A study of inline chemical coagulation/precipitation-ceramic microfiltration and nanofiltration for reverse osmosis concentrate minimization and reuse in the textile industry

Ca(OH)₂

Calcium hydroxide

Ca²⁺

Calcium

CMF

Ceramic microfiltration

COD

Chemical oxygen demand

EC

Electrical conductivity

FeCl₃

Ferric chloride

J

Wastewater flux

J/J₀

Flux decline

J₀

Wastewater initial flux

kDa

Kilodalton

MF

Microfiltration

Mg²⁺

Magnesium

MgO

Magnesium oxide

MWCO

Molecular weight cut-off

NaOH

Sodium Hydroxide

NF

Nanofiltration

Pt–Co

Platinum–Cobalt

R

Resistance

RO

Reverse osmosis

ROC

Reverse osmosis concentrate

Sal%

Salinity (%)

SiO₂

Silica

TOC

Total organic carbon

UF

Ultrafiltration

Due to rapid growth in population, urbanization, and industrialization, the reuse of industrial wastewater has become a trend for sustainable water management strategies to reduce freshwater consumption and minimize contamination of receiving water bodies (Yaqub & Lee 2019). The textile industry is one of the largest water-consumer industries, the reclamation of textile wastewater using a membrane-based reverse osmosis (RO) technology is widespread these days (Sahinkaya et al. 2018). This technology for the recovery of water and valuable salts especially in intensive water-consuming sectors like the textile industry has incited deep interest because of permeate quality (Joo & Tansel 2015), energy consumption, and safe operation (Yu et al. 2013). In addition, the best available technique reference documents (BREFs) emphasized an application of membrane technologies for effective water and salt recovery in the textile industry (Ozturk et al. 2016). However, one of the main limitations in RO application is the disposal of the RO concentrate (ROC) because it has a much higher concentration of contaminants as compared to the feed water (Zhao et al. 2019). The ROC ranges 15–50% of feed water which must be disposed of or treated for reuse (Warsinger et al. 2018). If untreated or improperly managed ROC is discharged to water bodies and it can cause adverse effects such as eutrophication, high salinity (Joo & Tansel 2015), and organic pollution (Tran et al. 2012) in the aquatic environment. Additionally, stringent environmental regulations are forcing the textile sector to reuse the ROC after further treatment to achieve zero liquid discharge (ZLD).

Membrane fouling is one of the most common issues, and it refers to the deposition or adsorption of contaminants on the membrane surface (Shon et al. 2002). Organic contaminants with particle sizes larger than the pores of the membrane are easily adsorbed on the membrane's surface during filtration (Elimelech & Wiesner 2002). This can happen due to the following reasons: (1) colloidal fouling; (2) organic fouling due to the deposition of organic substances; and (3) scaling, which is known as the formation of mineral deposits precipitating from the feed stream onto the membrane surface results in a loss of permeate flux through the membrane (Duranceau 2001). Due to these reasons, minimizing the scale-forming ions in ROC for further treatments is a significant issue. Since the presence of divalent cations is crucial in the membrane fouling process, it is critical to understand their function and mechanism. Calcium and magnesium ions, which are the most common and abundant cations in water settings, are the most representative divalent inorganic ions (Meng et al. 2020). In a recent study, the effect of Mg²⁺ on polysaccharide fouling during ultrafiltration was investigated, and it was discovered that gel layer formation was the dominant fouling mechanism in the presence of Mg⁺² (Wang et al. 2020). Therefore, pre-treatment before the membrane process is an attractive option to reduce the risk of fouling that ultimately increases treatment efficiency (Semblante et al. 2018). The primary target for pre-treatment of ROC is the removal of divalent ions (Sahinkaya et al. 2018) and organic matter (Umar et al. 2016) as studied by the addition of chemicals (Subramani & Jacangelo 2014). Coagulation has shown a great effect on organic matter removal but not sufficient for dealing with the high concentrations of scale-forming ions in ROC (Ho et al. 2015). The use of single treatment technology to achieve ZLD/near-ZLD is not possible for ROC treatment and thus a combination of technologies is essential (Subramani & Jacangelo 2014). The selection of the best pre-treatment method for further treatment of ROC to increase water recovery depends on the process requirements (Pérez-González et al. 2012). For instance, treatment of ROC through membrane filtration requires the removal of organics and scale precursors to prevent scaling and fouling (Semblante et al. 2018). As a result, a combination of coagulation with chemical precipitation could be useful for creating a synergetic pre-treatment (Ordóñez et al. 2012). By comparison, ceramic MF/UF membranes have the potential for pre-treatment when combined with oxidation processes to remove pollutants and prevent membrane fouling (Gray et al. 2015).

Some researchers have widely studied state-of-the-art technologies for RO concentrate minimization. Recent studies have investigated using a seeded precipitation stage that has been integrated into the ceramic ultrafiltration process for the removal of precipitated Ca²⁺ to prevent membrane fouling and increase water recovery (Sanciolo et al. 2012). Another simultaneous treatment hybrid system, consisting of coagulation with FeCl₃ + photocatalysis (UVC/TiO₂) was used and this system achieved 95% organic matter removal from the ROC within 6 hours (Zhou et al. 2011). Using a pellet reactor achieved 85% Ca²⁺ removal at a pH of 11.5, this removal was increased up to 95% when the reactor was combined with electrodialysis (ED) (Tran et al. 2012). The coagulation process in combination with the UV/H₂O₂ system achieved 17–27% additional DOC removal and suggested FeCl₃ as the most efficient coagulant (Umar et al. 2016). Some researchers achieved 80% SiO₂ removal from brackish water ROC using electro-coagulation under a current intensity of 0.5 (A) for 30 minutes of hydraulic retention time (Den & Wang 2008) and a seeded precipitation process was applied with a two-stage RO system, which presented 98% water recovery (Gabelich et al. 2007).

Generally, ROC studies have been carried out with groundwater, desalination, and municipal wastewater reclamation. Due to the presence of high color, organic matter, toxic compounds, and salt concentrations, textile ROC is quite difficult to treat (Holkar et al. 2016). The novelty of the research is the application of CMF and its combination with CP for the removal of Ca²⁺, Mg²⁺, and SiO₂ from textile ROC. This work aims to (1) investigate the CMF pre-treatment to remove the precipitated scaling ions, (2) explore the efficiency of CP and CMF integrated system from textile ROC streams, (3) prevent membrane fouling using CP-CMF pre-treatment, and (4) conduct cost–benefit analysis for economic feasibility to scale-up the system.

Characterization and filtration experiments were conducted with real ROC samples that were collected from the household linen processing textile industry. This industry is located in Tekirdag, Turkey with a capacity of 5,000 m³/d wastewater treatment. The treatment system consists of mechanical–physical treatment followed by biological treatment (activated sludge process). At this plant, biologically treated wastewater runs through the coagulation–pre-filtration unit and activated carbon adsorption unit, before undergoing UF and RO membrane processes. RO permeate is used for cooling tower water requirements. The existing wastewater treatment system scheme selected for this study is shown in Figure 1.

Ten samples of the ROC stream were collected and tested for COD, TOC, pH, EC, color, total hardness, Ca²⁺, Mg²⁺, and SiO₂ analysis. Measured characteristics were in the range as reported in the literature for pH, COD, conductivity, Na⁺, SO₄²⁻ (Sahinkaya et al. 2018), and TDS, TP, and Cl⁻ (Dhodapkar et al. 2007). Differences were observed in silica and Mg²⁺ concentrations, in the present study, silica concentration was higher and Mg²⁺ concentration was lower than similar effluent average values. The characteristics of textile ROC in comparison with literature studies are presented in Table 1.

All the chemicals used in the study were of analytical grade and purchased from Merck, Germany (FeCl₃, MgO, Ca(OH)₂), Sigma Aldrich (NaOH), and Carlo Erba (MgSO₄·7H₂O). The stock solution of FeCl₃ and NaOH was prepared with ultrapure water (18.18 mΩ·cm). In this study, we used tubular type ceramic membranes with three different molecular weight cut-off (MWCO) (0.2 μm, 15 kDa, 300 kDa), which were procured from TAMI Industries (Nyons, France) and polymeric NF membranes (NF270 and NF90) from Dow Filmtec™. The specifications of membranes are presented in Table 2.

NF270 and NF90 membranes were examined in terms of their removal efficiencies for color, TH, EC, and salinity on the permeate side. The membrane specifications selected for filtration tests are shown in Table 2.

The crossflow filtration through tubular ceramic membranes contains a peristaltic pump (Filltech, BT 100 2 J), a hot plate and magnetic stirrer (Daihan, MSH-30DSET), a holder, concentrate backpressure valve (1–10 bar), a precision scale (AND, FX-5000i) and a computer, as shown in Figure 2.

The Sterlitech HP 4750 stainless-steel stirred cell dead-end filtration unit was used for the NF experiments. The maximum operating pressure of the cell was 69 bar (1,000 psi). Stirring was applied to eliminate concentration polarization with a PTFE stir bar. The cell volume of the system was 300 mL and the membrane filtration area was 14.6 cm². In the experiment, pure water filtration tests were run before each experiment for compaction. Pre-treated and raw ROC filtration experiments were conducted with an 80% recovery rate. Membrane permeates’ flux and rejections rates were calculated using Equations (3) and (4) respectively.

The membrane fouling mechanisms were studied through the use of the resistance in-series model which illustrates the main fouling mechanism. Rm which is the hydraulic self-resistance of the clean membrane was identified by measuring the steady-state flux of pure water. Rt was determined by measuring the steady-state flux of the ROC filtration, Rc was specified by measuring the pure water flux after physically cleaning the membrane surface and Rp was specified by pure water flux after chemical cleaning. The sum of the Rm, Rp, and Rc indicated total resistance to the fouled membrane (Equation (6)).

COD, pH, EC, color, Ca²⁺, Mg²⁺, and SiO₂ analysis were performed for the ROC and membrane filtration permeate and concentrate samples. The EC and pH measurements were conducted using the WTW series Inolab pH-EC meter. The COD analyses were carried out using measurement test kits (Hach Lange, Germany) and the Hach Lange DR6000 UV/VIS spectrophotometer. A Thermo reactor system (Spectroquant TR320, Merck) was used for COD analysis. Color (as Pt–Co) analysis measurement was performed according to the Standard Methods (APHA 2005). The TOC was analyzed using a Shimadzu analyzer (Shimadzu, TOC-V CPH model) according to the SM-5310B method. The Ca²⁺, Mg²⁺, and SiO₂ analyses were conducted using an ICP-OES instrument (Perkin Elmer OPTIMA 8300 DV Model) with the ISO-11885 method.

To prevent organic and inorganic fouling on the membrane surface, the ROC was pre-treated using the coagulation process at laboratory scale. The coagulant was selected based on a literature study for similar textile ROC treatment (Umar et al. 2016), FeCl₃ was found to be the best coagulant and optimized with the dosage of FeCl₃ (1-2-3 and 4 mM) at pH values of 4.0, 5.0, 6.0, and 7.0. The temperature was kept constant at 25±1 °C and FeCl₃ was added into a jar with 2 L textile ROC while rapidly mixing at 200 rpm for 5 min. It was then mixed at 40 rpm for 15 min. The solution was then settled for an hour and filtered through the 0.2 μm tubular ceramic membrane filtration system. To simulate a CP–CMF process in a laboratory, as is depicted in Figure 2, the optimum dosage was found to be 3 mM at pH 7.

Depending on the type of wastewater; Semblante et al. (2018) stated that a range of pH from 3–7 could be used for removal of TOCs, Liu et al. (2019) found the optimum pH level to be 5 as well as Ho et al. (2015) who declared that pH 7 was optimum in the coagulation process. At neutral pH, removal of TOC, Ca²⁺, SiO₂, and Mg²⁺ were noted as 87, 45, 30, and 25%, respectively. At pH values of 7 or higher, there was a small decrease in COD removal. The distribution of iron ionic species may explain this result (Duan & Gregory 2003).

High pH is required for a successful softening process and pH ranges were selected based on literature values. Ordóñez et al. (2012) used CaCO₃ and Mg(OH)₂ in the range pH of 9.5–10.5 and (Sims 2015) used Ca(OH)₂ at pH >10. NaOH and Ca(OH)₂ solutions were investigated to compare their efficiencies and to optimize the required pH adjustment for softening processes in this study. The pH of the samples was adjusted by adding alkaline agents (NaOH and Ca(OH)₂) after 15 min. Of mixing at 200 rpm, and after the desired pH levels were reached, the solution was settled for 1 h and then filtered through CMF.

Use of extreme conditions for precipitation (Ca(OH)₂ and NaOH), high pH and conductivity values in the effluent were obtained. Ca(OH)₂ and NaOH processes produced alkalinity in the form of bicarbonates and these processes reduced COD (up to 43.7%) (Prazeres et al. 2020). In our case using Ca(OH)₂ instead of NaOH, enhanced COD removal from 10% to 30%, as is similar to in a study conducted by Latour et al. (2014b).

NaOH caused an important increase in the conductivity and salinity of the feed water in contrast with Ca(OH)₂, the conductivity of treated water increased by about 0.5 mS/cm similar to the literature result (Semerjian & Ayoub 2003; Latour et al. 2014a). Also with the use of NaOH, there was been no significant Mg²⁺ and SiO₂ removal. The optimum pH level was found to be using Ca(OH)₂ at pH 11.5, in terms of removal efficiency of Ca²⁺, Mg²⁺, and SiO₂ (Latour et al. 2015).

For SiO₂ removal during the softening treatment, two main mechanisms were regarded: precipitation as calcium and/or magnesium silicate or by adsorption (Latour et al. 2015). Precipitation as magnesium silicate is dominant at high pH (Rioyo et al. 2018). It was observed that SiO₂ can be reduced by precipitation as magnesium silicate when the molar ratio of Si/Mg²⁺ is higher than 1/6 (Parks & Edwards 2007). According to the data given in Table 3, the mass ratio of Si/Mg²⁺ was about 2/1 in the ROC, which means sufficient Mg²⁺ was not present. Due to the relatively low Mg²⁺ content of the ROC, Mg²⁺ compounds should be added to increase SiO₂ removal at high pH. Recent literature studies showed that 80–90% SiO₂ removal can be obtained by adding MgCl₂·6H₂O and MgSO₄·7H₂O at a high pH level with high dosages of chemicals (1,500 mg/L) at ambient temperature (Latour et al. 2014a). However, adding these Mg²⁺ compounds causes an increase in the conductivity of the ROC (Latour et al. 2015). In this study, a sparingly soluble Mg²⁺ compound (MgO) was used for SiO₂ removal. MgO was tested at four different dosages (from 100 to 400 mg/L) and different pH levels (9.5, 10.5, and 11.5). pH levels and MgO dosages were selected according to previous studies carried out by the most recent literature studies with similar effluent (Sims 2015). Using sparingly soluble compounds of Mg²⁺ has the advantage of adding less conductivity to water while causing no further scaling problems. The maximum SiO₂ removal rate was obtained 92% with the addition of 200 mg/L MgO at pH 10.5 and 11.5 as shown in Figure 3, while there was no significant removal efficiency at pH 9.5 from the raw ROC.

The CP–CMF process was conducted in the laboratory as depicted in Figure 2 and 0.2 μm, 300 kDa, and 15 kDa tubular type ceramic membranes were used. All the pre-treatment methods were applied both alone and in combination. The flux reduction fractions (J/J₀), was calculated based on the feed flux (J) values at a certain time divided by the initial feed flux (J₀) value generated during crossflow experiments. The initial and steady-state flux values obtained from the crossflow experiments are presented in Figure 4(a). The flux values that were measured for pre-treated ROC tested with 0.2 μm, 300 and 15 kDa membranes were, respectively, 345±45, 100±10, and 49±2 L/m²·h.

When evaluating 0.2 μm filtration studies, which have the highest flux for removal of precipitated divalent ions and organic matter, it was seen that the fluxes sharply decreased within 10 min. and then slowly lessened within the first 80 min. of filtration of the pre-treated ROC (Figure 6(b)). The major reason for this flux decline can be explained by the membrane pore size. In 300 kDa filtration studies 40% flux reduction was observed within 30 min. and total reduction was measured as 65%. For 15 kDa filtration studies, fast flux reduction was observed within 10 min. similar with the other membranes used in this study. This rapid flux decline can be explained by blocking of pores on membrane surfaces by divalent ions and organic matter.

Membrane fouling mechanisms have been examined with the resistance in-series model during ceramic MF and UF experiments. Generally, the membrane intrinsic resistance (Rm) of the membranes increases with decreasing pore size. With this knowledge, the Rm value of 0.2 μm membranes was found to be lower than 300 and 15 kDa. The Rt values of 0.2 μm, 300 kDa, and 15 kDa membranes during filtration were observed to be 7,32 E + 11, 3,93 E + 12, and 1,56 E + 13, respectively, and Rp values were increased with increasing pore size and found to be 5,07E+10, 5,64E+11, and 4,19E+12, respectively. The main fouling mechanism was physically removable membrane resistance according to the resistance in-series model for all MWCO of ceramic membranes Figure 4(b). Considering the results, the 0.2 μm value was found to be more appropriate with a high flux rate and lowest membrane resistance values among the three membranes tested for pre-treatment.

At neutral pH using 3 mM FeCl₃, the requisite amount of Ca(OH)₂ for achieving the targeted softening of pH and 200 mg/L MgO addition to the ROC, the 0.2 μm pore size of CMF was found as the best pre-treatment. The Ca²⁺, Mg²⁺, and SiO₂ removal performance by FeCl₃ was found to be 45, 23, and 30%, respectively, while 88, 71, and 92% removal was observed, respectively, with 200 mg/L MgO at pH 11.5. Adding MgO compound to ROC increased SiO₂ removal almost threefold (Figure 5). All pre-treatment methods were conducted using the CMF and were repeated at least duplicated.

A study was conducted with a UF membrane integrated pellet reactor that can remove scale-forming ions such as Ca²⁺, Mg²⁺, and Si from a pilot-scale textile industry ROC. The resulting effluent was further tested in a secondary RO process to reduce the amount of concentrate and increase overall water recovery. Unlike raw concentrate, after pellet treatment, more than 80% water recovery was provided without a dramatic flux decrease (Sahinkaya et al. 2018). In another study, resulting from the desalination unit, ROC was introduced to a ceramic UF membrane following chemical pre-treatment for enhancement total water recovery. A CUF membrane were used for increasing solid concentration and minimizing membrane concentrate (Sanciolo et al. 2012).

Technically, using ceramic MF and UF as pre-treatment steps before NF and RO is the optimum option for textile wastewater with high COD/BOD and TDS concentrations (Samaei et al. 2018).

NF membranes such as NF90 and NF270 have a tight polyamide separating layer (Zhou et al. 2015) and high rejections of polyvalent ions and low-to-moderate rejections of monovalent ions and small organic compounds (Lin et al. 2017). For salt removal compared to RO, NF90 might be the best option (Hilal et al. 2005; Zhou et al. 2015) as it ensures the key differentiating characters of NF are their higher flux, lower energy consumption, and longer membrane life (Mohammad et al. 2015). For these reasons, NF filtration tests were conducted with NF90 and NF270 membranes, and pre-treated samples were filtered through both membranes at least duplicated. A new membrane was used for each of the filtration stages and rinsed with pure water before the experiment and pre-treated ROC samples pH adjusted to 7–8 before the NF filtration experiments.

Raw ROC and pre-treated ROC concentrations after each treatment method are given in Table 3. Evaluating NF270 permeates, medium silica, Ca²⁺, and Mg²⁺ removal were observed. The NF concentrate showed a lack of Ca²⁺, the same average concentration of Mg²⁺ was found in the raw ROC, and silica values six times lower than the raw ROC concentration. While the NF90 permeate shows almost no organic pollutant and silica content, the low conductivity (0.74 ms/cm) and low salinity (0.38%) concentrations allow the permeate to be reused in the process. The NF90 concentrate COD, color, Ca²⁺, Mg²⁺, and silica concentration were measured as 389, 133, 4.05, 19.87, 31.12 mg/L, respectively, and, theoretically, recirculating NF concentrate is possible.

For MgO precipitation combined with NF270 filtration, the removal efficiencies were found to be: COD, color, Ca²⁺, Mg²⁺ and SiO₂; 92, 83, 97, 84 and 92%, respectively. Using MgO + NF90 tests, the removal efficiencies were COD, color, Ca²⁺, Mg²⁺ and SiO₂; 95, 85, 97, 81 and 98, respectively and lastly, for CP–CMF/NF90 tests, the removal efficiencies were COD, color, Ca²⁺, Mg²⁺ and silica, 98, 94, 97, 93, 99%, respectively (Figure 6(a)). When CP/CMF + NF270 tests were carried out, the removal efficiencies were COD, color, Ca²⁺, Mg²⁺ and SiO₂, 97, 89, 96, 87 and 98%, respectively (Figure 6(b)). The most efficient pre-treatment method was determined to be integrated FeCl₃ + MgO adding CMF. For further treatment, NF90 showed higher removal efficiency for COD, color, total hardness, and silica than NF270. Alongside decreased salinity and conductivity and increased permeate rates. NF90 membranes were found to be the best option for water recovery. These membranes were also examined for flux behavior in addition to their removal efficiencies, Figure 6 shows the steady-state flux values obtained after filtration tests for raw and pre-treated ROC for both two NF membranes.

The highest steady-state flux value for all membrane tests was obtained with the NF270, which has a large pore size. The lowest steady-state flux values for the two membranes were obtained after filtration tests with raw ROC. A significant increase in steady-state flux value for the NF270 membrane was obtained from the filtration tests after CP–CMF was used, while the steady-state flux of NF90 membrane increased approximately 1.5 times after the pre-treatment processes. While the fastest flux decline rate for the NF270 filtration test was measured with the pre-treated sample as 50% within the first 50 min., this rate was 70% with raw ROC. When the NF90 filtration tests were evaluated, within the first 50 min., 40% flux decline was observed for both pre-treated and raw ROC; after 150 min filtration, 50% flux decline was measured for the pre-treated ROC while the raw ROC flux decline was measured to be 75%.

In the NF90 filtration tests, no significant differences were found using FeCl₃ and MgO in terms of flux but their combination increased the flux 1.5 times. In NF270 filtration tests, with the addition of MgO, the scaling problem was eliminated at pH 11.5, and flux rate increased 1.5 times; for pre-treatment with FeCl₃ + MgO, this ratio was doubled (Figure 7).

A certain objective of the study was the utilization of ceramic membrane technology in order to achieve the standards recommended for reuse in textile processes. Addressing the quality of purified water for reuse purposes, it has to be noted that the specific requirements can be quite different according to a company's needs. Nevertheless, certain general criteria have been established that should be met by the treated water (Li & Zhao 1999; Rozzi et al. 1999). These are: COD <30 mg/l, a spectral absorption coefficient (SAC) at 426 nm <1.0 (m−1) and conductivity <1.8 mS/cm (Table 4).

The suggested pre-treatment process is a reasonable treatment option in combination with other advanced treatment technologies to achieve reuse standards. In particular, the permeate is suitable to be directly fed to the nanofiltration module. This two-stage process configuration has been tested and produces a water quality suitable for any textile process.

