Smart Meter Data for UK Solar Design: SMETS2

Last updated: November 2026

Two homes can use the same annual electricity total and need different solar designs. One may run appliances through the afternoon; the other may import most of its electricity after sunset. Interval meter data helps distinguish those patterns before an installer promises savings or chooses a battery.

This guide covers access routes for UK smart meter data, cleaning and aligning readings, PV and storage calculations, and handling customer information. The worked design example is illustrative, not a customer case study. Supplier prices, equipment specifications, and access arrangements need project-specific evidence.

SurgePV publishes this post and sells solar design software.

How SMETS2 Smart Meters Share Data

A smart-meter dataset records energy over defined intervals. For solar design, identify whether each series represents grid imports, grid exports, cumulative register readings, or a separate generation measurement before calculating a load profile.

SMETS2 means Smart Metering Equipment Technical Specifications, second generation. The meter and communications equipment are part of a wider data system; access through a supplier or authorized service is different from a local device reading inside the home. Do not assume a meter’s presence guarantees a complete historical download.

Hildebrand’s Bright documentation describes both a hardware route and delayed half-hourly data obtained through DCC after verification. That distinction matters: a CAD is not universally required for historical data, and a historical service should not be described as a live appliance monitor.

Import, Export, and the Existing-Solar Problem

Grid import is not total household consumption when PV or storage already operates behind the meter. A house may consume solar electricity without that energy appearing as an import. Export data alone also cannot reveal all the solar energy used inside the home.

For a site with PV but no battery, a simplified energy balance is load = import + PV generation - export, using aligned measurements at compatible boundaries. Check losses and meter locations before using that relationship. With a battery, charging, discharging, and losses add further terms.

If export readings are missing, ask the supplier or data provider to investigate the account, meter, and available series. A blank export file is not proof that the supplier simply forgot to activate a register. Do not replace missing exports with zeros and call the dataset complete.

Keep the data-access issue separate from an export payment contract. The supplier’s requirements for an export arrangement and the data available through an API may not be identical. Record both statuses so the financial model does not assume payment for a series that has not been verified.

Why Half-Hourly Data Changes Solar Design

Half-hourly data shows when energy crosses the meter, which helps compare it with modeled solar generation and tariff periods. It improves the evidence for a design without guaranteeing a particular sizing accuracy or payback.

A half-hourly energy value is not instantaneous power. For example, 0.5 kWh over half an hour corresponds to an average of 1 kW during that interval. A short appliance spike may be much higher than the interval average, so this data cannot replace electrical demand or starting-current analysis.

This comparison describes information content, not measured accuracy percentages. The earlier table claiming fixed PV and battery sizing errors for each input type was unsupported. A year’s data from the wrong meter can be less useful than a carefully understood shorter sample.

How Much History Is Useful?

Prefer a full representative year when available because it covers seasonal behavior. If less history is available, state the period and the conditions it represents. There is no universal minimum number of days that makes every solar sizing decision safe or accurate.

Ask about holidays, vacancies, work-from-home changes, electric heating, and EV charging. A dataset can be complete and still fail to represent the future customer. Keep planned loads in a separate scenario until their energy and operating schedules are agreed.

Compare winter and summer weeks, weekdays and weekends, and occupied and unoccupied periods where relevant. Avoid assuming that every UK home has the same seasonal multiplier. Electric heating, gas heating, and different occupancy patterns can produce quite different shapes.

Three Routes to Obtain Half-Hourly Data

Choose an access route that supports the actual meter and customer, with appropriate authorization. Verify available history, export coverage, units, and format before promising a particular delivery time or integration.

Route 1: n3rgy

n3rgy provides consent-management and meter-data services. Its consumer page describes visibility over organizations with access, consent withdrawal, and data download. Its support material explains that access requires end-customer consent and describes different ways business customers can establish that evidence.

Do not tell every homeowner to create an unrestricted standalone developer account. n3rgy also publishes a notice about closing consumer signups in October 2024 and access tied to active consent with business customers. Check the available route for the customer before specifying it in the installation workflow.

For a business integration, review the provider’s actual onboarding, consent evidence, permissions, and commercial terms. A free consumer download does not prove that unrestricted commercial API use is free or available. Keep the account owner and permitted purpose clear.

After access is established, inspect the available meters and series. Verify the dates returned rather than assuming a fixed amount of history. Save the retrieval date and the relevant metadata with the project so later changes can be understood.

Route 2: Hildebrand Glow and Bright

Hildebrand’s API guide, version 1.8 dated April 2026, describes account setup, verification, resource identification, and retrieval of time-series data. It distinguishes delayed DCC readings from Glow hardware access and gives a separate process for MQTT access.

