|
When making proved engineering decisions, it is necessary to consider relations between a geological formation, a well and a surface infrastructure. It requires conduction of multivariate calculations under uncertainty conditions. Gazprom Neft uses its own information system for integrated conceptual design – ERA:ISKRA. The technology that stands behind the system realizes batch calculation approach currently based on complete grid of all parameters combinations. When working with large datasets, the complete grid approach appears to be computationally expensive. Thus, time-effective algorithm for global optimization is needed. Current paper formulates the problem in terms of mathematical optimization. Different approaches were analyzed: classical design of experiments, derivative-free methods, metaheuristics, and metamodeling techniques. New metamodeling approaches are proposed. All algorithms were transformed to fit the current problem in terms of business constrains. Dataset of 864 pre-calculated options was used for the testing purpose. Two metrics were used: number of iterations to convergence and convergence at the certain iteration. Testing results show that proposed algorithms provide a significant improvement in the optimal solution search time for ERA:ISKRA. This should allow to enhance the process of conceptual design and rearrange time from routine operations to detailed research of the optimal solution. References 1. Batrashkin V.P., Ismagilov R.R., Panov R.A. et al., The integrated conceptual design as a tool of systematic engineering (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2016, no. 12, pp. 80–83. 2. Vlasov A.I., Mozhchilʹ A.F., Technology overview: from digital to intelligent field (In Russ.), PROneftʹ, 2018, no. 3(9), pp. 68–74. 3. Certificate of state registration of a computer program no. 2017610926 “ERA:ISKRA”, Authors: Zhagrin A.V., Khasanov M.M., Ismagilov R.R et al.. 4. Patent no. RU2670801C9, System of integrated conceptual design of hydrocarbon fields, Inventors: Ismagilov R.R., Panov R.A., Mozhchil' A.F., Gil'mutdinova N.Z., Dmitriev D.E., Kondakov D.E. 5. Khamidullin R.D., Ismagilov R.R., Kan A.V. et al., The choice of regional infrastructure development strategy in conditions of production uncertainty using software ERA:ISKRA (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2017, no. 12, pp. 64–67. 6. Ismagilov R.R., Maksimov Yu.V., Ushmaev O.S. et al., Integrated model for complex management of reservoir engineering and field construction (In Russ.), Neftyanoe khozyaystvo = Oil Industry, 2014, no. 12, pp. 71–73. 7. Shan S., Wang, G.G., Survey of modeling and optimization strategies to solve high-dimensional design problems with computationally-expensive black-box functions, Structural and Multidisciplinary Optimization, March 2010, V. 41, no. 2, pp. 219–241. 8. Rios L.M., Sahinidis N.V., Derivative-free optimization: a review of algorithms and comparison of software implementations, J. Global Optimization, 2013, V. 56, pp. 1247–1293. 9. Bergstra J., Yamins D., Cox D.D., Making a science of model search: hyperparameter optimization in hundreds of dimensions for vision architectures, Proceedings of the 30th International Conference on Machine Learning (ICML 2013), 2013. |
