Abstract
A chemical sensor array (consisting of eight conducting polymer (CP) sensors) was used to continuously monitor for sudden changes in the quality of wastewater over a 12-month period. The headspace gas generated from a sparged liquid sample within a temperature controlled flow-cell was directly transferred to the sensor chamber for analysis. Results from the field study at a wastewater treatment plant (using a fully automated system) provided high resolution profiles and showed the detection of accidental and simulated pollution events. A model was developed for the detection and identification of sudden changes in the sensor responses, and was successfully tested using large datasets acquired over several months. This simple data mining approach showed a great sensitivity and flexibility, independent of long-term drift effects, diurnal variations and changes in temperature and humidity levels. The findings demonstrate that a chemical sensor array can be operated under harsh environmental conditions and act as an upset early warning device for the detection and identification of wastewater treatment process failure.