A robust modeling framework for energy analysis of data centers

Abstract

Global digitalization has given birth to the explosion of digital services in approximately every sector of contemporary life. Applications of artificial intelligence, blockchain technologies, and internet of things are promising to accelerate digitalization further. As a consequence, the number of data centers, which provide the services of data processing, storage, and communication services, is also increasing rapidly. Because data centers are energyintensive with significant and growing electricity demand, an energy model of data centers with temporal, spatial, and predictive analysis capability is critical for guiding industry and governmental authorities for making technology investment decisions. However, current models fail to provide consistent and high dimensional energy analysis for data centers due to severe data gaps. This can be further attributed to the lack of the modeling capabilities for energy analysis of data center components including IT equipment and data center cooling and power provisioning infrastructure in current energy models. In this research, a technology-based modeling framework, in hybrid with a data-driven approach, is proposed to address the knowledge gaps in current data center energy models. The research aims to provide policy makers and data center energy analysts with comprehensive understanding of data center energy use and efficiency opportunities and a better understanding of macro-level data center energy demand and energy saving potentials, in addition to the technological barriers for adopting energy efficiency measures.

Geospatial assessment of water footprints for hyperscale data centers in the United States

“…This analysis applies and extends the thermodynamics-based models developed in references [5,7,9] to characterize the PUE and WUE Site of hyperscale DCs in the U.S. Three commonly implemented cooling system configurations in hyperscale DCs were considered: (1) airside economizers integrated with direct evaporative cooling, using air-cooled chiller as the supplemental mechanical cooling system; (2) airside economizers integrated with direct evaporative cooling, using water-cooled chillers as the supplemental mechanical cooling system; and (3) waterside economizers utilizing the indirect evaporative cooling capability of cooling towers, using water-cooled chiller as the supplemental mechanical cooling system. For each cooling system configuration, the physics-based model can reliably predict PUE and WUE Site under uncertainty by utilizing DC facility system variables and climate data (including dry bulb temperature, relative humidity, and atmospheric pressure) [5,7] However, the model's computational demands constrain the efficient exploration of the geospatial distribution of PUE and WUE Site for hyperscale DCs in the U.S. under various scenarios of climate and facility system operations.…”

Climate- and Technology-Specific PUE and WUE Predictions for U.S. Data Centers using a Physics-Based Approach

“…As a result, the electricity use of DCs has become a more frequent topic of research within the energy analysis community (Hintemann and Hinterholzer, 2019;Shehabi et al, 2018). DC electricity use can generally be divided into two primary categories: the electricity use of information technology (IT) equipment (i.e., servers, storage devices, and network devices) and the electricity use of power provision and cooling (i.e., infrastructure) equipment (Lei, 2020;Masanet et al, 2020a;Shehabi et al, 2016). Many energy analysts model infrastructure equipment energy use by assuming an average power usage effectiveness (PUE) value, where PUE is defined as the dimensionless ratio of a DC's total electricity use (in kilowatt-hours, or kWh) to its IT electricity use (in kWh) (Jaureguialzo, 2011).…”