A comparison of methods for early prediction of anaerobic biogas potential on biologically treated municipal solid waste.

Abstract

Anaerobic gas production tests, generically Biochemical Methane Potential (BMP) or Biogas Potential (BP) tests, are often used to assess biodegradability, though long duration limits their utility. This research investigated whether simple modelling approaches could provide a reliable earlier prediction of total biogas production. Data were assessed from a non-automated biogas test on a large number of both fresh and processed municipal solid waste (MSW) samples, sourced from a mechanical biological treatment (MBT) plant. Non-linear models of biogas production curves were useful in identifying a suitable test endpoint, supporting a test duration of 50 days. Biogas production at 50 days (B50) was predicted using the first 14 days of test data, using (a) linear correlation, (b) a new linearisation process, and (c) non-linear kinetic models. Prediction errors were quantified as relative root mean squared error of prediction (rRMSEP), and bias. Predictions from most models were improved by removing the initial exponential increase phase. Linear correlation gave the most precise and accurate predictions at 14 days (rRMSEP\u202f=\u202f2.8%, bias under 0.05%) and allowed acceptable prediction (rRMSEP <10%) both at 8 days, and at 6 days using separate correlations for each sample type. Of the other predictions, the new linearisation process gave the lowest rRMSEP (10.6%) at 14 days. More complex non-linear models conferred no advantage in prediction of B50. These results demonstrate that early prediction of anaerobic gas production is possible for a well-optimised test, using only basic equipment and without recourse to external data sources or complex mathematical modelling.