Date post: | 12-Apr-2015 |
Category: |
Documents |
Upload: | engineers-without-borders-uk |
View: | 36 times |
Download: | 0 times |
Individual Environmental Systems Project
Optimisation of Household scale Biosand Filters
Richard Outhwaite
Project Supervisor: Dr. Luiza Campos
Date: 08.09.2010
Optimisation of household Biosand filters Richard Outhaite
Acknowledgements
I would like to thank all those who have helped me carry out this project, in
particular to Dr Luiza Campos as the project supervisor and to Ian Sturtevant and
Judith Zhou for their assistance in the laboratory.
I would also like to thank Engineers without Borders for the funding provided they
have provided for the materials for this project, without which it would not have
been possible. Thanks also to WBB minerals for providing the sand used in the
project and Andy Skinstad for his transportation services.
Optimisation of household Biosand filters Richard Outhaite
Abstract
The household Biosand filter is a well-established example of ʻappropriateʼ technology used throughout the developing world to improve the quality of drinking water. This project studied two methods to optimize its performance and mitigate some of the risks that can arise from incorrect operating conditions and procedures. It also investigated the filters ability to remove pesticides from raw water. The first method to optimise the filter was by ʻseedingʼ one filter using sand from an existing Biosand filter to speed up the maturation of the filter. The second method used to improve the performance of the filter was the addition of a tap to
the outflow pipe, which would prevent the need for users to store water that had been filtered, thus removing the possibility of recontamination through storage. The ability of the filter to remove pesticides was studied to allow us to establish whether the filter could be used on waters contaminated by other pollutants. Total coliform counts were used as the indicator to quantify bacteriological removal and Gas Chromatography-Mass spectrometry was used to detect the residual pesticide, Metaldehyde, in the filtrate. Headlosses, pH, turbidity and Dissolved Oxygen were also measured and studied. The seeded filter reached 99% coliform removal after 16 days, 10 days before the unseeded filter. It showed significantly higher coliform removal rates than the unseeded filter throughout the study across the 3 samples taken during each test (p < 0.01). The filter with the tap showed that it could still maintain significantly high coliform removal rates (99.2%, p <0.01) even after pausing the flow of water. These two modifications to the installation and operating system indicate that they can both improve filter performance and reduce the risk of operational misuse. The filter did not show consistent removal of the Metaldehyde, as a wide spectrum of removal rates were recorded (0.00% - 73.5%). Whist the results were not conclusive to show that the filter could consistently remove the pesticide; it does however show some potential, which could be investigated in further studies.
Optimisation of household Biosand filters Richard Outhaite
Table of Contents
1. Introduction....................................................................................................1
2. Literature Review...........................................................................................3
2.1. Slow sand filtration....................................................................................4
2.2. Biosand filter .............................................................................................9
2.3. Pesticide removal....................................................................................12
3. Methodology ................................................................................................14
3.1. Experiment Design..................................................................................14
3.1.1. Filter maturation study ......................................................................15
3.1.2. Tap study..........................................................................................15
3.1.3. Pesticide removal study....................................................................20
3.2. Filter Construction...................................................................................22
3.2.1. Sand media ......................................................................................23
3.3. Dye tests .................................................................................................25
3.3.1. Tracer tests ......................................................................................25
3.3.2. Solution tests....................................................................................26
3.4. Raw water ...............................................................................................26
3.5. Bacteriological analysis & coliform counting ...........................................27
3.6. Culture Media..........................................................................................28
3.7. Metaldehyde sampling & analysis...........................................................28
3.8. pH meter .................................................................................................29
3.9. DO meter ................................................................................................29
3.10. Sterilization..............................................................................................29
4. Results and Discussion ..............................................................................30
