Power Quality, Demand & Metering: A Practical Guide

Power factor is the ratio between real power (kW, the power that actually does useful work) and apparent power (kVA, the total power the electrical system — transformers, cables, generators — actually has to be sized to deliver). A load with a low power factor draws more current for the same real power output than a load with a high power factor, because a larger share of that current is reactive current shuttling back and forth to sustain magnetic fields (in motors, transformers, and other inductive loads) without doing any net useful work. This is why poor power factor is fundamentally a capacity and current problem, not an energy-consumption problem in the usual sense — it doesn't necessarily waste kWh on the meter, but it does force every piece of upstream equipment (transformers, switchgear, cables) to be larger than it would otherwise need to be, and it's exactly why many utility tariffs apply a direct power-factor penalty (covered further below).

Power factor correction (PFC) capacitor banks work by supplying local reactive power right at the load, so the reactive current no longer has to be drawn all the way from the source — IEC 60831 governs the design and rating of the capacitor banks themselves, sized here to move the site from its existing power factor to a chosen target.

Capacitors and the system's own inductance (transformer and cable reactance) together form a resonant circuit, and every capacitor bank added to a system shifts that resonant frequency. If the resulting resonant point happens to land near a harmonic frequency that's already present on site — very plausible on a site with significant VFD, UPS, or other non-linear load — the capacitor bank can amplify that harmonic rather than simply improving power factor, sometimes dramatically. This is exactly why a harmonic resonance check (informed by IEC 61000-4-7 measurement guidance and the site's short-circuit MVA) belongs immediately after power factor correction sizing in this category's typical workflow, not as an unrelated, optional afterthought.

IEEE 519-2014 splits harmonic distortion limits into two genuinely different tables with different underlying logic. Table 1 sets voltage distortion limits at the point of common coupling — this is fundamentally the utility/system operator's responsibility, since voltage distortion reflects the overall health of the supply as seen by every customer connected to that point. Table 2 sets current distortion limits for an individual customer's injected harmonic current, referenced to Total Demand Distortion (TDD) against that customer's own maximum demand current (IL) — this is the individual site's responsibility, since it's specifically about how much harmonic pollution one customer is allowed to inject back into the shared system.

Table 2's limits also scale with the ratio of available short-circuit current to load current (Isc/IL) at the point of connection — a site connected to a comparatively 'stiffer' supply (high fault level relative to its own load) is allowed proportionally more harmonic current injection than one on a comparatively 'weaker' supply, because the same absolute harmonic current represents a smaller disturbance to a stiffer system. Mixing up which table applies to which measurement, or applying a current-injection mindset to what's actually a voltage-quality question (or vice versa), is a common and easy conceptual error.

The whole premise of demand-factor sizing is that not every piece of connected load in a category runs at its full rated power at the same instant — a demand factor for each load category (lighting, sockets, HVAC, heating, and so on) captures the realistic simultaneous-use fraction for that category, based on general practice and typically refined against a specific site's actual metered behavior over time. Lighting circuits, for instance, are commonly assigned a high demand factor (most installed lighting genuinely does run concurrently during occupied hours), while general socket outlets get a meaningfully lower one (most connected equipment on general-purpose outlets is not drawing full rated power simultaneously). Summing connected load at 100% demand factor across every category would systematically oversize the incoming supply, transformer, and switchgear relative to what the site will actually draw in practice.

Voltage unbalance (the NEMA-style definition: maximum deviation of any phase voltage from the average of all three, divided by that average) is disproportionately damaging to motors specifically because unbalanced supply voltage produces a negative-sequence current component, and the additional heating that negative-sequence current causes in a motor's rotor scales roughly with the square of the unbalance percentage — not linearly. This is exactly why the published NEMA MG1/ANSI C84.1 derating curve drops off increasingly steeply as unbalance rises (a jump from 1% to 2% unbalance costs meaningfully more derating than the jump from 0% to 1%), and why motors are explicitly flagged as needing manufacturer consultation above roughly 5% unbalance rather than having their derating simply extrapolated further along the same curve.

Balancing loads evenly across the three phases in a panel isn't only about not overloading any single phase — it also directly affects neutral conductor current. In a perfectly balanced linear three-phase system, the vector sum of the three phase currents returning through the neutral is close to zero; any imbalance between phases leaves a residual current that has to return through the neutral conductor instead, so an unbalanced panel doesn't just waste capacity on the lighter-loaded phases, it also increases neutral loading in proportion to that imbalance. This becomes particularly important on panels with significant non-linear (harmonic-producing) load, where triplen harmonics (3rd, 9th, 15th, and other odd multiples of 3) from single-phase non-linear loads can actually add together in the neutral rather than cancel — a phenomenon that can drive genuinely high neutral current even on a panel that looks reasonably balanced on the fundamental (50/60Hz) current alone.

Converting between real power, apparent power, reactive power, and current is straightforward algebra (kVA = kW / PF, kVAR = kVA × sin(the power factor angle), and so on) — but the current formula itself differs by system type, and using the wrong one is one of the more common practical errors in power-quality work: I = P / (V × PF) for single-phase, but I = P / (√3 × V × PF) for three-phase, where V is the line-to-line voltage. Missing the √3 factor (or double-counting it) produces a current figure off by a factor of roughly 1.73 — large enough to meaningfully mis-size a breaker or conductor if the error goes unnoticed.

Every other topic in this category traces back to a published engineering standard — IEC 60831, IEEE 519, NEMA MG1. Utility tariff structures don't have an equivalent universal reference: rate schedules, time-of-use period definitions, demand charge structures, and power-factor penalty formulas are all set independently by each utility and regulator, and vary enormously between them. A demand charge specifically bills for the site's peak kW drawn during the billing period — separately from, and often in addition to, straightforward energy (kWh) charges — which is precisely why smoothing out peak demand (through load scheduling, or the maximum-demand diversity concepts covered above) can reduce a bill even without reducing total energy consumption. Many tariffs also apply a direct power-factor penalty, which is the most concrete financial argument for the power factor correction work covered at the start of this category — poor power factor isn't just a capacity problem, it can be a direct line item on the bill.