Prediktion av massflöden vid förbehandling av substratblandningar före samrötning

Biogas production plays an important role in the transition toward renewable energy systems by converting organic waste into methane and nutrient-rich biofertilizer. This study developed two process models to describe and predict the behaviour of the pre-treatment facility at St1 Biokraft’s biogas plant in Södertörn, Sweden. An annual model was constructed to reproduce the mass balance of input materials like food waste, manure etc., water and rejected streams, including nutrient distributions. A weekly model was designed to evaluate short-term dynamics in the pretreatment step and to simulate variations in total solids (TS) for two different slurry fractions, thickened and unthickened. Both models were calibrated using plant data, laboratory analyses, and operational measurements. Although the available data on water use were uncertain, both models accurately captured the observed outcomes within their respective scopes. The annual model achieved a coherent mass balance and realistic production levels, while the weekly model successfully replicated TS fluctuations throughout the week. The weekly model, however, required more manual adjustments and effort, making it less practical for daily operational use compared to the annual model. One way to improve the accuracy of short-term forecasts, such as weekly predictions, is to install additional flow meters. This would reduce the need for corrective adjustments and increase data reliability. The results demonstrate that mass flows can be reliably modelled even with limited data certainty, providing a foundation for improved process understanding and data consistency at the plant. Extending the prediction timeframe reduce variance in the model’s outputs and can cover for shortages in the data collection.