PID Control for Pressure

PID pressure control with pressure transmitter VFD pump compressor and control valve

PID Control for Pressure: A Pressure Loop Is More Than a Sensor, PID Block and Valve

Pressure control is one of the most common closed-loop control applications in industrial automation.

It is used in:

water-pumping systems,

compressed-air systems,

process vessels,

gas pipelines,

hydraulic systems,

HVAC air systems,

chemical processes.

A simplified loop can be written as:

Pressure setpoint → Pressure measurement → PI/PID → Valve / VFD / pump / compressor → Process pressure

But the behaviour of the loop depends heavily on what is being pressurized, how the pressure is measured, and what equipment is used to change it.

Pressure-loop tuning is not just Kp, Ki and Kd. Pressure reference, transmitter range, sensor location, fluid compressibility, system volume and actuator behaviour all change the process dynamics.

For the general feedback principle, see What Is PID?.

What Is Pressure?

Pressure is fundamentally force acting over area:

P = F / A

Common industrial pressure units include:

Pa,

kPa,

MPa,

bar,

mbar,

psi.

But for control purposes, the numerical value is not enough.

The engineer must also know what reference the pressure is measured against.

Gauge Pressure, Absolute Pressure and Differential Pressure

Gauge pressure uses atmospheric pressure as its reference.

A normal industrial gauge-pressure transmitter may therefore indicate approximately:

0 barg at atmospheric pressure.

Absolute pressure uses perfect vacuum as its reference.

Approximately:

Pabs = Pgauge + Patm

Differential pressure measures the difference between two pressure points:

ΔP = P1 - P2

Differential pressure is widely used for:

filter monitoring,

flow measurement,

closed-tank level measurement,

duct and fan applications.

1 bar gauge and 1 bar absolute are not the same physical pressure. In vacuum and gas applications, confusing the pressure reference can put the entire control loop on the wrong engineering scale.

For the measurement family, see pressure sensors and what is a pressure sensor?.

The Pressure Sensor Is the Controller's View of the Process

The PID controller cannot know the real pressure independently.

It only knows the process variable reported by the measurement chain.

If real line pressure is:

6.0 bar

while the transmitter reports:

5.5 bar

the controller can continue increasing pump speed or closing a valve even though the real process pressure is already too high.

Before tuning, verify:

zero,

span,

measurement range,

engineering units,

calibration.

A badly calibrated transmitter cannot be corrected by better PID tuning.

How Is a 4-20 mA Pressure Transmitter Scaled?

A common example is:

4 mA = 0 bar

20 mA = 10 bar

In that case 12 mA corresponds approximately to the middle of the span, or 5 bar.

But if the transmitter is actually configured for:

0-10 bar

while the PLC scales the signal as:

0-16 bar

the PID process variable is wrong even though the current loop is electrically healthy.

This is why 4-20 mA scaling must be checked before tuning.

The signal fundamentals are covered in analog signals.

Do Not Choose a Pressure Range That Is Far Larger Than the Real Process

Using a 0-400 bar transmitter on a process that normally operates between 0 and 10 bar is not automatically good engineering.

The useful process range occupies only a small fraction of the transmitter span.

This can reduce:

usable resolution,

effective sensitivity,

ability to detect small pressure changes.

The selected range should account for normal operation, expected disturbances and safe maximum process pressure without making the actual control range unnecessarily small.

Sensor Location Can Determine Whether the Loop Is Useful

The controller regulates the pressure at the measurement point.

That point must represent the real process objective.

A transmitter installed directly at the pump discharge may see:

pump pulsation,

local turbulence,

a pressure that is higher than the pressure available at remote users.

In a long distribution system, the real question may be:

Do we want constant pump discharge pressure?

or:

Do we want adequate pressure at the most remote critical consumer?

Those are different control objectives.

A PID loop can be perfectly stable and still control the wrong pressure if the transmitter is installed at the wrong point.

Liquid and Gas Pressure Loops Behave Differently

Liquids are much less compressible than gases under normal industrial conditions.

A compressed gas volume can store significant pressure energy.

This means a compressed-air receiver can behave as a buffer:

short increases in air demand do not necessarily produce an immediate large pressure drop.

In gas systems, the pressure response depends on:

receiver volume,

source flow,

consumption,

temperature.

Liquid systems can respond much more sharply to very small volume displacement.

The same PID settings should not be copied blindly between a water-pressure system and a compressed-air receiver.

Pressure and Flow Are Coupled

Pressure does not exist independently from flow in a piping system.

As flow increases, friction and local losses increase.

A pump may show 6 bar at its discharge while a remote consumer receives much less pressure during peak flow.

This is why PID flow control and pressure control are often strongly connected.

Pressure tells us about the hydraulic or pneumatic energy level.

Flow tells us how much material is moving.