Cost estimation was carried based on optimum conditions determined in the proposed experimental studies. The plant was postulated to operate for 24 h per day and 365 days per year. The costs of the components were determined from market values and adapted from the available literature. For the cost analysis, the investment cost of ceramic membranes is significantly higher; the lifespan is on average of 15–20 years, approximately doubled their polymeric counterparts (Gitis & Rothenberg 2016). Based on the availability of real-scale data, the costs for CMF installation and operational costs were assumed as follows. Ceramic membrane costs are about 1,500–2,000 USD/m² and the module cost is 16–32 USD/m² (Dilaver et al. 2018). To treat 480 m³/d of water with 250 LMH, the required membrane area is 80 m² which accounts for 120,000 USD for the membranes and approximately 2,000 USD for the membrane module. In a recent study, polymeric NF and RO membrane cost was assumed to be 40 USD/m² (Panagopoulos 2021). Pannirselvam et.al. (2019) stated that the price of NF membranes was between 30–60 USD/m² with an average of 45/m² within their economic analysis studies for NF membrane costs (Pannirselvam et al. 2019). Vergili et al.(2012) estimated the NF membrane cost for integrated membrane scenarios as 30 USD/m² (Vergili et al. 2012). In our country market prices for NF membranes are about 50–100 USD/m². Thus far, there is no information on the cost of ROC treatment by coagulation. In this case, the costs were found for chemicals to be 0.26 USD/m³ (the total cost of used FeCl₃, MgO, and Ca(OH)₂ per m³). Disposal of chemical pre-treatment sludge cost was included in the total cost of CMF concentrate disposal adapted from Heijman & Bakker (2007). A summary of expenditures is given in Table 5. For CMF, the operational, specific, and maintenance costs were adapted from the TECHNEAU Report Heijman & Bakker (2007), a recent study conducted by Weschenfelder et al. (2016), Viegas et al. (2020), and NF costs adapted from Zhou et al. (2015). The costs of mixing and pumping energy requirements corresponding to 0.026 kWh/m³ for the CP-CMF process were included (Tompkins et al. 2019), in a recent work with ceramic MF membrane energy consumption was given as 0.4 kWh/m³ (Hakami et al. 2020), in our study it was calculated as 0.15 kWh. The energy consumption of the NF process is about 0.245 kWh/m³ by (Turek et al. 2009), in another study it was assumed to be 1.4 kWh/m³ for NF90 (Schäfer et al. 2007). In this study it is assumed to be 2.2 kWh. Major operating costs, the electricity, labor, and chemical consumptions costs are presented in the Supplementary information.

Annually, 175,200 m³ water will be saved through the proposed study. Both required wastewater treatment and fresh/process water consumptions will be reduced, and 87,600 USD are saved due to water recovery. The total net cost of the proposed study with the consideration of savings is about 0.47 USD/m³ (Table 5).

ROC management processes were evaluated; it was applied for treating textile RO concentrate using ED and evaporation alone and a combination of these processes. Only an evaporation process was used for ROC treatment and the cost of the method was found to be 3.88 USD/m³, while the ED and evaporation combination is approximately 0.55 USD (Praneeth et al. 2014). Using the ASP process it was 2.6 USD/m³ (Sanciolo et al. 2012), using RO–NF combination process cost was 0.57 USD/m³ (Chen & Chen 2004). A combination RO + ED + Crystallization method was conducted by Nayar et al. (2019) and the cost of the proposed method was 3.5 USD/m³ (Nayar et al. 2019). Şahinkaya et al. calculated the treatment cost of the pellet reactor as 0.83 USD for 1 m³ ROC (Sahinkaya et al. 2018). UV, ozonation, and evaporation processes were discovered to be the most expensive methods. Compared with the other methods, CP–CMF/NF may be one of the cost-effective processes for water recovery.

The pre-treatment method of this study is suitable in terms of removal efficiency and enhancement of flux rate. It was able to precipitate scaling cations and remove organic matters from textile industry ROC before the NF process. For further treatment, NF90 and NF270 membranes were used to reach desired water recovery values. The results show that NF90 was the effective NF membrane compared with NF270 in achieving desired removal efficiency and flux rate of almost 50%. Moreover, the NF concentrate analysis indicated that recirculation of the concentrate stream is technically possible. From an economic viewpoint, with the proposed method the return of the initial investment is expected within 2 years. As compared to the other methods, CP–CMF/NF may be a cost-effective approach and freshwater consumption can be reduced by 15% and the requirements for water reuse may be met. The CMF integrated CP process has the potential for pilot-/real-scale applications employing RO processes, municipal/industrial wastewater treatment, and reuse to achieve near-ZLD by CP–CMF/NF.

The authors would like to thank the TUBITAK-MRC for the laboratory infrastructure support offered during this study and Muhammad Yaqub for his insightful comments and suggestions.

All relevant data are included in the paper or its Supplementary Information.