How to: Cascade Control
Q1-2024
Let's start with the basics. Cascade control is a control strategy where you have two PID control loops. One PID loop feeds a setpoint into another loop.
To make things more practical to discuss, let's take two classic examples:
A level controller is cascaded to the flow of a tank's outlet
A heat exchanger's product temperature is cascaded to the steam flow
In Example 1, the level controller's output sets the setpoint of the flow controller.
In Example 2, the product temperature controller sets the setpoint of the steam flow controller.
In each example, we really only care about controlling one variable:
The tank level
The product temperature
You can start to see how these applications are similar. Each loop of a cascade pair has a name to refer to where it falls in the control strategy. There are actually a few naming conventions that are used interchangeably in the industry.
PID1
Outer Loop
Master Loop
Primary Loop
Example 1: Level
Example 2: Product Temperature
PID2
Inner Loop
Slave Loop
Secondary Loop
Example 1: Outlet Flow
Example 2: Steam Flow
So why do we add extra cost of the flow meters and complexity of a cascade control loop?
Because this is a feedback control, we can only respond to a process deviation. If our examples were not cascade control:
The level can only be corrected after an outlet stream causes the level to increase or decrease.
The process temperature can only be corrected after a steam pressure loss/spike causes the product to change temperature.
The primary reason for cascade control is to respond to disturbances before they impact the Outer Loop's measured variable. For plants that have very tight specifications and/or highly variable disturbances, a cascade control strategy allows disturbances to be controlled before they have large, if any, impact to the outer loop. This is why plants will pay the money to install the extra instrument.
Why doesn't feed forward work here?
At first these two might seem similar. The inner loops of a cascade control strategy measures disturbances, and feedforward also measures disturbances. A simple explanation is that feedforward should be implemented when the final control element (control valve, pump, etc.) does NOT have a direct impact on the measurement. If feedforward were to be used, it would create a feedback loop. For this reason, feedback control is used.
Can you recognize a good opportunity to implement cascade control?
If you see all these criteria, you should certainly consider it!
Your process measurement & control is slow, and is impacted by faster disturbances
There's an instrument you can add to measure those disturbances - even better if it's already installed!
Your final control element can be used to impact your disturbance measurement
Your disturbance measurement has an impact on your original process control measurement
Now you know the basics of cascade control. If your cascade loops are set up correctly and they still aren't responding well, please reach out to us! Tuning cascade loops can be tricky, and our expertise can help assist you in diagnosing the problems.