In conventional water treatment stations, the filter cleaning is performed with the use of filtered water. To save water and obtain higher production, the use of polystyrene (PS) beads has been proposed as a granular filter element because it is a granular element with a low specific mass. By being lightweight, this material requires a lower water velocity during backwash. The PS beads were applied in a descending rapid filter and compared to a conventional sand and anthracite filter, and its hydraulic performance was evaluated during the backwash with air and water interspersed. Although it presents a high fluidity, with lower rates (compared to conventional filters) of backwash, this fact does not necessarily represent an economy of backwash water, because it requires more time for cleaning. It was also observed that there is an optimal value for the removal of particles collected during the filtration without loss of material.

## INTRODUCTION

Sand, or a combination of sand and anthracite (dual layer), is widely used as a media filter. However, different materials may be used in order to improve the performance of the filters. Materials with different densities are usually used in order to enhance the filtration runs. Knudsen (1980) established relationships between different grain sizes and different densities of materials commonly used in descending rapid filtration (sand, anthracite and garnet), but did not mention anything about these backwashes. Davies & Wheatley (2012) also compare some media filters and highlight the influence of the shape of the grain in the filtration process.

Some authors have used alternative materials with different densities, to improve the performance of the filters, as did Farizoglu et al. (2003) and Sierra Filho et al. (2005), who used pumice as a filter element.

Other authors have used polymers for waste and water treatment to improve the performance of the filter. Pearce & Jarvis (2011) used polymers to treat effluent. Fabris et al. (2008) reported on the water treatment plant of Skullerud, Norway. This technology applies ETA descending direct filtration with triple-layer filters (two layers of different plastic media and a layer of sand). However, none of these authors mentions the use of such filter elements of different densities in order to increase the effective water production.

During the process of cleaning the filters, the backwash rate must be large enough to wash (remove) the material captured by the bed, but not so high that the filter material is pushed out of the filter (loss). To prevent loss of the filter material, it is necessary to determine the bed expansion that occurs with the fluidized material.

Akgiray & Saatçı (2001) showed that the Ergun equation is valid for fixed and expanded beds. However, in a fluidized bed, the headloss becomes constant.

The headloss of the filter material can be measured and/or evaluated by testing the material and expansion depending on the characteristics of the material.

This study was conducted in order to acquire a new filter element in which there is the prospect of reducing the consumption of water to backwash the filters.

Granular polystyrene (PS) was chosen because it has low specific gravity, similar to sand, a commercially available inert particle. This material was characterized by the authors as shown in Table 1 (Schöntag & Sens 2014).

Table 1

Comparison between characteristic of polymeric spherical PS, sand and anthracite (averaged)

CharacteristicsPolystyrene granulesSandAnthracite
Minimum size 0.50 mm 0.4 mm 0.6 mm
Maximum size 1.20 mm 2 mm 2 mm
Effective size 0.66 mm 0.62 mm 0.85 mm
Uniformity coefficient 1.36 1.69 1.65
Media grain diameter 0.87 mm 1.38 mm 0.95 mm
Porosity 0.387 0.550 0.665
Density 1.046 g/cm3 2.610 g/cm3 1.350 g/cm3
CharacteristicsPolystyrene granulesSandAnthracite
Minimum size 0.50 mm 0.4 mm 0.6 mm
Maximum size 1.20 mm 2 mm 2 mm
Effective size 0.66 mm 0.62 mm 0.85 mm
Uniformity coefficient 1.36 1.69 1.65
Media grain diameter 0.87 mm 1.38 mm 0.95 mm
Porosity 0.387 0.550 0.665
Density 1.046 g/cm3 2.610 g/cm3 1.350 g/cm3

The characterization of the material was possible through mathematical models of backwash presented by Turan et al. (2003), Naseer et al. (2011) and Mohammed et al. (2013) that simulate the behavior of the material during backwash. The authors found through mathematical models the need for special care during the process because the element is of low density (Schöntag & Sens 2014). Although low velocities are sufficient to maintain fluidization, resulting in smaller volumes of water during the cleaning process, it is necessary to increase shear during flushing in order to break down the flocs formed by the impurities retained during filtration. Furthermore, the rate of wash (water) needs to be big enough to drag the particles collected during filtration.

To evaluate the behavior of the material during backwashing, it was tested on an experimental scale. This study compares a single-layer filter with PS beads with another double layer of sand and anthracite filter, for they are the type of conventional descending filter most used today.

## MATERIALS AND METHODS

The characteristics of the materials that make up the filters were extracted and are presented in Table 1 (Schöntag & Sens 2014). The technology employed was direct filtration treatment (coagulation/filtration). Thus, two filters were constructed: a filter using expandable PS as a filter medium and one double layer of a sand and anthracite filter (A + A).

The filters were constructed according to Figure 1. The operation rate was 8.66 m/h (208 m3 m−2 d−1) (constant) with variable hydraulic load. The raw water source of Peri Lagoon (Santa Catarina, Brazil) was conducted for the rapid mixing unit, which received the coagulant PAC (poly aluminum) in a dose of 1.08 mg Al3+/L. The velocity gradient of the fast mixing chamber was 1,200 s−1. These parameters were defined in bench tests (jar-tests). After coagulation, the water was sent to the descending filters, which ran simultaneously.

Figure 1

Installation scheme of pilot filters.

Figure 1

Installation scheme of pilot filters.

Filtered water samples were collected once a minute in the first 30 min for recovery analysis. In recovery analysis, turbidity and apparent color rates were plotted on graphics in relation to time.

Piezometers were used for determining the timing of filtration runs. The filtration runs were completed when the headloss reached 2 metres. Ten filtration runs were performed, which reached different durations and established the average and standard deviation of the results.

The backwash of the filters was carried out after the end of the runs. Air and water were introduced in ascending order; the backwash processes were different, as shown below.

Backwash of A + A filter: the speed was at 66 m/h (1.1 m/min) for an expansion of 40%. The flow of compressed air was between 48 and 50 NL/min with a pressure of 8 kgf/cm2.

Backwash of PS filter: the minimum fluidization velocity was 0.89 ± 0.044 m/h (Schöntag & Sens 2014). To achieve an expansion of 40%, a velocity of water at 6.6 m/h (0.11 m/min) was necessary, which being very small did not carry dirt particles out of the filter. Thus, a speed of 22.8 m/h (0.38 m/min) was used for a growth of 200%. The flow of air used in the backwash of the PS filter was approximately 20 NL/min.

Ten backwash processes were performed, which obeyed the following criteria for the two filters: in the first 5 min, the air was introduced, even after an interval 1–2 min (for no loss of material), and 10 min of water immediately followed. The process was repeated, complying with a range of 1–2 min.

During the cleaning, the wash water samples were collected every minute and analyzed for turbidity, and in that way the backwash time could be optimized.

In a second step, other speeds of the backwash water were used in the PS filter (besides the 10 already described) such as:

1. 12.8 m/h – causing an expansion of 100% within 10 min of application;

2. 17.0 m/h – causing an expansion of 150% within 10 min of application;

3. 19.1 m/h – causing an expansion of 100% within 5 min of application;

4. 29.3 m/h – causing an expansion of 200% within 5 min of application.

These speeds were introduced after the application of air for 5-min and 2-min intervals, so there was no loss of material as in the previous process.

All these speeds of backwash were performed three times, where the backwash water samples were collected and the mean and standard deviation were extracted.

For all conditions, the value of the actual production was calculated and compared to the values of the A + A filter considering all optimized processes.

Once the duration of the filtration run and backwash processes are determined, it is possible to calculate the effective production according to Crittenden et al. (2011). The authors state that filters can be designed to achieve effective production of 95%, but that the rate of filtration must be at least 200 m3/m2. This study investigated the idea that with a different filter media, there could be a decrease in the volume of backwash.

In a second step, the thickness of the bed was reduced to 68 cm (Figure 2), whereas the depth of the bed was achieved in the region of 50 cm as in Schöntag et al. (2015). With less thickness, there is a lower initial headloss. Thus, the thickness of the bed was reduced in order to verify the behavior and productivity.