The two variables influence each other through the network resistance.

Should Pressure Be Controlled with a Valve or with Pump Speed?

Two common architectures are:

Control valve pressure regulation

and

variable-speed pump or fan pressure regulation.

The appropriate method depends on:

process requirements,

energy efficiency,

response speed,

equipment limitations.

For centrifugal pumping systems, variable-frequency drive control is widely used because pump speed can be adjusted directly instead of wasting pressure across a throttling valve.

Constant Pressure with a VFD

A typical pumping loop is:

Pressure SP → Pressure transmitter → PID → VFD speed command → Pump → Pressure

If pressure falls:

the PID increases pump speed.

If pressure rises:

the PID reduces pump speed.

This is one of the most common applications of PID control with a VFD.

But motor frequency or speed is not the process variable.

It is the manipulated variable used to change pressure.

A Higher Pump-Speed Command Does Not Guarantee More Useful Pressure Forever

Once the VFD reaches maximum speed, the controller has no more pump-speed authority.

If pressure is still below setpoint, possible causes include:

undersized pump,

blocked filter,

closed valve,

pipe leakage,

demand above design capacity.

If the PID is already at 100% and pressure cannot reach setpoint, increasing Kp does not make the pump larger.

This distinction between tuning and physical capacity is central to PID troubleshooting.

Centrifugal and Positive-Displacement Pumps Are Different

A centrifugal pump operates according to its pump curve and the connected system curve.

Changing speed moves the pump operating characteristic.

A positive-displacement pump behaves differently.

A gear, piston or screw pump can continue displacing fluid against a closed discharge path and create a very rapid pressure rise.

For positive-displacement systems, devices such as:

relief valves,

pressure limits,

independent trips

become especially important.

The PID controller must never be treated as a replacement for the independent relief protection required by a positive-displacement pumping system.

What Is Pump Deadhead?

A centrifugal pump can operate with the discharge path closed or nearly closed.

Flow falls toward zero while the pump continues rotating.

Heat can accumulate in the liquid and pump casing.

A pressure reading may still appear acceptable.

This is why pumping systems may require:

minimum-flow protection,

low-flow detection,

bypass arrangements,

dry-run detection.

Correct pressure alone does not prove that a pump is operating in a healthy hydraulic condition.

Multiple-Pump Constant-Pressure Systems

A single pump may not cover the entire required flow range efficiently.

A larger system may therefore use:

one lead pump under VFD control,

additional pumps that are staged on as demand rises,

pump staging off as demand falls.

This is often called:

pump staging

or

cascade pump control.

If stage-on and stage-off logic is poorly designed, pumps can repeatedly:

start,

stop,

restart.

Useful supervisory functions can include:

hysteresis,

minimum run time,

minimum stop time,

lead-lag alternation.

The PID loop controls pressure while supervisory logic decides how many pumps should be available.

Pressure Control with a Control Valve

A control valve is another very common pressure actuator.

But controller direction depends on where the valve is installed.

If the valve controls inflow to a vessel:

low vessel pressure may require the inlet valve to open farther.

If the valve controls venting or outlet flow:

high vessel pressure may require the outlet valve to open farther.

Therefore there is no universal rule such as:

“pressure rises, so the PID must always close the valve.”

Direct or reverse controller action is determined by the real process path. Inlet regulation and outlet/vent regulation can require opposite controller directions.

Fail-Open / Fail-Closed Is Not the Same as PID Direction

Valve terms such as:

air-to-open,

air-to-close,

fail-open,

fail-closed

describe actuator and failure behaviour.

PID direct/reverse action describes how controller output changes when the process variable changes.

These are related through the process, but they are not the same setting.

Controller direction should be proven with a controlled field test.

Valve Stiction Can Create Pressure Hunting

A valve may fail to respond to small command changes because of packing friction, linkage problems or actuator friction.

The sequence can become:

controller output increases slowly,

valve does not move,

output keeps increasing,

valve suddenly jumps,

pressure overshoots,

controller reverses direction,

valve sticks again.

The result can look exactly like bad PID tuning.

Valve stiction is a mechanical nonlinearity. PID tuning may hide it temporarily, but tuning does not repair the valve.

An Oversized Control Valve Can Make the Loop Too Sensitive

If a valve is much larger than required, the entire normal operating range may occur over a very small part of valve travel.

For example:

3% to 12% opening

may cover most of the process flow range.

A tiny position change can then create a large flow and pressure change.

The effective process gain becomes high.

Valve sizing and installed characteristic therefore matter just as much as controller gain.

Cascade Pressure-to-Flow Control

Instead of commanding a valve directly from the pressure controller, a pressure loop can generate a flow setpoint.

