The household biosand filter (BSF) is a highly utilized point-of-use water treatment tool. The effect of ambient temperature on the ability of the BSF to remove microbes from water is unclear. Model filters were distributed among different temperature-controlled laboratories and dosed daily with surface water amended with sewage. Comparison of the total coliform and Escherichia coli counts in the influent versus effluent revealed an immediate drop in the removal efficiencies of filters held in colder rooms. This performance difference, however, became less pronounced over the course of the experiment until no significant performance difference was detected between filters regardless of their ambient temperature, perhaps due to microbial adaptation within the BSFs. Subsequently, two-thirds of the filters were exposed to freezing temperatures, thawed, and re-tested for microbial removal. All filters exposed to freezing temperatures showed significant drops in microbial removal compared to control filters. Filters exposed to the most extreme temperatures showed the greatest drop in performance.

INTRODUCTION

According to the World Health Organization roughly 88% of all cases of diarrhea are caused by unsafe drinking water, substandard sanitation, and improper hygiene (Prüss-Üstün et al. 2008). The ideal scenario to counter unsafe drinking water would be to deliver treated water to each individual household via pressurized pipes from an unlimited, protected source such as already exists in some nations. However, with that goal still potentially decades away for many communities with fewer economic resources, point-of-use (POU) water treatment tools are effectively used to improve the quality of water in the home (Montgomery & Elimelech 2007). POU water treatment tools are used to treat water after delivery to the home and typically treat only the portion of the total flow designated for drinking (EPA 2015). When used properly, POU systems reduce the incidence of diarrhea by approximately half (Lenton et al. 2005; Sobsey et al. 2008). Common POU treatments include solar disinfection, chemical treatment, and filtration.

The household biosand filter (BSF) has recently become a popular POU filtration device, with more than 500,000 units distributed between 1991 and 2015 (CAWST 2015). The basic structure typically consists of a plastic or concrete shell, although they can be made of ferrocement when plastic shells are unavailable (Arnold 2015), filled mainly with a sand media, and a discharge tube that is raised to a level such that a shallow layer of water is maintained over the sand surface (Elliott et al. 2008) (Figure 1). When water is added to the filter, a pressure gradient is created which initiates flow out of the discharge tube. The flow rate decreases to zero as the water level within the filter approaches the same level as the discharge tube. In a properly operating unit, a volume of filtered effluent equal to the dose volume should exit the filter before flow completely stops (Young-Rojanschi & Madramootoo 2014), however, in order to improve water quality, an increased detention time of more than 24 hours can be used (Arnold 2015). Care should be taken, however, since extended pause periods in flow can cause decreased removal (Young-Rojanschi & Madramootoo 2014; Kennedy et al. 2012). During physical filtration pathogens are removed from water by trapping, natural die-off, sorption and predation by other microorganisms (CAWST 2015). During the first weeks of use the BSF has been observed to develop a schmutzdecke, or filter cake, in which a combination of particle buildup and biological growth develops in the uppermost portion of the media. This buildup corresponds to the peak pathogen removal efficiency of the BSF (Weber-Shirk & Dick 1997). Similar biological development occurs throughout the entire body of the BSF especially during its first weeks of use, further contributing to pathogen removal by biological action (Elliott et al. 2011; Bradley 2011). There can be issues with initial start-up, number of people served, and a period of non-use (Kennedy et al. 2012), but BSFs have been shown to reduce the number of diarrhea cases (Voth-Gaeddert et al. 2015a, 2015b).
Figure 1

BSF models: latest model (left) BSF (CAWST 2015) and vertically identical BSF model (right) used in the experiments.

Figure 1

BSF models: latest model (left) BSF (CAWST 2015) and vertically identical BSF model (right) used in the experiments.

A BSF is similar to a slow sand filter (SSF). Water is passed between small sand grains and pathogens are removed by way of physical filtration and biological action (Aslan & Cakici 2007). The two filter types are, however, different in that the SSF receives a constant flow of influent whereas the BSF is dosed intermittently. This difference has been shown to have a significant effect on the treatment capabilities of the two filter types (Young-Rojanschi & Madramootoo 2014).

Temperature depends on global location, elevation, season, and time of day. Global climate change also magnifies the importance of the effect of temperature on the global burden of water-related diseases, a subject that has gained more attention as a topic of study in recent years (IPCCWG 1998).

Exposing a biologically mature SSF to lower ambient temperatures has a significant initial effect on the filter's ability to remove coliforms due to the effect of the temperature change on the biological activity within the filter (Bellamy et al. 1985). In their study Bellamy et al. (1985) found that when matured filters, held at a temperature of 17°C, were shifted to constant temperatures of 5 and 2°C and then tested for coliform removal, 10% and 7.6% lower removal efficiency occurred, respectively. While one would expect temperature to have a similar effect on the microbial removal efficiency of the BSF, such a study has not been performed. Furthermore, the temperature study by Bellamy et al. (1985) did not report the effect of temperature on sand filters over time and thus did not test the filter's ability to adapt as a biological system to the changes in temperature.

