基于产量递减与LSTM耦合的常压页岩气井产量预测

doi:10.13809/j.cnki.cn32-1825/te.2023.05.012

摘要/Abstract

摘要：

针对常压页岩气井产气递减规律不明确、预测困难等问题，将产量递减模型与机器学习方法相结合，建立了一种新的页岩气井产量递减模型与LSTM（长短时间记忆神经网络）模型耦合预测方法。首先，依据页岩气井产水特点将南川常压页岩气井分为2类，类型1气井早期气水同产，后期产水明显减少，类型2气井长时间气水同产；其次，基于双对数诊断曲线和特征曲线明确气井的流动阶段，并采用7种常用的气井产量递减模型对不同类型页岩气井进行产量分析；最后，将递减模型的误差作为LSTM模型的输入，叠加得到耦合方法下产量预测。结果表明：拟稳态流动阶段的类型1的X1气井，优选递减模型为改进双曲递减模型和AKB递减模型，线性流动阶段的类型2的X2气井，优选SEPD（扩展指数递减）模型和Duong递减模型；对于递减模型误差较大时，耦合LSTM模型后，页岩气井产量预测精度明显提高，而递减模型误差较小时效果不显著。

关键词: 常压页岩气, 产量递减模型, LSTM模型, 流动阶段, 双对数诊断, 耦合方法

Abstract:

In order to address the challenges posed by the unclear production decline patterns and the difficulty in predicting production for normal pressure shale gas wells, a novel production prediction approach has been developed. This approach combines shale gas well production decline models with Long Short-Term Memory（LSTM） neural network models, leveraging machine learning techniques and different decline models for improved accuracy. Firstly, Nanchuan shale gas wells are divided into two types according to the characteristics of water production. For type 1, gas and water are produced simultaneously at the early stage, then water production decreases significantly in the later stage; while for type 2, gas and water are produced simultaneously for a long time. Secondly, double logarithmic diagnostic curves and characteristic curves are used to identify the flow stages of gas wells; then seven gas production decline models are used to analyze the production variety. Finally, the error of the decline models are used as the inputs of the LSTM model, meanwhile the yield prediction under the coupling method is obtained after superposition. The results show that a type 1 gas well, Well-X1, is in the pesudo-steady flow stage, its optimal decline model is the improved hyperbolic decline model or the AKB model; a type 2 gas well, Well-X2, is in the linear flow stage, the preferred models are SEPD decline and Duong decline model. When the error of the decline model is large, the production prediction accuracy of shale gas wells is effectively improved after coupling the LSTM model but the effect is not obvious when the error of the decline model is small.

Key words: normal shale gas reservoir, production decline method, LSTM model, flow stage, double logarithmic diagnosis, coupling method

中图分类号:

TE312

韩克宁, 王伟, 樊冬艳, 姚军, 罗飞, 杨灿. 基于产量递减与LSTM耦合的常压页岩气井产量预测[J]. 油气藏评价与开发, 2023, 13(5): 647-656.

HAN Kening, WANG Wei, FAN Dongyan, YAO Jun, LUO Fei, YANG Can. Production forecasting for normal pressure shale gas wells based on coupling of production decline method and LSTM model[J]. Petroleum Reservoir Evaluation and Development, 2023, 13(5): 647-656.

