Murali, R.; Dekhici, B.; Chen, T.; Zhang, D.; Short, M.

As the United Kingdom (UK) targets net-zero emissions by 2050, anaerobic digestion (AD) has become a cornerstone of renewable energy infrastructure. However, mathematical models, such as the Anaerobic Digestion Model No. 1 (ADM1), often struggle with high-solids agricultural feedstocks because they rely on Chemical Oxygen Demand (COD), a metric that introduces significant experimental error. To overcome this, this study applies an established mass-based ADM1 framework tailored for the co-digestion of maize silage and cow manure sourced from a UK AD site. This study uses a parallel reactor framework, using two identical laboratory-scale reactors to physically replicate the dynamic conditions of the full-scale site. A Global Sensitivity Analysis was first conducted, identifying biomass decay and carbohydrate breakdown rates as the most influential factors affecting system stability and model accuracy. The model was calibrated using data from the first reactor and then tested against an independent second reactor subjected to significant organic loading stress. Results show high predictive capabilities, with the model achieving a R2 of 0.81 for biogas production during calibration. The model maintained high predictive accuracy during the validation test of the second physical twin, achieving an R2 of 0.85, proving that the framework is robust and not overfitted to a single dataset. While predicting rapid fluctuations in pH and alkalinity remains challenging, the mass-based approach effectively forecasts gas yields and process stability. This methodology provides a reliable foundation for robust process modelling, offering a scalable tool for the UK biogas sector to optimise AD.