Fabrication of low-cost ceramic microfiltration membranes with the addition of fly ash to kaolin-natural zeolite structure for greywater treatment

Abstract

In this study, low-cost tubular ceramic microfiltration membranes were fabricated by extrusion using kaolin, natural zeolite, fly ash, and α-alumina for laboratory-scale greywater treatment. Fly ash content (5–20 wt %) was evaluated to tune membrane permeability, porosity, and mechanical behavior while avoiding crack formation. Among the fabricated membranes, kaolin–natural zeolite–fly ash (K–Z–F, 75–10–15) and kaolin–α-alumina–fly ash (K–Al–F, 80–10–10) showed the most favorable balance between permeate flux, COD reduction, and mechanical stability. The optimized membranes had average pore sizes in the microfiltration range (1.06–1.14 µm) and porosities of 33.07–41.26%. During 60 min greywater microfiltration at 1 bar and 25 °C, the K–Z–F (75–10–15) membrane achieved a permeate flux of 632.29 ± 3.63 L m⁻² h⁻¹ with 86.69 ± 0.18% COD removal, while the K–Al–F (80–10–10) membrane achieved 91.50 ± 0.03% COD removal with a permeate flux of 559.80 ± 15.53 L m⁻² h⁻¹. The COD decrease is interpreted as the combined removal of suspended and colloidal organic matter, oil/fat droplets, adsorption on aluminosilicate surfaces, and dynamic cake-layer filtration, rather than size exclusion of dissolved COD alone. The treated permeate satisfied the investigated Iranian discharge/reuse limits for the measured parameters; however, the study was based on single-source kitchen greywater and short-term filtration. Direct laboratory raw-material-and-energy costs, calculated using the complete furnace program, were estimated as 2.22 $ m⁻² for K–Z–F (75–10–15) and 4.27 $ m⁻² for K–Al–F (80–10–10). These values should not be interpreted as full industrial production costs because labor, equipment depreciation, maintenance, membrane replacement, infrastructure, and scale-up losses require pilot-scale techno-economic assessment.

Introduction

With the increasing population and growing demand for water resources, the reuse of wastewater as a new water source has received a great deal of attention. Greywater is a good candidate as a new water source and a potential resource for sustainable water management¹. Wastewater from kitchen, laundry, and bath, except for toilet water, is called greywater². Due to the low level of pollution in greywater, particularly in terms of suspended solids and total coliforms, it can be easily collected and treated. Greywater has very different specifications and includes 50% to 80% of domestic wastewater volume³. So far, various methods have been investigated for treating greywater, each having its own specific advantages and disadvantages⁴. Traditional methods like demulsification, flocculation, and gravity separation have been used to remove COD from wastewater⁵. However, these methods often produce large amounts of sludge, secondary pollutants, or require excessive time and resources^6,7. Electrochemical techniques can also cause equipment damage and generate hazardous materials⁸. To address these issues, researchers have turned to membrane separation, a cleaner, more efficient, and cost-effective alternative that doesn’t involve chemical additives^9,10.

Ceramic membranes, which outperform polymeric ones in greywater treatment due to higher flux and lower fouling, face challenges from high costs, fragility, and energy-intensive production using materials like alumina and zirconia¹¹. To address these, researchers have developed cost-effective microfiltration membranes from clay, phosphate, and recycled waste, which require lower processing temperatures but suffer from performance inconsistencies due to impurities, prompting ongoing research for optimization¹².

Recent advancements in greywater and wastewater treatment have explored a range of methods to enhance efficiency, reduce costs, and improve scalability¹³. These efforts include biological systems, membrane technologies, and alternative approaches, each offering unique benefits but facing challenges that limit widespread adoption¹⁴. Biological treatment systems have demonstrated high effectiveness for greywater purification¹⁵. For example, a rotating biological contactor (RBC) system was evaluated, achieving 93–96% BOD removal and 84–95% TSS removal across various influent concentrations¹⁶. Similarly, Fountoulakis et al.¹⁷ investigated a submerged membrane bioreactor for on-site greywater treatment, reporting approximately 87% COD removal, suitable for applications like toilet flushing. While these systems excel in pollutant removal, their high operational costs and space requirements often restrict their use to specific contexts¹⁸. Ceramic membrane technologies have gained prominence due to their durability and performance¹⁹. Researchers have focused on low-cost materials to overcome the high expense of traditional membranes. Rashad et al.²⁰ developed a membrane using China clay and alumina, achieving over 96% rejection for oil-in-water emulsions with high permeability. Belgada et al.²¹ synthesized a phosphate-based membrane, removing 73% total organic carbon (TOC) and 98% turbidity, while El Machtani Idrissi et al.²² created a kaolinite-perlite membrane with a low cost of ~ 7.7 $/m² and high permeability, suggesting potential for industrial scaling. Satyannarayana et al.²³ used bentonite clay for a tubular membrane, but its higher cost (~ 96 $/m²) limits competitiveness. These membranes offer cost savings, yet impurities in raw materials can lead to inconsistent performance. Fly ash-based ceramic membranes have emerged as a sustainable alternative due to their low cost and water-affinity properties. Fan et al.²⁴ combined fly ash with bauxite, achieving high permeability and mechanical strength, though requiring high sintering temperatures. Huang et al.²⁵ studied fly ash particle size effects, finding that broader distributions reduce porosity and strength, while larger particles increase permeability but risk structural cracks²⁶. Despite their environmental and economic benefits, fly ash membranes often face challenges with performance consistency, necessitating further optimization²⁷. Membranes enhanced with adsorbents have shown exceptional efficiency in specific applications²⁸. Jafari et al.²⁹ incorporated activated carbon and natural zeolite into kaolin membranes for oily wastewater treatment, achieving up to 99.99% TOC removal through improved adsorption and hydrophilicity. While highly effective, this approach has been less explored for greywater, limiting its immediate applicability to domestic wastewater treatment. Alternative treatment methods have also been investigated for greywater. Patil et al.³⁰ compared Fenton and cavitation-assisted Fenton processes, with the hybrid method achieving 92.8% COD removal, though at a high cost (~ 248 $/m³). Mane et al.³¹ used rice husk ash for greywater, reporting 94% COD and 90% BOD removal. These techniques are effective for specific scenarios but often lack the scalability and cost-efficiency of membrane-based systems.

This study addresses a critical gap in the development of cost-effective ceramic membranes for decentralized wastewater treatment through a systematic, laboratory-scale compositional screening of fly-ash-modified kaolin tubular membranes applied to real kitchen greywater. Rather than asserting broad novelty for low-cost ceramic membranes, the work delivers a focused, comparative evaluation of two promising formulations—kaolin–natural zeolite–fly ash and kaolin–α-alumina–fly ash—prepared and rigorously tested under identical conditions. The primary objectives are to elucidate the distinct roles of natural zeolite and α-alumina in governing membrane porosity, pore size distribution, surface roughness, flexural mechanical behavior, fouling morphology, greywater permeate flux, and chemical oxygen demand (COD) reduction. Furthermore, a transparent preliminary cost basis is presented, with clearly articulated limitations regarding its scope. This deliberate framing positions the study as a reproducible, laboratory-scale assessment of two optimized membrane families, explicitly distinguishing it from full field-scale validation or comprehensive industrial techno-economic analysis.

Materials and methods

Materials

Kaolin was sourced from the Zenouz mine, Marand County, East Azerbaijan, Iran (Iranian China Clay Company). The natural zeolite used in this study, obtained from the Semnan mine in Iran, had an average particle size of 15 μm (600 mesh), as measured by a Cilas 1064 laser particle size analyzer (LPSA). Its average pore size was 26 nm, and the specific surface area was 26.46 m²/g, as determined by BET analysis. Alpha alumina, with a median particle size of 75 µm and specific surface area of 5–10 m²/g, was procured from the Iranian Nano Materials Company, Mashhad, Khorasan Razavi, Iran. Fly ash, with a median particle size of 17.57 µm and specific surface area of 0.51 m²/g, was supplied by the Tehran Concrete Company, Tehran, Iran. The chemical compositions of these materials, determined by X-ray fluorescence (XRF), are presented in Table 1. All materials were used as received without further purification.

