Continuous blowdown water (CBW) from industrial boilers is of great quantity and energy, and therefore worth recycling. According to the data from a long period of monitoring, carbonate and sulfate were documented as the main contaminants in CBW. Herein, an attapulgite (ATP) membrane was prepared on a macroporous Al2O3 support through solid state sintering. The prepared membrane (attapulgite membrane (ATM)) was characterized by mercury porosimetry, scanning electron microscope, X-ray diffraction, and permselectivity analysis in dead-end mode. The optimal sintering temperature was 800 °C based on the morphology of the sintered active layer and its adhesive strength with support. The active layer exhibited a pore size distribution concentrated on 12.7nm with a thickness of about 80μm. In addition, crystal structures of ATP were retained in powder form after sintering. ATM obtained a pure water permeability of 1,411.87Lm–2h–1MPa–1 with the effective retention of carbonate and sulfate through electrical-related interactions. Several ATM filters were also developed and mounted on the boiler blowdown pipe, which had a high rejection rate of alkalinity and dissolved solids in CBW. The volume of wastewater emitted from the low-pressure boiler was reduced, since the filtrate could be recycled as part of the make-up water.

NOMENCLATURE

  • A

    effective filtering area, m2

  • J

    pure water permeability, L m–2 h–1

  • m1

    mass of the dried membrane, kg

  • m2

    mass of the solid-saturated membrane, kg

  • m3

    mass of the water-saturated membrane, kg

  • p

    imposed pressure, MPa

  • QCBW

    volume of CBW, m3 h–1

  • QS

    volume of steam product, m3 h–1

  • q

    porosity of the membrane, %

  • R

    ion rejection rate, %

  • r

    pore size of the membrane, nm

  • t

    filtering time, h

  • V

    volume of the filtrate, L

Greek letters

  • γ

    surface tension of mercury, 0.48 N/m

  • η

    blowdown rate of the industrial boiler, %

  • θ

    contact angle between the membrane and solution, degree

  • ρ0

    ion concentrations in the feed solution, mg/L

  • ρf

    ion concentrations in the filtrate, mg/L

Subscripts

  • CBW

    continuous blowdown water

  • f

    feed

  • S

    steam

INTRODUCTION

Boilers are used as the power source in most process industries (Luo et al. 2012). Thus, they are usually regarded as the ‘heart’ of industries. Based on the working principle of the steam engine, water can be boiled and transferred to pressurized steam by burning fossil fuels. The pressurized steam drives the related mechanical parts to achieve energy conversion. Boilers are the device for water reserve and steam production, which are usually treated as the power source in most process industries. Thereby, the intrinsic energy contained in fossil fuels can be converted to other forms of counterparts (electrical energy, heat, kinetic energy, etc.) for human utilization through the assistance of boilers. With accumulated operating experience, periodical and continuous blowdown have become the necessary procedures to prevent corrosion and deposition, and for secure operations. Insufficient blowdown will not only make the boiler water unqualified, but also threatens the safety of operation. Periodical and continuous blowdown are necessary procedures during boiler operation. Large quantities of water have to be emitted from the boiler body, resulting in high waste production. In order to maintain the water/vapor balance, equal amounts of fresh water to wastewater should be introduced into the boiler system. Another aspect deserving attention is that the wastewater emitted from the boiler contains lots of energy, which will be lost during the blowdown procedure.

In summary, we make the conclusion that industrial boilers are units of energy intensive, water consuming, and high waste production.

In general, boilers with lower pressure usually have a higher blowdown rate. According to the relevant reports in China (Li et al. 2009; Wei et al. 2013), the blowdown rate of low-pressure boilers (≤2.45 MPa) could reach 10% or even higher. In addition, boilers with different rated pressures have different types of working processes and components of contaminants in continuous blowdown water (CBW). (Li et al. 2009). For example, phosphate is the dominant pollutant in the CBW of medium-pressure boilers (∼3.82MPa); meanwhile, sulfate (SO42–), alkalinity (CO32–, HCO3 and OH) and metal cations (Na+) make up the main pollutants in the CBW of low-pressure boilers (Saidur 2011; Wei et al. 2013). Discharging CBW without proper treatment will undoubtedly pose an eco-toxicological threat to the organisms in receiving water bodies and soil. Besides, blowdown volume, especially from low-pressure boilers, is an obvious target for saving fresh or treated water, and chemical treatment.

