Porosity Conditioned Branch Encoding for Thermocline Tracking in Packed Bed Thermal Storage Using DeepONet
Authors
Predraga Lukić, Gavrila PajićFiles
Abstract
Thermocline displacement in packed bed thermal storage depends on inlet history and the pore volume available for fluid transport. This study proposes a DeepONet formulation with a porosity conditioned branch encoder and auxiliary front descriptors. The formulation separates inlet history encoding from bounded porosity modulation and couples temperature reconstruction with front position and thickness. Its evaluation uses analytic reference profiles and explicitly specified comparator perturbations to examine the behavior of field and front diagnostics. Five porosity levels from 0.32 to 0.48, six observation times, and 201 axial positions provide 6030 reference temperatures. Conditional comparator offsets are 0.008– 0.012 bed lengths, compared with 0.040–0.060 for the unconditioned series. At the central porosity, the specified offset decreases from 0.040 to 0.008 bed lengths, giving an 80% reduction within this controlled comparison. The corresponding width deviations are 4% and 15%, allowing position bias to be distinguished from transition broadening. These prescribed differences test diagnostic sensitivity; trained predictive accuracy remains a separate validation task. Profile reconstruction, threshold sensitivity, configuration variation, and computational cost scaling are examined through linked diagnostics. The contribution is an inspectable operator architecture and evaluation procedure rather than an experimentally established performance advantage. Subsequent validation should train matched baselines, withhold complete configurations and inlet histories, propagate sensor uncertainty, and measure inference latency under documented hardware and batching conditions.
References
Georgousis, N., Diriken, J., Speetjens, M. F. M., & Rindt, C. C. M. (2025). Comprehensive review on packed-bed sensible heat storage systems. Journal of Energy Storage, 121, Article 116516. https://doi.org/10.1016/j.est.2025.116516
Cheng, Q., Sahadath, M. H., Yang, H., Pan, S., & Ji, W. (2025). Surrogate modeling of heat transfer under flow fluctuation conditions using Fourier Basis-Deep Operator Network with uncertainty quantification. Progress in Nuclear Energy, 188, Article 105895. https://doi.org/10.1016/j.pnucene.2025.105895
Schwarzmayr, P., Birkelbach, F., Walter, H., & Hofmann, R. (2024). Exergy efficiency and thermocline degradation of a packed bed thermal energy storage in partial cycle operation: An experimental study. Applied Energy, 360, Article 122895. https://doi.org/10.1016/j.apenergy.2024.122895
Abddaim, E., Sakami, S., & Boukhattem, L. (2025). Numerical analysis of the impact of various parameters on the axial temperature distribution in a packed bed for thermocline energy storage. Thermal Science and Engineering Progress, 68, Article 104355. https://doi.org/10.1016/j.tsep.2025.104355
Yi, Y., & Nakayama, A. (2024). A three-energy equation model and estimation of effective thermal properties for transient analysis of bi-disperse packed bed thermocline storage system. Renewable Energy, 222, Article 119915. https://doi.org/10.1016/j.renene.2023.119915
He, J., Kushwaha, S., Park, J., Koric, S., Abueidda, D., & Jasiuk, I. (2024). Sequential Deep Operator Networks (S-DeepONet) for predicting full-field solutions under time-dependent loads. Engineering Applications of Artificial Intelligence, 127, Article 107258. https://doi.org/10.1016/j.engappai.2023.107258
Xu, C., Cao, B. T., Yuan, Y., & Meschke, G. (2024). A multi-fidelity deep operator network (DeepONet) for fusing simulation and monitoring data: Application to real-time settlement prediction during tunnel construction. Engineering Applications of Artificial Intelligence, 133, Article 108156. https://doi.org/10.1016/j.engappai.2024.108156
Huang, P., Leng, Y., Lian, C., & 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
Zheng, Z., Bai, X., & Nakayama, A. (2024). An analysis based on a three-energy equation model for forced convective heat transfer in bi-disperse packed bed channels and tubes. International Journal of Heat and Mass Transfer, 228, Article 125693. https://doi.org/10.1016/j.ijheatmasstransfer.2024.125693
