts to the control error in a different way.

That does not mean every industrial control loop uses all three terms. In practice, PI control is extremely common, while some applications use P only, PD or more specialised control structures.

The basic objective remains the same:

Measure what the process is doing, compare it with what it should be doing, and continuously correct the difference.

SP, PV and Control Output

Three values appear repeatedly in PID applications.

SP - Setpoint: the value the process is expected to maintain.

PV - Process Variable: the value currently measured from the process.

CV or MV - Control / Manipulated Variable: the output used to change the process.

Consider a compressed-air system that should operate at 7 bar.

The required 7 bar is the setpoint.

A pressure sensor or transmitter provides the actual pressure.

The controller may then change compressor speed through a variable-frequency drive.

If the measured pressure is 6.2 bar, the PID controller sees a control error and determines how much corrective action is required.

What Does the Proportional Term Do?

The proportional term responds to the error that exists right now.

A large error produces a larger proportional correction. As the process approaches the setpoint, the proportional contribution becomes smaller.

The strength of this response is generally associated with proportional gain, commonly written as Kp.

Increasing proportional action can make a loop respond more strongly, but an excessive value may cause overshoot, oscillation or instability.

For this reason, “increase Kp until the process becomes fast” is not a sensible general tuning method.

The full interaction of the three terms is explained in the P, I and D components of PID control.

Why Is Integral Action Needed?

A proportional controller can sometimes leave a small but persistent difference between the setpoint and the real process value.

Integral action responds to error that remains over time.

If a process continues to operate below its setpoint, the integral contribution gradually increases until additional corrective action is produced.

This is very useful for eliminating steady-state offset, but integral action introduces its own problems when used too aggressively.

The process may overshoot or develop slow oscillations.

Another important problem is integral windup.

Suppose a pump is already running at its maximum allowable speed but pressure still cannot reach the setpoint. More integral action cannot create additional pump capacity. If the integral term continues to accumulate error, a large stored correction may remain when operating conditions later improve.

Practical PID implementations therefore use output limits and, where supported, anti-windup or tracking functions.

What Does Derivative Action Actually Do?

Derivative action responds to the rate at which the relevant signal is changing.

It is sometimes described as “predicting the future error”, but that wording is misleading. The controller does not know the future. It reacts to the current rate of change and can provide additional damping in suitable systems.

Derivative action also has an important weakness: measurement noise contains rapid changes.

A noisy pressure, speed or position signal can therefore create a noisy derivative contribution.

For this reason, many industrial process loops use PI rather than full PID control. When derivative action is used, filtering and the exact controller implementation become important.

PID Is a Closed-Loop Controller

A typical feedback loop repeatedly performs the same sequence:

1. The sensor measures the process.

2. The controller compares PV with SP.

3. The control error is determined.

4. P, I and, where required, D actions are calculated.

5. A new control output is produced.

6. The actuator changes the physical process.

7. The sensor measures the result again.

This repeated measurement-and-correction structure is why PID is described as closed-loop control.

For a deeper look at the sequence itself, see how a PID controller works in industrial applications.

The Sensor Is Part of the Controller

A PID algorithm can only react to the process value it receives.

If the measurement is wrong, the controller will make perfectly logical decisions based on incorrect information.

Consider a pressure transmitter scaled for 0-10 bar with a 4-20 mA output.

If the PLC or controller interprets that signal incorrectly, the PID loop may believe the pressure is 5 bar when the real process is already at 6 bar.

The controller then increases its output even though no additional pressure is actually required.

This is why 4-20 mA and 0-10 V scaling should be verified before tuning begins.

A bad measurement cannot be repaired with better PID gains.

Example: Maintaining Pressure with a Variable-Speed Pump

Assume a water system must maintain 5 bar pressure while consumption changes throughout the day.

The control chain may be:

Pressure transmitter → PLC or VFD PID → Speed reference → Pump → Pipeline

If consumers open additional valves, line pressure begins to fall.

The controller detects the lower PV and increases the pump-speed command.

As pressure approaches 5 bar again, the required correction becomes smaller.

This is a classic process-control application because the controlled variable is pressure while the manipulated variable is pump speed.

The detailed process behaviour is covered in PID pressure control, while the drive implementation is covered in PID control with a VFD.

Example: Temperature Control Is a Different Type of Loop

An oven may also use PID, but its behaviour is very different from a pressure system.

The heater output can change immediately while the measured temperature responds slowly because of thermal mass, stored energy and sensor location.

Even after heater power is reduced, temperature may continue increasing for some time.

Consequently, a PID tuning that works well for pump pressure should not be copied into a temperature controller.

For temperature applications, see PID control with a temperature controller and our temperature sensor section.

Example: PID in Motion Control

PID terminology also appears in servo and motion systems, but the architecture is different from a slow process loop.

A servo drive may contain nested position, velocity and current or torque loops operating at different speeds.

Not every loop necessarily uses full P + I + D control.

The controlled variables, time scales and parameter units can also be completely different from those used in temperature or pressure control.

This distinction is discussed in PID control with servo motors.

Why Does PID Need Tuning?

A PID controller has to be adjusted for the dynamics of the actual process.

If the response is too weak, the process can be unnecessarily slow.

If proportional action is excessive, oscillation may appear.

If integral action is too aggressive, overshoot and long settling behaviour can result.

If derivative action is poorly applied, measurement noise can dominate the output.

Good tuning therefore depends on:

process speed,

dead time,

sensor response,

actuator limits,

measurement quality,

required response speed,

acceptable overshoot.

These relationships are examined in PID gain and tuning.

Time Is Part of the PID Controller

PID is often shown as a mathematical formula without much discussion of time.

In a digital controller, however, the process value is sampled and the output is recalculated at specific intervals.

