Optimum conditions for high-speed solid–liquid separation by ballasted flocculation

Coagulation–sedimentation is an important solid–liquid separation process (Lee et al. 2014; Jiang 2015; Dao et al. 2016; Khettaf et al. 2021a, 2021b) in water treatment and has been adopted in several water treatment plants. To remove suspended solids from water, a coagulant is used for efficient solid–liquid separation. In general, coagulants widely used in water treatment are aluminum-based inorganic coagulants, including aluminum sulfate and polyaluminum chloride (PAC) (Sinha et al. 2004; Wang et al. 2009). In particular, PAC has excellent treatment capacity compared to other coagulants (Sinha et al. 2004; Wang et al. 2009). Coagulation plays important removing colloidal and fine particles from surface water during water treatment (Wang et al. 2009). However, the flocs formed only with inorganic coagulant include water in the flocs, and the amount of water increases as the size of the flocs increases, resulting in a small density, low settling velocity, and low strength of the flocs (Matsui et al. 2003; Sinha et al. 2004; Wang et al. 2009). Therefore, optimizing the floc size and the strength is a key in coagulation–sedimentation to limit the settling velocity of coagulated flocs formed by inorganic coagulants such as PAC. In a water purification plant for processing large amounts of tap water, the area of the settling basin must be large to reduce surface loading. However, it is difficult to secure a large installation area for settling basins in urban areas due to land restrictions (Kumar et al. 2016). Therefore, to date, there is a need to develop a solid–liquid separation technique that is faster than the conventional coagulation–sedimentation.

Ballasted flocculation (trade name: ACTIFLO) was developed in the 1990s as an innovative solid–liquid separation technique (Desjardins et al. 2002). This process uses microsand and polymer flocculants to accelerate the sedimentation process by forming large flocs with a high specific density (Lapointe & Barbeau 2016). Thus, it makes floc sedimentation faster than coagulation–sedimentation based on conventional inorganic flocculants (Brahmi et al. 2018). Ballasted flocculation has been employed in various solid–liquid separation processes, including water purification (Lapointe & Barbeau 2018; Kumar et al. 2019; Qasim et al. 2019; Murujew et al. 2020).

In the early 2000s, ballasted flocculation was used as a solid–liquid separation process in water purification systems for low-turbidity water (turbidity of 10 nephelometric turbidity unit (NTU) or less) (Desjardins et al. 2002; Lapointe & Barbeau 2016, 2018). On the other hand, in monsoon regions, such as Japan, surface water, which is the raw water for tap water, is highly turbid due to heavy rainfall, exceeding 500 turbidity unit (TU) (unit as kaolin standard) most times (Saito et al. 2013). Such high turbidity of raw water must be considered to maintain high-standard tap water. As a pretreatment process for such high-turbidity water, ballasted flocculation has been studied and employed in some facilities in practice. Considering water whose turbidity significantly varies, it is difficult to achieve suitable conditions for ballasted flocculation. Based on experience, such conditions have been adjusted. However, the treatability in underconditioning and overconditioning, high turbidity, and extensive ballast addition is unclear. In fact, onsite adjustment of dosage conditions is still an essential task at water treatment plants to ensure the quality of the treated water.

In this study, we prepared three turbid water samples composed of kaolin clay (kaolin concentrations: 20, 200, and 500 mg/L) and examined the factors that affect the processability and floc settling velocity during ballasted flocculation. In addition, the mechanism of ballasted flocculation was investigated based on the results of the water treatment under short, optimum, and excess dosage conditions to provide a stable supply of treated water even for large fluctuations in the turbidity of raw water.

