Search

This study investigated the use of an electronic nose for on-line anaerobic reactor state monitoring at the pilot-scale level and then upscaling to the full-scale level. E-nose indicator was compared to classical state indicators such as pH, alkalinity, volatile fatty acids concentration and to other gas phase compounds. Multivariate statistical process control method, based on principal component analysis and the Hotelling's T(2) statistics was used to derive an indicator representative of the reactor state. At the pilot-scale level, the e-nose indicator was relevant and could distinguish 3 process states: steady-state, transient and collapsing process. At the full-scale level, the e-nose indicator could provide the warning of the major disturbance whereas two slight disturbances were not detected and it gave one major false alarm. This work showed that gas phase relation with anaerobic process should be deeper investigated, as an e-nose could indicate the reactor state, focusing on the gas phase., (Copyright © 2014 Elsevier Ltd. All rights reserved.)