In this study, the competitive mechanism of ammonia, iron and manganese for dissolved oxygen (DO) in a biofilter was investigated, and a new start-up method of a biofilter for ammonia, iron and manganese removal was approved, which can effectively shorten the start-up period from 3–4 months to 51 days. The results demonstrated that when DO was sufficient (about 8 mg · L−1), ammonia, iron and manganese could be completely removed. When DO decreased from 6.5 to 4 mg · L−1, the concentration of ammonia in the effluent increased accordingly, though iron and manganese were removed efficiently. When DO was as low as 3 mg · L−1, only iron was removed, whereas most of the ammonia and manganese still existed in the effluent. In addition, the oxidizing rates of the pollutants were not affected significantly with DO decrease. Turbidity removal in the biofilter was also investigated, and the results demonstrated that the turbidity decreased to less than 0.5 NTU at 0.4 m depth of the filter.

INTRODUCTION

The presence of ammonia in potable water supplies is often accompanied by the presence of iron and manganese in concentrations above the permitted limits (Gouzinis et al. 1998). Continuously increasing ammonia concentration in groundwater has been observed in the past several years (Mustika 2004), owing to the discharge of waste from both industry and bank-side residents without adequate pre-treatment and the sub-optimal condition of the catchments (Akkera et al. 2008). The presence of ammonia, iron and manganese in drinking water should be avoided and the maximum contaminant levels (MCLs) for ammonia of 0.5 mg · L−1, total iron of 0.3 mg · L−1 and manganese of 0.1 mg · L−1 have been established in China (GB 5749-2006). The presence of iron and manganese in potable water treatment can result in aesthetic and operational problems, such as giving water an unpleasant metallic taste, staining on laundry and plumbing fixtures (Tekerlekopoulou & Vayenas 2008), causing defects in industrial products, and substrates for the growth of bacteria (Kontari 1988). In addition, when manganese exceeds the permitted limit, it has been found to affect the central nervous system (Sharma et al. 2011). The presence of ammonia in potable water treatment will affect the chlorination process (Hasan et al. 2013), because ammonia reacts with chlorine to form disinfection by-products (Charrois & Hrudey 2007; Richardson & Postigo 2012), which can damage the human nervous system (Nieuwenhuijsen et al. 2000), cause deterioration of the taste and odor of the water (Richardson et al. 2007), and reduce disinfection efficiency (WHO 1996).

Biological iron, manganese and ammonia removal is preferable compared with chemical removal, since it is not necessary to provide the addition of extra chemicals, and the volume of the generated sludge is appreciably smaller and hence it is easier to handle (Tekerlekopoulou & Vayenas 2007). Redox is the key parameter which determines the sequence of the biological oxidation of iron, manganese and ammonia. Iron is oxidized at low redox values (<200 mV) and ammonia at higher redox values (200–400 mV), while manganese needs even higher redox values (>400 mV) at a pH of 7 (Mouchet 1992). Thus simultaneous removal of iron, manganese and ammonia, which are present in a water system, requires several step processes or spatial redox variation.

The start-up period of a biofilter for manganese removal is typically 1–2 months, but when raw water contains manganese and ammonia, the start-up period of a biofilter for manganese and ammonia removal may be 3–4 months (Frischherz et al. 1985), which is obviously longer than the start-up period of a biofilter for manganese removal, since biological manganese removal only takes place after complete nitrification because of the necessary evolution of the redox (Rittmann & Snoeyinck 1984; Mouchet 1992; Vandenabeele et al. 1995). Therefore, research into shortening the start-up period is significant.

During the biological ammonia, iron and manganese removal process, ammonia oxidizing bacteria (AOB), iron oxidizing bacteria (IOB) and manganese oxidizing bacteria (MnOB) utilize dissolved oxygen (DO) to oxidize the pollutants. So DO is a very important operational parameter, which determines the efficiency of ammonia, iron and manganese removal. Moreover, energy is needed in general, when oxygen dissolves in water, so the higher the DO in the water, the higher the energy consumption and operational cost. Many researchers have investigated simultaneous removal of ammonia, iron and manganese from different perspectives, such as the interactions between iron, manganese and ammonia removal using a pilot-scale trickling filter (Gouzinis et al. 1998), biological removal of iron, manganese and ammonia from potable water by single-step filtration (Tekerlekopoulou & Vayenas 2007), and the reaction dynamics of biological iron, manganese and ammonia removal (Tekerlekopoulou & Vayenas 2008). DO in their filters was high, meaning that DO was 7–8 mg · L−1 or sufficient for ammonia, iron and manganese oxidization. However, few researchers have investigated the competitive mechanism of ammonia, iron and manganese for DO at different DO conditions, and the efficiency of ammonia, iron and manganese removal when DO was insufficient.

