AI-Powered Parametric Life Cycle Assessment & Eco-design

This four-day, hands-on workshop introduces an integrated workflow for environmental assessment and sustainable product design. Using a bio-based packaging case study, participants will build a baseline LCA model, explore manufacturing scenarios, identify influential parameters, and develop machine learning models to predict selected environmental impacts. Guided activities use openLCA, Google Colab, and KNIME Analytics Platform to connect environmental modeling with data analysis and design evaluation. Participants will compare alternative configurations, examine environmental trade-offs, and verify a preferred design through LCA calculations. Supplied inventories, datasets, and calculation models will support practical learning, with attention to data quality, uncertainty, and prediction limits.

To develop practical skills in combining Life Cycle Assessment (LCA), parametric modeling, and artificial intelligence to assess environmental impacts, identify improvement opportunities, and support informed ecodesign decisions.

• Understand the LCA framework, functional units, system boundaries, and inventory requirements.
• Build a baseline product-system model in openLCA and identify environmental hotspots.
• Convert inventory inputs into parameters with realistic ranges and manufacturing constraints.
• Generate alternative scenarios and assess how input changes affect environmental impacts.
• Apply sensitivity analysis to identify influential material, energy, and process variables.
• Train and evaluate a Random Forest regression model and explore a prepared XGBoost model.
• Interpret predictions using feature importance and guided SHAP analysis.
• Compare ecodesign options using consistent boundaries, environmental indicators, and explicit decision criteria.
• Verify selected design improvements through LCA while recognizing uncertainty and model limitations.

Day 1: LCA Foundations, Baseline Modeling & Environmental Hotspots

• Life Cycle Assessment (LCA) framework: goal and scope definition, life cycle inventory, impact assessment, and interpretation. Functional units and system boundaries: cradle-to-gate versus cradle-to-grave. Mapping material, energy, water, transportation, waste, and emission flows. Foreground and background data, inventory sources, and data quality. Introduction to allocation, recycling assumptions, and biogenic carbon. Software setup check and baseline product-system modeling in openLCA. Carbon footprint calculation and assessment of selected environmental impact categories. Contribution analysis and identification of environmental hotspots.
• Hands-On Activity: Build a Baseline LCA Model
• Model a selected bio-based packaging component using a supplied inventory and identify the processes and inputs contributing most to its environmental impacts.
• Tools: openLCA | Excel/CSV | Process-flow and inventory templates

Day 2: Parametric LCA, Manufacturing Scenarios & Sensitivity Analysis

• Converting fixed inventory inputs into variable parameters. Parameterizing material mass, recycled content, energy consumption, scrap rate, transport distance, and electricity mix. Defining realistic parameter ranges, linked inputs, and manufacturing constraints. Creating baseline and alternative manufacturing scenarios. Using a supplied Python calculation model and checking its consistency with the openLCA baseline. Understanding the differences between scenario analysis, sensitivity analysis, and uncertainty. Performing one-variable-at-a-time sensitivity analysis. Guided demonstration of Morris sensitivity screening. Preparing an LCA scenario dataset for machine learning.
• Hands-On Activity: Explore Scenarios & Identify Key Impact Drivers
• Vary selected manufacturing parameters, compare the resulting environmental impacts, and rank influential variables.
• Tools: openLCA | Google Colab

Day 3: AI-Based Environmental Impact Prediction & Interpretation

• Using LCA scenario results to develop a predictive model. Selecting input features and environmental impact targets. Preparing data, separating training and test datasets, and preventing data leakage. Training a Random Forest regression model. Guided comparison with a prepared XGBoost model. Cross-validation and model evaluation using R², mean absolute error (MAE), and root mean squared error (RMSE). Feature importance and guided SHAP interpretation. Distinguishing predictive associations from causal explanations. Checking selected predictions against LCA calculations and understanding prediction limits.
• Hands-On Activity: Predict & Explain Environmental Impacts
• Train and evaluate a Random Forest regression model using the supplied scenario dataset, then interpret its predictions and assess its limitations.
• Tools: KNIME Analytics Platform

Day 4: Ecodesign, Sustainability Improvement & Decision-Making

• Translating hotspot and sensitivity findings into ecodesign strategies. Evaluating material substitution, lightweighting, and recycled content. Exploring renewable energy use, process efficiency, and waste minimization. Waste valorization, circular design, and end-of-life scenarios. Comparing bio-based and conventional alternatives using equivalent functions and consistent system boundaries. AI-assisted screening of feasible design options. Assessing trade-offs across environmental impacts and manufacturing requirements. Guided uncertainty demonstration for selected assumptions. Building an ecodesign decision matrix with explicit criteria, weights, and constraints. Verifying the preferred scenario through LCA and preparing sustainability recommendations.
• Hands-On Activity: Redesign, Compare & Select a Preferred Scenario
• Compare baseline and redesigned options, quantify changes in environmental impacts, and recommend a preferred configuration using defined decision criteria.
• Tools: openLCA | KNIME Analytics Platform

• Develop abaseline LCA modelwith a documented inventory and defined system boundaries.
• Produce ahotspot analysisidentifying major environmental contributors.
• Prepare aparameter table and scenario datasetfor manufacturing alternatives.
• Generatesensitivity plotsand rank key environmental impact drivers.
• Build amachine learning workflowfor predicting selected LCA impact indicators.
• Evaluate model performance usingR², MAE, RMSE, and cross-validation.
• Interpret model outputs while distinguishing predictive associations from causal explanations.
• Prepare anecodesign comparison and decision matrixthat accounts for environmental trade-offs and manufacturing constraints.
• Present apreferred design recommendation, supported by LCA verification and clearly stated assumptions.

Master’s students, PhD scholars, and researchers in environmental science, engineering, materials science, biotechnology, chemistry, manufacturing, and related disciplines. Academicians and educators interested in practical LCA, environmental data analysis, and sustainable design workflows. LCA practitioners, sustainability analysts, and environmental consultants seeking to explore parametric modeling and machine learning. Product designers, packaging professionals, and manufacturing engineers working on material selection, resource efficiency, and environmental improvement. Industry and R&D professionals involved in sustainable products, circular economy initiatives, and process development.