Twenty-four laboratory glass filters with a nominal diameter of 25 mm, an internal diameter of 19 mm and a length of 40 cm were used. The lower part of the filter was supported by a metal mesh to prevent clogging of the filter outlet and loss of the filter medium, which was 15 cm high [5,23,24]. To provide a hydraulic load to ensure that water was present throughout the filter medium, the effluent was collected through a silicone hose attached to the bottom of the filter. The other end of the hose was located above the upper part of the filter medium [5]. The filtration columns were placed in a support structure, which allowed proper operation of the filters (Fig. 1).

Figure 1 Source: Los Autores

The experimental unit was fed by a multi-partition distribution system. Gravity flow was distributed to each laboratory filter through silicone hoses, which had flow control valves at their ends to allow the flow to be distributed through constant dripping to the filters. This distribution system was supplied from a temporary storage tank in which the water level was kept constant to avoid variations in the inlet flows to the filters.

The laboratory-scale study used filtered water from the Puerto Mallarino WTP, which had undergone coagulation, flocculation, clarification and downstream sand and anthracite filtration [25,26]. Table 1 shows the characteristics of the filtered water.

Table 1

Source: Los Autores

The characteristics of the GAC and sand filter media used in the test are given in Table 2.

Table 2

Source: Los Autores

The configurations correspond to the height proportions of the filter media that consist of GACVEG and sand, GACMIN and sand and only sand. These configurations are presented in Table 3 and were analyzed in triplicate.

Table 3

Source: Los Autores

Filtered water from the WTP, was distributed to each laboratory filters at a constant filtration rate of 61 m³/m²d for a flow rate of 12 ml/min and a filtration time of six hours [5, 24]. The follow-up variables are shown in Table 4 and were measured every 5 minutes during the first half hour of filtration. Subsequently, they were measured at 10 minute intervals for the remainder of the first three hours and every 15 minutes in the last three hours of the test. The physical and chemical variables were measured according to the guidelines established in the standard method [27].

Table 4

Source: Los Autores

Turbidity is an indirect parameter that indicates the potential of microbiological risk in drinking water. Low values of turbidity in filtered water (<0.30 NTU) indicate greater efficiencies in the removal of protozoa (Giardia and Cryptosporidium) during filtration and favor the elimination of bacteria and viruses during disinfection [15]. Fig. 2 shows the behavior of the turbidity during the test for the two types of carbon (vegetable/mineral) and sand with their respective filter media configurations.

Figure 2 Source: Los Autores

Fig. 2 shows that all of the media configurations resulted in similar turbidity behavior as a function of time. The turbidity values are below the initial turbidity value obtained in the first filtration stage. As indicated by Wiecheteck et al. [19] and Silva et al. [20], double filtration provides greater turbidity removal than conventional filtration.

Ninety-seven percent of the data for all configurations resulted in turbidity values less than 0.3 NTU, which is the threshold value of the WHO [2] and the EPA [15] before disinfection to eliminate chlorine-resistant pathogens and ensure the effective elimination of Giardia, Cryptosporidium, and other material. All of the turbidity values measured for each configuration in this study complied with the national regulations for drinking water (<2 NTU) [7].

Graese et al. [28] founded turbidity <0.2 NTU in the effluent with GAC filtration, which is consistent with this study. The median values for the filters with the GACVEG configurations were between 0.199 and 0.238 NTU, those for the GACMIN filters were between 0.206 and 0.236 NTU, and those with sand had the lowest value of 0.181 NTU.

The median values obtained for each configuration show that the configuration with the sand filter resulted in the best reduction of turbidity; however, all configurations with GAC provided effective turbidity removal.

Although the GAC configurations showed lower turbidity removal than the sand filter, these results demonstrate that the implementation of activated carbon ensured the quality of the final effluent.

Organic matter is an important constituent of water that affects the performance of treatments in drinking water processes and the quality of drinking water. As a result, it requires the extensive use of coagulants, oxidants and disinfectants in addition to being a precursor to the formation of disinfection byproducts [29]. Several studies have used the UV₂₅₄ absorbance test as a parameter to evaluate the efficiency of the GAC filtration process for the removal of dissolved organic matter, which is capable of absorbing UV light [6, 20, 29]. Figure 3 shows the behavior of UV₂₅₄ during the tests for the two types of carbon (vegetable/mineral) and sand with their respective filter medium configurations.

Fig. 3 shows that the configurations with GACMIN and GACVEG were below the UV₂₅₄ absorbance value reported in the first filtration stage, whereas the sand filter remained very similar to the initial UV₂₅₄ value. Similar results were obtained by other studies, which concluded that sand does not effectively remove compounds associated with organic matter [30,31].

Figure 3 Source: Los Autores

Median values of UV₂₅₄ between 0.010 and 0.019 cm^-1 were obtained with the GACVEG configurations, median values between 0.007 and 0.008 cm^-1 were obtained with the GACMIN configurations, and a median of 0.032 cm^-1 was obtained with the sand. The results indicate that the sand did not significantly affect the reduction of organic matter because its values oscillated near the initial absorbance of 0.034 cm-1, and the configurations with GACMIN had a greater effect. The GAC filters clearly yielded a reduction of organic compounds as measured by UV₂₅₄. Therefore, the use of granular activated carbon as a filter media is recommended not only for reducing odor and flavor but also in the adsorption of organic compounds [9].

Silva et al. [20] demonstrated the removal of UV₂₅₄ using GAC in double filtration, and Pham et al. [22] achieved removal efficiencies of organic matter greater than 68%. These results were similar to those found in this study; the removal percentages for the GACVEG configurations ranged from 46% to 72%. The GACMIN configurations resulted in the highest removal percentages, between 75% and 78%, whereas the sand configuration only removed 6% of the organic matter.

The results indicate that filters with GAC in double filtration technology can remove organic substances, which reduces the possible risk associated with the formation of disinfection byproducts [17,32].

Additionally, Kim and Kang [33] found that replacing sand filters with dual-medium GAC-sand filters represents an ideal choice for the removal of organic matter compared to turbidity removal. This was confirmed by the results of this study, which showed small differences in the turbidity decrease between the GAC and sand configurations, whereas the configurations of the GAC filters eliminated organic matter more effectively than the sand filter alone.