For example:

Outer loop: Pressure PI → Flow SP

Inner loop: Flow PID → Control valve

The inner flow loop can reject:

supply-pressure disturbances,

valve nonlinearities,

fast hydraulic disturbances

before they strongly affect the outer pressure loop.

The inner loop should be substantially faster than the outer loop.

This is a practical application of cascade control.

Feedforward Can Improve Pressure Control

Some disturbances can be measured before pressure has fully changed.

Suppose main-line flow demand suddenly rises.

Instead of waiting for pressure to fall, measured flow can be used to provide an immediate feedforward increase in:

pump speed,

valve command,

compressor loading.

The pressure PI loop then provides the final correction.

A possible architecture is:

Measured flow disturbance → Feedforward command

plus:

Pressure PI → Trim correction

This is another example of combining feedback with feedforward rather than asking the PID to correct every disturbance after it has already affected the process.

Receiver Volume Changes Compressed-Air Dynamics

A compressed-air receiver stores air and provides temporary buffering.

A large receiver can absorb short changes in demand so pressure changes more slowly.

A very small receiver can produce fast pressure swings for the same load disturbance.

Important variables include:

receiver volume,

compressor capacity,

air demand,

pressure band.

If receiver volume changes significantly, the old PID tuning should not automatically be assumed optimal.

Compressor Pressure Control

Compressed-air systems may use:

load/unload control,

VFD speed control,

inlet modulation,

multi-compressor sequencing.

A VFD compressor can use a pressure controller to generate motor-speed demand.

But the compressor may still have:

minimum speed,

cooling requirements,

lubrication limits,

operating-envelope limits.

The PID output has to respect those physical limits.

A compressor is not an unlimited 0-100% pressure actuator.

Multiple Compressor Sequencing

In a large compressed-air station, running every compressor under independent aggressive pressure control can create poor interaction.

A better architecture may use:

base-load compressors,

a trim compressor,

pressure-band logic,

minimum run times.

One VFD machine can trim pressure while additional machines are brought online when total demand increases.

This is no longer just a PID problem.

It is:

supervisory sequencing + pressure regulation.

Hydraulic Pressure Control

Hydraulic systems can have very fast pressure dynamics.

A positive-displacement pump primarily produces flow.

Pressure then develops according to load and flow restriction.

Pressure may be regulated using:

proportional valves,

servo valves,

variable-displacement pumps,

pressure-compensated pumps.

But the hydraulic relief valve remains a separate protection device.

A relief valve should not be treated as the normal modulating element of a PID loop.

What Does a Hydraulic Accumulator Do?

An accumulator can provide:

short-term energy storage,

pressure support,

pulsation damping.

Accumulator volume and gas precharge change pressure dynamics.

If precharge is lost or changed, the pressure-loop behaviour can change even though the PID settings remain identical.

This is a good example of why physical system condition must be checked before retuning.

Pressure Pulsation Can Mislead the Controller

Reciprocating pumps and compressors can produce natural pressure pulsation.

If the PID attempts to react to every pressure peak and trough, the actuator can move unnecessarily.

Possible solutions include:

pulsation dampers,

better transmitter mounting,

snubbers,

signal filtering.

But filtering introduces delay.

Too much filtering can reduce loop stability, which connects directly to time delay and system dynamics.

Why Derivative Is Often Used Carefully in Pressure Loops

A pressure transmitter can see fast signals caused by:

pump ripple,

compressor pulsation,

valve turbulence,

electrical noise.

Derivative action can amplify those fast components.

For many pressure applications, PI control is a sensible starting point.

D should be added only when process testing demonstrates a useful improvement.

D also does not literally predict future pressure.

It responds to the rate of change of error or measurement.

The general terms are explained in P, I and D components.

Integral Windup in Pressure Control

Suppose a pump has reached 100% speed but pressure remains below setpoint.

If the integral term continues accumulating error, the controller stores a large correction demand.

When consumption suddenly falls, the pump command may remain unnecessarily high and produce a large pressure overshoot.

The same problem occurs with:

a fully open control valve,

a compressor at maximum capacity.

Important functions include:

output limiting,

anti-windup,

tracking.

Minimum VFD Speed Can Create a Different Saturation Problem

Suppose a pump must not run below 20 Hz.

The PID may request less output, but the drive remains clamped at its minimum frequency.

If process demand continues falling, pressure can rise even though controller output is already at its minimum useful value.

Possible supervisory solutions include:

sleep mode,

pump stop,

restart pressure threshold.

If the pressure PID spends long periods against the drive's minimum-speed limit, the problem may be operating sequence rather than PID gain.

Sleep / Wake Control

In a water-pressure system, nighttime demand may fall almost to zero.

Instead of keeping the pump running continuously at minimum speed, the controller can:

stop the pump when pressure is satisfied and demand is very low,

restart it when pressure falls below a defined wake threshold.

