Efficient Operation of Indirect Evaporative Data Center Cooling Systems Via Newton-Like Extremum-Seeking Control

Abstract

Nowadays the data center industry is playing a leading role in the world economic development and it is growing rapidly and constantly. On the other hand, this sector consumes a significant amount of energy and it contributes to Global Greenhouse Gas Emissions (GHGE). For these reasons, there are concerted and significant efforts to efficiently operate data centers. In particular, efficient cooling for data centers is essential for reducing operation costs because cooling equipment is energy intensive. In this context, we consider the problem of real-time efficient operation of an environmentally friendly cooling technology. In particular, we use a sinusoidally perturbed Newton-Like Extremum-Seeking Control (ESC) scheme, which benefits of the input-output plant map curvature estimate, to optimal operation of an indirect evaporative data center cooling system in a model-free fashion. Finally, by using a Matlab-based computer room air conditioning simulation environment, the effectiveness and performance of the proposed ESC scheme are compared in silico to those of standard ESC algorithm.