One of the most efficient ways to treat wastewater is membrane technology, which is an effective and reliable method for the combined removal of a broad range of pollutants (Rana & Matsuura 2010). Taking into account the different separating mechanisms, rejection processes and pore size distributions, membranes can be classified as different types: microfiltration, ultrafiltration, nanofiltration and so forth (Shirazi et al. 2013). Most nano-scale membranes are usually made from polymers for high separation efficiency (Agana et al. 2013). Polymeric membranes can separate aqueous mixtures containing strong electrolytes and/or uncharged solutes through the effects of electrostatic interactions (Luo & Wan 2013). Recently, ceramic membranes have gained increasing popularity for water treatment due to many inherent advantages over their polymeric counterparts such as high mechanical and chemical stability, and enhanced hydrophilicity (Chang et al. 2014). However, ceramic membranes, which could interact with charged species through electrical-related effects, are usually made from amphoteric material (Al2O3, TiO2 or ZrO2) by a sol-gel method (Chang et al. 2014). Moreover, organic solutions and nucleating agents are necessities during preparation for the integrity of membranes. Therefore, it is desired to develop a ceramic membrane with electrokinetic properties based on clay minerals via simple sintering.

Attapulgite (ATP) is one of the cost-effective natural minerals, which can be surface mined in eastern and southern parts of China (Deng et al. 2013). Because of its permanent negative surface charge and non-swelling properties, ATP powder has become an attractive material in water treatment with different degrees of success (Shi et al. 2013; Yin & Kong 2014; Liu et al. 2015). Adsorption is a widely accepted way for ATP to purify the discharged effluent. Unfortunately, alkali metal ions, which are inevitable in industrial wastewater, can adversely affect the adsorption capacity of ATP (Shi et al. 2013). Thus, an increasing number of studies have concentrated on the modification of ATP powder for enhanced contaminant retention and retarded contaminant migration. For instance, natural ATP was activated by dilute hydrochloric acid and further functionalized by ethylenediamine for lead removal in water (Deng et al. 2013); a porous material based on polyurethane and ATP could be considered as an efficient adsorbent for dyeing wastewater treatment (Dong et al. 2013). A survey of the literature reveals that little attention has been paid to the preparation of an active membrane layer made of ATP for purifying saline wastewater.

The present study will of necessity encapsulate the preparation of an attapulgite membrane (ATM) on macroporous Al2O3 support through solid state sintering. The performance of the ATM is evaluated by laboratory-scale experiments in dead-end mode. Several membrane filters were developed based on ATM, and these were applied to treat the CBW from an industrial boiler.

MATERIALS AND METHODS

Chemicals and reagents

ATP was purchased from the Chunqiu Ceramic Workshop (Zibo, China). X-ray fluorescence (Shimadzu, Japan) was used to determine the chemical composition of the ATP, which is summarized in Table 1. Commercially available Al2O3 supports were from Fengxi Porcelain Factory (Chaozhou, China). The specific parameters of the supports are listed in Table 2. Analytical grade hydroxyethyl cellulose, Na2CO3, Na2SO4, NaOH, NaCl and hydrochloric acid were obtained from Beijing Chemical Works (Beijing, China) and used as received. Except where otherwise stated, all the solutions mentioned in this study were prepared using deionized (DI) water with a conductivity of 7.6μS/cm, which was prepared through an HA-10LB style ultra-pure water system of Berclean Company (Chongqing, China) and used at its native pH (6.0 ± 0.2).

Table 1

Compositions of natural clay. LOI is loss on ignition

Material Chemical composition (wt.%) 
ATP SiO2 MgO Al2O3 Fe2O3 TiO2 Na2SO3 P2O5 MnO LOI 
55.13 21.57 17.41 3.75 1.43 0.31 0.10 0.05 0.02 0.23 
Material Chemical composition (wt.%) 
ATP SiO2 MgO Al2O3 Fe2O3 TiO2 Na2SO3 P2O5 MnO LOI 
55.13 21.57 17.41 3.75 1.43 0.31 0.10 0.05 0.02 0.23 
Table 2

Main parameters and hydraulic performance of purchased Al2O3

Item Unit Values 
Diameter mm 90 
Thickness mm 
Porosity 37.32 
Pure water flux L m–2 h–1 8,341.67 under 0.1 MPa 
Item Unit Values 
Diameter mm 90 
Thickness mm 
Porosity 37.32 
Pure water flux L m–2 h–1 8,341.67 under 0.1 MPa 

Sampling of CBW

Online monitoring of practical CBW discharged from a local WNS20–1.57YQ style boiler (Xinzheng Cigarette Factory, China) was carried out for 8 months. The samples were stored through the methods described in GB/T 6907-2005 for experiments in the laboratory (Standard GB/T 6907-2005). The characteristics of CBW are recorded in Table 3, which were determined according to the methods described by the American Public Health Association (APHA/AWWA/WEF 1998).

Table 3

Water quality of CBW from the WNS20-1.57YQ style boiler (August 2013–April 2014)

Parameters Unit Mean values 
pH (25 °C)  11.51 
Calcium (Ca2+mg L–1 Not detected 
Magnesium (Mg2+mg L–1 Not detected 
Sodium (Na+mg L–1 1,283.18 
Chloride (Clmg L–1 6.07 
Sulfate (SO42–mg L–1 405.63 
Carbonate (CO32–mg L–1 1,415.08 
Suspended solids mg L–1 Not detected 
Temperature °C 109 
Pressure MPa 0.14 
Parameters Unit Mean values 
pH (25 °C)  11.51 
Calcium (Ca2+mg L–1 Not detected 
Magnesium (Mg2+mg L–1 Not detected 
Sodium (Na+mg L–1 1,283.18 
Chloride (Clmg L–1 6.07 
Sulfate (SO42–mg L–1 405.63 
Carbonate (CO32–mg L–1 1,415.08 
Suspended solids mg L–1 Not detected 
Temperature °C 109 
Pressure MPa 0.14 

Preparation of ATP membrane

Before the coating procedure, pre-treatment of Al2O3 supports was carried out by grinding with emery paper of grade number 800, 1,000, 1,200, 1,600 and 2,000, rinsing, ultrasonic cleaning in DI water under 40 kHz for 15 min, and oven-drying at 80 °C.

The ATP powder was mixed with hydroxyethyl cellulose and water under vigorous stirring until a stable turbid solution (ATP, 8 wt.%) was harvested. A Petri dish was utilized to hold the ATP suspension. One side of the Al2O3 support was then dipped vertically into the suspension for 10 s and a wet ATP coating layer formed because of the capillary effect. All the coated substrates were transferred into a desiccator over silica gel at room temperature (27 °C, overnight) for natural drying. Afterwards, they were oven-dried at 90 °C to remove the residual moisture, and then sintered at the desired temperatures for 3 h with a heating rate of 1 °C min–1 in a SX2-2.5-12 style muffle furnace (Beijing Furnace Factory, China). After natural cooling, each as-prepared membrane was carefully collected for subsequent tests.

Characterization

According to the method described in GB/T 9286-1998 (Standard GB/T 9286-1998), the adhesive strength level between the ATP active layer and Al2O3 support was determined. The strength contains six levels: the lower the level, the higher the strength. The porosity (q, %) of the ATM was determined based on the Archimedes drainage method (Jakobs & Koros 1997): 
formula
1
where m1, m2 and m3 are the mass of the dried, solid-saturated and water-saturated membranes, respectively, in kg. It is noteworthy that the active layer cannot be totally detached from the support; thereby, the result of the ATP active layer should be subtracted from the contribution of the background (Al2O3 support).

The pore size distribution was measured through an AutoPore IV9500 style mercury porosimeter (Micromeritics, USA) based on Washburn's equation (Chibowski & Holysz 1992):

 
formula
2
where r is the pore size, nm; θ is the contact angle between the membrane and solution, degrees; γ is the surface tension of mercury (0.48 N/m) and p is the imposed pressure, MPa.

The surface morphology was observed through an FEI Quanta FEG 250 style field emission environmental electron microscope; before scanning, the samples were gold coated through a sputtering device. Crystalline properties of the as-prepared samples were characterized through a D/max2500VB 2 + /PCX style X-ray diffractometry (XRD, Rigaku Corporation, Japan). Zeta potentials of the membrane were determined based on streaming potential analysis and the Helmholtz–Smoluchowski model (Rodemann & Staude 1995); the pHs of influent solutions were adjusted to a desired value by adding negligible volumes of 0.1 M HCl or 0.1 M NaOH solution.

Laboratory-scale filtration experiment

The filtration experiments were conducted using a stirred cell with a volume of 400 mL described in our previous study (Wei et al. 2013). The DI water or model solutions were pressurized through the flat membrane with the help of a nitrogen cylinder. Pressures were monitored by a manometer of the treatment chamber.

Prior to filtration, a compaction procedure was performed as follows: the fresh membrane was immersed in DI water for at least 2 h and then was conditioned by filtering DI water under 0.6 MPa for 1 h. The membrane was also subjected to equilibrate in the test solution for a while before each experimental run. The pure water permeability (J, L m–2 h–1) could be mathematically described by Equation (3): 
formula
3
where A denotes the effective filtering area, m2; t denotes the filtering time, h; and V is the volume of filtrate, L.
The ion rejection performance (R, %) of the ATM was evaluated according to Equation (4): 
formula
4
where ρ0 and ρf represent the ion concentrations in the feed solution and filtrate, respectively, mg/L. All the laboratory-scale tests were conducted at room temperature (27 °C), and run in duplicate or more to check the reproducibility.

