Multivariate interaction analysis and optimisation of start-up parameters for maximising biohydrogen production via continuous dark fermentation of food waste

• A model simulation for dark fermentation developed and successfully validated. • Optimising the start-up operating parameters of dark fermentation is critical for maximising biohydrogen yield. • pH interactions with temperature, substrate concentration, and start-up hydraulic retention time are not significant. • Robust process design must consider integrated and multivariate optimisation rather than single-parameter adjustments. • Start-up phase optimisation is crucial for enabling a larger scalability of continuous dark fermentation systems. The growing demand for low-carbon energy has increased interest in hydrogen production through pathways that enable simultaneous waste valorisation and renewable energy production. Among these, dark fermentation of food waste presents an opportunity; however, its large-scale implementation remains challenged by low hydrogen yields, and limited understanding of parameter interactions and synergistic optimisation, particularly during the start-up phase of continuous systems. Optimising the start-up phase of dark fermentation systems is critical for achieving sustainable scale-up and economic viability. Therefore, this study performed a sensitivity analysis and response surface methodology optimisation of dark fermentative hydrogen production considering parameters such as start-up hydraulic retention time, substrate concentration, temperature, and pH based on food waste composition in terms of total solids, volatile solids, carbohydrates, proteins, and lipids. Through an innovative model simulation using Aspen Plus and a response surface methodology optimisation framework, results show that synergistically optimising start-up operating parameters of dark fermentation significantly maximises biohydrogen yield. The interactions between temperature and start-up hydraulic retention time as well as between start-up hydraulic retention time and substrate concentration were significant whereas those between temperature and substrate concentration, and between pH and the other parameters were not statistically significant within the investigated design space. More practically, the findings suggest that operating under thermophilic conditions (∼54 °C) with sufficiently long start-up retention times (∼15 days) at near-neutral pH (∼6) and moderately high substrate concentrations (∼42 gVS/L) can maximise hydrogen yield, thereby ultimately reduce production cost. Given the steady-state nature of the model, further validation under dynamic conditions and at pilot scale is necessary. Relevant future research should consider dynamic modelling and techno-economic assessments.