Preparing a large scale experimentation for causality analysis of textual data from social media influence on electrical consumption

Predicting electricity consumption represents a major methodological challenge. Integrating social media data could refine these estimates and increase their reliability. While previous work has established correlations between social media activity and electricity consumption, identifying causal effects between the prominence of specific topics on social media and energy use remains extremely difficult. This project aim to work on this challenge by analysing the causal impact of social media information.

BELL-Flex : Building Energy Living Lab (BELL-GreEn-ER) Energy Flexibility through forecasting

Internship Tasks:

• Study a real physical system (Building Energy Living Lab in GreEn-ER building)
• Gather existing historical data from InfluxDB and real time data from Schneider PLC
• Gather exogeneous information such as room occupation data
• Model HVAC systems and indoor comfort to create a forecasting model
• Participate to FlexReady pilot project

Analysis of Energy Flexibility and Indoor Air Quality of The GreEn-ER Building

This internship aims to analyse the impact of energy flexibility on indoor air quality, in response to modifications made to the default pressure set-points of air handling units(AHUs) in the GreEn-ER building in Grenoble, France. The building incorporates advanced HVAC systems, natural ventilation, and energy-recovery features, making it a suitable case study for energy flexibility research.

The experiment conducted for this internship (denoted neo-Flex) builds on key ideas from a previous experiment conducted in the same GEn-ER building (denoted CUBEFLEX), but introduces a new series of controlled interventions on selected AHUs (locally called CTAs). During the neo-Flex experiment, the supply and return pressure set-points of CTAs were reduced by 25% or 50% for four-hour periods on selected days when the building met our occupancy criteria.

Energy performance on these selected days was analysed by studying the electrical consumption data of the building, while indoor air quality and occupant comfort were analysed by studying the \(CO_2\) concentration data. Occupancy and window/door status were monitored manually to provide context.

The results show that reducing CTA pressure set-points can lower electrical consumption without appreciably degrading indoor air quality when the reduction is limited to 25%. At a 50% reduction of pressure setpoints of AHUs, some rooms experienced elevated \(CO_2\) levels but generally remained within acceptable occupant comfort limits. These findings confirm and extend the trends observed during CUBEFlex and illustrate the potential for controlled ventilation flexibility in tertiary buildings.

Contributions to the development of the Power Price Projection model for simulating electricity price trends in Europe

This internship was part of the development of the Power Price Projection (PPP) tool, which aims to create scenarios for electricity price trends in Europe until 2050 with hourly resolution. The main task was to build a model for forecasting hourly electricity demand, based on a bottom-up approach by usage and scenario. These demand profiles are used as inputs for the overall price simulation model.

In addition, the intern worked on several other developments such as: the integration of climate data, the collection and processing of data for Great Britain, the analysis of the Spanish market, and big data methods to optimize the management of large files. These various tasks helped to strengthen the robustness of the model and prepare for its deployment in new European countries.