Scientific papers

Research paperPDF

arXiv preprint (cs.AI), Harvard T.H. Chan School of Public Health · 2026

Assessing the Carbon Emissions and Energy Consumption of U.S. Hyperscale Data Centers

In fleet or facility energy estimation, use capacity × 8,760 × facility-load coefficient (0.48-0.70, central 0.58) and report the range (67.7-98.6 TWh) explicitly.

• State-level intensities (central scenario): Virginia 535, Oregon 440, Iowa 645, Nebraska 740, Wyoming 1035 gCO2/kWh; Virginia had 142 HDCs, 21.0 TWh, 11.5 Mt.

EnergySustainabilitySite planning

Read / Download

Research paperPDF

Springer Science and Business Media LLC (preprint) · 2026

Battery reserve autonomy and diesel generator mission reliability

When presenting UPS battery autonomy as a way to improve generator reliability, compute the reserve threshold against the mission duration and report that autonomy of minutes adds little to long-outage generator reliability.

• For a 72 h mission with Fehr parameters: 0.95617 with one generator, 0.95847 with an ideal 4 h battery reserve, 0.99808 with two independent generators.

GeneratorsReliabilityUPS

Read / Download

Research paperPDF

CIGRE US National Committee, 2025 Grid of the Future Symposium (arXiv preprint) · 2025

Best Practices for Large Load Interconnections: A North American Perspective on Data Centers

In a large-load interconnection application, verify the relevant entity's large-load threshold (10-75 MW) and ramp limit (Southern ≤20 MW/min) early; definitions differ widely between entities.

• Southern Company is the only entity in this review with a numeric ramp cap: ≤20 MW/min under normal operation.

ElectricalReliabilitySite planning

Read / Download

Research paperPDF

arXiv preprint (cs.CY; formatted for ACM e-Energy '26) · 2026

Beyond PUE: A Local-Impact Audit Framework for Data Center Environmental Accountability

In siting and community-acceptance files, present PUE, WUE and CUE together with regional water stress, grid headroom and verification maturity (L0-L4); back project commitments with contractual limits and independent metering (L2-L3).

• Compliance gate: any unresolved major regulatory violation caps the composite score at a ceiling of 40 (suggested, subject to calibration).

Performance metricsWaterSite planning

Read / Download

Research paperPDF

arXiv preprint (cs.SE), Google · 2024

Carbon accounting in the Cloud: a methodology for allocating emissions across data center users

In allocating carbon to data-centre users, give idle power to whoever holds the resource allocation and share dynamic power by actual usage; because an idle server draws about 45% of peak power, the idle share is large.

• Resources with equivalent busy power used to weight idle-power allocation: 1 Intel Skylake N1 CPU core = 20 GiB RAM = 1 TiB SSD = 6 TiB HDD.

SustainabilityEnergyPerformance metrics

Read / Download

Research paperFree to read

Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2 (ACM) · 2023

Carbon Explorer: A Holistic Framework for Designing Carbon Aware Datacenters

Add an hourly 24/7 CFE coverage calculator to site and energy design that sizes wind, solar, BESS (in hours of IT load) and server headroom together; use hourly or sub-hourly supply profiles, never daily averages.

• Grid lifecycle intensities (gCO2eq/kWh): wind 11, nuclear 12, hydro 24, solar 41, biofuels/other 230, natural gas 490, oil 650, coal 820.

SustainabilityEnergySite planning

Research paperPDF

IEEE Transactions on Power Systems (Institute of Electrical and Electronics Engineers (IEEE)) · 2021

Carbon-Aware Computing for Datacenters

Consider shifting temporally flexible workloads (completable within 24 hours) to hours of lower carbon intensity and lower power peak as an option in data-centre grid-connection negotiation and power capacity planning.

• A piecewise-linear model of power versus CPU usage achieves daily MAPE below 5% for more than 95% of power domains.

EnergySustainabilityOperations

Read / Download

Research paperPDF

Journal of Engineering Research and Sciences (Journal of Engineering Research and Sciences) · 2026

CFD Analysis of Data Center Hall Cooling Performance under Normal and Failure Modes with Control Strategies and Airflow Leakages

Do thermal verification of failure mode (two ACUs offline) with realistic air leakage (cabinet gaps, HAC panel joints); a leak-free model can comply while the leaky model exceeds 32 °C.

