In-line Coagulation/Ultrafiltration for Silica removal from brackish water as RO membrane pretreatment

Shiao-Shing Chen, Hsu-Hui Cheng and Shu-Ru Yang Institute of Environmental Engineering and Management, National Taipei University of Technology, No. 1, Sec. 3, Chung-Hsiao E. Rd. Taipei 106, Taiwan
*Corresponding author, email f10919@ntut.edu.tw

ABSTRACT In this study, a brackish water containing silica concentration (30 mg/L as SiO2) was studied for silica removal using in-line coagulation/ultrafiltration processes to prevent scaling problem for RO membrane. Other than pH and coagulant dosage, mixing intensity (velocity gradient, G) is also a critical factor on silica removal. Different velocity gradient (G), pH and alum dose were experimented and contour plots were used to determine the optimum silica removal. The optimum silica removal of 55 % was determined with G=2000 S-1, pH=7.1 and alum dosage=30 mg/L as Al2O3, corresponding to a maximum of 87 % recovery for RO membrane plant. Silica was removed by co-precipitation of alumimium hydroxide, supported by the turbidity and silica concentration. Flux loss was observed for the cross-flow UF membrane after filtration. Streaming potential by ElectroKinetic Analyser (EKA) was used to analyze the UF membrane before and after filtration, and slightly reduction of negative charge was observed at medium pH, resulted from the positive charge coagulated particle adsorbed on the negative charge UF membrane surface. Scanning electron microscopy & Energy Dispersive X-ray Spectroscopy (SEM-EDS) also showed slight aluminum and silica signal on the used UF membrane surface. KEYWORDS Silica scaling; in-line coagulation; ultrafiltration; EKA; EDS INTRODUCTION Silica scaling is a common problem encountered in membrane separation processes due to its low solubility of about 120 mg/L in amorphous form. (Sheikholeslami and Tan, 1999; Sheikholeslami and Zhou, 2000; Koo et al., 2001; Semiat et al., 2001; Sheikholeslami et al., 2001; Al-Rehaili, 2003; Chen et al., 2006). Presence of silica in water is due to the dissolution of silica to become silicate based on the following reaction: SiO2 + 2H2O Si(OH)4. The most common method of silica removal is by precipitation with polyvalent metal hydroxide, such as: Fe(OH)3, Al(OH)3, and Mg(OH)2, indicating both softening and coagulation are capable of removing silica (Darton, 1999; Sheikholeslami and Tan, 1999; Bremere et al., 2000; Sheikholeslami and Zhou, 2000; Hafez et al., 2002; Sheikholeslami and Bright, 2002; Taleb-Ahmed et al., 2002; Semiat et al., 2003; Chen et al., 2006). Although softening is capable of removing silica, it is not feasible unless raw water contains enough hardness. Coagulation is one the best options due to its low cost. However, applying coagulation along with sedimentation facilities requires a lot of footprints, which offsets the benefit of fewer footprints required for RO membrane process. Consequently, in-line coagulation/ultrafiltration is a significant methodology for silica pretreatment, where optimum silica removal was achieved in the in-line coagulation process, the remaining particle was removed by ultrafiltration and fewer footprints were required. In the present study, an in-line static mixer containing internal vanes to change the velocity pattern as well as momentum reversals were utilized to achieve the effect of in-line coagulation. Besides, the permeability of ultrafiltration would not be significantly reduced since the coagulated particle is not known as a fouling material. Very few studies have reported silica removal by coagulation and none of them applied in-line coagulation. Al-Rehaili, (2003) reported precipitation aids such as alum and ferric chloride were found necessary for improving performance of the lime-soda ash process to reduce silica to the acceptable level. Ma et al. (2007) used enhanced coagulation and UF process and found out silica was removed from 10 mg/L to less than 2 mg/L. Chuang et al. (2007) reported using coagulation process for industrial wastewater containing silica and silica was reduced from 30 mg/L to 10 mg/L. Chen et al. (2006) demonstrated that coagulation is
1

