In a significant advance for sustainable construction, researchers have combined machine learning and probabilistic life cycle assessment (PLCA) to accurately predict the carbon emissions of magnesium silicate hydrate cement (MSHC), offering a powerful new tool for evaluating its potential as a low-carbon alternative to ordinary Portland cement (OPC).
The teams' findings were recently published in Smart Construction.
Study: Assessing the low-carbon potential of magnesium silicate hydrate cement: a probabilistic life cycle approach. Image Credit: New Africa/Shutterstock.com
Why MSHC?
MSHC has drawn attention as an environmentally friendlier option to OPC, which is responsible for a substantial share of global carbon dioxide (CO2) emissions. Made from lightly calcined magnesium oxide, siliceous materials, water-reducing agents, and water, MSHC boasts low pH, high strength, thermal resistance, and corrosion resistance—all while potentially reducing emissions.
Unlike OPC, which requires high calcination temperatures, MSHC uses lightly calcined MgO, significantly cutting energy use and carbon output. The incorporation of industrial by-products such as silica fume and fly ash can further reduce its environmental impact. However, until now, there's been limited quantitative data assessing MSHC's emissions across its full life cycle.
Bridging the Knowledge Gap
To fill this gap, the research team applied a PLCA framework to measure both direct and indirect emissions throughout MSHC's life, from raw material extraction and transport to production. Emissions were expressed in CO2 equivalents per kilogram and benchmarked against OPC.
They compiled emission factor data from a broad set of global sources covering 2005 to 2024. Thirteen MSHC mix designs from published literature were analyzed using the Monte Carlo method to assess emissions variability, capturing the uncertainty inherent in material sourcing and production practices.
To better understand which factors drive emission variability, the researchers performed statistical distribution analysis and visualized the data. This foundation enabled the application of four machine learning models—linear regression, decision tree, random forest, and extreme gradient boosting—to predict carbon intensity based on mix composition. Feature importance analysis highlighted the most influential variables, and the results were integrated into two interactive graphical user interfaces (GUIs) for fast analysis.
What the Models Reveal
The findings underscore a critical insight: MSHC’s carbon emissions are highly dependent on its magnesium-to-silicon (Mg/Si) ratio. Mixes with a ratio below 0.667 consistently showed low emissions. Between 0.8 and 1.0, MSHC emissions closely matched those of OPC, while ratios above 1.0 often exceeded OPC levels.
L-MgO emerged as the primary emissions contributor across all mix ratios, with its variability tied to differences in purity, production processes, and usage levels. When the Mg/Si ratio was low, emissions from siliceous materials and transport also carried high uncertainty.
All ML models performed well, with R2 values exceeding 0.95. The random forest model proved to be the most accurate and consistent, confirming the Mg/Si ratio as the most decisive variable influencing emissions. Using this model, the team simulated 390 hypothetical mixes with Mg/Si values ranging from 0.1 to 4.0. Results showed a clear trend: emissions rise steadily as the Mg/Si ratio increases.
Why it Matters
This research provides the clearest picture yet of how MSHC performs environmentally, not just in theory, but across real-world variables and production conditions. While MSHC has the potential to significantly cut emissions, it’s not inherently low-carbon. Its impact hinges on how it’s formulated, particularly in relation to the Mg/Si ratio.
The two GUIs developed through this study offer practical tools for engineers, designers, and researchers to assess and optimize MSHC mixes for lower carbon impact. The methodology also sets a new benchmark for evaluating alternative building materials using a combination of life cycle modeling and machine learning.
Journal Reference
Li, Y., Luo, X., Liu, X., Zhang, Z., Meng, K., & Mu, J. (2025). Assessing the low-carbon potential of magnesium silicate hydrate cement: a probabilistic life cycle approach. Smart Construction. DOI: 10.55092/sc20250012, https://www.elspublishing.com/papers/j/1894665248839471104.html
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.