Intelligent design and modelling of natural gas storage facilities

Underground gas storage reservoirs provide an efficient and economical way to match the constant supply of gas from long-distance pipelines to the variable, weather-driven demand of the natural gas market. The most important design specifications for underground natural gas storage facilities in terms of startup and operation costs are its capacity, maximum reservoir pressure, number of wells required for drainage, flowing wellhead pressure, and the ratio between working gas and cushion gas. The relationships between these variables are quite complex, and a need exists for a reliable predictive tool. Despite these complexities, the optimal combination of these design parameters can be reached using artificial neural network (ANN) technology. In this study, ANN technology creates an intelligent system capable of learning the complex relationships between input parameters and output responses, which can quantify the importance of each relationship and design variables critical to the determination of optimal design.