SoSo CAD

Topic guide

Energy

Discover publications on data centre energy consumption, power demand and efficiency. Investigate measurement methods, operating conditions and improvement approaches through summaries and original references.

Sources
416
Downloadable PDFs
122
Years
2007–2026
Search and filter in the library

All sources

193–216 / 416

Research paperPDF

arXiv preprint (math.OC), MIT Operations Research Center 2025

Locational Marginal Emissions for Carbon-Aware Data Center Operations in Large-Scale Power Grids

In data-centre carbon accounting, consider hourly locational marginal emissions (LME) rather than an annual-average grid factor; the LME-based intensity (about 672 kgCO2/MWh) was about 3 times the average factor (202).

  • Emission rates assumed in the simulated grid: coal 800 kgCO2/MWh, gas 450 kgCO2/MWh, wind/solar/nuclear/hydro 0.
SustainabilitySite planningEnergy
Read / Download
White paperFree to read

Schneider Electric 2012

Making Large UPS Systems More Efficient

Store UPS losses as a three-coefficient model (no-load, proportional, square-law) or a 0/25/50/75/100% curve, and record whether quoted efficiency includes input filters or transformers.

  • Cooling adds 0.4 kW per kW of loss.
UPSEnergy
Research paperPDF

arXiv preprint (cs.SE); Skyworkz / Sustainable Digital Infrastructure Alliance (SDIA) 2023

Measuring IT Carbon Footprint: What is the Current Status Actually?

Compute IT carbon as (Es + Eo) × I + M and allocate embodied carbon (M) for shared resources by time and resource share; define scope, idle-time and shared-resource attribution explicitly.

  • Operational carbon = PUE × sum over intervals of grid-mix intensity × energy, constrained by energy-mix data granularity (for example 30-minute updates); average versus marginal emission factors are discussed.
SustainabilityPerformance metricsEnergy
Read / Download
Research paperPDF

IEEE Transactions on Sustainable Computing (Institute of Electrical and Electronics Engineers (IEEE)) 2017

Metrics for Sustainable Data Centers

In data-centre comparisons, do not use PUE as a cross-facility metric; normalise for IT load, weather and location, or state the same conditions.

  • Tier availability figures quoted: Tier I 99.671%, Tier II 99.741%, Tier III 99.982%, Tier IV 99.995%.
Performance metricsEnergyAirflow
Read / Download
Research paperPDF

Frontiers in Environmental Science (Frontiers Media SA) 2026

Net sustainability assessment of AI-driven data centers in the GCC: an MCDA approach framework

In AI data-centre investment assessment in hot, arid regions, weigh carbon, water, social and governance criteria together with explicit weights (0.40/0.25/0.20/0.15); indirect water (desalination energy) was about 0.68 times direct cooling water.

  • In the extended boundary, embodied carbon adds 15-25% to the operational footprint.
SustainabilityWaterEnergy
Read / Download
Research paperPDF

Challenges (MDPI AG) 2015

On Global Electricity Usage of Communication Technology: Trends to 2030

In data-centre electricity-demand projections use traffic-based extrapolation (traffic × annually improving energy intensity) as an upper-bound reference; the worst case is very high and the method was later criticised for overestimating demand.

  • In the worst-case scenario communication technology could use as much as 51% of global electricity in 2030 and contribute up to 23% of global GHG emissions.
EnergySustainability
Read / Download
Research paperPDF

Energy Strategy Reviews (Elsevier BV) 2024

On the means, costs, and system-level impacts of 24/7 carbon-free energy procurement

In a data-centre renewable procurement strategy, weigh that annual volumetric matching does not deliver hourly clean supply (54% to 83%) and that 85-95% hourly CFE targets cut emissions more at a relatively small premium.

  • Adding long-duration storage (salt-cavern hydrogen, 2.5 EUR/kWh) cuts the 100% CFE premium to 42% above annual matching; also adding clean firm generation cuts it to 15%.
SustainabilityEnergySite planning
Read / Download
Technical guideFree to read

Consulting-Specifying Engineer 2008

Outdoor air in data centers

In the tool's economizer studies, report hours below the humidity floor alongside energy savings, per climate.

  • NEBS levels 1-3 per SR-3580; level 3 applies to central-office equipment; key documents GR-63-CORE and GR-1089-CORE.
CoolingEnergySite planning
Research paperPDF

Buildings (MDPI AG) 2024

Performance and Applicability Analysis of Indirect Evaporative Cooling Units in Data Centers Across Various Humidity Regions

Assess an IEC unit by climate: natural cooling covered 97% of the year in a dry climate and 43% in a humid one; check the wet-bulb distribution (19 °C threshold) against project climate data.

  • Against packaged air-cooled DX units, IEC energy savings were 57.46% (Urumqi), 48.26% (Beijing) and 39.41% (Shanghai).
CoolingWaterEnergy
Read / Download
Research paperFree to read

arXiv preprint (Microsoft Azure) 2023

POLCA: Power Oversubscription in LLM Cloud Providers

In the power planner, apply different diversity and oversubscription factors by workload: inference rows can be provisioned below sum-of-nameplate (for example peak about 80% of provisioned), while training rows should assume near-coincident peaks with only about 10% headroom.

  • POLCA allows 30% more servers in the same inference GPU cluster while maintaining SLOs.
ElectricalRacksOperations
Research paperPDF

Geothermal Energy (Springer Science and Business Media LLC) 2026

Potential of aquifer thermal energy storage (ATES) for data centres in Germany

Before linking data-centre waste heat to nearby building heating, check that site subsurface geology suits ATES and that the site is outside water-protection zones; only 13 of 19 sites did.

  • Annual waste heat available for storage was Eth = Qth × F_load × (1 − F_free) × (1 − F_down), with a conservative mean annual load factor of 0.6.
SustainabilitySite planningEnergy
Read / Download
Research paperFree to read

arXiv preprint (Microsoft / OpenAI / NVIDIA) 2025

Power Stabilization for AI Training Datacenters

For AI-training halls, carry a 'synchronous load' flag and capture utility time-domain (MW/s ramp up/down, dynamic range) and frequency-domain (critical band, spectral cap) limits as interconnection inputs.

  • GPU power smoothing with the floor at 90% of TDP gave a total energy overhead of 10.5% in simulation.
ElectricalRacksReliability
Research paperFree to read

arXiv preprint (Emerald AI / EPRI / National Grid / NVIDIA / Oracle; submitted to IEEE) 2026

Power-Flexible AI Data Centers: A New Paradigm for Grid-Responsive Compute

Add a 'flexible load' profile in the grid-connection module: committed reduction depth (%), response time (s), duration (h) and recovery, derived from the workload priority mix.

  • Achieved 100% compliance across 200+ distinct National Grid power events on a 130 kW cluster of NVIDIA Blackwell Ultra GPUs.
EnergyOperationsElectrical

All content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.

Monthly newsletter · free

Keep your reading list current

Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.