Figure 2

Schematic of the PS filter: (a) provision of piezometers; (b) thickness of the bed.

Figure 2

Schematic of the PS filter: (a) provision of piezometers; (b) thickness of the bed.

Five filtration runs were performed with this configuration. The same operation controls and quality of the filters were applied. The filtration runs also reached different durations. The average results for each filter were compared by statistical hypothesis testing, as well as the percentage of removal water quality parameters as: turbidity, apparent color, conductivity, residual aluminum and removal of cyanobacteria.

Five backwash processes were used that obeyed the following criteria: within the first 3 min, the air was added at 20 NL/min, and after an interval of 1–2 min (no loss of material), 3 min of water (21.4 m/h) was applied, then in sequence 1 min of air, a 2-min interval and 4 min more of water, reaching an expansion of 182%.

During cleaning, the samples of wash water were collected every minute and analyzed for turbidity, and in that way the backwash time could be optimized.

## RESULTS AND DISCUSSION

In the first tests, the filtration run of the A + A filter had an average duration of 9.6 h, while the PS filter had an average duration of 5.8 h, or almost half the time to the same filtration rate as seen in Figure 3, which relates to the run length with a headloss.

Figure 3

Average and deflection of headloss of PS and A + A filters during filtration runs.

Figure 3

Average and deflection of headloss of PS and A + A filters during filtration runs.

This is because the porosity of the beads through the PS is less than the porosity of the sand and anthracite filter, as verified and shown in Table 1. Furthermore, the large amount of cyanobacteria (Cylindrospermopsis raciborskii) present in the water of Peri Lagoon causes the penetration (depth filtration) to be low, around 50 cm. The presence of two filter elements of larger effective diameters (according to Table 1) causes the impurities to spread and reach greater depths, at around 65 cm (Schöntag et al. 2015).

The recovery period was observed. The mean and standard deviation of the apparent turbidity and first 30 min of the color filter are presented in Figure 4.

Figure 4

Mean and standard deviation for turbidity and apparent color in a recovery period of 10 filtration runs for the A + A and PS filters.

Figure 4

Mean and standard deviation for turbidity and apparent color in a recovery period of 10 filtration runs for the A + A and PS filters.

The recovery time is the time it takes for the filter to stabilize the characteristics of the filtered water. With 95% confidence, the recovery period of the PS filter and the A + A filter shown in Figure 5 are the same, i.e., 12 min. It was also observed that the quality of the water treated by the PS filter was slightly better than the water of the A + A filter, from the initial 4 min of filtration, becoming equivalent in a few moments, until the end of the initial 30 min (Schöntag et al. 2015).

Figure 5

Curve of turbidity obtained during backwash of the double layer of the anthracite and sand filter (mean values and standard deviation of a sample of 10 runs).

Figure 5

Curve of turbidity obtained during backwash of the double layer of the anthracite and sand filter (mean values and standard deviation of a sample of 10 runs).

The backwash process was performed at the end of every run. Initially, the entire process of flushing lasted 37 min (air + water).

The results from the A + A filter are shown in Figure 5. The turbidity of the washing water stabilized after 3 min of flushing. This time, the results were sufficient for effective cleaning without wasting the treated water.

For the PS filter, it was observed that an increase of 40% would not be sufficient to remove the dirt particles. In order to increase the drag, the superficial velocity of the water was increased to 10.8 m/h (0.18 m/min), i.e., an expansion of 100% after the application of air. It was found that the particles were entrained with the expansion, but not enough to exit the filter. Thus, the superficial velocity again was increased to 22.8 m/h (0.38 m/min), i.e., an expansion of 200%. Only this superficial velocity of water was sufficient to remove collected particles along the filter, to the exit point of the filter.

The backwash process was achieved with an expansion of 200%. The results can be seen in Figure 6. For this speed, the material takes 10 min to reach this level of growth, which requires longer to clean the filter.

