Volume 19 Issue 2
May  2026
Turn off MathJax
Article Contents
Article Navigation >  Water Science and Engineering  > 2026  >  19(2): 302-314
Lu-hua Gu, Xi-jun Lai. 2026: Comparative performance evaluation of DeepONet architectures for dam-break hydrodynamic simulations. Water Science and Engineering, 19(2): 302-314. doi: 10.1016/j.wse.2026.01.004
Citation: Lu-hua Gu, Xi-jun Lai. 2026: Comparative performance evaluation of DeepONet architectures for dam-break hydrodynamic simulations. Water Science and Engineering, 19(2): 302-314. doi: 10.1016/j.wse.2026.01.004
shu

Comparative performance evaluation of DeepONet architectures for dam-break hydrodynamic simulations

doi: 10.1016/j.wse.2026.01.004

  • a. State Key Laboratory of Lake and Watershed Science for Water Security, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 211135, China;

  • b. University of Chinese Academy of Sciences, Beijing 100049, China;

  • c. Jiangsu Provincial Academy of Environmental Science, Nanjing 210036, China

Funds:

This work was supported by the National Key Research and Development Program of China (Grant No. 2024YFC3211700), the National Natural Science Foundation of China (Grant No. 42171012), and the Science and Technology Planning Project of the Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences (Grant No. NIGLAS2022GS07).

  • Received Date: 2025-08-18
  • Accepted Date: 2026-01-05
    • Available Online: 2026-05-30
    Abstract
  • Dam-break flows pose significant challenges for accurate and efficient hydrodynamic prediction. This study comprehensively evaluated three DeepONet architectures (vanilla DeepONet, physics-informed DeepONet (PI-DeepONet), and Markov DeepONet (M-DeepONet)) for simulating dam-break hydrodynamics. Extensive numerical experiments exhibited that vanilla DeepONet achieved the highest accuracy, with errors corresponding to only 37.9% and 17.5% of the errors produced by M-DeepONet and PI-DeepONet, respectively. Meanwhile, vanilla DeepONet maintained superior computational efficiency, with a computational duration equivalent to 17.19% of that required by the conventional numerical algorithm. However, its performance was highly dependent on the quantity of training data, with errors increasing by 95.08% as the sampling rate decreased from 100% to 10%. PI-DeepONet exhibited exceptional robustness under data-scarce conditions, with errors amounting to only 24.75% of those observed in vanilla DeepONet at 10% sampling, and it remained stable under high-noise conditions, showing just a 21.9% error increase. M-DeepONet balanced accuracy with sequential prediction capability but suffered from error accumulation in noisy scenarios. These findings reveal distinct trade-offs: vanilla DeepONet excels in data-rich high-accuracy applications, whereas PI-DeepONet is optimal when data are limited or corrupted by noise. This study provides actionable guidelines for model selection in disaster prediction systems, advances operator learning in hydrodynamics, and bridges critical gaps between theoretical development and practical engineering applications.

     

