Insulation Coordination
A metal oxide arrester’s protective characteristic is the voltage across it while it discharges a specified surge current — and that is the voltage the protected equipment may experience. Part Two explains the three standardised levels (the 8/20 μs discharge voltage, the 0.5 μs fast-front level and the switching impulse level), why a faster surge front gives a higher arrester voltage, the conservative 0.5 μs scaling method, and the IEEE and CIGRE dynamic arrester models.
Section 1
The protective characteristics of a metal oxide arrester describe the voltage that appears across the arrester while it discharges a specified surge current.
Protective characteristic = arrester voltage at a specified discharge current.
This matters because that voltage is also what the protected equipment may experience — depending on arrester location, lead length, reflections and station layout. The lower the arrester protective voltage, the better the insulation protection. But the arrester must still have enough MCOV, TOV and energy capability to survive the system duty, so selection always requires both survival capability and protective performance.
This is Part Two of the metal oxide surge arrester series. Part One covered the characteristics, ratings and durability tests; this part covers how the arrester protective voltage is specified and modelled.
Section 2
The arrester rating tells whether the arrester can remain connected and survive. The protective level tells how much voltage appears across the arrester during a surge. These are different concepts, and a correct application satisfies all of them.
| Concept | Main Question | Typical Data |
|---|---|---|
| Arrester rating | Can the arrester survive the system duty? | MCOV, TOV capability, energy capability |
| Protective characteristic | What voltage appears during discharge? | 8/20 μs voltage, 0.5 μs voltage, SI protective level |
| Insulation coordination | Is the protected equipment safe? | Arrester voltage plus lead / reflection effects versus equipment withstand |
An arrester rating proves survival. A protective characteristic defines protection. Both must be checked.
Section 3
Standards describe three main protective characteristics. They are not arbitrary values — each is a standardised way to describe how the arrester behaves for a different surge-current waveform:
| Characteristic | Current Waveform | Main Use |
|---|---|---|
| 8/20 μs discharge voltage | Lightning impulse current | Standard lightning protection level |
| 0.5 μs discharge voltage | Very fast-front condition | Steep incoming surge / front-of-wave assessment |
| Switching impulse protective level (SI) | \(45\)–\(60\ \mu\text{s}\) front current | Switching-surge insulation coordination |
Section 4
The 8/20 μs discharge voltage is one of the most important lightning protective characteristics — the voltage across the arrester when it discharges a current impulse with an \(8/20\ \mu\text{s}\) waveshape (about \(8\ \mu\text{s}\) front time, about \(20\ \mu\text{s}\) time to half value). The manufacturer tabulates the arrester voltage for several discharge-current crests:
| Discharge-Current Crest | Requirement |
|---|---|
| \(1.5\ \text{kA}\) | Tabulated |
| \(3\ \text{kA}\) | Tabulated |
| \(5\ \text{kA}\) | Tabulated |
| \(10\ \text{kA}\) | Tabulated |
| \(20\ \text{kA}\) | Tabulated |
| \(15\ \text{kA}\) | Required for arresters applicable to 500 kV systems |
| \(40\ \text{kA}\) | Often provided by manufacturers |
When a lightning surge reaches the arrester, the arrester conducts and clamps the voltage — but the equipment does not see zero volts. It sees a voltage related to the arrester residual voltage plus other effects.
Equipment stress is linked to the arrester discharge voltage. A high \(10\ \text{kA}\) discharge voltage means a higher surge on the transformer or breaker; a low discharge voltage improves the insulation margin.
Section 5
Older standards call this the front-of-wave (FOW) protective level — a carry-over from gapped silicon carbide arresters, where sparkover behaviour dominated the response. For gapless metal oxide arresters there is no sparkover gap, so the term is less physically clear. The precise term is the 0.5 μs discharge voltage.
For MO arresters, prefer “0.5 μs discharge voltage” over “front-of-wave” — it tells the engineer exactly what the quantity means.
It is defined as the voltage across the arrester when the voltage time to crest is \(0.5\ \mu\text{s}\), while the arrester discharges the lightning impulse classifying current (which depends on class and system voltage). It is not read from a normal 8/20 μs test — it is obtained by discharging current impulses with times to crest of \(1\), \(2\) and \(3\ \mu\text{s}\), plotting the arrester voltage against the voltage time to crest, and reading the value at \(0.5\ \mu\text{s}\).
A faster surge front produces a higher arrester discharge voltage — steep incoming surges can give higher arrester voltages than the standard 8/20 μs condition.
Section 6
The switching impulse protective level (\(\text{SI}\)) is the voltage across the arrester when it discharges a switching impulse current with a \(45\)–\(60\ \mu\text{s}\) time to crest, at a magnitude equal to the switching impulse classifying current. It is not required for distribution arresters. The classifying current depends on class and maximum system voltage:
| Arrester Class | Maximum System Voltage | SI Classifying Current |
|---|---|---|
| Station | \(3\)–\(150\ \text{kV}\) | \(500\ \text{A}\) |
| Station | \(151\)–\(325\ \text{kV}\) | \(1000\ \text{A}\) |
| Station | \(326\)–\(900\ \text{kV}\) | \(2000\ \text{A}\) |
| Intermediate | \(3\)–\(150\ \text{kV}\) | \(500\ \text{A}\) |
The switching impulse protective level is especially important for EHV systems, where switching surges may control the insulation coordination.
