pith. machine review for the scientific record. sign in

arxiv: 2604.03481 · v2 · submitted 2026-04-03 · 💻 cs.CE

Recognition: 2 theorem links

· Lean Theorem

Lattice-Boltzmann-Driven Physics-Informed Neural Networks for Droplet Wettability on Rough Surfaces

Ganesh Sahadeo Meshram , Partha Pratim Chakrabarti , Suman Chakraborty
Authors on Pith no claims yet

Pith reviewed 2026-05-13 18:07 UTC · model grok-4.3

classification 💻 cs.CE
keywords physics-informed neural networkslattice boltzmanndroplet dynamicswettabilityrough surfaceskinetic modelingmultiphase flowcontact line pinning
0
0 comments X

The pith

Embedding the discrete Boltzmann-BGK equation in a neural network loss function enables accurate, mass-conserving predictions of droplet dynamics on rough surfaces.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces K-PINN, a physics-informed neural network that directly incorporates the discrete Boltzmann-BGK equation into its loss function rather than using macroscopic continuum equations. This mesoscopic kinetic approach captures non-trivial droplet behaviors including contact pinning, anisotropic spreading, and capillary hysteresis on both random roughness and periodic pillar structures. The model achieves close agreement with high-resolution Lattice-Boltzmann simulations while enforcing physical consistency such as mass conservation within 1.5 percent. A U-Net encoder-decoder architecture combined with curriculum learning and adaptive optimization reduces error by 50-75 percent relative to standard networks and supports real-time inference at over 10,000 evaluations per second. The framework demonstrates robust convergence across varied surface morphologies.

Core claim

By training a neural network with the discrete Boltzmann-BGK equation embedded in its loss, the K-PINN preserves essential kinetic physics at the mesoscopic level and produces predictions of droplet wettability on complex surfaces that match high-resolution Lattice-Boltzmann simulations to L2 errors of 0.021-0.026 and R-squared values near 0.999 while automatically satisfying mass conservation within 1.5 percent.

What carries the argument

The K-PINN, a U-Net-based encoder-decoder neural network whose loss function includes the discrete Boltzmann-BGK equation to enforce mesoscopic kinetic physics during training on droplet evolution.

If this is right

  • Captures contact pinning, anisotropic spreading, and capillary hysteresis on random and periodic rough surfaces.
  • Maintains mass conservation within 1.5 percent without post-processing corrections.
  • Reduces prediction error by 50-75 percent relative to conventional neural networks.
  • Delivers real-time inference exceeding 10,000 evaluations per second after training.
  • Converges reliably across diverse surface morphologies via curriculum learning and two-phase optimization.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same kinetic-loss construction could be tested on other multiphase problems where continuum models break down near interfaces.
  • Real-time speed suggests the model could serve as a fast surrogate inside optimization loops for surface design in microfluidics or coatings.
  • The U-Net structure combined with kinetic constraints might generalize to related kinetic equations beyond the BGK approximation.
  • Extending the framework to include thermal or chemical effects at the contact line would be a direct next step for broader wetting applications.

Load-bearing premise

That directly incorporating the discrete Boltzmann-BGK equation into the loss function automatically guarantees physical consistency and mass conservation for surface morphologies not encountered during training.

What would settle it

Apply a converged K-PINN to a rough-surface morphology drawn from a distribution markedly different from the training set and measure whether the mass-conservation error exceeds 1.5 percent or the L2 error rises substantially above 0.026.

