The article discusses automatic interpretation of results of transient well tests of oil and gas wells. Pressure-transient test under build-up and production-transient test in the production wells and fall-off test in the injection wells are considered. Machine learning techniques are applied to solve the problem. Based on the log-log plots of the pressure change curve and its logarithmic derivative, the proposed algorithm allows determining the most suitable model of the well-reservoir system. This problem, in machine learning terms, is a multilabeled classification problem, since the same input data can be assigned to one or more classes. A one-dimensional convolutional neural network model was selected based on the results of cross-validation. After determining the model of the well-reservoir system, the analytical algorithm makes it possible to calculate the parameters of the reservoir, well completion parameters and distances to the boundaries of the reservoir and surrounding wells. Algorithms for automatic interpretation of pressure buil-up curve, pressure fall-off curve and pressure leveling curve are implemented as a separate functionality in the program complex RN-VEGA, which ensures the execution of a wide range of tasks related to the processing of source data, analysis and interpretation of various well testing technologies. The automatic interpretation functionality in the RN-VEGA expands the capabilities of an expert in well testing interpretation by generating a list of relevant models of the well-reservoir system and solving the problem of calculating the parameters for each of these models, which is impossible when processing dynamic well operation data manually. The functionality was tested on synthetic and field data from fields in Western Siberia and Volga-Ural region. The results of comparison with similar functionality in foreign software showed that the new algorithm allows obtaining the required reservoir and well completion parameters with more than 8 % greater accuracy.
References
1. Urazov R.R., Davletbaev A.Ya., Sinitskiy A.I. et al., Rate transient analysis of fractured horizontal wells (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2020, no. 10, pp. 62–67 , DOI: https://doi.org/10.24887/0028-2448-2020-10-62-67
2. Asalkhuzina G.F., Davletbaev A.Ya., Salakhov T.R., Applying decline analysis for reservoir pressure determination (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2022, no. 10, pp. 30-33, DOI: https://doi.org/10.24887/0028-2448-2022-10-30-33
3. KAPPA: Saphir – analiz GDIS (KAPPA: Saphir – well test analysis), URL: https://www.kappaeng.com/software/saphir/
4. Sarapulova V.V., Davletbaev A.Ya., Kunafin A.F. et al., The RN-VEGA program complex for well test analysis and interpretation (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2023, no. 12, pp. 124 –129, DOI: https://doi.org/10.24887/0028-2448-2023-12-124-129
5. Afanas'ev I.S., Sergeychev A.V., Asmandiyarov R.N. et al., Automatic well test data processing: a time series wavelet analysis approach (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2012, no. 11, pp. 34-37.
6. Ivaschenko D., Davletbaev A., Baikov V. et al., Wavelet-based transform analysis for non-darcy gas flow noisy data interpretation (In Russ.), SPE-166909-MS, 2013, DOI: http://doi.org/10.2118/166909-MS
7. Allain O.F., Horne R.N., Use of artificial intelligence in well-test interpretation, Journal of Petroleum Technology, 1990, V. 42 (03), pp. 342–349,
DOI: http://doi.org/10.2118/18160-PA
8. Daolun Li, Xuliang Liu, Wenshu Zha et al., Automatic well test interpretation based on convolutional neural network for a radial composite reservoir, Petroleum Exploration and Development, 2020, V. 47(3), pp. 623–631, DOI: https://doi.org/10.1016/S1876-3804(20)60079-9
9. Arubi S.L., Ikporo B., Igbani S., Obuebute A., Well test analysis and interpretation: the use of artificial neural network, International Journal of Engineering Applied Sciences and Technology, 2020, V. 4(11), pp. 438–446.
10. Ahmadi R., Shahrabi J., Aminshahidy B., Automatic well-testing model diagnosis and parameter estimation using artificial neural networks and design of experiments, Journal of Petroleum Exploration and Production Technology, 2016, V. 7(3), pp. 759–783, DOI: https://doi.org/10.1007/s13202-016-0293-z
11. Ivakhnenko A.G., Lapa V.G., Cybernetic predicting devices, New York: CCM Information Corp, 1966, 256 p.
12. Specht D.F., A general regression neural network, IEEE Transactions on Neural Networks, 1991, V. 2(6), pp. 568–576, DOI: https://doi.org/10.1109/72.97934
13. Specht D.F., Generation of polynomial discriminant functions for pattern recognition, IEEE Transactions on Electronic Computers, 1967, V. EC-16(3), pp. 308–319,
DOI: https://doi.org/10.1109/PGEC.1967.264667
14. Gringarten A.C., From straight lines to deconvolution: The evolution of the state of the art in well test analysis, SPE-102079-PA, 2008, DOI: http://doi.org/10.2118/102079-PA
15. Davletbaev A.Ya., Asalkhuzina G.F., Urazov R.R., Sarapulova V.V., Gidrodinamicheskie issledovaniya skvazhin v nizkopronitsaemykh kollektorakh (Hydrodynamic studies of wells in low-permeability reservoirs), Novosibirsk: DOM MIRA Publ., 2023, 176 p.
