Lake water from a Shanghai theme park faces restrictive phosphorus discharge limits (≤0.02 mg/L) that are technically challenging. Laboratory bench top and pilot plant testing were conducted. The results show that a coagulation–ultrafiltration (UF) hybrid process is able to reliably achieve the objective, while a coagulation–sand filter process cannot meet the requirement. Therefore, mature technologies such as Veolia Actiflo® and GE Zeeweed® are preferred for full-scale design. In addition, alum performs better than poly alum chloride on phosphorus removal. It was also suggested the efficiency of anion polyacrylamide for the filtered samples was reduced compared with non-filtered samples. Simultaneously, soluble PO4-P was also eliminated by UF resulting from the flocs and precipitates with greater surface area and the agglomeration of residual aluminum ions and orthophosphate. Membrane filtration helps reduce chemical dosage and 6 mg/L Al3+ (alum) concentration was enough to meet the requirement for the influent TP around 0.3 mg/L. Finally, several considerations were given to future scale-up design, such as treatment target of coagulation precipitation system, species of coagulant aid, flocculants, and online analyzers.

INTRODUCTION

Eutrophication has become a major environmental problem in the world. It is has led to increased algal growth, death of fish, increase in sedimentation, decrease in dissolved oxygen, and reduction of water transparency (Hullebusch et al. 2002). Many researchers have attributed this phenomenon to excessive nutrients inputs. Nitrogen and phosphorus are significant limiting factors to control eutrophication, especially phosphorus, as it is better than nitrogen to measure and regulate in preserving an aquatic ecosystem (Correll 1998). Therefore, the US Environmental Protection Agency recommends limiting total phosphorus (TP) to under 0.02 mg/L (Yoon et al. 2004). In addition, a ratio of nitrogen to phosphorus (N:P) above 15:1 has resulted in a highly eutrophic lake (Havens et al. 2001). To prevent eutrophication, a water treatment plant (WTP) is planned to be built to treat and circulate lake water in a Shanghai theme park. The final emission standard for the WTP is ammonia (NH3-N) less than 0.5 mg/L, TP less than 0.01–0.02 mg/L.

Different methods of achieving ultra-low phosphorus were compared with regards to the effectiveness and feasibility. Ultrafiltration (UF) is a type of membrane filtration. The solvent and other dissolved components can pass through the membrane, while particles bigger than filter pores would be detained and finally form the sludge. It is an attractive process due its outstanding characters of small area, high water quality, convenient and automatic operation, and easy maintenance (Laine et al. 2000). It has been widely used in drinking water treatment (Tian et al. 2008), municipal wastewater treatment, industrial wastewater treatment (Zhang et al. 2006), water reuse, and the pre-treatment of reverse osmosis (Chon et al. 2013), among other applications. However, the UF process alone may not eliminate phosphorus to the required level so that an additional phosphorus removal system, like a coagulation process, could be required. Adding a coagulation process not only changes the floc size and the surface electrical properties, but also lowers transmembrane pressure (TMP) and improves the removal rate of pollutants (Ha et al. 2004).

Many researchers have focused on the relationship between membrane fouling and floc characteristics (Wang et al. 2008; Stoller 2009) and the impact of different coagulants on the performance of organics and nutrient removal (Feng et al. 2015). As for phosphorus removal by coagulation–UF process, it is reported to reduce TP concentration from 5.0 mg/L to 0.3 mg/L by adding alum and ferric chloride into secondary effluent (Citulski et al. 2009). Lower phosphorus concentration has been achieved at pilot scale. For example, some researchers have proven the ability to produce effluent TP less than 0.05 mg/L (Benisch et al. 2007; Hugh & Tozer 2007). Hook & Ott (2001) selected six filter products to test TP removal using the Metro Wastewater Treatment Plant effluent as the experimental feed. These results illustrate that the 0.02 mg/L TP target can be obtained using continuously backwashed filter technology in series with high-rate flocculated settling or a membrane filter. DeBarbadillo et al. (2010) also fed four filtering processes (GE ZeeWeed, Blue Water, Veolia Actiflo and Parkson DynaSand) with secondary effluent and verified that all four technologies performed well and all achieved effluent TP lower than 0.03 mg/L.

In the face of higher end-user expectations, stricter regulations, stiffer economic pressure, a more demanding workforce, and increased environmental concerns, more attention has been paid to restrictions on effluent phosphorus concentration. Several regulations in North America have imposed phosphorus discharge limits of 0.1 mg/L and a number of full-scale tertiary filtration plants are using an effluent TP target of 0.05 mg/L during the design stage (Peeters et al. 2010). However, many domestic municipal wastewater treatment plants are just required to comply with the Environmental Quality Standard for Surface Water (GB3838-2002) Class IV (TP ≤ 0.3 mg/L). Even around environmentally sensitive lakes like Taihu in China, the effluent is to meet the Discharge Criteria for WWPTs & Key Industries of Taihu Area (DB32 1072-2207) (TP ≤ 0.5 mg/L). All of these discharges still belong to micro-polluted water and would have a high environmental risk (e.g., eutrophication as a result of excess nutrients) (Hu et al. 2004). Therefore, WTP in a Shanghai theme park is the first plant in China to be required to achieve TP 0.01–0.02 mg/L in a full-scale environment (the total capacity of the WTP is 24,000 m3/d).