Table 1.

Chemical analysis of kaolin, zeolite, alpha alumina, and fly ash.

Composition	Kaolin (wt %)	Zeolite (wt %)	Alpha alumina (wt %)	Fly ash (wt %)
Al2O3	24.00	11.412	99.000	28.239
SiO2	63.00	70.937	0.012	62.860
Fe2O3	0.55	1.479	0.012	3.774
TiO2	0.04	–	–	1.886
CaO	1.20	3.251	0.012	0.787
MgO	0.40	1.000	–	0.820
Na2O	0.30	2.177	0.430	–
K2O	0.40	0.907	–	1.128
P2O5	–	–	–	0.508
L.O.I	9.00	8.200	0.500	1.200
Moist	–	–	0.300	–

Greywater collection and characterization

The greywater was collected from a single residential kitchen sink in southern Iran. Samples were gathered in sterilized glass bottles, transported to the laboratory within 1 h, and stored at 4 °C until analysis (performed within 2 h). Physicochemical characteristics were analyzed according to standard methods (APHA, 2017), and the results are summarized in Table 2. COD was measured in triplicate and is reported as mean ± SD because it was used as the main performance indicator in the filtration experiments. The remaining parameters listed in Table 2 were determined once and are reported as single measurements for initial feed characterization. These values characterize the specific kitchen greywater sample used in the experiments and should not be interpreted as representing the temporal, seasonal, or household-to-household variability of domestic greywater. This sampling limitation is explicitly considered in the interpretation of the results and in the future-scope section. The greywater exhibited a COD of 718.69 ± 8.4 mg L⁻¹, BOD of 332 mg L⁻¹, and turbidity of 37.60 NTU, indicating moderate organic pollution.

Table 2.

Physicochemical characteristics of greywater used in this study.

Parameter	Value
COD (mg L⁻1)	718.69 ± 8.4
BOD (mg L⁻1)	332
pH	7.43
Electrical conductivity (µS cm⁻1)	2260
Chloride (mg L⁻1)	815
Detergent (mg L⁻1)	1.78
TSS (mg L⁻1)	179
Sulfate (mg L⁻1)	113
Nitrate (mg L⁻1)	3.40
Phosphate (mg L⁻1)	16.90
Turbidity (NTU)	37.60

COD is reported as mean ± SD (n = 3); all other parameters are single measurements.

Membrane fabrication

Tubular ceramic microfiltration membranes were fabricated using kaolin, natural zeolite, alpha alumina, and fly ash, with compositions detailed in Table 3. Fly ash content was varied (5, 10, 15, and 20 wt %) to optimize performance, while higher fly ash content (> 20 wt.%) caused cracking. The crack formation observed at fly ash contents above 20 wt.% mainly occurred during the air-drying stage of the extruded green membranes. This behavior is attributed to the high proportion of fine particles and glassy phases in fly ash, which increase water demand and drying shrinkage. Rapid moisture loss and shrinkage mismatch between fly ash-rich regions and the surrounding matrix generate internal stresses that exceed the mechanical strength of the green body, leading to crack development before sintering. Corn starch (8 wt.%) was used as a temporary binder and pore-forming agent and was completely removed during the sintering process; therefore, it is not included in the membrane composition notation, which refers only to the inorganic ceramic constituents.

Table 3.

Batch compositions of the fabricated membranes (wt.% of inorganic ceramic powders); corn starch was added separately at 8 wt.% relative to the ceramic powder mass.

Membrane	Kaolin	Zeolite	Alpha alumina	Fly ash	Corn starch (wt.% of ceramic powder mass)
K (100)	100	–	–	–	8
K-Z (90-10)	90	10	–	–	8
K-Z-F (85-10-5)	85	10	–	5	8
K-Z-F (80-10-10)	80	10	–	10	8
K-Z-F (75-10-15)	75	10	–	15	8
K-Z-F (70-10-20)	70	10	–	20	8
K-Al (90-10)	90	–	10	–	8
K-Al-F (85-10-5)	85	–	10	5	8
K-Al-F (80-10-10)	80	–	10	10	8
K-Al-F (75-10-15)	75	–	10	15	8
K-Al-F (70-10-20)	70	–	10	20	8

The raw materials were mixed in the specified proportions, homogenized using a magnetic stirrer (SH-2, 700 rpm, 15 min), and combined with distilled water (40 wt.% of mixture) to form a paste. The paste was extruded using a laboratory-scale extruder, recycled once for uniformity, and shaped into tubular membranes (length: 23–25 cm, inner diameter: 10 mm, outer diameter: 14 mm). A schematic representation of ceramic membrane preparation is shown in Fig. 1. membranes were air-dried at 25°C for 48 h, then sintered in an electric furnace (Nabertherm RS 80/300/13, Germany) following the temperature profile in Table 4.

Fig. 1.

A schematic representation of ceramic membrane preparation.

Table 4.

Sintering profile for ceramic membranes.

Temperature range (°C)	Rate (°C/min)	Hold time (min)
25–550	5	–
550	–	60
550–975	5	–
975	–	60
975–1250	3	–
1250	–	180

In a complementary study utilizing the same class of low-cost ceramic membranes developed here, specifically the kaolin–alumina–fly ash (K–A–F) formulation, we investigated their performance in the treatment of aquaculture wastewater³². The membranes demonstrated consistent structural integrity, high permeability, and effective separation of suspended solids and organic pollutants, achieving COD removal efficiencies exceeding 85%. Those results align with and further substantiate the findings presented in this work, particularly regarding the role of fly ash in enhancing porosity, improving hydrophilicity, and sustaining flux under fouling conditions. This cross-validation underscores the robustness, versatility, and scalability of the developed membrane materials for cost-effective water and wastewater treatment applications beyond greywater.

Experimental setup

Microfiltration experiments were conducted using a cross-flow filtration system comprising a feed tank, a Pentax pump (Italy), a Teflon tubular membrane module (length: 25 cm, inner diameter: 10 mm, outer diameter: 14 mm), a permeate collection vessel, and a digital scale. Greywater was pumped at a constant pressure of 1 bar and temperature of 25 °C, with a recirculation loop to enhance mixing and regulate flow with a valve to control outlet pressure. The schematic of the setup is shown in Fig. 2. Each greywater microfiltration experiment was independently repeated three times under identical operating conditions (n = 3). For each run, permeate mass was continuously recorded using a digital balance at fixed time intervals, and permeate flux values at each time point were calculated. The reported PF values represent the mean of the three independent runs, and the error bars indicate the corresponding standard deviation. The 60 min filtration tests were designed for comparative membrane screening under controlled laboratory conditions; they do not establish long-term stability, cleanability, regeneration behavior, or irreversible fouling resistance.

Fig. 2.

A schematic representation of the experimental setup.

Analytical methods

Water quality parameters (COD, BOD, nitrate, sulfate, phosphate, detergent, TSS, turbidity) were measured before and after filtration. Filtration ran for 60 min, with permeate samples collected during the final 10 min. COD and BOD were quantified using a Hach HQ40d and DRB200, respectively. Nitrate, sulfate, phosphate, and detergent concentrations were analyzed with a Hach DR 3900 UV–Vis spectrophotometer. The detergent concentration was measured as anionic surfactants using the methylene blue active substances (MBAS) colorimetric method based on standard procedures for water and wastewater analysis. COD removal efficiency was calculated using Eq. 1³⁰:

1

where C_f and C_i denote the final and initial COD concentrations, respectively.

Characterization of the fabricated membranes

Membrane morphology was examined using field emission scanning electron microscopy (FESEM, Tescan-Vega 3, Czech Republic) and atomic force microscopy (AFM, Brisk AFM, Ara Research, Iran) for surface and cross-sectional analysis. X-ray diffraction (XRD) patterns were recorded using a D8 Advance diffractometer (Bruker, Bremen, Germany) equipped with Cu Kα radiation (λ = 1.5406 Å) over a 2θ range of 10–80° with a step size of 0.07° and a counting time of 1 s per step. Thermal stability was assessed via thermogravimetric analysis (TGA, Q600, USA) from 30 to 800°C at 10°C/min under atmospheric pressure.

The porosity of each membrane was calculated using the following equations^33,34:

2

3

In Eq. 2, \varepsilon represents the percentage of porosity, v_s represents the volume of empty space in the membrane in cm³, v_t represents the total volume of the membrane in cm³. In Eq. 3, w₁ represents the weight of the wet membrane in grams, w₂ represents the weight of the dry membrane in grams, and _w represents the water mass density in g/cm³.

To calculate the average pore size, first, the pure water flux (PWF) of the membranes was measured at transmembrane pressures between 1 and 4 bar, along with the membrane porosity. The average pore size was then calculated using the Guerout–Elford–Ferry equation^35,36.

4

In Eq. 4, r_m represents the average pore size in µm, \varepsilon represents the percentage of porosity, µ represents the viscosity of water at the operating temperature, L represents the thickness of the membrane in meters, j represents the pure water flux (PWF) in L/m².h, and TMP represents the trans-membrane pressure in kPa.

For comparison, the average pore size was also estimated using a Hagen–Poiseuille-based capillary bundle model, which relates permeate flux to pore radius assuming straight cylindrical pores²⁰:

5

where r_m is the average pore size in µm, J is the pure water flux (PWF) in L m⁻² h⁻¹, ε is the porosity, TMP is the trans-membrane pressure in kPa., μ is the viscosity of water (Pa.s), L is the thickness of the membrane (m), and τ is the tortuosity value. In the absence of an independently measured tortuosity, τ = 1 was assumed to provide a first-order estimate of r_m. It should be noted that the Hagen–Poiseuille-based model assumes straight cylindrical pores and does not explicitly account for pore tortuosity, whereas the GEF equation indirectly incorporates tortuosity effects through permeability and porosity terms.

The flexural behavior of the fabricated membranes was characterized by a three-point bending test using a universal testing machine (Santam STM-20, Iran) in accordance with ASTM D790. The tests were carried out at 25 ± 2 °C with a crosshead speed of 0.5 mm min⁻¹. The flexural stress (σ_f) and flexural strain (ε) were calculated using Eqs. (6 and 7), respectively³⁷.

6

7

where σ_f is the flexural stress (MPa), F is the applied load at the fracture point (N), L is the support span (mm), d_out and d_in are the outer and inner diameters of the membrane (mm), D is the mid-span deflection (mm), and ε is the flexural strain (%).

Flexural properties, porosity, pore size, pure water flux, greywater permeate flux, and COD analyses were conducted in triplicate (n = 3), as described in the corresponding experimental sections. Other water quality parameters were determined once for influent and effluent samples.

Results and discussion

Characterization of the prepared ceramic membranes

Field emission scanning electron microscope (FESEM)

In this section, detailed microstructural and surface characterizations (FESEM, XRD, and AFM) are first presented for the optimized membranes to provide insight into their morphology and phase composition. Subsequently, porosity, average pore size, and permeation performance are discussed for all prepared membranes to enable a systematic comparison of the effects of composition and fly ash content. FESEM images of the top surfaces and cross-sections of the optimized kaolin-alpha alumina-fly ash (K-Al-F, 80-10-10) and kaolin-natural zeolite-fly ash (K-Z-F, 75-10-15) membranes are presented in Fig. 3. Samples were gold-coated and imaged at 5 kV. The top-surface and cross-sectional FESEM images shown in Fig. 3a–d were acquired at a magnification of 1.50 k× , whereas the additional cross-sectional images in Fig. 3c′, d′ were acquired at 60 × to provide a lower-magnification view of the membrane structure.

Fig. 3.

FESEM images of the membrane top surfaces before microfiltration: (a) K-Al-F (80-10-10) and (b) K-Z-F (75-10-15). Cross-sectional FESEM images before microfiltration: (c, c′) K-Al-F (80-10-10) and (d, d′) K-Z-F (75-10-15).

Figure 3a, b show the top surfaces of K-Al-F (80-10-10) and K-Z-F (75-10-15) membranes, respectively, revealing crack-free, porous structures with a relatively homogeneous microstructure.

The K-Al-F (80-10-10) membrane exhibits larger particle aggregates (up to 10 µm) compared to K-Z-F (75-10-15) (2–5 µm), attributed to the larger median particle size of α-alumina (75 µm) compared to the natural zeolite particle size (~ 15 µm). This difference contributes to the higher porosity of K-Al-F (80-10-10) (41.26 ± 0.21%) compared to K-Z-F (75-10-15) (33.07 ± 0.25%), as quantified in Section “Porosity and average pore size”. The incorporation of fly ash contributes to the overall structural integrity of the membranes, consistent with findings by El Machtani Idrissi et al.²², who reported improved homogeneity in fly ash-based ceramic membranes.

The cross-sectional images shown in Fig. 3c, c′ for K-Al-F (80-10-10) and Fig. 3d, d′ for K-Z-F (75-10-15) reveal a well-integrated, continuous ceramic matrix. The K-Z-F (75-10-15) membrane displays a denser, more compact layered structure compared to the more open, porous cross-section of K-Al-F (80-10-10). This denser packing in K-Z-F (75-10-15) correlates with its higher flexural strength (30.87 ± 0.35 MPa vs. 9.69 ± 0.27 MPa for K-Al-F, Section “Flexural behavior analysis”), likely due to enhanced particle interlocking facilitated by the smaller zeolite particles. The observed morphology indicates that the K-Al-F (80-10-10) membrane possesses larger pores and higher porosity, whereas the denser structure of K-Z-F (75-10-15) contributes to enhanced mechanical stability. However, permeation performance is governed not only by pore size but also by pore connectivity and structural compactness, as discussed in Section “Pure water flux (PWF) through ceramic membranes”.

FESEM images of the fouled kaolin -alpha alumina-fly ash (K-Al-F, 80-10-10) and kaolin -natural zeolite-fly ash (K-Z-F, 75-10-15) membranes after 60 min of greywater microfiltration are shown in Fig. 4. Samples were gold-coated and imaged at 5 kV. The top-surface and cross-sectional FESEM images shown in Fig. 4a–d were acquired at a magnification of 1.50 k × , whereas the additional cross-sectional images in Fig. 4c′, d′ were acquired at 60 × to provide a lower-magnification view for assessing fouling-related structural features. Figure 4a, b for K-Al-F (80-10-10) and K-Z-F (75-10-15), respectively, reveal significant fouling, with oil and fat deposits clogging pores on both membranes. The K-Z-F (75-10-15) membrane exhibits more extensive foulant accumulation compared to K-Al-F (80-10-10), likely due to its higher surface roughness (8.43 nm vs. 4.61 nm). Rougher surfaces enhance foulant adhesion by increasing surface area and trapping organic particles, as noted by Jafari et al.³⁵. This is consistent with the K-Z-F (75-10-15) membrane’s denser structure and smaller average pore size (1.06 ± 0.02 µm vs. 1.14 ± 0.02 µm, Section “Porosity and average pore size”), which may promote pore blocking over cake layer formation.

Fig. 4.

FESEM images of the top surfaces of fouled membranes after microfiltration: (a) K-Al-F (80-10-10) and (b) K-Z-F (75-10-15). Cross-sectional FESEM images after microfiltration: (c, c′) K-Al-F (80-10-10) and (d, d′) K-Z-F (75-10-15).

Figure 4c, c′ for K-Al-F (80-10-10), as well as Figs. 4d, d′ for K-Z-F (75-10-15), indicate that fouling occurs predominantly on the membrane surface, with minimal penetration into the internal structure.

Under the applied imaging conditions, a clearly distinguishable separate fouling layer is not observed. In contrast, the K-Z-F (75-10-15) membrane’s compact cross-section suggests greater foulant retention, potentially impacting long-term flux stability (Section “Permeate flux of greywater through ceramic membranes”). These findings highlight a trade-off: K-Al-F (80-10-10) offers better fouling resistance, while K-Z-F (75-10-15) achieves higher mechanical strength, critical for practical applications. Similar observations regarding surface-associated fouling and the influence of surface roughness and pore structure on foulant deposition have also been reported for ceramic microfiltration membranes treating oily and greywater streams, as noted by Jafari et al.²⁹, Twibi et al.³⁷, and Jafari et al.³⁸.

Atomic force microscopy (AFM)

AFM images of the kaolin -alpha alumina-fly ash (K-Al-F, 80-10-10) and kaolin -natural zeolite-fly ash (K-Z-F, 75-10-15) membrane surfaces, obtained in tapping mode over a 5 × 5 µm² scan area, are shown in Fig. 5. Before microfiltration, the K-Al-F (80-10-10) membrane exhibits a root-mean-square (RMS) roughness of 4.61 nm (Fig. 5a), while the K-Z-F (75-10-15) membrane has a higher roughness of 8.43 nm (Fig. 5b). The elevated roughness of K-Z-F (75-10-15) may be related to the incorporation of natural zeolite particles (~ 15 µm), which influence surface texture development during sintering. In contrast, the larger α-alumina particles (75 µm median size) tend to promote comparatively smoother surface features. Natural zeolite also exhibits a nanoporous structure (mean pore diameter ~ 15.68 nm), which is associated with its adsorption characteristics rather than direct particle packing effects.

Fig. 5.

AFM images of the surface of membranes before the microfiltration operation: (a) K-Al-F (80-10-10) membrane, (b) K-Z-F (75-10-15) membrane. AFM images of the inner surface of membranes after the microfiltration operation: (c) K-Al-F (80-10-10) membrane, (d) K-Z-F (75-10-15) membrane.

This increased roughness enhances foulant adhesion, as rough surfaces provide more sites for oil and fat deposition, consistent with observations by Jafari et al.²⁹ for zeolite-based membranes in oily wastewater treatment.

Post-microfiltration AFM images (Fig. 5c, d) reveal reduced roughness values of 4.09 nm for K-Al-F (80-10-10) and 4.62 nm for K-Z-F (75-10-15), respectively. This corresponds to a surface roughness reduction of approximately 11.3% for the K-Al-F (80-10-10) membrane and 45.2% for the K-Z-F (75-10-15) membrane after microfiltration. The significant decrease in K-Z-F (75-10-15) roughness (from 8.43 to 4.62 nm) indicates greater foulant accumulation, likely due to its denser structure and smaller pore size (1.06 ± 0.02 µm vs. 1.14 ± 0.02 µm), which promote pore blocking and surface smoothing by organic deposits. In contrast, the minimal roughness change in K-Al-F (80-10-10) suggests lower fouling propensity, attributed to its higher porosity (41.26 ± 0.21% vs. 33.07 ± 0.25%) and smoother initial surface, which reduce foulant entrapment. These findings underscore the trade-off between K-Z-F (75-10-15)’s higher flexural strength (30.87 ± 0.35 MPa) and K-Al-F (80-10-10)’s superior fouling resistance, critical for optimizing greywater treatment efficiency. Previous studies have demonstrated that increased surface roughness enhances foulant adhesion and accelerates membrane fouling, whereas smoother ceramic membrane surfaces exhibit improved fouling resistance during wastewater and oily water filtration, as noted by Jafari et al.^29,35 and Twibi et al.³⁷.

X-ray diffraction (XRD)

XRD analysis was performed to identify the crystalline phases present in the sintered kaolin–alpha alumina–fly ash (K–Al–F, 80–10–10) and kaolin–natural zeolite–fly ash (K–Z–F, 75–10–15) membranes. The corresponding XRD patterns are presented in Fig. 6.

Fig. 6.

XRD patterns of (a) raw mixed materials and (b) sintered tubular ceramic membranes prepared with K–Al–F (80–10–10) and K–Z–F (75–10–15) compositions.

The XRD results provide qualitative phase identification of the sintered membranes. Kaolin-derived phases transformed during sintering at 1250 °C, leading to the presence of mullite together with silica-based phases such as quartz and cristobalite. For the K–Al–F (80–10–10) membrane, mullite was observed together with residual quartz and corundum originating from the added α-alumina, while the K–Z–F (75–10–15) membrane showed stronger contributions from silica-rich phases, particularly cristobalite and quartz, consistent with the high SiO₂ content of natural zeolite and fly ash. Because no Rietveld refinement or quantitative phase analysis was performed, the discussion is limited to qualitative identification of the main crystalline phases and should not be interpreted as a quantitative phase balance. The phase assemblage helps explain differences in microstructure and flexural behavior, but it does not by itself prove long-term chemical or operational stability in real greywater.

Thermogravimetric analysis (TGA)

TGA was performed to evaluate the thermal stability, moisture content, and decomposition behavior of the kaolin -alpha alumina-fly ash (K-Al-F, 80-10-10) and kaolin -natural zeolite-fly ash (K-Z-F, 75-10-15) membranes. Results are presented in Fig. 7.

Fig. 7.

TGA curve for K-Al-F (80-10-10) membrane and K-Z-F (75-10-15) membrane.

The TGA curves show a minor initial weight loss below 200 °C for both membranes, attributed to the evaporation of physically adsorbed moisture. A subsequent mass-loss region between approximately 200 and 650 °C is associated with dehydroxylation of residual kaolinite-related phases and metakaolin formation, as reported in previous studies^38,39. The K–Z–F (75–10–15) membrane exhibits a higher total weight loss of 1.85% compared with 1.43% for the K–Al–F (80–10–10) membrane, likely because natural zeolite has higher moisture retention and loss-on-ignition than α-alumina.

As shown in Fig. 7, the low weight loss observed in TGA indicates limited residual mass-loss events within the tested temperature range. However, TGA up to 800 °C is a thermal characterization method and should not be interpreted as proof of long-term operational or chemical stability in real greywater. Although sintering was carried out at 1250 °C, the TGA measurement was limited to 800 °C because the main mass-loss processes in kaolin-based systems, including moisture removal and dehydroxylation, are commonly completed below this range. Higher-temperature sintering mainly involves solid-state reactions and densification rather than large additional mass loss. Dedicated acid/base resistance, leaching, cleaning-cycle, and long-term filtration tests are required before practical chemical durability can be claimed.

Flexural behavior analysis

The flexural behavior of the fabricated membranes was evaluated by a three-point bending test, and the corresponding flexural stress–strain curves are presented in Fig. 8. The flexural strength values are reported as mean ± SD (n = 3). The K-Z-F (75-10-15) membrane exhibited a significantly higher flexural strength (30.87 ± 0.35 MPa) than the K-Al-F (80–10-10) membrane (9.69 ± 0.27 MPa). This difference is mainly attributed to the lower porosity of K-Z-F (75–10-15) (33.07 ± 0.25%) compared with K-Al-F (80–10-10) (41.26 ± 0.21%), as well as its smaller average pore size (1.06 ± 0.02 µm vs. 1.14 ± 0.02 µm). Lower porosity and finer pores promote stronger particle interlocking and reduce stress concentration. Moreover, comparison with similar low-cost MF ceramic membranes reported in the literature indicates that the flexural strength of the K-Z-F (75–10-15) membrane (30.87 ± 0.35 MPa) exceeds the mechanical strength value of 21.00 MPa reported for kaolinite/sludge membranes, whereas the K-Al-F (80-10-10) membrane (9.69 ± 0.27 MPa) remains below this value³⁹. In addition, the presence of cristobalite and other silica-rich phases in K-Z-F (75-10-15) likely contributed to the formation of a denser matrix, thereby improving its flexural performance relative to the kaolin-dominated K-Al-F (80-10-10) membrane. As shown in Fig. 8, both membranes exhibited a typical brittle response, with an approximately linear increase in flexural stress up to fracture followed by a sharp drop. The K-Z-F (75-10-15) membrane fractured at a higher stress and strain than the K-Al-F (80-10-10) membrane, confirming its superior resistance to bending-induced failure.

Fig. 8.

Flexural stress–strain curves of the fabricated ceramic membranes obtained from three-point bending tests.

Porosity and average pore size

Porosity and average pore size were measured to assess the impact of fly ash addition on the microstructure of kaolin-based membranes. Porosity was determined using the water absorption method (Eqs. 2 and 3), and pore size was calculated via the Guerout-Elford-Ferry equation (Eq. 4). Results for membranes with varying compositions are presented in Table 5.

Table 5.

Average pore size, calculated using the GEF equation, and porosity of the fabricated membranes.

Membrane	Porosity (%)	Average pore size (rm) (µm)
K-100	32.97 ± 0.21	0.47 ± 0.003
K-Z (90-10)	37.14 ± 0.25	0.54 ± 0.008
K-Z-F (85-10-5)	34.87 ± 0.25	0.805 ± 0.02
K-Z-F (80-10-10)	34.5 ± 0.2	0.98 ± 0.01
K-Z-F (75-10-15)	33.07 ± 0.25	1.06 ± 0.02
K-Z-F (70-10-20)	39.41 ± 0.27	0.96 ± 0.008
K-Al (90-10)	35.34 ± 0.21	1.035 ± 0.02
K-Al-F (85-10-5)	40.36 ± 0.32	0.966 ± 0.01
K-Al-F (80-10-10)	41.26 ± 0.21	1.14 ± 0.02
K-Al-F (75-10-15)	42.27 ± 0.25	1.13 ± 0.03
K-Al-F (70-10-20)	40.42 ± 0.3	1.15 ± 0.02

Porosity values for the membranes are as follows: kaolin (K, 32.97 ± 0.21%), kaolin -natural zeolite (K-Z, 90-10, 37.14 ± 0.25%), kaolin -natural zeolite-fly ash (K-Z-F, 85-10-5, 34.87 ± 0.25%; 80-10-10, 34.50 ± 0.2%; 75-10-15, 33.07 ± 0.25%; 70-10-20, 39.41 ± 0.27%), kaolin -alpha alumina (K-Al, 90-10, 35.34 ± 0.21%), and kaolin -alpha alumina-fly ash (K-Al-F, 85-10-5, 40.36 ± 0.32%; 80-10-10, 41.26 ± 0.21%; 75-10-15, 42.27 ± 0.25%; 70-10-20, 40.42 ± 0.3%). The K-Al-F (75-10-15) membrane exhibits the highest porosity (42.27 ± 0.25%), while K-Z-F (75-10-15) has the lowest among optimized compositions (33.07 ± 0.25%). Average pore sizes for K-Z-F (75-10-15) and K-Al-F (80-10-10) are 1.06 ± 0.02 µm and 1.14 ± 0.02 µm, respectively. For comparison purposes, the average pore size of the two optimized membranes, K-Z-F (75-10-15) and K-Al-F (80-10-10), was also estimated using a Hagen-Poiseuille-based capillary model. For the K-Z-F (75-10-15) membrane, the pore size calculated using the Hagen-Poiseuille approach (0.70 ± 0.01 µm) is smaller than that obtained from the GEF equation (1.06 ± 0.02 µm). Similarly, for the K-Al-F (80-10-10) membrane, the Hagen–Poiseuille-based pore size (0.77 ± 0.01 µm) is lower than the corresponding GEF value (1.14 ± 0.02 µm). These differences, on the order of 0.3–0.4 µm, are attributed to the simplifying assumptions of the Hagen–Poiseuille model, which assumes straight cylindrical pores and does not explicitly account for pore tortuosity and pore-shape complexity in real ceramic membranes.

The addition of fly ash (up to 20 wt.%) increases porosity in both K-Z-F and K-Al-F membranes, with a more pronounced effect in K-Al-F compositions. For K-Al-F, porosity peaks at 15% fly ash (42.27 ± 0.25%) before slightly decreasing at 20% (40.42 ± 0.3%), likely due to particle agglomeration. The higher porosity observed in K-Al-F membranes may be related to the larger particle size of α-alumina (75 µm) compared to natural zeolite (~ 15 µm), which can influence particle packing and the development of interparticle voids during sintering.

Alpha alumina’s thermal stability at this temperature minimizes volume changes, preserving pore structure, as noted by Jafari et al.³⁵. In contrast, natural zeolite undergoes phase transformations (such as cristobalite formation) below 1250 °C, resulting in a denser matrix and lower porosity in K-Z-F membranes. The influence of fly ash content, α-alumina, and natural zeolite on pore size may be associated with variations in their average particle sizes and thermal behavior during sintering. Increasing fly ash content generally promotes pore development due to the presence of fine particles and the generation of interparticle voids; however, excessive fly ash addition (≥ 20 wt.%) may lead to particle agglomeration and partial pore collapse, resulting in non-uniform pore structures.

Alpha alumina, owing to its relatively large particle size and high thermal stability, helps maintain interparticle spacing during high-temperature sintering, leading to the formation of larger and more uniform pores. As a result, K-Al-F membranes exhibit higher average pore sizes compared to zeolite-containing membranes at similar fly ash contents.

In contrast, natural zeolite possesses a finer microstructure and undergoes phase transformations during sintering, which promote densification and pore shrinkage. Consequently, K-Z-F membranes generally show smaller average pore sizes despite comparable compositions. Similar trends regarding the influence of alumina, zeolite, and fly ash additives on pore size evolution in ceramic membranes have been reported in the literature^29,33,34. The lower porosity and smaller pore size of the K-Z-F (75-10-15) membrane enhance its flexural strength (30.87 ± 0.35 MPa) but also increase its fouling propensity. In contrast, the higher porosity and larger pores of the K-Al-F (80-10-10) membrane promote higher permeate flux at the expense of reduced flexural strength (9.69 ± 0.27 MPa). These findings highlight fly ash’s role in tailoring membrane microstructure for specific greywater treatment applications.

Table 5 presents the average pore size and porosity of the fabricated membranes. As can be clearly observed, the average pore size of the membranes varies between 0.47 ± 0.003 to 1.15 ± 0.02 µm and can be considered as a microfiltration membrane.

Pure water flux (PWF) through ceramic membranes

Pure water flux (PWF) through kaolin–natural zeolite–fly ash (K-Z-F) and kaolin–alpha alumina–fly ash (K-Al-F) membranes was measured in a cross-flow microfiltration system (Section “Experimental setup”) at transmembrane pressures of 1–4 bar and 25 °C. The results, plotted as PWF versus transmembrane pressure, are shown in Fig. 9. All PWF measurements were performed in triplicate (n = 3) and are reported as mean ± SD.

Fig. 9.

Pure water flux (PWF) as a function of transmembrane pressure for (a) K-Z-F membranes and (b) K-Al-F membranes. Symbols represent experimental data (mean ± SD, n = 3), and solid lines indicate linear fits.

