Statistical analysis of air-gap membrane desalination experimental data: Hypothesis testing

Membrane distillation (MD) is a process in which the driving force for mass transfer is temperature gradient rather than conventional ones based on density, static pressure, chemical nature, affinity, and freezing point gradients. Using a porous hydrophobic membrane, MD comes into four configurations; direct contact, air gap, sweeping gas, and vacuum MD. The current technical literature shows a growing interest in experimental investigation of MD processes. In this work, a complete set of experimental data on air gap membrane distillation is analyzed using statistical methods. The experimental data involves a study of the effects of salt concentration on permeate flux for MgCl2, Na2SO4, and NaCl using three commercial membranes in AGMD unit. Hypothesis testing regarding the mean permeation flux under different salt concentrations is implemented. The objective is to gain an idea about the statistical significance of performance differences among these membranes. Several statistical techniques, i.e., F-test, Fisher's LSD test, Bonferroni and Tukey's test for multiple comparisons are applied. The F-test predicts that all three membranes handle the three salts at their low salt concentration levels in a comparable manner with no significant differences in permeate fluxes but handle the same salts differently at the higher level of salt concentrations. © 2015 Elsevier B.V. All rights reserved.

A complete set of experimental data on AGMD has been analyzed using the methods of statistical experimental design. The experimental data involves a study of the effects of salt concentration (at pre-set conditions of different input parameters such as feed temperature, coolant temperature, feed flow rate) on permeate flux for three different inorganic salts (MgCl2, Na2SO4, and NaCl) using three different commercial membranes (TF-200, TF-450 and TF-1000). Several statistical techniques, i.e., F-test, Fisher's (least significant difference) LSD test, Bonferroni and Tukey's test for multiple comparison were applied. All different tests converge to the same conclusion of the presence of significant differences between all membranes for the different salts. The effect of membrane type on permeate flux is found much more important in the case of low salt concentration of MgCl2, but the opposite result is indicated for NaCl. The TF-200 membrane handles the three different salts at their low and high concentration levels in a comparable manner with no statistical difference. The uncertainty in the measured data proved to have dramatic effects on the statistically-based decision-making process related to the choice of a suitable membrane for handling certain salts.