Kaolin (FUJIFILM Wako Pure Chemical Co., Osaka, Japan) was used as suspended clay particles to prepare a 10-g/L slurry-like stock solution. PAC (concentration as Al₂O₃, 10–11%; basicity, 45–65%; density, ≥1.19; Asahi Kasei Advance Co., Tokyo, Japan) was used as an inorganic flocculant and was diluted 10-fold immediately before use to prepare a stock solution. The concentration of sodium hydrogen carbonate (NaHCO₃, FUJIFILM Wako Pure Chemical Co.), as an alkali, was adjusted to 100 mg/L in distilled water, and the kaolin stock solution was added to obtain turbid water. In this study, turbidity was evaluated by an integrating sphere type turbidimeter with a turbidity standard solution made of standard kaolin as specified in the Japanese Waterworks Law. The turbidity is defined as that the turbidity equivalent to 1 degree of TU when 1 L of purified water contains 1 mg of standard kaolin. Kaolin turbidity of 1.00 is 0.57 as the equivalent turbidity NTU. Three levels of kaolin turbid water (low: 20 mg/L; medium: 200 mg/L; high: 500 mg/L) were prepared to investigate the effect of turbidity loads. Silica sand (SiO₂ content: 98%; specific gravity: 2.63; particle size: 100–300 μm; Mikawa Keiseki Co., Okazaki, Japan) was used as microsand. The silica sand is industrially available with uniform specific gravity and high SIO₂ purity. The particle size was selected to be dispersed in the beaker under the stirring conditions during the jar test. Six polymer flocculants (anion: AP335PWS, AP120PWS, AP410PWS, Mitsubishi Chemical Co., Tokyo, Japan; nonion: NP800PWS, Mitsubishi Chemical Co.; SS-500PWG, HYMO Co., Tokyo, Japan; cation: KP1200H, Mitsubishi Chemical Co.) were examined to obtain the best suited for ballasted flocculation (Table 1). The stock solution of the polymer flocculant was dissolved in distilled water immediately before use to adjust the content to 0.1%. In addition, 1-mol/L hydrochloric acid (HCl) and 1-mol/L sodium hydroxide (NaOH) were added to adjust the pH of the solution.

To investigate the relationship between PAC dosage and the pH of the solution, we conducted conventional coagulation–sedimentation tests using a jar test apparatus (MJ-8, Miyamoto Co., Osaka, Japan). First, PAC was added to a turbid water sample and rapidly mixed (150 rpm, the estimated gradient velocity 201/s) for 3 min. The PAC dosages were as follows: low-turbidity water 20 mg/L, PAC 0–50 mg/L; medium-turbidity water 200 mg/L, PAC 0–100 mg/L; high-turbidity water 500 mg/L, PAC 0–120 mg/L. The pH of the solutions was controlled by adding HCl or NaOH. Then, the mixtures were slowly stirred (40 rpm, 28/s) for 15 min. After coagulation, the mixtures were allowed to settle for 10 min, after which 100 mL of the supernatant water was gently collected as the treated water using a syringe. The turbidity of the untreated and treated water was measured using a turbid meter (PT-200, Nittoseiko Analytech Co., Kanagawa, Japan) to determine the purification efficiency. The pH and electrical conductivity of the treated water were measured using a pH bench meter (LAQUA F-74, 9615-10D, HORIBA Co., Tokyo, Japan).

To ensure the uniform stirring and mixing of the suspensions obtained using PAC, microsand, and polymer flocculant, four baffle plates (width: 15 mm; height: 140 mm) were fixed inside the beaker (Okoro et al. 2021). The PAC stock solution was added to 1,000 mL of the sample in a 1,000 mL-beaker set in the jar tester and rapidly stirred (200 rpm, 302/s) for 1 min. Thereafter, microsand (0–30 g/L) was added directly under rapidly stirred (200 rpm), and after 2 min, the predetermined dosage (0–10 mg/L) of the polymer flocculant was added to the samples. Thereafter, the mixture was further stirred for 3 min at 140 rpm. After the flocculation, the upper treated water layer and the floc layer formed an interface, and the upper interface of the floc layer settled uniformly. Therefore, immediately after the stirring, the settling velocity of the flocs was measured as the settling depth at the interface between the floc suspension and the supernatant water per unit time (cm/s). The mixture was allowed to stand for 3 min, after which 100 mL of the supernatant water (treated water) was gently collected using a syringe.

Coagulation of suspended particles by PAC is the most important for ballasted flocculation.

Figure 5(a) shows the effect of microsand and flocculant dosages on the treatment efficiency and settling velocity of the low-turbidity water (20 mg/L kaolin). For the sample with no polymer flocculant, the turbidity of the treated water was high (8–14 TU), and the floc settling velocity was extremely low (0.003–0.008 cm/s) for all microsand dosages. The PAC microflocs could not be enlarged because they did not aggregate with the microsand without a polymer flocculant. When both microsand and polymer flocculant were added, the floc settling velocity markedly increased. As the dosage of the polymer flocculant increased, the settling velocity increased. Owing to the addition of a polymer flocculant, the microflocs and microsand aggregated and agglomerated to form large flocs with high tare weight. The optimum amount of microsand was 3.0 g/L, at which the turbidity of the treated water was less than 0.6 TU, and the floc settling velocity was the highest (1.34 cm/s). The average standard settling velocity for the ACTIFLO process is 1.11 cm/s (Desjardins et al. 2002), which is comparable with the value obtained herein. With the further addition of microsand above the optimum dosage, the settling velocity decreased. Sediments were observed when excess microsand was added, confirming that microsand was not included in the flocs. The unaggregated microsand contacted with the flocs due to rapid mixing and fragmented the flocs. The decrease in the settling velocity is attributed to the fragmentation of the flocs. When the dosage of the microsand was 10 g/L, the turbidity of the treated water was more than 1 TU for all flocculant dosages, which is attributed to the residual floc fragmentation. For the low-turbidity water, the highest floc settling velocity was achieved at microsand and polymer flocculant dosages of 3.0 g/L and 2.0 mg/L, respectively, with 97% turbidity removal efficiency.

