The water-energy nexus in the United States: environmental footprints of cities and data centers

Abstract

Rapid changes in regional water cycle, and accelerated water use by the energy sectors have highlighted the need for holistic understanding of "Water-Energy nexus". Along with water consumption, high amount of green house gas (GHG) emission associated with energy generation serves as the polluting strength of any energy consumer. Environmental footprinting methods at sectoral or geographical level provide a means to relate the environmental externalities of electricity production to electricity consumers. Though several methods have been developed to connect the environmental footprint of electricity generation to end users, estimates produced by these methods are inherently uncertain due to the impossibility of actually tracing electricity from the point of generation to utilization. Previous studies rarely quantify this uncertainty, even though it may fundamentally alter their findings and recommendations. Here, we evaluate the sensitivity of water and carbon footprints estimates among seven commonly used methods to attribute electricity production to end users. We assess how sensitive water and carbon electricity footprint estimates are to attribution method, how these estimates change over time, and the main factors contributing to the variability between methods. We evaluate the water and carbon footprints of electricity consumption for every city across the contiguous United States for all assessed methods. We find significant but spatially heterogeneous variability in water and carbon footprint estimates across attribution methods. No method consistently overestimated or underestimated water and carbon footprints for every city. The variation between attribution methods suggest future studies need to consider how the method selected to attribute environmental impacts through the electrical grid may affect their findings.We have implemented the general understanding and findings of the water-energy nexus at the sectoral level for thorough investigation at industry scale. Spatial dependency of sectoral demands for limited environmental resources require adequate attention as the competition for ever shrinking resources are on the peak. Data centers comprised of computer systems and related components represent one of the largest and fastest growing energy users in the United States. Vigorous effort from the researchers have been able to restrict the energy requirement growth to 6% compared to a sixfold growth in workload and computing demand in the past decade. Predicted a more ferocious bloom in near future, comprehensive study on the environmental stress exerted at a higher spatial resolution by these data centers requires imminent attention. The quantitative analysis found that more than 500 million m3 water is consumed annually to support the operational stage of data centers with a high dependency on arid south western region. Geographical distribution of the servers coinciding with water stressed subbasins have almost tripled the water scarcity footprint (WSF) of the consumed water. Furthermore, present state of data centers is considerable source of greenhouse gas (GHG) emission, accounting for almost 0.6% of overall emission in the US. The results are validated by sensitivity analysis based on a set of electricity attribution approaches commonly found in existing literature. Finally, a comparative approach optimizes relative environmental stress of a hypothetical data center with no added technological innovation across the US subbasins to support decision making for future installation of server bases.