An Integrated Experimental‐CFD‐RSM‐ANN Investigation of Arsenic Removal From Groundwater by Porous Geopolymer

ABSTRACT

A novel, low‐cost, porous geopolymer adsorbent incorporating silicon metal powder as a pore‐forming agent and graphene oxide for arsenic (As) removal from aqueous systems was synthesized, characterized, and evaluated. Extensive characterizations were conducted by scanning electronic microscopy (SEM), X‐ray diffraction (XRD), energy‐dispersive X‐ray spectroscopy (EDX), and Fourier transform infrared spectroscopy (FTIR), revealing its performance in a fixed‐bed, adsorption column. XRD and SEM results indicated the formation of geopolymer at 27–29° of 2θ, achieved at a 50 ⁰ C curing temperature with 1wt% silicon metal powder and 1 wt% graphene oxide flakes, forming regular, uniform, interconnected pores. After the preliminary experimental investigations, COMSOL Multiphysics was used to simulate the adsorption process. The model was satisfactorily validated against experimental data, demonstrating its reliability under different operating conditions. The simulations were extended over an inlet concentration (C _i ) range of 50–150 ppb, bed height (Z _b ) of 2–6 cm, and feed flow rate (Q _f ) of 0.005–0.015 L/min. Results showed that increasing Z _b and C _i , enhanced arsenic removal efficiency (As%) by more than 99% and bed saturation time (t _s ) from 4000 to 1200 min, while at higher Q _f , both were reduced. Subsequently, response surface methodology (RSM) was employed to optimize operating parameters and understand the main and interaction effects of design factors on As% and t _s . ANN‐based predictive modeling was also conducted. Findings of RSM and artificial neural network (ANN) were closely aligned with experimental and computational fluid dynamics (CFD) results, indicating their reliability. Overall, Z _b exhibited a profound effect as compared to other factors. This study provides valuable insights for enhancing adsorption efficiency and system scalability.