A decision support system for predicting settling velocity of spherical and non-spherical particles in Newtonian fluids

Abstract

An artificial intelligence-based system was developed to efficiently predict settling velocity (SV) using a large dataset comprised of 2726 samples. The ranges of particle size and fluid viscosity were 0.212 − 98.59 mm and 0.02 − 92800 mPa.s, respectively. Properties of particle and fluid were fed to a model as the inputs to obtain SV as the output. Six machine learning algorithms were tested for the prediction. The random forest (RF) performed better than other algorithms with a coefficient of determination of 0.98 and a mean square error of 0.0027. A simple decision support system was developed using the RF model. The current study demonstrates the complete methodology of modeling SV with ML.

Keywords:

• Artificial intelligence
• model
• machine learning
• random forest
• particle shape

Acknowledgment

We acknowledge the kind supports we received from King Faisal University (Saudi Arabia), The People’s University of Bangladesh (Bangladesh), and Islamic University (Bangladesh). We would like to acknowledge the assistance we received from Dr. Sanders and his team. A large part of the data used for the current study was collected from Breakey et al. (2018), which was produced from his lab. In addition, Mr. Majdi Al-Faiad (Lecturer, Department of Chemical Engineering, King Faisal University) deserves our gratitude for his kind assistance in collecting the data from the literature.