Figure 6

Turbidity curve obtained during the backwash filter with PS beads and an increase of 100%, 150% and 200% in 20 min with backwash water, i.e., 12.8 m/h, 17 m/h and 22.8 m/h, respectively (mean values and standard deviation of a sample of 10 runs).

Figure 6

Turbidity curve obtained during the backwash filter with PS beads and an increase of 100%, 150% and 200% in 20 min with backwash water, i.e., 12.8 m/h, 17 m/h and 22.8 m/h, respectively (mean values and standard deviation of a sample of 10 runs).

When the time of backwash water is optimized, i.e., 3 min for the A + A filter and 10 min for the PS filter, the values of effective production may be calculated according to Table 2.

Table 2

Effective production of the sand and anthracite filter and PS beads, average values obtained from five filtration runs with optimal results

A + A filterPS filter
tF (h) 9.6 VF (m/h) 8.67 UVCF (m3/m283.2 tF (h) 5.8 VF (m/h) 8.7 UVCF (m3/m250.3
tR (h) 0.05 VR(m/h) 66.7 UVR (m3/m23.3 tR (h) 0.16 VR(m/h) 22.8 UVR (m3/m23.8
trec (h) 0.18 Velocity = VF (m/h) 8.67 Uvrec (m3/m21.6 trec (h) 0.18 Velocity = VF (m/h) 8.7 Uvrec (m3/m21.6
Total time (h) (tF + tR + ar + interv.) 9.76 UVCF – UVR (m379.9 qEF (m/h) 8.2 Total time (h) (tF + tR + air + interv.) 6.1 UVCF – UVR (m346.5 qEF (m/h) 7.6
Effective production (%) 93.8% Effective production (%) 90%
A + A filterPS filter
tF (h) 9.6 VF (m/h) 8.67 UVCF (m3/m283.2 tF (h) 5.8 VF (m/h) 8.7 UVCF (m3/m250.3
tR (h) 0.05 VR(m/h) 66.7 UVR (m3/m23.3 tR (h) 0.16 VR(m/h) 22.8 UVR (m3/m23.8
trec (h) 0.18 Velocity = VF (m/h) 8.67 Uvrec (m3/m21.6 trec (h) 0.18 Velocity = VF (m/h) 8.7 Uvrec (m3/m21.6
Total time (h) (tF + tR + ar + interv.) 9.76 UVCF – UVR (m379.9 qEF (m/h) 8.2 Total time (h) (tF + tR + air + interv.) 6.1 UVCF – UVR (m346.5 qEF (m/h) 7.6
Effective production (%) 93.8% Effective production (%) 90%

Note: VF is the filtration velocity (m/h); VR is the velocity of backwash (m/h); tF is the length of filtration run, (h); tR is the time of backwash (h); trec is the time of recovery, (h); UVCF is the volume during the run; UVR is the volume of the backwash of the filter; UVRec is the volume used in the recovery period; QEF is the effective rate of filtration, (m/h).

Although the PS filter has a backwash flow rate lower than that used in backwashing the A + A filter, the backwash time is longer and the filtration run is less, which leads to the lowest effective production.

To improve the values of effective production of the PS filter, changes were made. The exit point of the washing water was reduced to 20 cm above the point of expansion of 100%.

Therefore, other backwash speeds were applied. The results of the turbidity of the backwash water for different rates of expansion can be seen in Figure 6 and Table 3.

Table 3

Average values for each backwash speed applied

Speed of backwash water (m/h)Degree of expansion (%)Time (min)Duration of run (h)Turbidity of final backwash (NTU)Qualitative evaluation of cleaningEffective production (%)
29.3 200 6.3 24.5 Good 89
19.1 100 14.7 Good 93
17 150 10 21.6 Good 92.4
12.8 100 10 6.3 15.9 Good 93
50 10 6.6 49.1 Bad 93.8
Speed of backwash water (m/h)Degree of expansion (%)Time (min)Duration of run (h)Turbidity of final backwash (NTU)Qualitative evaluation of cleaningEffective production (%)
29.3 200 6.3 24.5 Good 89
19.1 100 14.7 Good 93
17 150 10 21.6 Good 92.4
12.8 100 10 6.3 15.9 Good 93
50 10 6.6 49.1 Bad 93.8

The cleaning was extended by 20 min of water application. However, it was noted that this was not necessary, because it can be reduced to only 10 min, as it was observed that with such a degree of expansion, there was good cleaning.