The objective of this study, therefore, was to determine by laboratory testing the effect of temperature and freezing on the BSF's ability to remove coliforms and Escherichia coli.

METHODS

Filter preparation

Twelve BSF models were constructed using 6.4-cm diameter PVC. In the vertical direction they were built identically to the most current BSF model, the Version 10 (V10) filter, according to the Center for Affordable Water and Sanitation Technology (CAWST 2015) (Figure 1). The bottom drainage gravel ranged in size from 0.64 to 1.3 cm in diameter and the smaller gravel located between the base gravel layer and sand was between 0.076 and 0.64 cm. The filters were filled with distilled water before the media was added to avoid the trapping of air. Given an average measured media porosity of 47%, as well as the interior volume of the exit pipe, and the 5.1 cm of standing water above the top layer of sand, the total volume of water stored in each filter between doses was 1.2 litres.

To seed and establish biological growth, the filters were stored at 18°C during the first 35 days of the experiment with influent values of 155 to 9,550 colony-forming units (CFUs) of total coliform and from 41 to 2,275 CFUs of E. coli. The filters received 1-litre doses of water daily, which was a mixture of a local fresh water source amended with raw sewage. Filters were tested every 3 to 7 days for coliform and E. coli removal efficiencies. This sampling period was long enough to detect a measurable amount of removal, but short enough to document the removal in a detailed manner.

Temperature separation

To isolate the effect of temperature on filter performance, the filters were randomly separated into four groups of three and held at temperatures of 4, 12, 18, and 27°C starting on day 35 until day 63 of the experiment. These temperatures were selected because they evenly spanned the temperature range from just above freezing to the warmest laboratory setting available. In the temperature-controlled laboratories the filters continued to be dosed daily with the same fresh water–sewage mixture and were tested every 3 to 7 days for 25 days.

On day 64 of the experiment, to determine the effect of freezing on filter performance, two filters from each of the four temperature-controlled laboratories (eight of the 12 filters) were randomly chosen to be frozen for 10 hours and then re-thawed. Four of the eight selected filters were frozen at a temperature of −1°C while the four others were frozen at −22°C. These temperatures were selected to represent both a partial and a complete freeze, respectively. All eight of the filters selected for freezing were then placed back in the temperatures from which they had been removed and allowed to thaw for 24 hours before being tested. The filters that were not selected for freezing were held at their respective temperature values as controls.

Water analysis

Daily dosing of the BSFs consisted of 14 litres of river water at 18°C inoculated with 5 ml of raw sewage, and then vigorously stirred for three minutes before immediately dosing the filters. As a control measure, 2 litres of the water–sewage mixture were set aside and tested for coliforms and E. coli immediately (0-hour), and again after a 24-hour period, to determine the natural die-off of the bacteria.

Throughout the 66-day experiment filters were tested 15 times for total coliform and E. coli removal and eight times for turbidity reduction and flow rate. A Hach 2100P portable turbidity meter was used to measure the turbidity of the filter influent and effluent (Hach 2015). Flow rate was measured as the time it took for the first 100 ml to exit the filter. This volume was chosen as the minimum available, since the flow rate through the BSF continuously decreases as the water level above the exit pipe drops.

Membrane filtration was used to measure the coliform and E. coli colony-forming units (CFUs) in 100 ml grab-samples taken from the total effluents of each filter. Membranes were saturated with 2 ml of m-coli Blue broth and incubated for 24 hours at 37°C in accordance with EPA laboratory method 10029 (EPA 2003). Following incubation visible blue and red dots were counted as total coliform CFUs and blue dots as E. coli.

RESULTS AND DISCUSSION

The average turbidity of the influent mixture of natural surface water and sewage was 1.71 NTU with a standard deviation of 0.18. The average reduction in turbidity during filtration was 29.8% with a standard deviation of 7.7. The average flow rate throughout the experiment was 71.1 ml/min with a standard deviation of 8.45 ml/min. The average 24-hour natural die-off of coliforms and E. coli in samples stored alongside the filters during the first 30 days of the experiment was 10.8% and 15.3%, respectively.

In the first 35 days of the experiment, during which the filters were housed together in the 18°C laboratory, after an initial dip, the through-filter removals of coliforms and E. coli increased as a function of time to be similar by day 30 (Figure 2). On day 37, 2 days after the filters were separated into temperature-controlled laboratories, a clear trend toward lowered performance initially accompanied lower temperatures. By day 52 of the experiment, however, the average coliform and E. coli removal efficiencies of all filters had improved to reach similar values. This convergence of performance over time, regardless of varying temperature values, is perhaps due to the adaptation of the in-filter microbial strains to their respective temperatures. The greatest improvements were observed in filters held at lower temperatures (Figure 3).
Figure 2

Removal percentages of coliforms (top) and E. coli (bottom) during the first 35 days of testing. Error bars denote the standard error.

