Situation
- The cooling system of server farms are designed for the maximum load case but the thermal load varies with installed server capacity and load distribution
Task
- To determine the energy-saving potential of a data-driven control approach
Our Contribution
- Modelling of a cooling operation cycle
- Correlation with operation and thermal boundaries to generate optimisation value for different meteorological profiles
- Historic data analysis shows unusual component break downs and life time limits
- Pattern detection to optimise computational centres annual energy consumption & performance
- Improvement of overall Power usage Effectiveness (PuE) and rise of free cooling shift to quantify the optimisation potential