Review on Modeling, Simulation, and Intelligent Control of Anaerobic Digestion for Food Waste Management

Abstract

Anaerobic digestion (AD) is a promising technology for the valorization of food waste into renewable energy and nutrient-rich digestate while reducing greenhouse gas emissions. This review focus on the evolution of analysis models from mechanistic models like ADM1 and its extensions to data-driven and hybrid approaches, with a strong focus on their application to the highly variable and rapidly biodegradable food waste stream. Recent advances in fast-solving ADM1 implementations, C-N-P-S extensions, physics-informed machine learning, soft-sensing, and reinforcement-learning-based control are studied. Study cases are used to demonstrate that a properly designed hybrid frameworks and intelligent controllers can increase the methane yield up to 25%, raise stable organic loading rates above 6 kg VS m⁻³ d⁻¹, and also virtually eliminate acidification events in full-scale food waste digesters. Remaining bottlenecks limited deep hybrid integration, poor online monitoring, parameter uncertainty, and scarce long-term validation of digital twins are discussed in detail. Some future directions are proposed to transform AD into a truly autonomous pillar of the circular bioeconomy.