Other values with a lower speed were tested, with 7 m/h and an increase of 50% in 10 min, but the cleaning was not good, reaching only 49.1 NTU at the end of the backwash, and beyond that the flakes adhered to the grains and the heavier particles were sediments at the bottom of the bed.

It was observed that PS beads take longer (other than the conventional) to achieve a given degree of expansion. This is due to their spherical and light features. However, the backwash process is long and the cleaning needs a long time, which is not favorable, because the effective production will be lower. To assist the process in reducing the time, the backwashing speeds were set in such a manner that the material reached the same degree of expansion in half the time, i.e., instead of 10, 5 min was used, as shown in Figure 7.

Figure 7

The average curves of expansion (%) time (min) of the PS with speeds of 19.1 and 29.3 m/h.

Figure 7

The average curves of expansion (%) time (min) of the PS with speeds of 19.1 and 29.3 m/h.

Thus, three backwash processes were performed for each speed after the filtration runs. The recovery period of the filter was the same. The results of the turbidity of the backwash water for those expansion velocities are shown in Figure 8.

Figure 8

Curve of turbidity of backwash water of the filter with PS beads, with expansions of 100% and 200% in 5 min of backwashing with water, i.e., at 19.1 m/h and 29.3 m/h, respectively (mean values and standard deviation of a sample of three runs).

Figure 8

Curve of turbidity of backwash water of the filter with PS beads, with expansions of 100% and 200% in 5 min of backwashing with water, i.e., at 19.1 m/h and 29.3 m/h, respectively (mean values and standard deviation of a sample of three runs).

For each speed, a degree of expansion was obtained at any given time, with results of backwash quality and effective production as shown in Table 3.

According to Table 3, the best results were achieved at speeds of 19.1 m/h or 12.8 m/h, or 100% expansion, with good cleaning, reaching levels of turbidity in the order of 15 NTU and an effective production of 93%. Compared to the actual production obtained by the A + A filter of 93.8%, although this value is close, it is not satisfactory.

To analyze the performance of PS beads, it was decided to reduce the thickness of the bed to 68 cm. The filtration runs had an average duration of 8.1 h for the PS filter. The filtration runs were terminated when the headloss reached 2 metres. The optimal backwash of the PS filter with thickness 68 cm was performed making a total of 15 min of backwashing, as shown in Figures 9(a) and 9(b).

Figure 9

(a) Expansion curve at a time of 5 min and a speed of 21.4 m/h in the 68-cm bed. (b) Turbidity obtained during filter backwashing with PS beads, an increase of 182% in 5 min of backwashing with water, i.e., 21.4 m/h (mean values and standard deviation of a sample of three runs).

Figure 9

(a) Expansion curve at a time of 5 min and a speed of 21.4 m/h in the 68-cm bed. (b) Turbidity obtained during filter backwashing with PS beads, an increase of 182% in 5 min of backwashing with water, i.e., 21.4 m/h (mean values and standard deviation of a sample of three runs).

The characteristics of the water of the PS filter with 68 cm bed thickness were practically the same as with 97 cm of bed (Schöntag et al. 2015) as shown in Table 4. Transpassivity was not observed, as shown in Figure 10.

Table 4

Characteristics of water filtered by the PS filters with 97-cm and 68-cm beds (average percentage removal)

ParametersPS filter (97 cm)PS filter (68 cm)
Turbidity (uT) 75% 75.9%
Apparent color (uH) 67% 70.9%
Conductivity (μS/cm) (average value) 84.0 85.0
Residual aluminum (mg/L) (average value) 0.1 0.12
Removal of cyanobacteria (Cylindrospermopsis raciboskii50% 53.3%
ParametersPS filter (97 cm)PS filter (68 cm)
Turbidity (uT) 75% 75.9%
Apparent color (uH) 67% 70.9%
Conductivity (μS/cm) (average value) 84.0 85.0
Residual aluminum (mg/L) (average value) 0.1 0.12
Removal of cyanobacteria (Cylindrospermopsis raciboskii50% 53.3%
Figure 10

Behavior of the headloss and turbidity of the filtered water by the PS filter throughout the filtration run.

