Insulation Coordination

Station Lightning Coordination: Insulation Strength & BIL Selection

Part Three of the series. The other side of insulation coordination — turning calculated surge stress into a required insulation strength. This self-study covers safety factors for non-self-restoring and self-restoring insulation, how a non-standard station waveshape is compared with the standard BIL (the 1.15 BIL rule and its full-wave correction), the transformer chopped-wave, full-wave and long-tail criteria, the suggested transformer BIL criterion and strength curve, transformer bushings, and phase-ground / phase-phase air clearances with altitude correction.

Reading time ≈ 45 min · Part Three of the series

Section 1

From Surge Stress to Insulation Strength

Parts One and Two calculated the surge voltages at points in the station — the electrical stress. This part moves to the other side of insulation coordination: the insulation strength, and how the calculated non-standard surge is compared with equipment withstand to choose a BIL.

The equipment includes transformers and their bushings, circuit breakers, disconnecting switches, bus support insulators, external air clearances and other apparatus. The fundamental distinction throughout is between non-self-restoring and self-restoring insulation — the two are treated very differently.

From Part Two to Part Three

Part Two explained how surge voltages are calculated at the transformer, breaker and other station locations. Part Three now answers the next question: once the surge voltage is known, how should it be compared with the insulation strength of each item of equipment?

The basic comparison

Insulation coordination is not finished when the surge voltage has been calculated. That voltage must be compared with the appropriate withstand strength — \(\text{calculated surge stress} \le \text{insulation strength}\) — and the correct comparison depends on the type of insulation and the waveshape of the surge.

What Part Three covers

safety factors for non-self-restoring and self-restoring insulation;
how a non-standard surge waveshape is compared with the standard BIL;
the \(1.15\,\text{BIL}\) self-restoring rule and its full-wave correction;
the transformer chopped-wave, full-wave and long-tail criteria;
the suggested transformer BIL criterion and strength curve;
transformer bushings, and phase-ground / phase-phase air clearances.

Section 2

Self-Restoring and Non-Self-Restoring Insulation

Non-self-restoring insulation

Non-self-restoring insulation does not recover after breakdown — the key example is transformer internal insulation (also internal bushing insulation, oil-paper and solid insulation). Failure is permanent and usually expensive, so a safety margin is normally required.

Self-restoring insulation

Self-restoring insulation recovers after flashover — external air gaps, phase-to-ground and phase-to-phase clearances, external bushing parts, bus supports, and the external insulation of disconnectors and breakers. Flashover is still undesirable, but the consequence is far less severe than internal transformer failure, so safety margins are treated differently.

Table 1 — The two insulation categories compared.

Insulation Type	Examples	Behaviour	Typical Comparison
Non-self-restoring	transformer internal insulation, internal bushing insulation	damage may be permanent	BIL, chopped-wave and switching-impulse strength, with a safety factor
Self-restoring	air gaps, bus supports, disconnectors, circuit breakers, external bushings	recovers after flashover	CFO / BIL with altitude correction and a smaller or no safety margin

Section 3

Safety Factor for Non-Self-Restoring Insulation

Why a safety factor is needed here

Transformer internal insulation is non-self-restoring. If it fails, the damage may be permanent and repair can mean a major outage or replacement. For that reason a safety factor is applied when the calculated surge voltage is compared with the transformer withstand strength.

15% or 20%?

In practice a 15% margin is often used as a minimum engineering value, while 20% gives a more conservative criterion. On this page the 20% margin (\(SF = 1.20\)) is used for transformer internal insulation.

Before probabilistic coordination, large margins were used between strength and stress, because the incoming surge and station response were not calculated in detail. With modern analysis — a probabilistically selected surge (from an MTBF or MTBS criterion) and an EMTP® study — very large historical margins are no longer appropriate. But margins should not all be removed; the required margin depends on the insulation type.

