A row of digital electricity smart meters with lit display screens on a wall, the nearest one in sharp focus.
Smart meters are now the largest source of raw energy readings by sheer count. Each one records the energy passing a property and stores a fresh reading at the close of every half-hour. Photo: Pexels

Where Energy Data Begins

Every useful energy dataset starts as a physical observation. A smart meter counts energy passing a home. A substation sensor watches voltage and current. A transmission synchrophasor measures the waveform fast enough to see the system wobble. Each reading then makes a first handoff into the systems that carry it onward.

Scope: the sensors that produce the readings, the readings themselves, and each reading's first handoff. The carriage of that data, the settlement run and the data rules are linked at the points they matter, not repeated here.

Sources and standards

Quantitative and regulatory claims resolve to primary publications from Ofgem, NESO, DESNZ, Elexon and BSI, to instruments on legislation.gov.uk, and to the named measurement standards: DLMS/COSEM for meters, IEC 60870-5-104 and DNP3 for substation telemetry, IEEE C37.118 for synchrophasors, and IEC 61968-13 and IEC 61970-301 for the asset model.

Where Energy Observations Begin

An observation is a single recorded fact: a number, a time, and the thing it describes. In the GB electricity system almost every observation begins at one of three kinds of physical sensor. The first is the smart meter, the small box that records how much electricity a home or business uses. The second is substation telemetry, the instruments inside the buildings and yards where the grid changes voltage and routes power. The third is the synchrophasor, a high-speed sensor on the high-voltage network that measures the precise shape and timing of the alternating current.

These three sources differ in four ways that matter for everything downstream: what they measure, how often they measure it, who owns the resulting data, and how quickly that data needs to be used. A smart meter records one number every half-hour and a person may never look at it. A substation sensor records voltage and current every few seconds and a control engineer may act on it within the minute. A synchrophasor records fifty samples a second and an automatic system may respond before any human sees it. The slower, settle-it-later readings end up on bills; the faster, act-on-it-now readings keep the lights on. Reading the system correctly means knowing which kind of observation is in hand.

Two recent changes set the scene. The rollout of smart meters reached about 37 million meters operating in smart mode by the end of 2025, roughly 71 percent of the meter base, so the smart meter is now the dominant source of raw readings by sheer count.8 And the network's asset records moved onto a shared model: the Long Term Development Statement Stage 2 publication of 29 May 2026 sets out distribution network assets in a common data format, so a reading taken at a substation can be tied to a stable record of that substation.1 The operational and published data feeds that sit on top of these sensors are catalogued by NESO and the network companies.7

The three sensor families, the reading each one takes, and where the reading is first handed off

Based on the smart metering device specifications under the Smart Energy Code, the substation telemetry standards IEC 60870-5-104 and IEEE C37.118, and the operational data feeds published by NESO and Elexon. The figure sets out the three sensor families that produce almost every raw reading, what each one measures and how often, and the first handoff where the measurement leaves the device that took it.

The three sensor families, the reading each takes, and its first handoff A three-row table-style diagram. Each row follows one sensor family across three stages: the equipment, the reading it takes, and the first handoff. Row one: a smart meter, a SMETS2 electricity meter in a home or small business, produces energy used in kilowatt-hours as one reading every half-hour, handed first to the communications hub at the property and then onto the smart-data pipe. Row two: substation sensors, RTU and IED units in grid and local substations, produce voltage, current, power and switch status every one to ten seconds, handed first to the operator control room running SCADA-EMS. Row three: transmission sensors, synchrophasors or PMUs that are GPS time-locked, produce voltage and current phase angle at fifty samples every second, handed first to wide-area monitoring at the national control centre. A footer band notes that every reading is pinned to a fixed asset identity, an mRID, in the LTDS and CIM network model so the same meter point or substation can be compared across publications over time. Where each observation begins: the sensor, the reading it takes, and the first handoff off the device. Equipment The reading First handoff 1. Smart meter SMETS2 electricity meter, in a home or small business Energy used, in kilowatt-hours one reading every half-hour Comms hub at the property, then onto the smart-data pipe produces handed to 2. Substation sensors RTU and IED units in grid and local substations Voltage, current, power, status every 1 to 10 seconds The operator control room running SCADA-EMS produces handed to 3. Transmission sensors Synchrophasor (PMU), GPS time-locked Voltage and current angle 50 samples every second Wide-area monitoring at the control centre produces handed to Asset records: every reading is pinned to a fixed asset identity (an mRID) in the LTDS and CIM network model, so the same meter point or substation can be compared across publications over time.

