Ultrafiltration of Car Wash Wastewater: Pilot-Scale Studies

Abstract

Currently, the world faces serious challenges in meeting the growing demand for clean water. The present paper demonstrates the possibility of using the ultrafiltration (UF) process to reuse water from wastewater generated in car washes. Car washes commonly use foaming agents with dyes, which, although they are not necessary for washing cars, may hinder water reuse. For this reason, the aim of this work was to investigate the effect of the dyes present in car wash wastewater on the membrane fouling intensity. For this purpose, experimental tests were conducted with the application of a pilot plant with an industrial PCI B1 membrane module. The module was equipped with tubular FP100 (100 kDa) polyvinylidene fluoride (PVDF) membranes. For the feed, two types of cleaning agents and synthetic wastewater were used. The results obtained in the current study demonstrated that the UF membranes allowed the obtainment of the permeate characterized by high quality. In addition, it has been shown that the presence of Indigo carmine dye in the wastewater led to an increase in the fouling intensity. To sum up, it should be pointed out that the findings presented in the current study may be of key importance in the design of pilot installations used for the treatment of car wash wastewater.

1. Introduction

The world is currently facing significant challenges in meeting the growing demand for clean water. Only 3% of the Earth’s water is available as freshwater resources suitable for drinking [1]. In line with global trends in environmental protection and the circular economy, many countries have made efforts in wastewater recovery, establishing stringent regulations to encourage water reuse. As it has been pointed out in [2], in most developing countries, the provision of clean water is prioritized over other services. Reused water is defined as the wastewater that has been treated with the use of various treatment processes for meeting target water quality [3]. Worldwide, the total volume of reused water is approximately 14.2 billion m³ per year [4]. The literature analysis has shown that the number of published articles focused on this issue has significantly increased in recent years (Figure 1).

At the same time, the car wash industry is known as an industry that produces a high amount of wastewater. Moreover, it is expected to grow [5]. Hence, due to legislation and water scarcity, there is a requirement for the application of solutions ensuring the reuse of water in this industry. Accordingly, modern technologies for car washing including water recycling and reuse are expected to be implemented. Nevertheless, it is a great technological challenge since wastewater released from car wash centers can contain many pathogenic microorganisms [6] and a wide range of contaminants, such as the following: (i) suspended particles, (ii) oil, (iii) oil and grease, (iv) anionic surfactants, (v) fuel residues, and (vi) metals [7,8,9,10,11]. It is worth noting that the composition of wastewater may vary depending on the type and size of car being washed and the chemical products, soaps, and detergents used. The characteristics of car wash wastewater have been presented in a recently published review paper [12]. Wastewater also contains dyes that are added to the foaming solution to make the foam more attractive, yet removing dyes from wastewater is a great challenge [13,14] and a serious threat to the environment [15].

Obviously, untreated or insufficiently cleaned car wash wastewater discharged into natural water bodies may contaminate the land and water [16]. Therefore, the main aim of an appropriate treatment of this wastewater type is to reduce the negative impact of car washing on the environment and to reuse water resources [7,8]. To date, several studies have proven that the treatment of car wash wastewater may be ensured by numerous conventional technologies. The advantages and disadvantages of the above-mentioned techniques have been described in several studies [11,17,18,19].

