The maintenance of static equipment is conditional on accuracy, sufficiency and up-to-dateness of information which is used to make qualitative conclusions about the technical condition and residual life of a production unit. The effectiveness of maintenance measures, on the one hand, depends on regulatory requirements, and on the other hand, on the human factor. Significant costs are normally expected. The installation of advanced telemetry systems for static equipment units of relatively low cost and criticality is often not economically reasonable. Moreover, some pieces of acquired information are not applied advantageously. The article presents a solution to effective data acquisition via hybrid modelling to collect additional information on technical condition and residual life. This method is illustrated on 50 units of field tanks storing oil and produced water. The proposed approach fully complies with the regulatory documentation of the Russian Federation and involves the joint use of three main models: 1) stress-strain state, 2) defect growth accounting, and 3) defect prediction model. The approach is universal; its application is possible for various types of static equipment. The article presents, firstly, particularities of raw data processing modelling; secondly, training of cross-functional neural network stress-strain state models; thirdly, formation of a defect prediction model based on general recommended practice; and lastly, development of a machine learning model for defect prediction. There are also results of verification process of a model which was applied on tanks that have been in operation since 1980s. Considering the complexity of this model, verification was carried out separately for each individual type of the hybrid model. Validation of the model and feedback from operating organizations confirmed the prospect of using the developed hybrid model as part of a recommendation system for more thorough control of the weakest units; more accurate assessment of the technical condition and residual life of a unit (taking into account probable defects); assessing the risks of negative developments and planning budgets for various types of diagnostics and maintenance.
References
1. Golikov A.V., Slozhenkin G.E., Overview of types and analysis of the causes of defects and damages in the bearing structures of steel tanks (In Russ.), Vestnik Volgogradskogo gosudarstvennogo arkhitekturno-stroitel’nogo universiteta. Ser. Stroitel’stvo i arkhitektura, 2021, no. 4(85), pp. 14–28.
2. Smolyago G.A., Frolov N.V., Applied method for predicting corrosion damages and remaining resource of bendable reinforced concrete elements taking into account operating experience of similar projects (In Russ.), Vestnik Belgorodskogo gosudarstvennogo tekhnologicheskogo universiteta im. V.G. Shukhova = Bulletin of Belgorod State Technological University named after. V. G. Shukhov, 2019, no. 2, pp. 49–54, DOI: https://doi.org/10.12737/article_5c73fc0ef063c3.60645861
3. Gaysin E.Sh., Bakhtizin R.N., Gabdrakhmanova N.T., Frolov Yu.A., Mathematical model of estimation of the residual resource of vertical steel tanks (In Russ.), Problemy sbora, podgotovki i transporta nefti i nefteproduktov, 2017, no. 3(109), pp. 113-122.
4. Yumaguzin U.F., Bashirov M.G., Forecasting of equipment remaining life in the oil and gas industry (In Russ.), Fundamental’nye issledovaniya, 2014, no. 3–2, pp. 277–280.