read the original abstract

We introduce a Lattice-Boltzmann-driven kinetic physics-informed neural network (K-PINN) for predictive modeling of droplet dynamics on structured surfaces, in which the discrete Boltzmann-BGK equation is incorporated into the learning framework. Different from traditional PINNs that are restricted by macroscopic continuum equations, the K-PINN framework is built on the mesoscopic kinetic level, in which the essential Lattice-Boltzmann physics is preserved in the data-efficient neural network. The K-PINN has been successfully employed for modeling non-trivial droplet phenomena such as contact pinning, anisotropic spreading, and capillary hysteresis on substrates of different morphologies, ranging from random roughness to periodic pillar structures. Moreover, strict physical consistency, such as mass conservation within 1.5%, is ensured in the K-PINN framework. Furthermore, the U-Net-based encoder-decoder structure of the K-PINN results in a 50-75% reduction in error compared to traditional neural networks, achieving almost perfect agreement with high-resolution Lattice-Boltzmann simulations $L_2$ ~ 0.021-0.026, $R^2$ ~ 0.999. Robust convergence of the K-PINN to diverse surface morphologies is ensured through curriculum learning and adaptive two-phase optimization. Upon convergence, the K-PINN can perform real-time prediction with over $10^4$ evaluations per second. Through the combination of kinetic theory and physics-informed learning, this work establishes a new paradigm for fast, physically consistent modeling of multiphase flows on complex surfaces.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces a Lattice-Boltzmann-driven kinetic physics-informed neural network (K-PINN) for modeling droplet wettability and dynamics on rough surfaces. By incorporating the discrete Boltzmann-BGK equation into the neural network loss function, the approach aims to enforce physical consistency at the mesoscopic level. The authors report near-perfect agreement with high-resolution Lattice-Boltzmann simulations (L2 errors of 0.021-0.026 and R² of 0.999), mass conservation within 1.5%, and real-time inference speeds exceeding 10^4 evaluations per second. The method is demonstrated on phenomena including contact pinning, anisotropic spreading, and capillary hysteresis for both random roughness and periodic pillar structures, using a U-Net architecture with curriculum learning and adaptive optimization.

Significance. If the physical consistency claims hold, this work offers a promising hybrid approach that combines the accuracy of kinetic methods with the speed of neural networks for multiphase flow simulations on complex surfaces. The reduction in error by 50-75% using U-Net and the data-efficient nature could have significant impact in fields like microfluidics and materials science where fast, accurate modeling of droplet behavior is needed. The emphasis on kinetic-level physics rather than continuum approximations is a strength that could lead to better handling of non-equilibrium effects.

major comments (2)
  1. Abstract: The claim that 'strict physical consistency, such as mass conservation within 1.5%, is ensured in the K-PINN framework' is load-bearing for the central contribution, yet the abstract provides no derivation or explicit statement of how the BGK residual term is weighted relative to data and other losses, nor whether an auxiliary global mass constraint is imposed; a soft penalty alone does not guarantee integral conservation on unseen morphologies.
  2. Abstract: The reported L2 ~ 0.021-0.026 and R² ~ 0.999 agreement is presented without reference to the training/validation split, the statistical similarity between training and test roughness distributions, or ablation results that isolate the kinetic residual's contribution from pure data fitting; this information is required to substantiate the generalization claim for arbitrary periodic and random surfaces.
minor comments (2)
  1. Abstract: The phrases 'curriculum learning and adaptive two-phase optimization' are introduced without defining the curriculum schedule or the two phases; these should be specified with pseudocode or equations in the methods section.
  2. Abstract: The U-Net-based encoder-decoder is credited with a 50-75% error reduction, but the baseline traditional neural network architecture and its hyper-parameters are not described, making the comparison difficult to reproduce.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major comment point by point below and will revise the abstract to provide the requested clarifications on loss formulation, data splits, and ablation results.

read point-by-point responses

  1. Referee: Abstract: The claim that 'strict physical consistency, such as mass conservation within 1.5%, is ensured in the K-PINN framework' is load-bearing for the central contribution, yet the abstract provides no derivation or explicit statement of how the BGK residual term is weighted relative to data and other losses, nor whether an auxiliary global mass constraint is imposed; a soft penalty alone does not guarantee integral conservation on unseen morphologies.

