PID Control for Power

PID power control active reactive apparent power current loop SCR inverter and power electronics

PID Control for Power: First Define Which Power You Are Trying to Control

Power control is used in electric heating, power supplies, DC/DC converters, motor drives, inverters, grid-connected converters and many other industrial systems.

But before tuning any controller, one question has to be answered:

Which power?

An electrical system may contain:

active power - kW,

reactive power - kvar,

apparent power - kVA,

DC electrical power,

mechanical shaft power.

These are not interchangeable quantities.

If the PID setpoint is “10 kW”, the engineer must still define where that 10 kW is measured, whether power flow can reverse, and which actuator is physically capable of changing it.

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

Power and Energy Are Not the Same Thing

Power describes the rate at which energy is transferred.

Energy is the accumulated result of power over time.

For example:

10 kW is an instantaneous or averaged power level.

If a 10 kW load operates for one hour, it consumes approximately:

10 kWh

of energy.

Maintaining 10 kW with a control loop is therefore a different problem from limiting total daily energy to 100 kWh.

kW is power. kWh is energy. A power loop normally regulates an instantaneous or filtered power variable; energy-management logic integrates that result over time.

DC Power

For a simple DC system:

P = V × I

For example:

100 V × 20 A = 2,000 W = 2 kW

But even in a DC converter, the power controller does not necessarily manipulate both voltage and current directly.

The actual manipulated variable may be:

PWM duty cycle,

current reference,

voltage reference,

switching command.

This is why the outer objective and inner control variable must be separated.

Active, Reactive and Apparent Power in AC Systems

For a basic sinusoidal single-phase AC load:

Active power: P = V × I × cosφ

Reactive power: Q = V × I × sinφ

Apparent power: S = V × I

where:

P represents real power converted into useful work or loss,

Q represents reactive power associated with energy exchanged with electric or magnetic fields,

S represents the apparent electrical loading seen by the source.

For a balanced sinusoidal three-phase system, active power is commonly written as:

P = √3 × VLL × IL × cosφ

10 kW and 10 kVA do not mean the same thing.

Power Factor Is More Than Just cosφ in Nonlinear Loads

For a clean sinusoidal load, power factor is closely related to the phase angle between voltage and current.

But modern nonlinear loads such as:

rectifiers,

VFD front ends,

switch-mode power supplies

can draw distorted current.

In that case, true power factor is affected by both phase displacement and waveform distortion.

A serious power-control system may therefore need real measurements of:

RMS voltage,

RMS current,

active power,

reactive power,

apparent power,

true power factor.

Is the Controller Regulating Power or Current?

This distinction is fundamental.

The outer objective may be:

Power SP = 20 kW

while the faster inner controller actually regulates current.

A common structure is:

Outer loop: Power error → Current reference

Inner loop: Current error → PWM or switching command

The current loop runs much faster than the outer power loop.

In high-performance power electronics, the “power PID” often does not switch the semiconductor devices directly. It generates a reference for a much faster current-control loop.

Cascade Control in Power Electronics

Nested control loops are extremely common in converters.

For example, a DC/DC converter may use:

Outer loop: Output voltage → Current reference

Inner loop: Inductor current → PWM

Another system may use:

Outer loop: Power → Current SP

Inner loop: Current → Duty cycle

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

This is a direct power-electronics application of cascade and other control strategies.

Why Current Is Usually Controlled Inside the Power Loop

Current directly determines the electrical stress on:

semiconductors,

inductors,

transformers,

motor windings.

Excessive current can damage power devices much faster than a slow supervisory power controller can respond.

Fast current regulation and current limiting therefore belong inside the slower power loop.

The outer power controller should not be expected to provide semiconductor short-circuit protection.

Full PID Is Not Always Necessary

Power measurement often contains:

switching ripple,

ADC noise,

RMS calculation delay,

averaging,

digital filtering.

Derivative action can respond strongly to these fast measurement changes.

For that reason many converter control loops use:

PI control,

two-pole/two-zero compensators,

other purpose-designed compensation structures.

Derivative also does not literally predict future power. It reacts to the rate of change of error or measurement.

The general behaviour of the terms is explained in P, I and D components.

A Slow Power Meter Cannot Support a Fast Power Loop

Suppose a network power meter updates its value once per second.

That device may be excellent for:

SCADA,

energy management,

demand monitoring,

slow supervisory limiting.