For non-self-restoring insulation a margin is clearly needed, because of ageing, degradation, repeated low-magnitude surges, test-method limits, manufacturing tolerances, uncertain internal response, possible power-frequency-voltage effects, and the permanent consequence of failure. Transformer margins have ranged from 10% to 30% (the larger values when the surge was not calculated directly); when the actual transformer surge is computed, typical margins fall to 15% to 20%. The IEC and IEEE guides suggest 15%; the text suggests 20% for general use:

\[ SF = 1.20 \qquad (\text{non-self-restoring insulation}) \]

Section 4

Effect of Power-Frequency Voltage on Transformer Insulation

Power-frequency voltage can influence oil-paper insulation strength. In referenced tests, an oil-paper sample had a negative-polarity lightning-impulse CFO of \(177\ \text{kV}\) with no power-frequency voltage; when a continuous power-frequency voltage was applied and an impulse superimposed at different points on the wave, the CFO changed — the opposite-polarity case being the important one. For \(138\ \text{kV}\) and \(500\ \text{kV}\) systems, the ratio of power-frequency line-to-ground voltage to BIL is about \(0.20\) to \(0.28\), which may reduce the CFO by roughly 16% to 18%. These observations need further verification, but they show power-frequency voltage can affect impulse strength — supporting the case for a margin on non-self-restoring insulation.

Which voltage is compared

The value used for insulation coordination should be the voltage to ground at the equipment terminal, including the power-frequency component where it applies. That component must not be accidentally ignored or double-counted.

Section 5

Safety Factor for Self-Restoring Insulation

Why self-restoring insulation needs less margin

Self-restoring insulation, such as air gaps and external insulation, can recover after flashover, and its withstand is usually described statistically through CFO. So the large safety factor used for transformer internal insulation is not normally applied here — though altitude correction and practical design margins may still be required.

Margins of 15% to 20% were historically used for self-restoring insulation too. But the text questions combining a probability-based incoming surge with an additional fixed voltage margin: if higher reliability is required, the consistent method is to increase the required MTBF, which produces a more severe design surge. So a large safety factor is generally not recommended for self-restoring insulation — the suggestion is no safety factor for general use, with a small value of about 5% only where a margin is deliberately required:

\[ SF = 1.05 \qquad (\text{self-restoring, only if a margin is deliberately wanted}) \]

Section 6

Hidden Margins

Margins can also be hidden. BIL is usually specified as the lower of the positive- and negative-polarity standard withstand — often the positive-polarity value. But the incoming surge in station lightning studies is usually negative polarity, whose actual strength may be higher than the specified BIL.

Why this matters

A hidden margin may already exist because the negative-polarity strength exceeds the specified BIL. This is one reason applying additional large margins to self-restoring insulation is questionable — the conservatism may already be built in.

Section 7

The Core Problem — Non-Standard Waveshapes

The surge voltages calculated in the station usually do not have the standard \(1.2/50\ \mu\text{s}\) waveshape — yet BIL is defined for that standard wave. So the central question is: how should a non-standard surge waveshape be compared with BIL?

The station waveshapes depend on the incoming surge, the arrester V–I characteristic, the layout, the inter-equipment distances, the arrester location, equipment capacitance and reflections. Behind the arrester the voltage may have an initial spike then decay to the arrester discharge voltage; ahead of the arrester it may be damped-oscillatory then decay — both unlike a standard impulse.

Section 8

Methods for Evaluating Non-Standard Waveshapes

Two principal methods are used: subjective evaluation, and mathematical methods based on the leader progression model (the direct leader-progression equations and the destructive-effect method derived from them). The leader-progression method is more exact but is mainly practical inside a computer program; for a simplified design method a simplified approximation is needed — which the following sections provide for self-restoring and then non-self-restoring insulation.

Section 9

Self-Restoring Strength — the \(1.15\,\text{BIL}\) Rule

For self-restoring insulation, a non-standard waveshape can be converted to an equivalent strength using line-insulation analysis. For a surge with an exponential tail time constant \(T\), the non-standard CFO is \(CFO_{NS}\): for \(T = 10\ \mu\text{s}\) and \(20\ \mu\text{s}\) it is about \(1.26\times CFO\) and \(1.12\times CFO\). A practical approximation is therefore:

\[ CFO_{NS} \approx 1.15 \times CFO \;\;\Rightarrow\;\; \text{strength} \approx 1.15 \times \text{BIL} \]

This corresponds approximately to the \(3\ \mu\text{s}\) chopped-wave level. It applies to self-restoring apparatus — disconnecting switches, bus supports, external breaker insulation and air clearances. For breakers it is built into the chopped-wave test; for disconnectors and bus supports the test may not be required, but similar external strength is reasonable to expect.

