{"title":"利用基于逆问题求解的径向基函数网络估算油藏渗透率和储量场","authors":"V. P. Kosyakov, D. Yu. Legostaev","doi":"10.1134/s199508022460225x","DOIUrl":null,"url":null,"abstract":"<h3 data-test=\"abstract-sub-heading\">Abstract</h3><p>In the oil industry, there is a noticeable trend towards using proxy models to simulate various levels of complexity in order to make operational predictions. In particular, machine learning techniques are being actively developed in the context of the digitalization and automation of production processes. This paper proposes a method for combining a physically relevant fluid flow model with machine learning techniques to address the challenges of history-matching and prediction. The approach is demonstrated using synthetic oil reservoir models. The synthetic model has significant zonal inhomogeneities in the permeability and storativity fields. The simplified single-phase flow through a porous medium model was used for the proposed approach. This model was matched to the historical values of the development parameters by restoring the fields of the reservoir parameters. Properties fields were reconstructed using a radial basis functions network and a fully connected linear layer. Based on the reconstructed field, interwell connectivity coefficients were calculated, which corresponded qualitatively and quantitatively to the true interwell connectivity coefficients. The predictive characteristics of the proposed approach were evaluated by split the historical dataset into training and test time intervals.</p>","PeriodicalId":46135,"journal":{"name":"Lobachevskii Journal of Mathematics","volume":null,"pages":null},"PeriodicalIF":0.8000,"publicationDate":"2024-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Estimation of Oil Reservoir Transmissivity and Storativity Fields Using a Radial Basis Function Network Based on Inverse Problem Solving\",\"authors\":\"V. P. Kosyakov, D. Yu. Legostaev\",\"doi\":\"10.1134/s199508022460225x\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<h3 data-test=\\\"abstract-sub-heading\\\">Abstract</h3><p>In the oil industry, there is a noticeable trend towards using proxy models to simulate various levels of complexity in order to make operational predictions. In particular, machine learning techniques are being actively developed in the context of the digitalization and automation of production processes. This paper proposes a method for combining a physically relevant fluid flow model with machine learning techniques to address the challenges of history-matching and prediction. The approach is demonstrated using synthetic oil reservoir models. The synthetic model has significant zonal inhomogeneities in the permeability and storativity fields. The simplified single-phase flow through a porous medium model was used for the proposed approach. This model was matched to the historical values of the development parameters by restoring the fields of the reservoir parameters. Properties fields were reconstructed using a radial basis functions network and a fully connected linear layer. Based on the reconstructed field, interwell connectivity coefficients were calculated, which corresponded qualitatively and quantitatively to the true interwell connectivity coefficients. The predictive characteristics of the proposed approach were evaluated by split the historical dataset into training and test time intervals.</p>\",\"PeriodicalId\":46135,\"journal\":{\"name\":\"Lobachevskii Journal of Mathematics\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.8000,\"publicationDate\":\"2024-08-28\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Lobachevskii Journal of Mathematics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1134/s199508022460225x\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"MATHEMATICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Lobachevskii Journal of Mathematics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1134/s199508022460225x","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MATHEMATICS","Score":null,"Total":0}
Estimation of Oil Reservoir Transmissivity and Storativity Fields Using a Radial Basis Function Network Based on Inverse Problem Solving
Abstract
In the oil industry, there is a noticeable trend towards using proxy models to simulate various levels of complexity in order to make operational predictions. In particular, machine learning techniques are being actively developed in the context of the digitalization and automation of production processes. This paper proposes a method for combining a physically relevant fluid flow model with machine learning techniques to address the challenges of history-matching and prediction. The approach is demonstrated using synthetic oil reservoir models. The synthetic model has significant zonal inhomogeneities in the permeability and storativity fields. The simplified single-phase flow through a porous medium model was used for the proposed approach. This model was matched to the historical values of the development parameters by restoring the fields of the reservoir parameters. Properties fields were reconstructed using a radial basis functions network and a fully connected linear layer. Based on the reconstructed field, interwell connectivity coefficients were calculated, which corresponded qualitatively and quantitatively to the true interwell connectivity coefficients. The predictive characteristics of the proposed approach were evaluated by split the historical dataset into training and test time intervals.
期刊介绍:
Lobachevskii Journal of Mathematics is an international peer reviewed journal published in collaboration with the Russian Academy of Sciences and Kazan Federal University. The journal covers mathematical topics associated with the name of famous Russian mathematician Nikolai Lobachevsky (Lobachevskii). The journal publishes research articles on geometry and topology, algebra, complex analysis, functional analysis, differential equations and mathematical physics, probability theory and stochastic processes, computational mathematics, mathematical modeling, numerical methods and program complexes, computer science, optimal control, and theory of algorithms as well as applied mathematics. The journal welcomes manuscripts from all countries in the English language.