Importantly, in recent years, the use of membrane processes, especially ultrafiltration (UF) for this purpose, has received significant research attention, e.g., [20,21,22,23,24,25]. This is due to the fact that UF membranes allow the removal of pathogens, contaminants, and suspended particles presented in a feed [26]. Moreover, compared to most conventional technologies, membrane technology provides high-quality permeation, offers a more holistic solution, is cheaper, and is easier to use and manage [27,28,29,30]. Polyvinylidene fluoride (PVDF) is widely used for the UF process due to its good thermal stability, high mechanical strength, and excellent chemical and aging resistance [13,31]. However, it is well known that one of the challenges for full-scale applications of UF technology is fouling. This is defined as the phenomenon which causes the accumulation of suspended or dissolved solids on the membrane surface and/or its pores [32,33]. It is a complex issue which may include adsorption, pore blocking, and cake formation [34,35,36]. Obviously, it leads to a decline in the process performance and may affect the economic profitability of the membranes technology. For these reasons, the application of the UF process in a car wash requires the use of an effective cleaning agent which allows the restoration of the maximum permeate flux. It should be pointed out that in addition to detergents, cleaning agents used in car washes often contain dyes. This includes, for instance, Indigo carmine dye, that may lead to the human health deterioration and several environmental issues [37,38]. For this dye, the most intensive fouling of polypropylene (PP) and polytetrafluoroethylene (PTFE) membranes was observed during the filtration of car wash wastewaters containing yellow, green, red or blue dye [39]. The intensive adsorption of dyes on the surface of PVDF membranes was also demonstrated in the study [13]. It has been widely documented that UF may be an effective process for the separation of dyes present mainly in textile wastewater [26,40,41,42,43,44]. However, performing the literature review allowed us to find that up to now, the impact of dyes present in car wash wastewaters on UF process performance has not been determined. Therefore, in the current study, the impact of Indigo carmine dye, which is often added to cleaning agents used for washing cars, was investigated.

In a few studies [13,14], the influence of UF membrane performance, wastewater composition, and dye types on dye rejection degree was investigated. However, reclaimed water in a car wash station does not have to be completely free of all the components (e.g., dye or surfactants) which are present in wastewater [45]. Indeed, from a practical point of view, limiting the fouling intensity to maintain high permeate flux is of significant importance.

It is recognized that the studies, with the use of a pilot-scale installation, are crucial for considering the possibility of the industry implementing the UF process for the treatment of car wash wastewater. Nevertheless, in most of the studies, the experimental investigations with the use of UF membranes for this purpose were performed on a laboratory scale. On such a scale, the representation of industrial conditions is a great challenge, which makes it difficult to investigate the effect of the dyes’ presence in car wash wastewater on the fouling intensity. It was assumed that the dyes present in wastewater adsorb on membranes and change their properties, which affects the intensity of fouling during the separation process. To imitate the UF process and evaluate the effects of the applied membrane washing procedure in the current study, an industrial membrane module was used.

2. Materials and Methods

2.1. Experimental Set-Up

In the current work, a pilot-scale installation was used (Figure 2). It was equipped with 18 tube FP100 membranes made of polyvinylidene fluoride (PVDF) (PCI Membranes, Kostrzyn Wielkopolski, Poland). The membranes’ diameter and length were equal to 1.25/1.27 cm and 120 cm, respectively. The total membrane area was 0.9 m². The molecular weight cut-off (MWCO) declared by the manufacturer was 100 kDa. The membranes were mounted in a stainless steel PCI B1 module with a diameter of 10 cm (29% packing density).

The UF process was performed at transmembrane pressure (TMP) in the range from 0.05 to 0.2 MPa. The feed was pressurized by a centrifugal pump (type VNR-8, Grundfos, Bjerringbro, Denmark) and flowed inside the tubular membranes with a velocity equal to 1.2 m/s. A parallel flow head was mounted in the B1 module, which allowed for the obtainment of similar process conditions in each FP100 tube. As a result, the obtained test results are the average of 18 membrane samples.

The Pilot UF was equipped with a feed tank with a capacity of 200 L, and the feed line volume was 20 L. The valve system in the permeate line allowed for the continuous collection of permeate or its return to the feed tank (constant concentration option). For all types of the solutions tested, the feed temperature was 300 ± 1 K. Temperature control was ensured by a shell-and-tube cooler (made of AISI 321 stainless steel) through which the feed flowed.

The pilot installation was operated in a university laboratory; therefore, due to safety regulations, UF tests were conducted with a night break. After completing a given series of tests, the installation was rinsed twice with tap water softened in the nanofiltration process (50–60 µS/cm). Additionally, chemical cleaning was also used periodically, for 10 or 30 min. For washing the tested membranes, the PCI manufacturer recommends P3 Ultrasil 11 cleaning agent, which contains NaOH and detergents (SUTURAMED, Szczecin, Poland). For membrane washing, 0.1 wt.% solutions (pH = 11.7) were used.