Figure 10

Behavior of the headloss and turbidity of the filtered water by the PS filter throughout the filtration run.

In addition, the recovery time of the PS filter was lower, about 6 min. Thus, the effective production rose to 95.2%.

By reducing the thickness of the bed, there was an increased length of filtration run and this amounted to effective production without the need for increments in the cleaning process. It is believed that this occurred because the loss of the initial load was low. In addition, there was a larger spread of impurities throughout the filter bed, causing it to take longer to reach 2 metres, which can be seen in Figure 11, for different initial headlosses and the inclination of headloss curves. Importantly, the material was quite homogeneous (equal diameter) and did not suffer backwash stratification, as does sand or anthracite. In a conventional sand and anthracite filter, after repeated washings, the smaller grains stay on top of the filter bed, which does not occur with PS beads.

Figure 11

Curve of headloss along the filtration run, the different thicknesses of the PS bed of 97 and 68 cm.

Figure 11

Curve of headloss along the filtration run, the different thicknesses of the PS bed of 97 and 68 cm.

Considering the effective production of 95.2%, a production curve of the PS and A + A filters was developed according to the duration of the run, as seen in Figure 12.

Figure 12

Curve of production (L/h) according to the duration of the filtration runs for the A + A and PS filters, considering the effective production of 93.8% and 95.8%, respectively.

Figure 12

Curve of production (L/h) according to the duration of the filtration runs for the A + A and PS filters, considering the effective production of 93.8% and 95.8%, respectively.

Note that the production curve of the PS filter is slightly larger than the A + A filter, in the initial production, but with increasing filtration time the tendency is to stabilize production at the same value. However, the volume produced by the PS filter, although small, is larger.

Thus, the long-term production was estimated. In real terms of production volume in a year or 8,760 h of production, the PS filter produces 70,165 m3/m2, while the A + A filter produces 70,030 m3. The difference is low, i.e., 135 m3/m2. However, one should consider the size of the filter, which in this case has an area of 0.04 m2. If estimating the results for a filter of 25 m2, for example, the production increase would be 3,375 m3, and the revenue gain would be $7,434.00. In fact, the PS beads are more expensive than sand. The price of PS beads is approximately$0.92 (1 dollar) per kilo. For 17 m3 (0.68 m × 25 m2), it would take 18 tons of PS, which would cost $16,450.00. As for the sand, for the same 17 m3, 45 tons would be required, or approximately$14,884.60 making a difference of \$1,565.40. That is, at 2.5 months of use of the PS beads, the material is paid for.

## CONCLUSIONS

The aim of this study was to determine whether the PS beads, as a filter element, represented a reduction of the volumes of water used in backwashing filters, because they are lighter and thus require lower speeds to liquefy the material. Other studies have shown the ability to backwash this material (Schöntag & Sens 2014).

Some configurations of backwash processes were tested and compared to the backwash processes of a conventional sand and anthracite filter. It is important to note the cleaning process of conventional filters, not only how they can be optimized, because often the time and flow rates are agreed upon without concern for verification, causing wastage of treated water with excessive cleaning and increased recovery time of the filter. Therefore, once optimized values of conventional backwash filters are achieved, these can be compared to the backwash configurations suggested for the PS filter.

First, the PS filter was backwashed equally to the A + A filter. However, there were no advantages in regards to effective production because despite using a lower washing speed, it required more time to drag the particles. In addition to the filtration run, the time for this filter was considerably lower, at 5.8 h on average, when the thickness was 97 cm.

Other speeds and degrees of expansion were tested with less time, but showed no advantages.

