5.3. Effect of Crossflow Velocity (CFV) and Transmembrane Pressure (TMP)
Figure 7a shows the decline in permeate flux and the resulting permeate oil concentrations for a 40 nm UF membrane during the treatment of an OMS with an initial oil concentration of 30 ppm. We tested the effect of two different CFVs (4.1 and 7.1 m·s
−1) at a constant TMP of 2.0 bar in fed-batch mode and observed different trends in the decline of permeation flux and oil rejection during the first 150 min. This probably reflects the initial formation of a variable fouling layer composed of oil droplets on the membrane surface [
55]. During the first 150 min, the permeate flux declined by 75% (from 878 to 223 L·h
−1·m
−2) at a CFV of 4.1 m·s
−1 and by 65% (from 1085 to 390 L·h
−1·m
−2) at a CFV of 7.1 m·s
−1. Subsequently, the flux declined at a slow and steady rate in both experiments. If the permeation flux trends are considered for the total duration of the filtration experiment rather than a limited running time, the CFV shows an even more significant positive influence on membrane performance. The average membrane permeate flux over the entire duration of the filtration experiment (350 min) was 361 L·h
−1·m
−2 for a CFV of 7.1 m·s
−1, which is ~50% higher than the 244 L·h
−1·m
−2 observed at a CFV of 4.1 m·s
−1 [
60].
Figure 7a also shows that during the treatment of OMS, increasing the CFV reduces the probability of oil particles accumulating on the membrane surface, reducing the degree of fouling and achieving a higher quasi-steady-state permeate flux (165 L·h
−1·m
−2 at a CFV of 4.1 m·s
−1 compared to 240 L·h
−1·m
−2 at a CFV of 7.1 m·s
−1 after a filtration time of 250 min) [
15]. At a constant TMP of 2.0 bar, the higher CFV increases the performance of the SiC UF membrane without limiting its oil retention capacity over the entire duration of the filtration experiment (~6 h). The UF membrane therefore achieves an excellent oil removal efficiency of >98% at both CFVs (
Table 5).
Figure 7b shows the effect of CFV on the permeate flux through a 200 nm SiC MF membrane in total recycle mode, as well as oil adsorption on the membrane surface/pores and total membrane resistance. At a constant TMP of 1.0 bar, increasing the CFV from 0.25 to 1.0 m·s
−1 causes more shear stress at the membrane surface during the first 30 min of the filtration run, reducing the average rate of adsorption/deposition of oil particles from 23% to <7%. This in turn results in a quasi-constant trend of total membrane resistance (1.63·10
9 m
−1) and an average membrane flux of 221 L·h
−1·m
−2 by increasing the CFV to >0.5 m·s
−1.
Next, we investigated the influence of CFV and TMP on the performance of a 40 nm SiC UF membrane while changing the feed composition and characteristics in filtration experiments with both OMS and TDPW.
Figure 8a shows the typical decline in permeate flux through the membrane and the associated oil concentration in the permeate stream over time during the treatment of TDPW using two combinations of TMP and CFV (1.0 bar/4.1 m·s
−1 and 3.0 bar/7.1 m·s
−1). Under the lower TMP/CFV conditions, the initial membrane flux was 270 L·h
−1·m
−2, whereas under the higher TMP/CFV conditions it was ~2.7-fold higher at 750 L·h
−1·m
−2, indicating that a fouling layer accumulates during the first hour of filtration. Accordingly, the decline in membrane flux during the first hour was significantly steeper (~50%) when the higher TMP/CFV conditions were applied, compared to ~12% under the lower CFV/TMP conditions.
The online oil-in-water sensor recorded an average residual permeate oil concentration <0.6 ppm during the filtration of TDPW at a TMP of 1.0 bar and a CFV of 4.1 m·s−1. However, in experiments with the TMP increased to 3.0 bar, the oil retention capacity showed a different trend, and a higher oil content was observed in the permeate samples. The oil content in the permeate stream at a TMP of 3.0 bar increased from an average of 0.6 ppm to 3.8 ppm at times, but stayed constant at 0.6 ppm when the TMP was lower. This indicates that, under the higher TMP/CFV conditions, the critical TMP and CFV values have already been exceeded with regard to the oil retention capacity of the membrane.
The deformation of oil droplets occurs at higher TMPs, increasing the probability that oil droplets will pass through the membrane. Numerical simulations have been carried out to investigate the phenomenon of oil droplet deformation and breakup, and the critical pressure of permeation for oil droplets trapped in membrane pores with a circular cross-section [
60]. During the membrane-based treatment of oily wastewater, oil droplets can deform and pass through membrane pores if the TMP and CFV exceed the critical pressure and critical crossflow shear, respectively [
61]. Accordingly, oil droplets blocking the membrane pores can penetrate the membrane if the TMP is high enough. A similar phenomenon appears to be responsible for the higher oil concentrations found in the permeate samples in our experiments.
