Virtual flowmeter development for Zarubezhneft JSC

UDK: 681.518:622.276
DOI: 10.24887/0028-2448-2023-2-55-58
Key words: digitalization, digital transformation, digital field, digital ecosystem, digital twin, virtual flow meter, machine learning, virtual flow meter, big data
Authors: T.I. Mullagaliev (VNIIneft JSC, RF, Moscow), D.N. Kochanov (VNIIneft JSC, RF, Moscow), M.D. Trifonov (VNIIneft JSC, RF, Moscow), E.I. Magafurov (VNIIneft JSC, RF, Moscow), A.V. Chorny (Zarubezhneft JSC, RF, Moscow), A.V. Gubaev (Zarubezhneft JSC, RF, Moscow), A.A. Lubnin (Vietsovpetro JV, the Socialist Republic of Vietnam, Vung Tau)

Since 2019, the Zarubezhneft Group of Companies has been actively developing the direction of digital transformation. The tasks of digital transformation are solved through the use of innovative technologies and solutions in the areas of activity, including through the implementation of the Digital Field concept.

As part of the this concept, a digital ecosystem is being formed with the implementation of the principles of unification of the software used, their centralization and the creation of software lines by business segments. One of the first software products developed on the software infrastructure of VNIIneft JSC using machine learning methods and big data analysis is a virtual flow meter for Vietsovpetro JV. As part of the virtual flowmeter project, an algorithm for working with Big Data has been developed, which makes it possible to increase the responsiveness to changes in fluid flow rates and reduce their losses due to operational monitoring of data "from the wellhead". Machine learning models are considered as the basis of the algorithm. Data were collected and analyzed for the Vietsovpetro JV field, and a review was made on machine learning approaches and techniques, big data processing and model generation using the Python programming language libraries. A mechanism for automating the collection and filtering of initial data has been developed. Various types of machine learning models were tested to solve the regression problem. The software infrastructure has been improved for automatic additional training of virtual flow meter models upon receipt of new data. The uniqueness of the proposed approach lies in the fact that the wellhead parameters are only a part of the influencing indicators on the desired fluid flow rate, in addition to this, reservoir indicators, which are not promptly measured, are used. The liquid flow rate prediction error of the virtual flow meter prototype does not exceed 10 m3/day, which is sufficient to solve the tasks set for prompt response to flow rate changes at the wells of Vietsovpetro JV. Thus, the implementation of the Digital Field concept in the Zarubezhneft Group of Companies allows increasing the speed of information processing, improving the quality of planning and increasing the economic efficiency of field development.


1. Heddle R., Foot J., Rees H., ISIS Rate&Phase: Delivering virtual flow metering for 300 wells in 20 fields, SPE-150153-MS, 2012, DOI:10.2118/150153-MS

2. Gobel D., Briers J., Yee Men Chin, Architecture and Implementation of an Optimization Decision Support System, Proceedings of International Petroleum Technology Conference, Beijing, China, March 2013, DOI:10.2523/IPTC-17009-MS

To buy the complete text of article (a format - PDF) or to read the material which is in open access only the authorized visitors of the website can. .

Mobile applications

Read our magazine on mobile devices

Загрузить в Google play