Why \(1.15\,\text{BIL}\) for external apparatus

For external apparatus exposed to a short, spike-like surge, the withstand can be higher than the standard full-wave BIL — the short-duration withstand is about \(1.15\,\text{BIL}\). So when the waveform is fast and spike-like, the required BIL may be estimated by dividing the calculated crest voltage by \(1.15\).

Section 10

Spike-Like versus Full-Wave — the Self-Restoring BIL Estimate

For short separation distances the surge may have no significant initial spike and look more like a full wave — then using \(1.15\,\text{BIL}\) would be unconservative. So if the ratio of crest voltage to arrester discharge voltage is below \(1.15\), the strength is set to BIL rather than \(1.15\,\text{BIL}\):

\[ \frac{E_b}{E_d} > 1.15 \;\Rightarrow\; \text{BIL} = \frac{E_b}{1.15\,\delta} \qquad\qquad \frac{E_b}{E_d} \le 1.15 \;\Rightarrow\; \text{BIL} = \frac{E_b}{\delta} \]

\(E_b\): crest voltage to ground at the equipment
\(E_d\): arrester discharge voltage
\(\delta\): relative air density

In words: a spike-like waveshape justifies \(1.15\,\text{BIL}\); a full-wave-like waveshape should use BIL.

When not to use \(1.15\,\text{BIL}\)

If the external-insulation voltage is not a short spike, the \(1.15\,\text{BIL}\) allowance should not be used. For a full-wave-like stress the calculated voltage should be compared directly with BIL, with altitude correction where it applies.

Section 11

Relative Air Density Correction

External (self-restoring) insulation strength falls with altitude as the air density decreases. For altitude \(A\) in km:

\[ \delta = e^{-0.121\,A} \qquad (A\text{ in km}) \]

Because \(\delta < 1\) at altitude, dividing \(E_b\) by \(\delta\) in the BIL equation increases the required BIL.

Altitude, in plain terms

At higher altitude the air density is lower, and lower air density reduces the flashover strength of external insulation. External insulation and air clearances must therefore be increased or corrected when the station is above sea level.

What altitude does and does not affect

Altitude correction applies to external air insulation — air clearances, external bushing insulation, bus-support insulators, disconnectors and external circuit-breaker insulation. It does not apply in the same way to transformer internal oil-paper insulation.

Section 12

The IEC Approach for Self-Restoring Insulation

The IEC application guide uses a simpler approach: it sets the strength for all waveshapes equal to BIL, with a 5% safety factor:

\[ \text{BIL} = \frac{1.05\,E_b}{\delta} \]

This is simpler and more conservative than using \(1.15\,\text{BIL}\) for chopped or spike-like waveshapes.

Section 13

Why Transformer Evaluation Is Different

The self-restoring approach does not apply directly to transformers. Transformer internal insulation is non-self-restoring — it does not recover — and its response to non-standard waveshapes is more complex. So a subjective evaluation is usually required: the calculated transformer voltage waveshape must be compared with the standard test voltages and their waveshapes.

Section 14

Transformer Test Voltages

The transformer can be compared with its full-wave lightning impulse (BIL), chopped-wave, switching impulse (BSL) and induced power-frequency test voltages. In the USA, transformer strength has historically been taken as the chopped-wave value \(1.10\,\text{BIL}\), because chopped waves stress turn insulation and represent fast-front conditions. But this is not always appropriate — the time to crest of the actual transformer surge must also be considered.

Section 15

The Chopped-Wave and Time-to-Crest Criterion

The chopped-wave criterion, in words

If the transformer voltage is a fast spike that crests within about \(3\ \mu\text{s}\), it may be compared with the transformer chopped-wave strength, which is approximately \(1.10\,\text{BIL}\). Because a safety factor of \(1.20\) is applied, the required BIL is the calculated transformer voltage multiplied by \(1.20\) and divided by \(1.10\).

