All Issue

2025 Vol.62, Issue 5 Preview Page

General Remarks

Evaluating LSTM-Based Surrogate Models for Predicting Productivity at UBGH2-6 Gas-Hydrate Site

LSTM 기반 대리모델을 이용한 UBGH2-6 가스 하이드레이트 사이트 생산성 예측 평가

Seunghwan Baek1
,
Jihoon Kim2*
백 승환1
,
김 지훈2*
1Earth Systems Science Energy and Environment Directorate, Pacific Northwest National Laboratory, 902 Battelle Boulevard, Richland, WA, 99352, USA
2Harold Vance Department of Petroleum Engineering, Texas A&M University, 3116 TAMU Richardson Building, College Station, TX 77843, USA
1퍼시픽 노스웨스트 국립연구소, 에너지 환경 본부
2텍사스 A&M 대학교, 석유공학과
*Corresponding Author
31 October 2025. pp. 598-611
PDF XML Citation
Abstract

Data-driven, site-specific surrogate models are developed to enable rapid and accurate prediction of gas and water production at the UBGH2-6 gas-hydrate site. Synthetic datasets are generated under various depressurization scenarios. Considering the high computational cost and time requirements of full-physics simulations, long short-term memory (LSTM) networks are employed as surrogate models to efficiently approximate production behavior. LSTM surrogate models per phase (water and gas) are trained on datasets of differing sizes—basic and expanded—with the expanded dataset encompassing a broader range of operational conditions. Water-production predictions are highly accurate even under limited input features, whereas gas-production predictions improve significantly when trained on the expanded dataset. Sensitivity analyses highlight critical architectural components, including batch normalization, learning rate, and batch size, as key drivers of model performance. These findings indicate that LSTM-based surrogates are a computationally efficient and reliable alternative to conventional full-physics simulations for forecasting production in complex gas-hydrate reservoirs.

Keywords
gas hydrate
LSTM
ulleung basin
productivity
deep learning

본 연구에서는 UBGH2-6 가스 하이드레이트 사이트에서 가스 및 물 생산을 신속하고 정확하게 예측하기 위한 데이터 기반 현장 맞춤형 대리모델들을 개발하였다. 다양한 감압 시나리오 하에서 합성 데이터 세트를 생성하였으며, 전물리 시뮬레이션의 높은 계산 비용과 시간 소요를 고려하여 장·단기 메모리(LSTM) 신경망을 대리모델로 활용하였다. 상(phase)별(물, 가스) LSTM 대리모델은 기본 및 확장 데이터 세트를 이용해 학습되었으며, 확장 데이터 세트는 보다 넓은 운전 조건 범위를 포함한다. 입력 변수의 제약이 존재함에도 불구하고 물 생산 예측은 매우 높은 정확도를 보였으며, 가스 생산 예측은 확장 데이터 세트으로 학습 시 유의미하게 향상되었다. 민감도 분석 결과, 배치 정규화, 학습률, 배치 크기 등 주요 모델 구조 요소들이 성능 향상의 핵심 요인으로 확인되었다. 이러한 결과는 LSTM 기반 대리모델이 복잡한 가스 하이드레이트 저류층의 생산 예측을 위해 기존 전물리 시뮬레이션보다 계산 효율적이고 신뢰할 수 있는 대안임을 보여준다.

키워드
가스 하이드레이트
장·단기 메모리 신경망
울릉분지
생산성
딥러닝
References
1

Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G.S., Davis, A., Dean, J., Devin, M., Ghemawat, S., Goodfellow, I., Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, L., Kudlur, M., Levenberg, J., Mane´, D., Monga, R., Moore, S., Murray, D., Olah, C., Schuster, M., Shlens, J., Steiner, B., Sutskever, I., Talwar, K., Tucker, P., Vanhoucke, V., Vasudevan, V., Vie´gas, F., Vinyals, O., Warden, P., Wattenberg, M., Wicke, M., Yu, Y., and Zheng X., 2016. TensorFlow: Large- Scale Machine Learning on Heterogeneous Systems, arXiv:1603.04467.

