Joint inversion for heat transfer parameters of porous media with geothermal-seepage coupling based on a physics-informed neural network

Authors

• ZHANG Xiao, National Institute of Natural Hazards, Ministry of Emergency Management, Beijing 100085, ChinaFollow
• YANG Duoxing, National Institute of Natural Hazards, Ministry of Emergency Management, Beijing 100085, ChinaFollow
• CONG Lin, CCTEG Xi’an Research Institute (Group) Co., Ltd., Xi’an 710077, China
• CHENG Yimei, School of Mechanics and Civil Engineering, China University of Mining and Technology (Beijing), Beijing 100083, China
• ZHANG Lianzhong, School of Physics, Nankai University, Tianjin 300071, China

Abstract

Background The thermal diffusivity and vertical seepage velocity of porous media are critical to characterizing the mechanisms underlying subsurface heat transfer. These two parameters are typically determined through inversion using temperature data since they are difficult to measure directly. Fiber-optic distributed temperature sensing (FO-DTS) can yield high-resolution, continuous temperature profiles, providing data for parameter identification. However, conventional numerical inversion suffers from limitations such as mesh dependency and sensitivity to initial values when used to process continuous monitoring data. Methods This study proposed a joint geothermal parameter inversion method that integrates FO-DTS data with a physics-informed neural network (PINN). Specifically, the steady-state convection-diffusion equation was embedded as a soft regularization term into the loss function of a PINN to provide a physical constraint. In this manner, the parameter inversion problem was transformed into a physics-guided optimization process, enabling the simultaneous estimation of vertical seepage velocity and thermal diffusivity. [Results and Conclusions] The results indicate that the temperature field derived from PINN-based inversion exhibited a root mean square error (RMSE) of 0.20 ℃ compared to observed temperature data. The thermal diffusivity and vertical seepage velocity derived from the inversion were roughly consistent with the results from the upwind finite difference-based inversion. Furthermore, the sensitivity analysis of key hyperparameters confirms the high robustness of the PINN-based inversion framework. The results of this study provide a novel approach characterized by both physical consistency and data adaptability for identifying parameters in geothermal-seepage coupling systems.

Keywords

physics-informed neural network (PINN), heat transfer, parameter inversion, porous medium, fiber-optic distributed temperature sensing (FO-DTS)

DOI

10.12363/issn.1001-1986.25.10.0804

Recommended Citation

ZHANG Xiao, YANG Duoxing, CONG Lin, et al. (2026) "Joint inversion for heat transfer parameters of porous media with geothermal-seepage coupling based on a physics-informed neural network," Coal Geology & Exploration: Vol. 54: Iss. 4, Article 17.
DOI: 10.12363/issn.1001-1986.25.10.0804
Available at: https://cge.researchcommons.org/journal/vol54/iss4/17

Reference

[1] BEAR J. Dynamics of fluids in porous media[M]. New York：American Elsevier Pub. Co，1972.

[2] 庞忠和,段忠丰. 华北赋煤区地热资源富集模式及开发利用方向[J]. 煤田地质与勘探,2024,52(9):14−22

PANG Zhonghe,DUAN Zhongfeng. Accumulation patterns and exploitation and utilization targets of geothermal resources in the coal–bearing area in North China[J]. Coal Geology & Exploration,2024,52(9):14−22

[3] ANDERSON M P. Heat as a ground water tracer[J]. Groundwater,2005,43(6):951−968.

[4] 何丽娟,胡圣标,杨文采,等. 中国大陆科学钻探主孔揭示的大陆地壳生热模型[J]. 岩石学报,2006,22(11):2808−2814

HE Lijuan,HU Shengbiao,YANG Wencai,et al. Heating generation model for the continental crust based on the main hole of the Chinese Continental Scientific Drilling Project[J]. Acta Petrologica Sinica,2006,22(11):2808−2814

[5] SAFFER D M,TOBIN H J. Hydrogeology and mechanics of subduction zone forearcs:Fluid flow and pore pressure[J]. Annual Review of Earth and Planetary Sciences,2011,39:157−186.