When to use full-wave BIL instead

If the transformer voltage crests more slowly, or the oscillation is not rapidly attenuated, the waveform is closer to a full lightning impulse than a chopped wave. The calculated voltage should then be compared directly with BIL, and the \(1.10\,\text{BIL}\) chopped-wave allowance should not be used.

The chopped-wave test has a chopping time near \(3\ \mu\text{s}\). So if the transformer voltage reaches crest before \(3\ \mu\text{s}\), comparing the crest with the chopped-wave strength \(1.10\,\text{BIL}\) is reasonable; if it crests after \(3\ \mu\text{s}\), the chopped wave does not represent the stress, and the strength should be BIL. With \(t_T\) the time to crest of the transformer voltage:

\[ t_T \le 3\ \mu\text{s} \;\;\Rightarrow\;\; \text{BIL} = \frac{SF\cdot E_t}{1.10} \qquad\qquad t_T > 3\ \mu\text{s} \;\;\Rightarrow\;\; \text{BIL} = SF\cdot E_t \]

\(t_T\): time to crest of the transformer voltage
\(E_t\): crest voltage to ground at the transformer
\(SF\): safety factor (suggested \(1.20\))

Section 16

The IEEE Arrester-Guide Alternative

The IEEE arrester application guide uses the time to crest of the voltage at the arrester instead of at the transformer, which avoids computing \(t_T\) directly: if the arrester time to crest is \(\le 2\ \mu\text{s}\), the transformer strength is \(1.10\,\text{BIL}\); if greater, it is BIL. This is a practical simplification — though the transformer-terminal time-to-crest method relates more directly to the actual stress at the transformer.

Section 17

Transformer Coordination Cases (230 kV)

What the ratio \(E_t/E_d\) tells you

The ratio \(E_t/E_d\) shows how spike-like the transformer voltage is relative to the arrester discharge voltage. If the transformer crest is much higher than the arrester discharge voltage, the waveform behaves more like a short spike and the chopped-wave comparison is reasonable. If the ratio is not high enough, the waveform should be treated more conservatively as a full-wave stress.

Case 1 — the usual case

With \(E_t/E_d > 1.10\) and \(t_T < 3\ \mu\text{s}\), the transformer voltage is fast enough that the chopped-wave test is relevant, so \(1.10\,\text{BIL}\) is used. The crest surge is set equal to the chopped-wave test voltage, and the full-wave test voltage should still remain above the surge tail; if both hold, coordination is acceptable.

Case 2 — small separation distance

With \(E_t/E_d \le 1.10\) and \(t_T < 3\ \mu\text{s}\), the crest is not much above the arrester voltage and the waveshape is more full-wave-like. Setting the chopped-wave voltage equal to the crest could push the full-wave voltage below the surge tail, so coordination becomes marginal — the strength should be BIL rather than \(1.10\,\text{BIL}\): \(\text{BIL} = SF\cdot E_t\).

Case 3 — high crest but slow

With \(E_t/E_d > 1.10\) but \(t_T > 3\ \mu\text{s}\), the crest is high enough but the time to crest is too long, so the chopped wave is no longer representative and full-wave BIL is used: \(\text{BIL} = SF\cdot E_t\).

Section 18

Why Long-Tail Surges Need an Extra Check

The previous cases assumed a relatively short tail, typical of a backflashover surge. But a surge from a shielding failure without flashover can have a much longer tail, keeping the transformer voltage elevated for a long time. Even if the crest is coordinated with the chopped-wave test, the tail may not be coordinated with the full-wave or switching-impulse withstand — so long-tail surges require an additional check.

Why a long tail can govern the BIL

Not every incoming surge produces a short spike. A shielding failure without flashover can create a longer-duration surge, so the transformer insulation is stressed more like a switching impulse than a chopped lightning impulse. The switching-impulse strength is approximately \(0.83\,\text{BIL}\), so this long-tail check can decide the selected transformer BIL.

