Browsing by Author "Zhang, Hongwei"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Deep learning-based reconstruction of the structure of heterogeneous composites from their temperature fields
    Wu, Haiyi; Zhang, Hongwei; Hu, Guoqing; Qiao, Rui (2020-04-01)
    Inverse problems involving transport phenomena are ubiquitous in engineering practice, but their solution is often challenging. In this work, we build a data-driven deep learning model to predict the heterogeneous distribution of circle-shaped fillers in two-dimensional thermal composites using the temperature field in the composite as an input. The deep learning model is based on convolutional neural networks with a U-shape architecture and encoding-decoding processes. The temperature field is cast into images of 128 x 128 pixels. When the true temperature at each pixel is given, the trained model can predict the distribution of fillers with an average accuracy of over 0.979. When the true temperature is only available at 0.88% of the pixels inside the composite, the model can predict the distribution of fillers with an average accuracy of 0.94, if the temperature at the unknown pixels is obtained through the Laplace interpolation. Even if the true temperature is only available at pixels on the boundary of the composite, the average prediction accuracy of the deep learning model can still reach 0.80; the prediction accuracy of the model can be improved by incorporating true temperature in regions where the model has low prediction confidence.
  • Thumbnail Image
    A GNN-Based QSPR Model for Surfactant Properties
    Ham, Seokgyun; Wang, Xin; Zhang, Hongwei; Lattimer, Brian; Qiao, Rui (MDPI, 2024-11-19)
    Surfactants are among the most versatile molecules in the chemical industry because they can self-assemble in bulk solutions and at interfaces. Predicting the properties of surfactant solutions, such as their critical micelle concentration (CMC), limiting surface tension (γcmc), and maximal packing density (Γmax) at water–air interfaces, is essential to their rational design. However, the relationship between surfactant structure and these properties is complex and difficult to predict theoretically. Here, we develop a graph neural network (GNN)-based quantitative structure–property relationship (QSPR) model to predict the CMC, γcmc, and Γmax. Ninety-two surfactant data points, encompassing all types of surfactants—anionic, cationic, zwitterionic, and nonionic—are fed into the model, covering a temperature range of [20–30 °C], which contributes to its generalization across all surfactant types. We show that our models have high accuracy (R2 = 0.87 on average in tests) in predicting the three parameters across all types of surfactants. The effectiveness of the QSPR model in capturing the variation of CMC, γcmc, and Γmax with molecular design parameters are carefully assessed. The curated dataset, developed model, and critical assessment of the developed model will contribute to the development of improved surfactants QSPR models and facilitate their rational design for diverse applications.
  • Thumbnail Image
    Nanoscale Transport of Multicomponent Fluids in Shales
    Zhang, Hongwei (Virginia Tech, 2025-01-02)
    CO2 injection has demonstrated significant potential for enhanced oil recovery techniques in unconventional reservoirs, but there exists many challenges in optimizing its operations due to the limited understanding of CO2-oil transport mechanisms in these systems. This dissertation addresses these challenges using molecular dynamics (MD) simulations by investigating the gas and oil transport behaviors and properties within single nanopores under reservoir conditions. The first study examines the exchange dynamics of decane with CO2 and CH4 in a 4 nm-wide calcite nanopore. It is shown that both gases form distinct adsorbed and free molecular populations upon entering the pores, leading to different extraction dynamics. Notably, CO2-decane exchange is initially driven by adsorbed populations, with a transition to free populations later; whereas CH4 -decane exchange follows the opposite pattern. Despite these differences, the transport of both gases apparently follows the same diffusive behavior, with CH4 exhibiting higher effective diffusivities. By calculating self-diffusivities at various relevant compositions, it is found they do not always align well with their effective diffusivities. The second study therefore focuses on Maxwell-Stefan (M-S) diffusivities as a more comprehensive framework to describe the diffusion of CO2-decane mixtures in the first study. It is found that D12 (CO2-decane interactions) remains relatively constant across compositions, unlike bulk mixtures, while D1,s (CO2-wall interactions) increases sharply with CO2 loading. In contrast, D2,s (decane-wall interactions) shows a nonmonotonic trend and, unexpectedly, becomes negative under certain compositions. These phenomena are linked to the strong adsorption of CO2, causing significant density heterogeneity and reduced mobility. Using a multi-task Gaussian process regression model, the M−S diffusivities can be predicted with a relative root mean square error below 10%, significantly reducing computational demand for their practical usage. The third study examines concentration gradient driven diffusio-osmosis of oil-CO2 mixtures within silica and calcite nanopores. Despite higher CO2 enrichment near calcite walls, diffusio-osmotic is only marginally stronger than in silica pores, which is attributed to the variations in interfacial fluid structures and hydrodynamic properties in different pores. Continuum simulations suggest that diffusio-osmosis becomes increasingly significant compared to Poiseuille flow as pore width decreases. The fourth study investigates the oil mixture (C10+C19) recovery from a 4 nm-wide calcite dead-end pore with and without CO2 injection. It was found that CO2 accelerates oil recovery and reduces selectivity for lighter components (e.g., C10) compared to the recovery without CO2. Such improvements are influenced by interfacial and bulk phenomena, including adsorption competition and solubilization effects. Together, these studies provide quantitative insights into CO2-oil transport mechanisms and properties in nanopores. Such insights can help develop better reservoir simulators to guide the optimization of CO2 injection-based enhanced oil recovery in unconventional reservoirs.