Figure 9a, b present the PWF behavior of K-Z-F and K-Al-F membranes, respectively, as a function of applied pressure for different fly ash contents (0–20 wt.%). For all membranes, PWF increases linearly with pressure, indicating Darcy-type transport through the porous structure. Among the K-Z-F membranes (Fig. 9a), the highest PWF of approximately 8620.69 ± 344.83 L m⁻² h⁻¹ at 1 bar was obtained for the K-Z-F (75-10-15) composition, which can be attributed to its optimized porosity (33.07 ± 0.25%) and average pore size (1.06 ± 0.02 µm). Similarly, for K-Al-F membranes (Fig. 9b), the maximum PWF of approximately 6182.437 ± 133.39 L m⁻² h⁻¹ at 1 bar was observed for the K-Al-F (80-10-10) membrane, consistent with its higher porosity (41.26 ± 0.21%) and larger pore size (1.14 ± 0.02 µm). To further clarify this observation, although the K-Al-F (80-10-10) membrane exhibits higher porosity and slightly larger average pore size, the K-Z-F (75-10-15) membrane shows higher pure water flux at 1 bar. This indicates that permeation behavior is influenced not only by pore size and porosity but also by microstructural features such as pore connectivity, structural compactness, and surface characteristics. FESEM cross-sectional images suggest a more compact and well-integrated matrix for the K-Z-F membrane, which may promote more continuous water pathways and reduce hydraulic resistance. In addition, AFM analysis revealed higher surface roughness for K-Z-F, which can enhance wettability and facilitate water transport. Therefore, the higher flux of K-Z-F is attributed to its optimized pore connectivity and microstructural organization rather than pore size alone. The linear regression of PWF versus pressure yielded high coefficients of determination (R² > 0.98) for all membranes, confirming stable permeation behavior and allowing membrane permeability to be quantified from the slope of each fitted line. The presence of fly ash, rich in SiO₂ and Al₂O₃, enhances membrane hydrophilicity and water transport, as also reported for fly ash–based ceramic membranes in the literature³⁴. At higher fly ash contents (20 wt.%), a slight reduction in PWF is observed for both membrane systems, which is attributed to particle agglomeration and partial pore blockage, in agreement with the porosity trends discussed in Section “Porosity and average pore size”.

The membrane water permeability values were calculated from the slope of the linear fits of PWF versus transmembrane pressure, and the corresponding regression coefficients (R²) are summarized in Table 6.

Table 6.

Water permeability and linear regression parameters of ceramic membranes.

Membrane	Permeability (L m⁻2 h⁻1 bar⁻1)	R2
K-100	983	0.9931
K-Z (90–10)	1696.2	0.9865
K-Z-F (85-10-5)	3139	0.9912
K-Z-F (80-10-10)	4552.9	0.9893
K-Z-F (75-10-15)	5731.8	0.9831
K-Z-F (70-10-20)	5482.1	0.9866
K-Al (90-10)	4013.6	0.9995
K-Al-F (85-10-5)	4474	0.9993
K-Al-F (80-10-10)	6401.8	0.9998
K-Al-F (75-10-15)	5904.9	0.9989
K-Al-F (70-10-20)	6070.5	0.9992

Permeate flux of greywater through ceramic membranes

Permeate flux values are presented as mean ± SD based on three independent filtration runs, as described in Section “Experimental setup”. PF of the greywater (COD: 718.69 ± 8.4 mg L⁻¹) was measured using a cross-flow microfiltration system at 1 bar and 25°C for 60 min. Flux was recorded at regular intervals for kaolin-based ceramic membranes containing different additives and varying fly ash content (0–20 wt.%). Results are shown in Fig. 10.

Fig. 10.

Permeate flux (PF) of greywater as a function of time for (a) K-Z-F membranes and (b) K-Al-F membranes. Each curve represents the mean of three independent filtration runs (n = 3), and error bars correspond to the standard deviation among replicate experiments.

Figure 10a presents PF over time for kaolin (K), kaolin -natural zeolite (K-Z, 90-10), and kaolin -natural zeolite-fly ash (K-Z-F, 85-10-5, 80-10-10, 75-10-15, 70-10-20) membranes. Initial PF values are 908.21 ± 5.21, 1581.675 ± 9.07, 3228.263 ± 31.96, 3623.336 ± 20.78, 4647.337 ± 26.65, and 4282.124 ± 48.795 L m⁻² h⁻¹, respectively, decreasing to 66.58 ± 0.38, 358.35 ± 2.055, 405.79 ± 4.62, 537.12 ± 3.08, 632.29 ± 3.63, and 561.17 ± 3.22 L m⁻² h⁻¹ after 60 min due to fouling from oil and fat deposition. The K-Z-F (75-10-15) membrane exhibits the highest final PF (632.29 ± 3.63 L m⁻² h⁻¹), attributed to its optimized porosity (33.07 ± 0.25%), pore size (1.06 ± 0.02 µm), and fly ash hydrophilicity. Increasing fly ash up to 15% enhances PF by enlarging pores and improving water affinity, but PF declines at 20% due to particle agglomeration reducing pore connectivity.

Figure 10b shows PF for kaolin (K), kaolin-alpha alumina (K-Al, 90-10), and kaolin-alpha alumina-fly ash (K-Al-F, 85-10-5, 80-10-10, 75-10-15, 70-10-20) membranes. Initial PF values are 908.21 ± 5.21, 1646.46 ± 9.44, 2215.318 ± 25.89, 3102.09 ± 22.255, 2356.196 ± 24.55, and 2437.77 ± 20.53 L m⁻² h⁻¹, respectively, dropping to 66.58 ± 0.38, 461.58 ± 12.77, 526.57 ± 11.31, 559.798 ± 15.53, 531.47 ± 10.81, and 533.77 ± 13.26 L m⁻² h⁻¹ after 60 min due to cake layer formation. The K-Al-F (80-10-10) membrane achieves the highest final PF (559.798 ± 15.53 L m⁻² h⁻¹), driven by its higher porosity (41.26 ± 0.21%) and larger pore size (1.14 ± 0.02 µm). As shown in Fig. 10b, fly ash addition up to 10 wt.% in K-Al-F membranes increases permeate flux (reaching a maximum of 559.798 ± 15.53 L m⁻² h⁻¹ for K-Al-F (80-10-10)) due to enhanced porosity (41.26 ± 0.21%) and hydrophilicity; however, further addition to 15% and 20% leads to a slight decrease in flux (531.47 ± 10.81 and 533.77 ± 13.26 L m⁻² h⁻¹, respectively), likely due to increased fouling from particle agglomeration.

Compared to the baseline kaolin membrane (66.58 ± 0.38 L m⁻² h⁻¹), composite membranes with fly ash maintained significantly higher PF after 60 min, with K–Z–F (75–10–15) and K–Al–F (80–10–10) performing best under the tested conditions. The higher fouling tendency observed for K–Z–F (75–10–15) in FESEM/AFM analysis is partly offset by its higher short-term flux, while K–Al–F (80–10–10) provides a more balanced combination of flux and fouling resistance. These results should be interpreted as short-term, composition-dependent screening data; continuous operation, repeated cleaning cycles, flux-recovery analysis, and testing under variable greywater composition are required to establish long-term process stability.

Selecting of the most effective membranes

To identify the most effective membranes for greywater treatment, PF and COD removal efficiency were compared after 60 min of microfiltration. An optimal membrane balances high PF with effective pollutant removal. Results for all tested membranes are presented in Fig. 11, with COD reduction and effluent quality detailed in Fig. 12 and Table 7.

Fig. 11.

Comparison of COD removal efficiency and permeate flux (PF) of all membranes after 60 min of microfiltration. Error bars represent standard deviation (n = 3).

Fig. 12.

Filtration performance of the ceramic membranes based on COD during greywater microfiltration.

Table 7.

Quantitative analysis of wastewater, effluent discharge standards, and membrane-treated effluent.

Sample	COD (mg L⁻1)	BOD (mg L⁻1)	Detergent (mg L⁻1)	TSS (mg L⁻1)	Sulphate (mg L⁻1)	Nitrate (mg L⁻1)	Phosphate (mg L⁻1)	Turbidity (NTU)
Feed	718.69 ± 8.4	332	1.78	179	113	3.40	16.90	37.60
Effluent discharge standard to surface waters40	100	50	1.50	60	400	50	6	50
Effluent discharge standard to absorption wells40	100	50	0.50	–	400	10	6	–
Effluent discharge standard for reuse in irrigation and agriculture40	200	100	0.50	100	500	–	–	50
K-Z-F (75-10-15) the treated wastewater	95.67 ± 2.34	48	0.27	3	49	2.50	2.80	2.33
K-Al-F (80-10-10) the treated wastewater	61.08 ± 0.94	34	0.45	2	40	2.30	0.90	1.55

Figure 11 shows that the kaolin-natural zeolite-fly ash (K-Z-F, 75-10-15) and kaolin-alpha alumina-fly ash (K-Al-F, 80-10-10) membranes achieve the best balance of PF and COD removal. Since permeate flux and COD removal were used as the primary performance indicators for membrane selection, these parameters were evaluated in triplicate (n = 3) and are reported as mean ± SD. Error bars in Fig. 11 represent the corresponding SD values.