[6] 王亚宁，陆川，王贵玲. 岩石导热热阻对裂隙对流换热的影响机制[J]. 煤田地质与勘探，2023，51(3)：113–122.

WANG Yaning，LU Chuan，WANG Guiling. Influencing mechanism of rock thermal resistance on convective heat transfer of fracture[J]. Coal Geology & Exploration，2023，51(3)：113–122.

[7] 张延庆，谷同祥. 基于深度学习的扩散方程扩散系数反问题求解[J]. 计算物理，2025，42(2)：146–159.

ZHANG Yanqing，GU Tongxiang. Deep learning method for solving inverse problem of diffusion coefficients for diffusion equation[J]. Chinese Journal of Computational Physics，2025，42(2)：146–159.

[8] REITER M. Using precision temperature logs to estimate horizontal and vertical groundwater flow components[J]. Water Resources Research,2001,37(3):663−674.

[9] DAKIN J P,PRATT D J,BIBBY G W,et al. Distributed optical fibre Raman temperature sensor using a semiconductor light source and detector[J]. Electronics Letters,1985,21(13):569−570.

[10] TANGUDU R,SAHU P K. Review on the developments and potential applications of the fiber optic distributed temperature sensing system[J]. IETE Technical Review,2022,39(3):553−567.

[11] HENNINGES J,HUENGES E,BURKHARDT H. In situ thermal conductivity of gas–hydrate–bearing sediments of the Mallik 5L–38 well[J]. Journal of Geophysical Research:Solid Earth,2005,110(B11):2005JB003734.

[12] FREIFELD B M,FINSTERLE S,ONSTOTT T C,et al. Ground surface temperature reconstructions:Using in situ estimates for thermal conductivity acquired with a fiber–optic distributed thermal perturbation sensor[J]. Geophysical Research Letters,2008,35(14):2008GL034762.

[13] READ T,BOUR O,BENSE V,et al. Characterizing groundwater flow and heat transport in fractured rock using fiber–optic distributed temperature sensing[J]. Geophysical Research Letters,2013,40(10):2055−2059.

[14] TARANTOLA A,VALETTE B. Inverse problems = Quest for information[J]. Journal of Geophysics,1981,50(1):159−170.

[15] RAISSI M,PERDIKARIS P,KARNIADAKIS G E. Physics–informed neural networks:A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics,2019,378:686−707.

[16] TARTAKOVSKY A M,MARRERO C O,PERDIKARIS P,et al. Physics–informed deep neural networks for learning parameters and constitutive relationships in subsurface flow problems[J]. Water Resources Research,2020,56(5):e2019WR026731.

[17] CAI Shengze,WANG Zhicheng,WANG Sifan,et al. Physics–informed neural networks for heat transfer problems[J]. Journal of Heat Transfer,2021,143(6):060801.

[18] RASHT–BEHESHT M,HUBER C,SHUKLA K,et al. Physics–informed neural networks (PINNs) for wave propagation and full waveform inversions[J]. Journal of Geophysical Research:Solid Earth,2022,127(5):e2021JB023120.

[19] ALBERS A，GLATTING F，MENBERG K，et al. Depth-specific thermal conductivities of Triassic sandstones[J]. Geotherm Energy，2025，13(1)：34. https://doi.org/10.1186/s40517-025-00359-0.

[20] NEGRETE–ARANDA R,NEUMANN F,CONTRERAS J,et al. Transport of heat by hydrothermal circulation in a young rift setting:Observations from the Auka and JaichMaa Ja’ag’vent field in the Pescadero Basin,southern gulf of California[J]. Journal of Geophysical Research:Solid Earth,2021,126(8):e2021JB022300.