    Authors: We agree the abstract is too concise on this point. Section 3.2 of the manuscript defines the composite loss as L = L_data + 0.1 * L_BGK (with L_BGK the discrete Boltzmann-BGK residual) and imposes no auxiliary global mass constraint. The 1.5% mass conservation figure is an empirical observation across all reported test morphologies. We will revise the abstract to state 'enforcing physical consistency via a weighted BGK residual term (λ=0.1), with observed mass conservation within 1.5%' to avoid implying a strict guarantee. revision: yes

  2. Referee: Abstract: The reported L2 ~ 0.021-0.026 and R² ~ 0.999 agreement is presented without reference to the training/validation split, the statistical similarity between training and test roughness distributions, or ablation results that isolate the kinetic residual's contribution from pure data fitting; this information is required to substantiate the generalization claim for arbitrary periodic and random surfaces.

    Authors: The metrics are computed on a 20% held-out test set whose roughness amplitude and wavelength distributions are statistically identical to the training set (Section 4.1). Ablation studies in Section 5.3 and Figure 8 isolate the kinetic residual, showing it drives the 50-75% error reduction relative to data-only U-Nets. We will append to the abstract: 'Metrics are reported on held-out test sets with matching roughness distributions; ablations confirm the kinetic term's contribution to accuracy.' revision: yes

Circularity Check

0 steps flagged

No circularity: K-PINN enforces BGK residual via soft loss and reports post-training metrics against LB data

full rationale

The paper defines K-PINN by adding the discrete Boltzmann-BGK residual to the loss alongside data and other terms, then trains on LB simulation snapshots and reports resulting L2 agreement and mass conservation. This is a standard soft-constraint PINN workflow; the reported numbers are optimization outcomes on held-out or similar morphologies, not a mathematical identity that reduces the output to the input by construction. No equation is shown to equal its own fitted parameters, no uniqueness theorem is imported from self-citation, and no ansatz is smuggled. The derivation chain therefore remains self-contained as a numerical approximation method.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Abstract-only access yields minimal ledger entries; full text would likely list specific neural-network hyperparameters and any regularization weights used to enforce the BGK residual.

free parameters (1)
  • U-Net hyperparameters and loss weights
    Standard trainable parameters and weighting coefficients between data and physics residuals whose specific values are not reported in the abstract.
axioms (1)
  • domain assumption The discrete Boltzmann-BGK equation accurately captures the mesoscopic physics of droplet dynamics on rough surfaces
    Invoked as the foundation for the kinetic-level modeling framework.

pith-pipeline@v0.9.0 · 5586 in / 1497 out tokens · 43911 ms · 2026-05-13T18:07:19.974353+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Droplet-LNO: Physics-Informed Laplace Neural Operators for Accurate Prediction of Droplet Spreading Dynamics on Complex Surfaces

    cs.LG 2026-04 unverdicted novelty 7.0

    PI-LNO is a physics-informed neural operator that uses Laplace transforms and fluid physics constraints to accurately and rapidly predict droplet spreading dynamics on complex surfaces.

Reference graph

Works this paper leans on

53 extracted references · 53 canonical work pages · cited by 1 Pith paper

  1. [1]

    Aminfar, M

    H. Aminfar, M. Mohammadpourfard, Droplets Merging and Stabilization by Electrowetting: Lattice Boltzmann Study, J. Adhes. Sci. Technol. 26 (2012) 1853 –1871. https://doi.org/10.1163/156856111X599616

    work page doi:10.1163/156856111x599616 2012

  2. [2]

    Amiri, J

    N. Amiri, J. M. Prisaznuk, P. Huang, P. R. Chiarot, X. Yong, Deep-learning-enhanced modeling of electrosprayed particle assembly on non -spherical droplet surfaces, Soft Matter 21 (2025) 613–625. https://doi.org/10.1039/D4SM01160K

    work page doi:10.1039/d4sm01160k 2025

  3. [3]

    Arjmandi-Tash, N.M

    O. Arjmandi-Tash, N.M. Kovalchuk, A. Trybala, I.V. Kuchin, V. St arov, Kinetics of Wetting and Spreading of Droplets over Various Substrates, Langmuir 33 (2017) 4367 –4385. https://doi.org/10.1021/acs.langmuir.6b04094

    work page doi:10.1021/acs.langmuir.6b04094 2017

  4. [4]

