AI shortens time taken to measure the sustainability impact of a product

Choices made in the earliest stages of product design can influence the environmental impact of a product for years to come. From the materials used to the manufacturing method, these decisions can lock in consequences that ripple across supply chains and ecosystems. Yet, the very tool intended to guide such decisions, life cycle assessment (LCA), is often out of reach for the people who need it most.

 

LCA offers a detailed picture of a product’s environmental footprint, from raw material extraction, through use, and eventual disposal. In practice, it demands months of data gathering, expert knowledge, and considerable costs. It is therefore not surprising that many small and medium-sized enterprises, and even larger companies working on fast product cycles, cannot realistically apply it during design. By the time a conventional assessment is complete, it is often too late to change course.

 

“Product designers face many challenges: difficulty in assessing the impact of different materials because of a lack of reliable data, limited leverage with supply chains to obtain information, and incomplete understanding of energy consumption. Without clear guidelines, they often end up making choices in the dark,” explained Associate Professor Arlindo Silva from the Singapore University of Technology and Design (SUTD).

 

To address this, Assoc Prof Silva and his team developed a Streamlined Life Cycle Assessment (SLCA) approach, described in their research paper, “Bridging the gap: streamlining life cycle assessment for practical application in product development.” The method combines artificial intelligence (AI), 3D modelling, and secondary databases to cut through the complexity of traditional assessments while keeping results trustworthy.

 

Rather than starting from zero, the streamlined approach draws on prior studies and databases to identify the components most likely to drive the environmental impact of a product. These major contributors are then modelled in 3D to automatically extract their weight and volume. AI systems assist by assigning typical manufacturing processes and selecting appropriate data from repositories, such as Ecoinvent. The resulting assessment requires far fewer inputs, can be completed in a fraction of the time, and still provides a reliable picture of environmental hotspots.

 

“SLCA builds on prior knowledge to understand what matters most, instead of demanding every last detail. It uses 3D modelling to derive basic part characteristics and AI to match them with the most likely processes and materials,” added Assoc Prof Silva.

 

To validate the method, the team tested it on a case study of a small electronic hearing aid. A traditional full LCA of the device took three months and required 86 separate data inputs. By contrast, the SLCA took one week and used only 26 inputs, cutting input requirements by nearly 70 percent and time by over 90 percent. The streamlined results matched the full assessment with an average accuracy of 90 percent.

 

According to Assoc Prof Silva, this balance is key: “We ensured that the full LCA served as our ‘ground truth’. What we found is that a huge saving in time spent leads to only a minimal deviation in results—beyond a certain point, more effort does not translate into much greater accuracy.”

 

With SLCA, designers could test alternative concepts rapidly and iteratively, identifying which materials or processes are most environmentally burdensome before committing to them. Industries where products evolve quickly, such as consumer electronics or wearables, could benefit most immediately, while other sectors may adapt the method to their own contexts.

 

“Our approach is especially suited for early-stage design, where uncertainty is high. It enables teams to spot hotspots without waiting for every specification to be finalised, avoiding surprises later when a full LCA shows the impact is higher than intended,” explained Assoc Prof Silva.

 

Looking ahead, the research team plans to extend testing to more product types and refine the approach to make it more user-friendly. They also see opportunities to explore how AI might continue to evolve in this space, balancing automation with transparency. Ultimately, the goal is to make environmental impact assessment part of routine design practice rather than a mere afterthought.

 

“Right now, LCA is extremely difficult to integrate at the design stage—it is usually done when it is too late to do something about it,” said Assoc Prof Silva. “We hope this work contributes to embedding sustainability into design from the very start, where it can make the biggest difference.”

---

---