A Machine Learning Study of Predicting Mixing and Segregation Behaviors in a Bidisperse Solid-Liquid Fluidized Bed

Article Engineering, Chemical

Machine learning accelerated discrete element modeling of granular flows

Liqiang Lu et al.

Summary: The research accelerated the computation speed of DEM using a convolutional neural network, demonstrating its accuracy and efficiency in the simulation of granular flows through experiments.

CHEMICAL ENGINEERING SCIENCE (2021)

Add to Collection

Article Engineering, Chemical

CFD-DEM modelling of the migration of fines in suspension flow through a solid packed bed

Zhouzun Xie et al.

Summary: In this work, a smoothed volume distribution model (SVDM) was developed to simulate the migration of fines in suspension flow through a solid packed bed based on the unresolved CFD-DEM framework. The SVDM method allows a high-resolution mesh and can achieve a high efficiency and applicability compared to the DPVM method. The effects of inlet fine concentration and solid/fine size ratio on fines migration behavior were also studied, revealing insights into fine powder holdup and migration mechanisms.

CHEMICAL ENGINEERING SCIENCE (2021)

Add to Collection

Article Green & Sustainable Science & Technology

Short-term self consumption PV plant power production forecasts based on hybrid CNN-LSTM, ConvLSTM models

Ali Agga et al.

Summary: Global electricity consumption has increased in the last century, leading to a growing importance of green energy sources like PV technology. This study introduces two hybrid models to predict power production for PV plants, showing higher accuracy compared to a standard LSTM model.

RENEWABLE ENERGY (2021)

Add to Collection

Article Energy & Fuels

A multi-dimensional parametric study of variability in multi-phase flow dynamics during geologic CO2 sequestration accelerated with machine learning

Hao Wu et al.

APPLIED ENERGY (2021)

Add to Collection

Article Mechanics

A priori assessment of convolutional neural network and algebraic models for flame surface density of high Karlovitz premixed flames

Jiahao Ren et al.

Summary: Accurate modeling of unresolved flame surface area is essential for flame surface density method. This study evaluates algebraic and convolutional neural network models performance in high Ka premixed flames using direct numerical simulation database. Results suggest that algebraic models perform well in low Ka cases, while CNN model outperforms in high Ka flames with correlation coefficient over 0.91.

PHYSICS OF FLUIDS (2021)

Add to Collection

Article Multidisciplinary Sciences

Machine learning-accelerated computational fluid dynamics

Dmitrii Kochkov et al.

Summary: Numerical simulation of fluids is crucial in modeling physical phenomena like weather and aerodynamics, but solving the Navier-Stokes equations at scale remains challenging due to computational costs. This study uses end-to-end deep learning to improve approximations in computational fluid dynamics for modeling two-dimensional turbulent flows, achieving computational speedups and stability while generalizing well to different scenarios. Leveraging machine learning and hardware accelerators, this approach showcases the potential to enhance simulations without sacrificing accuracy or generalization.

PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA (2021)

Add to Collection

Article Engineering, Chemical

CFD-DEM study of segregation and mixing characteristics under a bi-disperse solid-liquid fluidised bed

Zhouzun Xie et al.

Summary: In this study, a recent smoothed CFD-DEM model is used to investigate segregation and mixing behaviors in a bi-disperse solid-liquid fluidized bed with different superficial liquid velocities and wide size ratios. Results show that the mixing index effectively quantifies segregation degree, with circulation motion of particles and liquid confirmed in the transition zone, and hydrodynamics and collision being the main underlying mechanism of segregation behavior in the bi-disperse SLFB.

ADVANCED POWDER TECHNOLOGY (2021)

Add to Collection

Article Energy & Fuels

The Sliding Window and SHAP Theory-An Improved System with a Long Short-Term Memory Network Model for State of Charge Prediction in Electric Vehicle Application

Xinyu Gu et al.

Summary: The prediction of SOC of an electric vehicle battery pack is crucial for its reliability, efficiency, and lifespan. Various techniques and statistical systems have been proposed in the past for improving prediction accuracy. The introduction of machine learning techniques in recent years has further enhanced SOC prediction accuracy and adaptability.

ENERGIES (2021)

Add to Collection

Article Energy & Fuels

A Priori Modeling of NO Formation with Principal Component Analysis and the Convolutional Neural Network in the Context of Large Eddy Simulation

Jiahao Ren et al.

Summary: A machine learning-based model utilizing PCA and CNN was proposed to predict NO reaction rate in turbulent flames within the framework of LES. Evaluation using DNS showed good performance in predicting thermal, prompt, and N2O pathways, but larger errors were observed for the NO2 pathway.

ENERGY & FUELS (2021)

Add to Collection

Article Engineering, Chemical