But it is not a suitable direct feedback device for a switching converter that requires millisecond or microsecond control action.

Fast converter loops normally use measurement chains such as:

current shunts,

Hall sensors,

isolated current sensors,

voltage dividers,

isolated amplifiers,

fast ADC channels.

If measurement bandwidth is far below the required control bandwidth, increasing PID gain does not make the measurement faster. It only makes the controller react more aggressively to old information.

This is a practical example of time delay and system dynamics.

Power Control of Resistive Heaters

For a resistive load:

P = V × I

and, under an ideal constant-resistance assumption:

P = V² / R

P = I² × R

But many real heater elements do not maintain exactly the same resistance over temperature.

Examples include:

tungsten lamps,

silicon-carbide heaters,

infrared emitters,

transformer-fed heaters.

Cold-start behaviour and hot operating behaviour can therefore be very different.

A 50% controller command does not automatically equal 50% real electrical power for every type of load.

SSR Burst-Firing and Time-Proportioning Control

For suitable resistive loads, a zero-cross SSR can regulate average power by switching complete AC cycles on and off.

Conceptually:

50% demand → some complete cycles ON + some complete cycles OFF

This approach may be called:

burst firing,

cycle proportioning,

zero-cross control.

It is closely related to the actuator behaviour discussed in PID temperature control.

Phase-Angle SCR Control

With phase-angle firing, the SCR is not triggered immediately at every voltage zero crossing.

The firing angle is delayed so that only part of each AC half-cycle is delivered to the load.

This can provide finer cycle-by-cycle power control.

Phase-angle control may be useful for selected:

infrared heaters,

transformer loads,

high-inrush loads,

special heating elements.

But waveform distortion also increases, which can affect:

harmonics,

EMI,

power factor.

Zero-Cross or Phase-Angle?

Zero-cross / burst firing can be an efficient and relatively simple choice for suitable resistive thermal loads.

Phase-angle firing provides more continuous control within each AC cycle and may be required for certain special loads.

The correct method depends on:

load type,

transformer use,

inrush current,

thermal response,

power-quality requirements.

There is no universal rule that one firing method is always better than the other.

Why Soft Start Matters

Some heating elements have much lower resistance when cold.

Applying full voltage immediately can produce very high inrush current.

A power controller may therefore increase output gradually during startup.

This soft-start function can reduce electrical and thermal stress on:

the heater,

transformer,

SCR,

supply system.

Current Limit Can Override the Power Demand

A 20 kW setpoint does not mean the controller is always allowed to draw whatever current is necessary to reach 20 kW.

If the required current exceeds:

semiconductor rating,

transformer rating,

cable rating,

process limit,

the current limit has priority.

A practical hierarchy may be:

Protection limit → Current limit → Normal power regulation

This is where current control and power control become inseparable.

Temperature PID and Power Control Are Different Layers

In an industrial furnace, the true process objective may be:

200 °C

rather than 20 kW.

The temperature controller can generate:

63% heater demand

and a lower-level power controller converts that demand into electrical power.

The architecture may be:

Temperature PID → Power reference → SCR power controller → Heater

The temperature loop controls the thermal process.

The power controller manages the electrical actuator.

This distinction also applies to PID with temperature controllers.

Mechanical Power in Motors

For a rotating mechanical system:

P = T × ω

where:

P is mechanical power,

T is torque,

ω is angular speed.

The same mechanical power can therefore result from:

high torque at low speed

or

lower torque at higher speed.

This directly connects power with speed control and torque control.

Electrical Motor Power and Shaft Power Are Different

If a drive draws 20 kW of active electrical power from the supply, that does not mean exactly 20 kW is available at the shaft.

Losses exist in:

the drive,

motor copper,

motor iron,

mechanical friction and ventilation.

Approximately:

Pmechanical = Pelectrical × η

with efficiency η representing the appropriate system stage.

Therefore a signal called “motor power” should clearly indicate whether it means:

drive input power,

motor electrical power,

estimated shaft power.

Constant-Torque and Constant-Power Motor Regions

A motor driven by a VFD can often operate below base speed in a region where substantial torque is available.

Above the voltage-limited base-speed region, the system can enter field weakening.

Flux is reduced because additional motor voltage is no longer available in the same proportion.

As speed rises, available torque normally falls and the operating characteristic approaches a constant-power region.

Constant power does not mean constant torque. Because P = T × ω, torque must decrease as speed increases if mechanical power remains approximately constant.

