Predictive Analytics
The Challenge
The demand for computing capacity is rising every year and so is the energy consumption of data centers.
A big part of this energy is transformed into heat, which can be reused to heat apartments and offices.
To make this as efficient as possible and to provide the necessary amount of heat, the demands for heating and computing need to be forecasted and synchronized.
The Solution
Smart Load Shifting for Data-centers
1. A self-learning forecast algorithm predicts the heat demand for locations that are connected in the system to exchange heat with data centers.
2. An optimization algorithm derives a schedule for the computing job execution that adapts location- and timeslots according to the predicted demands and other parameters, like job priority or energy efficiency of the data center.
3. The load shifter executes computing jobs according to the derived schedule, maximizing the reuse of the data center's heat.
Benefits
- Maximized energy efficiency of data centers
- Reduced costs for heating and computing
- Flexible and self-learning approach that can be extended for multiple data centers and locations
Further Use Cases
- Energy Optimization: Shift a lot of energy-intensive processes to the use of renewables
- Smart Buildings: Control heating, cooling, and EV-charging based on the building's usage and available resources
- Transportation: Derive smart schedules for public transportation systems.
... and many more
