Leak detection in gas flows and determination of flow pattern and gas fraction in multiphase flows by acoustic measurements and neural networks

• Grant number: 14/09347-6
• Support Opportunities: Research Grants - Visiting Researcher Grant - Brazil
• Duration: August 04, 2014 - August 03, 2015
• Field of knowledge: Engineering - Mechanical Engineering - Transport Phenomena
• Principal Investigator: Ana Maria Frattini Fileti
• Grantee: Ana Maria Frattini Fileti
• Visiting researcher: Ricardo Dias Martins de Carvalho
• Visiting researcher institution: Universidade Federal de Itajubá (UNIFEI). Pró-Reitoria de Pesquisa e Pós Graduação, Brazil
• Host Institution: Faculdade de Engenharia Química (FEQ). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

One goal of this research proposal is to further develop the technique for sonic leak detection in gas flows, initially developed under FAPESP grant Nº 2011/23571-8. In this previous project sonic signals characteristic of gas leaks were acquired and analyzed and a neural network developed for the data reduction. However, the experiments were rather simple as the main goal was to develop a foundation for the development of the new technique. In the current project the scope of the experiments will be extended to be more representative of practical applications of domestic gas supply facilities. Simultaneously, the neural model will also be extended and tested. Another goal of this research proposal is to further develop the technique for flow pattern identification and gas fraction determination of oil-water-gas flows using acoustic measurements and neural networks. For the past several years this technique has been the focus of a research team that includes the proponents of the current research proposal. The technique has been proven to provide good results in the laboratory setting and should now be tested with real flows from the oil industry. These research activities will be carried out at Faculdade de Engenharia Química (FEQ-UNICAMP) and Núcleo Experimental de Atalaia (NEAT-PETROBRAS). As an overall outcome of this research proposal, both techniques will be brought closer to the real applications. (AU)