Portable ion mobility spectrometry and partial least squared discriminant analysis for odour source discrimination in wastewater treatment plants

Abstract

Odour emissions from Wastewater Treatment Plants are a relevant issue concerning environmental and social impact, regulatory compliance, and plant management. Instrumental Odour Monitoring Systems are widely used for real-time odour emissions monitoring, but seasonal and plant variability limit their long-term reliability. Therefore, new sensing technologies and approaches are being studied to improve their reliability and the transferability of predictions between different plants and seasons. In this context, this work investigates the suitability of portable Ion Mobility Spectrometers to discriminate the main odour sources in Wastewater Treatment Plants. Two measurement campaigns were carried out in different seasons, considering different odour sources in two independent plants. Through a proper data analysis approach, based on the importance of ionic information, portable Ion Mobility Spectrometry proved effective in discriminating odour sources from the two main process lines: water and sludge treatment. In the first phase, conducted in the same plant and season, a balanced classification rate of 94 % (95 %CI: 82 %–100 %) was achieved. Subsequently, including seasonal and plant variability, a model trained on one plant was applied to the second. The direct transfer of the calibration achieved a balanced classification accuracy of 96 % (95 %CI: 86 %–100 %), confirming the relevance of the selected ions for odour assessment across different plants. These results suggest that portable Ion Mobility Spectrometry is a technology that deserves further attention for instrumental odour monitoring. The consistent classification rates obtained both within a single plant and when transferring the model demonstrate that Ion Mobility Spectrometry, combined with feature selection, can reliably identify ions specifically relevant for odour emission assessment.