4.1. Dye Testing.............................................................................................30
4.2. Filter maturation study.............................................................................33
4.2.1. Coliform removal efficiency...............................................................33
4.2.2. DO Consumption ..............................................................................37
4.2.3. Headloss development .....................................................................39
4.2.4. Turbidity reduction ............................................................................41
4.3. Tap study ................................................................................................45
4.3.1. Coliform removal efficiency...............................................................45
Optimisation of household Biosand filters Richard Outhaite
4.3.2. DO consumption...............................................................................48
4.3.3. Turbidity Reduction...........................................................................50
4.3.4. pH reduction .....................................................................................52
5. Pesticide removal ........................................................................................53
6. Conclusions and Recommendations.........................................................57
6.1. Maturation Study.....................................................................................57
6.2. Tap study ................................................................................................58
6.3. Pesticide Study .......................................................................................59
7. References ...................................................................................................60
8. Appendices ..................................................................................................64
Optimisation of household Biosand filters Richard Outhaite
List of Figures
Figure 2-1 – Relationship between grain diameter and pore size (Huisman &
Wood, 1974) ...................................................................................................6
Figure 2-2 - Structural formula of Metaldehyde (WHO, 1996) ..............................12
Figure 3-1 Schematic of test procedure and sampling frequency.........................17
Figure 3-2 - Schematic of test procedure and sampling frequency for tap study..19
Figure 3-3 - Schematic diagram of experiment variables & sampling frequency..21
Figure 3-4 - Plan and section views of constructed Biosand filters ......................22
Figure 3-5 - Particle size distribution - Unseeded sand........................................24
Figure 3-6 - Particle size distribution - Seeded sand............................................24
Figure 3-7 - Raw water sampling location – Regents Park...................................27
Figure 4-1 - Graph of Light absorption against Time – Unseeded filter ................30
Figure 4-2 - Light absorption against Time - Seeded filter....................................31
Figure 4-3 - Light absorption against Time - Unseeded & seeded solution tests .32
Figure 4-4 - Dye Tracer tests in Unseeded filter at (a) 5 mins & (b) 70 mins........33
Figure 4-5 - Graph of Coliform removal rate against time - Unseeded filter .........34
Figure 4-6 - Graph of Coliform removal rate against time - Seeded filter .............35
Figure 4-7 - Dissolved oxygen consumption against time (Unseeded filter).........38
Figure 4-8 - Dissolved oxygen consumption against time (Seeded filter).............39
Figure 4-9 - Headloss development for seeded and unseeded filters ..................40
Figure 4-10 - Turbidity of raw water .....................................................................42
Figure 4-11 - Turbidity of Filtrate - Unseeded filter ...............................................43
Figure 4-12 - Turbidity of Filtrate - Seeded Filter..................................................43
Figure 4-13 – Possible positions of seeded sand within a filter............................44
Optimisation of household Biosand filters Richard Outhaite
Figure 4-14 - Coliform removal rate against time – Filter without Tap..................47
Figure 4-15 - Coliform removal rate against time – Filter with Tap.......................48
Figure 4-16 - Coliform removal rate against DO consumption – No Tap..............49
Figure 4-17 - Coliform removal rate against DO consumption - Tap ....................49
Figure 4-18 - Turbidity of filtrate - Filter without tap..............................................51
Figure 4-19 - Turbidity of filtrate - Filter with tap...................................................52
Figure 5-1 – Concentration of Metaldehyde in filtrate at 180µg/L – Filter without
Tap................................................................................................................54
Figure 5-2 Concentration of Metaldehyde in filtrate at 1000µg/L – Filter without
Tap................................................................................................................54
Figure 5-3 Concentration of Metaldehyde in filtrate at 180µg/L – Filter with Tap .55
Figure 5-4 Concentration of Metaldehyde in filtrate at 1000µg/L – Filter with Tap55
Figure 7-1 - Gas Chromatography parameters ....................................................70
Figure 7-2 - Chromatogram for Metaldehyde sample with internal standard........71
Figure 7-3 - Integral of chromatogram to determine quantity of Metaldehyde ......72
Optimisation of household Biosand filters Richard Outhaite
