Desalination and Water Treatment www.deswater.com
63 (2017) 227–240 February
doi: 10.5004/dwt.2017.20195
Phosphate removal and recovery using lime-iron sludge: adsorption, desorption, fractal analysis, modeling and optimization using artificial neural network-genetic algorithm Beverly S. Chittoo, Clint Sutherland* Project Management and Civil Infrastructure Systems, The University of Trinidad and Tobago, San Fernando Campus, Trinidad (WI), Tel. 868 497 5744; email:
[email protected] (C. Sutherland); Tel. 868 491 6879; email:
[email protected] (B.S. Chittoo) Received 7 April 2016; Accepted 23 August 2016
a b s t r ac t An artificial neural network (ANN) was developed to predict the adsorption of phosphate by lime-iron sludge. A fitness function derived from the ANN was incorporated within a genetic algorithm (GA) to elucidate the most optimal combination of operational parameters. The adsorbent characteristics were examined through SEM imagery and analyzed by fractal analysis. Batch experiments were conducted and modeled to expound the mechanisms of adsorption. Kinetic data were best simulated using the diffusion-chemisorption model while equilibrium data followed the Langmuir isotherm. Film and intraparticle diffusion were the dominant transport mechanisms while physisorption was the dominant attachment mechanism. Lime-iron sludge exhibited a maximum adsorption capacity of 15.3 mg/g and compared well with other reported adsorbents. ANN-GA optimization revealed maximum adsorption at an initial phosphate concentration of 59 mg/L, sludge dose of 3 g and temperature of 325 K. The ANN-GA prediction was subsequently verified through laboratory experiments which revealed an excellent prediction. Keywords: Adsorption; Fractal analysis; Lime-iron sludge; Phosphate; Thermodynamics; Artificial neural network; Genetic algorithm
* Corresponding author. 1944-3994/1944-3986 © 2017 Desalination Publications. All rights reserved.