After these tests, it was decided to reduce the thickness of the filter bed to 68 cm. The length of the filtration run, which was on average 5.8 h, increased to 8.1 h, or 40% higher. The cleaning process of air and water was applied and optimized, for a total of 15 min to process. Furthermore, there was a decrease in the recovery time of the filter for an average of 6 min. With this, the effective production was 95.2%. At first glance, this advantage may seem small, or almost inexpressible, not justifying the use of the more expensive material. However, as seen by estimating the utilization of the material of larger filters in the long-term, it can be advantageous. In accordance with the results observed in this experiment the higher production and consequently higher sale of the water brings a profit of 0.20% per annum. This profit pays the material cost (more expensive than the sand and anthracite) in 2 months. Moreover, being lightweight has other benefits, such as for building reservoirs or decreasing the components and connections necessary to perform the backwash.

Some observations should be taken into consideration for the use of this material as a media filter, such as the process of shearing and the exit point of the wash water. The shear or detachment of particles adhered to the media filter must be performed before the introduction of the backwash water, as there is a risk of loss of the filter material during the process. The exit point of the wash water should not get much beyond 20 cm above the upper limit of the expansion, so that the backwash water can flush out dirt particles. However, this elevation of the expansion is not instantaneous due to the characteristics of the filter material, so there is a higher effective production of water. The time for the backwash water should be optimized according to the characteristics of the filter and the raw water to be treated.

## RECOMMENDATIONS AND CLOSING REMARKS

For the PS beads to be used advantageously as a filter element, it is necessary for other types of cleaning systems, such as dual air injection into different sites with one injection below and another above the expanded filter material, to be tested for the output of the filter particles, similar to flotation systems. Another possibility is to use another type of raw water, because it is believed that the system was impaired because of the water features of Peri Lagoon.

## ACKNOWLEDGEMENTS

The authors acknowledge the support of the FINEP/HABITARE, FAPESC/PRONEX and CNPq (National Council of Technological and Scientific Development).

## REFERENCES

REFERENCES
Akgiray
Ö.
Saatçı
A.
2001
A new look at filter backwash hydraulics
.
Water Supply
1
(
2
),
65
72
.
Crittenden
J.
Trussell
R.
Hand
D.
Howe
K.
Tchobanoglous
G.
2011
Water Treatment Principles and Design
.
MWH and John Wiley & Sons
,
Hoboken, NJ
.
Davies
P. D.
Wheatley
A. D.
2012
.
Water Science and Technology
66
(
12
),
2779
2784
.
Fabris
R.
Chow
C. W. K.
Drikas
M.
Eikebrokk
B.
2008
.
Water Research
42
(
15
),
4188
4196
.
Farizoglu
B.
Nuboglu
A.
Yildiz
E.
Keskinler
B.
2003
Performance of pumice as a filter bed material under rapid filter
.
Filtration and Separation
40
(
3
),
41
46
.
Knudsen
P.
1980
The effect of media selection on filtration performance
.
Process Engineering – Sydney Australia
8
(
4
),
41
43
.
Mohammed
R.
Abualhail
S.
Lu
X. W.
2013
.
Desalination and Water Treatment
51
(
22–24
),
4768
4778
.
Naseer
R.
Alhail
A.
Xi-Wu
L.
2011
Fluidization and optimum backwashing conditions in multimedia filter
.
Research Journal of Applied Sciences, Engineering and Technology
3
(
11
),
1302
1307
.
Pearce
P.
Jarvis
S.
2011
.
Water and Environment Journal
25
(
2
),
200
207
.
Schöntag
J. M.
Sens
M. L.
2014
.
Desalination and Water Treatment
(June)
.
Schöntag
J. M.
Sens
M.
Hynmo
F.
Pizzolatti
B.
V.
2015
.
Journal of Water Process Engineering
5
,
118
126
.
Sierra Filho
A.
Mattos
A.
Calò
F.
2005
Meio Filtrante à base de Pomes e tela Termoplástica – Alternativa para a filtração de água potável
.
Turan
M.
Sabah
E.
Gulsen
H.
Celik
M.
2003
Influence of media characteristics on energy dissipation in filter backwashing
.
Environmental Science & Technology
37
(
18
),
4288
4292
.