Simulation-derived electricity consumption dataset for standardized TOKİ mass housing across 11 provinces affected by the 6 February 2023 earthquakes in Türkiye
Description
This dataset contains simulation-derived energy, climate, and thermal performance variables for a standardized TOKİ mass housing typology located in provinces affected by the 6 February 2023 earthquakes in Türkiye. The data were generated using dynamic building energy simulations conducted in DesignBuilder with the EnergyPlus simulation engine. Simulations were performed on an annual basis using province-specific EPW climate files to capture regional climatic variations. The dataset includes electricity and gas consumption indicators, heating and cooling loads, thermal comfort parameters, environmental variables, and CO₂ emission-related outputs. All building geometry, envelope properties, and system configurations were kept constant across simulations to isolate the impact of climatic conditions on energy performance. The dataset is structured to support binary electricity consumption tier classification (LOW/HIGH) based on daily average electricity usage, in accordance with Türkiye’s tiered electricity tariff framework defined by the Energy Market Regulatory Authority (EPDK). Raw simulation outputs were organized and preprocessed through standardization and binary label transformation to facilitate reuse in data-driven analyses, including classification, feature selection, and dimensionality reduction studies. This dataset is intended to support research on residential energy consumption analysis, post-disaster housing assessment, energy efficiency evaluation, and machine learning–based modeling without requiring the repetition of computationally intensive building energy simulations.
Files
Steps to reproduce
Building model setup Create a standardized TOKİ mass housing building model in DesignBuilder. Use identical building geometry, envelope properties, occupancy schedules, and HVAC system definitions for all simulations to ensure consistency across locations. Climate data assignment Obtain province-specific EPW climate files corresponding to the provinces affected by the 6 February 2023 earthquakes in Türkiye. Assign the appropriate EPW file to each simulation scenario within DesignBuilder. Simulation configuration Configure the simulation to run on an annual basis (January 1–December 31) using the EnergyPlus simulation engine. Ensure that all output variables related to energy consumption, thermal comfort, environmental conditions, and emissions are enabled. Execution of simulations Run the simulations separately for each province while keeping all non-climatic parameters constant. Export the resulting simulation outputs for electricity consumption, gas usage, heating and cooling loads, indoor temperature metrics, and CO₂ emissions. Data compilation Compile the exported outputs into a structured tabular format (e.g., Excel or CSV). Each row should represent a simulation instance, and each column should correspond to a specific energy, climate, thermal, or environmental variable. Data preprocessing Convert annual or monthly outputs into daily average values where applicable. Apply Z-score standardization to continuous variables and perform binary label transformation to define electricity consumption tiers (LOW/HIGH) based on daily average electricity usage thresholds. Dataset validation Verify data completeness, consistency, and numerical integrity. Check for missing values, unit consistency, and correct alignment of feature labels prior to final dataset release. IMPORTANT Citation requirement: Users of this dataset must cite the associated research article published in Energy: Kavuran, G., Özer Yaman, G., Başarır, B., Doğan, E., İnce, B., & Dağteke, G. AI-supported decision framework for sustainable reconstruction: Case study on TOKİ housing after the 2023 Kahramanmaraş earthquake. Energy, Volume 344, 1 February 2026, Article 139891. This article provides the complete methodological framework, simulation configuration, electricity consumption tier labeling strategy, and analytical context underlying the generation and intended use of this dataset.
Institutions
- Malatya Turgut Ozal UniversitesiMalatya, Battalgazi