2.2. Feed Solutions and Process Conditions

In the first stage of the experimental research, tests were carried out with the use of two types of cleaning agents (EuroEcol, Łódź, Poland):

(i)

Euro Turbo Foam (‘White’), without added dye;

(ii)

Euro Turbo Foam Color Blue (‘Blue’), which has the same composition as White; however, this is enriched with Indigo carmine dye (5,5′-indigodisulfonic acid sodium salt). It is among the most commonly employed dyes [46]. Indeed, it is popularly used in food and other consumables (E132), cosmetics, and as a medicine contrast agent. It is also used as a fabric dye, which indicates its strong adsorption properties [47].

The composition of the above-mentioned cleaning agents is presented in [48]. The manufacturer supplies them in the form of liquid concentrates containing diethylene gly-col butyl ether, benzenesulfonic acid, and polymers (each component in the amount of 1–5 wt.%).

The UF tests were conducted for TMP = 0.1 MPa. The feed tank (200 L) was filled with water (NF permeate, 55 μS/cm) and the process was conducted for at least 1 h until a stable permeate flux was obtained. Then, ‘White’ or ‘Blue’ concentrate was added to the feed to obtain a 0.5 vol.% solution. Importantly, this concentration is similar to that used in car washes. After the permeate flux had stabilized, the feed solution was concentrated. The process was completed when its volume decreased from 200 to 50 L.

Additionally, during the UF process, the effect of pressure on the permeate flow was determined for water and solutions (VCR = 1, 2, and 4). In this case, the pressure was reduced to TMP = 0.05 MPa and increased every 5 min to 0.1, 0.15, and 0.2 MPa. After taking readings, the pressure was reduced to TMP = 0.1 MPa and the UF process was continued.

In a final experiment, the UF process was carried out with the use of synthetic wastewater consisting of a mixture of 0.5 vol.% foaming agent solution (‘White’ or ‘Blue’) and 0.2 vol.% Hydrowax (EuroEcol, Łódź, Poland). This agent is used to polish the car and contains diethylene glycol monobutyl ether polymer (10–20 wt.%) and polymeric waxes. In this case, the studies were performed at a constant feed concentration, which was achieved by returning the permeate to the feed tank. In order to increase the fouling intensity, the TMP value was increased to 0.2 MPa.

2.3. Analytical Methods

The turbidity of the test solutions was determined using a portable turbidimeter, model 2100 AN IS (Hach Company, Loveland, CO, USA). The values of electrical conductivity were measured using a 6P Ultrameter (Myron L Company, Carlsbad, CA, USA).

The Hach cuvette tests were used to determine the concentration of surfactants (LCK 334—nonionic, LCK 344—anionic) and chemical oxygen demand—COD (LCK 1014).

The rejection efficiency [%] was determined as follows:

$$\mathrm{R}=\left(1-{\displaystyle \frac{{\mathrm{C}}_{\mathrm{P}}}{{\mathrm{C}}_{\mathrm{f}}}}\right)·100\mathrm{\%},$$

(1)

where C_p [mg/L] and C_f [mg/L] are the measured oil concentrations of the permeate and feed, respectively.

The water recovery coefficient [%] was calculated with the following equation:

$$\mathrm{VCR}={\displaystyle \frac{{\mathrm{V}}_{\mathrm{F}}}{{\mathrm{V}}_{\mathrm{F}}-{\mathrm{V}}_{\mathrm{P}}}}$$

(2)

where V_F and V_P are the volumes of the feed and permeate, respectively.

The membrane surface was examined by atomic force microscopy (AFM). A multi-Mode 8 AFM apparatus equipped with a Nanoscope V converter from Bruker (Billerica, MA, USA) characterized the membrane roughness in the scanasyst mode. The R_a and R_q parameters were evaluated on a basis of at least five AFM images (10 μm × 10 μm).