This can reduce:

energy consumption,

unnecessary low-flow pump operation,

mechanical running hours.

The sleep/wake sequence is supervisory logic around the normal PID loop.

Water Hammer Is Not a Normal Steady-State PID Problem

Rapid valve closure or sudden pump stopping can create a hydraulic pressure transient known as water hammer.

The transient can produce pressures much higher or lower than normal steady-state operating pressure.

Possible engineering measures include:

controlled valve closing profiles,

pump deceleration ramps,

surge vessels,

air chambers,

proper check-valve selection.

A line normally controlled at 6 bar can still experience a much larger transient pressure during water hammer. Normal PID regulation should not be assumed to protect against fast hydraulic surge.

Vacuum Control Makes Sign Conventions Important

In everyday language, “more vacuum” means lower absolute pressure.

That can create confusing controller-direction terminology.

For example:

-0.8 barg

and approximately:

0.2 bara

can describe a similar physical operating point using different references.

Before configuring direct or reverse action, define:

the engineering unit,

the sign convention,

whether increasing PV means increasing absolute pressure or increasing vacuum level in everyday language.

Vacuum control should always be configured from the real numerical pressure variable rather than ambiguous words such as “more vacuum”.

Differential Pressure Across Filters

The pressure difference across a filter is a useful condition-monitoring signal.

As a filter becomes dirty, differential pressure can increase.

But PID is not always the correct response.

Differential pressure may be used primarily for:

maintenance monitoring,

warning,

alarm.

If fan speed continuously increases to maintain airflow through a dirty filter, the controller can temporarily hide the increasing resistance.

Fan output and filter ΔP should therefore be evaluated together.

A controller should compensate for normal process disturbance without hiding a maintenance failure indefinitely.

Hydrostatic Pressure and Level

Liquid-column pressure is approximately related to level by:

P = ρ × g × h

where:

ρ is liquid density,

g is gravitational acceleration,

h is liquid height.

This allows bottom pressure to be used for level measurement.

But the relationship becomes more complicated when:

density changes,

the tank is pressurized,

headspace pressure changes.

Closed tanks may therefore use differential-pressure measurement.

This links pressure measurement directly with PID level control.

Temperature Can Affect Gas Pressure

In a closed gas volume, pressure can change because temperature changes even when no additional gas enters.

A rising pressure therefore does not always prove that mass inflow has increased.

Gas pressure control may have to consider:

process temperature,

heat input,

gas volume.

This creates a direct physical connection with PID temperature control.

What Can a PLC Do?

A PLC can:

read a 4-20 mA pressure transmitter,

scale the signal into engineering units,

execute a PI/PID loop,

command a control valve,

send a speed reference to a VFD,

stage pumps or compressors,

manage high/low pressure alarms.

This is one of the most common architectures covered by PID control with PLCs.

But normal process regulation should remain separate from safety protection.

What Should Be Trended?

A useful SCADA trend can include:

Pressure SP

Pressure PV

PID output

Pump / fan / compressor speed

Valve-position command

Actual valve position

Flow

Motor current or load

High / low pressure alarms

Auto / Manual mode

Pressure alone often cannot tell whether the actual problem is:

PID tuning,

valve condition,

pump capacity,

transmitter error.

High-Pressure Alarm and Safety Trip Are Not the Same Function

A normal pressure controller may regulate a vessel at:

6 bar

while operating logic defines:

8 bar high alarm

and:

9 bar high-high trip.

Depending on the process risk, additional protection may include:

mechanical relief valves,

PSVs,

rupture discs,

independent pressure switches,

safety PLC functions.

The normal pressure PID is a process-regulation function. It does not replace the independent mechanical or safety-rated over-pressure protection required by the process.

Common Pressure-Control Problems

Pressure continuously oscillates around SP: inspect proportional gain, integral action, valve stiction, oversized valve, pump dynamics and pressure pulsation.

Pressure continues falling while PID output is 100%: inspect physical capacity, leakage, filter blockage and excessive process demand.

Pressure rises quickly and overshoots: integral windup, excessive pump capacity, small system volume or measurement delay may be involved.

Valve command changes but pressure responds late: inspect stiction, gas compressibility, process volume and transport delay.

Pressure signal is noisy: inspect pulsation, transmitter mounting, snubber design and filtering.

VFD remains at minimum frequency while pressure is high: sleep/stop logic may be required.

The pump repeatedly starts and stops: inspect pressure band, accumulator/receiver size, sleep/wake thresholds and minimum run time.

Remote pressure is low while pump discharge pressure looks normal: inspect piping losses and actual flow demand.

Fan speed increases continuously as a filter becomes dirty: the controller may be compensating for maintenance-related pressure drop.

Hydraulic pressure spikes occur: inspect valve and pump transient behaviour,