TY - JOUR

T1 - From digital to mathematical models

T2 - A new look at geological and hydrodynamic modeling of oil and gas fields by means of artificial intelligence

AU - Zakharian, A. Z.

AU - Ursegov, S. O.

PY - 2019

Y1 - 2019

N2 - The article shows that the traditional version of geological and hydrodynamic models of oil and gas fields based on a computer approach is not the only possible one and it prevents the development of modeling as a whole, since it is not truly mathematical. Considering that computers do not work with images, but with numbers, a new methodology for construction of geological and hydrodynamic models of oil and gas fields is presented, which have an unusual appearance and are not intended for visual analysis, but they are more effective for computer forecasting. New mathematical geological and hydrodynamic models are cascades of fuzzy-logical matrices. The matrices of the geological model are formed from spatial coordinates and geological parameters; the time coordinate is additionally included in the matrices of the hydrodynamic model. The number of fuzzy-logical matrices can reach several thousands. Using the obtained matrices, one can construct membership functions and predict the values of evaluated parameters, for example, the efficiency of new drilling, the distribution of remaining reserves, and the levels of hydrocarbon production.The proposed approach for geological and hydrodynamic modeling from a set of matrix cascades may seem complex. However, the calculation of these cascades is carried out completely automatically, and no one should control it. The matrix cascades are mathematical functions, not illustrations of the geological structure of the studied objects and they are directly used for forecasting calculations. The matrix cascades are a new form of machine learning. To do this, it is advisable to use the big amount of data. The new machine learning method based on the matrix cascades opens up new possibilities for the application of artificial intelligence in the geological and hydrodynamic modeling of oil and gas fields.

AB - The article shows that the traditional version of geological and hydrodynamic models of oil and gas fields based on a computer approach is not the only possible one and it prevents the development of modeling as a whole, since it is not truly mathematical. Considering that computers do not work with images, but with numbers, a new methodology for construction of geological and hydrodynamic models of oil and gas fields is presented, which have an unusual appearance and are not intended for visual analysis, but they are more effective for computer forecasting. New mathematical geological and hydrodynamic models are cascades of fuzzy-logical matrices. The matrices of the geological model are formed from spatial coordinates and geological parameters; the time coordinate is additionally included in the matrices of the hydrodynamic model. The number of fuzzy-logical matrices can reach several thousands. Using the obtained matrices, one can construct membership functions and predict the values of evaluated parameters, for example, the efficiency of new drilling, the distribution of remaining reserves, and the levels of hydrocarbon production.The proposed approach for geological and hydrodynamic modeling from a set of matrix cascades may seem complex. However, the calculation of these cascades is carried out completely automatically, and no one should control it. The matrix cascades are mathematical functions, not illustrations of the geological structure of the studied objects and they are directly used for forecasting calculations. The matrix cascades are a new form of machine learning. To do this, it is advisable to use the big amount of data. The new machine learning method based on the matrix cascades opens up new possibilities for the application of artificial intelligence in the geological and hydrodynamic modeling of oil and gas fields.

KW - Artificial intelligence

KW - Computer forecasting

KW - Fuzzy-logical matrices

KW - Geological and reservoir modeling

KW - Machine learning

KW - Oil and gas fields

UR - http://www.scopus.com/inward/record.url?scp=85077365590&partnerID=8YFLogxK

U2 - 10.24887/0028-2448-2019-12-144-148

DO - 10.24887/0028-2448-2019-12-144-148

M3 - Article

AN - SCOPUS:85077365590

VL - 2019

SP - 144

EP - 148

JO - Neftyanoe Khozyaystvo - Oil Industry

JF - Neftyanoe Khozyaystvo - Oil Industry

SN - 0028-2448

IS - 12

ER -