With Nigeria's massive proven natural Gas reserves and potential to provide a sustainable and economically viable system, a significant challenge has been the need for development strategies that enhance safety, growth, and investments in the gas sector. To tackle these challenges, the data generated in oil and gas, which is a valuable tool, has to be harnessed for stakeholders to implement life-changing solutions. The Nigerian economy has faced a significant drawback in the gas transportation and storage sector. The challenge in gas transportation can be seen in gas pipeline leakages which have resulted in the loss of lives, properties, and the country's revenue. Thus, early leak detection gas of pipelines remains critical for economic and safety reasons. This paper uses artificial intelligence to build models that utilize the available gas flow data to detect potential gas leakages across the pipeline. Machine learning algorithms which include Recurrent Neural Networks, and K-nearest neighbourhood are built and trained with operational data to derive the optimal learning model. Also, each model's performance metrics were evaluated to measure the model's accuracy and precision. Furthermore, an economic model is then developed to show the monetary benefits of implementing AI solutions to gas leakages. Thus, we provide a stepwise comparative analysis of the gas revenue, gas leakage detection cost, and the cost of providing an answer from an AI-based architecture to a non-AI-based one. The results showed that recurrent neural network outperforms the K-nearest neighbors in leak detection in pipelines as a result of the framework of neural network that allows the algorithm to learn without human supervision a and sift through the data set and label the data point. However, all the machine learning models possess high reliability. The accuracy and reliability of these models upon economic analysis proved to be a cost-effective solution lowering cost and increasing revenue. These models can be employed by companies and engineers to tackle the problem of pipeline leakage detection.

You can access this article if you purchase or spend a download.