The objectives of this paper are to carry out jar and pilot plant testing to determine the optimum coagulant species and chemical dosage to validate the feasibility and reliability of the coagulation–UF hybrid process to meet the phosphorus limit for the full-scale plant design and to make a comparison between sand filter and UF on the performance of phosphorus removal at the pilot scale.

MATERIALS AND METHODS

Feed water

Water quality around the park was continuously monitored from December 2012 to March 2013. The average measurements are shown in Table 1. It indicates that nutrients (nitrogen, phosphorus) are considered to be important limiting factors to meet the emission standard for WTP. A biological aerated filter (BAF) is to be used to mainly remove ammonia, Chemical Oxygen Demand (COD), and Biological Oxygen Demand (BOD5). Reducing TP to 0.01–0.02 mg/L is a technical challenge, requiring jar and pilot plant testing before full-scale design.

Table 1

Feed water quality and emission requirements for WTP

  Unit Influent WTP emission requirement 
pH – 8.1 ± 0.4 6.5–8.5 
CODcr mg/L 21.0 ± 6.0 <20.0 
BOD5 mg/L 2.85 ± 0.25 <6.0 
SS mg/L 18.7 ± 11.8 <10.0 
NH3-N mg/L 1.25 ± 0.99 <0.50 
NO3-N mg/L 1.7 ± 1.3 5.0–10.0 
TN mg/L 3.19 ± 1.53 – 
TP mg/L 0.16 ± 0.04 0.01–0.02 
  Unit Influent WTP emission requirement 
pH – 8.1 ± 0.4 6.5–8.5 
CODcr mg/L 21.0 ± 6.0 <20.0 
BOD5 mg/L 2.85 ± 0.25 <6.0 
SS mg/L 18.7 ± 11.8 <10.0 
NH3-N mg/L 1.25 ± 0.99 <0.50 
NO3-N mg/L 1.7 ± 1.3 5.0–10.0 
TN mg/L 3.19 ± 1.53 – 
TP mg/L 0.16 ± 0.04 0.01–0.02 

The dashes in Table 1 stand for no requirements for the indicator.

Jar testing

In order to make the chemical coagulation system technically feasible and economically rational, jar testing was performed in six 1.0 L plexiglass beakers using a programmable jar-testing apparatus (ZR4-6, Zhongrun Water Industry Technology Development Co. Ltd., China). Twenty seconds of rapid mixing at 400 rpm was followed by 10 minutes of slow mixing at 100 rpm, and then settling for 30 minutes. Coagulant was dissolved first and dosed into the beaker before rapid mixing. During some tests, micro-sand was dosed at the end of slow mixing as a seed for floc formation. After 30 minutes of settling, a sample was collected 2.0 cm below the surface for subsequent measurements of TP and PO4-P.

The goals of the jar testing included: (1) Finding out the different removal rates of alum and poly alum chloride (PAC, 28% aluminum oxide) under the same alum iron (Al3+) concentration. (2) Determining the relationship between chemical dosage and the TP removal rate. (3) Analyzing the 0.22 μm fiber membrane filtration influence on effluent phosphorus. (4) Investigating the impact of the coagulant polyacrylamide (PAM) on phosphorus removal.

Pilot plant

The pilot plant was in operation from June 2012 to August 2012. It was set up in the park with a maximum capacity of 6.0 m3/h. A schematic flow diagram of the pilot plant is shown in Figure 1. Before entering BAF, an equalization tank was prepared for the convenience of diluting or increasing influent loadings. Coagulants were dosed before the pump, which played an important role on rapid stirring and achieving complete mixing. Thus, at the end of the inclined plate precipitation, there were two valves to split the flow into either a sand filter or a membrane filter.
Figure 1

Process flow of pilot plant. Solid line refers to main process pipe.

Figure 1

Process flow of pilot plant. Solid line refers to main process pipe.

The mixing tank was designed for producing violent turbulence after adding coagulant so as to diffuse it evenly into the whole water body in a short time. This was achieved with a mechanical mixer. The effective volume of the mixing tank was 0.05 m3 with a 30 second retention time.

