Process optimisation through Response Surface Methodology for treatment of acid mine drainage using carbon nanotubes-infused thin film nanocomposite membranes

The current work illustrates the use of Central Composite Design (CCD) and Response Surface Methodology (RSM) to gain insight into the effect of five process factors (i.e. pressure, MgSO4 concentration, Fe2(SO4)3 concentration, Al2(SO4)3 concentration and carbon nanotubes loading) on the treatment of synthetic Acid Mine Drainage (AMD) solution with a Thin Film Nanocomposite (TFN) membrane. Performances of polyamide TFN membranes with different multi-walled carbon nanotube (MWCNT) loading were tested to determine optimum loading under different operating conditions. The treatment efficiency was examined in terms of flux and rejection, and based on these results, the optimum region from RSM was selected. The five previously mentioned process variables were investigated, and regression models were built, while data on permeate flux behaviour, MgSO4 rejection, Fe2(SO4)3 rejection, and Al2(SO4)3 rejection as the response values were collected. Predicated models were in good agreement with experimental data and the results showed that pressure and carbon nanotubes (CNT) loading are the main effects that influence the process. The optimal conditions were a pressure of 20 bars and a CNT loading of 0.3%. Under these conditions, a permeate flux of 83%, 94% MgSO4 rejection, 92% Fe2(SO4)3 rejection, and 89% Al2(SO4)3 rejection were observed. RSM models demonstrate the ability to overcome limitations of conventional experimental methods by accounting for interactions between the main factors which affect the process. RSM is applied as a novel method to optimise process conditions of CNT-infused TFN membranes in the removal of heavy metal ions from AMD. •CNT-infused thin film nanocomposite membranes used in the treatment of AMD.•Treatment process affected by pressure, metal concentration, and CNT loading.•CNT loading impacted membrane permeability and heavy metal rejection.•RSM models revealed optimum CNT loading at 0.3%, in agreement with experiments.•RSM models demonstrate ability to overcome limitations of experimental methods.