Creepage distance is a surface measurement, not an air gap
Creepage distance, printed as “leakage distance” on a lot of datasheets, is the shortest path between the two electrodes that runs along the surface of the housing. You measure it with a tape laid against the sheds, not a ruler through the air. The difference shows up fast. Two 36 kV post insulators can both carry 700 mm of clearance and 1,000 mm of surface path, and one gets there with deep sheds while the other uses a slimmer profile with more of them. Both are buildable.
What the longer surface path buys is a longer wetted film for the leakage current to run along. Clean and dry, 800 mm of surface and 800 mm of air behave much the same. With salt and dust holding a film on the sheds, that current heats the film, dries a band out somewhere in the middle, and forces the discharge back into the air gap. Once the band appears, flashover follows the shortest dry route and the rest of the string stops having a say.
Ordering happens in mm per kV of system voltage, or mm/kV, although plenty of utilities still buy on total leakage length. Ask which one the quotation answers: a supplier quoting 400 mm of total on a 36 kV line reads very differently from one quoting 11 mm/kV on the same part.
Where the pollution class numbers come from
Pollution severity is not something you judge from a photograph. IEC 60071-1 covers how insulation coordination treats polluted sites, IEC 60507 gives the artificial pollution test method and the field procedure for turning a real deposit into an equivalent salt deposit density figure, and IEEE 1128 covers the same question for polymeric housings in North American practice. Editions move, so check the bands against the current text and your utility’s own pollution map rather than a PDF someone forwarded in 2021.
The field procedure, in the order IEC 60507 lays out (verify the current edition before you write it into a spec):
- Take the windward face, and the top, middle and bottom positions of the string. Salt lands on one side of a tower far more than the other.
- Sample at least five units per position. Five tells you the class. One unit is a coincidence.
- Rinse each housing with distilled water, collect the run-off separately, and measure the conductivity of every sample.
- Convert conductivity to salt mass with the calibration curve in the standard, then divide by the washed area.
- Report the median of the five. An average gets dragged down by whichever unit a passing shower just cleaned.
The step people skip is the repeat visit. The class a utility wrote down three years ago, on a stretch of coast that has since grown a port terminal, is a guess wearing a hard hat.
The mm/kV bands you will see in tender documents
| Site condition | Typical class | Creepage band | Why it lands there |
|---|---|---|---|
| Inland rural, occasional rain, no heavy industry | Light | 14 to 20 mm/kV | Rain washes the sheds, so deposits stay thin |
| Towns, cement works, road dust, light rain | Light to medium | 20 to 25 mm/kV | Dust builds up between rain events |
| Coast within roughly half a kilometre, estuary, monsoon plus salt spray | Medium to heavy | 25 to 31 mm/kV | Wind carries salt inland, rain keeps clearing it |
| Salt works, chemical plant, offshore platform, little rain | Heavy | 31 to 43 mm/kV | Deposits accumulate and never dissolve |
Treat those as a starting point. The bands differ between editions and between utilities, and the class in the original design note is often higher than the site really is, which is why half the strings on an old line still perform.
The band is a minimum, and some utilities buy above it on purpose: 28 mm/kV to 40 mm/kV on a polymer housing costs a few centimetres of moulded length, far less than a washing programme. The counter-case matters too. Where rain never falls, extra creepage helps less than the table suggests, because nothing flushes the sheds. Heavy desert sites often end up specified with more leakage distance and a washing schedule at the same time.
Doing the arithmetic on a 66 kV line
Worked example, so the table turns into an order. A 66 kV line along a coastal corridor, the utility’s pollution map says medium, their table reads 25 to 31 mm/kV.
- 66 x 25 = 1,650 mm at the bottom of the band, 66 x 31 = 2,050 mm at the top
- Housing with 500 mm of leakage distance per unit: 4 units give 2,000 mm, about 30 mm/kV, inside the band with a little margin
- Housing with 400 mm per unit: 5 units give 2,000 mm, same total, same mm/kV
- 6 units of 400 mm give 2,400 mm, or 36 mm/kV, above what the band asks for
Two ways to land on 2,000 mm, and they are not the same part. The 400 mm unit carries smaller sheds, more of them, and a better chance of self-cleaning under light rain. The 500 mm unit is stiffer and easier to clean by hand. Which one suits depends entirely on whether that corridor ever sees rain. Nobody asks that question, so the factory decides.
What the quotation needs to state is “4 units x 500 mm = 2,000 mm total, 30 mm/kV at 66 kV”. The reviewer then checks one line. A quote that arrives as “36 kV unit, 8 sheds” makes you redo all of the above.
Reading the number wrong, and paying for it
Four things I keep finding in RFQs:
- Total length quoted where creepage per kV was asked for. A 435 mm housing on a 33 kV line is 13 mm/kV, below the light-pollution floor. The same housing on a 132 kV line is 3 mm/kV, and it would not hold a wet test in a clean laboratory. Part number correct, assignment wrong.
- Wet flashover reported without the creepage that produced it. 90 kV of wet flashover on a 400 mm unit and 120 kV on a 550 mm unit are both solid. Printed side by side with no leakage distance, the first one looks like the better product.
- Dry arc distance confused with leakage distance. The standards measure the straight air gap for lightning and switching impulse performance, and the surface path for pollution performance. A spec calling for 1,000 mm of “distance” without naming which one has been checked by nobody.
- Assuming a heavy-pollution porcelain profile drops onto a polymer replacement. The electrical length may match. The coupling and the gland usually do not, and the string stops being interchangeable with the spares in the store.
What to write into the order, and which standard backs it
Which standard covers which part. Check scope and edition before citing them.
| Product | Standard usually cited |
|---|---|
| Composite line insulators, above 1 kV | IEC 61109 |
| Composite post insulators, above 1 kV | IEC 61952 |
| Composite line post insulators | IEC 62231 |
| General construction and tests for polymer insulators | IEC 62217 |
| Insulation coordination, pollution severity classes | IEC 60071-1, IEC 60071-2 |
| Artificial pollution testing, ESDD measurement | IEC 60507 |
| US practice, polymeric line and post insulators | ANSI C29.11, IEEE 1128 |
Then the purchase order itself. Short list, everything on it checkable before the shipper loads:
- Leakage distance per unit, and the total for the string
- The creepage value in mm/kV, with the voltage it was divided by
- The pollution class, or the ESDD figure, plus the date it was measured
- The standard and edition for type tests and routine sample tests
- The sampling plan in words, for example ISO 2859-1 general inspection level II with the AQL stated
- Witness items separated from desk-review items. Wet power frequency flashover, thermal mechanical cycling and the interface test belong on the first list. Visual and dimensional checks belong on the second.
- The coupling drawing number for the existing pole top or tower hardware
Sizing against your numbers instead of a catalogue page
The creepage requirement is one line on a datasheet, and it is the line that decides whether a string survives its first winter. Get the class from a measurement, convert it to millimetres with the system voltage, and put both on the order. After that the factory can argue about shed profile, and you will not be arguing about whether the housing was long enough.
Send over the system voltage, the pollution class or the ESDD figure you measured, and the drawing for the pole top or tower fittings, and the quotation should come back naming the housing length, shed profile and unit count against those inputs. Vichit sizes polymer housings against the class you give us rather than the one we hold on the shelf, which is a slower way to quote.
