Cascade Control

You may find the answer in the following book:

"Automatic Control Systems"

By KUO

My first question is what you are trying to accomplish with two flow loops in cascade control. The typical reason for using cascade control is to control the setpoint of a fast loop from a slower loop. The classic example is controlling the temperature of a steam-jacketed vessel by controlling steam pressure. The temperature control is a very slow reacting value compared to the pressure loop. By cascading two flow loops that have very similar reaction times it is much more difficult to tune. At some valve positions the inner loop may actually be slower than the outer loop.

How are air flow and oil flow dependent? Is this a boiler application where you need to vary oil flow based on air flow? In that case you should use a ratio controller instead of a cascade. A cascade loop in your case would indicate that you are controlling air flow by controlling oil flow. In other words you would have one final element (valve) instead of the two you have.

Hope this helps, but let me know if I can help further.

Dear Andrew , This is a cracker where the raw material is oil , we are regulating the oil flow according the air flow . ie according to the air rate into the reactor we have to control the oil flow.

I would still use ratio control in that case, but reverse the oil and air that was suggested earlier. Determine what the proper ratio of oil to air is and control your oil setpoint based on air flow. When you use a cascade, you are controlling your oil setpoint from your air valve position not the actual flow rate. Butterfly valves are notoriously non-linear and the position is a poor guide to flow rate.

Be aware that the ratio of oil to air may be non-linear as well. Your ratio controller may not work well with a straight multiplier but may require calculating a slope and offset to the air flow to determine oil setpoint.

If air flow is meant to be matched to oil flow (for example, to match the proper quantity of combustion air to fuel-oil to be burnt), then what you need is ratio control, not cascade control.

Measure your oil flow, calculate the quantity of air needed, and se the calculated quantity as the setpoint for a PID controller that will control your air flow.

Simple as that.

Cheers! DZ

Get rid of the butterfly valve...now! First B/Fs are notorious for not seating, build up a lot of hysteresis in the controller which results in "chasing" the setpoint. Could try spit-range. Using a high gain with a sizeable derivative action, but Nothing is going to work with the butterfly in a cascade operation

Like Mr. Ward, I wonder whether this is a combustion application, where ratio control is needed. If that's the case, ratio control (rather than cascade), and characterizing the butterfly performance to match the fuel valve's response could provide much tighter control.