2

Baek, S., Bacon, D.H., and Huerta, N.J., 2023. Enabling site-specific well leakage risk estimation during geologic carbon sequestration using a modular deep-learning-based wellbore leakage model, International Journal of Greenhouse Gas Control, 126, 103903.

10.1016/j.ijggc.2023.103903
3

Baek, S., Hibbard, L, Mitchell, N, Appriou, D, Dilmore, R., and Mamud, M., 2025. Deep-learning-enhanced assessment of wellbore barrier effectiveness in geologic storage systems with intermediate aquifers, International Journal of Greenhouse Gas Control, 147, 104487.

10.1016/j.ijggc.2025.104487
4

Bahk, J.J., Kim, G.Y., Chun, J.H., Kim, J.H., Lee, J., Ryu, B.J., Lee, J.H., and Son, B.K., 2013. Characterization of gas hydrate reservoirs by integration of core and log data in the Ulleung Basin, East Sea, Marine and Petroleum Geology, 47, p.30-42.

10.1016/j.marpetgeo.2013.05.007
5

Bousquet, O. and Elisseeff, A., 2002. Stability and generalization, Journal of Machine Learning Research, 2, 499-526.

6

Chaikine, I.A. and Gates, I.D., 2021. A machine learning model for predicting multi-stage horizontal well production, Journal of Petroleum Science and Engineering, 198, 108133.

10.1016/j.petrol.2020.108133
7

Dauphin, Y.N., Fan, A., Auli, M., and Grangier, D., 2017. Language Modeling with Gated Convolutional Networks. D. Precup and Y.W. Teh, eds., Proceedings of the 34th International Conference on Machine Learning, volume 70 of Proceedings of Machine Learning Research, p.933-941, PMLR.

8

Goodfellow, I., Bengio, Y., and Courville, A., 2016. Deep Learning, MIT Press, Cambridge, MA, USA, 775p.

9

Ioffe, S. and Szegedy, C., 2015. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift.

10

Ji, M., Kwon, S., Kim, M., Kim, S., and Min, B., 2022. Generation of Synthetic Compressional Wave Velocity Based on Deep Learning: A Case Study of Ulleung Basin Gas Hydrate in the Republic of Korea, Applied Sciences, 12(17).

10.3390/app12178775
11

Kim, A.R., Kim, H.S., Cho, G.C., and Lee, J.Y., 2017. Estimation of model parameters and properties for numerical simulation on geomechanical stability of gas hydrate production in the Ulleung Basin, East Sea, Korea, Quaternary International, 459, p.55-68.

10.1016/j.quaint.2017.09.028
12

Kim, H.S., Cho, G.C., Lee, J., and Kim, S.J., 2013. Geotechnical and geophysical properties of deep marine fine-grained sediments recovered during the second Ulleung Basin Gas Hydrate expedition, East Sea, Korea, Marine and Petroleum Geology, 47, p.56-65.

10.1016/j.marpetgeo.2013.05.009
13

Kim, J. and Lee, J.Y., 2024. Sequential formulation of all-way coupled finite strain thermoporomechanics for largely deformable gas hydrate deposits, International Journal for Numerical and Analytical Methods in Geomechanics, 48, p.1613-1633.

10.1002/nag.3696
14

Kim, J., Moridis, G.J., and Rutqvist, J., 2012a. Coupled flow and geomechanical analysis for gas production in the Prudhoe Bay Unit L-106 well Unit C gas hydrate deposit in Alaska, Journal of Petroleum Science and Engineering,92-93, p.143-157.

10.1016/j.petrol.2012.04.012
15

Kim, J., Moridis, G.J., Yang, D., and Rutqvist, J., 2012b. Numerical studies on two-way coupled fluid flow and geomechanics in hydrate deposits, SPE Journal, 17(2), p.485-501.

10.2118/141304-PA
16

Lee, J., Kim, G.Y., Kang, N., Yi, B.Y., Jung, J., Im, J.H., Son, B.K., Bahk, J.J., Chun, J.H., Ryu, B.J., and Kim, D., 2013a. Phys- ical properties of sediments from the Ulleung Basin, East Sea: Results from Second Ulleung Basin Gas Hydrate Drilling Expedition, East Sea (Korea), Marine and Petroleum Geology, 47, p.43-55.