3. Results and Discussion

3.1. UF of White Foam Solution

Before carrying out tests using cleaning agents as a feed, the UF installation was cleaned with P3 Ultrasil 11 solution and then rinsed with water twice (100 L each). Subsequently, the installation was filled with a new portion of water (200 L) and the UF process was carried out for 2 h at TMP of 0.1 MPa, obtaining a stable water flux of 240 LMH.

It has been noted that adding the ‘White’ agent concentrate to water (0.5 vol.% solution) caused the flux decline to 95 LMH (Figure 3, ‘S1 Series’). The initial flux was equal to 39% of the water flux, which indicates pore blocking [49]. For the first 260 min of the process run, the obtained permeate was recycled to the feed tank (VCR = 1) and a stable permeate flux of 88 LHM was obtained. Then, the permeate was successively collected, which led to the subsequent flux decline to 67 LHM (VCR = 4). This could be due to the increasing feed concentration and a significant increase in its turbidity which led to an increase in the intensity of concentration polarization and fouling. This finding is consistent with the results presented in several previous studies wherein it has been documented that during the pressure-driven membrane processes, the permeate flux is influenced by the feed turbidity. This result was recorded for the following feed types: reactive dye [40], seasonal high-turbidity surface water [50], water [51], and fermentation broths [52].

The changes in the membranes’ performance that occurred during series S1 were analyzed by measuring the dependence of permeate flux on TMP (Figure 4). Overall, since TMP is a driving force of the UF process, if the membrane permeability does not change, the permeate flux increases linearly with an increase in TMP. In the figure, some of the points for TMP = 0.1 MPa lie above the straight line marked by the extreme points. This indicated the formation of a fouling layer, which, due to compression, reduced the flux for TMP = 0.15 MPa. It is worth noting that this effect increased with increasing feed concentration (Figure 4, ‘VCR = 2 and VCR = 4’). This change for the initial water flux (Figure 4, ‘Water’) was the result of another factor. There were also large pores (0.1 μm) on the surface of the FP100 membranes, which were cleaned by washing with P3 Ultrasil 11, which increases the initial flux [15]. During UF, these pores become blocked by deposits, which is facilitated by increasing TMP [49]. Module rinsing with water does not remove deposits from the pores’ interior, hence, in this case, a good convergence of the measurement points with the straight line was obtained (Figure 4, ‘rinsed’). The results of the study confirmed that under conditions favoring the increase in fouling and gel layer thickness (Figure 4, ‘White, VCR = 2 and VCR = 4’), the increase in TMP caused a smaller increase in permeate flux.

After each completed measurement series, the system was thoroughly cleaned three times using water. As a matter of fact, it did not allow for the restoration of the initial membranes performance (Figure 4, ‘rinsed’). For instance, after the membranes were rinsed, the flux for water at TMP of 0.1 MPa was equal to 150 LMH. From this analysis, it can be clearly seen that the main reason of the flux decline during the UF process was the irreversible fouling phenomenon. For this reason, before the second series of ‘White’ solution separations, the membranes were washed with P3 Ultrasil 11 solution (Figure 3, ‘W’), which allowed for an increase in the water flux to 175 LMH; however, after adding the ‘White’ agent, it decreased to 103 LMH (series S2). Moreover, it is worth noting that during the second series of the UF process (Figure 3, ‘S2 series’), the initial flux of 103 LMH decreased to 98 LHM when the feed concentration was started. In this case, a higher flux was obtained initially than during the S1 series (Figure 3, ‘S1 Series’). This can be explained by the fact that in this case, the feed was characterized by the lower turbidity (Figure 5, ‘White 2’).