图/表 17

图1

图2

图3

图4

图5

表1

图6

图7

图8

表2

图9

图10

图11

表3

表4

图12

图13

参考文献 28

[1]	何希鹏, 张培先, 房大志, 等. 渝东南彭水—武隆地区常压页岩气生产特征[J]. 油气地质与采收率, 2018, 25(5): 72-79.
	HE Xipeng, ZHANG Peixian, FANG Dazhi, et al. Production characteristics of normal pressure shale gas in Pengshui-Wulong area, southeast Chongqing[J]. Petroleum Geology and Recovery Efficiency, 2018, 25(5): 72-79.
[2]	姜宇玲, 陈晓宇, 包汉勇. 页岩气分段压裂水平井产量递减快速预测新模型——以涪陵页岩气田为例[J]. 天然气地球科学, 2021, 32(6): 845-850.
	JIANG Yuling, CHEN Xiaoyu, BAO Hanyong. A new model for rapid prediction of horizontal well production decline in shale gas staged fracturing: Case study of Fuling shale gas field[J]. Natural Gas Geoscience, 2021, 32(6): 845-850.
[3]	付天宇, 刘启国, 岑雪芳, 等. 碳酸盐岩三重介质气藏NPI产量递减分析研究[J]. 油气藏评价与开发, 2021, 11(6): 905-910.
	FU Tianyu, LIU Qiguo, CEN Xuefang, et al. Normalized pressure integral production analysis of triporate-uniphase parallel inter-porosity flow model[J]. Petroleum Reservoir Evaluation and Development, 2021, 11(6): 905-910.
[4]	YU S Y. Best practice of using empirical methods for production forecast and EUR estimation in tight/shale gas reservoirs[C]// Paper SPE-167118-MS presented at the SPE Unconventional Resources Conference Canada, Calgary, Alberta, Canada, November 2013.
[5]	FULFORD D S, BLASINGAME T A. Evaluation of time- rate performance of shale wells using the transient hyperbolic relation[C]// Paper SPE-167242-MS presented at the SPE Unconventional Resources Conference, Calgary, Alberta, Canada, November 2013.
[6]	闫建平, 罗静超, 石学文, 等. 川南泸州地区奥陶系五峰组—志留系龙马溪组页岩裂缝发育模式及意义[J]. 岩性油气藏, 2022, 34(6): 60-71.
	YAN Jianping, LUO Jingchao, SHI Xuewen, et al. Fracture development models and significance of Ordovician Wufeng Silurian Longmaxi shale in Luzhou area, southern Sichuan Basin[J]. Lithologic Reservoirs, 2022, 34(6): 60-71.
[7]	王静怡, 周志军, 魏华彬, 等. 基于页岩孔隙网络模型的油水两相流动模拟[J]. 岩性油气藏, 2021, 33(5): 148-154.
	WANG Jingyi, ZHOU Zhijun, WEI Huabin, et al. Simulation of oil-water two-phase flow based on shale pore network model[J]. Lithologic Reservoirs, 2021, 33(5): 148-154.
[8]	CÉDRIC F G, OUASSIM K, SOHEIL E. Improved time series decline curve analysis for oil production using recurrent neural network[C]// Paper presented at the Artificial Intelligence Principles and Techniques, Stanford University, USA, 2019.
[9]	SUN J L, MA X, KAZI M. Comparison of decline curve analysis DCA with recursive neural networks RNN for production forecast of multiple wells[C]// Paper SPE-190101-MS presented at the SPE Western Regional Meeting, Garden Grove, California, USA, April 2018.
[10]	温康, 闫建平, 钟光海, 等. 川南长宁地区五峰组—龙马溪组页岩气评价新方法[J]. 岩性油气藏, 2022, 34(1): 95-105.
	WEN Kang, YAN Jianping, ZHONG Guanghai, et al. Application of modified unbiased grey model to the prediction of oil and gas field production[J]. Lithologic Reservoirs, 2022, 34(1): 95-105.
[11]	SONG X Y, LIU Y T, XUE L, et al. Time-series well performance prediction based on Long Short-Term Memory (LSTM) neural network model[J]. Journal of Petroleum Science and Engineering, 2020, 186: 106682. doi: 10.1016/j.petrol.2019.106682
[12]	SAGHEER A, KOTB M. Time series forecasting of petroleum production using deep LSTM recurrent networks[J]. Neurocomputing, 2019, 323: 203-213. doi: 10.1016/j.neucom.2018.09.082
[13]	LEE K, LIM J, YOON D, et al. Prediction of shale-gas production at Duvernay formation using deep-learning algorithm[J]. SPE Journal, 2019, 24(6): 2423-2437. doi: 10.2118/195698-PA
[14]	BAGHERI M, ZHAO H R, SUN M Y, et al. Data conditioning and forecasting methodology using machine learning on production data for a well pad[C]// Paper OTC-30854-MS presented at the Offshore Technology Conference, Houston, Texas, USA, May 2020.
[15]	IMAMVERDIYEV Y, ABDULLAYEVA F. Development of oil production forecasting method based on deep learning[J]. Statistics Optimization & Information Computing, 2019, 7(4): 826-839.
[16]	LIU W, LIU W D, GU J W. Forecasting oil production using ensemble empirical model decomposition based Long Short-Term Memory neural network[J]. Journal of Petroleum Science and Engineering, 2020, 189: 107013. doi: 10.1016/j.petrol.2020.107013
[17]	KARASU S, ALTAN A, BEKIROS S, et al. A new forecasting model with wrapper-based feature selection approach using multi-objective optimization technique for chaotic crude oil time series[J]. Energy 2020, 212: 118750. doi: 10.1016/j.energy.2020.118750
[18]	ARPS J J. Analysis of decline curves[J]. Petroleum Transactions, 1945, 160(1): 228-247.
[19]	SESHADRI J, MATTAR L. Comparison of power law and modified hyperbolic decline methods[C]// Paper SPE-137320-MS presented at the Canadian Unconventional Resources and International Petroleum Conference, Calgary, Alberta, Canada, October 2010.
[20]	ILK D, PEREGO A D, RUSHING J A, et al. Integrating multiple production analysis techniques to assess tight gas sand reserves: Defining a new paradigm for industry best practices[C]// Paper SPE-114947-MS presented at the CIPC/SPE Gas Technology Symposium 2008 Joint Conference, Calgary, Alberta, Canada, June 2008.
[21]	ILK D, RUSHING J A, PEREGO A D, et al. Exponential vs. hyperbolic decline in tight gas sands: understanding the origin and implications for reserve estimates using Arps' decline curves[C]// Paper SPE-116731-MS presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, September 2008.
[22]	VALKO P P, LEE W J. A better way to forecast production from unconventional gas wells[C]// Paper SPE-134231-MS presented at the SPE Annual Technical conference and Exhibition, Florence, Italy, September 2010.
[23]	VALKO P P. Assigning value to stimulation in the Barnett Shale:A simultaneous analysis of 7 000 plus production histories and well completion records[C]// Paper SPE-119369-MS presented at the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, USA, January 2009.
[24]	DUONG A N. Rate-decline analysis for fracture-dominated shale reservoirs[J]. SPE Reservoir Evaluation & Engineering, 2011, 14(3): 377-387.
[25]	DUONG A N. Rate-decline analysis for fracture dominated shale reservoirs: Part2[C]// Paper SPE-171610-MS presented at the SPE/CSUR Unconventional Resources Conference-Canada, Calgary, Alberta, Canada, September 2014.
[26]	YU S Y, LEE W J, MIOCEVIC D J, et al. Estimating proved reserves in tight/shale wells using the modified SEPD method[C]// Paper SPE-166198-MS presented at the SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, USA, September 2013.
[27]	ANSAH J, KNOWLES R S, BLASINGAME T A. A semi-analytic(p/z)rate time relation for the analysis and prediction of gas well performance[J]. SPE Reservoir Evaluation & Engineering, 2000, 3(6): 525-533.
[28]	HOCHREITER S, SCHMIDHUBER J. Long Short-Term Memory[J]. Neural Computation, 1997, 9(8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276