It should be noted that although the concentration of iron and manganese may be lower than the permitted limits, iron and manganese can accumulate and be deposited into the water distribution system when presented over a long period of time (Hasan et al. 2011), thus the turbidity may exceed the MCL for a turbidity of 1 NTU in China (GB 5749-2006). In the filter system, support materials serve two purposes: bioconversion of iron, manganese and ammonia by the biomass attached to the large surface of the support materials, and physical removal of suspended particles by support material filtration (Vandenabeele et al. 1995). Therefore, utilization of small-sized support materials in the filter may overcome the issue, because it would not only increase the removal rate of iron and manganese (Cho 2007), but also intercept more suspended particles.

In this study, a pilot-scale biofilter for the simultaneous removal of ammonia, iron and manganese was established to propose a new method of quick-start strategy, and investigate the competitive mechanism of ammonia, iron and manganese for DO at different DO conditions (DO was about 8 mg · L−1, 6.5 mg · L−1, 6 mg · L−1, 5 mg · L−1, 4 mg · L−1 and 3 mg · L−1, respectively, and theoretical oxygen demand to completely oxidize Fe2+, Mn2+ and NH4+-N in raw water was calculated as follows (Li et al. 2013), [O2] = 0.14[Fe2+] +0.29[Mn2+] +4.57[NH4+-N] = 7.10 mg · L−1). The main objectives of this study were to find an effective method to shorten the start-up period and investigate the competitive mechanism of ammonia, iron and manganese for DO at different DO conditions to reduce energy consumption and operational cost.

MATERIALS AND METHODS

A pilot-scale biofilter was developed in a groundwater treatment plant (GWTP), which is located in Harbin city, P.R. China. The biofilter consisted of a 147.19 L transparent rigid plexi-glass column with an inner diameter of 250 mm and a height of 3,000 mm, which was designed according to a full-scale industrial filter in P.R. China to make the loadings (hydraulic and contaminants) per unit cross-sectional areas directly applicable to full-scale operations (Figure 1). At the top of the filter, the incoming water was firstly mixed in the mixing chamber and then it flowed into the filter. Meanwhile, at the bottom of the filter, an underdrain system was used to collect the treated water and any biological solids which detached from the media. Along the filter depth, there were 20 sampling ports at 10 cm intervals for DO, iron, manganese, ammonia and turbidity concentration measurements. Tank 1 (volume 2,000 L) was used to collect aerated groundwater with a DO concentration of about 8 mg · L−1, which was obtained from the GWTP. Tank 2 (volume 50 L) was used to collect the effluent water of the filter in the start-up process and raw groundwater with a DO concentration of about 0.2 mg · L−1 from the GWTP in the experiment to investigate the competitive mechanism of ammonia, iron and manganese for DO at different DO concentrations.
Figure 1

Schematic drawing of the pilot-scale filter.

Figure 1

Schematic drawing of the pilot-scale filter.

The biofilter was packed with two kinds of new support materials to a height of 150 cm. The upper part of the filter bed (30 cm) was filled with anthracite, which were cylinders with an average diameter of 1 mm and a height of 5 mm, while the lower part of the filter bed (120 cm) was loaded with manganese sand with a diameter of 0.8–1 mm.

Real groundwater, which was extracted from wells with a depth of 40–50 m, in Harbin city, P.R. China, was used throughout this experiment. The concentration of iron, manganese and ammonia in the raw groundwater was about 7–10 mg·L−1, 1.1 mg · L−1 and 1.2 mg · L−1, respectively. The temperature was about 8 °C. In order to avoid the collision of bubbles with the deposited sludge, the aeration process was not performed in the filter, which could cause a disturbance of the system and increase iron concentration in the effluent (Katsoyiannis et al. 2002; Katsoyiannis & Zouboulis 2004). Downward gravity flow was adopted in this biofilter, and the amount of flow was controlled at the entry point. Due to pore clogging from bacteria growth and iron and manganese precipitation on the support material surfaces, regular backwashing was performed about every 2 days to wash out dead bacteria and maintain the activity of the system at a high level. The sampling time was 8:30 am every day, and the backwashing time was 10:00 am.