Before performing the research presented in the current paper, the installation (Figure 2) was used to investigate other issues related to the separation of car wash wastewater. Despite the repeated cleaning of the set-up, after resuming its use, the remaining contaminants were probably detached from the pipe surfaces by the cleaning agent in the S1 series. In the next series focused on the separation of ‘Blue’ solution, the feed turbidity was already at a similarly lower level (Figure 5, ‘Blue’).

As it has been indicated by Yang et al. [4], the permeate quality is critical for the realization of water reuse. In the present study, despite the differences in the feed turbidity, the permeate turbidity was equal to 0.1–0.2 NTU (Figure 5). Slightly higher permeate turbidity values were recorded for the washed membranes. As shown in a previous work [21], the chemical cleaning of membranes successfully removes foulants from their surface and opens larger membrane pores. With respect to these findings, it should be clearly pointed out that the UF membranes used in this study allowed the obtainment of the permeate characterized by high quality. Indeed, for all measurement series, regardless of the feed type, the permeate turbidity did not exceed 0.3 NTU. Accordingly, a turbidity reduction higher than 98% was noted.

During S2 series (Figure 3), after receiving 150 L of permeate (VCR = 4), the process efficiency decreased to 65 LMH, which is similar to that obtained during the S1 series. However, after rinsing the installation with water, the obtained flux (TMP = 0.15 MPa) was 185 LMH (Figure 6, ‘rinsed’), which was lower than that obtained during the S1 series tests (220 LMH, Figure 4). This value increased to 265 LMH after washing the installation with P3 Ultrasil 11 solution. Increasing the membrane cleaning time to 30 min allowed the removal of the deposits from the large pores, which resulted in a significant increase in the permeate flux. During the TMP effect measurements, these pores were blocked again, which limited the flux value for TMP = 0.15 MPa. As a result, for similar reasons as for the initial water flux (Figure 4, ‘Water’), the point for TMP = 0.1 MPa also lies above the line (Figure 6, ‘washed’).

3.2. UF of Blue Foam Solution

In the next step of the presented research, the UF process of the ‘Blue’ solution was conducted (Figure 7). It has been found that the separation of the ‘Blue’ solution allowed us to receive a slightly higher permeate flux than that noted during the separation of the ‘White’ solution. Hence, the concentration of the feed to VCR = 4 was obtained after 108 min, while for ‘White’ this was after 130 min. This result was influenced by the fact that after chemical washing, the initial efficiency (TMP = 0.1 MPa) during the separation of the ‘White’ and ‘Blue’ solutions was 103 LMH and 125 LMH, respectively.

The second reason could be differences in the fouling mechanism. The solutions tested contained anionic and nonionic surfactants. The surfactants present in the feed form a gel/fouling layer which reduces the permeability of UF membranes [49]. Nonionic surfactants adsorbed on hydrophobic membranes cause a more pronounced flux decline than ionic surfactants [53]. Anionic dyes with surfactants form agglomerates with a negative charge, which limits their deposition on the surface of PVDF membranes [13]. This phenomenon may explain the slight increase in permeate flux for the ‘Blue’ feed.

It is necessary to point out that after concentrating the ‘Blue’ solution, washing the system with water allowed the obtainment of a much larger flux (Figure 8) than in the case of testing the ‘White’ solution (Figure 6). This can be attributed to the fact that the outer layer PVDF membrane is similar to the anionic structure of the studied blue dye carrying the negative charges [13]. The separation mechanisms of the binary dye mixtures using a PVDF ultrafiltration membrane led to the Donnan effect and their intermolecular character [13]. In contrast to the ‘White’ solution tests, during the separation of ‘Blue’ solutions, all points for TMP = 0.1 MPa were above the line. This indicates that in this case, the gel/fouling layer formation was greater.

The membrane selectivity is of crucial importance based on the intention of the separation performed [30]. Hence, in the study, the retention degree for suspensions and dissociable compounds has been investigated with the use of both feed types (Figure 9). It has been found the suspension (turbidity) retention in both cases was equal to 98%. In addition, it has been noted that the permeate conductivity was less than two times lower compared to the conductivity feed. This can be explained by the fact that generally, UF membranes ensure ions are retained to a limited extent [54]. Anionic dyes are retained by PVDF membranes to a greater extent due to electrostatic interactions [13]. In the case studied, however, it was not found to cause a significant increase in the level of rejection calculated for conductivity. Higher values were obtained for the separation of surfactants and COD (

Table 1. UF separation results: COD and surfactants (anionic and nonionic). P—permeate, F—feed, R—rejection, S1—series 1, S2—series 2.

Feed	COD [mg/L]			Anionic [mg/L]			Nonionic [mg/L]
	F [%]	P [%]	R [%]	F [%]	P [%]	R [%]	F [%]	P [%]	R [%]
White S1	2312	1014	56.2	776	269	65.3	28	4.0	85.7
VCR = 2	3435	1125	67.2	916	270	70.5	41.8	4.0	90.4
VCR = 4	4932	1195	75.8	1415	296	79.1	65.2	6.0	90.8
White S2	2330	1030	55.8	790	275	65.2	27.4	1.20	95.6
VCR = 2	3480	1135	67.4	940	277	70.5	45	1.55	96.6
VCR = 4	5044	1200	76.2	1468	284	80.6	71	1.35	98.0
Blue	2448	1024	58.2	780	270	65.4	30	1.30	95.7
VCR = 2	3375	1126	66.6	920	279	69.7	44.8	1.88	95.8
VCR = 4	5028	1198	76.2	1438	288	80.0	63.6	2.47	96.1

).

The concentration of the feed caused the content of organic solutes to systematically increase, in most cases more than twice (Table 1). Despite this, their content in the permeate increased only by 10–17%. This indicates that the growth of the gel/fouling layer during concentration increased the retention rate. This is also evidenced by the fact that for the membranes used with a MWCO of 100 kDa, a reduction in the COD value was obtained that was 10–20% higher than that obtained during the separation of dye solutions by a more dense polyethersulfone (PES) membrane (MWCO of 10 kDa) [49].

The AFM studies showed that there were no significant differences between the characteristics of the new membrane and membranes used for the separation of ‘White’ and ‘Blue’ solutions (Figure 10). This also confirms the above presented result demonstrating that there was no significant difference in the permeate flux obtained (Figure 7) and in the separation degree (Figure 9, Table 1). Surface roughness analysis showed only a slight increase, especially for the membrane used for the separation of the ‘Blue’ solution (

Table 2. Membrane surface roughness.

Parameter	New Membrane	Membrane Used for Separation of ‘White’ Solution	Membrane Used for Separation of ‘Blue’ Solution
Rq	37 ± 4.9	43.1 ± 5.1	45 ± 8.1
Ra	29.7 ± 4	34 ± 4.8	35.5 ± 5.9

).

This result indicated that the dye could permanently adsorb locally on the surface of the tested PVDF membranes. To confirm this, the FP100 membrane sample was immersed for 24 h in a 0.5 vol.% ‘Blue’ solution. After this period, the blue-dyed membrane was intensively rinsed in distilled water. Despite this, the membrane was still slightly blue, which indicated that small amounts of the dye remained on the membrane surface. This irreversible fouling caused by the dyes was also found in another work, which resulted in an approximately 8% decrease in the initial permeate flux [49].

3.3. UF of Synthetic Wastewater

In the present study, the synthetic car wash wastewater was prepared from a mixture of 0.5 vol.% foaming agent (‘White’ or ‘Blue’) solution with 0.2 vol.% Hydrowax. The Hydrowax solution used in the final stage of car washing contains substances that create a protective hydrophobic layer on the car surface. As shown in previous works, the filtration of such a mixture led to a significant fouling phenomenon [48,55].