分析方法	检验平均误差/ %	5 a后日产气量/ m³	10 a 累产气量/ 10⁸ m³	可采储量/ 10⁸ m³	剩余可采储量/ 10⁸ m³
指数递减	5.43	15 068.31	1.01	1.06	0.76
双曲递减	6.41	16 330.13	1.02	1.65	1.35
改进双曲递减	3.75	18 449.74	1.08	1.79	1.49
SEPD	10.98	41 951.91	1.87	28.14	27.84
PLE	5.94	32 128.20	1.52	2.88	2.59
Duong递减	3.95	31 257.25	1.50	5.18	5.17
AKB递减	3.85	17 832.17	1.08	1.36	1.07

分析方法	检验平均误差/%	5 a后日产气量/ m³	10 a累产气量/ 10⁸ m³	可采储量/ 10⁸ m³	剩余可采储量/ 10⁸ m³
指数递减+LSTM	3.14	13 874.96	0.96	0.98	0.68
双曲递减+LSTM	3.84	20 096.66	1.14	1.96	1.44
改进双曲递减+LSTM	5.38	15 099.31	0.96	1.45	1.32
SEPD+LSTM	4.33	35 492.72	1.63	25.34	25.07
PLE+LSTM	6.23	29 230.75	1.42	2.75	2.41
Duong递减+LSTM	4.15	30 748.86	1.48	5.03	4.98
AKB递减+LSTM	3.99	19 444.62	1.13	1.53	1.36

序号	分析方法	检验平均误差/%	5 a后日产气量/m³	10 a累产气量/10⁸ m³	可采储量/10⁸ m³	剩余可采储量/10⁸ m³
1	指数递减	18.94	140	0.17	0.17	0.04
2	双曲递减	17.07	2 519	0.24	0.28	0.17
3	改进双曲递减	19.01	2 459	0.24	0.27	0.16
4	SEPD	13.30	2 075	0.23	0.23	0.12
5	PLE	15.43	1 579	0.21	0.21	0.1
6	Duong递减	13.54	4 327	0.43	0.43	0.31
7	AKB递减	23.47	1 257	0.20	0.20	0.08

序号	分析方法	检验平均误差/%	5 a后日产气量/m³	10 a累产气量/10⁸ m³	可采储量/10⁸ m³	剩余可采储量/10⁸ m³
1	指数递减+LSTM	3.14	3 147	0.17	0.19	0.06
2	双曲递减+LSTM	3.84	2 887	0.25	0.31	0.19
3	改进双曲递减+LSTM	5.38	2 798	0.25	0.30	0.17
4	SEPD+LSTM	4.33	3 051	0.26	0.27	0.13
5	PLE+LSTM	6.23	1 146	0.19	0.19	0.10
6	Duong递减+LSTM	4.15	4 660	0.44	0.46	0.35
7	AKB递减+LSTM	3.99	1 524	0.20	0.21	0.10