    • FullText(HTML)
  • loading
    • References(39)
  • [1]
    Abdelwahed, H.G., Abdelrahman, M.A.E., Alsarhan, A.F., Mohamed, K., 2024. Numerical simulating the blood flow model via nonhomogeneous Riemann solver scheme. Partial Differential Equations in Applied Mathematics 11, 100845. https://doi.org/10.1016/j.padiff.2024.100845.
    [2]
    Aureli, F., Maranzoni, A., Petaccia, G., 2024. Advances in dam-break modeling for flood hazard mitigation: Theory, numerical models, and applications in hydraulic engineering. Water 16(8), 1093. https://doi.org/10.3390/w16081093.
    [3]
    Barreau, M., Shen, H., 2025. Accuracy and robustness of weight-balancing methods for training PINNs. arXiv arXiv:2501.18582. https://doi.org/10.48550/arXiv.2501.18582.
    [4]
    Bullwinkel, B., Randle, D., Protopapas, P., Sondak, D., 2022. DEQGAN: Learning the loss function for PINNs with generative adversarial networks. arXiv arXiv:2209.07081. https://doi.org/10.48550/arXiv.2209.07081.
    [5]
    Cao, Q., Goswami, S., Tripura, T., Chakraborty, S., Em Karniadakis, G., 2024. Deep neural operators can predict the real-time response of floating offshore structures under irregular waves. Computers and Structures 291, 107228. https://doi.org/10.1016/j.compstruc.2023.107228.
    [6]
    Chen, Z., Badrinarayanan, V., Lee, C., Rabinovich, A., 2017. GradNorm: Gradient normalization for adaptive loss balancing in deep multitask networks. In: Proceedings of the 35th International Conference on Machine Learning. PMLR, Stockholm.
    [7]
    Cheung, K.C., See, S., 2021. Recent advance in machine learning for partial differential equation. CCF Transactions on High Performance Computing, 3, 298-310. https://doi.org/10.1007/s42514-021-00076-7.
    [8]
    De Jesus Crespo, R., Wu, J., Myer, M., Yee, S., Fulford, R., 2019. Flood protection ecosystem services in the coast of Puerto Rico: Associations between extreme weather, flood hazard mitigation and gastrointestinal illness. Science of The Total Environment 676, 343-355. https://doi.org/10.1016/j.scitotenv.2019.04.287.
    [9]
    Di Leoni, P.C., Lu, L., Charles, M., Em Karniadakis, G., Zaki, T.A., 2023. Neural operator prediction of linear instability waves in high-speed boundary layers. Journal of Computational Physics 474, 111793. https://doi.org/10.1016/j.jcp.2022.111793.
    [10]
    Formisano, A., Tucci, M., 2024. Machine learning approaches for inverse problems and optimal design in electromagnetism. Electronics 13(7), 1167. https://doi.org/10.3390/electronics13071167.
    [11]
    Han, J., Arnulf, J., Weinan, E., 2018. Solving high-dimensional partial differential equations using deep learning. Proceedings of the National Academy of Sciences of the United States of America 115, 8505-8510. https://doi.org/10.1073/pnas.1718942115.
    [12]
    He, J., Koric, S., Kushwaha, S., Park, J., Abueidda, D., Jasiuk, I., 2023. Novel DeepONet architecture to predict stresses in elastoplastic structures with variable complex geometries and loads. Computer Methods in Applied Mechanics and Engineering 415, 116277. https://doi.org/10.1016/j.cma.2023.116277.
    [13]
    He, J., Kushwaha, S., Park, J., Koric, S., Abueidda, D., Jasiuk, I., 2024. Predictions of transient vector solution fields with sequential deep operator network. Acta Mechanica 235, 5257-5272. https://doi.org/10.1007/s00707-024-03991-2.
    [14]
    Hu, L., Wang, X., 2024. A shock-stable numerical scheme accurate for contact discontinuities: Applications to 3D compressible flows. Communications in Nonlinear Science and Numerical Simulation 128, 107602. https://doi.org/10.1016/j.cnsns.2023.107602.
    [15]
    Hu, Y., Jiang, Y., Huang, X., Zhang, W., 2024. High-order weighted compact nonlinear scheme for solving degenerate parabolic equations. Computational and Applied Mathematics 43, 40. https://doi.org/10.1007/s40314-023-02551-z.
    [16]
    Jiang, H., Qiu, Y., Zhu, J., Wang, J., Zhang, C., Du, X., 2025. A SEM-FEM-SPH framework for physics-based source-to-slope simulation of earthquake-induced landslides. Engineering Geology 353, 108127. https://doi.org/10.1016/j.enggeo.2025.108127.
    [17]
    Kumar, V., Goswami, S., Kontolati, K., Shields, M.D., Karniadakis, G.E., 2025. Synergistic learning with multi-task DeepONet for efficient PDE problem solving. Neural Networks 184, 107113. https://doi.org/10.1016/j.neunet.2024.107113.
    [18]
    Kvocka, D., Falconer, R.A., Bray, M., 2016. Flood hazard assessment for extreme flood events. Natural Hazards 84, 1569-1599. https://doi.org/10.1007/s11069-016-2501-z.
    [19]
    Lai, X., Jiang, J., Liang, Q., Huang, Q., 2013a. Large-scale hydrodynamic modeling of the middle Yangtze River Basin with complex river-lake interactions, Journal of Hydrology 492, 228-243. https://doi.org/10.1016/j.jhydrol.2013.03.049.
    [20]
    Lai, X., Jiang, J., Yang, G., Lu, X.X., 2013b. Should the Three Gorges Dam be blamed for the extremely low water levels in the middle-lower Yangtze River? Hydrological Processes 28(1), 150-160. https://doi.org/10.1002/hyp.10077.
    [21]