Start-up process

About 50 L backwashing water of the GWTP, containing an abundance of IOB, MnOB and AOB, was collected and sedimentated for 3 min to remove iron and manganese oxide, then the supernatant water (about 40 L) was discharged into tank 1 as inoculated water. Finally, the inoculated water and the aerated groundwater in tank 1 were mixed and pumped into the filter. Meanwhile, the effluent water of the filter was recycled into tank 1. This circulation condition between tank 1 and the filter was operated for 3 days to improve the effect of inoculation in this study. After that, new inoculated water and aerated groundwater were added in tank 1, and the biofilter was operated in the same way as mentioned above for 2 days.

Since the concentration of iron, manganese and ammonia in the aerated groundwater was relatively high, part of the effluent water of the filter was returned to the influent, and a low flow rate was adopted in the early days of the start-up period. Five days later, the effluent water of the filter was collected in tank 2. The aerated groundwater in tank 1 and the effluent water in tank 2 were pumped into the filter with a proportion of 0.2:0.8. When the concentrations of total iron, manganese and ammonia in the effluent were lower than 0.2 mg · L−1, 0.05 mg · L−1 and 0.1 mg · L−1, respectively, the above-mentioned proportion was changed to 0.4:0.6, and then to 0.6:0.4, 0.8:0.2 and 1:0 in the same way. The flow rate of the filter was maintained at 2 m · h−1 in this stage.

When the concentrations of total iron, manganese and ammonia in the effluent were decreased to lower than 0.2 mg · L−1, 0.05 mg · L−1 and 0.1 mg · L−1, respectively, the flow rate was first increased to 3 m · h−1 and then to 4 and 6 m · h−1 in the same way. When the flow rate was 2 or 3 m · h−1, the backwashing water of the biofilter was collected and sedimentated for 3 min, then the supernatant was discharged into tank 1 as inoculated water. It should be indicated that iron, manganese and ammonia were measured from the 6th day; when the concentrations of total iron, manganese and ammonia in the effluent were all lower than 0.2, 0.05 and 0.1 mg · L−1, and the flow rate was 6 m · h−1, the biofilter was started up successfully.

Competitive mechanism of ammonia, iron and manganese for DO at different DO concentrations

The aerated groundwater in tank 1 and the raw groundwater in tank 2 were pumped into the filter in a suitable proportion to ensure the DO concentration in the influent was about 6.5 mg · L−1, 6 mg · L−1, 5 mg · L−1, 4 mg · L−1 and 3 mg · L−1, respectively. In addition, the flow rate of raw groundwater was almost constant, because DO in the raw groundwater was so low that ferrous iron could not be oxidized, and iron hydroxide could not accumulate in the pipelines to reduce the pipe diameter and flow rate. Thus the flow rate of the aerated groundwater was regulated to ensure the proportion was suitable. It should be noted that the flow rate of the filter was fixed at 4 m · h−1 during this experiment, because iron in the influent was high.

Simultaneous removal of turbidity, ammonia, iron and manganese

The aim of this experiment was to investigate the removal efficiency of turbidity simultaneously with ammonia, iron and manganese. The turbidity in the influent and effluent was monitored, and the variation of turbidity along the filter depth was also monitored in this experiment.

Analysis methods

The pH, DO and turbidity measurements were conducted using a pH meter (Ultra BASIC UB-10), a DO meter (Oxi 315i-WTW), and a turbidimeter (2100Q-HacH), respectively. Concentrations of iron, manganese and ammonia were measured according to Standard Methods for the Examination of Water and Wastewater (Rittmann & Snoeyinck 1984).

RESULTS AND DISCUSSION

The start-up process

The biofilter had a high efficiency of iron removal from the 6th day during the start-up period, and its concentration in the effluent was all below 0.1 mg · L1 (data are not shown). The concentration of ammonia in the influent fluctuated drastically in the early days of the start-up period (Figure 2), because part of the effluent water was returned to be influent, and the return rate of the effluent water was varied. Ammonia in the effluent was decreased to lower than 0.1 mg · L1 on the 18th day, and then increased significantly due to the decreased return rate of the effluent water. Finally, ammonia in the effluent dropped quickly to lower than 0.1 mg · L1. The effluent water was not returned to be influent after the 31st day, and the flow rate was increased to 3 m · h1, 4 m · h1 and 6 m · h1 on the 33rd, 40th and 47th days, respectively. Ammonia in the effluent was increased as the return rate of the effluent water was decreased and the flow rate was increased except when the flow rate was increased to 6 m · h1; ammonia in the effluent dropped to below 0.1 mg · L1 in less than 24 h when the flow rate was increased from 4 to 6 m · h1.
Figure 2

Long-term performance of the biofilter with respect to simultaneous removal of Mn2+ and NH4+-N in the start-up period.