Use the documented resource definitions to identify consumption, export, and units. Do not assume every account contains every series or that a resource label has the same meaning as a field in another provider’s export. Keep the original metadata available during cleaning.

If the design question involves short-duration equipment operation, discuss an appropriate monitoring route with the provider. Do not infer a guaranteed update interval or history length from the existence of a CAD. The original article’s hardware price and latency claims were not retained without matching current documentation.

Route 3: Supplier Download or Supported API

Ask whether the customer’s supplier can provide an interval-data export or a supported API route. Octopus customers, for example, can consult the supplier’s developer documentation for the account and meter access available to them. Use the actual instructions for that account rather than assuming a particular credential prefix or response format.

Where an API is used, protect credentials and restrict access to authorized people. A customer-supplied file may be sufficient for a one-off design and avoids retaining ongoing account access. Select the least access needed for the agreed task.

Check whether a response is paginated and whether the requested date range is complete. The presence of recent readings does not mean an entire year was returned. Compare start and end timestamps, interval count, and totals before handing the file to the designer.

Comparing Access Routes

No route is ranked fastest or most accurate here. Choose based on verified coverage and the project question, then retain a fallback when a provider cannot retrieve the required history.

Parsing the Export and Building a Clean Load Profile

Preserve the original file and create a separate normalized dataset. Record every correction, exclusion, and estimate so the designer can distinguish measured readings from assumptions.

A useful normalized record includes meter identifier, register type, interval start, interval end, energy value, unit, source, and quality status. Do not discard provider flags simply because a spreadsheet template lacks a column for them.

Issue 1: Missing Intervals

A gap may reflect communications, retrieval, account coverage, meter replacement, or another issue. Investigate the cause before filling it. Do not attribute every gap to a HAN dropout or apply one interpolation rule to all missing periods.

For an isolated gap, an estimate may be acceptable for a stated feasibility purpose if clearly flagged. Compare plausible alternatives and show whether the design decision changes. For substantial missing periods, obtain additional data or keep the result explicitly provisional.

Never turn missing values into zero consumption without evidence. That can exaggerate exports and change battery dispatch. Preserve a quality mask so totals can be reported for measured and estimated periods separately.

Issue 2: Meter Replacement and Register Changes

Identify meter serial changes and the period covered by each meter. A cumulative register reset should not be interpreted as negative interval consumption. Conversely, an interval-energy file should not be differenced as though it were a cumulative register.

Check supplier communications and metadata around abrupt changes. Do not automatically remove a fixed number of days after migration or replacement. The appropriate treatment depends on the actual data quality and measurement boundary.

Issue 3: British Summer Time and Timestamps

Use timezone-aware timestamps and retain the original offset. Normalize the calculation timeline to a consistent basis, then convert to local time for customer-facing profiles and time-dependent tariffs as needed.

A local clock-change day can have fewer or more half-hour periods than an ordinary day. Do not force every local day to contain exactly the same number of rows by deleting duplicates or inventing missing records. Compare timestamps and interval boundaries before deciding that a row is duplicated.

Issue 4: Missing Export or Generation Data

On an existing-PV site, determine which measurements are required to reconstruct load. Grid import alone generally understates total demand during solar production. Missing generation data may require another meter or a qualified reconstruction method rather than an arbitrary uplift.

Check whether a zero export interval is plausible. The home might be consuming all generation, the battery might be charging, an export limit might apply, or the data might be incomplete. The series itself does not identify the cause.

Build Daily and Seasonal Profiles Correctly

Sum energy within each day or time window before averaging across days. An average half-hourly reading is not a whole-window energy total. State the window, timezone, included days, and missing-data treatment next to the result.

For example, a six-hour daytime window contains twelve half-hours on an ordinary day. If each interval averages 0.42 kWh, the window energy is 5.04 kWh, not 2.5 kWh. This corrects the dimensional error in the previous example.

A low percentile of interval demand can help describe quiet periods, but it is not a direct measurement of permanently running appliances. Seasonal heating, cycling refrigeration, and occupancy can affect it. Use the metric as a descriptive indicator and avoid labeling it a verified appliance baseload.

For evening demand, report a distribution rather than only a selected worst-day average. A battery designed around rare peaks may be underused on most days. Compare the customer’s objective with the cost and frequency of those peaks before selecting capacity.

Sizing the PV Array Against Real Consumption

Compare feasible array sizes using the same load profile, weather assumptions, tariff, and connection constraints. Maximum self-consumption percentage is not the same objective as maximum financial value or maximum annual energy production.