Integral and derivative action depend directly on that timing.

This means a PID block expected to run every 100 ms should not be executed at unpredictable intervals without considering the effect on the controller.

The relationship between dead time, sample time and physical process response is covered in time delay and process dynamics in PID control.

Kp, Ki, Kd and Kp, Ti, Td Are Not Interchangeable

Different controllers do not always expose PID parameters in the same form.

One PLC may use Kp, Ki and Kd.

Another controller may use Kp, Ti and Td.

Integral gain and integral time are not the same parameter.

Different controller families may also use dependent or independent gain equations, different time units and different derivative structures.

PID values should never be copied between controllers simply because the parameter names look similar.

Always check how the actual manufacturer defines the control equation.

Where Can the PID Algorithm Run?

A PID calculation can be implemented in many devices.

A PLC can control pressure, temperature, flow and level.

A VFD can run an internal process PID for pumps and fans.

A dedicated temperature controller may close a heating loop without a PLC.

A servo drive can run fast internal motion-control loops.

An industrial PC or microcontroller can also execute PID where its timing and I/O are suitable.

The important design question is not simply where can PID be programmed?

It is where should the control loop be closed for the required speed, reliability and maintainability?

PLC and PID

PLCs are widely used for process PID because measurement, machine logic, interlocks, actuators and communication can be integrated in one industrial controller.

A PLC can also coordinate several related loops and make operating decisions that go beyond the PID calculation itself.

For example, a pressure-control PLC may start a second pump when the first pump remains near maximum output for a defined period.

This wider control architecture is explained in PID control with PLCs.

What Is the Role of SCADA?

A SCADA system can show SP, PV, controller output, operating mode, alarms and historical trends.

It may also allow authorised users to change setpoints or tuning parameters.

That does not mean the fast PID calculation must run on the SCADA server.

In many industrial systems, the control loop remains local in a PLC, RTU, drive or process controller while SCADA performs supervision and data collection.

See PID control with SCADA for this division of responsibility.

Manual and Automatic Operation

Industrial PID loops frequently provide both Manual and Automatic operating modes.

In Manual mode, an operator or control program establishes the output directly.

In Automatic mode, the PID controller determines the output.

Manual operation is particularly useful during commissioning because it allows the engineer to check whether:

the actuator responds,

the process moves in the expected direction,

the sensor reports the change correctly.

Only after this chain has been confirmed does automatic tuning make sense.

Output Limits Matter

A mathematical controller may ask for more output than the physical machine can provide.

A valve cannot open beyond its mechanical travel.

A heater cannot provide more than its rated power.

A VFD cannot exceed its configured speed limit.

PID output limits must therefore represent the usable range of the actuator.

These limits are also closely related to integral windup and startup behaviour.

PID Is Not the Same Thing as Machine Safety

A PID loop controls normal operation.

It should not automatically be considered an independent protective layer for dangerous overpressure, overtemperature or personnel hazards.

A process may normally operate at 6 bar under PID control while a separate protective function exists for an unsafe high-pressure condition.

Good PID control does not replace safety engineering.

Is PID Always the Best Control Method?

No.

PID solves a very large number of industrial problems, but other strategies may be useful when a process has significant dead time, strong nonlinear behaviour, measurable disturbances, changing dynamics or several interacting variables.

Feedforward control can act on a known disturbance before a large feedback error appears.

Cascade control can use a faster inner loop.

Gain scheduling can use different tuning in different operating regions.

Adaptive and model-based methods can address problems that are difficult for one fixed PID controller.

These approaches are compared in PID and other control strategies.

Why Is PID Still So Common?

PID remains popular because it provides a very useful balance between performance and complexity.

It can control many processes without requiring a detailed mathematical plant model.

It is available in PLCs, drives, temperature controllers and motion-control systems.

Maintenance and commissioning engineers are generally familiar with its behaviour.

Its parameters can often be adjusted directly from measured process response.

And when full PID is unnecessary, the same family naturally reduces to P or PI control.

This practical flexibility is one of the main reasons PID remains fundamental in industrial automation.

Where Does PID Commonly Fail?

Problems are often blamed on PID even when the controller is not the real cause.

The output is at maximum but the setpoint is never reached: check actuator and process capacity.

The controller output changes but nothing happens: inspect the valve, drive, relay or mechanical equipment.

The process measurement moves randomly: inspect the sensor, wiring, analog input and grounding.

The system overshoots badly: examine tuning, delay, integral action and stored process energy.

The loop works at one operating point but poorly at another: the process dynamics may be changing.

Our PID troubleshooting page covers these symptoms in more detail.

Before Tuning a PID Loop

1. Verify the physical process measurement.

2. Check sensor range and engineering scaling.

3. Confirm SP and PV use compatible units.

4. Test the actuator manually.

5. Confirm the control direction.

6. Define useful minimum and maximum output limits.

7. Check sensor and input filtering.

8. Verify the controller execution interval.

9. Decide what should happen if the sensor fails.

10. Tune the loop only after the complete measurement-to-actuator chain works correctly.

Related PID Topics

PID fundamentals: P, I and D Components - How PID Works - PID Gain and Tuning

Dynamics and stability: Time Delay and Process Dynamics - PID for Different Process Types - Stable PID Operation - PID Limitations

Industrial controllers: PID with PLCs - PID with VFDs - PID with Servo Motors - PID with SCADA

Process applications: Pressure Control - Flow Control - Level Control - Temperature Control

Beyond basic PID: PID and Other Control Strategies

PID equations, parameter names, time units, derivative structures and Manual/Auto behaviour vary between manufacturers. Always verify the exact implementation in the documentation for the PLC, drive or process controller being commissioned.