By Jahangir Chowdhury
Filtration is the most common treatment process used in all water treatment plants to remove suspended particles and pathogens. A roughing filter uses media that is more than 2 mm in diameter. These can be of horizontal-flow, up-flow or down-flow.
Horizontal-flow roughing filtration (HRF) for pre-treatment has been applied successfully in many facilities for raw water of medium to low turbidity. The main characteristics of the process are its horizontal flow direction and three to four gravel packs, ranging from coarse to fine grain. HRF technique is a natural purification process and no chemicals are necessary.
Installation of such filters requires only local resources such as construction material and manpower. Furthermore, no mechanical parts are required to operate or clean the filters.
Subscribe to our Newsletter!
The latest environmental engineering news direct to your inbox. You can unsubscribe at any time.
Due to high deposits storage capacity and considerably longer service period, the HRF became a promising alternative pretreatment method prior to slow sand filters (SSF) and rapid sand filters (RSF). Its horizontal-flow direction allows construction of shallow and structurally simple filters.
However, at high raw water turbidity (> 200 NTU) or relatively high filtration rates (> 1 m/h), the turbidity removal efficiency and filter run length of HRF drop drastically. Consequently, it needs a large surface area. Removal of sludge by flushing remains critical where there is organic matter in raw water.
Collected silt in many cases is sticky, and is difficult to flush out. This makes HRF technology unattractive for application in semi-urban and urban areas. Therefore, further modification of HRF was carried out to overcome these drawbacks.
To overcome some of HRF’s limitations and to allow higher filtration rates and longer filter run time, application of a coagulant to the influent water before HRF is an interesting process modification. This new process modification is called direct horizontal-flow roughing filtration (DHRF). Investigation shows that the size of DHRF can be reduced substantially and hence capital cost may be lowered.
The feasibility of DHRF, so far, has proved to be an attractive and promising pretreatment process at bench scale. Its application on full plant scale prior to SSF and RSF in semi-urban areas and the small towns of developing countries needs to be investigated. The construction cost of DHRF was estimated to be about two times lower than a conventional flocculation-sedimentation system.
Main Features
Direct horizontal-flow roughing filtration is a promising pretreatment process with an appropriate level of technology.
The filter bed is composed of a simple box filled with gravel of different sizes varying from coarse (20 mm) to fine (5 mm) over two to three compartments in the direction of flow. The compartments at the inlet and outlet site of the filter box establish an even flow distribution and maintain a certain water level across the filter bed.
The total length of the filter bed ranges from 9 m to 12 m. Height is limited to 1.5 m to allow comfortable manual cleaning. The width of the filter box depends on the filter capacity and normally varies from 2 m to 5 m.
Water flows in a horizontal direction through the filter bed. Treated water is collected by an outlet chamber and then discharged over a weir. The bulk of solid matter is retained in the filter bed.
With progressive accumulation of solid matter in the media filter, efficiency decreases, so periodic cleaning must be done. After several months or years of service, periodic cleaning may not be sufficient to reestablish filter efficiency. Therefore, filter media must be manually removed, washed and replaced.
Removal mechanisms
With respect to the removal mechanism of the first coarse grain (~ 20 mm diameter) compartment of DHRF, several researchers have reported that sedimentation is the main removal process and in-pore flocculation also plays an important role. The filter acts as a multiple-plate settler and provides a very large net surface area on which sludge may accumulate.
Sludge accumulates on top of grains and grows into dome-shaped aggregates with advanced filtration time. The size and shape of the heaps formed is controlled by their slope stability. Some of the small heaps drift toward the filter bottom.
This drifting partly regenerates the filter efficiency of the upper filter layer, but gradually clogs the filter bed from the bottom.
Besides the vertical downwards drift of the sludge there is also a horizontal drift in the direction of flow. This drifting process is advantageous as the removal capacity of the upper layers is restored to a certain extent. (See Figure 1)
Preliminary studies on the removal mechanism of the subsequent compartment with medium size grains (~ 8 mm diameter) showed characteristics similar to deep-bed filtration. It can be said to act as a number of vertical filtration layers perpendicular to the flow direction.
