Physics-informed reservoir surrogates are often evaluated by state error or equation residual alone, although neither establishes local conservation, symmetry consistency, or reliability under distribution shift. We separate these properties in a controlled two-dimensional incompressible oil-water benchmark. A flux-predictive U-Net outputs pressure, saturation increment, and oriented intercell total fluxes. Soft finite-volume residual regularization, exact minimum-norm cellwise flux projection, and horizontal/vertical reflection canonicalization are varied in a three-seed 2 × 2 × 2 factorial design across seven IID/OOD regimes. Exact local projection reduces saturation error in every regime and enforces cellwise balance to numerical precision but does not guarantee accurate pressure or long-horizon dynamics. Canonicalization gives large additional gains only for symmetry-equivalent well shifts: under combined geology plus symmetry OOD, saturation nRMSE decreases from 0.0216 without canonicalization to 0.00994 with the full method, whereas its incremental benefit is not significant for non-symmetry well or hard compound OOD after Holm correction. Converged FNO and PINO-style baselines remain less accurate in saturation/front metrics. Projector-forensics controls show learned fluxes reduce saturation error by 25% - 56% relative to a zero-flux projector-only baseline, while reference simulator fluxes require only approximately 10？6 relative correction. On a separately generated 56-case holdout, pre-projection correction magnitude predicts failure (Spearman rho = 0.648; AUROC = 0.894 for saturation nRMSE > 0.025; severe-OOD AUROC = 0.971). The results establish that conservation, symmetry, and residual consistency address distinct failure modes, and that conservation correction provides an interpretable physics-distance reliability signal.
References
[1]
Rasmussen, A.F., Sandve, T.H., Bao, K., Lauser, A., Hove, J., Skaflestad, B., et al. (2021) The Open Porous Media Flow Reservoir Simulator. Computers & Mathematics with Applications, 81, 159-185. https://doi.org/10.1016/j.camwa.2020.05.014
[2]
Raissi, M., Perdikaris, P. and Karniadakis, G.E. (2019) Physics-Informed Neural Networks: A Deep Learning Framework for Solving Forward and Inverse Problems Involving Nonlinear Partial Differential Equations. Journal of Computational Physics, 378, 686-707. https://doi.org/10.1016/j.jcp.2018.10.045
[3]
Karniadakis, G.E., Kevrekidis, I.G., Lu, L., Perdikaris, P., Wang, S. and Yang, L. (2021) Physics-Informed Machine Learning. Nature Reviews Physics, 3, 422-440. https://doi.org/10.1038/s42254-021-00314-5
[4]
Lu, L., Jin, P., Pang, G., Zhang, Z. and Karniadakis, G.E. (2021) Learning Nonlinear Operators via DeepONet Based on the Universal Approximation Theorem of Operators. Nature Machine Intelligence, 3, 218-229. https://doi.org/10.1038/s42256-021-00302-5
[5]
Azizzadenesheli, K., Kovachki, N., Li, Z., Liu-Schiaffini, M., Kossaifi, J. and Anandkumar, A. (2024) Neural Operators for Accelerating Scientific Simulations and Design. Nature Reviews Physics, 6, 320-328. https://doi.org/10.1038/s42254-024-00712-5
[6]
Subedi, U. and Tewari, A. (2026) Operator Learning: A Statistical Perspective. Annual Review of Statistics and Its Application, 13, 123-148. https://doi.org/10.1146/annurev-statistics-042424-070908
[7]
Goswami, S., Kontolati, K., Shields, M.D. and Karniadakis, G.E. (2022) Deep Transfer Operator Learning for Partial Differential Equations under Conditional Shift. Nature Machine Intelligence, 4, 1155-1164. https://doi.org/10.1038/s42256-022-00569-2
[8]
Lara Benitez, J.A., Furuya, T., Faucher, F., Kratsios, A., Tricoche, X. and de Hoop, M.V. (2024) Out-of-Distributional Risk Bounds for Neural Operators with Applications to the Helmholtz Equation. Journal of Computational Physics, 513, Article 113168. https://doi.org/10.1016/j.jcp.2024.113168
