Control Loop

What is a control loop?

A control loop is a closed chain of measuring, comparing and correcting with which an automation system automatically holds a process variable such as temperature, pressure, level or flow at a desired value, the Setpoint. A Sensor measures the actual value, a Controller calculates the difference from the setpoint and an Actuator adjusts the process, after which the measurement starts again. Control loops are the foundation of the process industry, HVAC, machinery and industrial automation in general. A typical refinery or chemical plant runs hundreds to thousands of these loops at the same time, usually in a PLC or DCS. The aim is always the same: stable and safe operation, efficient use of energy and raw materials, and preventing deviations, errors or damage without constant human intervention.

🔧 How does a control loop work?

A control loop compares the measured value, the Process Value (PV), with the desired value, the Setpoint (SP). The difference is called the control error: e = SP − PV. From that error the controller calculates an output signal, the Manipulated Variable (MV), which goes to the actuator. A PLC or DCS repeats this calculation at a fixed cycle time, often a fraction of a second for fast loops, which makes control a real-time task.

The chain looks like this:

Setpoint → comparator → controller → Actuator → process → Sensor → back to the comparator

Example: an oven. The setpoint is 180 °C. A PT100 resistance thermometer measures 175 °C, so the control error is +5 °C. The controller increases the power to the heating element. As the temperature approaches the setpoint, the error shrinks and the controller reduces the power again, so the oven does not overshoot.

A controller in a PLC or DCS also has operating modes: in manual (MAN) the operator sets the output directly, in automatic (AUTO) the controller calculates it, and in cascade (CAS) the setpoint comes from a higher-level controller.

🔄 What is the difference between open-loop and closed-loop control?

Control engineering distinguishes between open-loop control, which acts without feedback, and closed-loop control, which uses feedback. Dutch has two separate words for this: sturen (open-loop) and regelen (closed-loop).

Closed-loop control is the industry standard because it also corrects disturbances nobody anticipated. Feedforward is not a synonym for open-loop: it is an addition that in practice almost always runs alongside a feedback loop.

🧠 What types of controllers are there?

PID is by far the most widely used algorithm. In a Honeywell study of more than 11,600 controllers at eighteen sites in refining, chemicals and pulp and paper, the median site used a PID algorithm for more than 97 percent of its controllers. Advanced control builds on top of this: an MPC usually manipulates the setpoints of existing PID loops rather than driving the valves directly.

🕰️ Where does control engineering come from?

• 1788 - James Watt applies the centrifugal (flyball) governor to his steam engine: spinning balls automatically regulate the engine speed.
• 1922 - Nicolas Minorsky analyses the three-term controller (P, I and D) for automatic steering of US Navy ships, based on how a helmsman steers.
• 1942 - John Ziegler and Nathaniel Nichols of Taylor Instruments publish tuning rules for pneumatic controllers that are still used as a reference.
• 1970s–1980s - controllers move from pneumatics and analogue electronics to digital DCS and PLC systems, with signals over 4-20 mA.
• Today - control loops run in software at level 1 of the Purdue Model and are connected to IT systems via SCADA and the historian.

🏭 Which control loops do you find in industry?

According to the same Honeywell analysis, the largest share of all loops controls a flow. Each type has its own dynamics and therefore its own approach:

The tags come from ISA-5.1, published in October 2024 as Instrumentation and Control Symbols and Identification: the first letter is the measured variable (F, L, P, T, A), the following letters are the function (I for indication, C for controller). On a P&ID you can therefore see at a glance which loop controls what.

Not every level loop has to hold the level tightly. For a surge vessel the goal is to absorb fluctuations in the outgoing flow (surge attenuation); a level controller tuned too tightly then passes the disturbance on to the rest of the plant.

Dutch and European examples:

• Water management - pumping stations control polder water levels with start and stop levels; drinking water companies control pressure in the distribution network and chemical dosing based on measured water quality.
• Buildings - building automation controls room temperature, air quality and flow temperature. Weather compensation shifts the flow temperature setpoint with the outdoor temperature along a heating curve. EU Regulation 811/2013 classifies such controllers into temperature control classes I to VIII, which add 1 to 5 percentage points to the seasonal efficiency of a heating package.

📊 How do you judge whether a control loop performs well?