• The grid study selected 7,288,454 cells with the standard k-epsilon model and wall functions.

AirflowCoolingReliability

Read / Download

Research paperPDF

Energies (MDPI AG) · 2022

CFD Modeling of Pressure Drop through an OCP Server for Data Center Applications

When modelling a server as a porous block in room-scale CFD, use Darcy-Forchheimer coefficients fitted to a resolved pressure-drop curve rather than generic assumptions such as porosity 0.5.

• Twenty simulations were run for flow rates of 0.003 to 0.06 m³/s, and steady-state results matched time-averaged transient results.

AirflowRacks

Read / Download

Research paperPDF

Resources, Conservation and Recycling (Elsevier BV) · 2022

Climate- and technology-specific PUE and WUE estimations for U.S. data centers using a hybrid statistical and thermodynamics-based approach

In early data-centre design, set PUE and WUE targets by size class, cooling system and climate zone using this study's ranges (for example 1.71-2.22 for small rooms, 1.12-1.25 for large scale).

• Airside economisers with adiabatic cooling (case 1) use less water than airside economisers alone (case 3).

CoolingWaterEnergy

Read / Download

Research paperPDF

IEEE Open Journal of Industry Applications (Institute of Electrical and Electronics Engineers (IEEE)) · 2020

Comparative Analysis of Experimental DC Arc Flash Results to Industry Estimation Methods

Keep bus gap as an input in arc flash calculations for DC panels and battery busbars; do not rely on the maximum-power method alone.

• Tests covered about 130 VDC, bolted fault currents of 1.9-27.5 kA and bus gaps of 12.7-63.5 mm.

ElectricalUPS

Read / Download

Research paperPDF

CIGRE US National Committee, 2026 Grid of the Future Symposium (arXiv preprint) · 2026

Comparative Analysis of Large Load Interconnection Requirements Across U.S. Transmission Entities

Design data-centre UPS and load control to the most restrictive entity's requirement: 6.0 s ride-through at 0.70-0.90 pu (CAISO), six fault-clearing attempts within 90 s (SPP), six 50-70 ms reclosings (Dominion), and the ban on constant-power control (SPP).

• MISO's proposal requires withstanding three voltage disturbances within 10 s and phase-angle jumps up to 25 degrees.

ElectricalReliabilitySite planning

Read / Download

Research paperPDF

Energies (MDPI AG) · 2019

Comparing Performance Metrics of Partial Aisle Containments in Hard Floor and Raised Floor Data Centers Using CFD

In airflow assessment do not rely on RCI alone; RCI of 100% coexisted with RTI of 40-50%, showing oversupply and bypass. Report RTI and per-rack Capture Index together.

• With supply cut by 50%, RTI rose above 80%, and recirculation appeared in all containment variants while RCI still looked acceptable.

AirflowPerformance metricsCooling

Read / Download

Research paperPDF

Energies (MDPI AG) · 2023

Conjugate Heat Transfer Modeling of a Cold Plate Design for Hybrid-Cooled Data Centers

In hybrid-cooled racks, represent a cold-plate server as a compact model in room-scale CFD: model the share of CPU heat going to the cold plate (~97%) and the residual escaping to air separately.

• A cited micro data centre with 1.2 kW IT capacity reached a maximum outlet water temperature of approximately 50 °C.

Liquid coolingAirflowRacks

Read / Download

Research paperCitation

Energy (Elsevier BV) · 2020

Cooling architecture selection for air-cooled Data Centers by minimizing exergy destruction

When comparing room, row and rack cooling architectures, report chiller, airspace and heat-exchanger losses separately.

• About 33% of overall data centre energy consumption is attributed to cooling infrastructure.

CoolingAirflowEnergy

Research paperPDF

Entropy (MDPI AG) · 2019

Cooling Effectiveness of a Data Center Room under Overhead Airflow via Entropy Generation Assessment in Transient Scenarios

Use in the design rationale that aisle containment cut air-side entropy generation to negligible levels (0.15% and 0.03%), against 34% for an uncontained legacy room.

• With containment, air-side entropy generation was only 0.15% (downward flow) and 0.03% (upward flow) of the total, against an earlier estimate of 34% for an uncontained legacy room.

AirflowCooling

Read / Download

Research paperPDF

Energies (MDPI AG) · 2023

Cooling Technologies for Internet Data Center in China: Principle, Energy Efficiency, and Applications

In China, define project design targets from GB 50174-2017 temperature and humidity limits (18-27 °C, dew point 5.5-15 °C, RH at most 60%) and provincial PUE caps (for example Guangdong 1.25).

• Cooling accounts for about 35-40% of total IDC energy consumption, the second largest consumer after IT equipment.

CoolingEnergyLiquid cooling

Read / Download

Research paperPDF

IEEE Transactions on Cloud Computing (Institute of Electrical and Electronics Engineers (IEEE)) · 2026

Coordinated Cooling and Compute Management for AI Datacenters

Design AI-hall cooling control together with workload management: when GPU count and tensor parallelism, DVFS and supply air temperature and airflow were optimised jointly, cooling energy fell by 31.2%.

• Cooling is cited as contributing approximately 45-50% of total energy consumption in AI data centres.

CoolingOperationsEnergy

Read / Download

Research paperPDF

IEEE Transactions on Components, Packaging and Manufacturing Technology (Institute of Electrical and Electronics Engineers (IEEE)) · 2021

CPU Overclocking: A Performance Assessment of Air, Cold Plates, and Two-Phase Immersion Cooling

For high-heat-flux chips, compare the point at which air cooling throttles (here 3.3-4 GHz) with the clock-rate gain from cold plates or 2P immersion (41-51%) in cooling technology selection.

• Compared with air cooling, cold plates gave up to 41% and 2P immersion up to 51% higher CPU clock rates.

Liquid coolingCoolingRacks

Read / Download

Research paperPDF

Frontiers in Energy Research (Frontiers Media SA) · 2022

Current Status and Challenges for Liquid-Cooled Data Centers

Design the CDU as redundant in liquid cooling and arrange the circulating pump as one duty plus one standby; a CDU failure can paralyse the whole system.

• Cooling systems are cited as consuming 30-50% of total data-centre energy.

Liquid coolingReliabilityCooling

Read / Download

Research paperPDF

arXiv (University of Michigan preprint) · 2026

Data Center Cooling Choices Shift Water Impacts Across the Grid: An Integrated Water-Energy Model for Sustainable Data Center Development

In cooling selection, calculate direct and indirect water together: an air-cooled chiller cut direct water 97% but raised the indirect share to 96.6% and electricity by 32%.

• Indirect water intensity varied by about 50% across months, depending on which generators supplied the added load.

WaterCoolingEnergy

Read / Download

Research paperPDF

Buildings (MDPI AG) · 2024

Data Center Energy Evaluation Tool Development and Analysis of Power Usage Effectiveness with Different Economizer Types in Various Climate Zones

Choose economiser type by climate zone: consider indirect economisers in 0A/0B and wet zones (PUE 0.03-0.07 lower) and adding evaporative cooling in dry zones (0.02-0.05).

• More than 95% of a data centre's total cooling load comes from heat generated by IT equipment.

CoolingEnergyPerformance metrics

Read / Download

Research paperFree to read

arXiv preprint (University College Dublin) · 2025

Data Center Model for Transient Stability Analysis of Power Systems

Expose UPS transfer thresholds (voltage %, frequency Hz), return-to-mains delay and staggering across UPS modules as design parameters, and warn when all modules share identical thresholds and delays, since synchronised reconnection can re-trip the grid.

• Cooling accounts for about 30% of total power of standalone data centers and is modelled as an induction motor.

ElectricalUPSReliability

Research paperPDF

npj Clean Water (Springer Science and Business Media LLC) · 2021

Data centre water consumption

In cooling selection, add the roughly 25.5 million litres per year per MW of evaporative cooling water need and the potable share (57-65%) to the siting and water supply decision.

• Raising chilled-water temperature from the usual 7-10 °C to 18-20 °C can cut expenses by 40%.

WaterCoolingPerformance metrics

Read / Download

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.