A fixed capacitor bank corrects power factor for one load level, but real facilities swing between light and heavy load all day, and correcting for one point means over- or under-correcting for every other. Automatic power factor correction, driven by an APFC controller, solves this by switching capacitor steps in and out to follow the reactive demand as it changes. This guide explains how the controller senses reactive power and decides which steps to switch, the role of the C/k sensitivity setting and the switching deadband, why steps are rotated for even wear, and where detuned reactors fit in.
APFC Step Control in one line: Automatic power factor correction is a control scheme in which an APFC controller continuously measures a facility's reactive demand and switches capacitor bank steps in or out to hold the power factor at a target value. It uses a C/k setting to know the smallest step's current relative to the sensing current so it can judge when adding or removing one step is worthwhile, and a switching deadband to avoid hunting between steps. It typically rotates which steps switch to spread wear evenly, and on harmonic-rich sites the capacitor steps include detuning reactors.
The APFC controller's job is to keep the measured power factor near a target, commonly a modestly lagging value chosen to avoid utility penalties without over-correcting into a leading condition. To do that it needs to know the reactive power the facility is currently drawing, which it derives from voltage and current measurements at the point of connection. Comparing the present power factor against the target tells the controller whether it needs to add reactive support, remove it, or leave the bank as it is.
The correction is delivered in discrete chunks. The capacitor bank is divided into steps, each a block of capacitance switched by a contactor or, for fast duty, a thyristor switch. When the reactive demand rises and power factor drops below target, the controller switches in more steps to supply additional reactive power; when demand falls and the power factor threatens to go leading, it drops steps out. Because the steps are finite in size, the controller can never hit the target exactly, only bring the power factor within one step of it, which is why step sizing matters as much as total bank size.
Some banks use equal steps and others use binary or weighted steps to get finer resolution from fewer physical stages, and the controller's switching logic accounts for the available step sizes when it decides which combination to energize. Throughout, the controller is balancing responsiveness against switching wear: it wants to track the load closely enough to hold the target, but not so eagerly that it switches contactors constantly, since each switching operation stresses the capacitors and the switchgear.
The pivotal setting in an APFC controller is C/k, which tells the controller the reactive current of the smallest capacitor step as seen through the measuring current transformer. It defines the sensitivity of the whole scheme: how big a change in reactive demand justifies switching one step. Set C/k too low and the controller reacts to trivial changes, switching steps needlessly and hunting; set it too high and the controller becomes sluggish, tolerating a power factor well away from target before it finally acts. Getting C/k right, matched to the smallest step and the CT ratio, is the single most important commissioning step for stable operation.
Working alongside C/k is a switching deadband, a band around the target inside which the controller holds still. The deadband exists because the correction is quantized: if the controller tried to switch the instant the power factor crossed the target, it could add a step, overshoot to the leading side, immediately drop it again, and oscillate forever. The deadband is set so that a single step change moves the power factor comfortably within the band rather than to the far edge, so the controller makes a correction and then rests instead of chattering between two states.
Because the same few steps would otherwise take all the switching duty, controllers rotate the order in which steps are switched, often on a first-in-first-out basis, so that operations and therefore wear are spread evenly across the contactors and capacitors. Rotation also allows discharge time between re-energizing any given step, which matters because a capacitor must discharge before it is switched back in to avoid a large inrush. Even wear extends the life of the whole bank and prevents one hard-worked step from failing far ahead of the rest.
On sites with significant harmonic-producing loads, such as variable frequency drives and rectifiers, plain capacitor steps are risky because the capacitors and the system inductance can resonate near a harmonic and amplify it, overheating the capacitors and distorting the supply. The standard remedy in an APFC bank is a detuning reactor in series with each capacitor step, sized to move the resonance below the lowest significant harmonic. Detuned steps also limit switching inrush, so on any harmonic-rich installation the APFC bank is normally specified as detuned rather than plain, and the controller still switches those detuned steps in exactly the same way.
An APFC bank fails quietly and expensively. A blown capacitor can, a stuck or welded contactor, a failed detuning reactor, or a controller with drifted settings can all leave the correction partly or wholly inoperative while the process runs on undisturbed. The first anyone notices is often a power-factor penalty reappearing on the utility bill, weeks after the correction silently stopped tracking, by which time the cost has already accrued.
This is why the health of an APFC scheme is worth monitoring continuously rather than checking on the occasional site visit. Controllers expose the measured power factor, reactive power, and the number of steps in service, and a cloud SCADA platform such as Merobix can pull those values from banks across every facility and trend them against the loads that create the reactive demand. A power factor that never reaches target, a step count that has stopped changing, or a bank that has drifted leading at light load all become visible in real time, so a failing step is caught and repaired before the next penalty lands on the bill.
C/k tells the controller the reactive current of the smallest capacitor step as seen through the measuring current transformer, which sets the sensitivity of the whole scheme. It determines how large a change in reactive demand justifies switching one step. Set too low it causes hunting and needless switching; set too high the controller is sluggish and tolerates poor power factor. Matching C/k to the smallest step and CT ratio is the key commissioning step.
Without rotation the same first steps would take almost all the switching duty, wearing their contactors and capacitors far faster than the rest. Controllers rotate the switching order, often first-in-first-out, so operations and wear spread evenly across the bank. Rotation also gives each step time to discharge before it is re-energized, which is necessary to avoid a large inrush when switching a capacitor back in.
On sites with harmonic-producing loads like variable frequency drives, yes. The capacitors and system inductance can resonate near a harmonic and amplify it, overheating the capacitors and distorting the supply. A detuning reactor in series with each step shifts the resonance below the lowest significant harmonic and also limits switching inrush. On harmonic-rich installations detuned steps are standard, and the controller switches them the same way as plain steps.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.