Li Jia

With the rapid expansion of artificial intelligence infrastructure and cloud computing services, data centres are evolving from rigid electricity loads into flexible resources capable of contributing to renewable energy integration, grid regulation and cross-regional computing power allocation. Addressing the shortcomings in existing research regarding the differences between various types of computing tasks, the mechanisms of green migration under network constraints, and the characterisation of curtailment risks for renewable energy, this paper proposes a risk-aware collaborative optimisation and scheduling method for a power–computing–network–storage coupled system across remote data centres. Firstly, a hierarchical model of multi-type computing tasks is constructed, classifying data centre loads into fixed real-time tasks, online inference tasks, long-duration AI training tasks, and opportunistic elastic tasks, to characterise the differences between these tasks in terms of latency, time-shift, migration, and completion volume constraints. Secondly, a graph-attention-inspired green affinity prior is proposed, mapping grid topological distance, renewable energy availability, data centre PUE, and energy storage regulation capacity into interpretable migration signals, thereby guiding flexible computing power to migrate towards nodes with abundant green electricity and favourable grid support conditions. Subsequently, we introduce the CVaR metric to quantify the tail risk of renewable energy curtailment, establishing a multi-scenario stochastic linear optimisation model that incorporates DC power flow, unit output, renewable energy utilisation, campus energy storage, task SLAs, and cross-node migration constraints. A 24 h simulation based on the IEEE 10-machine, 39-node system demonstrates that the proposed method can reduce the expected curtailment volume from 176.939 MWh to 0 MWh, lower the CVaR curtailment risk from 694.085 MWh to 0 MWh, and increase the proportion of green computing power by 9.283 percentage points compared to the fixed-load baseline, whilst improving the five-tier collaborative score by 4.885 points.

The common root characteristic of CO₂ and air pollutants in the power industry, both derived from fossil fuel combustion, provides a natural basis for their synergistic emission reduction. However, existing studies suffer from the lack of a multi-pollutant synergistic evaluation system and an imperfect emission reduction technology database, which hinder their ability to support low-cost and high-efficiency emission reduction practices in the industry. Targeting the minimization of synergistic emission reduction costs and the maximization of emission reduction effects, this study integrated the process and economic parameters of 11 power generation technologies and 55 pollutant control technologies to establish a full-chain energy conservation and emission reduction technology database for the power industry, through literature research, industry surveys, and data mining. Based on the definition of pollution equivalent in the Environmental Protection Tax Law, we innovatively developed an air pollutant equivalent normalization evaluation method and constructed a two-dimensional coordinate system comprehensive evaluation system for CO₂ and air pollutants, enabling quantitative analysis and visual evaluation of the synergistic emission reduction effects of various technologies. The results show that new energy power generation technologies such as nuclear power and wind power, as well as O₂/CO₂ cycle combustion, ammonia-based desulfurization, and SNCR-SCR combined reduction technologies, exhibit excellent synergistic emission reduction performance for CO₂ and multiple pollutants. In contrast, some conventional pollutant control technologies, such as the limestone-gypsum method and traditional electrostatic precipitation, have significant CO₂ emission increase antagonistic effects. This study also completed the two-dimensional classification of 66 emission reduction technologies based on “emission reduction efficiency-economic cost”, identified application scenarios for different types of technologies, and proposed optimized paths for synergistic emission reduction adapted to the development of the power industry. The research findings fill the gap in quantitative standards for multi-pollutant synergistic emission reduction, provide theoretical support and detailed technical references for emission reduction technology selection and environmental policy formulation in the power industry, and help the industry achieve the dual development requirements of the “double carbon” goal and air quality improvement.

Heavy metals (HMs) pollution limits large-scale utilization of sewage sludge (SS)–high ash coal (HC) co-combustion as an efficient solid waste treatment strategy. This study employed ICP–MS, the BCR sequential extraction method, and time-resolved combustion experiments (0–540 s) to investigate the migration of Pb, Cd, Zn, Cu, Cr, and Ni during co-combustion. The results showed that synergistic interactions regulated HM migration through mineral-mediated pathways and clarified the dynamic coupling between HMs and targeted mineral and organic components. Co–combustion suppressed HMs volatilization and promoted conversion to more stable forms, with 50% SS blend at 900 °C achieving the highest stabilization efficiency, accompanied by a decline in ecological risk. HMs stabilization was attributed to the synergistic effects of physical encapsulation and chemical bonding: SS promoted high–temperature fuel fusion and the formation of a glassy phase that encapsulates HMs, while mineral components chemically adsorbed and stabilized HMs in bottom ash. CaO showed a three-stage effect on HM release, while FeS2 increased Pb enrichment by 22.96% by suppressing sulfate decomposition. Pb and Cd volatilized mainly at 0–60 s with KCl decomposition, whereas Cu and Zn peaked at 120–240 s, concurrent with Na, Mg, and fixed carbon combustion.

Blade icing in cold climates poses significant risks to operational stability and results in substantial power generation deficits. This study establishes and validates an integrated multiscale framework, CFD-OpenFAST-Stacking, to characterize the complex aeroelastic behavior of iced wind turbines and facilitate high-fidelity power forecasting. The methodology utilizes high-fidelity CFD to quantify the aerodynamic degradation of simulated iced airfoils. These data are subsequently coupled with the OpenFAST aeroelastic platform for full-scale turbine simulations to evaluate the system’s dynamic response. A Stacking ensemble learning model is developed by synthesizing these simulation results with historical SCADA data through an innovative data-fusion approach. Numerical findings indicate that icing severely compromises aerodynamic efficiency, inducing a 17.65% reduction in the maximum lift coefficient and a 34.07% escalation in drag at the aerodynamically sensitive blade tip. Consequently, the rated power point is shifted from 10.5 m/s to 13 m/s, with performance degradation most prominent in the low-to-medium wind speed regime. Model validation demonstrates that the data-fusion technique significantly improves predictive robustness, increasing the R² from 0.75 to 0.84 while reducing the RMSE from 37.69 to 17.04. SHAP analysis further identifies generator speed and wind speed as the primary determinants of power variability. This research substantiates the efficacy of bridging physical simulations with data-driven methodologies, providing a robust theoretical framework for performance evaluation in extreme weather environments.