Carbon negative scheduling of direct air capture and hydrogen production in PV powered AI data centers under uncertainty

The rapid growth of artificial intelligence (AI) computing infrastructure, exemplified by OpenAI’s data centers, has significantly intensified electricity demand and carbon emissions. To mitigate this environmental impact, this paper proposes a novel photovoltaic (PV)-powered microgrid system that integrates Direct Air Capture (DAC) and hydrogen production to simultaneously achieve carbon removal and renewable energy storage. Unlike conventional microgrid scheduling frameworks, this study develops a multi-objective Distributionally Robust Optimization (DRO) model that addresses uncertain renewable generation, fluctuating electricity prices, and volatile hydrogen and carbon credit markets. The optimization simultaneously minimizes total operational costs, maximizes CO \(_2\) capture efficiency, and ensures hydrogen production aligns with both storage targets and market conditions. The uncertainty modeling is handled by DRO, capturing worst-case deviations within an ambiguity set derived from historical data, ensuring operational resilience. Furthermore, the Non-dominated Sorting Genetic Algorithm III (NSGA-III) is employed to derive Pareto-optimal trade-offs, providing comprehensive decision support across financial, environmental, and technical dimensions. The proposed optimization framework is applied to a realistic case study modeled on OpenAI’s energy infrastructure. A 100 MW PV farm supplies renewable energy, supported by a 50 MWh battery system and a 60 MW electrolyzer for hydrogen production. The DAC system operates within a 3.5 MW baseline demand, scaling dynamically with surplus solar power. Historical data from NREL, CAISO, EU-ETS, and California hydrogen markets inform the realistic energy prices, carbon credits, and hydrogen values. The optimization is implemented using Pyomo, solved with Gurobi 10.0, and combined with NSGA-III via Platypus. Sensitivity analysis, including Monte Carlo simulations with 10,000 trials, evaluates system resilience under varying solar irradiance, market price volatility, and operational uncertainties. The results demonstrate that PV generation directly governs DAC efficiency and hydrogen production scalability, with clear synchronization between high solar availability and optimal carbon capture performance. Pareto-optimal solutions reveal that increasing CO \(_2\) removal efficiency from 65% to 85% requires a 27% increase in operational costs, highlighting a trade-off between environmental and economic objectives. Furthermore, hydrogen storage utilization peaks at 85% when carbon credit prices surpass $90/ton, indicating strong market-driven operational shifts. Grid dependency reduces by 46% in scenarios prioritizing CO \(_2\) capture, underscoring the system’s ability to enhance renewable self-sufficiency.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-12
DOI
https://doi.org/10.1038/s41598-026-70059-7
Primary Topic
Hybrid Renewable Energy Systems
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Carbon negative scheduling of direct air capture and hydrogen production in PV powered AI data centers under uncertainty

Zhanzhi Liu, Zhen Yang, Shijian Zhang, Linhao Ye
Scientific Reports
Hybrid Renewable Energy Systems
article

Carbon negative scheduling of direct air capture and hydrogen production in PV powered AI data centers under uncertainty

Zhanzhi Liu, Zhen Yang, Shijian Zhang, Linhao Ye
article en

Abstract

The rapid growth of artificial intelligence (AI) computing infrastructure, exemplified by OpenAI’s data centers, has significantly intensified electricity demand and carbon emissions. To mitigate this environmental impact, this paper proposes a novel photovoltaic (PV)-powered microgrid system that integrates Direct Air Capture (DAC) and hydrogen production to simultaneously achieve carbon removal and renewable energy storage. Unlike conventional microgrid scheduling frameworks, this study develops a multi-objective Distributionally Robust Optimization (DRO) model that addresses uncertain renewable generation, fluctuating electricity prices, and volatile hydrogen and carbon credit markets. The optimization simultaneously minimizes total operational costs, maximizes CO \(_2\) capture efficiency, and ensures hydrogen production aligns with both storage targets and market conditions. The uncertainty modeling is handled by DRO, capturing worst-case deviations within an ambiguity set derived from historical data, ensuring operational resilience. Furthermore, the Non-dominated Sorting Genetic Algorithm III (NSGA-III) is employed to derive Pareto-optimal trade-offs, providing comprehensive decision support across financial, environmental, and technical dimensions. The proposed optimization framework is applied to a realistic case study modeled on OpenAI’s energy infrastructure. A 100 MW PV farm supplies renewable energy, supported by a 50 MWh battery system and a 60 MW electrolyzer for hydrogen production. The DAC system operates within a 3.5 MW baseline demand, scaling dynamically with surplus solar power. Historical data from NREL, CAISO, EU-ETS, and California hydrogen markets inform the realistic energy prices, carbon credits, and hydrogen values. The optimization is implemented using Pyomo, solved with Gurobi 10.0, and combined with NSGA-III via Platypus. Sensitivity analysis, including Monte Carlo simulations with 10,000 trials, evaluates system resilience under varying solar irradiance, market price volatility, and operational uncertainties. The results demonstrate that PV generation directly governs DAC efficiency and hydrogen production scalability, with clear synchronization between high solar availability and optimal carbon capture performance. Pareto-optimal solutions reveal that increasing CO \(_2\) removal efficiency from 65% to 85% requires a 27% increase in operational costs, highlighting a trade-off between environmental and economic objectives. Furthermore, hydrogen storage utilization peaks at 85% when carbon credit prices surpass $90/ton, indicating strong market-driven operational shifts. Grid dependency reduces by 46% in scenarios prioritizing CO \(_2\) capture, underscoring the system’s ability to enhance renewable self-sufficiency.

Scientific Reports
China Energy Engineering Group Guangdong Electric Power Design Institute Co., Ltd. (China) (CN), China Southern Power Grid (China) (CN)
Affordable and clean energy, Climate action, Industry, innovation and infrastructure
Hybrid Renewable Energy Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon