Parametrização das operações de filtração e retrolavagem de um filtro de areia comercial

Abstract

The sand filter has two independent but complementary modes of operation, known as filtration and backwashing. Studies show that a range of factors interfere with these operations, making correct use of the equipment difficult. Thus, divided into two general parts, this research addresses technical and scientific aspects regarding the definition of filtration and backwash parameters for a Hidrosolo commercial sand filter model FA07. The first part presents a general introduction to the topic under study with the objectives and hypotheses. The second part consists of three scientific articles. In the first article, the objective was to use artificial neural networks to model pressure loss and filtration removal efficiency. Experimental tests covered three grain sizes of silica sand (fine, medium, and coarse) and four filtration rates (20, 40, 60, and 75 m h⁻¹). Three artificial neural networks (ANN1, ANN2, and ANN3) were built to define the duration of the filtration cycle, the pressure loss, and the filtered volume. ANN1 was based on modeling the pressure loss through the filter, while ANN2 and ANN3 were based on suspended solids in the effluent. The variables used were the filtration rate, the filtered volume, grain size, suspended solids, and pressure loss. ANN1 performed well with a 5-4-1 architecture. Increasing the filtration rate reduced the duration of the filtration cycle, while increasing grain size prolonged it. Networks ANN2 and ANN3 performed unsatisfactorily and were therefore discarded. In the second article, the objective was to evaluate the influence of the drainage system’s constructive design on the hydraulic behavior of the backwash process. Two filter bed heights were used (0.15 and 0.30 m) and three drain models: cylindrical drain — corresponding to the manufacturer’s original model; conical; and a proposed drain — corresponding to the result of development using computational fluid dynamics. Bed expansion was similar between the original and proposed drains for both bed heights. The greatest bed expansion efficiency was observed for the conical drain at the 0.15 m height. The proposed drain showed lower pressure loss with an empty filter, but values were similar when filled with sand. The minimum fluidization velocity was the same for the original and proposed drains (18.0 m h–1 ), and higher for the conical drain (24.0 m h–1 ). In the third article, the aim was to evaluate the influence of the drainage system's constructive design, the backwash duration, and the expansion of the filter layer on the layer's cleaning efficiency. Two levels of filter bed expansion (25% and 50%), three backwash durations (three, eight, and fifteen minutes), and two drainage system designs were evaluated: 1) cylindrical drain; and 2) proposed drain. The proposed drain model outperformed the original model, especially at 25% expansion. Overall, 50% expansion provided better performance for both drain models, with the proposed drain standing out. Cleaning efficiency was lower in regions near the filter walls, regardless of the drain model, the layer expansion level, or the backwash duration. In general, there was a trend of increased total bed cleaning efficiency with longer backwash duration. The turbidity behavior of the backwash water showed that the greatest variation occurs in the first minutes of the backwash, indicating that most solids are removed at the beginning of the process. Finally, the proposed drain had lower water and energy consumption, with higher efficiency rates compared to the commercial model