In the first coarse compartment, the turbidity decreases gradually along the length, in a pattern similar to the particle removal pattern in a sedimentation basin. In the finer media compartment, a progressive clogging front is developed with filter run time, which is assumed to be similar to vertical deep bed filtration.
Experimental results
Six filter runs under different process conditions were considered. Performance of the first and second compartments is shown in Figure 2. Experimental data used in this Figure have been calculated as weighted average value during the “working stage” of the filter.
The bigger size particles present in the influent are settled by sedimentation (as in settling basin) in the first compartment of DHRF. Influent of the second compartment (i.e., effluent of the first compartment) consists of particles of comparatively smaller size and of less settling velocities.
However, due to higher head loss (~ 3 cm/m compared to 1 mm/m in the first compartment) results in higher G – value and in-pore flocculation, the particles may subsequently grow in size again in the second compartment.
The earlier breakthrough with higher filtration rates and influent turbidity occurs because of higher shear stresses, which cause particle dislodgement. The increase of influent turbidity provides a higher concentration of particles, which contributes to higher floc formation.
The pore size or the storage capacity of the filter reduces faster. Therefore, after a certain filter run time, the particles move horizontally with the bulk of the flow instead of settling. This causes an early breakthrough.
Although the breakthrough or filter run time reduces tremendously with increasing filtration rates, the cumulative specific deposits value was found to be nearly the same.
The accumulation of deposits for various filtration rates and influent turbidity show that approximately the same amount of deposits accumulate (~ 25 g/l in first and ~ 11 g/l in second compartment) in the pores of the grains before breakthrough.
Head loss increases with filter run time. However, its development in the first compartment is negligible and, in the second compartment, it is of minor importance for filter operation as it will only be within a few centimetres.
Modelling the first compartment
In this model, plain sedimentation is the main process mechanism in the gravel bed of the first coarse compartment of DHRF. The filter bed acts as a multiple plate settler. The model applies sedimentation theories from a multiple plate settler system. To determine the removal efficiency of the filter, the results of quiescent column settling tests are incorporated in the model.
The assumption of the model is that the total volume of the model settler is equal to that of the first filter compartment. The thickness of the settler plates and the spaces between the plates are equal to the average grain size dg and the initial porosity ε₀ of the filter respectively.
Modelling the second compartment
A mathematical model was developed to assess the main process mechanisms and to predict the removal efficiency along the filter length of the second compartment of DHRF with medium sized grain (~ 8 mm diameter). In the model, three different approaches are made to assess the dominant process mechanism.
The first approach treats the process as sedimentation, as in the case of the first compartment. Deep bed filtration is assumed in the second approach. A combination of sedimentation and deep bed filtration is considered in the third approach.
According to the optimization of parameters of the first and second compartment of DHRF, a tentative design guideline for initial turbidity ranges from 200 NTU to 400 NTU at the influent of the first compartment.
Conclusions
The process modification done on HRF by adding coagulant prior to filtration has proved to yield good effluent quality with longer run time. Higher filtration rates (3 m/h – 5 m/h) can be applied in DHRF compared to HRF (0.5 m/h – 1 m/h).
Because of the relatively low construction and operation cost, DHRF could be an appropriate pretreatment technology. Findings also indicate that it has good potential to substitute for the flocculation-sedimentation process and be used before the slow and rapid sand filtration.
The dominant particle removal mechanism in the first compartment of DHRF with coarse grain (~ 20 mm diameter) is sedimentation, while in the second compartment with medium sized grains (~ 8 mm diameter) it is deep bed filtration.
As filtration has much higher particle removal efficiency than sedimentation, the removal efficiency of the second compartment is higher (~ 96 %) than the first (~ 75 %) for the same filter dimensions.
For filter runs with medium (200 FNU) and highly (400 FNU) turbid water and filtration rate varying from 3 m/h to 7 m/h, the optimum filter length for the first compartment is determined to be 4 m and for the second compartment is 2 m to 3 m.
Jahangir Chowdhury, MScEng, P.Eng., is with Aecom. This article appears in ES&E Magazine’s April 2019 issue.