Section 19

Long-Tail Surges and the Switching-Impulse Level

The long tail is checked against the switching-impulse test voltage, approximately \(0.83\,\text{BIL}\) — suitable for long durations (front \(> 100\ \mu\text{s}\), time-to-zero \(> 1000\ \mu\text{s}\), time above 90% crest \(> 200\ \mu\text{s}\)). With \(E_t/E_d > 1.10\) and \(t_T < 3\ \mu\text{s}\), the crest may look acceptable against the chopped wave, but the tail may exceed the full-wave tail; if the switching-impulse voltage is greater than the arrester discharge voltage, coordination may be acceptable.

For a short separation distance (\(E_t/E_d \le 1.10\)) with a long-tail surge, the full-wave voltage may be below the tail and the switching-impulse voltage may be below the arrester voltage — then coordination is not acceptable, and the arrester voltage should be set equal to the switching-impulse test voltage, giving a more conservative BIL.

Section 20

When Is a Tail "Long"?

Let \(t_c\) be the time when the arrester voltage significantly decreases. The long-tail criterion is applied if:

\[ t_c > 60\ \mu\text{s} \]

This approximately separates the two surge origins: a backflashover surge has a shorter tail, while a shielding failure without flashover has a longer tail.

Section 21

The Suggested Transformer BIL Criterion

Two checks are applied, and the highest required BIL is selected. Check 1 is for a short-tail surge (\(t_c < 60\ \mu\text{s}\)); Check 2 is the long-tail criterion (\(t_c > 60\ \mu\text{s}\), or in practice shielding failure without flashover).

Table 2 — The transformer BIL checks (select the highest).

Check	Condition	Required BIL
1A (short tail)	\(t_T \le 3\ \mu\text{s}\) and \(E_t/E_d > 1.10\)	\(\dfrac{SF\cdot E_t}{1.10}\)
1B (short tail)	\(t_T \le 3\ \mu\text{s}\) and \(E_t/E_d \le 1.10\)	\(SF\cdot E_t\)
1C (short tail)	\(t_T > 3\ \mu\text{s}\)	\(SF\cdot E_t\)
2 (long tail)	\(t_c > 60\ \mu\text{s}\) / shielding failure	\(\dfrac{SF\cdot E_d}{0.83}\)

Here \(0.83\,\text{BIL}\) represents the switching-impulse test voltage, and the suggested safety factor for transformer insulation is \(SF = 1.20\). Because the long-tail check is conservative, it is often applied generally.

Always take the highest required BIL

Both the short-tail and long-tail criteria should be checked where relevant. The selected transformer BIL must be based on the highest required value, then rounded up to the next available standard BIL.

Section 22

The Transformer Insulation Strength Curve

A continuous strength curve represents transformer strength across time regions, connecting the standard and non-standard test points so strength can be viewed as a function of time:

Table 3 — Test points on the transformer strength curve.

Test	Level	Approx. Time	Standard?
Front-of-wave	\(1.3\)–\(1.5\,\text{BIL}\)	\(\approx 0.5\ \mu\text{s}\)	non-standard (by agreement)
Chopped-wave	\(1.10\,\text{BIL}\)	\(3\ \mu\text{s}\)	standard
Full-wave	\(\text{BIL}\)	\(\approx 8\ \mu\text{s}\)	standard
Switching impulse	\(\text{BSL} = 0.83\,\text{BIL}\)	\(\approx 300\ \mu\text{s}\)	standard
One-hour induced	\(1.5\,V_{LG,\max}\)	\(\approx 10^{3.4}\ \text{s}\)	power-frequency region

Front-of-wave and chopped-wave mainly stress turn insulation; full-wave BIL stresses both turn and ground-wall insulation; switching-impulse and power-frequency mainly stress ground-wall insulation. The region between BIL and BSL is less clearly defined. The curve aids understanding but does not change the practical criteria — chopped-wave when fast enough, full-wave BIL when slower or full-wave-like, the long-tail / switching-impulse check where appropriate, with a suitable safety factor.