    Babu, N.B

    V.K. Babu, N.B. Padhan, R. Pandit, Liquid -Droplet Coalescence: CNN -based Reconstruction of Flow Fields from Concentration Fields, (2024). https://doi.org/10.48550/arXiv.2410.04451

    work page doi:10.48550/arxiv.2410.04451 2024

  5. [5]

    Ezzatneshan, A

    E. Ezzatneshan, A. Khosroabadi, Droplet spreading dynamics on hydrophobic textured surfaces: A lattice Boltzmann study, Comput. Fluids 231 (2021) 105063. https://doi.org/10.1016/j.compfluid.2021.105063

    work page doi:10.1016/j.compfluid.2021.105063 2021

  6. [6]

    M. Miwa, A. Nakajima, A. Fujishima, K. Hashimoto, T. Watanabe, Effects of the Surface Roughness on Sliding Angles of Water Droplets on Superhydrophobic Surfaces, Langmuir 16 (2000) 5754–5760. https://doi.org/10.1021/la991660o

    work page doi:10.1021/la991660o 2000

  7. [7]

    Nakajima, Design of hydrophobic surfaces for liquid droplet control, NPG Asia Mater

    A. Nakajima, Design of hydrophobic surfaces for liquid droplet control, NPG Asia Mater. 3 (2011) 49–56. https://doi.org/10.1038/asiamat.2011.55

    work page doi:10.1038/asiamat.2011.55 2011

  8. [8]

    H. Liu, L. Nan, F. Chen, Y. Zhao, Y. Zhao, Functions and applications of artificial intelligence in droplet microfluidics, Lab. Chip 23 (2023) 2497 –2513. https://doi.org/10.1039/D3LC00224A

    work page doi:10.1039/d3lc00224a 2023

  9. [9]

    X. Dong, Z. Li, X. Zhang, Contact -angle implementation in multiphase smoothed particle hydrodynamics simulations, J. Adhes. Sci. Technol. 32 (2018) 2128 –2149. https://doi.org/10.1080/01694243.2018.1464092

    work page doi:10.1080/01694243.2018.1464092 2018

  10. [10]

    Ahangar, M.B

    E.K. Ahangar, M.B. Ayani, J.A. Esfahani, K.C. Kim, Lattice Boltzmann simulation of diluted gas flow inside irregular shape microchannel by two relaxation times on the basis of wall function approach, Va cuum 173 (2020) 109104. https://doi.org/10.1016/j.vacuum.2019.109104

    work page doi:10.1016/j.vacuum.2019.109104 2020

  11. [11]

    Ibrahim, A.S

    M. Ibrahim, A.S. Berrouk, T. Saeed, E.A. Algehyne, V. Ali, Lattice Boltzmann-based numerical analysis of nanofluid natural convection in an inclined cavity subject to multiphysics fields, Sci. Rep. 12 (2022) 5514. https://doi.org/10.1038/s41598-022-09320-8

    work page doi:10.1038/s41598-022-09320-8 2022

  12. [12]

    X. Shan, H. Chen, Lattice Boltzmann model for simulating flows with multiple phases and components, Phys. Rev. E 47 (1993) 1815–1819. https://doi.org/10.1103/PhysRevE.47.1815

    work page doi:10.1103/physreve.47.1815 1993

  13. [13]

    Chai, Baochang Shi, B

    Zhenhua Chai, Z. Chai, Baochang Shi, B. Shi, Multiple -relaxation-time lattice Boltzmann method for the Navier -Stokes and nonlinear convection -diffusion equations: Modeling, analysis, and elements., Phys. Rev. E 102 (2020) 023306. https://doi.org/10.1103/physreve.102.023306

    work page doi:10.1103/physreve.102.023306 2020

  14. [14]