In the conducted research, three measurement series were performed (Figure 11). During the downtimes (12 h—night break), no action was taken and the tested solution remained in the installation. It is clear from these details that for both of the synthetic wastewater mixtures, a significant decrease in the flux was observed. However, a greater decrease in the process performance flux was noted for the mixture with the ‘Blue’ solution. Indeed, a slightly higher UF process performance after the end of Series S3 was noted for the mixture with a ‘White’ solution. Hence, it can be concluded that the presence of Indigo carmine dye in the wastewater led to an increase in the fouling intensity. This can be explained by the fact that it has adsorption properties which caused an increase in the feed–membrane surface interactions, contributing to the formation of a filter cake and the blocking of the membrane pores. During the UF process of the clean ‘Blue’ solution, the decrease in permeate flux was much less significant (Figure 7). This indicated that in complex mixtures, there may be interactions between the dye and its components, the products of which may intensify the fouling phenomenon.

Finally, it is interesting to note that the plant downtimes of the installation did not have a significant impact on the UF process performance. This finding may be of key importance in the design of pilot installations intended for the treatment of car wash wastewater.

The results presented in Figure 12 confirm that the addition of ‘Hydrowax’ to a feed led to the more significant decline in the permeate flux. In the case of the ‘White’ solution for TMP = 0.15 MPa, the initial flux was 140 LHM (Figure 6, ‘White’), while in the mixture with Hydrowax, the flux at the beginning of the S1 series, it was only 110 LHM (Figure 12, ‘White + Hydrowax S1’). This finding clearly indicates that in this case, the fouling intensity became more severe. Moreover, the ‘Hydrowax’ presence decreased the efficiency of the membrane rinsing. After the UF process of the “White” and ‘Blue” solutions, the rinsing module with water always resulted in an increase in permeate flux (Figure 6 and Figure 8, ‘rinsed’). Meanwhile, for Hydrowax mixtures, it has been found that the lowest value of permeate flux (feed–water) has been obtained for membranes which have been used for the UF process of the wastewater that contained the ‘White solution’. Lower water flux values were recorded after the UF process of wastewater that contained the ‘Blue solution’ (Figure 12, ‘Blue + Hydrowax’ (rinsed)). Therefore, it can be concluded that the presence of Indigo carmine dye led to an increase in the irreversible fouling intensity.

After completing the measurement series, the UF installation was rinsed with water, washed with P3 Ultrasil 11 solution (30 min), and then washed twice with water. The performing of this procedure allowed the obtainment of a water flux close to the initial one.

4. Conclusions

The main aim of this study was to determine the effect of the dyes present in car wash wastewater on the fouling intensity and the efficiency of the UF process using an industrial membrane module. The results presented in the current study may contribute to more efficient and sustainable solutions for car wash wastewater on a pilot scale.

It has been demonstrated that the PVDF tubular membranes used in the current study allowed the obtainment of the UF permeate characterized by a turbidity that did not exceed 0.3 NTU. Importantly, water of this purity can be successfully used for washing cars.

The applied FP100 membranes rejected surfactants at the level of 65%, which due to the formation of a fouling layer, increased to over 80%, which resulted in a reduction in COD by 76%. Adding dye (Indigo carmine) to the foaming solution did not cause any significant changes in the obtained results.

Moreover, it has been found that the application of 75% water recovery (VCR = 4) caused the permeate flux to decrease to 60–70 LHM (TMP = 0.1 MPa), both for foaming solution with and without dye. Washing the membranes with water allowed the recovery of the initial permeate flux at about 75% (White) and 82% for the solution containing dye (Blue). It is important to note that the initial flux was recovered after performing membrane washing (30 min) with P3 Ultrasill 11 solution (pH = 11.7).

After the UF of synthetic wastewaters containing the tested foaming solution and the mixture of polymeric waxes (Hydrowax), rinsing the membranes with water resulted in a slight increase in water flux, which indicated a higher intensity of the irreversible fouling phenomenon. It has been shown that the presence of Indigo carmine dye in the wastewater led to the increase in the membrane fouling intensity. However, in both cases, the initial permeate flux was recovered after intensive membrane washing with P3 Ultrasill 11 solution.

Moreover, it has been noted that the plant downtimes of the pilot installation did not have a significant impact on the UF process performance.