Automated system based on electronic tongue system for urban water supply monitoring.

Author(s)

Guilherme de Souza Braga

• Document type: Doctoral Thesis
• Press: São Paulo.
• Institution: Universidade de São Paulo (USP). Escola Politécnica (EP/BC)
• Defense date: 2011-12-12

Examining board members

Fernando Josepetti Fonseca; Daniel Souza Corrêa; Leonardo Giordano Paterno; Antonio Riul Júnior; Ana Lúcia Silva

• Advisor: Fernando Josepetti Fonseca; Leonardo Giordano Paterno

Abstract

Water from São Paulo metropolitan area comes from water reservoirs, which are prone to algae bloom (eutrophication). These algae can be toxic and produce compounds, such as geosmin (GEO) and 2-methylisoborneol (MIB) that give unpleasant taste and odour to the water supplied to consumers. In fact, their presence in water is one of the major complains of consumers to São Paulo\'s water company (SABESP). Despite that, technologies that can be applied for monitoring these substances in real time and in locu are still not available. In this sense, the present thesis work has focused on the development of an electronic tongue system (ET) and evaluate its performance on detecting MIB, GEO and isoborneol (ISO) in water samples. The ET is based on an array of non-specific chemical sensors, which are made of gold interdigitated microelectrodes covered with nanostructured polymeric films deposited via the layer-by-layer technique and interrogated in alternate current regime. The ET calibration was conducted by analyzing distilled, mineral and tap water samples tainted with known concentrations of MIB, GEO and ISO. The samples were solutions with a single tainting compound and solutions containing MIB and GEO mixed together. The sensors\' electrical response (capacitance measurements) was interpreted by principal component analysis (PCA) and Fuzzy logic. Among the main features of this work, the present ET presents a detection limit of about 20 ng L-1 for both tainting compounds. The ET can also discriminate with great easiness and confiability individual and mixture samples. PCA plots show that samples are grouped into relatively small and well-separeted clusters (no overlapping). The ET is also capable of detect, based on changes on the first principal component, changes in the nGEO/nMIB molar ratio of mixed samples. Also, the ET can discriminate with great easiness, water samples supplied by SABESP, collected at Guarapiranga\'s reservoir, before and after the water treatment plant. The influence of the water type on the sensors\' response has little significance when the Fuzzy controller is employed. The ET with the Fuzzy controller presents an excellent performance with very low error rates: 0.25% (distilled water), 5% (mineral water) and 6.66% (tap water). Besides that, Fuzzy logic results can be displayed in simpler plots which ensures easier readings of hit and error rates of the ET. Also, the numeric values (compound concentration) can be replaced by linguistic variables such as proper and improper. Finally, it can be concluded that the ET developed herein, when considering sensor and data analysis, shows a great potential for monitoring MIB and GEO in water reservoirs, providing in real time and in locu information about water quality in treatment stations. This thesis also presents the study on the assembly of molecularly imprinted polymeric films to be used in specific chemical sensors, carried out during the internship at the Universidade Autonoma de Barcelona. (AU)

• FAPESP's process: 08/00631-2 - Automated system for urban water supply network monitoring based on the electronic tongue
• Grantee: Guilherme de Souza Braga
• Support Opportunities: Scholarships in Brazil - Doctorate