effective in a jar testing experiment to enhance the silica removal, and coagulation along can remove silica to 50 % at medium pH to significantly reduce the possibility of silica scaling. All the evidence demonstrates that coagulation is a feasible process for silica removal. For in-line coagulation process, one of the most important factors is mixing intensity. However, very few studies have demonstrated this issue. Tchobanoglous et al (2003) report 1500-7500 s-1 and Jones et al (2002) has reported that 3000-5000 S-1. However, these reported values are wide ranged and no specific value can be followed. Since it is an important factor when applying in-line coagulation, the present study also intends to evaluate the optimum mixing intensity for silica removal by in-line coagulation. Besides, the impacts of these coagulated particles on ultrafiltration, including the zeta potential variation on the membrane surface, are also reported. Therefore, three major issues are discussed in this article: (1) Determination of the silica removal requirement for RO membrane by in-line coagulation/ultrafiltration process from a removal- saturation- recovery curve (2) Effects of pH, coagulant dosing and mixing intensity on silica removal for in-line coagulation (3) Impact on ultrafiltration membrane fouling due to preceding in-line coagulation. Methods The raw water is obtained from a brackish groundwater source for a 600 CMD drinking water treatment plant located in Penghu County, Taiwan. This water treatment plant uses RO membrane process, but silica is the limitation for water recovery due to the possibility of scaling, and the raw water qualities were listed in Table 1. Table 1 Raw water qualities for the tested brackish water SiO2 Magnesium Calcium Alkalinity pH mg/L mg/L as CaCO3 mg/L as CaCO3 mg/L as CaCO3 7.8 30 333.3 221 225

chloride mg/L 1550

TDS mg/L 3451

Figure 1 is the schematic diagram for in-line coagulation/ultrafiltration process in this study. A static mixer containing internal vanes to change the velocity pattern as well as momentum reversals was allocated in a 10 mm diameter tube. Flowrate was varied to achieve desire mixing intensity or velocity gradient (G). For static mixer, the degree of mixing is directly related to the headloss (i.e. pressure drop) through the mixer, and the velocity gradient is calculated from Equation (1), where G=average velocity gradient (s-1), P=power input(W), =dynamic viscosity (N.s/m2) and V is volume(m3). Q is the flowrate (m3/s) and P is the pressure drop (bar). The headloss or pressure drop through the mixer was measured by a differential electronic manometer meter (model UEI EM200, manufactured by Universal Enterprise Inc. in Oregon, USA) with measurement range of -150 to 150 mBar
G= P = V Qp V
(1)

PAC tank (coagulation)

UF membrane
P

In-line static mixer Permeate water

pH Settling tank Raw water

Figure 1 Schematic diagram for In-line coagulation/Ultrafiltration process PACl (polyalumnium chloride) was the selected coagulant in this study since it is the most popular used one in Taiwan. All chemicals used are of reagent grade. pH was adjusted by either caustic soda or hydrogen chloride. Silica was measured according to the methods 4500-SiO2 listed in the 20th edition of the Standard Methods (APHA, 1999) using a UV-Vis spectrophotometer (HACH Model DR-4000). Surface charge of the membrane, represented by zeta potential measured by a streaming potential electrokinetic analyzer EKA, was purchased from Brookhaven Instrument Corp, New Jersey. Particle charge, represented by zeta potential, is measured by Zetasizer 3000JS, manufactured by Malvern Instruments Ltd, UK. Scanning electron microscopy & Energy dispersive X-ray Spectroscopy (SEM-EDS) is a S-2400 model, manufactured by Hitachi, Ltd., Japan. RESULTS AND DISCUSSION Determining of Optimum silica removal The optimum operation for RO water recovery and silica removal was determined based on the solubility of silica of 120 mg/L. In other word, silica scaling will occur if the silica concentration is over 120 mg/L. From Equation (2), an optimum RO water recovery and silica removal can be determined accordingly.
Solubility (mg/L) = a[SiO 2 ] a 30 mg / L = = 120 mg / L 1- X 1 X

(2)

In Equation (2), the a value and X represent the silica removal efficiency and water recovery for RO membrane, respectively. Figure 2 is plotted from Equation (2) for better presentation. The result shows, in raw water or 20 % silica removal case, if the water recovery is high, the saturation (%) will exceed 100 % and silica scaling will occur. There are two ways to solve the possible silica scaling problems (1) Decrease the recovery to 73 %, where the saturation (%) will be less than 100 % for raw water (2) Remove the silica concentration at least 40 %, where the concentration will be less than 120 mg/L even the recovery is 85 %. Consequently, in order to achieve the desired recovery 85 %, the targeted SiO2 concentration before entering RO is set up at 18 mg/L, which is equal to 40 % removal of silica for the influent.

