Life cycle assessment and multicriteria decision making analysis of additive manufacturing processes towards optimal performance and sustainability
Abstract The pressing need for sustainable construction materials and processes has been driving research into the optimum environmental and economic efficiency of Additive Manufacturing (AM). Most models available for Life Cycle Assessment (LCA), however, do not capture the dynamism of real-time da...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-07-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-92025-5 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract The pressing need for sustainable construction materials and processes has been driving research into the optimum environmental and economic efficiency of Additive Manufacturing (AM). Most models available for Life Cycle Assessment (LCA), however, do not capture the dynamism of real-time data and the existing levels of uncertainty, and decision-making frameworks are not adaptive to evolving sets of criteria. In this paper, these described limitations are addressed through the introduction of an integrated approach that couples predictive Life Cycle Assessment (LCA) with Gaussian Process Regression (GPR), dynamic decision criteria weighting via Stochastic Forest for Multi-Criteria Decision Analysis (MCDA), and multi-objective optimization using Particle Swarm Optimization (PSO). In this study, GPR-based predictive LCA is conducted using historical and real-time environmental data for modeling impact categories of CO2 and energy use. This methodology makes estimates of not only the mean impact but also allows quantification of the uncertainties through confidence intervals and dynamic LCA. Stochastic Forest algorithm will enhance the traditional MCDA by weighting decision criteria like cost, environmental impact, and durability, in a more dynamic manner aligning to real-time manufacturing performance for better decision-making. Further, PSO will optimize material and process parameters to balance the multiple objectives of material strength, energy efficiency, and cost-effectiveness. In this way, this integrative novel approach of machine learning with bioinspired optimization contributes to the sustainability of AM. Experimental results prove that predictive accuracy can be achieved up to 85–90% by GPR, which reduces material wastage by 12%. By using Stochastic Forest, an improvement in decision accuracy can be attained to the extent of 15–20%, together with a cut in costs of about 10%. For its part, PSO optimizes design and manufacturing parameters for the materials, raising their efficiency by 10–15%, while the energy consumption goes down by 8–12%. The next framework is an important step toward integrating the steps reviewed for the development of sustainable Additive Manufacturing practices. This framework overcomes the present limitations of the LCA model by introducing dynamic predictive modeling using Gaussian Process Regression, real-time adaptive decision-making through Stochastic Forest, and multi-objective optimization through Particle Swarm Optimization. This integration of the techniques in the framework would help address the real-time data and uncertainties that are inherent for adaptive and sustainable solutions in additive manufacturing processes. |
|---|---|
| ISSN: | 2045-2322 |