A Framework for Predictive Life-Cycle Assessment of Catalytic Processes

Thesis Abstract

The rapid expansion of catalytic processes in chemical industries has heightened concerns about environmental impacts and resource sustainability, particularly in the life-cycle stages of feedstock extraction, synthesis, operation, and end-of-life disposal. Despite advances in process optimization and environmental assessment, there remains a gap in integrated, predictive frameworks that link catalyst design choices to holistically quantified life-cycle outcomes. This study proposes a framework for Predictive Life-Cycle Assessment (P-LCA) of catalytic processes, integrating material-, energy-, and emission-intensity dimensions with catalyst performance metrics to enable proactive decision-making during process development and scale-up. The aim is to develop a modular, data-driven framework that (i) forecasts cradle-to-grave environmental footprints of catalytic systems under varying operating conditions, (ii) quantifies trade-offs between catalytic efficiency, selectivity, and lifecycle emissions, and (iii) identifies design levers in catalysts and process conditions that minimize overall environmental impact without compromising economic viability. Specific objectives include (a) constructing a lifecycle database for representative heterogeneous and homogeneous catalytic systems across petrochemical and fine-chemicals sectors; (b) developing a hybrid modeling approach that couples first-principles reaction mechanisms with machine learning predictors for energy intensity and material use; (c) formulating a probabilistic uncertainty quantification scheme to address variability in feedstock quality, catalyst lifetime, and operating regimes; (d) validating the framework against benchmark processes with publicly available life-cycle data and internal industry datasets; and (e) delivering a decision-support tool that outputs comparative environmental performance, sensitivity analyses, and scenario-based optimization. The methodology adopts a sequential, mixed-methods design. The population comprises industrial catalytic processes from refinery, petrochemical, and specialty chemical facilities, with a purposive sample of 20 representative processes spanning oxide, supported metal, zeolite, and homogeneous catalysts. Data collection instruments include (i) process simulation files (Aspen HYSYS/PRO/UniSim) to extract energy and material balances, (ii) lifecycle inventory data collated from Ecoinvent, GaBi, and company reports, and (iii) experimental catalyst performance metrics gathered from pilot-scale testing and literature for parameterization. The study employs (i) first-principles kinetic modeling for reaction networks, (ii) machine learning regression and gradient-boosting methods to predict energy intensity and material consumption from catalyst descriptors, (iii) Monte Carlo simulations to propagate uncertainties, and (iv) scenario analysis to assess policy and market variations. Validation uses cross-validation for ML components, chi-square and goodness-of-fit tests for model alignment, and comparison with ISO 14040/44 compliant LCA results where available. The conceptual framework integrates a hybrid life-cycle model with a catalytic performance module, framed by a systems-thinking theory of industrial ecology and supported by the Theory of Planned Behavior to account for adoption of sustainable catalyst designs in industry. Expected findings include (i) robust relationships between catalyst lifetime, regeneration frequency, and cradle-to-gate emissions, (ii) quantifiable links between catalyst selectivity and overall process energy footprints, (iii) probabilistic estimates of environmental intensities under feedstock variability, and (iv) a transparent, user-ready decision-support interface that highlights the most effective design levers for reducing life-cycle impacts. The framework is anticipated to reveal non-linear interactions between catalyst design variables and lifecycle outcomes, underscoring the importance of integrated assessment during the early stages of process development. Contribution to knowledge encompasses (i) a novel, modular P-LCA framework tailored to catalytic processes that bridges detailed catalyst data with lifecycle-level environmental metrics, (ii) methodological integration of mechanistic kinetics with data-driven predictors within a lifecycle context, and (iii) a validated toolset for industry practitioners and policymakers to inform sustainable catalyst selection, process design, and investment decisions. The study concludes that predictive lifecycle insights can alter conventional optimization emphasis from solely isolated process efficiency to encompassing end-to-end sustainability, and recommends embedding the framework in corporate R&D workflows, expanding the dataset with real-time plant data, and extending the approach to include economic and social lifecycle dimensions for a comprehensive sustainable engineering assessment.