A good control loop responds fast enough, does not overshoot much and settles down. Common criteria for a setpoint step are overshoot (how far the PV goes past the setpoint), settling time (how long it takes for the PV to stay within a band around the setpoint), oscillation and variability at a constant setpoint.

In practice, performance is often disappointing. Honeywell assessed 26,000 PID controllers in the continuous process industries:

Only a third therefore performed acceptably or better. Common causes are a worn or sticking control valve (stiction), wrong or never-revised tuning, a badly located or fouled sensor, and operators who put a restless loop in manual. Every loop in manual means the automation is not doing its job.

🛠️ How do you tune a control loop step by step?

1. Check the hardware first - test whether the valve or drive responds linearly and without backlash, and whether the measurement is correct. Tuning does not fix mechanical problems.
2. Run a step test in manual - change the output by a few percent and record the PV response, for example in the historian.
3. Identify a simple model - process gain, time constant and dead time (a first-order-plus-dead-time model), or an integrating slope for level.
4. Choose a tuning method - Ziegler-Nichols (1942) gives fast but often aggressive tuning; lambda tuning lets you choose the desired closed-loop time constant λ, with a PI loop completing a setpoint step in roughly 4λ without overshoot; the SIMC rules by Sigurd Skogestad (2003) are a robust, simple alternative.
5. Test in automatic - apply a setpoint step and assess overshoot, settling time and valve movement.
6. Record and manage - document the parameters and route later changes through change management.

🔐 Why is the control loop a target for cyber attacks?

The control loop is where digital commands have physical consequences. Whoever can change a setpoint, a tuning parameter or the mode of a loop controls the process. MITRE ATT&CK for ICS describes this as, among others, technique T0836 Modify Parameter.

• Stuxnet manipulated the output frequency of the drives powering centrifuges, while operators were shown normal values.
• TRITON targeted the safety instrumented system of a petrochemical plant in Saudi Arabia in 2017: the layer that has to intervene when normal control fails.

That is why IEC 61511 separates the basic process control system (BPCS) from the safety system: an ordinary control loop may claim a risk reduction factor of no more than 10 as a protection layer, unless it is designed to the functional safety standards. IEC 62443 adds zones and conduits, restricted write access and logging. Enforce setpoint limits in the controller itself, not only on the HMI, and monitor write actions to control loops.

❓ Frequently asked questions

What is the difference between a control loop and a controller?

A control loop is the complete chain of measuring, comparing and correcting, including the sensor, actuator and process. The controller is one part of it: the device or software block that calculates the output signal from the control error. Without a sensor and an actuator, a controller cannot control anything.

What is a control error?

The control error is the difference between the setpoint and the measured process variable, e = SP − PV. A controller tries to make that error as small as possible. A P controller usually leaves a small steady-state offset, which the integral action of a PI or PID controller removes.

Why is PID used so often in control loops?

PID is used so often because the algorithm works well enough in almost every control loop, is easy to understand and is built into virtually every PLC and DCS. In a Honeywell study, the median plant used PID for more than 97 percent of its controllers. A different algorithm costs considerably more engineering and maintenance.

What is the difference between open-loop and closed-loop control?

Open-loop control executes a command without measuring whether the result is correct. Closed-loop control measures the result with a sensor and corrects the error automatically. As a result, only closed-loop control removes unexpected disturbances.

Why is a control loop in manual?

A control loop is in manual when an operator has switched off automatic control and sets the output directly. This often happens because the loop is restless due to poor tuning, a sticking valve or a faulty measurement. In a study of 26,000 controllers, more than a third were in manual or had their output saturated.

How do you protect a control loop against manipulation?

You protect a control loop by restricting write access to setpoints and parameters, enforcing hard limits in the controller and logging every change. Network monitoring can flag unexpected write commands. An independent safety instrumented system according to IEC 61511 prevents a manipulated control loop from leading to a hazardous situation.

📌 In summary

A control loop ensures that a system reaches and maintains its desired state, the Setpoint, by continuously measuring, comparing and correcting. In the process industries the median plant uses PID for more than 97 percent of its loops, yet only a third perform acceptably or better: good instrumentation, systematic tuning, managed changes and a separate safety system make the difference.