Section 23

The Transformer Bushing — Two Insulations in One

A transformer bushing contains both internal (non-self-restoring) and external (self-restoring) insulation, so it must be treated two ways. The internal insulation is treated like transformer insulation — chopped-wave strength \(1.10\,\text{BIL}\) when installed in the transformer. The external insulation is treated like other external apparatus — chopped-wave level \(1.15\,\text{BIL}\).

Because they are evaluated differently, two BIL requirements may result. If the external requirement is lower than the internal, the external BIL is set equal to the internal; if higher, both may be accepted — which can happen at high altitude, where external strength falls with air density while internal insulation is largely unaffected.

Check a bushing twice

A transformer bushing should be checked twice. The internal insulation is non-self-restoring and is coordinated like transformer insulation. The external porcelain or composite insulation is self-restoring and is coordinated like external air insulation, including altitude correction. The selected external bushing BIL should normally be equal to or greater than the internal bushing BIL.

A testing issue

If the external BIL exceeds the internal BIL, the external insulation cannot be tested at that higher level while installed in the transformer (the internal insulation might not withstand it). Tests on the separate bushing shell may then need to be accepted — a practical procurement and testing consideration.

Section 24

Phase-Ground Air Clearances

Air clearance, in words

Air clearance is selected by converting the maximum expected surge voltage into a required physical distance. A conservative withstand gradient is used, so the required clearance grows as the expected surge voltage grows.

The lightning-impulse strength of air gaps depends on geometry, polarity, electrode shape and atmosphere. The positive-polarity CFO gradient is about \(540\) to \(650\ \text{kV/m}\); the negative-polarity strength about \(540\) to \(750\ \text{kV/m}\) (the higher value for a rod-plane gap). Because negative-polarity surges dominate station lightning studies, a suggested practical value is \(605\ \text{kV/m}\). The \(3\ \mu\text{s}\) strength is about \(1.38\times CFO\), giving \(1.38 \times 605 = 835\ \text{kV/m}\) — but for conservatism the text uses \(605\ \text{kV/m}\) so as not to overestimate the gap strength. The clearance is then:

\[ d = \frac{V}{G} = \frac{V}{605} \qquad\text{and}\qquad d_{\text{alt}} = \frac{d_{\text{sea level}}}{\delta} \]

\(d\): clearance (m)
\(V\): relevant surge voltage (kV)
\(G\): CFO gradient (kV/m), suggested 605 (conservative)

Section 25

A Warning About Clearance From Bus-Support BIL

Some standards link air clearance to the bus-support BIL by dividing it by a minimum positive-polarity gradient (e.g. \(500\ \text{kV/m}\)). The text does not recommend this, because it introduces hidden margins: it uses the actual station BIL rather than the required minimum, and a positive-polarity gradient even though negative-polarity surges usually dominate. The method is conservative but not technically preferred for the simplified coordination method.

Section 26

Phase-Phase Clearances

When a line flashes over to one phase, a coupled voltage of the same polarity appears on the other phases. At the struck point the phase-phase voltage is normally less than the phase-ground voltage; propagation toward the station can increase it, but because the struck point is usually relatively close to the station, the phase-phase voltage seldom exceeds the phase-ground voltage.

When phase-phase must be checked separately

For the lightning-surge cases here, the phase-ground clearance generally provides a satisfactory basis for phase-phase clearance. For higher-voltage systems, or switching-surge-dominated cases, the phase-phase switching-impulse requirement should also be checked.

A practical simplification

The phase-ground clearance can generally also be considered adequate for phase-phase clearance — so a separate phase-phase clearance calculation is usually not required.

Section 27

Practical Design Logic

The strength-side logic, in order

calculate the surge voltage at the equipment;
external self-restoring — compare with a BIL-based strength, with atmospheric correction and usually no large safety factor;
transformer internal — compare the waveshape with the test withstands: \(1.10\,\text{BIL}\) only when fast and spike-like, otherwise BIL;
check long-tail surges against the switching-impulse level \(0.83\,\text{BIL}\);
apply \(SF = 1.20\) for transformer internal insulation;
treat bushings as two insulations; use a conservative \(605\ \text{kV/m}\) gradient for clearances, with altitude correction.