Step 1: Generate a PV Profile for the Roof

Use a documented model with location, orientation, tilt, equipment, shading, and losses appropriate to the site. The European Commission’s PVGIS tool provides resource and PV estimation functions, including hourly data options. Record the database, period, and output units.

Check whether an output column is power or energy before combining it with meter kWh. Convert using the interval duration where appropriate. Keep the model’s azimuth convention explicit; compass headings and a tool’s signed orientation convention may differ.

Model separate roof planes where their timing differs materially. A single annual total cannot represent the effect of east-facing and west-facing surfaces on time-of-day generation. Retain the shading assumptions with the production scenario.

Step 2: Align Generation and Load

Match interval boundaries and timezones before calculating overlap. If hourly generation is split into two equal half-hours, label that as a simplifying assumption. It cannot recover within-hour weather variability and may affect overlap estimates.

For a site without existing generation or storage, the measured import profile can serve as a historical load baseline, subject to known changes. For an existing solar or battery site, reconstruct or measure load first. Replaying a new system against net imports from an old system can double-count benefits.

For a no-battery calculation with aligned interval energy:

direct_use = min(load_energy, pv_energy)
export_energy = max(0, pv_energy - load_energy)
import_energy = max(0, load_energy - pv_energy)

These equations describe an energy balance at the chosen boundary. They do not include storage, export curtailment, power limits, or a tariff. Add those constraints explicitly when they apply.

Step 3: Compare Feasible Sizes

For each layout, calculate generation, direct use, exports, imports, and cost using consistent assumptions. A larger array can have a lower self-consumption percentage while still supplying more useful energy. Do not reject it solely because the percentage falls below an arbitrary threshold.

Check the marginal equipment and installation cost against the incremental modeled value. Use the actual proposed import and export prices, with sensitivity cases where future prices are uncertain. Do not assume one national export rate or a universal array-size cap.

Roof area, structural suitability, inverter compatibility, and network constraints remain necessary checks. Meter data does not make an electrically invalid layout acceptable. Keep the energy comparison separate from the engineering approval.

Sizing the Battery Against the Load Shape

Size storage by both usable energy and power, using a defined operating objective. Evening load is useful evidence, but it is not a universal battery-sizing formula.

Define the Operating Objective

Ask whether the battery is intended for solar shifting, time-of-use charging, backup, export management, or a combination. Those objectives may compete for state of charge. A reserve held for an outage is not simultaneously available for daily tariff savings.

Identify when charging energy is available and at what cost. A large evening load does not prove that enough surplus PV will fill the battery in winter. If grid charging is proposed, model its tariff and any relevant equipment or contract restrictions.

Check Energy and Power Separately

Usable energy determines how long a battery can supply a given load, while power limits determine what it can supply at once. Use the selected manufacturer’s usable-capacity definition and operating limits. Do not apply one depth-of-discharge percentage to every lithium iron phosphate product.

For a simplified energy calculation, required stored energy equals the intended AC energy delivery divided by the discharge-path efficiency. Add a reserve only if the objective requires it, and avoid counting losses twice when a product’s usable-energy figure already uses the relevant boundary.

Then review charge and discharge power, starting loads, temperature limits, aging assumptions, and backup switching where applicable. Half-hourly averages can hide brief high demands. Equipment selection needs more than a daily kWh total.

Compare Several Storage Sizes

Run feasible capacity options against the same timeline. Record how often each fills, how often it reaches its reserve, and how much additional import it avoids. A larger battery can deliver diminishing additional value without being technically wrong.

Include installation cost and any changed inverter or protection requirements. Compare the incremental battery investment with incremental savings rather than dividing the combined system cost by only the battery’s benefit. Present the objective clearly when backup value motivates a choice that is not justified by bill savings alone.

Validating the Design with an Interval Replay

Replay the proposed system across the representative dataset and retain the assumptions. A dispatch model must respect energy balance, power limits, usable capacity, reserve, losses, and the chosen operating schedule.

There is no universal target percentage that proves a UK home is correctly sized. Explain why a particular configuration suits the customer’s needs and constraints. A high self-consumption percentage can result from a small array that leaves substantial demand unmet.

Use generation and financial modeling to explore documented scenarios where supported. SurgePV’s published scope does not specify native integrations with the named meter providers or a specific automated half-hourly battery-dispatch workflow; evaluate those requirements directly in a demonstration.

Illustrative Design Example: A Surrey Household

The following figures are retained as illustrative assumptions, not a real customer dataset or a verified PVGIS run. The original location details, claimed customer decision, named battery performance, and payback claims were unsupported and have been removed.