10.1016/j.marpetgeo.2013.05.017
17

Lee, J.S., Lee, J., Kim, Y., and Lee, C., 2013b. Stress-dependent and strength properties of gas hydrate-bearing marine sediments from the Ulleung Basin, East Sea, Korea, Marine and Petroleum Geology, 47, p.66-76.

10.1016/j.marpetgeo.2013.04.006
18

Lee, K., Lim, J., Yoon, D., and Jung, H., 2019. Prediction of shale-gas production at duvernay formation using deep-learning algorithm, SPE Journal, 24(06), p.2423-2437.

10.2118/195698-PA
19

Lee, T., Lee, J., Ahn, T., and Son, H., 2021. Numerical simulation of gas hydrate production using the cyclic depressurization method in the Ulleung Basin of the Korea east sea, Applied Sciences, 11, 9748.

10.3390/app11209748
20

Lundberg, S. and Lee, S.I., 2017. A Unified Approach to Interpreting Model Predictions.

21

Luo, P., Wang, X., Shao, W., and Peng, Z., 2019. Towards Understanding Regularization in Batch Normalization, arXiv:1809.00846.

22

Malhotra, P., Vig, L., Shroff, G., Agarwal, P., et al. 2015. Long short term memory networks for anomaly detection in time series, Proceedings, 89, p.89-94.

23

Moridis, G.J., 2003. Numerical studies of gas production from methane hydrates, SPE Journal, 8(4), p.359-370.

10.2118/87330-PA
24

Moridis, G.J., Kim, J., Reagan, M.T., and Kim, S.J., 2013. Feasibility of gas production from a gas hydrate accumulation at the UBGH2-6 site of the Ulleung Basin in the Korean East Sea, Journal of Petroleum Science and Engineering, 108, p.180-210.

10.1016/j.petrol.2013.03.002
25

Moridis, G.J., Kowalsky, M.B., and Pruess, K., 2008. TOUGH+HYDRATE v1.0 User’s Manual: A Code for the Simulation of System Behavior in Hydrate-Bearing Geologic Media. Report LBNL-00149E, Lawrence Berkeley National Laboratory, Berkeley, CA.

10.2172/927149
26

Nagao, M., Yao, C., Onishi, T., Chen, H., Datta-Gupta, A., and Mishra, S., 2024. An efficient deep learning-based workflow for CO2 plume imaging considering model uncertainties with distributed pressure and temperature measurements, International Journal of Greenhouse Gas Control, 132, 104066.

10.1016/j.ijggc.2024.104066
27

Park, G., Kwon, S., Ji, M., Min, B., Nguyen, X.H., Kim, K.H., Kim, S., and Lee, K., 2021. A review on deep learning applications in logging data to model gas-hydrate-bearing sediments, Journal of the Korean Society of Mineral and Energy Resources Engineers, 58(2), p.161-178.

10.32390/ksmer.2021.58.2.161
28

Park, S.Y., Son, B.K., Choi, J., Jin, H., and Lee, K., 2022. Application of machine learning to quantification of mineral composition on gas hydrate-bearing sediments, Ulleung Basin, Korea, Journal of Petroleum Science and Engineering, 209, 109840.

10.1016/j.petrol.2021.109840
29

Patel, G.S. and Jana, A.K., 2024. Machine-Learning-Based prediction of gas hydrate dynamics: a comparison with a funda- mental model against experimental data, Energy & Fuels, 38(17), p.16272-16285.

10.1021/acs.energyfuels.4c02271
30

Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., and Duchesnay, E., 2011. Scikit-learn: machine learning in python, Journal of Machine Learning Research, 12, p.2825-2830.

31

Qin, Y., Song, D., Chen, H., Cheng, W., Jiang, G., and Cottrell, G., 2017. A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction.

10.24963/ijcai.2017/366
32

Rutqvist, J. and Moridis, G.J., 2009. Numerical studies on the geomechanical stability of hydrate-bearing sediments, SPE Journal, 14(2), p.267-282.