    Lai, X., Liang, Q., Yesou, H., Daillet, S., 2014. Variational assimilation of remotely sensed flood extents using a 2-D flood model. Hydrology and Earth System Sciences 18, 4325-4339. https://doi.org/10.5194/hess-18-4325-2014.
    [22]
    Li, H., Miao,Y., Khodaei, Z.S., Aliabadi, M.H., 2025. An architectural analysis of DeepOnet and a general extension of the physics-informed DeepOnet model on solving nonlinear parametric partial differential equations. Neurocomputing 611, 128675. https://doi.org/10.1016/j.neucom.2024.128675.
    [23]
    Liu, H., Wang, Z., Zhang, D., Xiang, L., 2024a. An integrated model for dam break flood including reservoir area, breach evolution, and downstream flood propagation. Applied Sciences 14(23), 10921. https://doi.org/10.3390/app142310921.
    [24]
    Liu, J., Song, T., Mei, C., Wang, H., Zhang, D., Nazli, S., 2024b. Flood risk zoning of cascade reservoir dam break based on a 1D-2D coupled hydrodynamic model: A case study on the Jinsha-Yalong River. Journal of Hydrology 639, 131555. https://doi.org/10.1016/j.jhydrol.2024.131555.
    [25]
    Lu, L., Jin, P., Pang, G., Zhang, Z., Em Karniadakis, G., 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.
    [26]
    Reichstein, M., Camps-Valls, G., Stevens, B., Jung, M., Denzler, J., Carvalhais, N., Prabhat, 2019. Deep learning and process understanding for data-driven Earth system science. Nature 566, 195-204. https://doi.org/10.1038/s41586-019-0912-1.
    [27]
    Reyhanian, E., Dorschner, B., Karlin, I., 2023. Exploring shock-capturing schemes for particles on demand simulation of compressible flows, Computers & Fluids 263, 105947. https://doi.org/10.1016/j.compfluid.2023.105947.
    [28]
    Saeed, A.M, Alfawaz, T.A.F., 2025. Finite volume method and its applications in computational fluid dynamics. Axioms 14(5), 359. https://doi.org/10.3390/axioms14050359.
    [29]
    Son, H., Cho, S.W., Hwang, H.J., 2023. Enhanced physics-informed neural networks with augmented Lagrangian relaxation method (AL-PINNs). Neurocomputing 548, 126424. https://doi.org/10.1016/j.neucom.2023.126424.
    [30]
    Vivarelli, G., Qin, N., Shahpar, S., 2025. A review of mesh adaptation technology applied to computational fluid dynamics. Fluids 10(5), 129. https://doi.org/10.3390/fluids10050129.
    [31]
    Wang, S., Teng, Y., Perdikaris, P., 2021. Understanding and mitigating gradient flow pathologies in physics-informed neural networks. SIAM Journal on Scientific Computing 43(5), A3055-A3081. https://doi.org/10.1137/20M1318043.
    [32]
    Wang, S., Wang, H., Perdikaris, P., 2022. Improved architectures and training algorithms for deep operator networks. Journal of Scientific Computing 92, 35. https://doi.org/10.1007/s10915-022-01881-0.
    [33]
    Wang, S., Perdikaris, P., 2023. Long-time integration of parametric evolution equations with physics-informed DeepONets. Journal of Computational Physics 475, 111855. https://doi.org/10.1016/j.jcp.2022.111855.
    [34]
    Wang, X., Zhu, X., Meng, X., Zhu, Z., Zhang, S., Song, T., 2025. Inference and training acceleration of deep learning partial differential equation solver. The Journal of Supercomputing 81, 733. https://doi.org/10.1007/s11227-025-07204-y.
    [35]
    Zhang, D., Guo, L., Em Karniadakis, G., 2020. Learning in modal space: Solving time-dependent stochastic PDEs using physics-informed neural networks. SIAM Journal on Scientific Computing 42(2), A639-A665. https://doi.org/10.1137/19M1260141.
    [36]
    Zhao, K., Yang, S., Yuan, P., Shi, F., Ming, F., 2025. Numerical investigation of aerated water entry by multiphase Riemann SPH method. Ocean Engineering 333, 121244. https://doi.org/10.1016/j.oceaneng.2025.121244.
    [37]
    Zhao, Y., Zhao, J., Wang, Z., Lu, S., Zou, L., 2024. A comprehensive comparison study between deep operator networks neural network and long short-term memory for very short-term prediction of ship motion. Journal of Hydrodynamics 36, 1167-1180. https://doi.org/10.1007/s42241-025-0106-2.
    [38]
    Zhou, C., Cui, W., Sun, R., Huang, Y., Zhuang, Z., 2024. Enhancing the assimilation of SWOT simulated observations using a multi-scale 4DVAR method in regional ocean modeling system. Remote Sensing 16(5), 778. https://doi.org/10.3390/rs16050778.
    [39]
    Zylka, M., Gorski, G., Zylka, W., Gala-Bladzinska, A., 2024. Numerical analysis of blood flow in the abdominal aorta under simulated weightlessness and earth conditions. Scientific Reports 14, 15978. https://doi.org/10.1038/s41598-024-66961-7.
    • Relative Articles
    • Supplements(0)
    • Cited By
  • 加载中

Catalog

    通讯作者: 陈斌, bchen63@163.com
    • 1. 

      沈阳化工大学材料科学与工程学院 沈阳 110142

    1. 本站搜索
    2. 百度学术搜索
    3. 万方数据库搜索
    4. CNKI搜索

    Figures(1)

    Get Citation
    PDF
    XML

    Article Metrics

    Article views (34) PDF downloads(0) Cited by()
    Proportional views
    Related

    Copyright © 2009 Editorial office of Water Science and Engineering 苏ICP备12023610号-1

    Supported by: Beijing Renhe Information Technology Co. Ltd   support: info@rhhz.net

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return