The volume of the flocculation tank was 2.0 m3, resulting in a reaction time of 10 to 20 minutes. Following the flocculation tank was an inclined plate precipitator with up-flow velocity. The velocity was 2.5 mm/s and the area and the height was 0.5 m2 and 0.87 m, respectively.

The UF process in the pilot plant was an external pressurizing membrane with hollow fiber modules. It consisted of influent, membrane cassette, aeration, back-wash, and effluent units. Aerators were laid on the bottom, and a pressure transfer-sensor and vacuum meter were installed on the top to track the change in TMP. Simultaneously, there were two DN40 mud holes in the bottom that needed to be cleaned regularly. The inner diameter of the hollow fiber membrane was 1.0 mm, the outer diameter was 1.6 mm, and the pore size was 0.01 μm with 50 KD cut-off level. The maximum suction pressure was 90 KPa, and 0–80 KPa was recommended. The cassette had a membrane surface area of 490 m2 with tank size of 1,198 × 805 × 2,685 mm (L × W × H). The flux of the membrane was 10–15 L/(m2 h). The operation could be divided into two phases, filtration and backwashing. The backwash cycle took place every 10 minutes and lasted for 15 seconds. The backwash water was discharged into the sludge equalization tank for further treatment. In addition to backwashing, maintenance cleaning and recovery cleaning were conducted periodically according to the manufacturer's recommendations to prevent fouling.

Analysis methods

The pH value was measured using a pH analyzer. CODcr, BOD5, TN, NH3-N in the influent and effluent were analyzed using standard laboratory methods. NO3-N concentration was measured by a segmented flow analyzer (Futura, Alliance). TP and PO4-P concentration were determined by a Mo-Sb anti-spectrophotometer with a minimum detection limit of 0.01 mg/L. A reliable analytical technique was vitally important when applied at this level of concentration that approached the minimum limit. All the sampling bottles were immersed in hydrochloric acid for 24 hours and rinsed three times by purified water and targeted water in the case of suffering significant variation and errors. The fitting degree of standard curve was controlled under 99.95–99.99%. To ensure testing accuracy, blank samples of ultra-pure water and parallel samples were also measured.

RESULTS AND DISCUSSION

Jar testing results

Conventional water clarification processes primarily involve the destabilization and subsequent removal of colloidal suspended solid materials that are not readily removed by gravity settling alone. Usually, a net negative surface charge causes individual particles to repel each other and remain in suspension. To counteract these repulsive forces, a chemical coagulant such as alum, ferric chloride, ferric sulfate, poly-aluminum chloride, lime (CaO or Ca(OH)2), or any other highly charged ionic chemical species is added to reduce the repulsive force. Ferric salt is characterized by good flocculation, but it would color the effluent and have a negative effect on the efficiency of downstream ultraviolet disinfection. For these reasons, it was not considered. Alum and PAC were chosen for jar and pilot plant testing. Two jar tests were performed. The first test was done on December 5, 2011 (Jar testing Phase I) and the second on April 15, 2012 (Jar testing Phase II).

Jar testing Phase I

During Phase I of the jar testing, PAC and alum were dosed to remove phosphorus in the water. It is noted that the unit of PAC and alum concentration were unified into aluminum ion (Al3+). pH, TP, and PO4-P parameters from the test are shown in Figure 2. Taking dosing alum as an example, starting with an initial TP and PO4-P concentration of 0.17 mg/L and 0.13 mg/L, respectively, the residual phosphorus of 0.02 mg/L was nearly achieved with alum 10 mg/L. Theoretically, one mole would precipitate one mole of phosphate. However, many competing reactions and their associated equilibrium constants, and the effects of alkalinity, pH, trace elements, and ligands must be taken into account (Metcalf & Eddy 2004). Therefore, the chemical dosage far outweighed phosphorus reduction in the jar testing.
Figure 2

The relation of pH, phosphorus, and chemical dosage (2011). TP and PO43- concentration are located in the left coordinate, with pH values in the right coordinate.

Figure 2

The relation of pH, phosphorus, and chemical dosage (2011). TP and PO43- concentration are located in the left coordinate, with pH values in the right coordinate.

Adding coagulants lowered the pH values due to the production of hydrogen ions by an alum hydrolysis reaction (Equation (1)). Simultaneously, phosphorus precipitation with aluminum lowers the pH as well, as seen in the following reaction (Equation (2)). Comparing the impact of alum and PAC on pH, a sharper decline in pH was recorded after alum addition. The variation of pH exerted different influences on phosphorus removal. The minimum solubility of AlPO4 occurred at about pH 6.3, and in practical application, the range of pH 5.0–8.0 still worked. As shown in Figure 2, the original pH value was about 8.3 when dosing PAC, being unfavorable for phosphorus removal. 
formula
1
 
formula
2
The removal rate of phosphorus leveled off when the alum dosage reached 10 mg/L. The effluent TP concentration was slightly variable by approximately 0.02 mg/L. When dosing PAC more than 10 mg/L, phosphorus was still reduced to 0.02–0.04 mg/L. It was thus illustrated that alum performed better on the removal of phosphorus, but close attention should be given to pH if dosing alum in future operations.

Jar testing Phase II

The results of the Phase II jar test are shown in Figure 3. It can obviously be seen that there was a strong positive correlation between the chemical dosage and the effluent TP and PO4-P concentration. In general, each chemical achieved a higher removal rate at a comparatively higher phosphorus concentration, thus became lower even when dosing more chemical. This may suggest that the chemistry and kinetics of chemicals species behaved differently at phosphorous concentration near the solubility limit and coagulant overdosing did not effectively improve the phosphorous removal efficiency as other competing reactions may have a more important influence on the phosphorous precipitation under the given thermodynamic equilibrium condition (Peeters et al. 2010). Therefore, it was a big challenge to achieve an ultra-low phosphorus limit. In addition, alum performed better than PAC on the phosphorus removal at the same Al3+ concentration whether the samples were filtered by a 0.22 μm membrane or not. If adding PAM into the samples, the phosphorus removal rate increased due to PAM helping to aggregate the smaller particles together into a larger floc with better settling characteristics. Floc formation was typically accomplished by forming inter-particle polyelectrolyte bridges by using chemical (flocculant aid) polymer. However, the efficiency of PAM was reduced when the samples were filtered by a 0.22 μm membrane. For example, in Figure 3(a) and 3(b), phosphorus reduction increased by more than 50% when adding PAM. Conversely, PAM addition did not result in additional phosphorus removal when using membrane filtration, as shown in Figure 3(c) and 3(d).
Figure 3

The experimental data from jar testing Phase II (2012). (a and b) Four combinations of chemicals, namely, PAC, PAC + PAM, Al2(SO4)3 and Al2(SO4)3 + PAM, were tested to determine the performance of phosphorus removal (PO43- or TP) at the same chemical dosage (Al3+). (c and d) To quantify the membrane filtration impact on phosphorus removal, comparison testings were carried out as well. Taking PAC (0.22 μm) for example, it meant that chemical PAC is dosed with the samples filtered by 0.22 μm membrane. Effluent PO43- and TP concentration were recorded respectively.

Figure 3

The experimental data from jar testing Phase II (2012). (a and b) Four combinations of chemicals, namely, PAC, PAC + PAM, Al2(SO4)3 and Al2(SO4)3 + PAM, were tested to determine the performance of phosphorus removal (PO43- or TP) at the same chemical dosage (Al3+). (c and d) To quantify the membrane filtration impact on phosphorus removal, comparison testings were carried out as well. Taking PAC (0.22 μm) for example, it meant that chemical PAC is dosed with the samples filtered by 0.22 μm membrane. Effluent PO43- and TP concentration were recorded respectively.

From these results, at least 12 mg/L Al3+ concentration (alum) should be dosed to ensure an effluent TP below 0.02 mg/L if water samples are not filtered by 0.22 μm membrane. The same results were also shown in jar testing Phase I. In contrast, if the samples were filtered, 6 mg/L coagulant concentration was enough to meet the requirement. In conclusion, membrane filtration helps enhance the phosphorus removal.

Pilot plant results

As noted in the onsite logs, the flow rate was 2.0 m3/h from June 28 to July 5, 2012, then increased to 5.0 m3/h. Chemical PAC (effective Al2O3 28%) was dosed before July 19, 2012, and then changed to dose liquid alum (effective Al2O3 15.7%). The concentration of the two chemicals is unified as Al3+ in the following figures.

Performance of coagulation–UF process

The coagulant was dosed in the inlet pipe, and from there the flow went into a mixing tank and then a flocculation tank to form greater particles and was finally settled by an inclined plate. TP concentration for the chemical dosing unit (CU), UF influent, and UF effluent is shown in Figure 4. During the pilot plant, the influent TP concentration to the chemical dosing unit (CU) was slightly variable, ranging from 0.21 mg/L to 0.41 mg/L. Effluent TP concentration from the CU, the same as the UF influent, reduced to 0.02–0.08 mg/L, mainly depending on the influent TP. The phosphorus removal rate of the CU was an average of 84.5%. It was further enhanced by the UF unit with a very fine pore size to filter out smaller particles and colloids. 72.2% of samples from the UF effluent were at or below 0.02 mg/L TP. The average effluent TP from the UF was 0.018 mg/L, demonstrating that a coagulation–UF hybrid process has the ability to achieve the TP limit.
Figure 4

TP concentration in the chemical dosing unit (CU) influent, UF influent, and UF effluent (2012).

Figure 4

TP concentration in the chemical dosing unit (CU) influent, UF influent, and UF effluent (2012).

