Dataset for study: ‘Efficiency in Building Energy Use: Pattern Discovery and Crisis Identification in Hot-Water Consumption data’
Description
This dataset contains time-series data on energy consumption for hot-water preparation from 10 residential apartment buildings in Kaunas, Lithuania. These buildings are connected to the city's central district heating system. The data collection period varies across buildings, with the longest spanning 10 years, from 2011 to 2021. Measurements are recorded hourly and include parameters such as inlet and outlet water temperature, temperature difference, flow rate, and power. Additionally pre-processed dataset is added, where power consumption (kWh) for DHW preparation is recalculated into DHW volume (m3) using synthetic time series data for cold-water temperature (CWT). Power consumption is also normalised by the useful area of each building and presented as a mean value. Additional features such as quarter, month, day of year, hour, minute, weekday, day, is weekend/holiday are added during feature engineering process. Dataset includes quarantine severity level for specific periods ranging form 0-5, determined based on open Google Mobility data (https://www.google.com/covid19/mobility/). Moreover, 3 periods are separated indicating pre-COVID, COVID during official lockdown in Lithuania and COVID while lockdown was not in-place.
Files
Steps to reproduce
The raw-data for 10 separate buildings data were collected from meters installed at the heating points of each building. In the pre-processed data, CWT is determined using moving average method based on the mean daily outdoor temperature, taking into account the inertia of the soil. The relationship between the CWT and the moving average outdoor temperature is based on the CWT measurements declared by the provider, where a lower limit of -4 ° C water temperature corresponds to an outdoor temperature of -20 ° C and an upper boundary of -16 ° C corresponds to +25 ° C. The cold water and outdoor air temperature measurements were used to apply a linear regression method. Power consumption is also normalised by the useful area of each building and presented as a mean value. Additional features such as quarter, month, day of year, hour, minute, weekday, day, is weekend/holiday are added during feature engineering process. Quarantine severity (during COVID-19 pandemic) determined based on open Google Mobility data (https://www.google.com/covid19/mobility/) taking retail/recreation and transit stations into consideration. Based on Google mobility data the severity is separated into 5 levels (5th being the highest) each representing 20% deviation from the base-line (normal conditions).