Protection Against Overvoltages

An overhead line may face a lightning surge of megavolts and a switching surge of two or three per unit; the transformer at its end is built to withstand far less. Insulating the transformer to survive the full surge would make it enormous and ruinously costly, so power engineering takes the opposite path: it limits the surge at the equipment to a level the equipment can comfortably bear. The number that matters is the protective level — the highest voltage a protective device will allow to appear — which must sit safely below the equipment's withstand voltage (its BIL for lightning, its SIL for switching).

So protection has two non-negotiable conditions. The device must do nothing at the continuous operating voltage — drawing negligible current and wasting no energy year after year — yet must act decisively the instant a surge arrives, clamping it below the withstand. A device that conducts too eagerly fails the first test; one that clamps too high fails the second. The whole evolution of arrester technology is the search for a device that meets both demands ever more cleanly.

The oldest protector is the simplest: a rod gap — two metal rods separated by an air gap, connected across the insulation to be protected, with the gap set so that it flashes over at a voltage below the equipment's withstand. When a surge arrives, the gap sparks over and short-circuits the surge to earth before the equipment can be harmed. It is cheap, rugged and entirely passive.

But its very simplicity is its weakness. The flashover voltage scatters widely with weather and surge shape, so the protective level is imprecise. Worse, once the gap has flashed over it does not self-extinguish: the power-frequency voltage keeps driving a follow current through the ionised gap, so the surge protection has created a line-to-earth fault that the breakers must clear, interrupting supply. And the abrupt collapse of voltage when the gap fires produces a steep chopped wave that stresses transformer winding insulation severely. For these reasons the plain gap survives only as a cheap backup; real protection needs a device that limits the voltage smoothly and then resets itself.

The ideal protector is a voltage-dependent resistor connected from line to earth. At the normal operating voltage it presents an enormous resistance and draws only a trickle of current — effectively an insulator. When the voltage rises toward a danger level, its resistance collapses, so it conducts the surge current to earth and, by Ohm's law, holds the voltage across itself nearly constant at a safe clamping value. When the surge passes and the voltage falls, its resistance climbs again and it reseals into the insulating state, leaving the system undisturbed. This is the surge arrester: not a switch that opens a fault path, but a clamp that pins the voltage to a ceiling and then quietly steps aside.

The first practical realisation used blocks of silicon carbide (SiC), a ceramic whose resistance falls as the voltage across it rises, following a non-linear law

An \(\alpha\) of about five is helpful but not enough on its own: at the continuous operating voltage a SiC block would still pass a substantial current and overheat. So the SiC arrester places series spark gaps in front of the blocks. In normal service the gaps stand off the operating voltage, isolating the blocks entirely. When a surge arrives the gaps spark over, connecting the SiC blocks, which then limit the discharge voltage; and crucially the same blocks throttle the power-frequency follow current after the surge down to a value the gaps can interrupt at the next current zero, so the arrester reseals without leaving a fault. This gapped SiC design protected the grid for decades, but it carries the gaps' scatter and the complication of follow-current interruption — burdens the next generation would shed.

The modern arrester is built from metal-oxide (zinc-oxide, ZnO) varistor blocks, whose non-linearity is so extreme that the gaps can be thrown away entirely. Its exponent is

The consequence is striking. With \(\alpha\) near thirty, the current changes by orders of magnitude while the voltage barely moves: at the maximum continuous operating voltage the ZnO column draws less than a milliampere — mostly harmless capacitive leakage — so no series gap is needed to isolate it. The arrester is therefore gapless, permanently connected, always watching. When a surge drives the voltage up by a modest fraction, the current leaps from microamps to kiloamps, and the steep characteristic holds the voltage almost flat at the residual (discharge) voltage. With no gaps there is no spark-over scatter, no follow current to interrupt, and no chopped wave — the response is instantaneous and perfectly repeatable. Simpler, faster and more reliable, the gapless metal-oxide arrester has become the universal choice for surge protection from distribution voltages to UHV.

An arrester is specified by a small set of ratings that together guarantee it meets the two demands of Section 24-1. The most important are:

Selection threads a needle. The MCOV must exceed the continuous line-to-earth voltage \(U_s/\sqrt{3}\), and the rated voltage must ride out the temporary overvoltage that the healthy phases see during an earth fault — push these too low and the arrester overheats or fails. Yet the residual voltage, which is the protective level, must stay comfortably below the equipment withstand — push the ratings too high and the protection is too loose. The quality of the protection is measured by the protective margin:

An arrester clamps the voltage at its own terminals — but the equipment it protects sits a few metres away, and that distance matters more than intuition suggests. An incoming surge with a steep front reaches the equipment, reflects off it (often an open or high-impedance point, where the wave tends to double as in Chapter 23), and so the voltage at the equipment can climb above the arrester's clamping level while the surge front is still rising. The faster the front and the greater the separation, the larger this excess. A useful estimate of the equipment voltage is

The practical rules follow directly. Mount the arrester as close as possible to the equipment, above all to a transformer; keep the connecting and earthing leads short, because their inductance adds an \(L\,\mathrm{d}i/\mathrm{d}t\) rise of its own; and provide a low-resistance earth so the diverted current flows away cleanly. For especially sensitive apparatus — rotating machines, whose winding insulation is thin and whose turn-to-turn insulation hates steep fronts — a surge capacitor is added in parallel to slope the wavefront, lowering its steepness \(S\) and so easing both the separation effect and the stress on the first turns.

The arrester gives the system a known, dependable ceiling on overvoltage — its protective level — and the protective-margin inequality ties that ceiling to the strength chosen for each piece of equipment. This is exactly the raw material of insulation coordination: arrange the withstand levels of every component, and the protective levels of the arresters that guard them, into one consistent, economical scheme in which the weakest planned path to flashover is always a safe, self-restoring one (a line gap or the arrester itself) rather than the costly internal insulation of a transformer. The arrester fixes the ceiling; the next chapter shows how the whole system's insulation is then designed beneath it — how the BILs and SILs are selected, how the margins are justified statistically, and how the standards turn this reasoning into the numbers that appear on every equipment nameplate.

For each item, first identify what it tests — the protective-level idea, the rod gap, the arrester concept, the SiC/ZnO characteristic, the ratings and margin, or the separation effect — then apply it. Difficulty rises down the list.