Year: 2025
Type: article
Abstract: In this study, a laboratory-scale negative pressure fluidization air gravity separation unit was employed to conduct air velocity tests. Subsequently, computational fluid dynamics (CFD) modelling was utilized to gain insights into the airflow profile and associated properties. The investigation examined three distinct airflow intensities corresponding to low, medium, and high suction fan settings of 10, 30, and 50 Hz. These airflow measurements were conducted across four sections of the suction nozzle: the fan-side (FS) inlet, other-side (OS) inlet, connecting pipe, and fluidizing box. The comparison between the modelled and experimental velocities reveals a significant level of agreement, as evidenced by high coefficients of determination (R2) which provide an indication of the extent to which the experimental variable explains the variation in the response variable. However, some deviations were observed in the predicted variables, as indicated by the root mean square error (RMSE), which indicates the predictive capability of the regression model in estimating the value of the response variable. The RMSE must be observed in the context of the actual airflows, which it represents in that if the RMSE is significantly lower than the observed airflow, the predictive capability of the regression model is considered accurate. Specifically, for the low flow system of 10 Hz, the R2 and RMSE values were determined as 0.95 and 1.13 m/s respectively. Similarly, for the medium flow system of 30 Hz, the R2 and RMSE values were found to be 0.93 and 4.09, respectively. Lastly, for the high-flow system, the R2 and RMSE values were obtained as 0.94 and 6.78, respectively. The comparison of the RMSE to the average observed velocity in the inlets and connecting separating units indicates that the predictive capability of the model is good, with minimal deviation in the predicted and observed variables. The similarity between the RMSE and the velocity in the fluidizing box shows a lower accuracy of the predicted values in this section.
Authors: Nikki Hughes, Marco le Roux, Quentin P. Campbell, Abraham F. van der Merwe
Institution: North-West University
Language: English
Subfield: Computational Mechanics
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7111005555
FWCI:
Cites:
Cited by:
Related to: