Public defence in Industrial Engineering and Management, M.Sc. Zeinab Farshadfar

Title of the thesis: Machine learning applications in enhancing sustainable supply chains—a foresighted empirical study

Thesis defender: Zeinab Farshadfar
Opponent: Associate Professor Anna Aminoff, Hanken University
Custos: Assistant Professor Siavash Khajavi Haghighat, Aalto University School of Science

Can Machine Learning Make Supply Chains Truly Sustainable?

Supply chains are responsible for more than 80% of global greenhouse gas emissions and a major share of global resource consumption. At the same time, artificial intelligence and machine learning are rapidly transforming industries. But do these digital technologies genuinely make supply chains more sustainable — or do they mainly improve efficiency on paper?

This doctoral thesis examines how machine learning can enhance sustainable and circular supply chains in practice. The purpose of the study is to move beyond conceptual discussions and simulation models and provide real-world, empirical evidence of how machine learning affects economic and environmental sustainability.

The research combines a systematic literature review with pioneering case studies in two industries: construction waste management and aviation spare part supply chains. The study uses the triple bottom line framework — economic, environmental, and social sustainability — to analyse how machine learning applications contribute to sustainable supply chain management.

The findings show that most previous research has used machine learning mainly as a decision-support tool, for example in forecasting or supplier selection. In contrast, this dissertation demonstrates the impact of operational machine learning applications embedded directly in physical processes.

In the construction industry, machine learning–enabled robotic waste sorting increases recycling rates and can become economically competitive in high-wage environments. Sensitivity analyses reveal that labour costs, discount rates, and equipment investment are critical factors determining long-term competitiveness.

In the aviation industry, machine learning–enabled generative design and additive manufacturing reduce CO₂ emissions through lightweight components and enable more efficient spare part inventory pooling. The research identifies the spare part's weight and durability thresholds under which these solutions become both economically and environmentally viable.

By developing cost models and conducting scenario and sensitivity analyses, this thesis quantifies the sustainability impacts of machine learning more directly than previous studies. The results show that machine learning can function not only as a planning tool, but as an operational technology that reduces waste, lowers emissions, improves resource efficiency, and supports circular material flows.

Thesis available for public display 7 days prior to the defence at Aalto University's public display page.