Assessment of Transient Extractable Power from Puga Geothermal Field Using Neural Network Model

Abstract

Three-dimensional conceptual model of Puga geothermal reservoir, which is developed using the available resistivity data is presented and the appropriate zones for the extraction of entrapped thermal energy are demarcated. Subsequently, the dynamic response of reservoir under doublet extraction scheme is simulated under various operational conditions by considering coupled fluid flow and heat transfer processes involved during the operation phase of the geothermal reservoir. Each operating condition is a combination of distinct well spacing, injection temperature, injection/extraction rate and injection depth. Subsequently, the transient temperature of thermal water that can be extracted over the lifetime of reservoir is estimated by solving the coupled governing equations, which is further used to be determine the extractable power. The heuristic knowledge between well spacing, injection/extraction rate, injection depth, extractable temperature and time is used to train and develop a feed forward neural network model which can further be used to predict the transient extractable temperature for any desired operating condition. The extractable temperature estimated using developed network model is observed to be satisfactory as the mean percentage average deviation is less than 0.5 %. The developed neural networks model, helps to estimate the extractable power from Puga geothermal reservoir for different operational conditions without performing simulation studies for intermittent conditions.