Figure 2

Long-term performance of the biofilter with respect to simultaneous removal of Mn2+ and NH4+-N in the start-up period.

Manganese in the effluent was lower than 0.05 mg · L−1 before the 41st day, and increased to 0.34 mg · L−1 in the 42nd day, but dropped to lower than 0.05 mg · L−1 3 days later. Then manganese in the effluent was increased as the flow rate increased, but dropped to lower than 0.05 mg · L−1 quickly. From the 51st day, total iron, manganese and ammonia in the effluent were all below 0.2 mg · L−1, 0.05 mg · L−1 and 0.1 mg · L−1, respectively, and the flow rate was 6 m · h−1, suggesting the biofilter was started up successfully. Manganese in the effluent was very low at the beginning of the start-up period, attributed to adsorption of manganese sand, dilution by the returned effluent, and biological oxidization by the inoculated MnOB. In the start-up period, manganese was removed by manganese sand adsorption and biological oxidation. The effect of the former became weaker while the latter was enhanced gradually as time went on, which meant if biological manganese oxidation was enhanced quickly enough, manganese in the effluent would be all below 0.05 mg · L−1.

In this experimental condition, the start-up period of the biofilter for ammonia, iron and manganese removal was 51 days at about 8 °C, which was much shorter than that of Frischherz et al. (1985). The reasons were as follows: (a) iron, manganese and ammonia in the influent were diluted by the returned effluent water and the bacteria in the returned effluent water were inoculated again; and (b) two kinds of support materials were used to avoid the redox reaction occurring between Fe2+ and Mn4+ after backwashing. Therefore, it was an effective strategy to shorten the start-up period of the biofilter that part of the effluent water was returned to be influent with varying return rates, two kinds of support materials were used, and a low flow rate was adopted in the early days of the start-up period. Besides, if the proportion of the aerated water and the flow rate were increased slowly, the concentration of total iron, manganese and ammonia in the effluent would be all below the permitted limits from the 17th day, and thus less substandard water would be produced.

Competitive mechanism of ammonia, iron and manganese for DO at different DO concentrations

When DO in the influent was sufficient (about 8 mg · L−1), the pollutants were simultaneously removed by the biofilter, and the concentrations of total iron, manganese and ammonia in the effluent were all below 0.1 mg · L−1, 0.05 mg · L−1 and 0.1 mg · L−1 (Figure 3) with average concentrations of 0.033 mg · L−1, 0.013 mg · L−1 and 0.033 mg · L−1, respectively. Total iron, manganese and ammonia (Figure 4) at depths 0.2 m and 0.4 m of the filter were 0.12 mg · L−1, 0.95 mg · L−1 and 0.24 mg · L−1; and 0.09 mg · L−1, 0.16 mg · L−1 and 0.17 mg · L−1, respectively, and manganese was reduced to lower than 0.05 mg · L−1 at depth 0.8 m. Iron and ammonia were mainly removed at depths of 0–0.2 m, while manganese was mainly removed at depths of 0.2–0.4 m, suggesting that iron and ammonia were simultaneously removed from the top of the filter depth, while manganese removal occurred when iron was completely removed.
Figure 3

Effect of filter on (a) total iron, (b) ammonia and (c) manganese removal for DO in the influent of about 8 mg · L−1, 6.5 mg · L−1, 6 mg · L−1, 5 mg · L−1, 4 mg · L−1 and 3 mg · L−1, respectively.

Figure 3

Effect of filter on (a) total iron, (b) ammonia and (c) manganese removal for DO in the influent of about 8 mg · L−1, 6.5 mg · L−1, 6 mg · L−1, 5 mg · L−1, 4 mg · L−1 and 3 mg · L−1, respectively.

Figure 4

Total (a) Fe, (b) NH4+-N and (c) Mn2+ (c) concentration profiles along the filter depth at different DO conditions (8, 6.5, 6, 5, 4 and 3 mg · L−1).

Figure 4

Total (a) Fe, (b) NH4+-N and (c) Mn2+ (c) concentration profiles along the filter depth at different DO conditions (8, 6.5, 6, 5, 4 and 3 mg · L−1).