The K-Z-F (75–10-15) membrane delivers a PF of 632.29 ± 3.63 L m⁻² h⁻¹ and 86.69 ± 0.18% COD removal, while the K-Al-F (80–10-10) membrane achieves a PF of 559.798 ± 15.53 L m⁻² h⁻¹ and 91.50 ± 0.03% COD removal. These performances are driven by their optimized porosity (33.07 ± 0.25% and 41.26 ± 0.21%, respectively), pore size (1.06 ± 0.02 µm and 1.14 ± 0.02 µm), and fly ash hydrophilicity, which enhance flux while maintaining rejection efficiency. The K-Al-F (80-10-10) membrane’s higher porosity reduces fouling (Section “Field emission scanning electron microscope (FESEM)”), contributing to its superior COD removal, whereas K-Z-F (75-10-15)’s denser structure supports higher flux but increases fouling propensity.

The high COD removal values should not be interpreted as molecular-scale rejection of dissolved organic compounds by microfiltration pores. Kitchen greywater contains suspended solids, emulsified oil/fat droplets, food residues, detergent-associated colloids, and particulate organic matter, all of which can contribute substantially to measured COD. Therefore, COD reduction in this study is attributed to the combined effects of particulate and colloidal rejection, retention of oil/fat droplets, adsorption on aluminosilicate surfaces, and the formation of a dynamic fouling/cake layer that decreases the effective pore size during filtration. The FESEM and AFM observations after filtration support surface-associated fouling and pore blocking; however, soluble COD/particulate COD fractionation, TOC fractionation, oil-and-grease analysis, adsorption-only controls, and a complete mass balance were not performed. The proposed mechanism should therefore be regarded as a scientifically plausible interpretation rather than a fully resolved mechanistic proof.

Figure 12 illustrates COD reduction across all membranes. For the two optimized membranes, the measured COD and BOD values of the treated permeate were below the investigated Iranian limits for agricultural reuse and for discharge into surface waters or absorption wells⁴⁰. This compliance statement is limited to the parameters measured in this study and should not be generalized to all possible reuse criteria, such as microbiological safety, trace organic contaminants, pathogens, or long-term salinity effects. The electrical conductivity (EC) of the permeate from the two optimized membranes was also measured. The EC of the raw greywater was 2260 µS cm⁻¹, while treated permeate values ranged from 2190 to 2210 µS cm⁻¹. As expected for microfiltration, no significant change in EC was observed, indicating that the membranes mainly removed suspended/colloidal and organic contaminants rather than dissolved ionic species.

Table 7 summarizes the effluent quality for the optimized membranes. For K–Z–F (75–10–15), COD decreased from 718.69 ± 8.4 to 95.67 ± 2.34 mg L⁻¹ (86.69 ± 0.18% removal) and turbidity decreased from 37.60 to 2.33 NTU (93.80% removal). For K–Al–F (80–10–10), COD decreased to 61.08 ± 0.94 mg L⁻¹ (91.50 ± 0.03% removal) and turbidity decreased to 1.55 NTU (95.87% removal). These results are comparable to values reported by Jafari et al.³⁸ for ceramic membranes applied to greywater treatment. Nevertheless, the present experiments do not constitute a dedicated chemical-stability or cleanability test. Chemical stability is therefore not claimed as experimentally proven in this work; acid/base soaking, mass-loss measurements, leaching analysis, permeability retention after chemical exposure, and repeated cleaning-cycle tests are required before practical durability can be established. Within the short-term filtration window tested here, the enhanced performance is attributed to fly ash-mediated microstructural tuning, aluminosilicate surface interactions, and dynamic cake-layer formation.

Comparison of removal efficiencies for various parameters in different greywater treatment processes

Table 8 presents a contextual comparison of removal efficiencies (%) reported for different greywater treatment processes in the literature and for the ceramic membranes developed in this study. Because the compared studies differ in feed composition, treatment configuration, hydraulic conditions, membrane geometry, filtration duration, pore size, and analytical basis, the values are not directly normalized and should not be interpreted as a one-to-one superiority ranking. Only parameters available in each reference are included, and N/A indicates that the corresponding data were not reported.

Table 8.

Contextual comparison of wastewater treatment performance obtained in this study with selected greywater treatment technologies reported in previous studies (N/A means not available; values are not normalized across operating conditions).

Process	Removal efficiency (%)
	COD	BOD	detergent	TSS	Sulphate	Nitrate	Phosphate	Turbidity
Biofilter system41	64	61	N/A	75	N/A	85	88	N/A
Ceramic membrane38	88	87	40.4	N/A	71.5	86.5	94.3	N/A
Horizontal flow sand filter (HFSF)42	54.5	79	55.6	81.2	N/A	N/A	66.4	N/A
Mulch tower treatment system43	24–28	N/A	N/A	52	7–12	24–28	7–12	N/A
Bioelectrochemical GAC-packed bed (MEC/MFC)44	86.18–97.87	N/A	Complete (SDS)	N/A	70–89	84–100	28–86	85–98.3
Membrane (this study)	91.5 ± 0.03	89.7	74.6	98.8	64.6	32.3	94.6	95.8

SDS refers to sodium dodecyl sulfate, an anionic surfactant used as a detergent model compound.

As shown in Table 8, conventional biofiltration and sand-based systems generally exhibit moderate removal efficiencies for organic matter and suspended solids, while advanced membrane and bioelectrochemical systems often achieve higher COD and turbidity reductions under their respective operating conditions. The membrane developed in this work demonstrates competitive short-term removal efficiencies for COD, BOD, TSS, phosphate, and turbidity under the tested single-source kitchen-greywater conditions. The comparison is intended to place the present results within the broader greywater-treatment literature rather than to claim universal superiority.

Furthermore, the treated effluent quality obtained in this study met the investigated local Iranian limits for the measured parameters⁴⁰. Broader reuse validation would require additional monitoring of pathogens, long-term salinity effects, trace organics, and performance stability under variable greywater composition.

Table 9 compares the physical properties of the ceramic membranes developed in this study with representative membranes reported in the literature. Because membrane geometry, pore-size measurement method, feed type, operating pressure, and filtration duration differ among studies, this table is used only for broad benchmarking. The fabricated membranes show relatively high short-term permeate flux within the microfiltration range while maintaining adequate mechanical strength and defined average pore size. The higher flux is mainly attributed to the optimized pore structure and porosity achieved through fly ash incorporation under the tested conditions.

Table 9.

Contextual comparison between physical properties of the fabricated ceramic membranes and representative membranes reported in the literature (N/A means not available; operating conditions are not normalized).