    Fei, Jiapei Yang, J

    Linlin Fei, L. Fei, Jiapei Yang, J. Yang, Yiran Chen, Y. Chen, Yiran Chen, Y. Chen, Huangrui Mo, H. Mo, Huangrui Mo, Huangrui Mo, Kai H. Luo, K.H. Luo, Mesoscopic simulation of three-dimensional pool boiling based on a phase -change cascaded lattice Boltzmann method, Phys. Fluids 32 (2020) 103312. https://doi.org/10.1063/5.0023639

    work page doi:10.1063/5.0023639 2020

  15. [15]

    Wang, Xiaolei Yuan, X

    Huili Wang, H. Wang, Xiaolei Yuan, X. Yuan, Hong Liang, H. Liang, Zhenhua Chai, Zhenhua Chai, Z. Chai, Baochang Shi, B. Shi, A brief review of the phase-field-based lattice Boltzmann method for multiphase flows, Capillarity 2 (2019) 33 –52. https://doi.org/10.26804/capi.2019.03.01

    work page doi:10.26804/capi.2019.03.01 2019

  16. [16]

    Barnes, A.R.(आीशर सोनवने) Sonwane, E.C

    C. Barnes, A.R.(आीशर सोनवने) Sonwane, E.C. Sonnenschein, F. Del Giudice, Machine learning enhanced droplet micro fluidics, Phys. Fluids 35 (2023) 092003. https://doi.org/10.1063/5.0163806

    work page doi:10.1063/5.0163806 2023

  17. [17]

    Deepak, S

    V. Deepak, S. Vengadesan, Droplet Velocity and Film Thickness Studies of an Elongated Taylor Droplet in a Microchannel and Characterization Using Machine Learning, Ind. Eng. Chem. Res. 64 (2025) 8908–8921. https://doi.org/10.1021/acs.iecr.5c00194

    work page doi:10.1021/acs.iecr.5c00194 2025

  18. [18]

    Hasan, Ahmed Ali, Predicting Solar Irradiance at Several Time Horizons Using Machine Learning Algorithms, Sustainability 15 (2023) 8927 –8927

    Chibuzor N Obiora, Ali N. Hasan, Ahmed Ali, Predicting Solar Irradiance at Several Time Horizons Using Machine Learning Algorithms, Sustainability 15 (2023) 8927 –8927. https://doi.org/10.3390/su15118927

    work page doi:10.3390/su15118927 2023

  19. [19]

    Dong, J.-X

    T. Dong, J.-X. Wang, Y. Wang, G. -H. Tang, Y. Cheng, W. -C. Yan, Development of machine learning based droplet diameter prediction model for electrohydrodynamic atomization systems, Chem. Eng. Sci. 268 (2023) 118398. https://doi.org/10.1016/j.ces.2022.118398

    work page doi:10.1016/j.ces.2022.118398 2023

  20. [20]

    Jafari Gukeh, S

    M. Jafari Gukeh, S. Moitra, A.N. Ibrahim, S. Derrible, C.M. Megaridis, Machine Learning Prediction of TiO2 -Coating Wettability Tuned via UV Exposure, ACS Appl. Mater. Interfaces 13 (2021) 46171–46179. https://doi.org/10.1021/acsami.1c13262

    work page doi:10.1021/acsami.1c13262 2021

  21. [21]

    Babu, N.B

    V.K. Babu, N.B. Padhan, R. Pandit, Convolutional neural network based reconstruction of flow- fields from concentration fields for liquid-droplet coalescence, Commun. Phys. 8 (2025) 1–13. https://doi.org/10.1038/s42005-025-02097-y

    work page doi:10.1038/s42005-025-02097-y 2025

  22. [22]

    N. Chen, S. Lucarini, R. Ma, A. Chen, C. Cui, PF-PINNs: Physics-informed neural networks for solving coupled Allen -Cahn and Cahn -Hilliard phase field equations, J. Comput. Phys. 529 (2025) 113843. https://doi.org/10.1016/j.jcp.2025.113843

    work page doi:10.1016/j.jcp.2025.113843 2025

  23. [23]