Figure 5(b) shows the effect of microsand and polymer flocculant dosages on the treatment efficiency and floc settling velocity of the medium-turbidity water (200 mg/L kaolin). Similar to the case of low-turbidity water (Figure 5(a)), the turbidity of the treated water was high, and the settling velocity was extremely low when no polymer flocculant was added, even with the addition of microsand. However, the floc settling velocity markedly increased with the flocculant and microsand dosages, reaching a maximum value of 1.2 cm/s at 10 g/L of microsand. With flocculant dosages of 0.25–2.0 mg/L and microsand of 1–10 g/L, the turbidity was less than 0.6 TU, indicating high treatment efficiency. However, excessive microsand (20–30 g/L) significantly reduced the floc settling velocity and increased the turbidity of the treated water, and unaggregated microsand was observed in the sediments.

Figure 5(c) demonstrates the effect of microsand and polymer flocculant dosages on the treatment efficiency and floc settling velocity for the high-turbidity water sample (500 mg/L kaolin). The treatment efficiency and floc settling velocity showed a trend similar to that of the medium-turbidity water, and the floc settling velocity significantly increased with the addition of microsand and polymer flocculant, reaching a maximum at 10-g/L microsand. However, even when the dosage of the polymer flocculant was 2.0 mg/L, the settling velocity was 1.0 cm/s, which is lower than that of the low- and medium-turbidity water samples. At high flocculant dosages of 7.0–10 mg/L, the floc settling velocity reached 1.7 cm/s with 10 g/L of microsand. In the case of the high-turbidity water, a high flocculant dosage was required to obtain a large amount of microflocs with microsand. The optimum amount of microsand and the decrease in the floc settling velocity with the excessive addition of microsand can be explained by the same mechanism as the cases of low and medium-turbidity water.

In an actual continuous treatment process of ballasted flocculation system called ACTIFLO (https://www.veoliawatertechnologies.com/en/technologies/actiflo), most of the sand is recovered from the settled sludge by a hydrocyclone and reused again. In our practical management experience for a ballasted flocculation system, the amount of sand replenishment is less than 1/1,000 of the set sand concentration in the system. The dosing cost (inorganic coagulant, polymer flocculant, and sand replenishment) is 0.015–0.029 USD/m³, depending on the water quality and other conditions. In the case of inorganic coagulation only, the cost is 0.007–0.022 USD/m³. In the actual system of ballasted flocculation in water treatment plants, it is also necessary to determine the aluminum (especially when pH is high) and polymer persistence in the final treated water, as well as the actual carryover of microsand.

We investigated the optimum ballasted flocculation treatment conditions for three water samples of different turbidity based on the variations in the turbidity of the treated water and floc settling velocity with the microsand and flocculant dosages. On the basis of the obtained results, we examined the factors affecting the ballasted flocculation treatment. The microflocs, which comprise suspended solids and PAC, did not aggregate with microsand when no polymer flocculant was added, and the size of the flocs did not increase; thus, high-speed separation can be achieved by ballasted flocculation. Microflocs are constituents of large flocs, which include microsand, and a polymer flocculant promotes the flocculation of microflocs and microsand by cross-linking. Since the formation of microflocs depends on the turbidity of the water, which is determined by the concentration of suspended particles and the dosage of PAC as a coagulant, microsand has an optimum amount when a polymer flocculant is added. The appropriate dosage of PAC varies with the concentration of suspended particles, and the amount of microflocs also varies; thus, the optimal amount of microsand is determined accordingly by turbidity. Excess microsand could not aggregate with the flocs and fragmented the flocs during agitation. Thus, the residual turbidity increased, and the floc settling velocity decreased. To achieve high-speed ballasted flocculation, the following are required: (1) a suitable dosage of coagulant to form microflocs of suspended solids (turbidity) as the constituent materials of flocs; (2) the maximum amount of microsand that can be included in the flocs; (3) a polymer flocculant required to maintain the strength of the flocs.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.