Insulation Coordination

Insulation coordination is the selection of the insulation strength of equipment, and of the protective devices that guard it, in relation to the voltages that can appear on the system and the environment in which it operates. The aim is a design that is strong enough to keep failures acceptably rare, yet economical enough to build — for insulation is expensive, and over-insulating every component "to be safe" would make a power system unaffordable. Coordination is the art of buying just enough strength, in just the right places.

The organising principle is a ladder of levels. At the bottom sits the continuous operating voltage; above it the temporary and switching overvoltages; and capping them, the arrester's protective level — the ceiling of Chapter 24. Every insulation in the station is then given a withstand strength that lies above that ceiling by a margin. Crucially, the levels are also ordered among themselves so that the weakest planned path is always an external, self-restoring one — an air gap or a line insulator that simply flashes over and recovers — rather than the internal, non-self-restoring insulation of a transformer, which would be destroyed. Coordination thus channels the rare failure to where it does least harm.

Coordination balances two columns: the stresses that appear and the strengths that resist them. The stresses are the overvoltage classes already met — continuous power-frequency, temporary, slow-front switching, and fast-front lightning (plus very-fast fronts in gas-insulated switchgear). The strengths are expressed as standardised rated withstand voltages, the voltages an insulation is guaranteed to survive in a defined test:

The BIL (basic lightning impulse insulation level) is the headline number that appears on nearly every nameplate; the SIL (switching impulse level) joins it at the highest voltages, where, as Chapter 23 showed, slow-front stress governs. Coordination is fundamentally the act of choosing these withstand numbers so that each comfortably exceeds the corresponding stress, with the arrester holding that stress down to a known protective level in the first place.

A distinction runs through the whole subject and decides which method applies. Self-restoring insulation — air gaps, the air around line insulators, external clearances — fully recovers its strength after a flashover; the arc clears, the air de-ionises, and the insulation is as good as before. Non-self-restoring insulation — the oil-paper inside a transformer, the solid dielectric of a cable, the gas-barrier of a bushing — is permanently damaged by a single breakdown; one puncture and the component is ruined.

The consequence is profound. For self-restoring insulation an occasional flashover is a nuisance, not a disaster, so we can accept a small, calculated probability of failure and design statistically. For non-self-restoring insulation no failure can be tolerated, so it is protected almost absolutely, by a firm deterministic margin above the arrester level. The two methods that follow are simply these two attitudes made quantitative.

The older, simpler approach treats both stress and strength as single fixed numbers. The highest overvoltage that can reach the equipment is held to the arrester's protective level, and the insulation's withstand is set above it by a fixed factor. The quality of the coordination is the protective ratio (or coordination factor):

A ratio of around \(1.2\) or more — equivalently a protective margin of \(20\%\) or more — is the usual target, with the surplus covering the separation effect, ageing and uncertainty. This deterministic method is exactly right for non-self-restoring insulation, where the idea of an "acceptable failure rate" is meaningless and a clear, assured margin above the worst credible stress is what is wanted. Its weakness is that, by always assuming the worst case, it tends to over-insulate — which is acceptable for the transformer but wasteful if applied to the vast air clearances of an EHV station.

For self-restoring insulation the deterministic worst case is needlessly expensive, because both the overvoltage and the withstand are really distributions, not single values. Switching overvoltages, being statistical (Chapter 23), are described by a probability distribution of crest values; and the flashover of an air gap is itself probabilistic, described by a \(50\%\) flashover voltage \(U_{50}\) with a spread, the conventional deviation \(\sigma\). The statistical method works with both distributions at once.

The risk of failure is the overlap: the probability that, on a given event, the overvoltage happens to be high and the gap happens to be weak at the same moment. Formally it is the integral of the overvoltage probability density times the gap's flashover probability across all voltages. Rather than demand zero overlap (infinite insulation), the designer accepts a small risk — perhaps one failure in a thousand events — and sizes the gap accordingly. Two statistical quantities anchor the calculation: the statistical overvoltage \(U_2\) (the value exceeded by only \(2\%\) of overvoltages) and the statistical withstand \(U_{10}\) (the value the gap withstands \(90\%\) of the time, i.e. \(U_{10}=U_{50}-1.28\sigma\)). Their ratio is the statistical safety factor:

To make coordination uniform worldwide, the standard IEC 60071 lays down a common vocabulary and a fixed menu of withstand levels. Equipment is identified by its highest voltage for equipment \(U_m\) — the maximum sustained rms line-to-line voltage it is built for — and the standard splits the voltage world into two ranges, because different stresses dominate in each:

In Range I (distribution and sub-transmission up to \(245~\mathrm{kV}\)) the insulation is verified by a lightning-impulse withstand and a one-minute power-frequency test; switching surges are mild enough to be covered implicitly. In Range II (EHV and UHV above \(245~\mathrm{kV}\)) the switching-impulse withstand becomes a separate, governing requirement alongside the lightning one. For each \(U_m\) the standard offers a short list of permissible withstand values — so the engineer does not invent a BIL, but selects the lowest standard value that satisfies the coordination calculation.

IEC 60071 turns all of this into a clear chain that carries the design from the overvoltages a point will see to the standard withstand voltage stamped on the equipment. Each step multiplies by a factor that accounts for one source of uncertainty:

Reading the chain: the representative overvoltage \(U_{rp}\) is what the location actually sees (set by the arrester for the protected equipment). Multiplying by the coordination factor \(K_c\) — which embeds the chosen risk — gives the coordination withstand voltage \(U_{cw}\), the strength needed in service. Multiplying again by the atmospheric correction \(K_a\) (for altitude) and the safety factor \(K_s\) (for ageing and manufacturing scatter) gives the required withstand voltage \(U_{rw}\). Finally the engineer selects the nearest standard withstand voltage \(U_w \ge U_{rw}\) from the IEC list — and that BIL or SIL is what the equipment is built and tested to.

One correction deserves singling out because it bites hardest in practice: altitude. The strength of external (air) insulation falls as the air thins, so a clearance that is ample at sea level may flash over in a mountain substation. IEC accounts for this with the atmospheric correction factor

where \(H\) is the altitude in metres and the exponent \(m\) depends on the voltage type (near \(1\) for lightning-impulse and power-frequency withstand of air gaps). At \(2000~\mathrm{m}\) this already demands roughly a quarter more withstand than at sea level — a real cost for high-altitude grids. Humidity and temperature carry their own smaller test corrections, but altitude is the dominant environmental term.

Pull the threads together and a coordinated substation emerges as a tidy hierarchy. The metal-oxide arrester, mounted hard against the transformer, sets the lowest ceiling — the protective level. The transformer, the precious non-self-restoring heart, is given a BIL above that ceiling by a firm deterministic margin. The bus, disconnectors and bushings are coordinated next, and the external air clearances are set highest and designed statistically, so that any overvoltage that somehow defeats the arrester flashes harmlessly across an air gap rather than puncturing the transformer. Every nameplate BIL and every metre of clearance in that station is the output of the chain in Section 25-7 — the quiet, unseen calculation that lets a power system ride out the sky's worst and its own switching, year after year, at a price the world can afford.

For each item, first identify what it tests — the coordination idea, the stress/withstand pairing, the self-restoring distinction, the deterministic ratio, the statistical method, the IEC framework, or the procedure chain — then apply it. Difficulty rises down the list.