What Have I Learned? – Cascade Control (Part 1)

A cascade control system has a secondary (inner or slave)) loop that gets a remote set point that is the output of a primary (outer or master) loop. The set point of the secondary loop is driven to meet the needs of the primary loop. Most of the benefits stem from the secondary loop correcting for disturbances, nonlinearities, and non-self-regulation before they affect the primary loop. A more obscure benefit is the speeding up of the primary loop by decreasing its natural frequency, particularly when there is a secondary process lag. All of these benefits depend upon the cascade rule that the secondary loop be sufficiently faster than the primary loop. The trend plot SlowSecondaryLoopOscillations.pdf shows how oscillations break out when the secondary loop is slowed down by a factor of five.

Where do you cascade control systems and how many do you already have in service?

How many control valves do you have? Would you believe you should have as many cascade control systems as you have control valves?

Every control valve connected to a digital control system should have a digital valve controller (DVC). The DVC is a high gain fast secondary loop that takes care of most of the non ideal stuff that can occur with valve position due to backlash, friction, and actuator response plus give you diagnostics on the valve’s health and dynamic capability, and feedback (readback) of the actual valve position. In the old days, with analog loops, putting a pneumatic positioner on a fast loop was stated to be a “no-no”! The solution (the use of a booster instead of a positioner) was worse than the problem as discussed in the chapter “Compressor Surge Control – Traveling in the Fast Lane” in the E-book A Funny Thing Happened on the Way to the Control Room and in the Chapter “Instrument Requirements” in the E-book Centrifugal and Axial Compressor Control. In reality, even in the old days, the analog flow loop was usually tuned so slow, the cascade rule was not really an issue. Of course, some positioners were poorly designed, particularly the spool type single stage positioners slapped onto on-off valves posing as throttle valves. Every now and then I see the question still asked; when should you use a valve positioner on a control valve? Academics and people stuck in the mindset of the days of analog controllers will say “important slow loops.” The right answer in my book is “every loop” if you are talking about an electronic high performance positioner (e.g. DVC) unless you really don’t care what the valve is doing.

The next most common secondary loop is the flow loop, which corrects for pressure upsets and valve characteristic nonlinearities before they affect the primary loop. Most of the common primary loops (e.g. composition, pressure, level, and temperature loops) can benefit from cascade control. If you are going to do flow ratio or flow feedforward control, secondary flow loops are almost essential. Most secondary flow loops should have secondary valve position loop forming a triple cascade control system.

There are exceptions as to when a secondary flow loop is useful. If the flow measurement has significantly less rangeability than the control valve or excessive noise or failure rate, a secondary flow loop can do more harm than good. In 3 element boiler drum level control, the level controller output switches from cascade control of a secondary flow loop to direct manipulation of boiler feed water valve at low loads because of the insufficient rangeability of the differential head flow meter on the feed water.

Liquid or polymer and some gas pressure loops are too fast to have a secondary flow or valve position loop. In general, the controller output of these extremely fast pressure loops should go directly to a variable speed pump via a high resolution input card with a suitably designed variable speed drive with minimal velocity limiting and no deadband. In some cases, the pressure loop should use an analog electronic controller or a DCS with special fast scan and execution time.

Inline (e.g. pipeline or static mixer) pH loops have a response almost as fast as the flow loop. The pH loop must consequently be detuned to be slowed down enough to satisfy the cascade rule. Also, the flow loop often lacks the rangeability needed for pH control and flow ratio control is inexact at best due to the extreme effect of immeasurable changes in feed concentration. Most inline pH loops perform better if their output goes directly to a final element with good resolution and minimal deadband. The exception is when there are Coriolis feed and reagent flow meters, a relatively constant feed composition, and the pH set point is on the relatively flat part of the titration curve making mass flow ratio control more sensitive than pH. If there is no flow feedforward, a “head start” to momentarily preposition the valve or the use of signal characterization helps the pH loop deal with startup and large load disturbances.

In some cases, the process gain of an equal percentage valve characteristic, which is proportional to throttle flow rate, compensates for a process gain that is inversely proportional to load (e.g. feed rate). The most common cases are inline concentration and pH control and heat exchanger temperature control. The use of a secondary flow loop removes this compensation of the process gain making the primary loop more nonlinear.

In part 2 we look in greater detail at the cascade rule, the use of reset in the secondary loop, and how dynamic reset limiting with external reset is a powerful tool for cascade control. In part 3 we conclude with the “Rules of Thumb” summary for cascade control.