[21] 左玉妹,成建梅,赵锐锐,等. 英山地热田区深大断裂的控水控热作用及地热系统成因研究[J]. 水文地质工程地质,2024,51(4):220−232

ZUO Yumei,CHENG Jianmei,ZHAO Ruirui,et al. The impacts of deep faults on fluid migration,heat accumulation with implication to genesis of Yingshan geothermal system[J]. Hydrogeology and Engineering Geology,2024,51(4):220−232

[22] KLEPIKOVA M V,LE BORGNE T,BOUR O,et al. A methodology for using borehole temperature–depth profiles under ambient,single and cross–borehole pumping conditions to estimate fracture hydraulic properties[J]. Journal of Hydrology,2011,407(1/2/3/4):145−152.

[23] LU Lu,MENG Xuhui,MAO Zhiping,et al. DeepXDE:A deep learning library for solving differential equations[J]. SIAM Review,2021,63(1):208−228.

[24] KARNIADAKIS G E,KEVREKIDIS I G,LU Lu,et al. Physics–informed machine learning[J]. Nature Reviews Physics,2021,3(6):422−440.

[25] TÓTH J. A theoretical analysis of groundwater flow in small drainage basins[J]. Journal of Geophysical Research,1963,68(16):4795−4812.

[26] STALLMAN R W. Steady one–dimensional fluid flow in a semi–infinite porous medium with sinusoidal surface temperature[J]. Journal of Geophysical Research,1965,70(12):2821−2827.

[27] INGEBRITSEN S E，SANFORD W E，NEUZIL C E. Groundwater in geologic processes (2^nd Edition)[M]. Cambridge：Cambridge University Press，2023.

[28] RUMELHART D E,HINTON G E,WILLIAMS R J. Learning representations by back–propagating errors[J]. Nature,1986,323(6088):533−536.

[29] HORNIK K,STINCHCOMBE M,WHITE H. Multilayer feedforward networks are universal approximators[J]. Neural Networks,1989,2(5):359−366.

[30] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313(5786):504−507.

[31] JIN Lei,CHENG Jingchun,ZHANG Chunxi. Infrared pedestrian tracking with graph memory features[J]. IEEE Signal Processing Letters,2021,28:1933−1937.

[32] KURYLYK B L,MACQUARRIE K T B,MCKENZIE J M. Climate change impacts on groundwater and soil temperatures in cold and temperate regions:Implications,mathematical theory,and emerging simulation tools[J]. Earth–Science Reviews,2014,138:313−334.

[33] HARTOG A. A distributed temperature sensor based on liquid–core optical fibers[J]. Journal of Lightwave Technology,1983,1(3):498−509.

[34] HURTIG E,GROßWIG S,JOBMANN M,et al. Fibre–optic temperature measurements in shallow boreholes:Experimental application for fluid logging[J]. Geothermics,1994,23(4):355−364.

[35] MWAKANYAMALE K,SLATER L,DAY–LEWIS F,et al. Spatially variable stage–driven groundwater–surface water interaction inferred from time–frequency analysis of distributed temperature sensing data[J]. Geophysical Research Letters,2012,39(6):L06401.

[36] BAO Bo,MELO L,DAVIES B,et al. Detecting supercritical CO₂ in brine at sequestration pressure with an optical fiber sensor[J]. Environmental Science & Technology,2013,47(1):306−313.

[37] PAULSSON B N P，THORNBURG J A，HE Ruiqing. A fiber–optic borehole seismic vector and acoustic sensor system for geothermal site characterization and monitoring[C]//Proceedings World Geothermal Congress. Melbourne，2015.

[38] CONSTANTZ J. Heat as a tracer to determine streambed water exchanges[J]. Water Resources Research,2008,44(4):2008WR006996.

[39] ANIBAS C,FLECKENSTEIN J H,VOLZE N,et al. Transient or steady–state? Using vertical temperature profiles to quantify groundwater–surface water exchange[J]. Hydrological Processes,2009,23(15):2165−2177.

[40] WANG S,WANG H,PERDIKARIS P. On the eigenvector bias of Fourier feature networks:From regression to PDE solving[J]. Journal of Computational Physics,2021,446:110662.

