Open energy consumption and smart meter datasets
For household-level smart meter data, UK Power Networks' Low Carbon London dataset and the UCI Individual Household Electric Power Consumption dataset are the most accessible starting points; for appliance-level disaggregation, UK-DALE and REFIT are the standard research benchmarks; and for grid-scale generation, demand and pricing, the US EIA API and Europe's Open Power System Data / ENTSO-E Transparency Platform are the reference sources. Below is what each one contains, at what resolution, and how to get access.
The short answer
Pick by scale and resolution: whole-household smart meter readings (Low Carbon London, UCI household power), appliance-level disaggregation for non-intrusive load monitoring research (UK-DALE, REFIT), circuit-level residential data with solar and EV charging (Pecan Street Dataport, gated access), and grid-scale generation/demand/price data (EIA API for the US, Open Power System Data and ENTSO-E for Europe). Most of the sources below are free with no more than a basic account; Pecan Street is the one exception requiring institutional verification.
Sources, one by one
1. Low Carbon London smart meter data (UK Power Networks)
Published on the London Datastore, this dataset covers half-hourly electricity consumption readings from 5,567 London households collected during the UK Power Networks Low Carbon London project, November 2011 to February 2014 — around 167 million rows when unzipped. A subset of roughly 1,100 households was exposed to dynamic time-of-use pricing in 2013, with day-ahead high/low/normal price signals, which makes this one of the few open datasets that lets you study demand response alongside raw consumption. A free London Datastore account is required to download; there is no cost and no institutional-affiliation requirement.
2. UCI Individual Household Electric Power Consumption
The UCI dataset records minute-level electricity consumption for a single household in Sceaux, France, over almost four years (December 2006-November 2010), including global active/reactive power, voltage, intensity and three sub-metered circuits. It is small, well documented and free with no login, which makes it the most common teaching dataset for time-series forecasting on energy data, though its single-household scope limits how far conclusions generalize.
3. UK-DALE (UK Domestic Appliance-Level Electricity)
UK-DALE is an open-access, appliance-level dataset originally from Imperial College London: one house recorded continuously for 655 days at a 16kHz whole-house sample rate plus 1/6Hz per-appliance readings, with data later extended to five houses. It is the highest-frequency open household energy dataset generally available and is the reference benchmark for non-intrusive load monitoring (NILM) research — inferring which appliance is running from the aggregate signal. It is free for research use; download links and documentation are on the dataset's own site and via the original Scientific Data journal publication.
4. REFIT Electrical Load Measurements
The REFIT dataset, from the Personalised Retrofit Decision Support Tools For UK Homes project, covers active power at an 8-second sample interval for 20 UK households recorded between 2013 and 2015, with per-appliance breakdowns for around nine circuits per home plus whole-house totals. Compared to UK-DALE, REFIT trades sampling frequency for more households and a longer, more consistent multi-home window, which makes it a common complement or cross-validation set alongside UK-DALE in NILM studies. It is published for free academic use through the University of Strathclyde's data repository.
5. Pecan Street Dataport
Dataport is described by Pecan Street as the largest available source of disaggregated residential energy data: circuit-level consumption (often at 15-minute or 1-minute intervals) covering whole-home use, solar generation, EV charging, HVAC and individual appliances, plus water-use data, from participating US and international homes. Unlike the other sources here, it is not a direct public download — access is granted to faculty, staff and students at accredited institutions who verify their affiliation, and other applicants may receive limited access at Pecan Street's discretion. It is free, but the vetting step means you should plan for an approval delay before you can pull data.
6. EIA Open Data (US Energy Information Administration)
The EIA API is the US government's official source for time-series energy data across electricity, petroleum, natural gas, coal, nuclear and international categories, including state- and grid-region-level consumption and generation. It is free; a free API key is required for the API itself, though bulk downloads are available without one. It is the right source for US macro-level questions — total consumption by state or fuel type, generation mix trends — rather than individual household behavior.