    Dreisbach, E

    M. Dreisbach, E. Kiyani, J. Kriegseis, G. Karniadakis, A. Stroh, PINNs4Drops: Convolutional feature-enhanced physics-informed neural networks for reconstructing two-phase flows, (2024). https://doi.org/10.48550/arXiv.2411.15949

    work page doi:10.48550/arxiv.2411.15949 2024

  24. [24]

    J. Pu, Y. Chen, Complex dynamics on the one-dimensional quantum droplets via time piecewise PINNs, Phys. Nonlinear Phenom. 454 (2023) 133851. https://doi.org/10.1016/j.physd.2023.133851

    work page doi:10.1016/j.physd.2023.133851 2023

  25. [25]

    X. Chu, W. Guo, T. Wu, Y. Zhou, Y. Zhang, S. Cai, G. Yang, Flow r econstruction over a SUBOFF model based on LBM -generated data and physics -informed neural networks, Ocean Eng. 308 (2024) 118250. https://doi.org/10.1016/j.oceaneng.2024.118250

    work page doi:10.1016/j.oceaneng.2024.118250 2024

  26. [26]

    Karniadakis, I.G

    G.E. Karniadakis, I.G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, L. Yang , Physics-informed machine learning, Nat. Rev. Phys. 3 (2021) 422 –440. https://doi.org/10.1038/s42254 -021- 00314-5

    work page doi:10.1038/s42254 2021

  27. [27]

    A. Roy, A. Mukherjee, B. Prasad, A.K. Nayak, A computational analysis of flow dynamics and heat transfer in a wavy patterned channel using physics-informed neural networks, Phys. Fluids 37 (2025). https://doi.org/10.1063/5.0264160

    work page doi:10.1063/5.0264160 2025

  28. [28]

    R. Sun, H. Jeong, J. Zhao, Y. Gou, E. Sauret, Z. Li, Y. Gu, A physics-informed neural network framework for multi -physics coupling microfluidic problems, C omput. Fluids 284 (2024) 106421. https://doi.org/10.1016/j.compfluid.2024.106421

    work page doi:10.1016/j.compfluid.2024.106421 2024

  29. [29]

    Sharma, W.T

    P. Sharma, W.T. Chung, B. Akoush, M. Ihme, A Review of Physics -Informed Machine Learning in Fluid Mechanics, Energies 16 (2023) 2343. https://doi.org/10.3390/en16052343

    work page doi:10.3390/en16052343 2023

  30. [30]

    Starnoni, Multiphase Flow and Coalescence Filtration in Fibrous Filters: A Review of Numerical and Machine Learning Approaches, Ind

    M. Starnoni, Multiphase Flow and Coalescence Filtration in Fibrous Filters: A Review of Numerical and Machine Learning Approaches, Ind. Eng. Chem. Res. 64 (2025) 22515 –22539. https://doi.org/10.1021/acs.iecr.5c03439

    work page doi:10.1021/acs.iecr.5c03439 2025

  31. [31]

    C. Tang, M. Qin, X. Weng, X. Zhang, P. Zhang, J. Li, Z. Huang, Dynamics of droplet impact on solid surface with different roughness, Int. J. Multiph. Flow 96 (2017) 56 –69. https://doi.org/10.1016/j.ijmultiphaseflow.2017.07.002

    work page doi:10.1016/j.ijmultiphaseflow.2017.07.002 2017

  32. [32]

    Cassie, S

    A.B.D. Cassie, S. Baxte r, Wettability of porous surfaces, Trans. Faraday Soc. 40 (1944) 546 –

    work page 1944

  33. [33]

    https://doi.org/10.1039/TF9444000546

    work page doi:10.1039/tf9444000546

  34. [34]

    Mondal, V

    N. Mondal, V. Arya, P. Sarangi, C. Bakli, Interplay of roughness and wettability in microchannel fluid flows—Elucidating hydrodynamic details assisted by deep learning, Phys. Fluids 36 (2024) 062014. https://doi.org/10.1063/5.0208554

    work page doi:10.1063/5.0208554 2024

  35. [35]

