Candidate: Rodrigo Rizzi Starr
Program: UFSC / POSMEC
Date: July, 2006
Advisor: Jonny Carlos da Silva
Abstract: This work deals with the implementation of an expert system prototype to refine existing leak detection methods. The presence of leak detection systems in oil or gas transportation pipelines is important because, due to large flow rates and kinds of fluid usually flowing through these pipelines, it is necessary to promptly detect leaks, so as to avoid human and financial losses and environmental damages. One of the most common method of leak detection is the analysis of information provided by sensors installed on the pipeline for other functions. This method of leak detection is usually called software-based leak detection. Despite being widely used in industry, this method has some drawbacks, among them a small sensitivity and, sometimes, a high rate of false alarms. On the other hand, in the daily contact with these systems, pipeline controllers end up developing heuristics to deal with the system deficiencies. Besides, pipeline controllers also use the results of the leak detector analysis while evaluating other operational conditions. Thus, this work presents the implementation of an expert system prototype able to carry out some of the analysis performed by pipeline controllers. This prototype was implemented in a partnership with a gas pipeline controller company. This system has been developed using the fusion of a traditional expert system with a fuzzy logic based system. The prototype was able to perform satisfactorily some of the required analysis, but the development process has shown itself very dependent on the company where the system will be installed. It has been identified that management support from this company is of fundamental importance to the success of the development process, specially during knowledge acquisition.
RIZZI STARR, R. Contribuições para a detecção de vazamentos em tubulações de gás natural: uma abordagem baseada em conhecimento. 2006. 112 p. Dissertação (Mestrado em Engenharia Mecânica). Universidade Federal de Santa Catarina, Florianópolis.