Silica cocentration (mg/L)

140 120 100 80 60 40 20 0 0 20 40 60 80 100 Recovery (%)

raw water 20% removal 40% removal 60% removal 80% removal

Figure 2 Silica removal efficiency-concentration-recovery curves Effects of Velocity gradient, pH and alum dosage on silica removal Velocity gradient, pH and alum dosage all affect the mixing intensity and coagulation, and furthermore affect the efficiency of silica reduction. Figure 3 demonstrates the isopleths for silica removal rate vs. pH and velocity gradient for the four different coagulant dosages. First of all, from the four diagrams, the optimum G value is around 2000 s-1 regardless of alum dosages. Tchobanoglous et al (2003) report 1500-7500 s-1 and this study suggests the optimum value for silica removal by in-line coagulation was around 2000 s-1. Secondly, the optimum pH was shifted down as the alum dosage was increased. The optimum pH was 7.1 for alum dosage of 30 mg/L (all alum dosages were as Al2O3 throughout the manuscript) decreased to 6.9 for dosage of 60 mg/L and to 6.4 for dosage of 90 mg/L. The reduction of optimum pH for the in-line coagulation for silica removal can be attributed to the generated proton when alum was added. Since the highest silica removal was observed at 30 mg/L of alum dosage, the optimum silica removal of 55 % was determined with G=2000 S-1, pH=7.1 and alum dosage=30 mg/L, corresponding to a maximum of 87 % recovery for RO membrane plant. Besides, silica removal was not increased as dosage increase, due to the destabilization of the coagulated particle to decrease the silica reduction.

8.5

(a)
8

(c)

7.5

pH

62

6.5

47

5.5 5.5 1 8.5

1.5

2.5 G value (*1000 1/s)

3.5

(b)

1.5

2.5

3.5

(d)

45
7.5

pH

6.5

41
6

5.5 1 1.5 2 2.5 3 3.5 4 1 1

1.5

2.5

3.5

G value (*1000 1/s)

G value (*1000 1/s)

Figure 3 Silica removal for different pH and velocity gradient for different coagulant dosages (a) 30 mg/L (b) 60mg/L (c) 90mg/L (d) 120mg/L The silica was removed by co-precipitation aluminum hydroxide, supported by the results of turbidity and silica removal in Figure 4. In Figure 4, using an example of alum dosage of 30 mg/L at G= 2000s-1, higher turbidity due to the formation of aluminum hydroxide resulted in higher silica removal, indicating silica was removed during the process of co-precipitation of aluminum hydroxide. From solubility diagram of freshly precipitated aluminum hydroxide, the lowest solubility was located at pH 7(Snoeyink and Jenkins, 1980); therefore, the turbidity was gradually increased as pH increased until around 7. But after pH 7, the turbidity was decreased since solubility was increased for pH >7 from solubility diagram.
80% 70% 60% 50% 40% 30% 20% 10% 0% 5.5 6 6.5 7 7.5 pH 4 3.5 3 2.5 2 1.5 1 0.5 0 9 9.5

SiO2 removal turbidity

8.5

Figure 4 Silica removal and remaining turbidity for different pHs at 30 mg/L and G=2000 s-1
5

turbidity (NTU)

SiO2 removal

Ultrafiltration fouling potential by coagulation Flux loss was observed for the cross-flow UF membrane after filtration, as shown in Figure 5. The gel layer resistance was evaluated using resistance-in-series model. Other than lower mixing intensity at G=1000 s-1, the resistance was increased significantly for the rest three mixing intensity. This phenomenon was evaluated by measurement of membrane and particle surface charge, and both were represented by their zeta potentials.

9.E+06 8.E+06 7.E+06 6.E+06 5.E+06 4.E+06 3.E+06 2.E+06 1.E+06 0.E+00 0

G=1000 s-1

G=2000 s-1

G=3000 s-1

G=4000 s-1

Resistance (1/m)

500

1000

1500

2000

Filtration time (min)

Figure 5 Membrane resistances versus filtration time throughout the experiment Figure 6 presents the membrane surface charge (zeta potential) measured by EKA for fresh and used membrane. In Figure 6, for the fresh membrane, the pH of zero charge (pHzpc) was 4.4 with higher negative charge at medium pH. After the membrane was used, the negative charge was less since the particle of silica -aluminum hydroxide was attached on the membrane surface. The indication of less negative charge on the membrane surface was supported by the particle zeta potential data, as reported in Figure 7. Figure 7 shows zeta potential for different dosages at G of 2000 s-1 before and after membrane filtration. For the water before membrane filtration, the particle zeta potential increases as the dosage increases, resulted from the positive charge coagulated particle. Therefore, the reduction of negative charge in Figure 6 was attributed to the positive charge coagulated particle attaching onto the membrane surface. After the membrane filtration, the zeta potential stayed negative, due to the removal of those positive particles and small amount of negative charge of silica particle remaining. Another evidence of the coagulated particle attachment is shown in Figure 8 by EDS for used membrane, where a significant peak was observed for aluminum and silica, indicating the attachment of the coagulated particle onto the membrane.