Figure 2

Removal percentages of coliforms (top) and E. coli (bottom) during the first 35 days of testing. Error bars denote the standard error.

Figure 3

Removal percentages of coliforms (top) and E. coli (bottom) after groups of filters were placed in four separate temperature-controlled laboratories. Error bars denote the standard error.

Figure 3

Removal percentages of coliforms (top) and E. coli (bottom) after groups of filters were placed in four separate temperature-controlled laboratories. Error bars denote the standard error.

Following the freezing and thawing of eight of the 12 filters, significant drops in the removal efficiencies of coliforms and E. coli were observed. The greatest reduction in removal efficiency was observed in the three filters which had been frozen at −22°C. Average drops of 29% and 35% were observed in the removal efficiencies of coliforms and E. coli, respectively, in those filters (Figure 4). Two filters were damaged by the freeze at −22°C and could not be further tested.
Figure 4

Removal percentages of coliforms and E. coli following exposure to 10 hours of freezing temperatures. Error bars denote the standard error.

Figure 4

Removal percentages of coliforms and E. coli following exposure to 10 hours of freezing temperatures. Error bars denote the standard error.

The results of this experiment suggest that the microbial community within the BSF is more adaptable to changes in temperature than was previously known. In order to determine what type of adaptation occurs within the BSF following exposure to different temperatures, the microbial community within the filters could be analyzed using molecular techniques as a function of time and ambient temperature. This would be useful in order to determine whether new types of microorganisms are repopulating the BSF when the temperature drops or if the original species are adapting to the cold.

One surprising result was that on day 35 of the experiment, following the initial separation of the filters, a small drop in performance was observed in filters which remained in the 18°C laboratory. This drop could be the result of small disturbances to the filters as they were moved approximately 3 metres and tethered to a new support shelf. This brief disturbance might be analogous to the cleaning of the top layer of the BSF during which water is agitated and drops in removal performance are commonly observed (Chiew et al. 2009). The decreased performance observed in the control filters at 18°C was, however, recovered over the course of the experiment. This recovery, as well as the recovery of the filters in different temperature laboratories, speaks to the adaptability of the BSF following exposure to many different types of environmental changes. The initial change in performance of the controls, however, demonstrates the vulnerability of the microbial community within the BSF, which can be negatively affected by seemingly minor disturbances. Therefore, disturbance of a BSF should be avoided.

Since freezing the filters led to markedly decreased performance, allowing the water within the filter to freeze should be avoided if possible in the field. The removals of coliforms and E. coli within the filters frozen solidly at −22°C dropped by approximately 29% and 35%, respectively. This is similar to the lowest removals observed during BSF development at the start of the experiment and suggests that the microbial community within the BSF may have been completely inactivated. In this case a longer period of redevelopment might be necessary before the BSF recovers its removal efficiency. Future study is necessary to investigate the length of the recovery time of the BSF following freezing events of differing severities. The combined observations of the BSF exposed to freezing temperatures indicate that, when possible, filters should be installed indoors when in environments where extremely low temperatures and freezing occur.

The effects of temperature changes on the performance of the BSF had not been previously observed in a laboratory setting. The results of this study indicate that the BSF has the ability to adapt to temperature changes within the range of those expected in most field conditions. Also, the adaptation of the BSF occurred on a time scale that indicates that it would be able to keep up with average seasonal temperature changes in the field as well as smaller global climate changes. However, the drops in BSF performance observed in this study, following very slight disturbances by temperature and movement, might be part of the explanation for the tendency toward higher removal efficiency in highly controlled laboratory studies as opposed to the field where diurnal temperature changes occur, a question pointed out by Stauber et al. (2006).

CONCLUSIONS

These pioneering laboratory experiments on temperature effects on a biological sand filter reveal that:

  1. BSFs are an effective, appropriate technology for water treatment;

  2. BSFs should be kept at warmer temperatures since BSFs at colder temperatures have less bacteria removal initially;

  3. BSFs can be effective at any temperature above freezing, however, since, given time, the microbial community adapts to the ambient temperature and bacteria removal levels reach similar levels at all temperatures;

  4. BSFs should not be physically disturbed or allowed to freeze since this can cause an initial decrease in performance; and

  5. BSF performance does recover from a physical disturbance over time.

ACKNOWLEDGEMENTS

This research was carried out with the financial assistance of the Peace Corps Masters International Program at Michigan Tech. The authors would also like to acknowledge the MTU biology department for lending its laboratory equipment, Emily Geiger for her expertise in microbiology, Casey Arnold for his guidance using AutoCAD, Dr Mark Rouleau and Dr Daisuke Minakata for their writing suggestions, and Candice Young-Rojanschi with the CAWST organization for their research suggestions.