List of Tables
Table 2.1 - Health effects of Metaldehyde exposure (INCHEM, 1996).................13
Table 4.1 – Effect of seeding on coliform removal rates for samples collected at
different residence time.................................................................................36
Table 4.2 – Effect of residence time on coliform removal rates for seeded and
unseeded filters.............................................................................................37
Table 4.3 - Coliform removal rates and significance factors for filters with and
without Tap ...................................................................................................45
Table 4.4 – Coliform removal rates on filters with tap and without tap .................46
Table 4.5 - Mean turbidity for filtrate from filters with & without tap ......................51
Table 4.6 - Mean filtrate pH for filter with & without tap........................................52
Table 5.1 - Reduction in Metaldehyde concentration at 180µg/L & 1000µg/L ......53
Optimisation of household Biosand filters Richard Outhaite
List of Equations Equation 1 - Uniformity Co-efficient......................................................................23
Equation 2 - Porosity of sand bed ........................................................................25
Optimisation of household Biosand filters Richard Outhaite
- 1 -
1. Introduction
In 1804, John Gibb of Paisley, Scotland established the first recorded example of
filtration as a means of water treatment. This was mainly for private use in his
bleachery, but some of the excess filtered water was sold off to the public. This
method was modified and adopted by others until James Simpson constructed the
first plant supplying water to the public in 1829 at the Chelsea Water works. Whilst
the processes involved in the prevention of disease by slow sand filtration were not
identified until John Snowʼs discovery of materes morbi, the benefits of slow sand
filtration were readily apparent. By 1885 the work of Pasteur and other leading
scientists of that era had led to regular bacteriological testing of the drinking water
supplies.
Due to advances in the technology of water treatment and vast improvements in the
knowledge of the process involved in water treatment, public health has gone from
strength to strength in the developed world. Water-borne disease has become a
thing of the past. However, many parts of the world are still desperately lacking in the
infrastructure and knowledge on how to provide themselves with clean and safe
drinking water. Children are the most vulnerable to a lack of access to clean water. It
is estimated that every 15 seconds a child dies from water related disease (UNDP,
2006) and that a total of 1.4 million children die from diarrhoea each year (Pruss-
Ustin et al., 2008). Out of all the people killed by diarrhoea each year, 90% are
children under 5 years old living in developing countries (UN Water, 2008).
Optimisation of household Biosand filters Richard Outhaite
- 2 -
Developments have been made in the appropriate household water treatment
technology in recent years. Developments such as the Biosand filter (Manz, 1997;
Buzunis 1995) and the Filtron ceramic water purifier (Potters for Peace, 2008) have
begun to address this problem, but there still remain a number of challenges.
As new technology develops and becomes available, new technical and socio-
technical challenges arise when these technologies are implemented. How people
use, and often more importantly misuse, these technologies create new challenges
to be solved. Incorrect or unanticipated use of the technology in the field can
increase the risk of exposure to contaminants (Baumgartner et al, 2007).
This study aimed to investigate whether modifications to the Biosand filter can
increase its performance in removing water-borne pathogens and avoid possible
pathways to misuse that can reduce its efficiency, as well as speeding up the time
for the filter to reach its optimum performance. It will also look at whether the filter is
able to remove other chemical compounds, such as pesticides, as the filter is often
used in rural, agricultural areas without protected water sources.
Optimisation of household Biosand filters Richard Outhaite
- 3 -
2. Literature Review
Diseases from poor sanitation and low quality drinking water currently affect more
than 2.2 billion people worldwide. It is estimated of those people over 1 billion people
still do not have access to an improved water source within one kilometre (WHO,
2000; Sphere, 2004).
Well known diseases such as diarrhoea and cholera; as well as trachoma,
schistosomiasis, amebiasis and giardiasis are prevalent in areas without access to
proper sanitation and drinking water (UNICEF, 2006). This greatly impacts on
mortality rates of populations, especially children. Not only are the health and
wellbeing of communities being impacted upon by the lack of critical infrastructure
but also the economic output of developing countries. People who are suffering from
any of those diseases cannot work and become a burden for their community.
The international community has recognized the issue, which as long ago as 1980
has been the subject of numerous international conventions. The most recent
international effort, The Millennium Declaration by the UN, created the Millennium
Development Goals (MDG). These outline major targets for improving sanitation and
water supply for people in developing countries. These aim to halve the number of
people unable to reach or afford safe drinking water and basic sanitation by 2015
(UN, 2000).