In this experiment, iron, manganese and ammonia were removed at a relatively low filter depth. The reasons were as follows. (a) Smaller-sized support materials could provide larger specific surface area for bacterial adherence, thus enhancing the efficiency of ammonia, iron and manganese removal (Cho 2007). Although small support materials could cause a pore clogging phenomenon (Kontari 1988; Tekerlekopoulou & Vayenas 2007 ), this could be solved by using a higher filter, which would produce a higher water pressure to enhance the influent flow rate. (b) The depth of support materials could increase gradually due to the accumulation of iron hydroxide and manganese oxides on the support materials, which were not washed out during the backwashing process. However, the sampling ports did not change in all the experiments for conveniently comparing the removal efficiency of ammonia, iron and manganese along the filter depth.

When DO in the influent was reduced to about 6.5 mg · L−1 on the 11th day (Figure 3), which was not enough to completely oxidize iron, manganese and ammonia simultaneously, iron could still be removed completely at the top of the filter depth where DO was sufficient, while manganese and ammonia removal were affected obviously. In detail, manganese was firstly completely removed, then exceeded the permitted limit, and finally, dropped to lower than 0.05 mg · L−1 again. The reason for manganese varying in the effluent could be explained as follows. Manganese oxidation needed two steps. Mn2+ was adsorbed on bacteria cells and oxidized to manganese oxide (Hasan et al. 2012). Therefore, manganese was completely removed firstly by adsorbing onto the surface of MnOB cells, and then the adsorption ability was saturated, leading to manganese exceeding the permitted limit. Eventually, MnOB adapted to the low DO condition (DO < 0.5 mg·L−1 in 0.4 m depth of the filter) and oxidized Mn2+ to manganese oxides efficiently, which made the manganese in the effluent drop to lower than 0.05 mg · L−1 again. In steady periods, the oxidizing rate of manganese was not affected significantly at the DO concentration of 6.5 mg · L−1 compared with the oxidizing rate when DO was sufficient (Figure 4(c)), while the oxidizing rate of ammonia was slightly affected (Figure 4(b)). In this condition, IOB utilized DO firstly to oxidize iron, and manganese was also oxidized, but ammonia was not completely oxidized.

When DO in the influent was reduced to about 6 mg · L−1, 5 mg · L−1 and 4 mg · L−1 on the 33rd, 57th and 83rd days (Figure 3), respectively, the removal order of ammonia, iron and manganese was the same as that when DO was about 6.5 mg · L−1. When DO was about 6 and 5 mg · L−1, iron, manganese and ammonia in the effluent were all below the permitted limits in steady periods, but when DO decreased to about 4 mg · L−1, ammonia in the effluent exceeded the permitted limit, although iron and manganese were still below the permitted limits. When DO in the influent was about 6, 5 and 4 mg · L−1 (Figure 4), the oxidizing rates of iron, manganese and ammonia were not affected significantly compared with the sufficient DO condition, although ammonia could not be completely oxidized. Therefore, biofilters for simultaneously removal of ammonia, iron and manganese could be operated at relatively low DO conditions to save energy consumption and operational cost.

When DO in the influent was reduced to about 3 mg · L−1 (Figure 3), iron was also completely removed, while the efficiency of ammonia removal was 30%–40%. Surprisingly, manganese in the effluent was about 20% higher than that in the influent. At 0.1 m depth of the filter, the removal rates of total iron and ammonia were 88% and 28%, respectively, but manganese was slightly higher than in the influent (Figure 4). In the rest of the filter depth, ammonia was hardly removed, while total iron decreased but manganese increased conversely. This phenomenon could be attributed to the reduction of manganese oxides by Fe2+ in the biofilter. Iron and ammonia could be simultaneously removed from the top of the filter depth, but manganese removal only took place after ferrous iron was oxidized completely, suggesting that IOB and AOB preferentially utilized DO to oxidize iron and ammonia, while MnOB could not use DO until the iron was completely removed.

The efficiency of turbidity removal in pilot-scale biofilter

Turbidity removal with simultaneous ammonia, iron and manganese removal in the biofilter was also investigated. As shown in Figure 5, turbidity in the influent fluctuated drastically around 2.73 NTU, because when raw groundwater was aerated, ferrous iron could be easily oxidized to ferric iron hydroxide. The turbidity in the effluent was all lower than 0.5 NTU, and the average concentration was 0.30 NTU.
Figure 5

Effect of the biofilter on turbidity removal.

Figure 5

Effect of the biofilter on turbidity removal.

