«Oilfield engineering»

Application of knowledge transfer method for predicting wells integrity failure

UDC: 622.276.5.054.3:004.896
DOI: -

Authors:

¹ TatNIPIneft PJSC "TATNEFT" named after V.D. Shashin, Almetyevsk, Russia
² Almetyevsk State Technological University "Higher School of Oil", Almetyevsk, Russia

Keywords: knowledge transfer method, machine learning, clustering, production casing leak, silhouette method

Annotation:

Evaluation of operational wells technical state is of current importance due to fields’ operation at the late stage of development, high water cut of products and wells aging. Operation of leaky wells results in oil losses, increased water cut, reduced equipment efficiency and environmental risks. To address the above challenges, the authors of the article have developed a new method for casing leak detection using machine learning algorithms. For this purpose, knowledge transfer method was used to train machine learning models amid lack of data about production targets under consideration. The approach is based on clustering of production targets (PT) to enable identification of similar targets to use bigger datasets for training less informative groups of data. To determine the optimal number of clusters, the silhouette method was applied to allow for grouping of production targets based on their characteristics. The main steps included clustering of production targets using silhouette method to determine the number of clusters, application of K-means algorithm for partitioning of production targets into clusters, and integration of data related to similar production targets to create more meaningful datasets. As a result, four main clusters for production targets with leak tight and leaky casings were identified. Clustering results were analyzed and compared to identify similar regions. Implementation of knowledge transfer method improved model training due to application of integrated data, which increased the accuracy of casing leak prediction. The authors of the article also compare quality metrics before and after application of knowledge transfer method to train casing leak prediction model. Knowledge transfer method is a promising solution to tackle lack of data about oil and gas assets for searching similar properties or dependencies between the targets of interest.

Bibliography:

1. K voprosu o negermetichnostyakh ekspluatatsionnykh kolonn / L.B. Khuzina [i dr.] // Stroitel’stvo neftyanykh i gazovykh skvazhin na sushe i na more. – 2023. – № 4(364). – S. 35–38. – DOI: 10.33285-0130-3872-2023-4(364)-35-38
2. Negermetichnost’ skvazhin − global’naya problema, a ne lokal’naya / D.P. Anikeev [i dr.] // Aktual’nye problemy nefti i gaza. – 2019. – № 4(27). – S. 1–14. – URL: https://oilgasjournal.ru/issue_27/anikeev-zakirov.pdf (data obrashcheniya: 28.10.2024).
3. Mukhametshin V.G., Dubinskiy G.S., Aver’yanov A.P. O prichinakh narusheniy germetichnosti ekspluatatsionnykh kolonn i meropriyatiyakh po ikh predotvrashcheniyu // Problemy sbora, podgotovki i transporta nefti i nefteproduktov. – 2016. – № 3. – S. 19–24.
4. Nazarov V.F., Mukhutdinov V.K. Kontrol’ germetichnosti obsadnoy kolonny i NKT v nagnetatel’nykh skvazhinakh po izmereniyam kompleksnoy apparaturoy // Innovatsionnaya nauka. – 2015. – № 12-2. – S. 107–112.
5. Metodika opredeleniya tekhnicheskogo sostoyaniya skvazhiny metodom mashinnogo obucheniya / I.M. Ishkulov, D.D. Takhauov, R.R. Vafin [i dr.] // Neftepromyslovoe delo. – 2024. – № 7(667). – S. 43–48.
6. O metodakh opredeleniya negermetichnosti ekspluatatsionnoy kolonny skvazhiny / I.M. Ishkulov [i dr.] // Neft. khoz-vo. – 2024. – № 7. – S. 56–60.
7. Opredelenie negermetichnosti kolonny skvazhin s ispol’zovaniem modeley mashinnogo obucheniya / I.M. Ishkulov [i dr.] // Neftegazovoe delo. – 2024. – T. 22, № 4. – S. 260–267.