ADRENALIN - Multi-National Time-Series Dataset of Sub-Metered Building Energy Use

Overview The ADRENALIN dataset offers valuable insights for a wide range of stakeholders, including researchers, energy analysts, building owners, and policymakers. Its diversity in building types, nations, and energy systems enhances the dataset’s applicability for studies in local energy system analysis. While it was originally developed for load disaggregation, it also supports broader research into energy flexibility and building energy use patterns. Applications The dataset supports a variety of use cases, including: Hourly energy load disaggregation. Forecasting of consumption and flexibility potential. Energy system modelling for both buildings and urban districts. Benchmarking algorithms. Evaluating demand response strategies. Developing intelligent control systems for HVAC and other energy-intensive operations. Dataset Structure Each ZIP file in the dataset corresponds to a specific building or location and contains structured subfolders for: Time-series energy data from main and sub-meters, recorded at various resolutions (e.g. 1 h, 30 min, 15 min, 5 min). Weather data aligned with the energy measurements, stored in separate subfolders. Raw data originally collected and reformatted from building management systems. Metadata files describing building characteristics (e.g., HVAC systems, floor area) and meter topology (e.g., relationships between main and sub-meters). File Naming Convention All data files are in .csv format and follow a consistent naming convention: adrenalin.<resolution>.<location-id>.<building-id>.<meter-id>.<quantity-key>.csv Example: adrenalin.15m.L01.B01.EM001.P.csv This file represents 15-minute resolution power data from the main electricity meter of building B01 in location L01. Data Format Each .csv file contains two columns: timestamp (in ISO 8601 UTC format) value (numerical measurement, e.g., power in kW) This structured and standardised format ensures ease of use for time-series analysis, machine learning, and integration into simulation or modelling workflows. Supporting Files Data keys for interpreting abbreviations of the measured quantities and their units are found in the file: ADRENALIN - Data keys.csv The PDF ADRENALIN_data_description.pdf provides a comprehensive description of the dataset, including background, data organization, preprocessing and harmonization procedures, metadata, and data quality assessment results. The dataset includes the Jupyter Notebook data_quality_assessment.ipynb which offers tools for assessing data completeness, consistency, and overall quality.