Regression analysis for the adsorption of oil using human hair as natural adsorbent


Citation

Kiew Peck Loo, . and Pavithira Sathinathan, . Regression analysis for the adsorption of oil using human hair as natural adsorbent. pp. 16-29. ISSN 2289-1692

Abstract

The abundance of human hair waste (known as keratin biomass) in most parts of the world and its accumulation in waste streams are causing many environmental problems. Human hairs potential as a natural adsorbent to remove oil from wastewater was emphasised in this study. Mathematical and deep learning approaches were adopted to develop the regression models of oil adsorption using gents and ladies hair wastes. The experimental results were obtained from literature review to perform regression analysis using Artificial Neural Network (ANN) in Matlab Microsoft Excel and Design Expert 6.0.6. The efficiency of these tools was compared in predicting the oil removal percentage within a specified range of adsorption parameters. This was done by comparing the R2 value of the established adsorption models. In this study the effect of different adsorption parameters namely pH contact time and adsorbent dosage on the oil removal percentage was included in the regression analysis. Subsequently using the regression model with the highest R2 value for both gents and ladies hair adsorbent a graphical user interface (GUI) was developed for the oil adsorption process to ease users in predicting the oil removal percentage within the specified range of adsorption parameters. The results showed that the adsorption model with the highest R2 value of 0.9570 for gents hair and 0.9650 for ladies hair was developed using the ANN tool implying its superiority in correlating the adsorption parameters to the oil removal percentage. However in creating the GUI layout for the oil adsorption process the regression model generated by the RSM regression tool (the second-highest R2 at 0.8290 for gents hair and 0.6158 for ladies hair) was adopted to the limitations in retrieving the regression model from the ANN tool. The outcome of this study is expected to benefit users without prior knowledge of the oil adsorption process using human hair adsorbent to predict the removal percentage within a specified range of adsorption parameters without the hassle of conducting experimental work.


Download File

Full text available from:

Abstract

The abundance of human hair waste (known as keratin biomass) in most parts of the world and its accumulation in waste streams are causing many environmental problems. Human hairs potential as a natural adsorbent to remove oil from wastewater was emphasised in this study. Mathematical and deep learning approaches were adopted to develop the regression models of oil adsorption using gents and ladies hair wastes. The experimental results were obtained from literature review to perform regression analysis using Artificial Neural Network (ANN) in Matlab Microsoft Excel and Design Expert 6.0.6. The efficiency of these tools was compared in predicting the oil removal percentage within a specified range of adsorption parameters. This was done by comparing the R2 value of the established adsorption models. In this study the effect of different adsorption parameters namely pH contact time and adsorbent dosage on the oil removal percentage was included in the regression analysis. Subsequently using the regression model with the highest R2 value for both gents and ladies hair adsorbent a graphical user interface (GUI) was developed for the oil adsorption process to ease users in predicting the oil removal percentage within the specified range of adsorption parameters. The results showed that the adsorption model with the highest R2 value of 0.9570 for gents hair and 0.9650 for ladies hair was developed using the ANN tool implying its superiority in correlating the adsorption parameters to the oil removal percentage. However in creating the GUI layout for the oil adsorption process the regression model generated by the RSM regression tool (the second-highest R2 at 0.8290 for gents hair and 0.6158 for ladies hair) was adopted to the limitations in retrieving the regression model from the ANN tool. The outcome of this study is expected to benefit users without prior knowledge of the oil adsorption process using human hair adsorbent to predict the removal percentage within a specified range of adsorption parameters without the hassle of conducting experimental work.

Additional Metadata

[error in script]
Item Type: Article
AGROVOC Term: Hair
AGROVOC Term: Adsorbents
AGROVOC Term: Regression analysis
AGROVOC Term: Experimental design
AGROVOC Term: Data collection
AGROVOC Term: Adsorption
Depositing User: Mr. AFANDI ABDUL MALEK
Last Modified: 24 Apr 2025 00:55
URI: http://webagris.upm.edu.my/id/eprint/9814

Actions (login required)

View Item View Item