Adm1 modeling and application for anaerobic conversion of fruit and vegetable waste

Abstract

Although the Anaerobic Digestion Model No. 1 (ADM1) is widely used to simulate anaerobic digestion (AD) processes, its application to fruit and vegetable waste (FVW) remains limited when substrate characteristics, reactor configuration, and alternative metabolic routes are considered. In this context, this doctoral research aimed to evaluate, calibrate, and structurally extend ADM1 for representing different anaerobic conversion pathways of FVW through an integrated experimental and computational approach. The substrate consisted of collected FVW subjected to pretreatment, generating two fractions: a solid fraction (S-FVW) and a liquid fraction (L-FVW). The latter was applied in the experimental and modeling stages of wet anaerobic digestion (W-AD) and dark fermentation (DF). The thesis was developed in three complementary stages. First, a bibliometric analysis focused on ADM1 applications in dry anaerobic digestion (D-AD) identified a clear predominance of studies dedicated to W-AD, revealing that high-solids systems still remain comparatively underexplored and often require specific kinetic adaptations, particularly in hydrolysis and disintegration processes. Second, ADM1 was applied, adjusted, and calibrated for W-AD of L-FVW in Upflow Anaerobic Sludge Blanket (UASB) reactors. The results demonstrated that adequate representation of this substrate under UASB operational conditions requires substrate-dependent adjustment of hydrolysis, disintegration, and microbial uptake parameters, leading to high predictive accuracy for biogas production and intermediate process dynamics, with coefficients of determination reaching 0.99. Third, for DF of L-FVW in an Anaerobic Structured-Bed Reactor (AnSTBR) with attached biomass, the standard ADM1 structure proved insufficient to represent metabolic pathways associated with hydrogen production under acidogenic conditions. To address this limitation, additional pathways related to lactate formation and consumption were incorporated into the model structure, together with kinetic adjustments adapted to reactor operation and substrate characteristics. The modified model significantly improved predictive performance for hydrogen yield, gas composition, volatile fatty acids (VFA), and biogas production, with coefficients of determination above 0.90. The main original contribution of this thesis lies in demonstrating that ADM1 can be progressively adapted beyond its conventional formulation, from parameter recalibration to structural modification, in order to represent distinct anaerobic valorization routes of FVW within a unified mechanistic framework. These findings expand the applicability of ADM1 to emerging waste-to-energy systems and provide a scientific basis for future virtual biorefinery strategies for organic waste valorization.