Tong, Z., Wang, H., Tong, S., Yang, Q., & Nie, T. (2025). Thermocline performance in a molten salt thermocline energy storage tank with annular-arranged and cross-arranged diffusers. Journal of Zhejiang University-SCIENCE A, 26(4), 339–358. https://doi.org/10.1631/jzus.a2400359
Khurana, H., Majumdar, R., & Saha, S. K. (2024). Improved realistic stratification model for estimating thermocline thickness in vertical thermal energy storage undergoing simultaneous charging and discharging. Journal of Energy Storage, 82, Article 110490. https://doi.org/10.1016/j.est.2024.110490
Liu, L., Nath, K., & Cai, W. (2024). A Causality-DeepONet for causal responses of linear dynamical systems. Communications in Computational Physics, 35(5), 1194–1228. https://doi.org/10.4208/cicp.oa-2023-0078
He, J., Pal, D., Najafi, A., Abueidda, D., Koric, S., & Jasiuk, I. (2024). Material-response-informed DeepONet and its application to polycrystal stress–strain prediction in crystal plasticity. JOM, 76(10), 5744–5754. https://doi.org/10.1007/s11837-024-06681-
Zhu, Y., Chen, J., & Deng, W. (2025). R-adaptive DeepONet: Learning solution operators for PDEs with discontinuous solutions using an R-adaptive strategy. Numerical Mathematics: Theory, Methods and Applications, 18(1), 31–65. https://doi.org/10.4208/nmtma.oa-2024-0124
Zhang, C., Gan, Z., Jiang, Z., Xu, Z., & Yang, Q. (2026). An integrated surrogate and material optimization framework for PCM-enhanced energy piles based on EC-DeepONet. Energy, 360, Article 141529. https://doi.org/10.1016/j.energy.2026.141529
Pérez-Gallego, D., Gonzalez-Ayala, J., Medina, A., & Calvo Hernández, A. (2025). Comprehensive review of dynamical simulation models of packed-bed systems for thermal energy storage applications in renewable power production. Heliyon, 11(4), Article e42803. https://doi.org/10.1016/j.heliyon.2025.e42803
Moya, C., Mollaali, A., Zhang, Z., Lu, L., & Lin, G. (2025). Conformalized-DeepONet: A distribution-free framework for uncertainty quantification in deep operator networks. Physica D: Nonlinear Phenomena, 471, Article 134418. https://doi.org/10.1016/j.physd.2024.134418
Han, Z., Su, B., Zeng, X., & Xu, Z. (2026). Dynamic performance analysis of thermocline packed bed thermal energy storage system with hybrid structure. Energy Conversion and Management, 349, Article 120832. https://doi.org/10.1016/j.enconman.2025.120832
Shokrnia, M., Cagnoli, M., Gaggioli, W., Liberatore, R., Russo, V., & Zanino, R. (2024). Geometrical and PCM optimization of a thermocline energy storage system. Journal of Energy Storage, 98, Article 113070. https://doi.org/10.1016/j.est.2024.113070
Sun, A. Y., Jiang, P., Shuai, P., & Chen, X. (2024). Bridging hydrological ensemble simulation and learning using deep neural operators. Water Resources Research, 60(10), Article e2024WR037555. https://doi.org/10.1029/2024wr037555
Yang, Y. (2026). DeepONet for solving nonlinear partial differential equations with physics-informed training. Neural Networks, 197, Article 108490. https://doi.org/10.1016/j.neunet.2025.108490
Kothari, R., La Seta, A., Hemmingsen, C. S., Desai, N. B., & Haglind, F. (2024). Numerical analysis of measures to minimize the thermal instability in high temperature packed-beds for thermal energy storage systems. Journal of Energy Storage, 94, Article 112431. https://doi.org/10.1016/j.est.2024.112431
Rabi’, A. M., Radulovic, J., & Buick, J. M. (2024). Packed bed thermal energy storage system: Parametric study. Thermo, 4(3), 295–314. https://doi.org/10.3390/thermo4030016
Thorpe, G. R. (2025). Ventilated packed beds: The non-hygroscopic/hygroscopic continuum. Applied Thermal Engineering, 258, Article 124588. https://doi.org/10.1016/j.applthermaleng.2024.124588
Mandl, L., Goswami, S., Lambers, L., & Ricken, T. (2025). Separable physics-informed DeepONet: Breaking the curse of dimensionality in physics-informed machine learning. Computer Methods in Applied Mechanics and Engineering, 434, Article 117586. https://doi.org/10.1016/j.cma.2024.117586
Ouambo Tobou, R., Baudin, N., Lontsi, F., Soto, J., & Fan, Y. (2026). Air-based packed-bed thermocline thermal energy storage: Numerical and experimental study over a charge-standby-discharge cycle. Journal of Energy Storage, 159, Article 121554. https://doi.org/10.1016/j.est.2026.121554