|
Freiman O.A.
|
|
Northern (Arctic) Federal University Arkhangelsk |
|
Eremin N.A.
|
|
Oil and Gas Research Institute RAS Moscow |
DOI: 10.24412/2076-6785-2022-7-118-120
Abstract
Reservoir properties of oil are necessary to justify the effective regulation of field development. Measurement accuracy in field development depends on reservoir data (eg material balance calculations, reserves estimation, predictive data analysis). Incorrect measurement of reservoir properties can lead to serious errors in the calculation results. In the literature, the influence of reservoir data uncertainty on test results was considered, for example, in material balance equations and estimates of hydrocarbon reserves and the release of more volatile fluids. In recent decades, various models have been developed to assess the reservoir properties of formation fluids, such as empirical, compositional and based on neural networks. In this study, a machine learning method will be used to predict the performance indicators of an oil field and calculate reservoir fluid properties.
Materials and methods
Reservoir fluid properties were taken from a open database for the Volve field, North Sea, Norwegian sector. Machine learning methods formed the basis for determining reservoir properties of fluids and calculating technological development indicators.
Keywords
reservoir properties, oil and gas, machine learning, artificial intelligence
