Mixed laundry and dishwashing greywater were collected from five households located in “Zogona” and “Wentenga” districts of Ouagadougou (Burkina Faso) and used for the treatment. The experimental device consisted of seven cylindrical polyvinyl chloride (PVC) columns with internal diameter of 16 cm and a height of 1 m in which were distributed different layers of filter media (Figure 1). Each column was filled from the bottom to top with a drainage layer (10 cm of granitic gravel with a size of 11 - 25 mm), a transition layer (10 cm of granitic gravel with a size of 4 - 10 mm) and a varying filtration layer consisting of either sand, granitic gravel or lateritic gravel (Figure 1). The columns were equipped with a PVC pipe at their base to allow the collection of the treated greywater.

In order to simulate household greywater production and discharge, the biofilters

were fed at 8 am, 1 pm and 6 pm on Saturday, Sunday and Monday and left to stand the rest of the time. A total of 4 L/d (1.5, 1 and 1.5 L at 8 am, 1 pm and 18 am respectively) corresponding to a hydraulic load of 200 L/m²/d was distributed into each biofilter.

Greywater samples (raw and pretreated) were collected during the second feeding period (1 pm) for analysis. During sampling, the pH was measured in situ from the effluent and influent using a pHmeter multi-340 (WTW GmbH, Germany). Some important physico-chemical parameters (5-day biological oxygen demand [BOD₅], chemical oxygen demand [COD], suspended solids [SS] were measured. The SS were determined by a gravimetric method using glass microfiber filters Whatman (porosity 1.5 μm). The BOD₅ was determined by the respirometric method using an Oxitop placed in an incubator set at 20˚C and operating in the dark for five (5) days. The COD was determined by the acidic oxidation method in an excess of potassium dichromate, at a temperature of 150˚C.

Escherichia coli and fecal coliforms were assessed as indicator bacteria. We used Chromocult Coliform Agar ES as culture medium for both bacteria. For enterococci, we used M-Enterococcus Agar as culture medium. The spread plate method was used after appropriate dilutions of the samples. All parameters were determined in accordance with the procedure described in standard methods for the examination of water and wastewater [12].

Since, maintenance is an important factor influencing the success and acceptability of a treatment system, the infiltration time of each treatment unit was measured and used as a basis to estimate the clogging level of the different biofilters. This infiltration time was measured as the time necessary for the greywater to pass throughout the system.

For each biofilter, we evaluated the removal efficiency (RE) of physico-chemical parameters and bacteria using equation 1 and 2 respectively.

R E ( % ) = ( X 0 − X X 0 ) ∗ 100 (1)

R E ( log .u ) = log ( X 0 ) − log ( X ) (2)

where X₀and X = concentration of a considered parameter in the raw greywater and the effluent respectively.

For all parameters, the RE was determined using Excel software. The effects of media type and particle size on the efficiency of the biofilters were compared using the t-test (α = 0.05).

The raw greywater showed high concentrations of organic matter and fecal indicators (Table 1). The pH values were slightly alkaline (7.47 and 8) with an average of 7.69. Similar results were reported by [7] in their work on the design of a “slanted soil” system for greywater treatment in rural Burkina Faso. The alkaline pH of greywater is related to the use of soaps and detergents in dishwashing and laundry activities [13] [14]. The concentration of organic matter (BOD₅, COD, SS) was high in the raw greywater. Similar results were obtained by [14] who found DCO value of up to 1120 mg/L on urban greywater in Lagos (Nigeria). The high concentration of organic matter observed in greywater could be explained by the presence of food residues from dishwashing, but also clothing fibers that could also be found in laundry greywater. However, our results are higher than those from developed countries like France where values of 110, 398 and 196 mg/L were reported for SS, COD and BOD respectively [15]. The low concentration of organic matter in the greywater from developed countries compared to that of developing countries could be the consequence of the dilution, owing to the large quantity of water used in domestic activities in developed countries.

In our study, we also found high contents of fecal bacteria in the raw greywater with values of 1.05 × 10⁷, 2.75 × 10⁸ and 7.35 × 10⁶ CFU/mL for E. coli, fecal coliforms and enterococci respectively. This finding shows that, greywater could be a risk for human health and must be treated before being reused. Previous studies have reported high contents of E. coli, ranging from 10¹ to 10⁸ CFU/100 mL in greywater in urban and peri-urban environments [16] [17]. These concentrations of fecal bacteria remain lower than those of greywater in rural areas, which are of the order of 10⁸ CFU/100 ml for E. coli and 10⁹ for fecal coliforms [7]. However, the values obtained in our study are higher than those reported by the [15], which were in the order of 2.6 × 10⁶ and 3 × 10⁵ respectively for Enterococci and E. coli obtained from raw greywater from French cities.