Field test

Several filters were developed based on the ATM, which were mounted on the continuous blowdown pipe of a WNS20-1.57YQ style boiler. The working pressure of this boiler was 1.1 ± 0.1 MPa with a saturated steam temperature of 183 ± 1 °C. The water quality of the CBW before and after treatment was recorded from an online monitoring device. The blowdown rate (η, %) was evaluated according to Equation (5): 
formula
5
where QCBW and QS represent the volume of CBW and steam product, respectively, m3 h–1. In normal operation, the average amount of steam product (QS) is about 18.70 m3 h–1.

RESULTS AND DISCUSSION

Determination of optimum sintering temperature

Ceramic coating on an Al2O3 support is a double sintering process. The connection between the active layer and support will be not tight enough under relatively low sintering temperature; however, surface flaws, such as cracks, will appear at excessively high temperatures, especially for the asymmetric membrane. To this end, it is worth determining the optimum sintering temperature for the ATP active layer.

Figure 1 shows the adhesive strength levels between the active layer and support under different sintering temperatures. In relation to asymmetric membranes, the adhesive strength between the active layer and support should be at least Grade 3 (Standard GB/T 9286-1998). From Figure 1, the adhesive strength shows an enhanced tendency with elevated temperature, and reaches Grade 3 at 800 and 850 °C, rising to Grade 2 at 900 and 950 °C. Therefore, the sintering temperature should be over 800 °C to achieve the specification requirement.
Figure 1

Adhesive strength levels between the ATP active layer and the Al2O3 support under different sintering temperatures.

Figure 1

Adhesive strength levels between the ATP active layer and the Al2O3 support under different sintering temperatures.

Moreover, porosity is another key parameter for ceramic membranes, since it can affect the hydraulic performance during the separation and purification process. To a certain degree, the decisive factor affecting porosity is the sintering temperature. Figure 2 shows the surface morphology and porosities of the as-prepared membranes at different sintering temperatures. It can be observed that the pores of the ATM seem to be dying out as the temperature increases. In addition, Figure 2(d) reveals that the porosity declines steadily with increasing temperature: the determined values are 36.75%, 27.83% and 20.04% at sintering temperatures of 800 °C, 850 °C and 900 °C, respectively. The decreasing tendency can be attributed to the melting of raw material under relatively high temperature, which is conducive to the formation of liquidoid. Although the surface adhesive strength is enhanced (Figure 1) because of the penetration of liquidoid into the membrane pores, the crystal structure of the ATP might be impaired for the fusion of adjacent mineral particles. This subsequently leads to severe blockage of pores, generating a low porosity. In general, the porosity of ceramics is usually more than 30% (Maarten Biesheuvel & Verweij 1999); therefore, from the standpoints of adhesive strength and porosity, the sintering temperature for the ATM is selected at 800 °C in this work.
Figure 2

Surface morphology (sintering temperature: (a) 800 °C; (b) 850 °C; (c) 900 °C) and porosity (d) of ATM.

Figure 2

Surface morphology (sintering temperature: (a) 800 °C; (b) 850 °C; (c) 900 °C) and porosity (d) of ATM.

Characterization of ATM

ATP powder has unique crystal characteristics, which enable it to exhibit permanent negative charges on the particle surface (Huang et al. 2007). Therefore, it is of value to analyze the crystal information of the ATP before and after sintering through XRD technology, and the results are shown in Figure 3. As can be observed, both ATP powder and the as-prepared membrane are composed of SiO2, Mg2SiO4, and a small content of Al4.5FeSi as well as Al8Si6Mg3Fe. The characteristic d spacing of 10.64, 4.48, 4.25, and 3.34 Å confirms the sample used in this work is ATP (Huang et al. 2007). After sintering, the peak intensity of d spacing at 10.64 Å lessens noticeably; conversely, that of d spacing at 3.34 Å enhances sharply. This is evidence that the membrane forming process only changes the crystallinity of certain phases rather than the charge-related parameters, such as the crystal lattice, lattice parameter and crystallite dimensions. Therefore, it is reasonable to assume that the ATM would keep a charge property similar to that of ATP powder.
Figure 3

XRD patterns of ATP before and after sintering at 800 °C.

Figure 3

XRD patterns of ATP before and after sintering at 800 °C.

Further study on the pore size distribution of the Al2O3 support and ATM is shown in Figure 4. It can be seen that the pore size of the support ranges from 7,178.85 to 8,239.74 nm; with respect to the active layer, the pore sizes are concentrated on 13.39 nm. Combining Washburn's theory (Chibowski & Holysz 1992) and Equation (2), the mean pore size of the active layer is observed to be 12.7 nm. Therefore, the much more open morphology of the support compared with that of the ATP layer is easily visible. This result is directly supported by SEM images shown in Figure 5(a) and 5(b). It is worth mentioning that the active layer is formed by irregular accumulation of fibrillar ATP particles, which would hardly intrude into the pores of the support during the dip-coating process, in contrast to their fine spherical counterparts. Hence, it is unnecessary to prepare the traditional intermediate layer for a high performance membrane (Fang et al. 2011). In addition, irregular accumulation of particles is conducive to obtaining an active layer with high porosity and tortuosity, which is prone to remove contaminants in the wastewater efficiently (De Angelis & de Cortalezzi 2013). According to Figure 5(c), the active layer can be clearly distinguished from the support, with an approximate thickness of 40μm from a cross-sectional view.
Figure 4

Pore size distribution of the support and ATP membrane.

Figure 4

Pore size distribution of the support and ATP membrane.

Figure 5

SEM images of Al2O3 substrate (a), ATP layer (b) and cross section (c).

Figure 5

SEM images of Al2O3 substrate (a), ATP layer (b) and cross section (c).

Rejection performance of ATM for synthetic CBW (laboratory scale)

Before developing the module based on the ATM, it is of primary importance to collect the hydraulic data for the ATM in the laboratory. To evaluate the membrane permeability under dead-end mode, DI water was subjected to permeation under different operating pressures as shown in Figure 6. It is observed that the ATM exhibits an almost linear relationship between flux and trans-membrane pressure (TMP) with a permeability of 1,411.87 L m–2 h–1 MPa–1 at pH 6.36 and a temperature of 27 °C. It is noteworthy that the permeability of the ATM is relatively lower than that of reported values with similar pore size distributions (Llanos et al. 2010; Xu et al. 2010; Muthukumaran & Baskaran 2014). This may be ascribed to the thick active layer as shown in Figure 5(c), which can increase the intrinsic membrane resistance of the ATM inextricably, and subsequently lead to a decline in permeability.
Figure 6

Variation of pure water flux as a function of driving pressure.

Figure 6

Variation of pure water flux as a function of driving pressure.

Furthermore, according to the practical operation of the WNS20-1.57YQ boiler, the outlet pressure of the CBW maintained at 0.14 ± 0.01 MPa after decompression and expansion through an expansion tank. Thus, except for special indications, the driving pressure for dead-end filtration is 0.14 MPa in this work. Figure 7 presents the hydraulic performance of the ATM for treating the single salt solution. As can be observed from Figure 7(a) and 7(b), water flux for each model solution deteriorates gradually as the ion concentration increases. In terms of ion retention, the rejection of CO32– seems to be maintained at a high level (∼91%) throughout the concentration range based on Figure 7(a). Also, the rejection of SO42– has an increasing trend from the initial concentration and finally stays stable at about 92% according to Figure 7(b).
Figure 7

Rejection performance of the ATP for a single electrolyte solution at 27 °C: (a) sodium carbonate; (b) sodium sulfate.

Figure 7

Rejection performance of the ATP for a single electrolyte solution at 27 °C: (a) sodium carbonate; (b) sodium sulfate.

Based on the data of the hydrated ionic radius, a preliminary judgment can be made that the rejection mechanism of the ATM for target ions could hardly be the physical sieve effect: the hydrated radius of SO42– and CO32– are 0.300 nm and 0.394 nm, respectively (Tansel et al. 2006), which are far smaller than the average pore size of the ATM (shown in Figure 4). Therefore, the characteristics of ion retention and water flux may be attributed to the surface electrical properties of the ATM, such as Donnan exclusion, dielectric effect and Coulomb repulsion. In addition, feed solutions in this study were prepared through sodium carbonate (Na2CO3) and sodium sulfate (Na2SO4). Hence, Na+ has a noticeable reserve in each feed, which can aggravate the flux of nano-scale membranes and generate a decreasing tendency with the increase in ion concentration (Kaewsuk et al. 2012).

In order to provide evidence for our assumption on the rejection mechanism and surface electrical properties of the ATM, the retention of salts (Na2CO3 and Na2SO4) at different pH values were conducted as shown in Figure 8. The concentrations of two ion species were selected in reference to that in the real CBW (listed in Table 1). Moreover, given that CO32– is a radical group in the acidic medium, the degree of ionization and speciation will be affected by pH. Thus, in order to reflect real ion retention of CO32–, the initial testing pH value was selected at eight for the rejection process. Figure 8 reveals that the ATM exhibits a pH-sensitive property for ion retention. It may be concluded that ion retention versus pH curves are the signature of a charged membrane that rejects ions principally by electrostatic interactions (Luo & Wan 2013). The zeta potential of the ATM provides further direct evidence for surface electrical properties of the ATM as shown in Figure 9. It is worth further attention that an indifferent electrolyte (NaCl) was used as the test medium in order to avoid a potential shift caused by the adsorption of asymmetric salts (Na2SO4 and Na2CO3). According to Figure 9, the ATM shows amphoteric behavior with an isoelectric point (IEP) at about pH 5.0, which is almost in line with the results obtained from Figure 8. The small deviation may result from the adsorption of divalent anions, which subsequently results in the early charge inversion of the membrane surface.
Figure 8

Retention of CO32– and SO42– under various pH values at 27 °C.

Figure 8

Retention of CO32– and SO42– under various pH values at 27 °C.

Figure 9

Variation of ζ potential of the ATM as a function of pH.

Figure 9

Variation of ζ potential of the ATM as a function of pH.

According to Figure 8, for sulfate, ion retention seems to be inefficient in the low pH range (less than 4.9); subsequently, the rejection rate shows an increased tendency, and finally keeps stable at around 90% as pH increases. For carbonate, the rejection rate increases with the elevated pH values, and a similar steady state appears when the pH exceeds 11. Based on conclusions made by other researchers (Van der Bruggen et al. 2001), the IEP of the ATM is expected to have a pH value of 4.9 for the simulated CBW.

Furthermore, allowing for the mean pore size of the ATM (12.7 nm), a little SO42– and CO32– can be directly rejected by the membrane through its surface electrical properties. Much work has indicated that the crystal lattice of ATP contained various cations (Na+, Ca2+, and Al3+) (Deng et al. 2013; Dong et al. 2013). According to the XRD patterns shown in Figure 3, the main crystal structure of ATP was not changed drastically after sintering. Thus, it is reasonable for cations to be still kept in the original lattice. Some cations in the lattice can exchange with Ca2+, Mg2+ and other cations, when the membrane contacts with polar feeds (Huang et al. 2007). However, this exchange process is hard for Na+ in the lattice based on the conclusion drawn by Liu et al. (2013). As a result, the concentration of Na+ in the membrane pores is lower than that in the bulk solution, and Donnan exclusion can be formed (Vezzani & Bandini 2002). For the rejection of SO42–, the ATM presents a positive surface charge when the pH values were less than the IEP, because of the confined cations in the lattices. Therefore, Na+ is rejected due to electrostatic repulsion; SO42– is also rejected for charge balance. With increasing pH, the rejection rate decreases because the membrane charge decreases as well, and reaches a minimum at a pH value corresponding to the IEP.

At pH >IEP, another mechanism may be responsible for the increasing rejection rate as the pH increases, which is depicted in Figure 10. Owing to the cations kept in the lattices of the ATM, anions in the feed can be adsorbed on the inner wall of pores through Coulomb's law. Adsorption of anions generates a negative charged surface, for which the concentration of anions in the pores of the ATM is larger than that in the bulk solution. In this case, the dielectric effect plays a key role in co-ion retention (Déon et al. 2009). In order to keep electrical neutrality, cations such as Na+ are forced to stay on the feed side. Moreover, hydroxyl ions (OH) will enhance the dielectric effect at the pH range over 7. Therefore, the rejection rate increases gradually, since the retention of anions is vigorous by the enhanced negatively charged surface.
Figure 10

Schematic diagram of the rejection mechanism of ATM (pH>4.9).

Figure 10

Schematic diagram of the rejection mechanism of ATM (pH>4.9).

The rejection process towards CO32– seems to be complicated due to the special dissociation constant of carbonate. At room temperature, monovalent ions (HCO3) are dominant in the feed at pH 8 or lower (Calderón et al. 2008). As pH increases, HCO3 will deprotonate, followed by the formation of CO32–. Divalent ions can be rejected following a mechanism akin to that of sulfate mentioned above.

In general, Donnan exclusion and the dielectric effect are the essential factors to determine the retention efficiency of the ATM. In detail, Donnan exclusion is more prominent than the repulsion of the dielectric effect for sulfate retention in the pH range below the IEP. The very reverse, the dielectric effect is the main cause of increasing retention when the pH value is over the IEP. Considering the particularity of CO32–, the dielectric effect might be the controlling factor during the limited ion retention.

