Energy and Exergy-Aware Workload Assignment for Air-Cooled Data Centers

The energy required to cool a data center (DC) depends on both thermal and workload management. Although existing energy- and temperature-aware workload assignment approaches reduce operational expenditure by minimizing cooling energy consumption, they do not address the loss of available cooling capacity (i.e., loss of thermodynamic exergy) due to unavoidable system inefficiencies. By quantifying the exergy efficiency, the actual cooling COP and maximum achievable COP during operation can be embedded in the DC control system. A multi-objective optimization problem is formulated that quantifies the tradeoff between energy and exergy efficiencies. By reducing the exergy loss, the available cooling capacity, and therefore the capital expenditure can be effectively utilized, potentially reducing the total cost of ownership. We identify scenarios when significant improvements in the utilization of cooling capacity are possible with only small increases in energy consumption.