The variations observed between the concentrations of fecal bacteria in urban and rural areas on the one hand, and in an African city and an European city, on the other hand, suggest that the behaviour of the inhabitants has an important influence on the characteristics of the greywater.

All treated greywater exhibited slightly neutral mean pH values compared to the raw greywater (Table 2). These values are in compliance with the WHO reuse in irrigation standards (pH of 6.5 to 8) [18]. Based on these values, if reused in irrigation, the treated greywater could promote bacterial growth in the soil since most bacteria prefer neutral or slightly alkaline conditions, around 6.5 to 8.5 [19]. The contents of organic matter from the biofilters using sand and granitic

( ) = standard deviation.

gravel of (1 - 2 mm) were the lowest. Low concentrations were also obtained in many studies using sand (1 - 4 mm) and gravel (2 - 4 mm) biofilters for greywater treatment in Palestine [20] [21] and Jordan [22]. From the organic matter point of view, sand (1 - 2 mm) and granitic gravel (1 - 2 mm) biofilters produced treated greywater with high reuse potential.

The effect of the type of filter media on the RE of organic pollutants was evaluated by considering the data obtained with sand (particle size of 1 - 2 mm, 2 - 4 mm), granitic gravel (particle size of 1 - 2, 2 - 4 and 4 - 6 mm) and lateritic gravel (particle size of 2 - 4 and 4 - 6 mm). The results obtained with the different type of biofilters operating at the same particle size were compared. It appeared that sand and granitic gravel of 1 - 2 mm in size significantly reduced BOD₅, COD and SS compared to the other filter materials used in this study. In most cases, the sand obtained the highest REs (for example, for sand and granitic gravel at the particle size of 1 - 2 mm, COD removal was 67.35% and 55.67% respectively) (Table 3). However, the statistical analysis showed that there was no

For a given parameter, values with different letters are significantly different.

significant difference between the REs of the organic pollutants by the different filter media (sand, granitic and lateritic gravel) operating at the same particle size (1 - 2 mm, 2 - 4 mm or 4 - 6 mm) for all of the tested parameters (COD, BOD₅ and SS).

The effect of the variation of the particle size for a given filter media was evaluated using the granitic gravel (particle size of 1 - 2 mm, 2 - 4 mm and 4 - 6 mm), sand (particle size of 1- 2 mm, 2 - 4 mm) and lateritic gravel (particle size of 2 - 4 mm, 4 - 6 mm). For all of the tested parameters (COD, BOD₅, SS), the biofilters using particle size of 1 - 2 mm mainly obtained the highest REs regardless of the filter media (sand, granitic or lateritic gravel). For BOD₅, the statistical analysis showed that, within a given biofilter media, the variation in particle size did not have a significant effect on the RE of the different biofilters. For COD, when granitic gravel was used as filter materials, it appeared that the decrease in particle size had a significant effect on the REs. Indeed, the value obtained with the smallest particle size (1 - 2 mm) was significantly higher than that of 2 - 4 mm and the largess particle size (4 - 6 mm). For SS, granitic gravel with a particle size of 1 - 2 mm on the one hand and 2 - 4 mm on the other hand obtained significantly higher organic pollutants REs than that of 4 - 6 mm in size.

When we consider the effect of the combination of the type and size of the filter media, some significant differences appeared. For BOD₅, the REs of sand biofilters 1 - 2 mm in size in the one hand and 2 - 4 mm in size in the other hand were significantly higher than that of lateritic gravel biofilter 4 - 6 mm in size. Regarding COD removal, the results obtained with sand biofilter 1 - 2 mm in size were significantly higher than that obtained with granitic gravel biofilters 2 - 4 mm and 4 - 6 mm in size. In addition, the sand biofilter of 2 - 4 mm in size exhibited significantly higher REs compared to the granitic and lateritic gravel biofilters, both of 4 - 6 mm in size. For SS, the REs of sand biofilters 1 - 2 mm and 2 - 4 mm in size and granitic gravel biofilter 1 - 2 mm in size were significantly higher than that of lateritic and granitic gravel biofilters, both of 4 - 6 mm in size.