[9]
Fang, Z., Wang, S. and Perdikaris, P. (2024) Learning Only on Boundaries: A Physics-Informed Neural Operator for Solving Parametric Partial Differential Equations in Complex Geometries. Neural Computation, 36, 475-498. https://doi.org/10.1162/neco_a_01647
[10]
Wen, G., Li, Z., Azizzadenesheli, K., Anandkumar, A. and Benson, S.M. (2022) U-FNO—An Enhanced Fourier Neural Operator-Based Deep-Learning Model for Multiphase Flow. Advances in Water Resources, 163, Article 104180. https://doi.org/10.1016/j.advwatres.2022.104180
[11]
Jiang, Z., Zhu, M. and Lu, L. (2024) Fourier-MIONet: Fourier-Enhanced Multiple-Input Neural Operators for Multiphase Modeling of Geological Carbon Sequestration. Reliability Engineering & System Safety, 251, Article 110392. https://doi.org/10.1016/j.ress.2024.110392
[12]
Badawi, D. and Gildin, E. (2025) Neural Operator-Based Proxy for Reservoir Simulations Considering Varying Well Settings, Locations, and Permeability Fields. Computers & Geosciences, 196, Article 105826. https://doi.org/10.1016/j.cageo.2024.105826
[13]
Ma, X., Zhong, R., Zhan, J. and Zhou, D. (2024) Enhancing Subsurface Multiphase Flow Simulation with Fourier Neural Operator. Heliyon, 10, e38103. https://doi.org/10.1016/j.heliyon.2024.e38103
[14]
Kadeethum, T., Verzi, S.J. and Yoon, H. (2024) An Improved Neural Operator Framework for Large-Scale CO2 Storage Operations. Geoenergy Science and Engineering, 240, Article 213007. https://doi.org/10.1016/j.geoen.2024.213007
[15]
Santos, E.S., Barros, G.F., Oliveira, A.C.N., Silva, R.M., Freitas, R.S.M., Valiveti, D.M., et al. (2026) Hybrid DeepONet Surrogates for Multiphase Flow in Porous Media. Journal of Computational Physics, 561, Article 114981. https://doi.org/10.1016/j.jcp.2026.114981
[16]
Huang, P., Leng, Y., Lian, C. and Liu, H. (2024) Porous-DeepONet: Learning the Solution Operators of Parametric Reactive Transport Equations in Porous Media. Engineering, 39, 94-103. https://doi.org/10.1016/j.eng.2024.07.002
[17]
Diab, W., Chaabi, O., Alkobaisi, S., Awotunde, A. and Al Kobaisi, M. (2024) Learning Generic Solutions for Multiphase Transport in Porous Media via the Flux Functions Operator. Advances in Water Resources, 183, Article 104609. https://doi.org/10.1016/j.advwatres.2023.104609
[18]
Yan, X., Lin, J., Ju, Y., Zhang, Q., Zhang, Z., Zhang, L., et al. (2025) A Finite-Volume Based Physics-Informed Fourier Neural Operator Network for Parametric Learning of Subsurface Flow. Advances in Water Resources, 205, Article 105087. https://doi.org/10.1016/j.advwatres.2025.105087
[19]
Zhang, Z., Yan, X., Liu, P., Zhang, K., Han, R. and Wang, S. (2023) A Physics-Informed Convolutional Neural Network for the Simulation and Prediction of Two-Phase Darcy Flows in Heterogeneous Porous Media. Journal of Computational Physics, 477, Article 111919. https://doi.org/10.1016/j.jcp.2023.111919
[20]
Yan, X., Lin, J., Wang, S., Zhang, Z., Liu, P., Sun, S., et al. (2024) Physics-Informed Neural Network Simulation of Two-Phase Flow in Heterogeneous and Fractured Porous Media. Advances in Water Resources, 189, Article 104731. https://doi.org/10.1016/j.advwatres.2024.104731
[21]
Hanna, J.M., Aguado, J.V., Comas-Cardona, S., Askri, R. and Borzacchiello, D. (2022) Residual-Based Adaptivity for Two-Phase Flow Simulation in Porous Media Using Physics-Informed Neural Networks. Computer Methods in Applied Mechanics and Engineering, 396, Article 115100. https://doi.org/10.1016/j.cma.2022.115100
[22]
Almajid, M.M. and Abu-Al-Saud, M.O. (2022) Prediction of Porous Media Fluid Flow Using Physics Informed Neural Networks. Journal of Petroleum Science and Engineering, 208, Article 109205. https://doi.org/10.1016/j.petrol.2021.109205
[23]