Section 28

Summary of Key Equations

Equation Summary

Relative air density

\( \delta = e^{-0.121A} \)

Non-standard CFO

\( CFO_{NS} \approx 1.15\,CFO \)

SR, spike-like

\( \text{BIL} = \dfrac{E_b}{1.15\,\delta} \)

SR, full-wave-like

\( \text{BIL} = \dfrac{E_b}{\delta} \)

IEC self-restoring

\( \text{BIL} = \dfrac{1.05\,E_b}{\delta} \)

Transformer, fast spike

\( \text{BIL} = \dfrac{SF\cdot E_t}{1.10} \)

Transformer, slow / full-wave

\( \text{BIL} = SF\cdot E_t \)

Transformer, long tail

\( \text{BIL} = \dfrac{SF\cdot E_d}{0.83} \)

Air clearance

\( d = \dfrac{V}{605} \)

Altitude-corrected clearance

\( d_{\text{alt}} = \dfrac{d_{\text{sea level}}}{\delta} \)

Section 29

Reader Should Remember

This part converts calculated station surge voltages into a required insulation strength. The key distinction is non-self-restoring versus self-restoring insulation. Transformer internal insulation needs a safety factor because failure is permanent — a practical \(SF = 1.20\). For self-restoring external insulation, large fixed factors are questionable when the surge is already chosen probabilistically; any margin should be small.

The station waveshape is usually not the standard \(1.2/50\ \mu\text{s}\) wave, so strength must be read by waveshape: a spike-like wave may justify \(1.15\,\text{BIL}\) for self-restoring insulation; the transformer chopped-wave level \(1.10\,\text{BIL}\) applies only when the transformer voltage crests within about \(3\ \mu\text{s}\) and is spike-like, otherwise BIL; and long-tail surges bring in the switching-impulse level \(0.83\,\text{BIL}\). Bushings are special — internal like a transformer, external like self-restoring insulation — and air clearances use a conservative \(605\ \text{kV/m}\) with altitude correction.

The single most important message

BIL selection is not only a crest-voltage comparison. The waveshape, time to crest, surge tail, insulation type, safety factor, altitude and equipment test basis must all be considered together.

Reader should remember

Different insulation types must not be coordinated the same way. Transformer internal insulation is non-self-restoring and needs a safety factor. External air insulation is self-restoring and is strongly affected by altitude. Bushings contain both internal and external insulation and must be checked for both. The correct BIL criterion depends not only on the voltage crest, but also on the surge waveshape and time to crest.

This is Part Three of the station lightning insulation coordination series, covering insulation strength, safety factors, transformer BIL selection, bushings and air clearances. Later parts continue with the remaining coordination steps and worked station examples.

Section 30

Key Symbols

Table 4 — Key symbols used on this page.

Symbol	Meaning
BIL / BSL	Basic Lightning Impulse Level / Basic Switching Impulse Level (\(\text{BSL} \approx 0.83\,\text{BIL}\))
CFO / \(CFO_{NS}\)	Critical Flashover Voltage; CFO for a non-standard waveshape
SF	Safety factor (\(1.20\) non-self-restoring; up to \(1.05\) self-restoring)
\(E_t,\ E_d,\ E_b\)	Crest to ground at transformer; arrester discharge voltage; crest to ground at external equipment
\(t_T\)	Time to crest of the transformer voltage
\(t_c\)	Time for the arrester voltage to significantly decrease (long-tail test)
\(\delta,\ A\)	Relative air density; altitude in km (\(\delta = e^{-0.121A}\))
\(G\)	Air-gap CFO gradient (kV/m), suggested \(605\)
\(d\)	Air clearance distance (m), \(d = V/605\) with altitude correction

Part Three Reading now

Insulation Strength & BIL Selection

Turning surge stress into strength — safety factors, the 1.15 BIL rule, the transformer chopped-wave / full-wave / long-tail criteria, bushings and clearances.

Series progress 3 of 8

Part Four

Standard BILs & Worked Examples

Applying the simplified method — standard BIL selection and worked 230 kV single-line, two-line, contingency and non-symmetrical 115 kV examples.