Winders Are a Classic Power/Torque Example

In a winding application, roll diameter increases while line speed and web tension may remain approximately constant.

As diameter increases:

motor rpm decreases,

required shaft torque increases.

Mechanical power can remain approximately constant.

This is one reason winding applications naturally connect:

speed control,

torque control,

power control.

Power Control in DC/DC Converters

A buck, boost or buck-boost converter may be operated to regulate:

voltage,

current,

power.

But the switching device itself is usually manipulated through PWM duty cycle.

A typical power-control architecture could be:

Power SP → Power error → Current SP → Current controller → PWM duty cycle

The fast inner current loop directly manages:

inductor current,

semiconductor stress,

fast transient behaviour.

MPPT Is Not the Same as Fixed-Power PID

In a solar system, Maximum Power Point Tracking - MPPT does not normally attempt to maintain one fixed kW setpoint.

Its objective is to find the operating point at which the PV array can deliver maximum available power under the current:

irradiance,

cell temperature.

The MPPT algorithm changes a voltage or current reference and evaluates:

P = V × I

to search for the maximum-power operating point.

The resulting reference is then implemented by faster converter loops.

MPPT is an optimization layer, not simply a “hold 10 kW” PID loop.

Active-Power Control in Grid-Connected Inverters

A grid-connected inverter does not generally increase active power by simply increasing output voltage.

The controller measures grid:

voltage,

frequency,

phase.

A phase-locked loop can track the grid angle, and controlled inverter current is then synchronized with the grid.

Active-power demand is commonly converted into an appropriate current reference.

Once again, the fast control layer is fundamentally current control.

Active and Reactive Power Are Separate Control Objectives

A grid-connected converter may regulate:

P - active power

and

Q - reactive power

independently within its current and apparent-power capability.

A plant may therefore export a defined number of kilowatts while also following:

a kvar reference,

a power-factor reference,

a grid-support command.

One generic “power PID” should not be assumed to control P, Q and power factor simultaneously.

Power-Factor Control Is Not Active-Power Control

A power-factor controller may alter:

reactive current,

capacitor-bank steps,

inverter Q command.

Its objective is not necessarily to change active kW.

The plant can therefore have correct active power while power factor is still poor.

Correct kW does not guarantee correct PF.

Power-Flow Direction Matters

Power can flow in either direction in many modern systems.

Examples include:

a motor drive consuming power during motoring,

a regenerative drive returning power during braking,

a solar inverter exporting power,

a battery converter charging and discharging.

The power measurement therefore needs a clearly defined sign convention:

import / export,

positive / negative,

motoring / regeneration.

A current transformer installed with reversed polarity can also make the indicated power direction wrong.

CT and Voltage-Measurement Errors Can Completely Mislead the Controller

Correct active-power calculation requires correct:

voltage measurement,

current measurement,

phase relationship.

Errors in:

CT ratio,

CT polarity,

VT ratio,

phase mapping

can produce incorrect power values.

For example, assigning Phase-A current to the wrong voltage phase can corrupt the calculated active power even though both individual RMS values look reasonable.

Verify the complete power-measurement chain before tuning the power controller.

0-10 V or 4-20 mA Power References

A power controller may accept a 0-10 V or 4-20 mA demand signal.

For example:

4 mA = 0 kW

20 mA = 100 kW

The PLC engineering-unit scale and the controller input scale must match.

A signal can be electrically correct while still representing the wrong power demand because of incorrect scaling.

See also analog signals.

Feedforward Can Greatly Improve Power Control

Suppose line voltage changes.

For an approximately resistive load:

P ≈ V² / R

so even a moderate voltage change can significantly alter real power at the same firing or duty command.

If supply voltage is measured, feedforward can estimate a better starting output command immediately.

The feedback PI loop then corrects the remaining error.

This is a strong example of feedforward plus feedback control.

Output Saturation and Integral Windup

A power converter has finite limits:

maximum current,

maximum voltage,

maximum duty cycle,

maximum firing angle range,

thermal capacity.

Once one of these limits is reached, additional controller error cannot create more physical power.

If integral action continues accumulating, integral windup occurs.

When load or supply conditions later return to normal, the stored integral can produce severe overshoot.

Anti-windup, output limiting and bumpless mode transfer are therefore important parts of PID troubleshooting.

Why Use a Power Ramp?