Figure 8b shows the transient permeate flux behavior of a 40 nm SiC UF membrane and the resulting oil concentration in permeate samples during the filtration of TDPW and OMS at a TMP of 1.0 bar and a CFV of 4.1 m·s
−1. The initial flux during the treatment of OMS was 1000 L·h
−1·m
−2, which is four times higher than the 270 L·h
−1·m
−2 recorded during the treatment of TDPW. Oil is the only fouling component in the OMS, whereas the complex makeup of TDPW (
Table 3) provides additional contaminants that cause immediate membrane fouling. This immediate fouling layer generates an additional barrier, thus the average oil content in the TDPW samples (initial C
oil = 200 ppm) was 0.5 ppm, compared to 1.0 ppm for the OMS samples (initial C
oil = 30 ppm). However, the total separation efficiency in both cases exceeded 98% (
Table 5).
Figure 9a shows the effect of gradually increasing the TMP (from 0.5 to 2.0 bar) on the permeation flux of two SiC MF membranes with molecular weight cut-off values of 200 and 500 nm. The total recycle mode experiments with an OMS (initial C
oil = 100 ppm) were carried out at a constant CFV of 1.0 m·s
−1 and a process temperature of 40 °C. Initially, the flux through both MF membranes increased with the increase in TMP, consistent with earlier reports concerning the treatment of oily wastewater with ceramic MF membranes [
10,
62]. However, a decrease in permeate flux was observed over the duration of the filtration (1 h) at both TMPs, which is the main challenge limiting the practical applications of MF in oily wastewater treatment [
61]. The decrease in membrane flux may vary depending on the process and feed conditions, membrane material, and pore size.
Figure 9a also indicates that the higher TMP of 0.5 bar caused fouling to occur more rapidly during each filtration run while significantly reducing the permeation rate of the fouled membrane. However, the increase in permeate flux at TMPs >1.0 bar was not linear, in agreement with an earlier study [
62]. Higher TMPs cause oil droplets to pass through the membrane pores, thus more oil droplets accumulate on the membrane surface and in the pores, leading to a higher fouling rate [
63]. Membrane fouling was more severe at the larger pore size of 500 nm (
Figure 9a).
When an OMS with an initial oil concentration of 100–200 ppm was processed using a 200 nm SiC MF membrane, increasing the TMP from 0.5 to 2.0 bar caused a significant increase in the initial permeate flux but also in the amount of oil adsorption on the membrane, which rose from 0 to 35% (
Figure 9b). This also led to a quasi-linear increase in the total membrane resistance from 5.6·10
8 to 8.0·10
8 m
−1 [
63]. At higher TMPs, more severe fouling was observed because oil and particulates that pass through the membrane are adsorbed and accumulate within the pore channels [
25,
61].
5.4. Membrane Chemical-Cleaning Procedure
Membrane fouling is classed as reversible if it can be controlled by physical or chemical cleaning and irreversible if it accumulates over time and cleaning cannot restore the original membrane performance [
64]. The main foulants are organic, colloidal and mineral, which account for 50%, 30% and 20%, of fouling respectively [
58]. The purpose of chemical cleaning is to reverse the loss of membrane permeability and restore flux and rejection by removing the foulants described above [
54,
59].
To specify the flux recovery efficiency
of the SiC membranes, we measured the pure water permeability of virgin (unused) membranes and fouled membranes before and after chemical cleaning. We then used linear regression to plot the pure water flux (
J) against the TMP in the range 0.5–2.0 bar. To determine whether TMP is a relevant parameter affecting membrane cleaning and the resulting membrane regeneration efficiency, we carried out MF/UF experiments with OMS under a range of TMP values between 0.5 and 3.0 bar (
Table 5).
Figure 10a,b compare the clean water fluxes (plotted against TMP) for a virgin 200 nm SiC MF membrane before the filtration run, the same membrane after the treatment of OMS (fouled membrane), and the same membrane after chemical cleaning. The membrane was fouled in total recycling mode under a gradually increasing TMP (from 0.5 to 2.0 bar) while the CFV and temperature were kept constant at 1.0 m·s
−1 and 40 °C, respectively. The initial pure water flux of the virgin SiC membrane was 347 L·h
−1·m
−2 at 0.5 bar and 1329 L·h
−1·m
−2 at 2.0 bar.
Our results showed that the higher the applied TMP (>2.0 bar) during the filtration of oily wastewaters, the greater the degree of irreversible fouling of the membrane and thus the lower the efficiency of chemical cleaning at the end of the filtration process (
Figure 10a). When the TMP was >2.0 bar, irreversible fouling of the membrane could not be eliminated completely using standard cleaning procedures based on NaOH alone, regenerating only 80% of the original membrane flux. An additional citric acid cleaning step was needed to increase the regeneration efficiency to >90%. In contrast, a moderate and constant TMP of ≤1.0 bar resulted in a membrane cleaning efficiency of >98% using standard cleaning procedures based on NaOH alone (
Figure 10b,
Table 5). At TMPs ≤1.0 bar, the limited fouling layer can be removed from the membrane surface and pores by standard chemical cleaning to restore the initial pure water flux.