Carbon emission modeling: an assessment of potential changes in energy demand for future HPC
Year: 2025
Type: article
Abstract: High-Powered Computing (HPC) systems are indispensable for modern technological development, yet their environmental impact demands critical attention. We address the environmental impact of HPC systems, focusing on their energy-intensive operations and associated carbon emissions, water usage, e-waste, and dependence on fossil fuels. The study aims to develop models and strategies to quantify and mitigate these impacts by integrating renewable energy and improving efficiency. A carbon emission model based on energy consumption and diverse energy combinations is proposed alongside an Energy Portfolio Matrix to assess future HPC energy demand and potential changes in energy structure. Through strategic modeling, we simulate scenarios with renewable energy proportions ranging from 30% to 100%, demonstrating significant emission reductions as renewable energy adoption increases. In evaluating resource optimization, liquid and air cooling systems are considered more sustainable than traditional methods. Sensitivity analysis using SALib reveals that the proportion of renewable energy is the primary factor influencing emissions, while growth rates and emission factors have a lesser impact. On the technical side, the study emphasizes energy-efficient hardware, advanced cooling systems, renewable energy integration, and optimization algorithms. Policy recommendations include implementing carbon taxes, energy subsidies, and green standards to foster sustainability. While the model effectively considers various environmental factors, it remains static and constrained by assumptions. That underscores the need for continuous innovation and effort in mitigating the environmental footprint of HPC systems.
Authors: XiaoLeng Wang, Ronglang Li, Qinyuan Lu, Jianan Wu
Language: English
Topic: Green IT and Sustainability
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4409238636