cascade control, feed forward control


Thread Starter

sandeep datar

Why and when a cascade control and feed forward controls?
what are the benifits of both types of controller?
In a cascade control why output of`master`is used to `drive`slave? what will happen if o/p of master is directly used as final control.

Bill Clemons

Cascade control looks at two process variables and applies two successive PID controls to effect the desired control operation. The master (primary) loop controls the desired process condition (for example, batch reactor temperature). The slave (secondary) control loop receives its set point (output from the primary loop, scaled to engineering units of the secondary loop) to effect the final control element (for example, steam flow valve to reactor jacket pressure). While the primary goal is to control batch reactor temperature, which could be performed by a single loop to control the steam flow valve (the answer to your last question). A secondary or cascade loop is added, which updates its set point around current temperature requirement. Also, the second loop compensates for variations in steam supply pressure, thereby increasing sensitivity to the temperature control.

Feed forward control also looks at two process variables. The desired control operation is effected by a single PID loop, wherein an additional bias is introduced to the loop output based on analysis of an upstream process variable (hence, the name feed forward). An example is an exit water temperature on a natural gas fired boiler. The primary loop measures boiler exit temperature and controls a gas flow valve, thereby adjusting the firing rate. An upstream measurement of the amount of methane in the gas stream to the boiler may be indexed and biased to primary PID loop output. The PID loop will add or subtract additional output (based on methane composition in natural gas) to trim the firing rate for improved temperature control out of the boiler.

John A. Shaw

Why cascade control?
The three most important reasons are:
1) It will allow the faster secondary (or slave) controller to handle disturbances in the slower secondary loop before they can affect the primary (or master) loop.

2) It will allow the secondary controller to handle non-linear valve and other final control element problems. The primary controller (often more difficult to tune and less robust) will more likely see a linear process).

3) It will allow the operator to directly control the secondary loop during certain types of operation, such as startup, when the primary loop is not in operation.

See for more information and an example.

Why is the output of master used to drive setpoint of the slave?
That is the definition of cascade. In a simple, non-cascade furnace temperature loop the output of the temperature controller would directly drive the fuel valve. In a cascade loop the temperature controller drives the setpoint of the fuel flow controller, asking the fuel flow controller to deliver a given amount of fuel.

If the output of the master were connected directly to the valve, then you would have simple non-cascade control.

Why and when feed forward control?
If there is a disturbance that can be measured, feed forward control, combined with the output of the feed back controller, will begin minipulating the valve before the feedback controller sees the effect of the disturbance. It can be used only when there is a disturbance that can be measured and that has a known effect on the controlled variable.


Two questions to ask yourself before going cascade:
Q1: Are the process dynamics of my slave loop at least three times faster than the master loop?
Q2: If the process dynamics are faster but not three times faster, are most of my disturbances coming from the slave loop controlled variable

If you answered NO to the two questions above, disregard cascade as this will less performant than a standard PI controller.

Ben Janvier, principal process control consultant
Emerson, [email protected]