The first handoff is the boundary of physical data generation. A smart meter reading enters the smart-data pipe at the property; substation telemetry enters the operator's control room; a synchrophasor sample enters wide-area monitoring at the national control centre. Everything after the first handoff, the carriage of the data, the settlement run and the data rules, belongs to the pages linked below.

Smart meters: tens of millions of sensors reading every half-hour

A smart meter is the largest single source of energy data by the number of readings it produces. By the end of 2025 about 41 million smart and advanced meters had been installed in homes and small businesses, of which more than 37 million were operating in smart mode, around 71 percent of the meter base. A meter is in smart mode when it is on a known communications path and exchanging messages with the central smart-metering system; the rest sit in a basic mode for reasons such as a poor signal or a configuration still to be completed.8

An electricity meter with a small digital display mounted on a white-painted brick wall, beside two grey isolator boxes.
An electricity meter records a running total of energy used and shows it on a small digital display. In smart mode the meter stores a sealed reading for each half-hour and waits for an authorised party to ask for it. Photo: Pexels

An electricity smart meter takes one active import reading at the close of each half-hour, time-stamped to its own clock, and stores it locally. That is 48 readings a day. The reading is signed by the meter so that a later reader can tell it has not been altered, and it waits until an authorised party, usually the energy supplier, asks for it. The default request is a daily pull of the previous day's 48 readings. A gas smart meter works differently because it runs on a battery: it wakes once a day, exchanges its reading, and returns to a low-power state, so gas stays on a daily rhythm rather than a half-hourly one.

How many readings a day the meter fleet produces

Count only the meters that settle half-hourly. About 33 million second-generation electricity meters each take one reading every half-hour:

33,000,000 meters × 48 readings/day = 1,584,000,000 readings/day

That is roughly 1.58 billion readings produced every day, before a single one of them moves off the meters. The number rises further as the move to half-hourly settlement for every meter point completes, which is covered on the data lifecycle and settlement page.

PopulationCountReading rhythmNote
All smart and advanced metersaround 41 millionvariesInstalled by end of 2025; about 71 percent of the meter base
Meters in smart modemore than 37 millionhalf-hourly (electricity)On a known communications path with the central system
Second-generation electricity metersaround 33 millionhalf-hourlyInteroperable when a household switches supplier
Gas smart meterslarge minoritydailyBattery powered; a daily wake-up rather than half-hourly

What a smart meter reading is not is also worth holding. It is not raw, continuous data; it is a tidy total for a fixed half-hour. It is not read by default at finer resolution than monthly for the supplier's own use: a household keeps the right to limit how granular the supplier's access is, with daily and half-hourly access taken only with consent. That consent layer is being put on a standard footing by the Consumer Consent Solution, which Ofgem decided on 29 April 2025 and appointed the Retail Energy Code Company to operate, so that a household can grant and manage consent to share its energy data.12 The wider privacy rules for meter data sit in the Data Access and Privacy Framework,10 and the statutory duty for smart-meter communications now sits in Schedule 16 of the Data (Use and Access) Act 2025.9 How a reading is then carried, and how it reaches settlement, is set out on the digital infrastructure page.

Network telemetry: substation sensors and transmission synchrophasors

Network telemetry is the second source of raw readings. It produces far fewer observations than the meter fleet because there are far fewer assets, but each asset is watched far more often. The instruments live in substations, the sites where the grid steps voltage up or down and routes power between circuits. A large transmission substation is heavily monitored; a small local substation that feeds a few streets may have almost no instruments at all.

High-voltage substation equipment under a clear sky, with porcelain insulators, instrument transformers, busbars and steel support towers.
Instrument transformers and switchgear inside a substation feed the sensors that measure voltage, current and switch status. The readings travel to the operator's control room every few seconds. Photo: Pexels

The workhorse device is the Remote Terminal Unit, often working alongside Intelligent Electronic Devices that combine protection and measurement. Together they record the voltage on each busbar, the current and power on each circuit, the position of each switch, the tap position and temperature of each transformer, and any fault event. At transmission level the readings flow on a one to two second rhythm; at distribution level it is more like two to ten seconds at a main substation, and event-driven or nothing at a small one. The data is carried to the operator's control room using standard protocols, chiefly IEC 60870-5-104 at transmission and DNP3 at distribution, where engineers see it and the control system runs its safety checks against it.7

Above ordinary telemetry sits the synchrophasor, produced by a Phasor Measurement Unit. A PMU measures the voltage and current waveform 50 times a second, works out the phase angle, and stamps every sample against the GPS time signal so that samples from sites hundreds of kilometres apart can be compared to the same instant. Around 80 PMUs across the transmission network feed a wide-area monitoring service that NESO and the transmission owners use to watch system stability and spot oscillations before they grow. The Grid Code sets the operational duties at transmission that this monitoring supports.5

Why the synchrophasor estate is small but data-heavy

Each PMU reports 50 samples a second, and each sample carries voltage and current magnitude and angle on several circuits. Across about 80 PMUs that is:

80 PMUs × 50 samples/second = 4,000 samples/second

That is only a few dozen sensors, yet at full reporting they produce a continuous stream of the order of hundreds of megabytes a second once every channel is counted, which is why the stream is sampled down and kept in full only around events.

