Journal
APPLIED ENERGY
Volume 291, Issue -, Pages -Publisher
ELSEVIER SCI LTD
DOI: 10.1016/j.apenergy.2021.116798
Keywords
Data centers; Data center electricity needs; System dynamic forecasting; Moore?s law; Internet of Things
Categories
Ask authors/readers for more resources
This study presents a forecasting model for data center electricity needs based on understanding usage growth, concluding that efficiency gains from technological innovation cannot fully compensate for this growth. Simulations for the end of Moore's law and the growth of Industrial IoT were run, showing significant increases in electricity consumption. Sensitivity analysis revealed the vulnerability of data center electricity consumption to behavioral usage trends, highlighting the need for further research on the impact of usage on energy consumption.
This article presents a forecasting model of data center electricity needs based on understanding usage growth and we conclude that this growth is not fully compensated by efficiency gains of data center technological innovations. We predict a combined growth of data center electricity needs of 286 TWh in 2016 until about 321 TWh in 2030, if all currently known growth factors remain the same. We next run simulations for the end of Moore?s law and the growth of industrial Internet of Things (IoT). The end of Moore?s law results in about 658 TWh for 2030 and an increase of the share of global data center electricity consumption from about 1.15% in 2016 to 1.86% in 2030. A rise of the Industrial IoT may result into total energy consumption of about 364 TWh (about 1.03%) in 2030. Moore?s law and IoT combined cause data center energy needs going up to 752 TWh in 2030, and about 2.13% of global electricity available. Our sensitivity analysis reveals that the future impact of the data centers? electricity consumption is vulnerable to behavioral usage trends, since the 95% confidence interval of [343, 1031] TWh is relatively wide. Our forecasts, however, exclude the energy needs of mobile devices, edge and fog computing. We offer a system dynamic model and simulation input data selected from the existing literature for replicating this study and applying alternative parameters to it. We further suggest multiple research directions on usage impact on data center energy consumption.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available