3.2. Physicochemical Analysis of Rainwater
The incorporation of impurities present on the collection surface into the rainwater occurs due to the direct contact of rainwater with the roof.
As the building where the two filters for the rainwater harvesting treatment were installed is located approximately 100 m from a highway with a high traffic load, it was verified through the pH data that the rainwater was within the standard for normality in relation to acidity and was not considered as acid rain (i.e., pH < 5, which was not the case).
Box-plot graphs containing median, minimum, maximum, 1st and 3rd quarters, and interquartile range of the parameters evaluated are shown in Figure 6
The filtration materials used in the sand filter had a degree of uniformity equal to 4.9 for the sand and 1.9 for the gravel, making them highly uniform. Sezerino [34
] obtained coefficients of uniformity equal to 5.70 for sand and 1.89 for gravel in the characterization of filtration materials.
According to Sezerino [34
], a coefficient of uniformity lower than or equal to 5 is recommended for sand. Coelho and Di Bernardo [8
] obtained coefficients of uniformity equal to 2.0 and lower than 1.7 for sand and granular activated carbon, respectively. Brinck [9
] obtained coefficients of uniformity equal to 1.76 and 1.97 for different sands and 1.30 and 1.96 for different anthracites. The lower the value for the coefficient of uniformity, the more uniform the granular material, the deeper the penetration of impurities, and the longer the filtration run time will be.
provides a statistical summary of the results for the physicochemical analysis of the rainwater collected from the sampling points and allows a comparison with the values established in guidelines for water quality, in Brazil, that is, CONAMA Resolution 357/2005 [29
], NBR 15527:2007 [32
], and MS Directive 2914/2011 [31
]. The results were also compared with the values given in the US EPA [30
It can be noted in Table 5
that the average and the standard deviation values for the pH of the rainwater samples collected post-treatment met the standards established by the abovementioned guidelines.
Regarding the ammonia and the nitrite concentrations, for all sampling points, the results remained within the values established by CONAMA Resolution 357/2005 [29
] for a class 2 water body. Although this resolution is used to set limits for contaminants in a different type of water body, it was used herein as a standard for comparison purposes. However, the turbidity values obtained were higher than the values established by NBR 15527:2007 [32
], mainly for the untreated rainwater. This could be due to the proximity of the building to highway BR 277.
Based on the turbidity results, both filters managed to retain particulate matter and therefore reduce the turbidity values. It was verified that the highest turbidity values for the untreated rainwater and the water after passing through the filters were obtained in the samples collected on 28/06/2016 and 13/07/2016. These results occurred after the longest period without rain, which allowed time for the accumulation of particulate matter in the atmosphere, thereby resulting in an increase in the turbidity of the rainwater.
Based on MS Directive 2914/2011 [31
], which deals with water potability, the highest ammonia values were above the limit, while the nitrite values were acceptable.
In the case of nitrate, only the highest value obtained for the untreated rainwater was above the limit of MS Directive 2914/2011 [31
] and CONAMA Resolution 357/2005 [29
]. Both filters were efficient in the removal of ammonia but not in the removal of nitrite. Acceptable levels for ammonia, nitrite, and nitrate are not specified in the US EPA [30
] or NBR 15527:2007 [32
The average values obtained for the pH at all collection points met the limits of the guidelines used for comparison. In relation to the turbidity, only the samples collected after passing through the two filters met the limits of all guidelines. In the case of the untreated rainwater, the average turbidity values met the limits only of CONAMA Resolution 357/2005 [29
The average ammonia, nitrite, and nitrate values were acceptable according to MS Directive 2914/2011 [31
] and CONAMA Resolution 357/2005 [29
]. These parameters are not addressed in the US EPA [30
] and NBR 15527:2007 [32
Honório et al. [35
] carried out a study in the western region of Amazonia, Brazil, in order to evaluate the quality of untreated rainwater. Samples were collected in the towns of Parintins, Itapiranga, Tabatinga, Boa Vista, and Apuí and in the city of Manaus (where one sample was taken in an area covered by vegetation and another in an urban area). For the seven sampling points, average nitrate concentrations in the range of 4.7 to 32.0 mg/L were observed. The highest values were obtained in Manaus, which could be due to the rapid urban development and increase in the use of fossil fuels, given nitrite and nitrate concentrations in rainwater have been attributed to the combustion of fossil fuels [35
Lee et al. [2
] carried out a study on the quality of rainwater in the city of Gangneung, South Korea, considering three main collection points: untreated rainwater, rainwater after flowing off a roof, and rainwater from a reservoir. They obtained the following average values: pH 5.3 and nitrate 2.2 mg/L for rainwater flowing through a roof and pH 7.8 and nitrate 7.6 mg/L for rainwater accumulated in a reservoir.
In this study, the physicochemical parameters of the water in the first flush device were not evaluated. However, in both systems, a 2-mm hole made for their self-cleaning became clogged due to the accumulation of dirt inside the device, verifying the presence of impurities in the first flush water. Silva [36
] also concluded that the first flush devices are more efficient at removing impurities than the filters.
For the sand filter, the removal efficiencies obtained were 13.0% for turbidity, 34.0% for ammoniacal nitrogen, and 10.0% for nitrate. In the case of the membrane filter, the corresponding values were 11.0%, 32.1%, and 13.6%, respectively.
For both filters, considering the average values, the treatment did not lead to significant changes in the pH and nitrite. In relation to pH, this is due to the fact that the values were close to the point of neutrality, and the filters did not reduce the nitrite levels because the concentrations in the untreated rainwater were low.
The sand and gravel filter did not require backwashing during the evaluation period as there was no change in the treatment efficiency. In relation to the membrane filter, the membrane was changed after a period of use of four months as it was worn and needed to be replaced in order to maintain the treatment efficiency.
On evaluating the results, it can be concluded that there were similar levels of efficiency in the rainwater treatment achieved with the sand and membrane filters. However, the sand filter can be considered more suitable for the rainwater treatment based on the maintenance criterion because it did not require any intervention during the period under study, maintaining the same treatment efficiency throughout the experiment.
The membrane filter, on the other hand, did require some maintenance, such as the adjustment of the membrane position and membrane replacement during the experimental period, to avoid a decrease in the treatment efficiency. Thus, considering the aspects of maintenance and treatment efficiency, the sand filter was found to be more suitable for rainwater treatment for the period evaluated in this study.