TierDeviceReading rhythmProtocolOwner
Wide-area transmissionPMU (synchrophasor)50 samples per secondIEEE C37.118NESO and the three transmission owners
Transmission substationRTU and IED1 to 2 secondsIEC 60870-5-104; IEC 61850NGET, SPT, SSEN-T
Distribution substation (main)RTU and IED2 to 10 secondsDNP3; IEC 60870-5-104The 14 network operators
Distribution substation (small)Limited or noneevent-drivenvariesThe 14 network operators

Telemetry does not normally leave the company that owns it. It is the live picture the operator uses to run the system, and only aggregated or derived figures are published: the national demand outturn, the generation mix behind the Carbon Intensity service, the wind output. None of these public feeds republishes the raw substation stream; each one is a summary computed against it inside the operator. The asset model that lets a reading at one substation be compared with the same substation later rests on the Common Information Model base in IEC 61970-301.4

Dynamic line ratings: turning local weather into extra network capacity

A dynamic line rating is a worked example of new sensors creating new data to solve a real problem. An overhead line can only carry so much current before it heats up and sags too far. The traditional limit is a fixed seasonal figure, set against a cautious assumption about a still, warm day. But a line in a cold wind can safely carry more, because the wind cools the conductor. A dynamic line rating measures the actual weather and the actual conductor temperature and works out, minute by minute, how much the line can really carry right now.

A pole-mounted weather sensor station with a spinning-cup anemometer and a ventilated radiation shield, set against green trees.
A weather sensor station measures wind speed, air temperature and sunlight near the line. These readings let a dynamic line rating work out how much current the conductor can safely carry in the conditions of the moment. Photo: Pexels

The data flow is an extension of ordinary telemetry. Weather sensors on the tower or the conductor measure wind speed, wind direction, air temperature and sunlight; a temperature sensor or a thermal model tracks the conductor itself. A small computation turns these into a live current rating that feeds back into the control room, where it informs how much power can be routed down that line. The reading rhythm is typically one to ten minutes.7

How much extra a cold wind buys

A conductor's safe current depends on the balance between the heat the current adds and the heat the wind and air carry away. Take a line whose fixed seasonal rating assumes still air at 20 degrees:

still-air rating: 100 units → with a 4 m/s crosswind: roughly 120 to 130 units

The same physical line safely carries 10 to 30 percent more on a cold, breezy day, with no new steel built. That headroom matters now because the Connections Reform Gate 2 results of April 2026 progressed about 283 GW of generation and storage and 99 GW of demand toward firm grid offers, far faster than new lines can be built.11 The operational rules for using a changing rating sit in the Distribution Code and the Grid Code.6

Asset records: giving every reading a stable identity in the network model

A reading is only useful when its subject is known. A voltage of 11,200 volts means little until it is tied to a named transformer, at a named substation, on a named circuit. Asset records are the part of the data layer that supplies that identity. They are not sensor readings; they are the slow-changing register of what exists on the network and how it connects, and the live readings hang off them.

The register is organised by the voltage cascade, the steps the grid takes from the highest transmission voltage down to the socket. Each step has its own population of assets, its own sensors and its own place in the shared data model. The table below ties each voltage tier to the kind of asset record that represents it.

TierNominal voltageSensing rhythmAsset record
Transmission super-grid400 kV, 275 kV1 to 2 seconds; 50 per second on monitored circuitsLine, busbar and substation classes in the transmission model
Distribution boundary132 kV2 to 10 secondsLine and transformer classes; distribution profile
High-voltage distribution33 kV and 11 kV2 to 30 secondsLine, transformer and switch classes; distribution profile
Low-voltage service230 V single-phase, 400 V three-phasehalf-hourly at the meterConsumer and usage-point classes; meter linkage

Two standards supply the shared vocabulary. IEC 61970-301 sets the Common Information Model base used for the transmission network,4 and IEC 61968-13 sets the distribution profile that represents the lower-voltage network.3 Each asset in the published model carries a stable identifier, an mRID, that is kept across every later publication, so the same transformer can be tracked from one statement to the next. The first large GB instance of this regulated, model-based publishing is the Long Term Development Statement Stage 2 release of 29 May 2026, with the agreed schedule and the move to a common model set out in Ofgem's derogation letter of 13 May 2026.1 The official location of record for the data-exchange definitions is the BSI CIM Engagement Hub.2 The full treatment of this model, and how it is validated, is on the LTDS explained page; the voltage cascade as an engineering subject is on the voltage page.