Section 7
Each protective characteristic represents a different stress condition. The engineer must use the one that matches the overvoltage being studied:
| Characteristic | Current Waveform | Main Application |
|---|---|---|
| 8/20 μs discharge voltage | Lightning impulse current | Lightning protection |
| 0.5 μs discharge voltage | Very fast-front condition | Steep incoming surge / front-of-wave assessment |
| SI protective level | \(45\)–\(60\ \mu\text{s}\) front switching impulse | Switching surge insulation coordination |
Section 8
Protective characteristics are often expressed in per unit of crest MCOV, which normalises them across different arrester voltage ratings. The crest of MCOV is:
so a protective voltage in per unit of crest MCOV is:
Section 9
Typical ranges, in per unit of crest MCOV, span a wide band — which is exactly why real manufacturer data must be used for practical selection:
| Arrester Class | FOW / 0.5 μs | 8/20 μs | SI |
|---|---|---|---|
| Station | \(2.01\)–\(2.48\) | \(1.97\)–\(2.25\) | \(1.64\)–\(1.85\) |
| Intermediate | \(2.38\)–\(2.85\) | \(2.28\)–\(2.55\) | \(1.71\)–\(1.85\) |
| Distribution — heavy duty | \(2.40\)–\(3.75\) | \(2.00\)–\(3.46\) | Not required |
| Distribution — normal duty | \(2.90\)–\(3.53\) | \(2.77\)–\(3.32\) | Not required |
| Riser pole | \(2.07\)–\(3.32\) | \(2.65\)–\(3.32\) | Not required |
Two arresters with the same MCOV may have different discharge voltages — which changes the insulation margin. Do not assume all arresters with the same MCOV have the same protective level. Use manufacturer data: 8/20 μs and 0.5 μs discharge voltages, SI protective level, the V–I curve, energy capability and TOV capability.
Section 10
The arrester discharge voltage depends not only on current magnitude but also on current steepness. A shorter time to crest generally produces a higher discharge voltage:
This happens because the arrester and its current path have dynamic effects — the arrester does not respond only as a static nonlinear resistor. The voltage crest also occurs slightly differently from the current crest:
| Current Front Time | Voltage Time to Crest |
|---|---|
| \(8\ \mu\text{s}\) current front | about \(7\ \mu\text{s}\) |
| \(1\ \mu\text{s}\) current front | about \(0.5\ \mu\text{s}\) |
Practical incoming-surge steepness may be in the range \(500\)–\(2000\ \text{kV/}\mu\text{s}\) — much faster than the standard 8/20 μs arrester test. For station insulation coordination it may therefore be necessary to use a protective characteristic constructed for the incoming-surge time to crest, or a dynamic arrester model that includes the effect of time to crest — especially for steep surges from nearby backflashovers.
Section 11
If detailed dynamic data are unavailable, a conservative fast-front model can be built by scaling the 8/20 μs data. For a \(318\ \text{kV}\) MCOV arrester, the source gives:
| Quantity | Value |
|---|---|
| 0.5 μs discharge voltage | \(1070\ \text{kV}\) |
| 15 kA 8/20 μs discharge voltage | \(915\ \text{kV}\) |
Multiplying all 8/20 μs discharge voltages by \(1.17\) gives a conservative 0.5 μs characteristic — the fast-front voltage is about \(17\%\) higher than the standard value for that specific arrester and current condition. It is not a universal constant.
The fast-front discharge voltage is higher than the standard 8/20 μs discharge voltage — but simple fast-front scaling, while conservative, may overestimate the equipment voltage and demand more insulation strength than necessary.
Section 12
During a current impulse, the arrester voltage–current path forms a dynamic loop because current and voltage do not peak at the same instant — the current time to crest precedes the voltage time to crest. The arrester is therefore not just a static V–I curve during fast transients; its apparent behaviour depends on current magnitude, current front time, arrester inductance, current-path inductance, nonlinear block behaviour and dynamic turn-on.
A proper arrester model must reproduce two things: the voltage–current trajectory during a surge, and the increase in crest voltage as the current time to crest decreases.
The model must know that faster surges produce higher arrester voltages. Both the IEEE and CIGRE models were developed to reproduce this dynamic behaviour.
