In recent years, more and more oil and gas companies have been getting interested in automated interpretation of geophysical well logging data. Thereat, the automation goals combine considerable acceleration of the data interpretation process and the opportunity to downsize the number of the data interpreters in research and service divisions of the companies with the interpretation quality and uniformity enhancement, knowledge and expertise preservation within the company, etc. Nowadays, the principal approaches to the digitalization of the data interpretation process are working out versatile or, to the contrary, highly special algorithms by the relevant professionals, and the use of machine learning means. However, the three approaches mentioned above have certain disadvantages: in the first case, the applicability of the algorithms to solving specific tasks may be problematic, while in the second the algorithms are too specific and hardly adaptable to changing conditions. At the same time, the application of neural networks is an extremely obscure method complicated by low controllability of the automated processes. Moreover, the overwhelming majority of the automation means offered herein require a geophysicist to have certain programming skills, and as the tasks to be solved are becoming more and more complicated, such restriction is becoming increasingly substantial.
As a possible solution of the problems described above, we offer a concept of a digital apprentice for log interpretation (DALI). An interactive system of such type provides a geophysicist with an opportunity to easily and quickly formalize his own observations and actions while working with well-log curves, interacting with the DALI directly in the course of the interpretation. The training results in such data processing script, all steps of which may be represented in a simple readable form; thereat, special attention is paid to visual patterns on well-log curves for an interpreter to be governed by. The DALI is particularly notable for high flexibility of the algorithms it generate, that is, their ability to be automatically adjusted and supplemented when new extraordinary situations occur in the course of the interpretation, as well as for the lookback analysis, thanks to which all the changes introduced into the data processing script are verified in real time basing on already processed material.
References
1. Dmitrievskiy A.N., Eremin N.A., Digital modernization of oil and gas ecosystems – 2018 (In Russ.), Aktual'nye problemy nefti i gaza, 2018, no. 2 (21), pp. 1–12.
2. Belozerov B.V., Bukhanov N.V., Egorov D.V. et al., Automatic well log analysis across Priobskoe field using machine learning methods (In Russ.), SPE-191604-18RPTC-RU, 2018, DOI:10.2118/191604-18RPTC-MS.
3. Minikeeva L.R., Nadezhdin O.V., Nugumanov E.R. et al., Development of methods for automation of multi-well logging data interpretation and core analysis (in Russ.), Neftyanoe khozyaystvo = Oil Industry, 2018, no. 6, pp. 54–57, DOI: 10.24887/0028-2448-2018-6-54-57
