Someone told me that high-voltage transformers need oil because it prevents arcing. However, in terms of dielectric breakdown, isn't air the strongest?

I remember from classes in college that as the dielectric constant increases, the breakdown voltage decreases. Is this correct?

  • 2
    \$\begingroup\$ Air is quite low actually extremeelectronics.co.uk/tables/… \$\endgroup\$
    – JonRB
    Dec 23 '18 at 0:33
  • 8
    \$\begingroup\$ Dry air is a good insulator (low conductivity) but does not have a high dielectric breakdown voltage. You may be mixing up those two concepts because they are related. A good insulator has high resistance to all voltages unless the dielectric withstanding voltage (or field strength) is exceeded. \$\endgroup\$
    – mkeith
    Dec 23 '18 at 1:30
  • \$\begingroup\$ @JonRB that page looks awful; I wouldn't trust it much, especially given the non-existent footnote references... \$\endgroup\$
    – Nick T
    Dec 23 '18 at 18:05
  • 1
    \$\begingroup\$ @NickT I agree, but it was the quickest reference showing there are other insulating materials with higher dielectric strength. Relying on the figures? no rough idea? yes since it show oil >> air. allaboutcircuits.com/textbook/reference/chpt-3/… \$\endgroup\$
    – JonRB
    Dec 23 '18 at 18:27
  • \$\begingroup\$ Just for your information, there are also SF6 isolated transformers SF6 is a gas called sulphur hexafluride. (GIT...gas insulated transformer), these transformers are non-flammable and non-explosive, but SF6 has a big impact as a greenhouse gas when it is released. There are several transformer stations whole isolated in SF6 technology, it is very compact and saves a lot of space in comparison to the classical construction method. \$\endgroup\$ Dec 26 '18 at 22:25

3 reasons: Much higher voltage breakdown, thermal conductivity and much lower contaminants including moisture from condensation that lead to Partial Discharge, which is cheaper to monitor and repair in oil than dry-epoxy-types.

  • I added the 3rd reason which is more complex as easier to remove foreign particles in oil and the viscosity reduces the kinetic energy of accelerating particles in an E-field hitting a conductor with enough energy to release hydrogen, a combustible gas from the water molecule.

  • Dry transformers do exist < 5MVA occupy a smaller footprint, quieter, safer, preferred for some urban areas, but less efficient, cost more and rely on more expensive insulation with Mica tape, and epoxy polymers to make moisture resistant. Dry transformers must combat the tendency to absorb molecules of moisture, which rapidly deteriorates breakdown voltage.

Transformer grade Oil is at least 8x and up to 25x better than air for dielectric breakdown and at least 6x better thermal conductivity in[W/mK].

Oil is predominantly used > 5MVA due to better electrical and cooling efficiency. Oil is necessary for cooling, thermal spreading of hotspots and for electrical insulation.

Partial Discharge (PD) is all about the flow of ions in plasma, like an aurora or corona. It needs some contaminants to collide and cause discharge.

From my experiments on Nydas Transformer oil in a transformer factory to exceed 25kV/mm. With typical results varying 25 to 40kV.

With more expensive processing to remove ppm level contaminants, it can reach 70kV/mm. Those that can afford the $50k+ machine, use them but some skill in invisible contamination processing and process quality controls in a clean room environment are necessary.

The test is done with about 1kV/s ramp with ultraclean large (~2cm) brass flat electrodes in a clean glass beaker with tapered smooth edges.

Like Air, it is mobile contaminants and pressure changes that can lead to partial discharge that causes variability in the breakdown voltage BDV of an insulator.

For transformer oil, the Partial Discharge also breaks down the large hydrocarbon chain into H2 which has a lower explosive threshold of 4% concentration.

Clean air breakdown is 3kV/mm while dirty moist air is < 500V/mm in flat to flat, while point to point is about 1/3 of these Voltage thresholds.

An ultra-low vacuum gives a high BDV but a partial vacuum very low as reduction of molecules allows for less drag and higher kinetic energy when an ion in the air hits the conductor. ( See Paschen law. )


Transformer oil not only prevents arcing but it also prevents the transformer from over heating at its operating temperature.

So isn’t air the best against voltage breakdown?

Well the answer to that is no. According to my experiments It shows me that yes air does have a voltage breakdown in the gap of 1 inch at a voltage of 20,000 volts. But transformer oil has a voltage breakdown of 70,000 volts per inch.

If you put it this way, as the distance from the two conductors gets larger and larger the dielectric breakdown voltage needed to arc across that distance gets larger.

Then yes you are correct.