The turbidity concentration profiles along the filter depth are shown in Figure 6. The turbidity in the influent was 4.13 NTU, and decreased to 0.24 NTU at 0.4 m depth of the filter, indicating that most of the turbidity could be removed in the upper part of the filter bed, because the smaller-sized support materials in the upper part of the filter bed (the smaller-sized support materials were arranged in the upper part of the filter bed after backwashing) had a high ability to intercept the iron hydroxide and other impurities.
Figure 6

Turbidity profiles along the filter depth.

Figure 6

Turbidity profiles along the filter depth.

CONCLUSIONS

Several operating measures including two kinds of support materials (anthracite and manganese sand), inoculated AOB, IOB and MnOB from the GWTP, and returning part of the effluent water to be the influent in the early days of the start-up period were proved to be effective at shortening the start-up period of the biofilter from 3–4 months to 51 days for simultaneous ammonia, iron and manganese removal.

IOB and AOB could utilize DO preferentially to oxidize iron and ammonia, respectively, at the top of the filter depth, while MnOB could not utilize DO to oxidize manganese until the iron was completely removed. However, MnOB would have an advantage over AOB in utilizing DO at low DO conditions (DO < 0.5 mg L−1). Furthermore, the oxidizing rates of ammonia, iron and manganese were not affected significantly with DO in the influent decreasing.

Small-sized support materials, which were used in the biofilter, not only improved the efficiency of ammonia, iron and manganese removal, but also exhibited a high efficiency for the removal of turbidity. Moreover, the turbidity was mainly removed at 0–0.4 m depth of the filter.

ACKNOWLEDGEMENTS

This work was supported by the Scientific Research Foundation of CUIT (KYTZ201511).

REFERENCES

REFERENCES
Cho
B. Y.
2007
Manganese removal using an aerated granular filter
.
Korean, J. Chem. Eng.
24
,
757
762
.
Frischherz
H.
Zibuschka
F.
Jung
H.
Zerobin
W.
1985
Biological elimination of iron and manganese
.
Water Supply
3
,
125
136
.
Hasan
H. A.
Abdullah
S. R. S.
Kamarudin
S. K.
Kofli
N. T.
2013
On–off control of aeration time in the simultaneous removal of ammonia and manganese using a biological aerated filter system
.
Pro. Safety Environ. Protect.
91
,
415
422
.
Katsoyiannis
I.
Zouboulis
A.
Althoff
H.
Bartel
H.
2002
As(III) removal from groundwaters using fixed-bed upflow bioreactors
.
Chemosphere
47
,
325
332
.
Kontari
N.
1988
Groundwater, iron and manganese: an unwelcome trio
.
Water Engineer. Manage.
135
,
25
26
.
Mouchet
P.
1992
From conventional to biological removal of iron and manganese in France
.
J. Am. Water Works Assoc.
84
,
158
162
.
Mustika
S. A.
2004
Potential optimisation of chlorination at Buaran WTP, Jakarta Water and Environmental Engineering
.
Degree of Master of Science, University of Surrey
,
Guildford, UK
.
Nieuwenhuijsen
M. J.
Toledano
M. B.
Eaton
N. E.
Fawell
J.
Elliott
P.
2000
Chlorination disinfection byproducts in water and their association with adverse reproductive outcomes: a review
.
Occup. Environ. Med.
57
,
73
85
.
Richardson
S. D.
Postigo
C.
2012
Drinking water disinfection by-products.
In:
Emerging Organic Contaminants and Human Health, Vol. 20
.
Springer
,
New York, NY
,
Vol. 20, Chapter 4
, pp.
93
137
.
Rittmann
B. E.
Snoeyinck
V. L.
1984
Achieving biologically stable drinking water
.
J. AWWA.
76
,
156
174
.
Sharma
S. K.
Kappelhof
J.
Groenendijk
M.
Schippers
J. C.
2011
Comparison of physicochemical iron removal mechanisms in filters
.
J. Water Supply: Res. Technol. Aqua.
50
,
187
198
.
Tekerlekopoulou
A. G.
Vayenas
D. V.
2007
Ammonia, iron and manganese removal from potable water using trickling filters
.
Desalination
210
,
225
235
.
Vandenabeele
J.
Beer
D. D.
Germonpré
R.
Sande
R. V.
Verstraete
W.
1995
Influence of nitrate on manganese removing microbial consortia from sand filters
.
Wat. Res.
29
,
579
587
.
WHO
1996
Ammonia in Drinking Water. Guidelines for Drinking-Water Quality, Health Criteria and Other Supporting Information, Vol. 2
, 2nd edn,
World Health Organization
,
Geneva, Switzerland
.