Integrated Leak Detection in Gas Pipelines Using OLGA Simulator and Artificial Neural Networks

Abstract

Gas pipeline leakages do not only represent loss of valuable non-renewable resources but also a potential source of environmental pollution and fire disaster, thus making it very important to have a quick and accurate leak awareness scheme in pipeline systems. In a novel approach of providing industry with more reliable, friendly and cost effective technique in handling pipeline leakage especially in countries were pipeline vandalism has been widely reported. This require urgent detection and intervention, the potential of an Artificial Neural Network (ANN) in detecting and locating leaks from everyday flow line measurements was explored in this research paper. Pipeline responses under zero (no) leak condition, and realistic different leak conditions were simulated on a gas pipeline using OLGA simulator. Simulated leaks were seen to be characterized with unique flow rate, velocity, pressure and temperature signatures that were identifiable by the neural network. The network during training was able to learn the correlation between these signatures and the leak parameters and then use the learning in predicting and locating leaks from new signature data. The network predictions showed final RMSE values of 6.14 and 0.977 for the single and multiple leaks model respectively. The correlations between the actual and predicted leak sizes gave R values of 0.95 and 0.60 for the single and multiple leaks respectively while those of the actual and predicted leak locations were 0.97 and 0.96. The system was able to locate 90% of the induced leaks to a distance that is less than 10 m away from the actual leak locations while 62% of the ANN predicted leak sizes differed from the actual value by less than 50% of the actual leak size and the remaining 38% differed by about 65 -400% the actual sizes. This thus implies that the set has a better precision in predicting leak locations than predicting leak sizes. Based on obtained results, it can be said that with sufficiently large number of measurements, neural networks are of great potential in predicting and locating leaks. Results also showed that leak detectability tends to improve with increasing leak size and increasing distance from the source and the relatively low correlation obtained for the multiple leaks model can be attributed to the partial masking of smaller downstream leaks by larger upstream leaks.These characteristics of leaks are predominant to the countries with cases of vandalism, and could be a potential approach in handling pipe leaks.

Adaptive thresholds‐based leak detection using real‐time transient modeling of two‐phase flows

“…This program can simulate a one-dimensional, extended two-fluid system in addition to a steady-state point model (OLGAS) and a complete transient computational code. [16][17][18] The GERG 2008 Equation of state model has been used in this current research. 19 The black oil composition was used as fluid and OLGA methodology has been discussed.…”

“…The OLGA simulator is utilized for many different design and operational problems, such as startup and shutdown transients, terrain slugging, changing production rates, pigging, gas lift, and many others. This program can simulate a one‐dimensional, extended two‐fluid system in addition to a steady‐state point model (OLGAS) and a complete transient computational code 16–18 . The GERG 2008 Equation of state model has been used in this current research 19 .…”

Leak Detection in Gas Mixture Pipelines under Transient Conditions Using Hammerstein Model and Adaptive Thresholds

“…Data for training may be acquired through supervisory control and data acquisition (SCADA) system (physical sensors) or from the mathematical models of the pipelines (virtual sensors) [19]. In this study, OLGA simulator is used to generate pipeline data which is based on transient mathematical models and used by several studies [24,40,41]. Transient conditions in actual pipelines are due to various reasons such as varying customer demand, the compressibility of a gas mixture, changes in atmospheric conditions, dynamic friction factor, line shutdown, start-up, compressor surges, etc.…”

Data driven leakage diagnosis for oil pipelines: An integrated approach of factor analysis and deep neural network classifier

“…The leakage detection problem is treated as a classification problem and was applied to detect the leaks of the Seraje pipeline (in Iran) (Valizadeh et al, 2009). These authors also used the OLGA software (Afebu et al, 2015) to acquire the inlet and outlet pressures, temperatures, and the flow rate signals which were needed for their methodology. They also extracted the data features and studied different classifiers; then, compared the corresponding (correct classification rate) CCRs.…”