Reactive power charges and power factor: what kVArh on your bill means, and when correction equipment pays back

Most arguments about a business electricity bill start with the pence per kWh. The line that quietly costs money on half-hourly sites is usually further down: a charge in pence per kVArh, sometimes labelled "reactive power", "excess reactive units" or "low power factor". It looks trivial per unit. On some sites it runs to thousands of pounds a year for no useful energy at all.

What kVArh actually measures

Motors, transformers, fluorescent lighting ballasts, welders and older refrigeration plant do not draw current in step with the voltage. Part of the current builds and collapses magnetic fields rather than doing useful work. That part is reactive power, measured in kVAr, and its accumulated total over time is measured in kilovolt-ampere reactive hours: kVArh.

Active power (kW) and reactive power (kVAr) combine to give apparent power (kVA), which is what the cables and transformers must actually carry. Power factor is the ratio of active power to apparent power. A site at 0.85 power factor needs roughly 18% more current than the same site at 1.0 to do the same work. Distribution networks size their assets for that extra current and lose energy carrying it, so they charge for it.

Worth telling clients early: reactive units are not energy billed twice. You do not pay commodity costs or levies such as the Climate Change Levy on kVArh. The charge is a network cost, passed through by the supplier.

How the charge is worked out

The common approach across GB distribution network operators is a threshold test applied half hour by half hour. Where a half-hourly metered site's reactive units exceed 33% of its active units in the period, the excess kVArh are charged at a published rate in pence per kVArh. That 33% figure is the equivalent of an average power factor of 0.95, as set out plainly in SSEN's 2026/27 use of system charging statement. Ofgem's charge design working paper describes the same mechanism under the Common Distribution Charging Model, and the convergence on a single approach goes back to Ofgem's 2006 review of charging for reactive power.

Four practical consequences follow from the half-hourly test. First, a site can have an acceptable annual average power factor and still be charged, because bad half hours are not netted off against good ones. Second, night and weekend periods with low load and idling transformers are often the worst offenders, since fixed magnetising demand sits against very little active load. Third, rates differ by DNO region and by tariff, so the only reliable number is the one in your network's current charging statement: see National Grid Electricity Distribution and Northern Powergrid%20-%20%202025-26%20LC14%20Charging%20Statement%20-%20v0.2.pdf) for published schedules. Fourth, where reactive data is missing, DNO statements reserve the right to estimate, typically assuming 0.95 lagging, and to charge users who fail to supply the data.

Separate reactive charges are most visible on CDCM half-hourly tariffs at LV and HV. Sites on site-specific EDCM tariffs, generally those connected at EHV or metered at an HV substation, may not see a distinct reactive line, because power factor is reflected in the capacity and asset-based elements instead. Check which model the MPAN sits under before promising a saving.

Finding the problem in half-hourly data

This is where a broker adds value cheaply. Half-hourly meters with current transformer metering record reactive registers alongside active ones, and DNOs require reactive import and export data for the relevant measurement classes. Ask the supplier or the data collector for the kVArh channel as well as the kWh channel, in the same 48-period format.

Then do the arithmetic in a spreadsheet:

• For each half hour, calculate kVArh divided by kWh. Anything above 0.33 is chargeable excess.
• Convert the ratio to power factor using cos(arctan(ratio)) if the client prefers seeing 0.87 rather than 0.56.
• Sum the excess: excess kVArh equals kVArh minus (0.33 times kWh), floored at zero, then totalled across the year.
• Plot the worst periods by time of day and day of week. A flat band of reactive demand overnight points at transformers or permanently energised plant. Spikes during production shifts point at motors and compressors.

Cross-check the result against the DUoS breakdown on the bill or the supplier's pass-through statement. Where the numbers do not match, the reactive registers may be misconfigured, the CT ratio may be wrong, or on-site generation may be exporting reactive power. Metering faults are common enough that ruling them out before quoting for equipment is time well spent.

One note on scope. Market-wide half-hourly settlement is making half-hourly data normal across the market, with Elexon expecting around 80% of meters migrated by October 2026 and completion in 2027, according to its MHHS pages. That does not mean every small smart-metered shop suddenly has reactive registers. Reactive measurement still belongs to the CT-metered population.

When correction equipment pays back

Power factor correction usually means an automatic capacitor bank at the main switchboard, switching steps of kVAr in and out as load varies. Where variable speed drives, LED drivers or UPS systems are present, the bank needs detuned reactors to avoid harmonic resonance, which adds cost.

The payback sum is straightforward. Take an illustrative site using 2,000,000 kWh a year with 1,200,000 kVArh recorded, giving a ratio of 0.6 and a power factor near 0.86. The 33% allowance covers 660,000 kVArh, so 540,000 units are chargeable. At a DNO rate of 0.5p per kVArh, that is £2,700 a year. If correction also lets the site cut agreed capacity by 60 kVA at a capacity rate of 3p per kVA per day, that is a further £657 a year. Against an installed quote of £9,000, payback lands under three years.

Use the client's real rates rather than these figures, because both the reactive rate and the capacity rate vary sharply by region and voltage. Three things change the answer materially:

• Capacity reduction is not automatic. You must apply to the DNO to lower agreed supply capacity, and the site needs headroom to take the reduction safely.
• Contract structure matters. On a fully fixed, all-inclusive price, reactive charges may already be baked into the unit rate, so the saving only appears at renewal or if the contract moves to pass-through.
• Maintenance is real. Capacitors degrade and contactors fail; a bank that is not serviced quietly stops correcting. Budget for annual checks and expect a working life of roughly ten to fifteen years.

Below about 100,000 kWh a year on half-hourly metering, a bank rarely justifies itself on reactive charges alone. Sites with large motor loads and oversized transformers running at low utilisation usually do.

What a broker can do this week

Run a quick screen across your half-hourly book. Pull twelve months of kWh and kVArh per MPAN, calculate the annual ratio, and flag anything above 0.33. On flagged sites, ask the supplier for the reactive line from the DUoS breakdown and confirm the current p/kVArh rate from the DNO's published statement for 2026/27.

Put a one-page note to each affected client showing annual excess kVArh, the cost at the published rate, the worst time bands, and a likely equipment cost range from two local quotes. Include the contract check: if the site is on a fixed bundled price, say when reactive costs can actually be removed. Where power factor is close to 0.95 and the excess is small, say so and park it. Recommending a capacitor bank against £400 a year of charges damages trust faster than doing nothing.

Keep an eye on the meter side too. A site showing zero kVArh where you would expect some, or a ratio above 1.0, needs raising with the data collector before anyone spends money on plant. Harmonic distortion and on-site generation both complicate the picture and deserve a separate conversation with an electrical engineer.