Section 13
The IEEE model uses two nonlinear elements, \(A_0\) and \(A_1\), separated by a resistance–inductance network, with additional inductance, resistance and capacitance elements. The two nonlinear characteristics are estimated from standardised data in per unit of the \(10\ \text{kA}\) 8/20 μs discharge voltage. Its parameters are:
| Parameter | Meaning / Basis |
|---|---|
| \(A_0,\ A_1\) | Nonlinear elements from tabulated V–I data |
| \(L_1,\ R_1\) | Depend on arrester height and number of parallel columns |
| \(L_0,\ R_0,\ C\) | Calculated from the model equations |
| \(d,\ n\) | Arrester height; number of parallel columns |
The model is built and calibrated by a practical procedure, valid for current times to crest from \(0.5\ \mu\text{s}\) to \(45\ \mu\text{s}\):
Section 14
The CIGRE model uses a different structure: a single nonlinear element \(R_i\) (developed from the 8/20 μs characteristic), a turn-on resistance \(R_T\) representing the dynamic conduction behaviour, and an inductance \(L\) (or an equivalent travelling-wave representation) representing the current path. Typical inductance values are:
| Arrester Type | Inductance | Equivalent Surge Impedance | Travel Time |
|---|---|---|---|
| Outdoor | \(1\ \mu\text{H/m}\) | \(Z = 300\ \Omega\) | \(3.33\ \text{ns/m}\) |
| GIS | \(0.33\ \mu\text{H/m}\) | \(Z = 100\ \Omega\) | \(3.33\ \text{ns/m}\) |
The lower inductance for GIS arresters reflects their more compact geometry.
Section 15
| Item | IEEE Model | CIGRE Model |
|---|---|---|
| Nonlinear elements | Two (\(A_0,\ A_1\)) | One (\(R_i\)) |
| Dynamic representation | R–L network between the nonlinear elements | Turn-on resistance + inductance / travelling path |
| Calibration | Adjust \(A_0,\ A_1,\ L_1\) | 8/20 μs data + turn-on resistance |
| Purpose | Reproduce response, \(0.5\)–\(45\ \mu\text{s}\) fronts | Reproduce the V–I dynamic response |
| Complexity | Relatively complex | Also requires parameter estimation |
Both models are more accurate than a simple static V–I curve for fast-front surge studies.
Section 16
Detailed IEEE or CIGRE models are not always necessary. They are most useful where the result is sensitive to fast-front voltage:
| Detailed model most useful when… | Simplified model acceptable when… |
|---|---|
| Insulation margins are small; incoming surge fronts are very steep; open-breaker stress is assessed; GIS is studied; arrester lead effects matter; accurate tail voltage is required | Detailed dynamic data are unavailable; the study is not very sensitive; a conservative insulation margin is acceptable; early design screening is being performed |
A practical simplified method is to use the 8/20 μs characteristic adjusted to the 0.5 μs fast-front value, which usually gives conservative results. Its limitation is that it can overestimate the tail of the voltage waveform at equipment locations:
Simple fast-front scaling is conservative, but not always economically optimal — for critical projects, use a dynamic arrester model.
Section 17
| Situation | Recommended Modelling Approach |
|---|---|
| Early design screening | Manufacturer 8/20 μs data + conservative 0.5 μs scaling |
| Standard AIS study, good margin | Simplified arrester model may be acceptable |
| Very steep incoming surge | Fast-front characteristic or dynamic model |
| GIS / compact substation | Detailed model if margins are tight |
| Open-breaker protection study | Dynamic model if the result is sensitive |
| Final coordination, critical equipment | Manufacturer data + validated model |
Section 18
| Misconception | Correct Interpretation |
|---|---|
| “The 8/20 μs voltage is enough for every lightning study” | Very steep surges may need the 0.5 μs voltage or a dynamic model |
| “‘Front-of-wave’ is the best name” | For MO arresters, “0.5 μs discharge voltage” is more precise |
| “Arrester voltage depends only on current magnitude” | It depends on magnitude and current time to crest |
| “A single static V–I curve is always sufficient” | Fast-front studies may need IEEE or CIGRE dynamic models |
| “Conservative modelling is always better” | It can overestimate stress and force unnecessary insulation |
Memory map. Protective voltage = arrester voltage at a discharge current → use the 8/20 μs level for lightning, the 0.5 μs level for steep fronts, the SI level for switching → faster fronts give higher voltages → normalise in per unit of \(\sqrt{2}\,\text{MCOV}\) → scale 8/20 by \(\approx 1.17\) for a quick fast-front estimate → use an IEEE or CIGRE dynamic model when the study is sensitive.
The protective characteristics define how much voltage appears across the arrester during discharge: the 8/20 μs level for lightning, the 0.5 μs level (or a dynamic model) for very steep surges, and the SI level for switching. The key lesson is that arrester protective voltage depends on both discharge-current magnitude and front time — so for accurate insulation coordination the arrester must not be represented by a single static voltage. Use manufacturer protective data, and for sensitive studies consider the IEEE or CIGRE dynamic arrester models.
Five-Part Technical Series
A five-part self-study of metal oxide surge arresters — characteristics and ratings, protective characteristics and models, rating determination, the lightning discharge current, and distribution-system and IEC application.
The 8/20 and 0.5 microsecond discharge voltages, the switching impulse protective level, per-unit crest MCOV, time-to-crest effects, conservative fast-front scaling, and the IEEE and CIGRE dynamic arrester models.
