8 th Brazilian Congress on Metrology, Bento Gonçalves/RS, 2015 A fuzzy approach to leak detection in liquid pipelines Thiago Lessa Aramaki 1,3 , Ricardo Tanscheit 2 , Marley Vellasco 2 , Carlos R. Hall Barbosa 3 1 Petrobras Transportes, Rio de Janeiro, Brazil; 2 Dept. of Eletrical Engineering, Pontifical Catholic University of Rio de Janeiro; 3 Posgraduate Programme in Metrology, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil E-mail: aramaki@petrobras.com.br Abstract: This paper presents a leak detection technique based on fuzzy logic. The developed system is connected via OPC protocol to a pipeline simulator for testing purposes. The model represents a real pipeline operated by Petrobras Transportes. In order to assess its performance, leaks are simulated, as well as operational maneuvers that usually cause leak detection systems to generate false alarms. Keywords: leak detection, fuzzy logic, opc, simulation 1. INTRODUCTION Leak detection is a major concern in oil industry. Among the many techniques found in the literature, industry still makes use of the more conservative ones, while novel techniques are not sufficiently exploited by software developers. The selection of a leak detection system depends on a variety of factors such as pipe and products characteristics, instrumentation, communication capabilities, including the data acquisition system, thermal insulation, if present, burial depth, operating temperature, background noise and operating conditions [1]. Due to economic reasons and in some cases vandalism, instrumentation is only available at pipeline ends. Uncertainties from instrumentation, pipeline characteristics and fluid characteristics affect the sensibility to leak detection [2]. Performance metrics and classification of methods [3] provide some guidance to the industry on the selection criteria. Other types of classification can be found in [4]. Regarding internal leak detection techniques, [5] provide a useful coverage. Leak detection scenarios can be one of the following types: shut-in condition, flowing condition and flowing or shut-in condition with slack line 1 . Most of leak detection systems do not work under slack line condition. The proposed work applies to flowing pipelines without slack line condition. A technique solely based on fuzzy logic for leak detection and localization can be found in [6]. The system is composed of two fuzzy systems, one for state recognition and the other for leak detection. As the state of the pipeline is important for the detection of leaks, the output of the first fuzzy system is also an input to the second one. The system presented here acts in a similar way but consists of only one fuzzy system for both state recognition and leak detection. 1 Slack line is a condition where the pipeline absolute internal pressure is below the steam pressure of the product at any point of the pipe. At this point the product begins to evaporate.