REFERENCES

REFERENCES
Arnold
N. B.
2015
Household Biosand Water Filters: Materials, Capacity, and Temperature Effects on Performance
.
MS Thesis
,
Dept. Civil and Environmental Engineering, Michigan Technological University
,
Houghton, MI
,
USA
.
Aslan
S.
Cakici
H.
2007
Biological denitrification of drinking water in a slow sand filter
.
Journal of Hazardous Materials
148
(
1–2
),
253
158
. doi:10.1016/j.jhazmat.2007.02.012.
Bellamy
W. D.
Hendricks
D. W.
Logsdon
G. S.
1985
Slow sand filtration: influences of selected process variables
.
Journal – American Water Works Association
77
(
12
),
62
66
. .
CAWST
2015
CAWST Center for Affordable Water and Sanitation Technology. Charitable foundation. CAWST Center for Affordable Water and Sanitation Technology, Calgary, Canada
. .
Chiew
H.
Sampson
M. L.
Huch
S.
Ken
S.
Bostick
B. C.
2009
Effect of groundwater iron and phosphate on the efficacy of arsenic removal by iron-amended biosand filters
.
Environmental Science & Technology
43
(
16
),
6295
6300
. .
Elliott
M. A.
Stauber
C. E.
Koksal
F.
DiGiano
F. A.
Sobsey
M. D.
2008
Reductions of E. coli, echovirus type 12 and bacteriophages in an intermittently operated household-scale slow sand filter
.
Water Research
42
(
10–11
),
2662
2670
. doi:10.1016/j.watres.2008.01.016.
Elliott
M. A.
DiGiano
F. A.
Sobsey
M. D.
2011
Virus attenuation by microbial mechanisms during the idle time of a household slow sand filter.
Water Research
45
(
14
),
4092
4102
. doi:10.1016/j.watres.2011.05.008.
EPA
2003
United States Environmental Protection Agency
. .
EPA
2015
Office of Web Communications and Office of Environmental Information. Point of Use (POU) Technologies| Radionuclides Decision Tree| US EPA
. .
Hach Company
2015
Ensuring water quality for people around the world. Hach Company website
. .
IPCCWG
1998
The Regional Impacts of Climate Change: An Assessment of Vulnerability. Cambridge University Press, Cambridge, UK
.
Kennedy
T. J.
Anderson
T. A.
Hernandez
E. A.
Morse
A. N.
2012
Determining the operational limits of the biosand filter
.
Water Science Technology Water Supply
13
(
1
),
56
65
.
Lenton
R. L.
Wright
A. M.
Lewis
K.
2005
Health, Dignity and Development: What Will It Take? Earthscan, London
.
Montgomery
M. A.
Elimelech
M.
2007
Water and sanitation in developing countries: including health in the equation
.
Environmental Science & Technology
41
(
1
),
17
24
.
Prüss-Üstün
A.
Bos
R.
Fiona Gore
J. B.
2008
Safer Water, Better Health: Costs, Benefits and Sustainability of Interventions to Protect and Promote Health. World Health Organization (WHO), Geneva, Switzerland
.
Sobsey
M. D.
Stauber
C. E.
Casanova
L. M.
Brown
J. M.
Elliott
M. A.
2008
Point of use household drinking water filtration: a practical, effective solution for providing sustained access to safe drinking water in the developing world
.
Environmental Science & Technology
42
(
12
),
4261
4267
. doi:10.1021/es702746n.
Stauber
C. E.
Elliott
M. A.
Koksal
F.
Ortiz
G. M.
DiGiano
F. A.
Sobsey
M. D.
2006
Characterisation of the biosand filter for E. coli reduction from household drinking water under controlled laboratory and field use conditions
. Water Science & Technology
54
(
3
),
1
7
.
Voth-Gaeddert
L. E.
Divelbiss
D. W.
Oerther
D. B.
2015b
Utilizing structural equation modeling as an evaluation tool for critical parameters of the biosand filter in a pilot study in Para, Brazil
.
Water Science Technology Water Supply
15
(
4
),
845
851
.
Weber-Shirk
M. L.
Dick
R. I.
1997
Physical–chemical mechanisms in slow sand filters
.
American Water Works Association Journal
89
(
1
),
87
.
Young-Rojanschi
C.
Madramootoo
C.
2014
Intermittent versus continuous operation of biosand filters
.
Water Research
49
(
February
),
1
10
. doi:10.1016/j.watres.2013.11.011.