10
fresh membrane

5 ZP (mV) 0 -5 -10 -15 -20 pH 2 3 4 5 6 7 8

uesd membrane

10

11

12

Figure 6 Alternation of membrane surface charge for fresh and used membrane
15 10 ZP(mV) 5 0 -5 -10 -15 Coagulant dosage (mg/L) 30 60 90 120 before filtration after filtration

Figure 7 Zeta potential for different dosages before and after membrane filtration (G=2000 s-1)

Si Al S

Figure 8 EDS for used membrane in the in-line coagulation/ultrafiltration processes Conclusions Silica was successfully removed in these in-line coagulation/ultrafiltration processes. Other than pH and coagulant dosage, mixing intensity (velocity gradient, G) is also a critical factor on silica removal. The
7

optimum silica removal of 55 % was determined with G=2000 s-1, pH=7.1 and alum dosage=30 mg/L as Al2O3, corresponding to a maximum of 87 % recovery for RO membrane plant. Flux loss was observed for the cross-flow UF membrane after filtration. Streaming potential by EKA was used to analyze the UF membrane before and after filtration and slightly reduction of negative charge was observed, resulted from the positive charge coagulated particle adsorbed on the negative charge UF membrane surface. EDS images also showed also showed slight aluminum and silica signal on the UF membrane surface. References Al-Rehaili A. M. (2003) Comparative chemical clarification for silica removal from RO groundwater feed. Desalination 159, 21-31. Bremere I., Kennedy M., Mhyio S., Jaljuli A., Witkamp G.-J., Schippers J. (2000) Prevention of silica scale in membrane systems: removal of monomer and polymer silica. Desalination 132, 89-100. Chen S., Chang T., Lin C. (2006) Silica pretreatment for a RO brackish water source with high magnesium. Water Science and Technology: Water Supply 6, 179-187. Chuang S. H., Chang T. C., Ouyang C. F., Leu J. M. (2007) Colloidal silica removal in coagulation processes for wastewater reuse in a high-tech industrial park. In: Water Science and Technology, Volume 55. Edited by Kim I. S., Kim S. and Cho J., pp. 187-195. Darton E. G. (1999) RO plant experiences with high silica waters in the Canary Islands. Desalination 124, 33-41. Hafez A. I., El-Manharawy M. S., Khedr M. A. (2002) RO membrane removal of unreacted chromium from spent tanning effluent. A pilot-scale study, Part 2. Desalination 144, 237-242. Jones S. C., Sotiropoulos F., Amirtharajah A. (2002) Numerical modeling of helical static mixers for water treatment. Journal of Environmental Engineering 128, 431-440. Koo T., Lee Y. J., Sheikholeslami R. (2001) Silica fouling and cleaning of reverse osmosis membranes. Desalination 139, 43-56. Ma W., Zhao Y., Wang L. (2007) The pretreatment with enhanced coagulation and a UF membrane for seawater desalination with reverse osmosis. Desalination 203, 256-259. Semiat R., Sutzkover I., Hasson D. (2001) Technique for evaluating silica scaling and its inhibition in RO desalting. Desalination 140, 181-193. Semiat R., Sutzkover I., Hasson D. (2003) Scaling of RO membranes from silica supersaturated solutions. Desalination 157, 169-191. Sheikholeslami R., Al-Mutaz I. S., Koo T., Young A. (2001) Pretreatment and the effect of cations and anions on prevention of silica fouling. Desalination 139, 83-95. Sheikholeslami R., Bright J. (2002) Silica and metals removal by pretreatment to prevent fouling of reverse osmosis membranes. Desalination 143, 255-267. Sheikholeslami R., Tan S. (1999) Effects of water quality on silica fouling of desalination plants. Desalination 126, 267-280. Sheikholeslami R., Zhou S. (2000) Performance of RO membranes in silica bearing waters. Desalination 132, 337-344. Snoeyink V. L., Jenkins D. (1980) Water Chemistry. John Wiley & Sons, New York. Taleb-Ahmed M., Taha S., Maachi R., Dorange G. (2002) The influence of physico-chemistry on the retention of chromium ions during nanofiltration. Desalination 145, 103-108. Tchobanoglous G., Burton F. L., Stensel H. D. (2003) Wastewater engineering: treatment and reuse. 4th ed. McGraw-Hill, . New York, USA.