Assume annual household load of 5,820 kWh and three candidate array configurations. The table uses assumed direct-use and export totals to show the calculation. It is not evidence that any Surrey roof will produce these results.

Exports equal generation minus direct use in each row. The larger candidates produce more total energy in this example, but the increase in direct use is small. That observation alone cannot select the best array: the export value and incremental installed cost remain necessary inputs.

Check the Arithmetic Before the Recommendation

If an illustrative export price of £0.12/kWh is used, export income is £212.40, £313.80, and £424.80 respectively. These are calculations from assumed prices and energy, not current supplier offers. Replace the price with the customer’s proposed contract before making a recommendation.

Likewise, a proposed twelve-module layout using 415 W modules has 4.98 kWp of nameplate capacity, not exactly 5 kWp. Recalculate the model using the final equipment count and rating. Keep rounding visible so layout, proposal, and electrical schedules remain consistent.

A Battery Starting Point, Not a Product Selection

Suppose the design objective is to deliver an assumed 5.4 kWh during an evening period. If a hypothetical discharge-path efficiency is 90%, that delivery requires 6.0 kWh at the stored-energy boundary before any separate reserve. The efficiency is an example input, not a product specification.

Compare that requirement with the actual battery’s usable capacity and power, then replay the charging schedule. Do not round down to a nearby product and claim the original target is met without checking the result. If the customer accepts partial coverage, state that explicitly.

The example cannot demonstrate a payback because it lacks verified equipment prices, a full dispatch calculation, and a contract-specific tariff. It shows where the evidence must enter the decision. A real recommendation should preserve the same calculation trail with actual inputs.

GDPR, Authorization, and the Smart Energy Code

Treat identifiable household energy data as customer information requiring lawful, transparent handling. Follow the access provider’s consent process and obtain appropriate advice on your own data-protection responsibilities.

n3rgy’s consent guidance requires end-customer permission for its service and distinguishes provider-managed evidence from its trusted-consent route. That is a provider-specific access requirement, not proof that one generic signed form resolves every UK GDPR obligation.

The government’s data-protection overview states the principles of lawful use, specified purposes, data minimization, accuracy, appropriate security, and keeping data no longer than necessary. It does not prescribe the universal 90-day solar-design retention rule claimed by the previous article.

Define the Purpose and Access

Explain what data is needed, why it is used, who receives it, and how long it is retained under the applicable policy. Separate a one-off design from ongoing performance monitoring. Do not assume permission for design also covers unrelated marketing or sharing with other businesses.

Use an approved access and storage process. Keep API credentials out of proposal documents, shared spreadsheets, and ordinary project notes. Remove access when it is no longer needed under the agreed process and provider rules.

Set a Justified Retention Policy

Choose and document retention periods based on purpose and applicable obligations, with professional advice where needed. Aggregated data can still be personal data when linked to an identifiable customer. Replacing raw intervals with daily totals does not automatically remove data-protection responsibilities.

Record what is retained, where, by whom, and how deletion or restriction requests are handled. Include backup copies and exported files in the policy. A software promise of automatic deletion needs evidence about its scope and configuration before it appears in a customer notice.

The earlier story about an unnamed installer receiving an ICO settlement was not sourced. It should not be treated as a real enforcement case or as evidence for a particular retention period.

Common Data Quality Problems and How to Fix Them

Investigate anomalies against the provider’s format and project history before altering values. A normalization process should preserve evidence and make assumptions reviewable.

Negative or Unexpected Values

Check whether a value is an interval quantity, a signed net flow, a cumulative reading, or a provider correction. Do not clamp every negative number to zero. That can erase a real correction or hide a parsing error.

Compare affected periods with meter changes, export behavior, and the provider’s quality flags. If the cause remains unresolved, mark the affected analysis as provisional and assess whether excluding that period changes the design conclusion.

Meter and MPAN Matching

Match the supply identifier and meter serial to the customer’s records. A building may have multiple meters or an export identifier distinct from its import identifier. Verify the actual series rather than assuming a successful login means the correct site was downloaded.

Keep the date range for each meter when equipment changes. Prevent overlapping datasets from double-counting energy, and check for gaps between old and new records. Reconcile totals with bills over matching dates where possible.

Units and Time Resolution

Distinguish W, kW, Wh, and kWh. Power requires a duration to become energy; energy should not be multiplied by duration again. A unit conversion error can dominate every later sizing result while leaving the chart visually plausible.

Check whether a timestamp marks the beginning or end of the interval. Align generation, consumption, and tariffs consistently. Save the conversion method with the normalized data so another reviewer can reproduce it.

