Context
The client wants to reduce energy consumption. The factory permanently maintains a temperature of 21°C in the clean room, all year round. Optimization work begins with the boiler room.
Challenge
Predict heating needs based on changing external conditions. This modeling step will determine whether boiler room adjustment actions improve energy consumption.
Technologies used
DBMS: SQL Server 2008R2 & 2016 • Python 3.6 • Cloud AWS • Pandas, Jupyter, Scikit-Learn, XGBoost
Methodology and implementation
Energy optimization project through predictive modeling:
**Data collection** - Temperature sensors (indoor/outdoor) - Gas consumption measurements - Weather data - Building occupation - Equipment status
**Modeling** 1. Creation of a building energy model (baseline) 2. Thermal inertia analysis 3. Heating needs prediction 24h in advance 4. Setpoint temperature optimization based on weather forecasts
**Optimization** - Reduction of setpoint temperature during favorable periods - Anticipation of cold peaks - Exploitation of building thermal inertia
**Measuring gains** Comparison of actual consumption vs reference model over 12 months
Results
Construction of an energy model of buildings, optimization of the setpoint temperature
Visualizations
Boiler efficiency forecast
Energy optimization analysis