Assessment of catalyst deactivation in biodiesel production plants under real-world feedstock variability

Thesis Abstract

The study addresses the critical challenge of catalyst deactivation in biodiesel production plants arising from real-world variability in feedstock quality and composition, which undermines process efficiency, increases operating costs, and impedes sustainable scale-up. The aim is to quantify the mechanisms and rates of catalyst deactivation under representative feedstock scenarios and to identify mitigation strategies to enhance catalyst longevity and process resilience. Specific objectives include (1) to characterize feedstock variability from diverse locally sourced lipid/oil streams over a 12-month operational cycle; (2) to quantify catalyst aging parameters (sorption, coking, sintering, acid site poisoning) using in-situ and ex-situ diagnostics; (3) to establish relationships between feedstock properties (free fatty acid content, moisture, iodine value, phosphorus compounds) and deactivation rates via regression modelling; (4) to evaluate the effectiveness of regeneration protocols and process adjustments on restoring catalyst activity; and (5) to develop a predictive framework for remaining useful life (RUL) of catalysts under variable feed conditions. A mixed-methods approach is employed in a field-embedded setting within three commercial biodiesel reactors operating with heterogeneous feedstocks. The population comprises nickel- or base-metal-supported catalysts used in transesterification/hydrolysis stages, with a purposive sample of 9 catalysts across 3 plants over 18 months. Data collection integrates reactor performance records, feedstock characterization (gas chromatography for fatty acid profiles, Karl Fischer for water content, ICP-OES for trace metals, titration for acidity), and catalyst health monitoring through temperature-programmed oxidation (TPO), solid-state NMR, BET surface area analysis, and X-ray fluorescence (XRF) mapping. In-situ monitoring includes online reactor effluent analysis and online gas chromatography–mass spectrometry (GC-MS) for by-product formation, complemented by periodic offline analyses. Instrument validity is established through calibration standards, spiked recovery tests, and duplicate measurements, with data triangulated across time, plant, and feedstock type. Quantitative data will be analyzed using multivariate regression, partial least squares (PLS) regression to link feedstock properties with deactivation metrics, and survival analysis to model catalyst remaining life under varying operating conditions. ANOVA and Tukey post hoc tests will compare deactivation rates across feedstock categories, while time-series analysis will capture dynamic aging trends. Mechanistic interpretations will be supported by kinetic modelling of coke formation and poison deposition, with parameters estimated from TPO and spectroscopic data. Qualitative insights from process operator interviews will be analyzed through thematic analysis to identify practical factors affecting catalyst performance and regeneration feasibility, with theory-informing the interpretation via the Theory of Planned Behavior and the Resource-Based View of firm capability. Expected findings include quantification of the dominant deactivation pathways under realistic feed variability, notably carbonaceous deposition and acid-site poisoning linked to feed impurities, and the identification of threshold feed properties beyond which deactivation accelerates markedly. The study anticipates that regeneration effectiveness and optimal maintenance scheduling are contingent on precise feedstock profiling and catalyst type, yielding a predictive RUL model with acceptable predictive error (<15%) for maintenance planning. The contribution to knowledge lies in providing empirically grounded evidence on how feedstock heterogeneity shapes catalyst longevity in biodiesel processes, integrating material science diagnostics with process data, and offering a practical framework for predictive maintenance and process optimization under real-world variability. The main conclusion expected is that tailored catalyst selection, proactive feedstock management, and calibrated regeneration protocols can significantly reduce deactivation rates and extend catalyst life, improving overall plant productivity and sustainability. Recommendations include implementing routine feedstock screening with predefined acceptance criteria, adopting catalyst pre-treatment or protective coatings to mitigate poisoning, refining regeneration cycles based on data-driven thresholds, and developing an enterprise-wide maintenance decision-support system that integrates the proposed RUL model with economic risk assessment.