Permeability Estimation and its Simulation to Determine the Reservoir Characteristics of Shurijeh Formation in One of the Reservoirs of Northeast Iran

Permeability is one of the most important parameters in hydrocarbon reservoirs. It is beneficial to have a correct understanding of the permeability and its distribution in the production management process. Due to the limitations, the coring process is performed on a small number of wells in the field, while most of the wells are subjected to well logging operations. Therefore, finding a way to estimate the characteristics of the reservoir by well-logging and modeling it on the field is a valuable technique. Therefore, in this study, the multilayer perceptron artificial neural network method (error back propagation) has been used to estimate the permeability of different parts of the Shurijeh Formation in the Kopeh Dagh sedimentary basin. Sonic logs, neutrons, density and the results of the formation evaluation, including porosity and saturation of useful water as input layer, and permeability data from core well analysis of two wells as output layer cells, were used to train the network. After training the network with the data of these two wells, the core analysis data of another well was used to test the network, which in the network test stage, a correlation coefficient of 98% for permeability was obtained. With the help of this neural network, permeability was estimated for other wells in the field that no core data were obtained from. After estimating the permeability using neural network, its distribution and expansion were determined using Sequential Gaussian Simulation algorithm (SGS) in the field scale. According to the obtained model, the sandstone areas, which are mainly in zones B and D, are separated as reservoir areas and also the central and northwestern areas of the field, due to the higher average permeability, are areas prone to further excavations of the field.