Mathematical Modelling for Predicting Methylene Blue Removal Efficiency Through Adsorption Using Activated Carbon of Parthenium hysterophorus

Excessive release of methylene blue (MB) in the environment and its potential infusion to human body may cause several health issues such as blurred vision, confusion, hallucinations and seizures. Based on experimental results on MB removal efficiency through adsorption using activated carbon of Parthenium hysterophorus, a simple mathematical model is developed for the prediction of MB removal efficiency for any combination of input conditions. Based on earlier experimental results, four independent variables such as pH, contact time, initial MB concentration and dose concentration were considered for the model. Based on individual experimental results, a factor is proposed for each independent variable. Eventually, all the factors are multiplied with maximum achievable removal efficiency to predict a removal efficiency under specific conditions. It is found that the developed model can accurately predict the higher range of efficiency while underestimates during lower range of efficiency. To overcome this drawback, the developed model was further fine-tuned with an adjustment factor, through which model predictions were turned out to be much better. Standard errors of the original model's estimations are RMSE = 10.09, MAE = 7.08 and RAE = 0.12. After the mentioned adjustment, the same errors were lowered by 60%, 57% and 83%, respectively. Such modelling approach is very useful for the decision-makers on deciding input variables for industry-scale implementations. Similar mathematical modelling can be adopted for other pollutants and for other adsorbents.

Automated summary

The study presents a mathematical model for predicting the removal efficiency of methylene blue on activated carbon derived from Parthenium hysterophorus. By considering four independent variables and fine-tuning the model with adjustment factors based on experimental results, the accuracy of predictions was significantly improved. Standard errors of the original model were reduced by 60%, 57% and 83% after adjustments, showing the effectiveness of the developed modelling approach.