    Upadhaya, Talinungsang, P

    D. Upadhaya, Talinungsang, P. Kumar, D.D. Purkayastha, Tuning the wettability and photocatalytic efficiency of heterostructure ZnO -SnO2 composite films wit h annealing temperature, Mater. Sci. Semicond. Process. 95 (2019) 28 –34. https://doi.org/10.1016/j.mssp.2019.02.009

    work page doi:10.1016/j.mssp.2019.02.009 2019

  36. [36]

    B. Yin, X. Xie, S. Xu, H. Jia, S. Yang, F. Dong, Effect of pillared surfaces with different shape parameters on droplet wettability vi a Lattice Boltzmann method, Colloids Surf. Physicochem. Eng. Asp. 615 (2021) 126259. https://doi.org/10.1016/j.colsurfa.2021.126259

    work page doi:10.1016/j.colsurfa.2021.126259 2021

  37. [37]

    S. Li, J. Yang, A. Ansell, Data-driven reduced-order simulation of dam-break flows in a wetted channel with obstacles, Ocean Eng. 287 (2023) 115826. https://doi.org/10.1016/j.oceaneng.2023.115826

    work page doi:10.1016/j.oceaneng.2023.115826 2023

  38. [38]

    Bhushan, M

    B. Bhushan, M. Nosonovsky, Y.C. Jung, Towards optimization of patterned superhydrophobic surfaces, J. R. Soc. Interface 4 (2007) 643–648. https://doi.org/10.1098/rsif.2006.0211

    work page doi:10.1098/rsif.2006.0211 2007

  39. [39]

    X. Zhu, X. Hu, P. Sun, Physics -Informed Neural Networks for Solving Dynamic Two -Phase Interface Problems, SIAM J. Sci. Comput. 45 (2023) A2912 –A2944. https://doi.org/10.1137/22M1517081

    work page doi:10.1137/22m1517081 2023

  40. [40]

    Siemenn, E

    A.E. Siemenn, E. Shaulsky, M. Beveridge, T. Buonassisi , S.M. Hashmi, I. Drori, A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation, ACS Appl. Mater. Interfaces 14 (2022) 4668–4679. https://doi.org/10.1021/acsami.1c19276

    work page doi:10.1021/acsami.1c19276 2022

  41. [41]

    Zhuang, Q

    Y. Zhuang, Q. Ye, N. Liu, X. Xie, H. Y an, L. Zeng, Hybrid physics-data-driven deep learning for pore -scale transport in microfluidic system, Phys. Fluids 37 (2025) 073363. https://doi.org/10.1063/5.0271043

    work page doi:10.1063/5.0271043 2025

  42. [42]

    Zhang, J

    S. Zhang, J. Tang, H. Wu, Simplified wetting boundary scheme in phase-field lattice Boltzmann model for wetting phenomena on curved boundaries, Phys. Rev. E 108 (2023) 025303. https://doi.org/10.1103/PhysRevE.108.025303

    work page doi:10.1103/physreve.108.025303 2023

  43. [43]

    Huang, C

    J.J. Huang, C. Shu, J.J. Feng, Y.T. Chew, A Phase -Field-Based Hybrid Lattice -Boltzmann Finite-Volume Method and Its Application to Simulate Droplet Motion under Electrowetting Control, J. Adhes. Sci. Technol. 26 (2012) 1825 –1851. https://doi.org/10.1163/156856111X599607

    work page doi:10.1163/156856111x599607 2012

  44. [44]

    Bariki, S

    S.G. Bariki, S. Movahedirad, A flow map for core/shell microdroplet formation in the co -flow Microchannel using ternary phase -field numerical model, Sci. Rep. 12 (2022) 22010. https://doi.org/10.1038/s41598-022-26648-3

    work page doi:10.1038/s41598-022-26648-3 2022

  45. [45]