Figure 11 shows the treatment effect towards practical CBW of the ATM. During filtration, both the samples of feed and filtrate were taken in triplicate, respectively; and the ion concentration of either SO42– or CO32– was measured to evaluate the retention efficiency. Based on Figure 11, a relatively stable rejection rate can be observed for both ions, which indicates the ATM still has the excellent separation performance for CBW. Another aspect which deserves attention is that small suspended solids and organic species in CBW also contribute to the high ion retention, because the ATM can hardly be fouled or clogged by solid impurities. The marginally downward trend of the two curves may be induced by reversible ion accumulation at the membrane vicinity. This phenomenon, normally known as concentration polarization, manifests itself as flux decline and decreased transmission over time. In this respect, however, it is prone to enhance the electroviscous effect caused by ion aggregation in narrow membrane pores. Thus, trans-membrane diffusion of the target ions can be retarded, yielding high ion retention (Bowen & Yousef 2003).
Figure 11

Treatment of CBW from the industrial boiler by ATM at 27 °C.

Figure 11

Treatment of CBW from the industrial boiler by ATM at 27 °C.

Application of ATM to the industrial boiler (field test)

Several filters based on the ATM were developed with an effective area of 0.7 m2, which were mounted onto the outlet pipe of an expansion tank serving the WNS20-1.57YQ low-pressure boiler in dead-end filtration mode. The process flowchart of our field test is depicted in Figure 12. It is noteworthy that alkalinity is usually used as a substitute for the concentration of CO32–, OH and HCO3.
Figure 12

Schematic diagram of the industrial boiler water/steam system.

Figure 12

Schematic diagram of the industrial boiler water/steam system.

Table 4 lists the monitored parameters of the CBW before and after treatment. On observing the four parameters of treated water, they are all improved in different degrees. In detail, alkalinity has a substantial decrement to be maintained at an average value of 1.22 mM L–1; whereas the pH of the treated water remained outside the required standard. This can be attributed to the insufficient rejection of monovalent anions (OH), which can keep the filtrate alkaline. In addition, the rejection of SO42– is in good agreement with the results of laboratory-scale experiments. Removal of dissolved solids in the CBW results in a pronounced decline of conductivity.

Table 4

Water quality parameters of CBW before and after treatment

Parameter Unit Before treatment After treatment National standard 
Alkalinity mM L–1 Max. 17.52 Min. 11.50 Mean 13.28 Max. 1.99 Min. 0.98 Mean 1.22 6–24 
pH – Max. 11.87 Min. 11.23 Mean 11.49 Max. 9.33 Min. 8.87 Mean 9.06 7–9 
SO42− mg L–1 Max. 419.88 Min. 391.29 Mean 407.11 Max. 38.16 Min. 25.36 Mean 27.52 10–30 
Conductivity μS cm–1 27 °C Max. 3,860.09 Min. 3,440.00 Mean 3,580.15 Max. 522.83 Min. 507.24 Mean 515.22 ≤550 
Parameter Unit Before treatment After treatment National standard 
Alkalinity mM L–1 Max. 17.52 Min. 11.50 Mean 13.28 Max. 1.99 Min. 0.98 Mean 1.22 6–24 
pH – Max. 11.87 Min. 11.23 Mean 11.49 Max. 9.33 Min. 8.87 Mean 9.06 7–9 
SO42− mg L–1 Max. 419.88 Min. 391.29 Mean 407.11 Max. 38.16 Min. 25.36 Mean 27.52 10–30 
Conductivity μS cm–1 27 °C Max. 3,860.09 Min. 3,440.00 Mean 3,580.15 Max. 522.83 Min. 507.24 Mean 515.22 ≤550 

In summary, based on the prescribed National Standards (Standard GB/T1576-2008), it can be concluded that all the parameters meet the criteria of the softened water except pH value, which is limited between 7 and 9. In order to make the pH value be appropriate to the standard, the demineralizer system out of the boiler should be operated at a low load to supply a small amount of softened water. In this way, the filtrate treated by ATM can be used as the make-up water and pumped into the boiler after mixing with the softened water.

After the application of ATM filters, the volume of the CBW has a noticeable decrement from 1.82 to 0.37 m3 h–1. According to Equation (5), the evaluated blowdown rate is about 1.98%, which is far lower than the original value (about 9.7%).

Besides reusing the treated water, another aspect which should be mentioned is that the heat contained in the CBW could also be recycled along with the filtrate. In this case, recycling the treated CBW can reduce the consumption of fuels, which will have various environmental benefits.

CONCLUSION

An Al2O3 supported ATP membrane was prepared for carbonate and sulfate removal. Based on the results of surface morphology, porosity and adhesion strength between the active layer and substrate, the sintering temperature for ATM was selected as 800 °C. The sintering process only changed the interplanar spacing rather than the crystal phase species of ATP according to the XRD analysis. The mean pore size of the ATP active layer was calculated as 12.7 nm with a thickness of about 40μm.