Chemical consumption

Except for the TP concentration, the coagulants’ dosage and flow rate are also shown in Figure 5. During the first 4 days, the influent TP did not vary greatly. The PAC dose was initiated at 4.45 mg/L (Al3+), which resulted in inconsistent achievement of the required effluent quality. Thus, the PAC dosage was increased to 5.92 mg/L. Combined with a reduction in the influent TP concentration, the effluent TP was reduced to 0.01 mg/L. On July 14, 2012 the flow rate was increased to 5.0 m3/h, and the influent TP concentration also increased to approximately 0.4 mg/L, which resulted in the effluent TP fluctuating and exceeding the limit again. Thus the PAC dose was increased to the maximum, 7.41 mg/L. On July 19, 2012 the coagulant was switched to alum. In the following few days, the effluent TP stabilized around 0.01 mg/L at a dosing rate of 5.92 mg/L for the influent TP 0.21–0.37 mg/L. It was nearly the same result as Figure 3(d). Therefore, it was suggested that 0.71 kg/day of Al3+ was the minimum dosage required at the flow rate 5.0 m3/h to meet the effluent limit.
Figure 5

Flow, chemical dosage, influent TP, and effluent TP from June 28 to August 2 (2012). The unit of flow rate (in x line) is m3/h. Two kinds of chemicals were dosed, PAC before July 19 and alum after that.

Figure 5

Flow, chemical dosage, influent TP, and effluent TP from June 28 to August 2 (2012). The unit of flow rate (in x line) is m3/h. Two kinds of chemicals were dosed, PAC before July 19 and alum after that.

Rejection coefficients for phosphorus

UF uses a membrane fine enough to retain colloidal particles, proteins, enzymes, viruses, or large molecules. Most particle phosphorus was removed by precipitation in the CU. UF influent and effluent TP and PO4-P results of the pilot plant are shown in Table 2. P coefficients for influent ranged from 62.5% to 100% (81.1% average). As a result of the detection limits, the UF effluent PO4-P concentrations were nearly the same as the effluent TP, and the ratio PEff increased to 83.3% on average.

Table 2

Rejection coefficients for TP and soluble PO4 (expressed as percentages). The statistical indexes were also observed as the minimum, maximum, average values, and standard deviation

Date TPInf (mg/L) TPEff (mg/L) PO4Inf (mg/L) PO4Eff (mg/L) PInf (%) PEff (%) fTP (%) fPO4 (%) 
2012.06.28 0.08 0.03 0.05 0.02 62.5 66.7 62.5 60.0 
2012.06.29 0.06 0.01 0.05 0.01 83.3 100.0 83.3 80.0 
2012.07.03 0.05 0.01 0.04 0.01 80.0 100.0 80.0 75.0 
2012.07.04 0.04 0.01 0.04 0.01 100.0 100.0 75.0 75.0 
2012.07.05 0.04 0.01 0.03 0.01 75.0 100.0 75.0 66.7 
2012.07.15 0.07 0.03 0.06 0.02 85.7 66.7 57.1 66.7 
2012.07.17 0.05 0.03 0.05 0.01 100.0 33.3 40.0 80.0 
2012.07.18 0.04 0.01 0.03 0.01 75.0 100.0 75.0 66.7 
2012.08.01 0.04 0.01 0.03 0.01 75.0 100.0 75.0 66.7 
2012.08.02 0.08 0.03 0.06 0.02 75.0 66.7 62.5 66.7 
Min 0.04 0.01 0.03 0.01 62.5 33.3 40.0 60.0 
Max 0.08 0.03 0.06 0.02 100.0 100.0 83.3 80.0 
Average 0.06 0.02 0.04 0.01 81.2 83.3 68.5 70.3 
STDEV 0.02 0.01 0.01 0.00 11.72 23.57 13.1 6.70 
Date TPInf (mg/L) TPEff (mg/L) PO4Inf (mg/L) PO4Eff (mg/L) PInf (%) PEff (%) fTP (%) fPO4 (%) 
2012.06.28 0.08 0.03 0.05 0.02 62.5 66.7 62.5 60.0 
2012.06.29 0.06 0.01 0.05 0.01 83.3 100.0 83.3 80.0 
2012.07.03 0.05 0.01 0.04 0.01 80.0 100.0 80.0 75.0 
2012.07.04 0.04 0.01 0.04 0.01 100.0 100.0 75.0 75.0 
2012.07.05 0.04 0.01 0.03 0.01 75.0 100.0 75.0 66.7 
2012.07.15 0.07 0.03 0.06 0.02 85.7 66.7 57.1 66.7 
2012.07.17 0.05 0.03 0.05 0.01 100.0 33.3 40.0 80.0 
2012.07.18 0.04 0.01 0.03 0.01 75.0 100.0 75.0 66.7 
2012.08.01 0.04 0.01 0.03 0.01 75.0 100.0 75.0 66.7 
2012.08.02 0.08 0.03 0.06 0.02 75.0 66.7 62.5 66.7 
Min 0.04 0.01 0.03 0.01 62.5 33.3 40.0 60.0 
Max 0.08 0.03 0.06 0.02 100.0 100.0 83.3 80.0 
Average 0.06 0.02 0.04 0.01 81.2 83.3 68.5 70.3 
STDEV 0.02 0.01 0.01 0.00 11.72 23.57 13.1 6.70 

