CleanZone Field Brief

Will the Lights Stay On? Grid Reliability by Postcode

SAIDI and SAIFI are not household names, but they predict whether your street goes dark when the wind picks up. These public metrics, plus substation age and storm frequency, are what separate a reliable postcode from one that lives by candlelight every winter.

Two streets sit three miles apart. One sits at the end of a single overhead line running eleven miles through forest before it reaches a substation; the other sits inside a looped urban network fed from two directions at once. The same October gale drops the same oak branch on both. On the urban street, a relay senses the fault, a switch opens, and power is back from the other direction before anyone has found a torch. On the rural street, the entire line is dead until a crew drives out, finds the branch, and clears it — sometimes within the hour, sometimes not until the next day. Nothing about the storm was different. Everything about the grid feeding each street was.

That difference has a name, and regulators measure it every year. In the United States, the Energy Information Administration (EIA) collects SAIDI (System Average Interruption Duration Index) and SAIFI (System Average Interruption Frequency Index) from utilities and publishes them by state and utility territory. In Europe, the Council of European Energy Regulators (CEER) runs the equivalent exercise through its Benchmarking Report on the Continuity of Electricity and Gas Supply. Between them, these bodies turn "does the power stay on here" from a rumour into a number sitting in a public filing, not a marketing brochure.

SAIDI and SAIFI: what they actually measure

SAIDI is the total duration of customer outages divided by the total number of customers served. It captures the aggregate pain of an outage event: if a storm takes out a substation and three thousand customers lose power for six hours, that is 1,080,000 customer-minutes into the SAIDI numerator. SAIFI is the total number of customer interruptions divided by the total number of customers. The same storm, contributing one interruption to three thousand customers, adds three thousand customer-interruptions to the SAIFI numerator instead.

The distinction matters because different failure modes produce different SAIDI/SAIFI fingerprints. A network plagued by frequent momentary blips — tree contacts on overhead lines, automatic reclosing operations — shows a high SAIFI but a modest SAIDI: annoying, rarely costly. A network with rare but catastrophic failures — a transformer fire, a storm-felled distribution backbone — shows a modest SAIFI but a brutal SAIDI. A buyer who reads only one of the two numbers is reading half the risk.

Capacity and reserve margin

Transmission and distribution operators publish equipment ratings, but the rating alone says nothing about comfort. What matters is the ratio of peak demand to rated capacity — the reserve margin. In the United States, EIA Form 860 collects generator and power-plant data, and the North American Electric Reliability Corporation (NERC) publishes anticipated reserve margins by region in its Long-Term Reliability Assessment. In Europe, ENTSO-E publishes the equivalent transmission adequacy data. A substation running above eighty percent of rated capacity at summer peak has no headroom for ordinary load growth, none for a heatwave's extra air-conditioning draw, and none for the load that transfers onto it automatically when a neighbouring substation trips.

Age and condition

The U.S. Department of Energy and the International Energy Agency (IEA) both publish reports on grid infrastructure age. In developed economies, a significant share of distribution transformers and substation equipment dates from the 1960s and 1970s. Age alone doesn't equal unreliability — maintenance regimes matter — but it does mean replacement cycles are approaching, and deferred investment shows up in SAIDI eventually. The UK's Ofgem publishes network asset age profiles by company; where that data is public, it's a leading indicator of tomorrow's SAIDI, not just a record of yesterday's.

Radial rural feeders vs meshed urban grids

Topology is the part buyers skip because it sounds like an engineering detail. It isn't. A radial feeder is a single line running out from a substation with no alternate path back: lose a segment, and every customer downstream is dark until the fault is physically repaired. A meshed network loops multiple substations together with tie switches that open and close automatically, so a fault on one segment can be isolated and the customers beyond it re-fed from a second direction — often in seconds, via automated switching, without a truck ever leaving the depot.

Rural areas end up on radial feeders for a reason that has nothing to do with neglect: looping every farmhouse and hamlet into a meshed network means building and maintaining a second set of lines across low-density terrain, for a customer count that can't support the cost. Urban cores get meshed networks because customer density justifies the redundant infrastructure. That's an economic decision made decades ago, and it still governs whether this winter's storm becomes a one-hour blip or a two-day ordeal.

Storm frequency and the exposure multiplier

Climate reanalysis datasets — ECMWF's ERA5, NASA's MERRA-2 — provide gridded wind-speed and lightning-strike density going back decades. The NOAA Storm Events Database and the equivalent national hydrometeorological services across Europe catalogue the actual storm tracks. Both point the same direction over the past two decades: more days per year crossing damaging wind-gust thresholds, and a growing share of utility-reported major-event days attributed to weather rather than equipment failure. A postcode doesn't need a catastrophe model to use this — it needs to know whether it sits in a corridor of above-median storm exposure, because that exposure multiplies whatever topology and asset-age risk is already there.

A radial feeder in a low-storm corridor can outperform a meshed network in a high-storm one. Topology sets the ceiling; weather decides how often you hit it.

What the data cannot say

SAIDI and SAIFI are backward-looking by construction. A postcode with a clean ten-year record and fifty-year-old equipment may be due a step-change the moment a major component fails. A postcode with a poor record and a recent capital upgrade may already be improving faster than its trailing average suggests. The indices need to be read alongside forward-looking filings — RIIO business plans in the UK, rate-case submissions in the United States, ENTSO-E's Ten-Year Network Development Plan process. Where those filings are public, and in most regulated markets they are, they contain the investment narrative that turns a historical number into a forecast.

What separates a resilient postcode from a fragile one

None of these four factors — index history, capacity margin, topology, storm exposure — is decisive alone. A resilient postcode stacks favourably on most of them: SAIDI and SAIFI at or below benchmark, serving-substation headroom below the tight zone, a meshed or partially looped local network, storm exposure at or below the regional median. A fragile postcode stacks the opposite way, and the stack compounds — an ageing, tightly loaded substation at the end of a long radial feeder in a high-storm corridor isn't four separate risks, it's one risk multiplied four times.

• Latest SAIDI and SAIFI figures pulled for the utility serving the target postcode
• State or national reliability benchmark identified for comparison
• Serving substation located and its peak-load ratio (reserve margin) estimated
• Substation and asset age checked from regulatory filings or utility reports
• Local network topology confirmed: meshed/looped vs long radial feeder
• Storm event history pulled from NOAA, Met Office, or national equivalent
• Wind-gust and lightning density checked from ERA5 or MERRA-2 reanalysis
• Utility investment plans reviewed for funded upgrades in the next five years
• Backup power requirement weighed against outage frequency and duration