  • 11
    \$\begingroup\$ In addition to cooling and isolation, the oil also allows the detetion of isolationm faults (it also mitigates the effects of an Isolation fault to some extent). When an isolation fault occurs, gas will appear in the transformer oil which can be detected by a Buchholz relay (en.wikipedia.org/wiki/Buchholz_relay). \$\endgroup\$
    – Klaws
    Dec 23 '18 at 8:27
  • 5
    \$\begingroup\$ Air has a dielectric strength of 3MV/m. If we want to compare that to hands, thumbs, or other ancient units of measurement that's fine, but we should at least include the scientific units. In the industry, standard BDV testers use a 2.5mm gap so you'll often hear technologist level people in said industry refer to transformer oil breakdown voltage without a distance qualifier - transformer oil is generally held to a 30kV BDV minimum (by this test @2.5mm) before needing replacement, giving a minimum dielectric strength of 12MV/m compared to 3MV/m for air. \$\endgroup\$
    – J...
    Dec 24 '18 at 12:00
  • 2
    \$\begingroup\$ @J... if you want different units you can simply google "70000 per inch" and there you have it. The point of the answer is the relative performance of air and oil, the units are barely relevant \$\endgroup\$ Dec 24 '18 at 20:20
  • 2
    \$\begingroup\$ @CarlKevinson I didn't ask how to convert the units, I suggested that standard units be used since it is one way in which this answer could be improved. \$\endgroup\$
    – J...
    Dec 24 '18 at 20:43

How do you keep the wires in place? Air won't help. It's done with cardboard soaked in oil.

In addition, gases have a nasty property: they have very low pressure so randomly ionized atoms are easily moved by the electric field. If the gas pressure is low enough for the voltage, the ions and free electrons reach the counter electrode before they could recombine so you have a conducting channel, and the heat from the current flow will ionize even more atoms. Oil prevents that simply by its viscosity.

If you want to have excellent isolation with a gas, you had to use a hard to ionize gas with high viscosity, for example Sulfur hexafluoride.

  • \$\begingroup\$ High pressures may be applied to gases as well as liquids. There is no nasty property of gases having very low pressures. \$\endgroup\$
    – Uwe
    Dec 25 '18 at 21:45

You need cooling. And clean air is hard.

The key to the question is where they put transformers - if they were indoors, that'd be neato keen, but in reality they are sited in foul places full of dust, snow, moisture and every other environmental contaminant known to man. And it must give a basically unmaintained service life in the decades.

The air isn't there only to insulate. They could do that with mica. They also need to cool the transformer so they can get more useful current rating out of the same amount of copper and iron.

So they can use air, but environmental issues make interchanging atmospheric air impracticable. So they'd need to hermetically seal it, bottling air inside at manufacture time, though of course it could be some other gas like nitrogen or argon.

The next question is the thermal efficiency of the material as heat transfer fluid. Heat is an excitation level of an atom. Neutrals and protons don't store heat, so mass doesn't store heat, atoms do. Oil is massively more atom-dense than air, being liquid. Oil also expands when heated (just look at your car's dipstick), so hot oil is lighter by volume than cold oil, gravity forces it to rise, and that causes convection circulation. This can be exploited to make it circulate through cooling fins, so no need for a coolant pump.


Look up "dry-type transformer," which today are in common use. There's one of these in a huge outdoor utility box near our building. At least in our case the transformer isn't up on a power pole, so it has less need to be extremely compact and low-weight. A heavy xfrmr with extra iron and thicker copper windings runs cooler, so doesn't need cooling oil and radiator loops. And, if small size isn't a major issue, then the EHT windings can be spaced far apart from the low-volt side, so that air-gap provides sufficient prevention of arc-breakdown.

Note that oil-transformer explosions and fires aren't unknown during thunderstorms and from windstorm-shorted lines. Dry-type transformers lack these failure mechanisms.

enter image description here

  • \$\begingroup\$ There are safety devices for oil transformers like a Buchholz relay.for shorted lines protection. \$\endgroup\$
    – Uwe
    Dec 24 '18 at 13:10

All of the above answers are good, but fail to mention the environmental benefits of dioxin that is present in the oil of many older transformers.

  • \$\begingroup\$ I think you mean poly-chloro-biphenols (PCBs) but that's not relevant to the question. Note that "above answers" doesn't work on StackExchange where answers float up and down by votes or user sort preferences - unless you intend to stay at the bottom somehow! \$\endgroup\$
    – Transistor
    Dec 27 '18 at 21:33
  • \$\begingroup\$ "Environmental benefits of dioxin"?? Surely you mean the environmental impact of dioxin! \$\endgroup\$
    – user39382
    Apr 30 '19 at 19:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.