Wang, N., Chang, H. and Zhang, D. (2022) Surrogate and Inverse Modeling for Two-Phase Flow in Porous Media via Theory-Guided Convolutional Neural Network. Journal of Computational Physics, 466, Article 111419. https://doi.org/10.1016/j.jcp.2022.111419
[24]
Yan, B., Harp, D.R., Chen, B., Hoteit, H. and Pawar, R.J. (2022) A Gradient-Based Deep Neural Network Model for Simulating Multiphase Flow in Porous Media. Journal of Computational Physics, 463, Article 111277. https://doi.org/10.1016/j.jcp.2022.111277
[25]
Yan, B., Harp, D.R., Chen, B. and Pawar, R. (2022) A Physics-Constrained Deep Learning Model for Simulating Multiphase Flow in 3D Heterogeneous Porous Media. Fuel, 313, Article 122693. https://doi.org/10.1016/j.fuel.2021.122693
[26]
Yan, B., Chen, B., Robert Harp, D., Jia, W. and Pawar, R.J. (2022) A Robust Deep Learning Workflow to Predict Multiphase Flow Behavior during Geological CO2 Sequestration Injection and Post-Injection Periods. Journal of Hydrology, 607, Article 127542. https://doi.org/10.1016/j.jhydrol.2022.127542
[27]
Li, J., Zhang, D., He, T. and Zheng, Q. (2023) Uncertainty Quantification of Two-Phase Flow in Porous Media via the Coupled-TgNN Surrogate Model. Geoenergy Science and Engineering, 221, Article 211368. https://doi.org/10.1016/j.geoen.2022.211368
[28]
Chakraborty, A., Rabinovich, A. and Moreno, Z. (2024) Physics-Informed Neural Networks for Modeling Two-Phase Steady State Flow with Capillary Heterogeneity at Varying Flow Conditions. Advances in Water Resources, 185, Article 104639. https://doi.org/10.1016/j.advwatres.2024.104639
[29]
Alhubail, A., Fahs, M., Lehmann, F. and Hoteit, H. (2024) Modeling Fluid Flow in Heterogeneous Porous Media with Physics-Informed Neural Networks: Weighting Strategies for the Mixed Pressure Head-Velocity Formulation. Advances in Water Resources, 193, Article 104797. https://doi.org/10.1016/j.advwatres.2024.104797
[30]
Basha, N., Arcucci, R., Angeli, P., Anastasiou, C., Abadie, T., Casas, C.Q., et al. (2024) Machine Learning and Physics-Driven Modelling and Simulation of Multiphase Systems. International Journal of Multiphase Flow, 179, Article 104936. https://doi.org/10.1016/j.ijmultiphaseflow.2024.104936
[31]
Zhu, H., Chen, Z., Gao, X., Chen, X., Yu, W. and Sepehrnoori, K. (2025) A Physics-Informed Adaptive-Wavelet Neural Network (PIAWNN) for Reservoir Simulation with Two-Phase Flow. Advances in Water Resources, 207, Article 105174. https://doi.org/10.1016/j.advwatres.2025.105174
[32]
Wang, Q., Zha, W., Li, D., Li, X., Shen, L. and Shi, Z. (2026) Parameterized Analytical Solution of Seepage Equation for Reservoir Simulation Using Physics-Informed Kolmogorov-Arnold Network without Labels. Journal of Computational Physics, 548, Article 114599. https://doi.org/10.1016/j.jcp.2025.114599
[33]
Wang, Q., Li, X., Li, D., Zha, W. and Shen, L. (2026) Asymptotic Solution Neural Network for Solving Multi-Well Seepage Equation with Deep Integration of Superposition Principle and Laplacian Distance Decay Weighting in Reservoir Simulation. Journal of Computational Physics, 558, Article 114845. https://doi.org/10.1016/j.jcp.2026.114845
[34]
Lin, J., Yan, X., Zhang, K., Zhang, Q., Zhang, Z., Zhang, L., et al. (2026) Deep Learning-Based Upscaling for Reservoir Models on Corner-Point Grids via Finite-Volume Physics-Informed Fourier Neural Operator. Petroleum Science, 23, 5662-5692. https://doi.org/10.1016/j.petsci.2026.04.037
[35]
Donnelly, J., Daneshkhah, A. and Abolfathi, S. (2024) Physics-Informed Neural Networks as Surrogate Models of Hydrodynamic Simulators. Science of The Total Environment, 912, Article 168814. https://doi.org/10.1016/j.scitotenv.2023.168814
[36]