Smart Meter Data for Small Business Solar

Use the same measurement discipline for business premises, but model the actual operating schedule and tariff. Do not assume every office, pub, shop, or workshop has a high daytime load or a fixed self-consumption rate.

Ask about opening hours, refrigeration, shutdowns, seasonal trade, and planned equipment. Compare workdays with weekends and closed periods. A large annual consumption total can still coexist with substantial midday export during closures.

Identify the metering and data-access arrangement with the supplier or authorized provider. Do not use a universal annual-kWh threshold to decide which meter technology must be present. Access permissions and available data can differ between sites.

Where the tariff includes demand-related charges, evaluate the actual measurement and billing rules. Half-hourly energy can support some analyses, but it does not prove the maximum instantaneous load or replace the tariff’s specified demand interval. Keep energy and power savings separate.

Linking Meter Data to UK Compliance Workflows

Interval data supports the energy and financial case; it does not replace electrical, network, or installation requirements. Use the relevant network operator and certification instructions for the project.

For a connection application, provide the required equipment and export-control information. A load profile may help explain an operating assumption, but it is not permission to exceed an approved export limit. Check the standard for any technical threshold rather than deriving it from this article.

For an export arrangement, follow the supplier’s metering and application requirements. For tax treatment, obtain the appropriate advice from the project facts. A household load profile alone cannot determine VAT eligibility or certify a mixed-use property’s tax allocation.

Keep commissioning and as-built records consistent with the model. If the installed equipment or operating schedule changes, revisit the customer forecast rather than leaving the original savings estimate untouched.

Market-Wide Half-Hourly Settlement

Market-wide Half-Hourly Settlement, or MHHS, changes electricity settlement arrangements; it does not make every design forecast a billing record. The government’s July 2026 update states that full meter migration remained scheduled for May 2027.

The previous claim that all suppliers would complete migration by the end of 2026 was therefore removed. The article also no longer predicts a separate future move to another interval without a supporting source.

For installers, retain the practical distinction between measured history, supplier settlement, and a prospective solar simulation. A forecast can use interval data well and still differ from later bills because weather, occupancy, equipment operation, or tariff terms change.

When Historical Data Needs a Separate Scenario

Historical readings are less representative when occupancy or equipment will change. Keep the evidence, but do not pretend it already includes future behavior.

For a new build, use an explicitly modeled load profile with stated appliance and occupancy assumptions. For a tenancy change, ask what is known about the new usage pattern. For an EV or heat pump addition, add a separately documented load model and schedule.

Avoid replacing the entire historical baseline with a generic annual uplift. Where possible, model the additional load in the periods it is expected to operate. Compare uncertainty cases so the customer understands which decision depends on future behavior.

Recommended Workflow and Next Steps

Start by agreeing the design objective and appropriate data access. Verify the site and meter, retrieve the available period, and preserve the raw file. Normalize units and timestamps while retaining quality flags and a record of corrections.

Build the load profile at the correct measurement boundary. Compare feasible PV layouts, then storage options if they serve the customer’s objective. Replay the assumptions consistently, reconcile energy totals, and value imports and exports using documented tariff inputs.

Present measured facts separately from future-load assumptions and modeled results. Keep the customer proposal clear about data coverage, missing information, and the conditions behind any savings estimate. Assign engineering approval and data-protection responsibilities to the appropriate people.

Bring those requirements to a SurgePV demonstration to evaluate its published design and financial-modeling capabilities. Ask for evidence of any specific meter-data import or dispatch function your workflow requires before including it in the purchasing decision.

Frequently Asked Questions

Why use smart meter data for solar design?

Interval data shows when the site imports and exports electricity. Comparing those periods with modeled PV helps assess direct use, storage operation, and tariff value, provided the measurement boundary is understood.

How much smart meter history is needed for solar sizing?

A representative full year is preferable when available, but there is no universal minimum that guarantees accuracy. State the available period, missing data, and any future occupancy or equipment changes.

Does Bright require a CAD for historical meter data?

Hildebrand’s documentation describes delayed DCC data after verification without requiring Glow hardware. Hardware and MQTT access are separate routes with their own setup requirements.

Can import data alone size solar on an existing PV home?

Import data alone generally cannot reconstruct total load on an existing PV or battery site. Obtain the other relevant energy flows and align their measurement boundaries before modeling an upgrade.

Must installers delete smart meter data after 90 days?

This article found no verified universal 90-day solar-design rule. Apply a justified retention policy under the relevant data-protection obligations and provider permissions, with qualified advice where needed.

Can evening consumption alone determine battery payback?

No, battery value also depends on charging energy, power limits, losses, cost, dispatch, and tariffs. Compare complete operating scenarios rather than applying one evening-kWh threshold.