The first handoff, and where the longer journey continues

The first handoff is the natural edge of physical data generation. Up to that point the story is about equipment and measurement; past it, the story is about carriage, trust, settlement and rules, and each of those has a page that owns it properly. Holding the boundary in mind keeps the picture from blurring into one undifferentiated stream.

  • A smart meter reading is handed to the communications hub at the property and then enters the smart-data pipe. How that pipe is licensed and how it carries the reading is on the digital infrastructure page.
  • Once carried, an electricity reading is reconciled into money through the settlement run. That seven-stage journey from a meter reading to a settled bill is on the data lifecycle and settlement page.
  • Substation telemetry is handed to the operator's control room and stays there, surfacing publicly only as the aggregated feeds that the wider energy data layer catalogues by type, owner and format.
  • Who may see which reading, and under what consent and privacy rules, is set out alongside the carriage on the digital infrastructure page.

Read in order, the three sensor families, their readings and their first handoffs are the foundation the rest of the data layer is built on. Everything a market, a regulator or a planner later does with energy data depends on a measurement that began at one of these sensors and was handed off, intact and identified, into a system that could carry it.

Primary sources

The most load-bearing sources for how energy observations are made and first handed off are listed below.

  1. LTDS CIM Stage 2 and 3 Extension (Derogation) Letter, Ofgem, dated 13 May 2026. Sets the Stage 2 publication of 29 May 2026 and the move to a common, model-based publication of distribution asset records. https://www.ofgem.gov.uk/sites/default/files/2026-05/LTDS-CIM-Stage-2-and-3-Extension-Derogation-Letter.pdf
  2. BSI CIM Engagement Hub. The UK location of record for the Common Information Model data-exchange definitions used by the LTDS publication. https://cim.bsigroup.com/
  3. IEC 61968-13 Edition 2.0 (BS EN IEC 61968-13:2021). The Common Distribution Power System Model profiles that represent lower-voltage network assets. https://webstore.iec.ch
  4. IEC 61970-301 Edition 7.0 with Amendment 1:2022. The Common Information Model base for the transmission network and energy management systems. https://webstore.ansi.org/standards/din/dineniec619703012025
  5. The Grid Code, NESO, Issue 6 Revision 37, 13 April 2026. Sets the operational duties at transmission that wide-area monitoring supports. https://www.neso.energy/industry-information/codes/grid-code-gc
  6. The GB Distribution Code, Issue 59, 24 April 2026, Distribution Code Review Panel. Sets the operational rules at distribution, including the use of changing line ratings. https://www.dcode.org.uk/
  7. NESO Data Portal. Operational and derived data feeds, including the demand outturn, generation mix and the inputs behind dynamic ratings, under the NESO Open Licence. https://www.neso.energy/data-portal
  8. Market-wide Half Hourly Settlement: key programme milestones, MHHS Programme. Tracks the smart-meter reading base and the move to half-hourly settlement for every meter point. https://www.mhhsprogramme.co.uk/programme-information/key-programme-milestones
  9. Data (Use and Access) Act 2025, Schedule 16. Holds the statutory duty for smart-meter communications licences in line with the Energy Act 2008. https://www.legislation.gov.uk/ukpga/2025/18/schedule/16
  10. Smart Metering Implementation Programme: review of the Data Access and Privacy Framework, DESNZ. The privacy framework over meter data, setting purpose limits and graded consumer consent. https://www.gov.uk/government/publications/smart-metering-implementation-programme-review-of-the-data-access-and-privacy-framework
  11. Connections Reform Gate 2 results, NESO, April 2026. The volume of generation, storage and demand progressed toward firm grid offers, which raises the operational case for dynamic ratings. https://www.neso.energy/industry-information/connections-reform/connections-reform-results
  12. Consumer Consent decision, Ofgem, 29 April 2025. Appoints the Retail Energy Code Company to operate an enduring Consumer Consent Solution for sharing energy data. https://www.ofgem.gov.uk/decision/consumer-consent-decision

The named measurement standards, DLMS/COSEM for meters, DNP3 and IEC 60870-5-104 for substation telemetry, IEEE C37.118 for synchrophasors, and IEC 61850 for substation automation, are cited inline as the technical instruments of the sensing layer.