Materials	Sintering temperature (°C)	Flexural strength (MPa)	Pore size (µm)	Porosity (%)	Permeate flux (L m⁻2 h⁻1)
Mullite monolith20	1400	43	0.45	64	100.5
Mixed clay-based membrane45	N/A	14.708 ± 2.603	0.092	57.27 ± 1.002	N/A
Saint-Gobain SiC (m) + SiC (s)46	2200	N/A	0.25	40	210
Al2O3 capillary monolith47	1350	15	0.4	47	11.6
This work	1250	(9.69 ± 0.27)–(30.87 ± 0.35)	(0.47 ± 0.003)–(1.15 ± 0.02)	(33.07 ± 0.25)–(41.26 ± 0.21)	(559.80 ± 15.53)–(632.29 ± 3.63)

Membrane cost assessment

To improve transparency, the membrane cost assessment was redrafted following the category-based structure used by Suresh et al.⁴⁸, in which raw materials, electricity, equipment/depreciation, labor, maintenance, and miscellaneous expenses are reported separately when available. In the present manuscript, the calculated numerical values are limited to direct laboratory raw-material and electricity costs because site-specific labor cost, equipment depreciation, maintenance, membrane replacement rate, infrastructure, and scale-up losses were not experimentally or economically quantified. Accordingly, Table 10 should be interpreted as a preliminary laboratory fabrication-cost basis rather than a complete industrial techno-economic analysis. Equations 8 and 9 were used to calculate raw-material and energy costs, respectively^22,48.

8

9

where C_material and C_energy are the raw material cost and energy cost, respectively. The cost contribution of each raw material was calculated from its weight fraction in the membrane composition and normalized to the membrane surface area. The energy cost includes electricity consumption associated with mixing, extrusion, and the complete electric-furnace sintering program. The experimentally applied thermal program in Table 4 corresponds to approximately 9.7 h of furnace operation, excluding cooling. Ambient drying at 25 °C was not counted as oven electricity consumption. Using this revised basis, the estimated direct raw-material-and-energy costs were 2.22 $ m⁻² for K–Z–F (75–10–15) and 4.27 $ m⁻² for K–Al–F (80–10–10). The higher cost of the K–Al–F membrane is mainly associated with the α-alumina content.

Table 10.

Revised laboratory-scale cost basis for optimized kaolin–natural zeolite–fly ash (75–10–15) and kaolin–α-alumina–fly ash (80–10–10) membranes.

Item	K–Z–F (75–10–15)	K–Al–F (80–10–10)
Length (cm)	25	25
Inner diameter (mm)	10	10
Outer diameter (mm)	14	14
Area used for cost normalization (m2)	0.011	0.011
Total raw material used per membrane (g)	36	38
Kaolin price ($ kg⁻1)	0.08	0.08
Fly ash price ($ kg⁻1)	0.25	0.25
Natural zeolite price ($ kg⁻1)	0.01	–
Corn starch price ($ kg⁻1)	0.015	0.015
α-alumina price ($ kg⁻1)	–	6
Raw material cost ($ m⁻2)	0.33	2.38
Magnetic stirrer energy cost ($ m⁻2)	0.02	0.02
Extruder energy cost ($ m⁻2)	0.055	0.055
Ambient drying energy cost ($ m⁻2)	0.00	0.00
Furnace energy cost, full schedule ($ m⁻2)	1.814	1.814
Total energy cost ($ m⁻2)	1.89	1.89
Direct laboratory fabrication cost: raw material + electricity ($ m⁻2)	2.22	4.27
Labor cost	Not included; site-specific	Not included; site-specific
Equipment depreciation	Not included; requires equipment-life basis	Not included; requires equipment-life basis
Maintenance / consumables / cleaning	Not included	Not included
Membrane failure, replacement, infrastructure, scale-up losses	Not included	Not included
Interpretation	Preliminary direct laboratory cost only	Preliminary direct laboratory cost only

The electricity tariff was assumed to be 0.1 $ kWh⁻¹. The cost values are normalized to the external geometric membrane area (0.011 m²). The furnace operating time was recalculated from the complete sintering schedule in Table 4 (~ 9.7 h, excluding cooling). Ambient drying was performed at 25 °C and was not counted as oven electricity. Labor, equipment depreciation, maintenance, membrane failure/replacement, infrastructure, cleaning, and scale-up losses were not included in the direct cost and require pilot-scale techno-economic analysis.

It should be noted that these values are preliminary direct fabrication-cost estimates and are lower-bound values because they do not include labor, depreciation of the extruder/furnace/module, maintenance, membrane breakage or rejection during scale-up, cleaning chemicals, infrastructure, quality control, financing, or end-of-life costs. Therefore, the values are used only to compare the two optimized formulations on the same laboratory basis. A realistic industrial assessment would require pilot-scale production data, membrane lifetime, replacement frequency, cleaning protocol, module packing density, and full techno-economic analysis. This cautious framing prevents direct comparison with studies that include broader cost categories unless their assumptions are explicitly aligned.

Limitations and future scope

Several limitations should be recognized when interpreting the results. First, the greywater was collected from a single kitchen source and most physicochemical parameters were measured once for feed characterization. Therefore, the results do not represent the full temporal, seasonal, or household-to-household variability of domestic greywater.

Second, the filtration tests were conducted for 60 min and were intended for controlled comparison among membrane compositions. They do not establish long-term operation, repeated fouling-cleaning behavior, irreversible fouling, flux recovery ratio, regeneration performance, or stability under variable greywater composition.

Third, the COD-removal mechanism was inferred from COD, turbidity, TSS, FESEM, and AFM results. Soluble/particulate COD fractionation, TOC fractionation, oil-and-grease analysis, adsorption-only controls, and complete mass-balance experiments were not performed. These tests should be included in future work to quantitatively separate size exclusion, adsorption, and cake-layer contributions.

Fourth, XRD and TGA were used for qualitative phase and thermal characterization only. Quantitative XRD/Rietveld analysis, chemical leaching tests, acid/base resistance, cleaning-chemical exposure, and mechanical-strength retention after operation are required before long-term chemical durability can be claimed.

Finally, the cost analysis provides a preliminary direct laboratory raw-material-and-energy basis only. Pilot-scale studies should include labor, equipment depreciation, maintenance, membrane replacement, module design, cleaning cost, infrastructure, and scale-up losses. Future work should therefore evaluate multiple greywater sources, long-term cyclic filtration, cleaning protocols, durability tests, pathogen removal, and pilot-scale techno-economic feasibility.

Conclusion

In this study, low-cost tubular ceramic microfiltration membranes based on kaolin, natural zeolite, fly ash, and α-alumina were fabricated and evaluated for laboratory-scale greywater microfiltration. Fly ash addition influenced porosity, pore connectivity, flux, and mechanical behavior, although excessive fly ash content (> 20 wt.%) promoted crack formation during drying due to shrinkage and internal stress development. Among the investigated compositions, K–Z–F (75–10–15) and K–Al–F (80–10–10) showed the most favorable short-term performance under the tested conditions. K–Z–F (75–10–15) provided the highest greywater permeate flux (632.29 ± 3.63 L m⁻² h⁻¹) and high flexural strength (30.87 ± 0.35 MPa), while K–Al–F (80–10–10) provided the highest COD reduction (91.50 ± 0.03%) with improved fouling resistance linked to its higher porosity. The optimized membranes had average pore sizes of 1.06–1.14 µm, confirming their microfiltration character. The observed COD decrease is attributed to combined rejection of suspended/colloidal organic matter, oil/fat droplets, adsorption, and dynamic cake-layer filtration rather than dissolved-COD size exclusion alone. The treated effluents met the investigated Iranian standards for the measured parameters, but broader reuse validation requires additional microbiological, chemical, and long-term stability assessments. The revised direct laboratory raw-material-and-energy cost estimates were 2.22 $ m⁻² for K–Z–F (75–10–15) and 4.27 $ m⁻² for K–Al–F (80–10–10); these values are preliminary and do not replace a full industrial techno-economic analysis. Overall, the results support the potential of fly-ash-modified kaolin ceramic membranes for decentralized greywater treatment, while also identifying the mechanistic, durability, sampling, and scale-up studies required before practical implementation.

Author contributions

Conceptualization, E.G. and R.S.; methodology, R.S., E.G., A.K; validation, M.A.; formal analysis, R.S.; investigation, E.G.; resources, N.D.; data curation, M. A. and R. S., N.D.; writing—original draft preparation, E.G., R.S.; writing—review and editing, M.A.; visualization, R.S.; supervision, M.A. and R. S., A.K; project administration, M.A., A.K., R.S.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.