Zur Hauptnavigation wechseln Zur Suche wechseln Zum Hauptinhalt wechseln

Measuring the Efficacy of AI-Assisted Life Cycle Assessment: A Systematic Review and Empirical Case Study in the AEC Sector

  • Antonio Pantusa

    Studienabschlussarbeit: Masterarbeit

    Abstract

    Life Cycle Assessment (LCA) is the most rigorous methodology for quantifying the environmental impacts of products and processes, yet its practical uptake remains limited by the very rigor that makes it valuable, particularly the resource-intensive compilation of primary inventory data. As European regulations increasingly demand rapid, comparable environmental disclosure, Artificial Intelligence (AI) and Machine Learning (ML) have been proposed to overcome these bottlenecks. However, few of these studies benchmark a complete AI-assisted workflow against a conventional manual LCA on the same product system, and the practical benefits and limitations of commercial, off-theshelf AI-LCA software remain largely unquantified. This thesis addresses that gap through a two-phase investigation. The first phase, a Systematic Literature Review conducted according to the PRISMA protocol, identified 29 primary studies applying ML across the four LCA phases, showing that integration concentrates in the Inventory and Impact Assessment phases despite chronic data scarcity and a persistent lack of uncertainty quantification. The second phase translates these findings into an empirical case study in the Architecture, Engineering and Construction sector, in which the commercial platform One Click LCA was applied to a small residential building across three constructive systems of differing material composition and benchmarked against a manual LCA conducted with SimaPro© and Ecoinvent©. Significant deviations emerged, driven by geographic proxy distortions, database bias, and a semantic mapping limitation for non-standard materials; the platform's automated benchmarking further rated the concrete system above the timber system, contradicting their actual performance. Overall, AI-assisted LCA offers a clear advantage in speed, but not in accuracy or uncertainty treatment, and remains dependent on the maturity of background data. The thesis concludes that commercial AI-LCA platforms should be treated as calculation engines rather than autonomous decision-makers, proposing a Human-in-the-Loop paradigm and future research directions toward probabilistic uncertainty and open-source data for nonconventional materials.
    Datum der Bewilligung2026
    OriginalspracheEnglisch
    Gradverleihende Hochschule
    • FH OÖ
    Betreuer/-inPeter Hehenberger (Betreuer*in)

    Studiengang

    • Innovation and Product Management

    Zitieren

    '