Under dead-end mode, ATM exhibited a good linear relationship between pure water flux and TMP with a permeability of 1,411.87 L m–2 h–1 MPa–1 at room temperature (27 °C). The rejection performance of ATM exhibited the pH-sensitive property for carbonate and sulfate solutions with an IEP at pH 4.9. Donnan exclusion and the dielectric effect could be responsible for the high divalent ion retention: when the pH was over 11, the rejection rate could maintain at about 90%.

Several filters were developed based on ATM to purify the practical CBW from a low-pressure boiler. According to the data from the field test, the parameters of water quality after treatment were akin to that of make-up water for low-pressure boilers except for the pH values. Thus, the treated water could be reused after mixing with a defined amount of softened water, which would reduce the operational load of the demineralizer system. In sequence, the continuous blowdown rate had a substantial decrease to about 1.98%. In addition, the obtained results revealed that recycling the treated CBW could essentially be significant in saving water and energy.

ACKNOWLEDGEMENTS

This work would not have been possible without the financial support from the National Natural Science Foundation of China (No. 31501435) and Research Initiation Funds for Young Teachers of Beijing Technology and Business University (QNJJ2015-30). The authors also thank the financial support for field tests from the Innovation Research Team of Polymeric Functional Film of Beijing Technology and Business University (No. 19008001071).

REFERENCES

REFERENCES
APHA/AWWA/WEF
1998
Standard Methods for the Examination of Water and Wastewater
.
20th edn
.
American Public Health Association/American Water Works Association/Water Environment Federation
,
Washington, DC
.
Bowen
W. R.
Yousef
H. N. S.
2003
Effect of salts on water viscosity in narrow membrane pores
.
J. Colloid Interf. Sci.
264
,
452
457
.
Calderón
J. A.
Barcia
O. E.
Mattos
O. R.
2008
Reaction model for kinetic of cobalt dissolution in carbonate/bicarbonate media
.
Corros. Sci.
50
,
2101
2109
.
Chang
Q.
Zhou
J.-E.
Wang
Y.
Liang
J.
Zhang
X.
Cerneaux
S.
Wang
X.
Zhu
Z.
Dong
Y.
2014
Application of ceramic microfiltration membrane modified by nano-TiO2 coating in separation of a stable oil-in-water emulsion
.
J. Membr. Sci.
456
,
128
133
.
Kaewsuk
J.
Lee
D. Y.
Lee
T. S.
Seo
G. T.
2012
Effect of ion composition on nanofiltration rejection for desalination pretreatment
.
Desalin. Water. Treat.
43
,
260
266
.
Liu
W.
Yang
T.
Xu
J.
Chen
Q.
Yao
C.
Zuo
S.
Kong
Y.
Fu
C.
2015
Preparation and adsorption property of attapulgite/carbon nanocomposite
.
Environ. Prog. Sustain. Energy
34
,
437
444
.
Rana
D.
Matsuura
T.
2010
Surface modifications for antifouling membranes
.
Chem. Rev.
110
,
2448
2471
.
Shirazi
M. M. A.
Kargari
A.
Bazgir
S.
Tabatabaei
M.
Shirazi
M. J. A.
Abdullah
M. S.
Matsuura
T.
Ismail
A. F.
2013
Characterization of electrospun polystyrene membrane for treatment of biodiesel's water-washing effluent using atomic force microscopy
.
Desalination
329
,
1
8
.
Standard GB/T 1576-2008
Water Quality for Industrial Boilers
.
General Administration of Quality Supervision, Inspection and Quarantine of China, Standardization Administration
,
China
.
Standard GB/T 6907-2005
Analysis of Water Used in Boil and Cooling System – The Sampling Method of Water
.
General Administration of Quality Supervision, Inspection and Quarantine, Standardization Administration
,
China
.
Standard GB/T 9286-1998
Paints and Varnishes – Cross Cut Test for Films
.
General Administration of Quality Supervision, Inspection and Quarantine
,
China
.
Tansel
B.
Sager
J.
Rector
T.
Garland
J.
Strayer
R. F.
Levine
L.
Roberts
M.
Hummerick
M.
Bauer
J.
2006
Significance of hydrated radius and hydration shells on ionic permeability during nanofiltration in dead end and cross flow modes
.
Sep. Puri. Technol.
51
,
40
47
.
Van der Bruggen
B.
Daems
B.
Wilms
D.
Vandecasteele
C.
2001
Mechanisms of retention and flux decline for the nanofiltration of dye baths from the textile industry
.
Sep. Purif. Technol.
22–23
,
519
528
.