A setpoint step such as:

0 kW → 500 kW

may produce severe transient stress on:

the converter,

transformer,

DC bus,

supply network,

heater.

A separate ramp generator can limit the rate of power change in:

kW/s

or

%/s.

The power controller then follows this ramped reference.

Ramp rate and PID tuning are different functions.

What Can a PLC Do?

A PLC can:

read power-meter data,

generate kW references,

run a slow outer PI/PID,

send an analog or fieldbus power demand,

manage load sharing,

apply ramp-rate limits,

supervise power limits.

But a converter current loop running at many kilohertz normally belongs in:

a DSP,

MCU,

FPGA,

the converter's dedicated control hardware.

This division of responsibility is important in PID control with PLCs.

Power-Electronics Control Is a Real-Time Problem

Fast current and converter loops require:

predictable sampling,

predictable calculation time,

predictable PWM update timing.

ADC sampling and PWM timing may even be deliberately synchronized.

Excessive execution jitter or delay reduces achievable bandwidth and stability margin.

This is one of the most important applications of real-time PID control.

SCADA Does Not Replace the Fast Power Loop

A SCADA system can display and trend:

kW,

kvar,

kVA,

power factor,

voltage,

current,

energy,

alarms.

It can also issue supervisory limits or setpoints.

But semiconductor switching and fast current regulation normally remain in the converter or dedicated controller.

SCADA supervises the power system; it should not be treated as the high-bandwidth semiconductor-control loop.

Power PID Is Not a Protection System

Normal power regulation does not replace protection against:

short circuit,

semiconductor overcurrent,

desaturation,

DC-bus overvoltage,

overtemperature.

These faults can develop much faster than the normal power controller is designed to respond.

Power PID is a normal-operation regulation function. Short-circuit, overcurrent, overvoltage and thermal protection must remain separate protection layers.

What Should Be Trended During Commissioning?

Useful signals include:

Power SP

Active power PV

Reactive power

Apparent power

Power factor

Voltage

Current

Current reference

Controller output

Current-limit state

DC-bus voltage

Device temperature

Import / export direction

Watching only the kW value can hide the actual reason the power loop is failing.

Common Power-Control Problems

Power SP rises but measured power does not: inspect current limit, voltage limit, source capacity and load impedance.

The controller remains at 100% output: physical power capacity may be insufficient; tuning is not automatically the problem.

Current is high but kW is lower than expected: check power factor and reactive current.

The power meter shows negative kW: this may be true regenerative power flow or incorrect CT polarity / phase mapping.

Power measurement jumps continuously: inspect harmonics, sampling, averaging and meter update rate.

SCR output is unstable: verify load type, firing mode, line synchronization and current limiting.

A heater produces excessive current at cold start: investigate cold resistance, soft start and current limits.

Motor power is constant while torque falls: motor speed may be increasing into a constant-power / field-weakening region.

A grid inverter produces correct kW but poor power factor: inspect reactive-power and current-vector control separately.

Reducing gain stabilizes the system but makes it very slow: inspect measurement delay, filtering, sampling time and inner-loop bandwidth.

Energy and power targets are being mixed: separate kWh management from the instantaneous kW loop.

Power-Control Commissioning Check

1. Is the controlled variable active kW, DC power, apparent power or mechanical power?

2. Are kW and kWh clearly separated?

3. Are voltage and current measurements correct?

4. Are CT/VT ratios and phase assignments correct?

5. Is power-flow direction defined correctly?

6. Is true active power measured correctly under nonlinear load?

7. Is there a faster current loop below the power loop?

8. Is the inner-loop bandwidth sufficiently greater than the outer-loop bandwidth?

9. Does the system repeatedly hit current, voltage or thermal limits?

10. Is anti-windup working correctly?

11. Is a power ramp-rate limit required?

12. If SCR control is used, is zero-cross, burst or phase-angle firing appropriate for the load?

13. Has heater cold resistance and inrush current been considered?

14. In motor applications, are electrical input power and shaft power distinguished?

15. In grid applications, are P, Q and PF treated as separate objectives?

16. Is measurement update time fast enough for the intended control loop?

17. Are short-circuit, overcurrent, overvoltage and overtemperature independently protected?

18. Has the system been tested at minimum, nominal and maximum realistic load?

Related PID and Power-Control Topics

PID fundamentals: What Is PID? - P, I and D Components - PID Gain and Tuning -