The use of electronic gas sensors and near-infrared spectroscopy (NIR) to monitor the dynamics in a biogas process was evaluated using multivariate data analysis. The digester, a completely stirred 8 l tank reactor fed with a mixture of cellulose, albumin and minerals, was exposed to an overload of glucose after which monitoring of electronic gas sensor responses, NIR spectra as well as traditional chemical variables and analysis of microbial community structure were done. The responses from an array of electronic gas sensors consisting of MOS and MOSFET-sensors were correlated against volatile compounds in the headspace using partial least square (PLS) regressions. The root mean square error of prediction (RMSEP) was 0.15 g/l for acetate in the range of 0.14–1.72 g/l and the RMSEP for methane was 2.3% in the range of 27–73%. Selected wavelengths from the second derivative of the original NIR spectra (400–2500 nm) resulted in a PLS-model for predicting microbial biomass, measured as total phospholipid fatty acids, with a RMSEP of 9 nmol/ml in the range of 163–293 nmol/ml. The NIR model developed for acetate had a RMSEP of 0.20 g/l within the range of 0.14–1.72 g/l. The results clearly show that both NIR and an array of electronic gas sensors can provide simultaneous non-invasive in situ monitoring of important process variables in anaerobic digesters.

This content is only available as a PDF.
You do not currently have access to this content.