7. Open Power System Data and ENTSO-E Transparency Platform (Europe)
Open Power System Data (OPSD) is a free platform that collects, cleans and republishes European electricity system data — generation, load, prices, capacity — sourced in part from the ENTSO-E Transparency Platform, the mandatory EU regulatory data feed for transmission-system operators. Going to ENTSO-E directly gives you the authoritative source, but its interface and download process are widely reported as slow and inconsistent across countries; OPSD's processed, harmonized files are usually the faster path for research use if you do not need the raw regulatory feed. Both are free with no payment required; ENTSO-E requires a free account for full API access.
Sources side by side
| Dataset | Best for | Resolution / scale | Access |
|---|---|---|---|
| Low Carbon London | Household smart meter + time-of-use pricing | Half-hourly, 5,567 households | Free London Datastore account |
| UCI Household Power | Teaching, single-household forecasting | Minute-level, 1 household, ~4 years | Free, no login |
| UK-DALE | High-frequency appliance disaggregation (NILM) | 16kHz whole-house, up to 5 houses | Free, research use |
| REFIT | Multi-household NILM cross-validation | 8-second interval, 20 households | Free, academic use |
| Pecan Street Dataport | Circuit-level data with solar/EV | 15-min/1-min, many homes | Free, institutional verification required |
| EIA Open Data | US grid-scale generation and consumption | State/regional time series | Free API key; bulk download free |
| Open Power System Data / ENTSO-E | EU grid-scale generation, load, prices | Country/zone time series | Free; ENTSO-E account for full API |
How to choose
Decide first whether your question is about individual behavior or the grid as a whole. Household- and appliance-level questions — forecasting one home's usage, detecting which appliance is running, studying demand response to price signals — belong with Low Carbon London, UCI, UK-DALE, REFIT or Pecan Street, roughly in order of how much sampling frequency and appliance granularity you need versus how many households you need to cover. Grid-scale questions — regional demand trends, generation mix, wholesale price dynamics — belong with EIA (US) or Open Power System Data / ENTSO-E (Europe). If your work is specifically about non-intrusive load monitoring, use UK-DALE and REFIT together: UK-DALE gives you the highest sample rate for method development, REFIT gives you more households to validate generalization.
Before you commit, check three things: the sample rate matches your task (16kHz whole-house data and 30-minute smart-meter readings answer very different questions), whether the dataset requires institutional verification that could add delay (Pecan Street) versus immediate download (most of the others), and whether the years covered are still representative — appliance mixes, EV adoption and tariff structures have all shifted meaningfully since some of these datasets were collected in the early-to-mid 2010s.
Frequently asked questions
What is the difference between UK-DALE and REFIT?
Both are UK appliance-level (non-intrusive load monitoring) datasets, but UK-DALE covers fewer households (originally one home for 655 days, later extended to five) at a very high 16kHz whole-house sample rate, while REFIT covers 20 households at a coarser 8-second interval over 2013-2015. UK-DALE suits high-frequency disaggregation research; REFIT suits longer-horizon, multi-household appliance studies.
Do I need to register to use Low Carbon London smart meter data?
You need a free account on the London Datastore to download it, but there is no cost and no research-affiliation requirement. The dataset covers half-hourly readings from 5,567 London households between November 2011 and February 2014, including a subset exposed to dynamic time-of-use pricing.
Is Pecan Street Dataport free to use?
Access is free but gated: university-affiliated researchers can register with proof of their institutional affiliation to get full access, and other applicants may be granted limited access at Pecan Street's discretion. It is not an open-download dataset in the same sense as UCI or the EIA API.
Can't find the resolution or region you need? Leave it to us
If none of the sources above match your required sample rate, region or metering type, you can give us your task goal and the conditions it must meet, and we run a data availability assessment (3 free credits on sign-up) first — searching reputable data platforms for real and judging matches and gaps against each required item. Even when no perfectly fitting dataset is found, we present the search directions, approximate sources and item-by-item findings honestly for your reference.
New accounts get 3 free searches; every result comes with verifiable source links.