The effectiveness of UF in the removal of phosphorus was evaluated by the rejection coefficients (Acero et al. 2010), which was defined by the following equations: 
formula
3
 
formula
4
in which, TPInf and PO4Inf represent the TP and PO4-P in the influent, while TPEff and PO4Eff represent the effluent. The observation of the rejection coefficients are also depicted in Table 2. The rejection coefficients for TP and PO4-P were similar with the range of 40.0–83.3% and 60–80%, respectively. The average fPO4 value was a little higher than fTP value. It can be deduced that soluble PO4-P was also removed by the UF membrane even though the molecule size was smaller than the membrane pore size. There were several reasons contributing to the interception of soluble PO4-P. In the pilot plant, the flocs were not completely removed by the inclined plate precipitator and some of these flowed into the following UF reactor. For the large particles with greater surface area, there were saturation differences on the surface possessing higher surface energy, which led to the growth of particles for attaining an equilibrium state (Jiang et al. 2013). The precipitates can also grow easily into new large precipitates through agglomeration of more residual aluminum ions and orthophosphate (in the UF) on the precipitates’ surface (Zhang et al. 2014a, 2014b). Chemical properties, electrostatic effect (Narong & James 2006), and other reasons may help remove soluble phosphorus. On the whole, higher phosphorus removal efficiency was achieved.

Phosphorus removal rate of sand filter

Sand filtration is a simple to operate, low-cost, efficient, and reliable technique and is used successfully to remove contaminants. It is a physical process of filtering out particulates from the water. The type of medium used and its grain size distribution determine how small a particle can be strained out. Coarser sands have larger pore spaces that have high flow-through rates but pass larger suspended particles, while finer sands have lower flow-through rates and pass smaller suspended particles. However, a recent study (Zhang et al. 2014a, 2014b) shows that coarse medium with higher filter layer was preferred, rather than fine medium with lower filter layer. The latter was characterized with more head loss and higher effluent NTU. Simultaneously, in order to achieve the desired flow-through rate of 8.0 m/h, silica sand with a grain size of 0.9–1.2 mm was used to fill the sand filter of the pilot plant.

To compare the performance of the sand filter versus UF, the sand filter was in operation during July 25 to July 31, 2012 (see Figure 6). During this period, the flow was 5.0 m3/h with liquid alum dosed as the coagulant. The influent TP to the sand filter varied from 0.027 mg/L to 0.052 mg/L, less than the maximum UF influent of 0.08 mg/L. All effluent TP of the sand filter exceeded the target value of 0.02 mg/L, ranging between 0.023 and 0.031 mg/L. Moreover, the average phosphorus removal rate was just 16.73%. Even when the coagulant Al3+ was increased from 5.8 mg/L to 6.7 mg/L, the effluent TP was still 0.026 mg/L. Thus, it is validated that a coagulation–sand filter process could not satisfy the ultra-low phosphorus limit.
Figure 6

Correlation of chemical dosage with influent and effluent TP of sand filter (2012).

Figure 6

Correlation of chemical dosage with influent and effluent TP of sand filter (2012).

Full-scale design criteria

The coagulation–UF process was verified to be feasible to achieve the phosphorus limit through jar and pilot plant testing. For the full-scale design, mature technologies like Veolia Actiflo and GE Zeeweed were preferred for their successful application in coagulation and filtration.

The Actiflo process is a compact, conventional-type clarification system that utilizes microsand as a seed for floc formation. The microsand provides a surface area that enhances flocculation and acts as a ballast or weight. The resulting sand ballasted floc displays unique settling characteristics, which allows for clarifier designs with high up-flow rates and short retention times. These designs result in system footprints that are between 5 and 50 times smaller than conventional clarification systems of similar capacity.

GE immersed Zeeweed is a variety of membrane filtration technologies in which forces like pressure or concentration gradients lead to a separation through a semipermeable membrane. It is specifically designed for difficult-to-treat water sources. In addition, GE optimizes Zeeweed to improve flux, achieve greater permeability, reduce energy consumption, and increase membrane lifespan. The specific design criteria for the full-scale design are shown in Table 3.

