What is cascade control?
Cascade control is a control structure in which two control loops are connected in series: the output of the outer controller (master) is the setpoint of the inner controller (slave). The inner loop controls a fast-responding intermediate variable, such as flow or pressure, and absorbs disturbances before they reach the main variable. The outer loop monitors the actual process variable, such as a temperature or level. After the single PID loop, cascade control is the most widely used control structure in the process industry.
🧠 How does cascade control work?
Example: temperature control of a steam-heated reactor
| Loop | Controls | Measurement | Output to |
|---|---|---|---|
| Master (outer loop) | Reactor temperature | Temperature transmitter | Setpoint of the steam flow controller |
| Slave (inner loop) | Steam flow | Flow transmitter | Control valve in the steam line |
Without cascade, the temperature controller drives the steam valve directly. If the steam header pressure drops, less steam flows through the valve, and only minutes later does the temperature controller see that the reactor is cooling. With cascade, the flow controller notices within seconds that the steam flow is dropping and opens the valve further, before the temperature changes.
🔄 What is the difference between single-loop and cascade control?
| Single control loop | Cascade control | |
|---|---|---|
| Number of controllers | One | Two (master and slave) |
| Response to inner-loop disturbances | Slow, only visible in the main variable | Fast, absorbed by the slave |
| Extra measurement needed | No | Yes, for the intermediate variable |
| Tuning | One controller | First the slave, then the master |
| Complexity | Low | Moderate |
🏭 Where is cascade control applied?
- Temperature → flow — reactor or furnace heating via steam, gas or oil flow
- Level → flow — tank level controlled via outflow, insensitive to pressure fluctuations
- Pressure → speed — line pressure controlled via the speed of a pump with a variable frequency drive
- Position → speed → current — in motion control and servomotors even three cascaded loops are common
- Distillation — product temperature via reflux or heating medium flow
🛠️ How do you tune cascade control?
- Put the master in manual and the slave in automatic
- Tune the slave — make the inner loop fast and stable, usually with a P or PI controller
- Check the speed ratio — the inner loop should be at least three to five times faster than the outer loop
- Put the slave in cascade mode, so that it receives its setpoint from the master
- Tune the master — the slave now behaves as a fast part of the process
- Test with disturbances — check that disturbances in the inner loop are absorbed
Pay attention to switching between manual, automatic and cascade: a good bumpless transfer prevents setpoint jumps when switching modes.
⚠️ What mistakes are commonly made with cascade control?
| Mistake | Consequence | Solution |
|---|---|---|
| Inner loop not faster than outer loop | Loops fight each other and oscillate | Choose an intermediate variable that responds at least 3–5× faster |
| Master tuned before the slave | Settings no longer fit once the slave changes | Always tune the slave first |
| No limits on the slave setpoint | Master demands unrealistic values, such as negative flow | Set limits on the master output |
| No anti-windup | The master’s integrator winds up when the slave saturates | Feed the slave status back to the master |
| Slave left in manual | Cascade does not work, master controls nothing | Show the mode on the HMI and raise an alarm |
These points are also relevant for change management: an unintentionally changed mode or limit in the DCS can silently disable a well-functioning cascade.
❓ Frequently asked questions
What is the advantage of cascade control?
The main advantage of cascade control is that disturbances in the inner loop, such as pressure fluctuations in a steam or gas line, are corrected before they affect the main variable. This keeps the process variable more stable. The inner loop also compensates for non-linear control valve behaviour.
When does cascade control not make sense?
Cascade control does not make sense if the inner loop is not clearly faster than the outer loop, or if the main disturbances do not occur in the inner loop. In those cases the extra controller adds complexity without better control. A reliable measurement of the intermediate variable must also be available.
What is the difference between cascade control and feedforward?
In cascade control, a disturbance is detected through the intermediate variable and corrected with feedback in the inner loop. In feedforward, the disturbance itself is measured and the control signal is adjusted in advance, before the process responds. The two techniques are often combined.
How is cascade control built in a DCS or PLC?
In a DCS or PLC, cascade control is built by linking two PID function blocks: the output of the master controller is connected to the external setpoint input of the slave controller. Modern systems have standard functions for cascade mode, bumpless transfer and feeding the slave status back to the master.
What is the difference between master and slave in cascade control?
The master is the outer controller and controls the actual process variable, such as temperature or level. The slave is the inner controller and controls a faster intermediate variable, such as flow or pressure. The master’s output is the slave’s setpoint; only the slave drives the control valve or motor directly.
Can cascade control consist of more than two loops?
Yes, cascade control can consist of three or more loops. In servo drives a cascade of position, speed and current control is standard, with each inner loop faster than the one above it. In the process industry it usually stays at two loops, because every extra loop makes tuning more complex.
📌 In summary
Cascade control connects two control loops in series: the master controls the main variable and sets the setpoint of a faster slave, which absorbs disturbances before they reach the process. It is the standard solution when a quickly measurable intermediate variable, such as flow or pressure, strongly affects the process variable.