Performance of the PFB-MF/UF hybrid process compared to that of conventional pre-

3.3.4.1 Characteristics of SFW

In this study, three different SFW characteristics, expected to have significant impact on the fouling of oxidized iron and manganese, were tested: 1) a soft water with negligible calcium concentration (hardness of 2 mg CaCO3/L), 2) a moderately hard water with a calcium concentration of 40 mg/L (hardness of 100 mg CaCO3/L) and 3) a moderately hard water (100 mg CaCO3/L) in the presence of humic acids (≈ 2 mg C/L). In the first step, an appropriate amount of sodium bicarbonate stock solution was added into the DM water to achieve the final alkalinity of 50 mg CaCO3/L. For moderately hard water conditions, the calcium chloride dihydrate stock solution was added into the water. For experiments in the presence of humic acids, the concentration of humic acids was adjusted via addition of humic acids stock solution (Sigma-Aldrich, CAS NO. 68131-04) prepared with Milli-Q^TM water. Finally, the pH was adjusted at 7.0 by bubbling CO2 and/or N2 gas into the water. A fresh Fe(II) and Mn(II) stock solutions, 150 mg Fe²⁺/L and 75 mg Mn²⁺/L, were separately prepared and continuously spiked into the SFW prior entering the columns to achieve a final concentration of 2.6 and 1.3 mg/L in the feed water, respectively. Characteristics of the SFW are summarized in Table 3.7.

Table 3.7. Characteristics of the SFW tested in this study.

Iron

The experiments were conducted in a small pilot plant (Q = 0.5 m³/d) located at Polytechnique Montreal. In the first set of the experiments, the MF and UF membranes were operated to establish the fouling behavior in the conventional peroxidation-MF/UF process. In the second set of the experiments, the MF/UF behavior was examined while using the PFB as a pretreatment process. A schematic of the PFB-MF/UF membrane experimental set-up is shown in Figure 3.4.

SFW

Backwash flow

Sodium hypochlorite injection point Dissolved iron/manganese injection point

Permeate Permeate

Compressed air valve

MF

Pressure probes

UF

Figure 3.4. Schematic of the pilot-scale PFB- MF/UF hybrid process.

A fresh KMnO4 (280 mg KMnO4/L) or sodium hypochlorite (200 mg Cl2/L) stock solutions were prepared for the pre-oxidation or PFB experiments, respectively, and spiked into the SFW immediately prior entering the columns. The oxidation column was a 2-meter clear PVC pipe with an inner diameter of 50.8 mm (2 in.) providing a retention time of 11.6 min for a total flow of 350 mL/min. The PFB column was a 2.5-meter clear PVC pipe with an inner diameter of 25.4 mm (1 in.), which was filled up to a height of 90 cm with pyrolucite media providing an empty bed contact time of 1.7 min for a total flow of 350 mL/min (HLR of 41 m/h (30% bed expansion)). In this process, the free chlorine dosage in the influent was adjusted such that to provide an effluent free chlorine residual concentration of 1.0 mg Cl2/L. The PFB was operated for a period of 10 h for each experiment and the samples were taken at various heights of the bed (0, 9, 25, 50, 75, 100 and 120 cm from the bottom of the column). Total and dissolved (filterable through a 0.45 µm syringe filter (PVDF, CAT. NO. CS-GLPV3045, Chem Science Inc.)) iron and manganese concentrations were measured at each sampling point. Effluent from the oxidation or PFB columns were introduced to the MF/UF membranes after 1 and 4 h of operation, respectively.

Membrane experiments were conducted in dead-end, inside-out and constant flux configuration.

Each membrane unit was equipped with a pressure transducer (Omega, PX409-030GUSB) that recorded the instantaneous pressure every 30 s using a data acquisition software (TRH Central).

Each MF/UF experiment successively consisted of: 1) filtration with DM water at the desired flux (150 or 300 LMH for the peroxidation-MF/UF process and 300 LMH for the PFB-MF/UF hybrid process), 2) filtration with particle-free SFW at the same flux, 3) filtration with the effluent of the columns for 6 h or until the TMP across the UF membrane increased to more than 170 kPa (25 psi), 4) physical cleaning (backwashing with particle-free SFW and air scouring), 5) determining the permeability of the particle-free SFW, 6) chemical cleaning (using oxalic acid and/or sodium hypochlorite solution), 7) rinsing with DM water, and 8) determining the clean water permeability of the membranes a day after the cleaning cycles. The same MF/UF membranes were used for all the experiments. Samples of membrane feed water and permeate were taken to measure the concentrations of iron, manganese, DOC and TOC.

3.3.4.3 Characterization of membrane fouling

Previous studies revealed that in many municipal and industrial solid-liquid separation via membrane filtration, the pressure increase profile in constant flux filtration mode can be described by the power-law compressible cake formation model (Sørensen and Sorensen, 1997; Chellam et al., 1998). Therefore, the constant flux blocking law model for compressible cake filtration (Eq.

(1.21)) (Chellam and Xu, 2006) was applied in this study to describe the pressure increase behavior and the compressibility of the cake layer. The compressibility parameters, β and n’, in this equation were estimated via fitting the experimental pressure data to Eq. (1.21) using a curve-fitting tool in Matlab (R2015a) with the smallest sum of squared residuals.

In addition, the resistance-in-series model (Eq. (2.1)) was also applied to provide further quantitative insight about the reversibility of the fouling and the effectiveness of cleaning methods.

𝐽 = ^∆𝑃

𝜇 𝑅𝑡 = ^∆𝑃

𝜇 (𝑅𝑚+ 𝑅𝑝𝑟+𝑅𝑐𝑟+𝑅𝑖𝑓) (2.1) where Rt is the total resistance (m^-1); Rpr is the physically reversible resistance (m^-1); Rcr is the chemically reversible resistance (m^-1); and Rif is the chemically irreversible fouling (m^-1).

These parameters were determined by:

Rm: measuring the particle-free SFW permeability of the membranes at the beginning of the experiment;

Rpr: the difference between the TMP of particle-free SFW after the physical cleaning step and the TMP at the end of filtration experiment;

Rcr: the difference between the stabilized TMP before and after the chemical cleaning step;

Rif: the difference between Rt and the sum of Rm, Rpr, and Rcr.

CHAPTER 4 ARTICLE 1 : SIZE AND ZETA POTENTIAL OF