The environmental impact, carbon emissions and sustainability of computing in the ATLAS experiment

TL;DR

This paper tackles the environmental footprint of the ATLAS computing ecosystem amid the HL-LHC era, where resource needs are projected to surge by up to $3$–$4\times$ and beyond. It presents a multi-pronged approach—raising awareness, policy adjustments, and site-level optimizations—to reduce both direct (operational) and embodied (Scope 2/3) carbon emissions, including data-reproduction strategies, automated waste reduction, and infrastructure improvements. The study outlines concrete methods with quantified energy and carbon implications (e.g., $16\%$ replication break-even, HS23 benchmarking, and heat-reuse savings) and underscores the role of open data and knowledge sharing to broaden environmental benefits. Overall, the work demonstrates actionable pathways for ATLAS and related facilities to sustain scientific output while mitigating climate impact, with scalable relevance to future large-scale collider computing.

Abstract

ATLAS, a general-purpose experiment at the Large Hadron Collider (LHC), makes use of a large internationally-distributed computing infrastructure, including over $10^6$ TB of managed data on disk and tape and almost one million simultaneously running CPU cores. Upgrades for the High-Luminosity LHC (HL-LHC) will increase the required computing resources by a factor of 3-4 by the beginning of the 2030s, and by an order of magnitude before the conclusion of data taking at the beginning of the 2040s. These resources are spread over around 100 computing sites worldwide. Efforts are underway within the experiment to evaluate and mitigate various aspects of the environmental impact of the sites, with the additional long-term goal of making recommendations to the sites that will significantly reduce the total expected environmental impact in the HL-LHC era. These efforts take several forms: building awareness in the experiment community, adjusting aspects of the computing policy, and modifications of data center configurations, either in ways that take advantage of particular features of ATLAS workloads or in generic ways that reduce the environmental impact of the computing resources. This paper describes the ongoing investigations and approaches that have already provided useful and actionable outcomes.

Figures (10)

• Figure 1: The worldwide distribution of ATLAS computing, based on the amount of CPU provided, in HS23 (see Section \ref{['sec:HS23']} for the precise definition of HS23), on average in 2023--2024. Countries in gray did not contribute significant CPU.
• Figure 2: An example email of a job report from PanDA including carbon footprint information. The carbon footprint information is boxed at the bottom of the example email, and includes a link to a website providing more information about the calculations PanDACarbon.
• Figure 3: The power consumption of a typical processor during the running of the HEPScore benchmarking suite. Each point represents a power measurement. Seven different sets of three runs of individual workloads chosen to be representative and described in Ref. HEPScore are distinguished with dashed vertical lines.
• Figure 4: The evolution of HS23 score with CPU load for (a) AMD and (b) Intel CPUs. The gray points show data from the WLCG sites, while the colored series connected with lines indicate specific choices of frequency governor on benchmark machines. The data are from Ref. HEPScoreSustainability.
• Figure 5: An extrapolation of renewable power sources in Germany and power demand to 2030, based on Refs. German-PowerFraunhofer. The load (demand), on-shore and off-shore wind, and solar power production are scaled based on data from an example week at the end of May, 2023 and projections for consumption and renewable energy construction projects. Other sources include hydroelectric, biomass, geothermal, and waste-to-energy production.