Table 3

Sizing for 24,000 m3/d Actiflo-UF facilities

Veolia-Actiflo system with chemical dosing followed by dual media filter (Unit1) 
No. of Actiflo clarifiers 
Actiflo design hydraulic load per train (average) 302 m3/h 
Actiflo design hydraulic load per train (minimum) 213 m3/h 
Actiflo design hydraulic load per train (maximum) 640 m3/h 
Microsand diameter ∼150 μm 
Coagulation/Flocculation retention time 2 min 
Total retention time (including settling) 5 min 
Up-flow velocity at settling tank 120 m/h 
No. of dual media filter 
Filter design hydraulic loading rates (average) 11.3 m/h 
Filter design hydraulic loading rates (peak hour) 13.0 m/h 
Filtration area per filter 26.07 m2 
Total filtration area 104.28 
Depth of anthracite 460 mm 
Depth of silica sand 305 mm 
Depth of gravel 75 mm 
GE Zeeweed Membrane Filtration (Unit 2) 
Design hydraulic loading rates 45.0 L/(m2*h) 
Membrane tank dimension (L*W*H) 5.96 m*2.6 m*3.5 m 
Type of membrane Zeeweed® 500 d 
General TMP −90 ∼ 90 KPa 
Module surface area 40.9 m2 
Number of trains 
No. of cassettes per train 
No. of modules per cassette 64 
Total cassettes 
Total modules 576 
Total membrane area, m2 23,558 
Veolia-Actiflo system with chemical dosing followed by dual media filter (Unit1) 
No. of Actiflo clarifiers 
Actiflo design hydraulic load per train (average) 302 m3/h 
Actiflo design hydraulic load per train (minimum) 213 m3/h 
Actiflo design hydraulic load per train (maximum) 640 m3/h 
Microsand diameter ∼150 μm 
Coagulation/Flocculation retention time 2 min 
Total retention time (including settling) 5 min 
Up-flow velocity at settling tank 120 m/h 
No. of dual media filter 
Filter design hydraulic loading rates (average) 11.3 m/h 
Filter design hydraulic loading rates (peak hour) 13.0 m/h 
Filtration area per filter 26.07 m2 
Total filtration area 104.28 
Depth of anthracite 460 mm 
Depth of silica sand 305 mm 
Depth of gravel 75 mm 
GE Zeeweed Membrane Filtration (Unit 2) 
Design hydraulic loading rates 45.0 L/(m2*h) 
Membrane tank dimension (L*W*H) 5.96 m*2.6 m*3.5 m 
Type of membrane Zeeweed® 500 d 
General TMP −90 ∼ 90 KPa 
Module surface area 40.9 m2 
Number of trains 
No. of cassettes per train 
No. of modules per cassette 64 
Total cassettes 
Total modules 576 
Total membrane area, m2 23,558 

CONCLUSIONS

Two jar testing and pilot plant studies demonstrated that a coagulation–UF process was effective at removing phosphorus and able to reliably achieve effluent TP concentrations less than or equal to 0.02 mg/L. A conventional sand filter suffered from a low phosphorus removal rate of 16.73% with chemical coagulation, and it failed to meet the limit.

PAC and alum were chosen to compare the performance of phosphorus removal. The results showed that alum was better than PAC. However, alum had a greater impact on pH values, so it is recommended to carry out real-time pH monitoring in future practices. Simultaneously, chemical achieved a higher removal rate at a comparatively higher phosphorus concentration, thus became lower even when dosing more chemical. The chemistry and kinetics of chemical species and other competing reactions may have more important influence on the phosphorous precipitation. In addition, PAM exerted a larger influence on PO4-P than TP and assisted alum to remove more phosphorus than PAC. The efficiency of PAM for the filtered samples was reduced compared with non-filtered samples.

In the pilot plant, 72.2% of samples from the UF effluent were at or below 0.02 mg/L TP, demonstrating that a coagulation–UF hybrid process has the ability to achieve the TP limit. Soluble PO4-P was removed by the UF membrane even though the molecule size was smaller than membrane pore size. The flocs and precipitates with greater surface area and the agglomeration of more residual aluminum ions and orthophosphate contributed to this phenomenon. As for the chemical consumption, at least 12 mg/L Al3+ concentration (alum) should be dosed to ensure an effluent TP below 0.02 mg/L if water samples are not filtered. In contrast, if the samples were filtered, 6 mg/L coagulant concentration was enough to meet the requirement for the influent TP around 0.3 mg/L.