Our results are in compliance with those of [30] who demonstrated that the particle size affects the optimal absorption of pollutants when wood chip and peanut shell biofilters were used to remove organic matter from domestic wastewater. The differences in RE between the same media filters of different particle sizes could mainly be attributed to the heterogeneous shape and compaction, with unpredictable particle size organization and distribution leading to preferential pathways [31]. The RE of sand biofilter (1 - 2 mm) and granitic gravel (1 - 2 mm) were directly related to their particle sizes but also to their long infiltration time. Indeed, according to [11], if a fine material is used, the retention time of wastewater in the filter is longer, which often leads to higher REs; however, the water takes longer time to infiltrate and the potential for clogging increases as shown in our study where after four weeks, both sand 1 - 2 mm and 1 - 2 mm granitic gravel biofilters were clogged. These results could be explained by a significant reduction in the porosity of the filter media due to organic loading [32] and biofilm formation [33]. The opposite effect is observed when coarser filter media are used: it leads to late clogging but lower REs. Indeed, with lateritic and granitic gravels (4 - 6 mm in size) any clogging was noticed, but the REs obtained were low (Table 3).

The assessment of the influence of the type of filter media and particle size on the removal of the microbial pollution was conducted based on the determination of the contents of E. coli,fecal coliforms and enterococci in the raw and pre-treated greywater.

When E. coli was considered, we noticed that whatever the type of the filter media, we obtained a decrease in the RE if the particle size was increased. The highest RE was observed using granitic gravel of 1 - 2 mm in size with a value of 2.38 log.u (Table 4). Overall, granitic gravel (1 - 2 mm) seemed to show better E. coli removal compared to sand (1 - 2 mm). However, the statistical analysis did not show any significant difference between the filter media when compared to each other, nor between the particle sizes of the same filter media.

For fecal coliforms, it is noted that an increase in the particle size of the filter materials resulted in a decrease in the RE. From 1 to 4 mm in size, the REs obtained by the different materials were sand > granitic gravel > lateritic gravel. The highest RE was obtained with sand at 1 - 2 mm in size with a value of 1.77 log.u. The REs obtained with sand and granitic gravel of 1 - 2 mm in size were both significantly higher than that of granitic gravel of 4 - 6 mm in size.

With enterococci, regardless of the type of the filter material, we noticed an increase in the REs when the particle size of the filter media was decreased. The highest RE was obtained with sand of 1 - 2 mm in size. The REs obtained with sand of 1 - 2 and 2 - 4 mm in size were both significantly higher than that of lateritic and granitic gravels of 4 - 6 mm in size.

For a given fecal indicator, REs without any letter in common are significantly different.

From our results, it appeared that:

1) Media of fine texture remove fecal bacteria better than coarse texture media. A similar trend was reported by [34] where unsaturated vertical filters with sand (1 - 4 mm) and gravel (4 - 8 mm) removed 1.9 and only 0.8 log.u of E. coli respectively.

2) The removal of fecal bacteria from greywater using biofilters depends on the interaction between the type of the filter media, the particle size and the type of microorganism. The reasons for high bacterial removal in sand biofilters were previously attributed to physical filtration, biofilm growth and accumulation of solids on the top surface which lead to a decrease in pore space, thereby, increasing the removal capacity of bacteria [27] [35].

The results of the direct observation of the biofilters showed that the early or late clogging depends on the type and the particle size of the filter media (Figure 4). Indeed, sand and granitic gravel of 1 - 2 mm in size were clogged the fifth weeks of operation, followed by sand with particle size of 2 - 4 mm the ninth week after operation. The lateritic gravel of 2 - 4 mm in size was clogged after the tenth week. In contrast, the lateritic and granitic gravels with a grain size of 4 - 6 mm, did not show any signs of clogging throughout the study (13 weeks). The clogging of biofilters was previously attributed to the accumulation of organic particles and the microbial growth that lead to the development of biofilms in the filter materials [36]. Indeed, kitchen greywater has the highest levels of organic substances and suspended solids [16]. This explains that the filters of fine to medium texture (1 - 2 mm and 2 - 4 mm) that clogged early, presented the accumulation of organic matter on their surfaces (Figure 4). The early clogging could be explained by the fact that fine media pile up and leave less free space for the water to percolate, which increases the time required for infiltration. In addition, organic matter fits into the pores, reducing them, increasing the infiltration time until clogging. In fact, the materials of fine texture showed an increase in

the infiltration time until clogging (33 times i.e. from 43 to 1440 min for sand 1 - 2 mm and 35 times i.e. from 58 to 2048 min for granitic gravel 1 - 2 mm (Figure 5). Besides, the coarse texture materials with grain sizes of 4 - 6 mm had an almost constant infiltration time during the experiment (20 min). The results of direct observation are correlated with the data on infiltration time assessment (Figure 5): the longer the infiltration time, the greater the degree of clogging.