Gao, H., Sun, L. and Wang, J. (2021) Phygeonet: Physics-Informed Geometry-Adaptive Convolutional Neural Networks for Solving Parameterized Steady-State PDEs on Irregular Domain. Journal of Computational Physics, 428, Article 110079. https://doi.org/10.1016/j.jcp.2020.110079
[37]
Ranade, R., Hill, C. and Pathak, J. (2021) Discretizationnet: A Machine-Learning Based Solver for Navier-Stokes Equations Using Finite Volume Discretization. Computer Methods in Applied Mechanics and Engineering, 378, Article 113722. https://doi.org/10.1016/j.cma.2021.113722
[38]
Sun, L., Gao, H., Pan, S. and Wang, J. (2020) Surrogate Modeling for Fluid Flows Based on Physics-Constrained Deep Learning without Simulation Data. Computer Methods in Applied Mechanics and Engineering, 361, Article 112732. https://doi.org/10.1016/j.cma.2019.112732
[39]
Wang, S., Teng, Y. and Perdikaris, P. (2021) Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks. SIAM Journal on Scientific Computing, 43, A3055-A3081. https://doi.org/10.1137/20m1318043
[40]
Yu, J., Lu, L., Meng, X. and Karniadakis, G.E. (2022) Gradient-Enhanced Physics-Informed Neural Networks for Forward and Inverse PDE Problems. Computer Methods in Applied Mechanics and Engineering, 393, Article 114823. https://doi.org/10.1016/j.cma.2022.114823
[41]
Wu, C., Zhu, M., Tan, Q., Kartha, Y. and Lu, L. (2023) A Comprehensive Study of Non-Adaptive and Residual-Based Adaptive Sampling for Physics-Informed Neural Networks. Computer Methods in Applied Mechanics and Engineering, 403, Article 115671. https://doi.org/10.1016/j.cma.2022.115671
[42]
Jagtap, A.D., Kharazmi, E. and Karniadakis, G.E. (2020) Conservative Physics-Informed Neural Networks on Discrete Domains for Conservation Laws: Applications to Forward and Inverse Problems. Computer Methods in Applied Mechanics and Engineering, 365, Article 113028. https://doi.org/10.1016/j.cma.2020.113028
[43]
Cardoso-Bihlo, E. and Bihlo, A. (2025) Exactly Conservative Physics-Informed Neural Networks and Deep Operator Networks for Dynamical Systems. Neural Networks, 181, Article 106826. https://doi.org/10.1016/j.neunet.2024.106826
[44]
Gerken, J.E., Aronsson, J., Carlsson, O., Linander, H., Ohlsson, F., Petersson, C., et al. (2023) Geometric Deep Learning and Equivariant Neural Networks. Artificial Intelligence Review, 56, 14605-14662. https://doi.org/10.1007/s10462-023-10502-7
[45]
Celledoni, E., Ehrhardt, M.J., Etmann, C., Owren, B., Sch？nlieb, C. and Sherry, F. (2021) Equivariant Neural Networks for Inverse Problems. Inverse Problems, 37, Article 085006. https://doi.org/10.1088/1361-6420/ac104f
[46]
Ma, G., Wang, Y., Lim, D., Jegelka, S. and Wang, Y. (2024) A Canonicalization Perspective on Invariant and Equivariant Learning. Advances in Neural Information Processing Systems 37, Vancouver, 10-15 December 2024, 60936-60979. https://doi.org/10.52202/079017-1948
[47]
Bulusu, S., Favoni, M., Ipp, A., Müller, D.I. and Schuh, D. (2021) Generalization Capabilities of Translationally Equivariant Neural Networks. Physical Review D, 104, Article 074504. https://doi.org/10.1103/physrevd.104.074504
[48]
Xu, Y., Kohtz, S., Boakye, J., Gardoni, P. and Wang, P. (2023) Physics-Informed Machine Learning for Reliability and Systems Safety Applications: State of the Art and Challenges. Reliability Engineering & System Safety, 230, Article 108900. https://doi.org/10.1016/j.ress.2022.108900
[49]
Psaros, A.F., Meng, X., Zou, Z., Guo, L. and Karniadakis, G.E. (2023) Uncertainty Quantification in Scientific Machine Learning: Methods, Metrics, and Comparisons. Journal of Computational Physics, 477, Article 111902. https://doi.org/10.1016/j.jcp.2022.111902
[50]
Guo, L., Wu, H., Wang, Y., Zhou, W. and Zhou, T. (2024) IB-UQ: Information Bottleneck Based Uncertainty Quantification for Neural Function Regression and Neural Operator Learning. Journal of Computational Physics, 510, Article 113089. https://doi.org/10.1016/j.jcp.2024.113089