For scale-up issues, design standards must be considered, but the pilot study results should be taken into account in the design. For example, alum was selected for the coagulant and PAM for the flocculant. The coagulation precipitation system was required to lower the TP concentration to no more than 0.08 mg/L. Online TP analyzers should be installed in the influent and effluent of coagulation precipitation systems for forward and feed-back control so as to optimize the chemical dosage. When the ultra-low phosphorus concentration of the UF effluent cannot not be measured or exceeds the minimum limit of online analyzers, an automatic sampler is recommended to form composite samples for lab testing. Accurate and reliable experimental data are vitally important for evaluation of the effluent concentration. In addition, sludge produced by chemical dosing and backwash water should be taken into account and a sludge system such as an equalization tank, a sludge storage tank, sludge dewatering, and sludge silo be set up in the full-scale WTP. It is estimated to have 680 m3/d sludge with 99.8% water percentage.

ACKNOWLEDGEMENTS

This work was supported by the National Natural Science Foundation of China (Grant No. 40903049 and Grant No. 41473094).

REFERENCES

REFERENCES
Acero
J. L.
Benitez
F. J.
Leal
A. I.
Real
F. J.
Teva
F.
2010
Membrane filtration technologies applied to municipal secondary effluents for potential reuse
.
J. Hazard. Mater.
177
,
390
398
.
Benisch
M.
Clark
D.
Neethling
J. B.
Sid Fredrickson
H.
Gu
A.
2007
Tertiary phosphorus removal pilot tests technology limits in Coeur D'Alene, ID
.
Water Pract.
1
(
5
),
1
13
.
DeBarbadillo
C.
Shellswell
G.
Cyr
W.
Edwards
B.
Waite
R.
Sabherwal
B.
Mullan
J.
Mitchell
R.
2010
Development of full-scale sizing criteria from tertiary pilot testing results to achieve ultra-low phosphorus limits at Innisfil, Ontario
. In:
Proceedings of the Water Environment Federation, Nutrient Recovery and Management
, pp.
976
997
.
Feng
L. J.
Wang
W. Y.
Feng
R. Q.
Zhao
S.
Dong
H. Y.
Sun
S. L.
Gao
B. Y.
Yue
Q. Y.
2015
Coagulation performance and membrane fouling of different aluminum species during coagulation/ultrafiltration combined process
.
Chem. Eng. J.
262
,
1161
1167
.
Havens
K. E.
Fukushima
T.
Xie
P.
Iwakuma
P.
James
R. T.
Takamura
N.
Hanazato
T.
Yamamoto
T.
2001
Nutrient dynamics and the eutrophication of shallow lakes Kasumigaura (Japan), Donghu (PR China), and Okeechobee (USA)
.
Environ. Pollut.
111
,
263
272
.
Hook
G.
Ott
R.
2001
The ultimate challenge for technology: 0.02 mg/l effluent total phosphors
. In
Proceedings of the Water Environment Federation, Session 41 to Session
50
, pp.
113
132
.
Hu
K. L.
Deng
L.
Wang
L. F.
2004
Analysis on the city wastewater treatment and the eutrophication of lakes and reservoirs
.
J. Kunming Univ. Sci. Technol.
29
,
168
172
.
Hugh
G.
Tozer
P. E.
2007
Study of five phosphorus removal processes select ComagTM to meet Concord, Massachusetts’ stringent new limits
. In
Proceedings of the Water Environment Federation, Nutrient Removal
, pp.
1492
1509
.
Jiang
C.
Jia
L. Y.
He
Y. L.
Zhang
B.
Kirumba
G.
Xie
J.
2013
Adsorptive removal of phosphorus from aqueous solution using sponge iron and zeolite
.
J. Colloid Interf. Sci.
402
,
246
252
.
Laine
J. M.
Vial
D.
Moula
S. P.
2000
Status after 10 years of operation-overview of UF technology today
.
Desalination
131
,
17
25
.
Metcalf & Eddy
2004
Wastewater Engineering-Treatment and Reuse
.
4th edn
.
McGraw-Hill
,
New York
,
USA
, pp.
491
502
.
Narong
P.
James
A. E.
2006
Effect of pH on the ζ-potential and turbidity of yeast suspensions
.
Colloid. Surfaces A: Physicochem. Eng. Aspects
274
,
130
137
.
Peeters
J.
Liu
M. G.
Thompson
D.
2010
UF membrane for the achievement of limit of technology effluent phosphorous concentrations
. In
Proceedings of the Water Environment Federation, Membrane Applications
, pp.
1014
1025
.
Tian
J. Y.
Liang
H.
Li
X.
You
S. J.
Tian
S.
Li
G. B.
2008
Membrane coagulation bioreactor (MCBR) for drinking water treatment
.
Water Res.
42
,
3910
3920
.
Zhang
J. F.
Zhang
D. Z.
Zhou
C.
Wang
X. M.
Yang
H. W.
Xie
Y. F.
2014a
Effect of filter media configuration characteristics on the performance efficiency in quartz sand filter
.
J.Xi'an Univ. Arch. Tech.
46
(
5
),
737
741
.