Publications

Physics-informed neural network for modeling force and torque fluctuations in a random array of bidisperse spheres

Published in International Journal of Multiphase Flow, 2023

An interpretable shared-block neural network (NN) was established using TensorFlow in Python to predict the force and torque fluctuations in any random arrays of spheres with different sizes. The $R^2$ score of this NN can reach up to 0.91, which is better than any existing NN that can only predict the cases of spheres with the same size. In the meantime, the number of model parameters in the present model is only around 20% in comparison with that of a fully-connected NN introduced by other researchers, which is a result of the delicate design of my NN architecture based on the prior knowledge of physics, symmetries and other constraints.

Recommended citation: Zihao Cheng and Anthony Wachs. Physics-informed neural network for modeling force and torque fluctuations in a random array of bidisperse spheres. International Journal of Multiphase Flow, 169:104603, 2023. https://www.sciencedirect.com/science/article/pii/S0301932223002239?via%3Dihub

Hydrodynamic force and torque fluctuations in a random array of polydisperse stationary spheres

Published in International Journal of Multiphase Flow, 2023

A probabilistic data-driven model was proposed based on a physics-guided Bayesian approach to predict the force and torque distributions in the random arrays of bidisperse spheres, which was the first attempt for modelling the bidisperse suspension hydrodynamics leveraging machine learning techniques. The probability maps of sphere distribution based on different flow conditions were constructed, and the data extracted from the probability maps were properly filtered, standardized and correlated using the SGD regression method with the Scikit-Learn library in Python. The optimal $R^2$ score of this model can reach up to 0.85.

Recommended citation: Zihao Cheng and Anthony Wachs. Hydrodynamic force and torque fluctuations in a random array of polydisperse stationary spheres. International Journal of Multiphase Flow, 167:104524, 2023. https://www.sciencedirect.com/science/article/pii/S0301932223001453?via%3Dihub

An immersed boundary/multi-relaxation time lattice Boltzmann method on adaptive octree grids for the particle-resolved simulation of particle-laden flows

Published in Journal of Computational Physics, 2022

An Immersed Boundary-Lattice Boltzmann (IB-LB) solver was presented to simulate the particle-laden flows, especially the large-scale dense suspension problems thanks to its superior parallel performance that has been demonstrated throughout both the strong and weak scaling tests (efficiency > 90% with 3000+ CPU cores). This solver was the first effort to implement LB method with an adaptive mesh refinement (AMR) technique for modelling the three-dimensional flow configuration. The algorithm was written into an open-source software Basilisk using the C programming language. A series of simulations were conducted, which yielded 160000+ data points for further analysis with machine learning methods.

Recommended citation: Zihao Cheng and Anthony Wachs. An immersed boundary/multi-relaxation time lattice Boltzmann method on adaptive octree grids for the particle-resolved simulation of particle-laden flows. Journal of Computational Physics, 471:111669, 2022. https://www.sciencedirect.com/science/article/pii/S002199912200732X?via%3Dihub

A numerical study of droplet dynamic behaviors on a micro-structured surface using a three dimensional color-gradient lattice Boltzmann model

Published in Soft Matter, 2018

A color-gradient LB method was proposed to solve the two-phase flow problems, with a special research interest in the hydrodynamic behaviors of a liquid droplet placed on the micro-structured surface which usually exhibits super-hydrophobic or super-hydrophilic features. The algorithm was developed in FORTRAN programming language with OpenMP parallel computing.

Recommended citation: Zihao Cheng, Yan Ba, Jinju Sun, Chao Wang, Shengchuan Cai, and Xiaojin Fu. A numerical study of droplet dynamic behaviors on a micro-structured surface using a three dimensional color-gradient lattice Boltzmann model. Soft Matter, 14(5):837–847, 2018. https://pubs.rsc.org/en/content/articlelanding/2018/SM/C7SM02078C

Development of a vortex method with penalization for modeling the complex fluid and moving/deforming solid interaction

Published in International Journal of Numerical Methods for Heat & Fluid Flow, 2018

The penalization technique is incorporated with the semi-Lagrangian VIC method, which makes the complex boundaries of moving/deforming bodies readily treated. An iterative algorithm is further proposed for the penalization and used to solve the Poisson equation, which enhances the vorticity solution accuracy at the body boundary. The complex vortical physics of the moving/deforming body flows are well revealed, and the propulsive mechanism of fish-like swimmer is well illustrated with the present method.

Recommended citation: Chao Wang, Jinju Sun, and Zihao Cheng. Development of a vortex method with penalization for mod- eling the complex fluid and moving/deforming solid interaction. International Journal of Numerical Methods for Heat & Fluid Flow, 28(8):1809–1826, 2018. https://www.emerald.com/insight/content/doi/10.1108/HFF-11-2017-0443/full/html