10.2118/126129-PA
33

Rutqvist, J., Moridis, G., Grover, T., and Collett, T., 2009. Geomechanical response of permafrost-associated hydrate deposits to depressurization-induced gas production, Journal of Petroleum Science and Engineering, 67, p.1-12.

10.1016/j.petrol.2009.02.013
34

Sak, H., Senior, A.W., and Beaufays, F., 2014. Long short-term memory recurrent neural network architectures for large scale acoustic modeling, INTERSPEECH, p.338-342.

10.21437/Interspeech.2014-80
35

Santurkar, S., Tsipras, D., Ilyas, A., and Madry, A., 2019. How Does Batch Normalization Help Optimization?, arXIv:1805.11604.

36

Sarkar, P., Yoon, S., Kim, J., Baek, S., Sun, A., and Yoon, H., 2025. Surrogate models for development of unconventional shale reservoirs by an integrated numerical approach of hydraulic fracturing, flow and geomechanics, and machine learning, Ge-omechanics for Energy and the Environment,38, 100691.

10.1016/j.gete.2025.100691
37

Wang, L. and Sun, A., 2020. Well Spacing Optimization for Permian Basin Based on Integrated Hydraulic Fracturing, Reser- voir Simulation, and Machine Learning Study. Proceedings of the 8th Unconventional Resources Technology Conference, American Association of Petroleum Geologists.

10.15530/urtec-2020-3104
38

White, M.D., Kneafsey, T., Seol, Y., Uchida, S., Lin, J., Myshakin, E., Gai, X., Gupta, S., Reagan, M., Queiruga, A., and Kimoto, S., 2020. An international code comparison study on coupled thermal, hydrologic and geomechanical processes of natural gas hydrate-bearing sediments, Marine and Petroleum Geology, 120, 104566.

10.1016/j.marpetgeo.2020.104566
39

Wu, H. and Li, C., 2025. The logging identification of gas hydrate occurrence state based on deep learning, Geoenergy Science and Engineering, 255, 214083.

10.1016/j.geoen.2025.214083
40

Yoon, H., Yoon, S., Lee, J., and Kim, J., 2021. Multiple porosity model of a heterogeneous layered gas hydrate deposit in Ulleung Basin, East Sea, Korea: A study on depressurization strategies, reservoir geomechanical response, and wellbore stability, Journal of Natural Gas Science and Engineering, 96, 104321.

10.1016/j.jngse.2021.104321
41

Yoon, H.C., Kim, J., Um, E.S., and Lee, J.Y., 2022. Integration of Electromagnetic Geophysics Forward Simulation in Coupled Flow and Geomechanics for Monitoring a Gas Hydrate Deposit Located in the Ulleung Basin, East Sea, Korea, Energies, 15, 3823.

10.3390/en15103823
42

Yoon, S., Yoon, H.C., Kim, J., and Lee, J.Y., 2024. The role of layer heterogeneity on productivity, geomechanical responses, and wellbore stability of a marine gas hydrate deposit in South Korea, Energy & Fuels, 38, p.11662-11674.

10.1021/acs.energyfuels.4c00557
43

Zhang, J., Moridis, G., and Blasingame, T., 2023. Effect of geomechanics and of grid discretization on the predictions of production from natural hydrate deposits and of the associated geomechanical system response, Gas Science and Engineering, 112, 204942.

10.1016/j.jgsce.2023.204942
Information
  • Publisher :The Korean Society of Mineral and Energy Resources Engineers
  • Publisher(Ko) :한국자원공학회
  • Journal Title :Journal of the Korean Society of Mineral and Energy Resources Engineers
  • Journal Title(Ko) :한국자원공학회지
  • Volume : 62
  • No :5
  • Pages :598-611
  • Received Date : 2025-09-19
  • Revised Date : 2025-10-20
  • Accepted Date : 2025-10-24
  • DOI :https://doi.org/10.32390/ksmer.2025.62.5.598
Journal Informaiton Journal of the Korean Society of Mineral and Energy Resources Engineers Journal of the Korean Society of Mineral and Energy Resources Engineers
Online Submission
  • NRF
  • KOFST
  • crossref crossmark
  • crossref cited-by
  • KCI 문헌 유사도 검사
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close