There has been a tendency to attribute disturbances and failures to power surges, a term often used by the media but rather ill-defined. The ambiguity results in part from an unfortunate dual definition of the word surge.
a) To some people, a surge is indeed the phenomenon being discussed here, that is, a transient voltage or current lasting from microseconds to at most a few milliseconds, involving voltages much higher than the normal (two to ten times).
b) To other people, a surge is a momentary overvoltage, at the frequency of the power system, and lasting for a few cycles, with voltage levels slightly exceeding the five to ten percent excursions that are considered normal occurrences.
The term swell has been adopted by this recommended practice to describe this second type of overvoltage; perhaps one day it will supplant the usage of surge for that meaning. It would be a mistake to attempt protection against these long-duration power frequency swells with a surge protective device that is designed to absorb large but short impulses of energy.
There is a growing recognition that the horror tales of surge protective device failures are more likely
to be caused by swells rather than by large surges.
Nonlinear loads draw nonsinusoidal currents from the power system, even if the power system voltage is a perfect sine wave. These currents produce nonsinusoidal voltage drops through the system source impedance, which distort the sine wave produced by the power plant generator.
A typical nonlinear load is a dc power supply consisting of rectifiers and a capacitor-input filter, such as used in most computers, drawing current only at the peaks of the voltage sine wave. This current has a high third harmonic content that has also created a new concern, that of insufficient ampacity of the neutral conductor in a three-phase system feeding power supplies.