[41] XIANG Zixue,PENG Wei,LIU Xu,et al. Self–adaptive loss balanced physics–informed neural networks[J]. Neurocomputing,2022,496:11−34.

[42] 邓龙传,李晓昭,吴云,等. 北山场址区不同尺度结构面导水特性研究[J]. 工程地质学报,2021,29(1):77−85

DENG Longchuan,LI Xiaozhao,WU Yun,et al. Study on water conductivity characteristics of different scale structure surfaces in Beishan site area[J]. Journal of Engineering Geology,2021,29(1):77−85

[43] STALLMAN R W. Computation of ground–water velocity from temperature data[C]//BENTALL R. Methods of collecting and interpreting ground–water data. U. S. Geological Survey Water–Supply Paper 1544–H，1963：36–46.

[44] CLAUSER C，HUENGES E. Thermal conductivity of rocks and minerals[M]//AHRENS T J. Rock physics & phase relations：A handbook of physical constants. Washington，D. C. ：American Geophysical Union，1995：105–126.

[45] 张德良. 计算流体力学教程[M]. 北京：高等教育出版社，2010.

[46] NOCEDAL J，WRIGHT S J. Numerical optimization[M]. New York：Springer，1999.

[47] MADDU S,STURM D,MÜLLER C L,et al. Inverse Dirichlet weighting enables reliable training of physics informed neural networks[J]. Machine Learning:Science and Technology,2022,3(1):15026.

[48] 杨多兴. 物理信息与光纤传感数据驱动地温场正反问题联合求解深度学习计算软件V1. 0[CP/CD]. 登记号：2025SR1442851，2025.

[49] WANG Sifan,YU Xinling,PERDIKARIS P. When and why PINNs fail to train:A neural tangent kernel perspective[J]. Journal of Computational Physics,2022,449:110768.

[50] HASHIMOTO S,LIN Weiren,ISHITSUKA K. Measurements of thermal conductivity of intact rocks by using cuttings[J]. Journal of the Society of Materials Science Japan,2023,72(11):828−835.

[51] 韩元红,吕俊,张育平,等. 秦岭山前眉县低温地热水及伴生氦气成因机制[J]. 煤田地质与勘探,2024,52(1):168−176

HAN Yuanhong,LYU Jun,ZHANG Yuping,et al. Genetic mechanisms of low–temperature geothermal water and its associated helium gas in Meixian County in the piedmont of Qinling Mountains[J]. Coal Geology & Exploration,2024,52(1):168−176

[52] 宋前进,王刚,许一川,等. 深层地热能开发及其对地热水流场的影响:以兰考县深层地热能开发为例[J]. 煤田地质与勘探,2021,49(2):184−193

SONG Qianjin,WANG Gang,XU Yichuan,et al. Development of deep geothermal energy and its influence on geothermal water flow field:Taking the development of deep geothermal energy in Lankao County for an example[J]. Coal Geology & Exploration,2021,49(2):184−193

[53] LIN Weiren,HIROSE T,TADAI O,et al. Thermal conductivity profile in the Nankai accretionary prism at IODP NanTroSEIZE site C0002:Estimations from high–pressure experiments using input site sediments[J]. Geochemistry,Geophysics,Geosystems,2020,21(7):e2020GC009108.

[54] 陈雯雯,饶松,孟博言,等. 基于三维地质建模的大安东部地区地热资源评价[J]. 煤田地质与勘探,2025,53(9):147−158

CHEN Wenwen,RAO Song,MENG Boyan,et al. Geothermal resource assessment of the eastern Da’an area based on 3D geothermal geological modeling[J]. Coal Geology & Exploration,2025,53(9):147−158

[55] JAGTAP A D,EM KARNIADAKIS G. Extended physics–informed neural networks (XPINNs):A generalized space–time domain decomposition based deep learning framework for nonlinear partial differential equations[J]. Communications in Computational Physics,2025,28(5):2002−2041.