    Hashemi, M

    Z. Hashemi, M. Gholampour, M.C. Wu, T.Y. Liu, C.Y. Liang, C.-C. Wang, A physics-informed neural networks modeling with cou pled fluid flow and heat transfer – Revisit of natural convection in cavity, Int. Commun. Heat Mass Transf. 157 (2024) 107827. https://doi.org/10.1016/j.icheatmasstransfer.2024.107827

    work page doi:10.1016/j.icheatmasstransfer.2024.107827 2024

  46. [46]

    Zhang, Junfeng Zhang, J

    Junfeng Zhang, J. Zhang, Junfeng Zhang, J. Zhang, Daniel Y. Kwok, Daniel Y. Kwok, D.Y. Kwok, D.Y. Kwok, Lattice Boltzmann Method (LBM), Introd. Lattice Boltzmann Method (2014). https://doi.org/10.1007/springerreference_67034

    work page doi:10.1007/springerreference_67034 2014

  47. [47]

    Zhang, L

    S. Zhang, L. Zhao, L. Han, S. Zhao, L. Song, M. Zhao, Y. Zheng, C. Liu, Neural Network Prediction of Micrometer -Scale Equivalent Contact Angle Mapping: From Microforce Measurements to Local Wettability Characterization, ACS Appl. Mater. Interfaces 17 (2025) 59923–59933. https://doi.org/10.1021/acsami.5c17871

    work page doi:10.1021/acsami.5c17871 2025

  48. [48]

    Suetrong, T

    N. Suetrong, T. Tosuai, H. Vo Thanh, W. Chantapakul, S. Tangparitkul, N. Promsuk, Predicting Dynamic Contact Angle in Immiscible Fluid Displacement: A Machine Learning Approach for Subsurface Flow Applications, Energy Fuels 38 (2024) 3635 –3644. https://doi.org/10.1021/acs.energyfuels.3c04251

    work page doi:10.1021/acs.energyfuels.3c04251 2024

  49. [49]

    J. Chen, F. Yang, K. Luo, Y. Wu, C. Niu, M. Rong, Study on contact spots of fractal rough surfaces based on three -dimensional Weierstrass -Mandelbrot function, in: 2016 IEEE 62nd Holm Conf. Electr. Contacts Holm, 2016: pp. 198 –204. https://doi.org/10.1109/HOLM.2016.7780032

    work page doi:10.1109/holm.2016.7780032 2016

  50. [50]

    Akkaya, O

    I. Akkaya, O. Arslan, J.P. Rolland, Automated and highly precise surface wetting contact angle measurement with optical coherence tomography based on deep learning model, Measurement 253 (2025) 117788. https://doi.org/10.1016/j.measurement.2025.117788

    work page doi:10.1016/j.measurement.2025.117788 2025

  51. [51]

    Chen, Gary D

    Shiyi Chen, S. Chen, Gary D. Doolen, G.D. Doolen, LATTICE BOLTZMANN METHOD FOR FLUID FLOWS, Annu. Rev. Fluid Mech. 30 (1998) 329 –364. https://doi.org/10.1146/annurev.fluid.30.1.329

    work page doi:10.1146/annurev.fluid.30.1.329 1998

  52. [52]

    Peng, L.F

    C. Peng, L.F. Ayala, O.M. Ayala, A thermodynamically consistent pseudo -potential lattice Boltzmann model for multi -component, multiphase, partially miscible mixtures, J. Comput. Phys. 429 (2021) 110018. https://doi.org/10.1016/j.jcp.2020.110018

    work page doi:10.1016/j.jcp.2020.110018 2021

  53. [53]

    Hajisharifi, R

    A. Hajisharifi, R. Halder, M. Girfoglio, A. Beccari, D. Bonanni, G. Rozza, A LSTM -enhanced surrogate model to simulate the dynamics of particle -laden fluid systems, (2024). http://arxiv.org/abs/2403.14283 (accessed August 12, 2024)

    work page arXiv 2024