«Automation and Informatization of the fuel and energy complex»

Surrogate modeling application to optimize liquified natural gas (LNG) production cycle by C3MR technology

UDC: 519.673+66.011
DOI: -

Authors:

¹ National University of Oil and Gas "Gubkin University", Moscow, Russia

Keywords: optimization of LNG production, surrogate modeling, natural gas, deep neural network, random search, process simulation

Annotation:

The authors of article present the results of stochastic optimization of the LNG production process using C3MR technology. Surrogate modeling was used for the optimization. The basic technological model was built in the UniSim Design R430 program. To create a synthetic database, the initial generation of initial data, their filtering and calculations based on the basic model were carried out. A deep neural network (DNN) was trained on the resulting database to create a surrogate model. The neural network architecture was selected automatically using the Hyperband hyper-parameter optimization algorithm. The goal of optimization based on the created surrogate model was to increase LNG production, taking into account restrictions on heat exchange and compressor equipment. The random search method was used as an optimization algorithm. In the course of optimization, optimal technological parameters were selected for the LNG production mode in Russia. The use of surrogate modeling made it possible to perform high-performance calculations with a high degree of reproducibility of results in minimal time. The calculation time did not exceed 1 minute, whereas the calculation using a technological model would have taken weeks. At the same time, the deviation of the macro-parameter UA, the fixation of which is the main limitation, was only 6 %. Based on the comparison results, the performance improvement was 6,44 % compared to previously published optimization results.

Bibliography:

1. Optimization of mixed fluid cascade LNG process using a multivariate Coggins step-up approach: Overall compression power reduction and exergy loss analysis / A. Nawaz, M.A. Qyyum, K. Qadeer [et al.] // Int. J. of Refrigeration. – 2019. – Vol. 104. – P. 189–200. – DOI: 10.1016/J.IJREFRIG.2019.04.002
2. Nikulin A.S., Kurkin D.S., Zhedyaevskiy D.N. Optimizatsiya energeticheskikh zatrat pri proizvodstve szhizhennogo prirodnogo gaza v arkticheskikh usloviyakh // Avtomatizatsiya i informatizatsiya TEK. – 2022. – № 7(588). – S. 5–14. – DOI: 10.33285/2782-604X-2022-7(588)-5-14
3. Nikulin A.S., Stepnov D.A., Kurkin D.S. Razrabotka surrogatnoy modeli kolonny stabilizatsii gazovogo kondensata // Gazovaya prom-st'. – 2022. – № 11(840). – S. 42–45.
4. Razrabotka i aprobatsiya metodicheskikh podkhodov i tsifrovykh tekhnologiy neyrosetevogo proksi-modelirovaniya ustanovivshegosya dvukhfaznogo techeniya mnogokomponentnoy smesi v sistemakh sbora i promyslovoy podgotovki gaza (na primere Chayandinskogo NGKM) / A.V. Belinskiy, V.A. Marishkin, V.V. Samsonova, P.V. Pyatibratov // Avtomatizatsiya i informatizatsiya TEK. – 2024. – № 4(609). – S. 44–59.
5. Kochueva O.N. Approksimatsiya koeffitsienta szhimaemosti gaza na osnove geneticheskikh algoritmov // Avtomatizatsiya i informatizatsiya TEK. – 2023. – № 11(604). – S. 59–68. – DOI: 10.33285/2782-604X-2023-11(604)-59-68
6. Lim Wonsub, Choi Kwangho, Moon Il. Current Status and Perspectives of Liquefied Natural Gas (LNG) Plant Design // Industrial & Engineering Chemistry Research. – 2013. – Vol. 52, Issue 9. – P. 3065–3088. – DOI: 10.1021/ie302877g
7. Primabudi E., Morosuk T., Tsatsaronis G. Multi-objective optimization of propane pre-cooled mixed refrigerant (C3MR) LNG process // Energy. – 2019. – Vol. 185. – P. 492–504. – DOI: 10.1016/j.energy.2019.07.035
8. Peng Ding-Yu, Robinson D.B. A New Two-Constant Equation of State // Industrial & Engineering Chemistry Fundamentals. – 1976. – Vol. 15, Issue 1. – P. 59–64. – DOI: 10.1021/i160057a011
9. Primabudi E. Evaluation and Optimization of Natural Gas Liquefaction Process with Exergy-Based Methods: A Case Study for C3MR: Doctoral Thesis. – Berlin, 2019. – 191 p. – DOI: 10.14279/DEPOSITONCE-8519
10. Impact of Sampling Technique on the Performance of Surrogate Models Generated with Artificial Neural Network (ANN): A Case Study for a Natural Gas Stabilization Unit / M. Ibrahim, S. Al-Sobhi, R. Mukherjee, A. AlNouss // Energies. – 2019. – Vol. 12, Issue 10. – P. 1906. – DOI: 10.3390/en12101906
11. A Python surrogate modeling framework with derivatives / M.A. Bouhlel, J.T. Hwang, N. Bartoli [et al.] // Advances in Engineering Software. – 2019. – Vol. 135. – Article No. 102662. – DOI: 10.1016/J.ADVENGSOFT.2019.03.005
12. Kingma D.P., Ba J.L. Adam: A method for stochastic optimization // 3rd Int. Conf. on Learning Representations (ICLR 2015), San Diego, CA, USA, May 7–9, 2015. – DOI: 10.48550/arXiv.1412.6980
13. Bergstra J., Bengio Y. Random Search for Hyper-Parameter Optimization // J. of Machine Learning Research. – 2012. – Vol. 13, Issue 10. – P. 281–305. – DOI: 10.5555/2503308.2188395
14. Močkus J. On bayesian methods for seeking the extremum // Optimization Techniques IFIP Technical Conf., Novosibirsk, July 1–7, 1974. – P. 400–404. – DOI: 10.1007/3-540-07165-2_55
15. Jamieson K., Talwalkar A. Non-stochastic Best Arm Identification and Hyperparameter Optimization // 19th Int. Conf. on Artificial Intelligence and Statistics (AISTATS), Cadiz, Spain, May 9–11, 2016. – P. 240–248. – DOI: 10.48550/arXiv.1502.07943
16. Hyperband: A Novel Bandit-Based Approach to Hyperparameter Optimization / L. Li, K. Jamieson, G. DeSalvo [et al.] // J. of Machine Learning Research. – 2018. – Vol. 18. – P. 1–52. – DOI: 10.48550/arXiv.1603.06560
17. Forrester A.I.J., Sóbester A., Keane A.J. Engineering Design via Surrogate Modelling: A Practical Guide. – John Wiley & Sons, 2008. – 210 p. – DOI: 10.1002/9780470770801