Optimisation of household Biosand filters Richard Outhaite
- 4 -
Achieving these goals is a monumental task. To be effective in developing countries
that do often have access to trained personal or specialized equipment, the
application of appropriate technology is essential. One of the most simple and
successful ways to treat water over the past 200 years has been filtration of water
through a sand bed.
2.1. Slow sand filtration
The basic operation of a slow sand filter is very simple. The raw water enters the
filter and remains above the sand bed for a few hours due to the slow flow rate of
water through the bed. During this time some separation and sedimentation occurs.
The water then passes through the sand bed, usually for 2 hours or more, where it is
exposed to a number of different processes that purify it. These will be covered in
more detail later.
In comparison to a rapid sand filter, in which the water only spends a few minutes in
the filter, the amount of bacterial removal and chemical improvement is very low.
However, rapid sand filters are usually used in conjunction with chemical
coagulation-flocculation followed by sedimentation or flotation. This removes a large
number of the contaminants before the water arrives at the rapid sand filter. To
maintain flow rates through the rapid sand filters, regular cleaning through
backwashing is required. The high flow rates in a rapid sand filter, between 100-200
m3/m2/day (Tebbut, 1998) and the frequent backwashing prevents the development
of a biofilm and schmutzdecke on the surface of the rapid filter, one the main
components of bacterial removal in slow sand filtration.
Optimisation of household Biosand filters Richard Outhaite
- 5 -
Typical flow rates of slow sand filters range from 1-4 m3/m2/day (Tebbut, 1998)
depending on their position in the treatment process. Slow sand filters in Amsterdam
have been run at rates of up to 6 m3/m2/day, as they were the final step of a multi-
treatment process (Visscher, 1990). It is this combination of slow flow rates and long
filter runs through a fine sand media that allow a host biological community to
establish in the filter and form a biofilm.
The schmutzdecke develops on the surface of the filter as the raw water is filtered
through it to create a layer of purifying bacteria and microorganisms. It is made up of
filamentous algae that host plankton, diatoms, rotifers, protozoa and bacteria
(Huisman & Wood, 1974). It is here that the purification process begins. After
passing through the schmutzdecke the water passes through the voids between the
sand grains where the water is subjected to further biological activity and other
removal mechanisms.
The processes that occur in slow sand filtration that removes pathogenic bacteria
and other microorganisms can be broadly placed into three categories: physical
straining, physico-chemical attachment and biological predation.
Physical straining occurs through the fine sand media used within the filter. Small,
suspended particles and other matter to which bacteria can often be attached are
filtered out at or near the surface of the sand bed. A sand bed can filter particles up
to approximately 15% of its grain diameter (Haarhoff & Cleasby, 1991, & Huisman &
Optimisation of household Biosand filters Richard Outhaite
- 6 -
Wood, 1974). This means that for a typical slow sand filter with an effective grain
size of 0.15mm, only particles larger than 20μm will be removed in theory. This is
shown in Figure 2-1. Colloidal particles, with diameters of 1μm or less, and bacteria
and viruses with lengths up to 15μm are much smaller than the pore sizes (Tebbut,
1998). Therefore the removal of viruses and bacteria within the filter must occur
through other processes.
Figure 2-1 – Relationship between grain diameter and pore size (Huisman & Wood, 1974)
The main processes of virus and bacteria removal from raw water are biological
predation, adsorption to the biofilm and to non-biological surfaces. (Wheeler et al.,
1988). Predatory bacteria and microorganisms that inhabit the biofilm and
schmutzdecke consume the pathogens as they pass through (Weber-Shirk & Dick,
1997). Whilst most of the biological action occurs within the uppermost areas of the
sand bed where the micro fauna and flora are most abundant (Wheeler et al., 1988),
biological action that occurs within the sand bed can occur up to 40cm deep into the
bed (ASCE, 1991).
Optimisation of household Biosand filters Richard Outhaite
- 7 -
Attachment to the biofilm and sand media takes place through of a series of physical
and chemical reactions between the sand grains, the biofilm and the bacterial cells.
Specific electrostatic charges cause an attraction between the organic matter and
the negative charged sand particles, which cause the bacteria or virus to bond to it
(Huisman & Wood, 1974). The same effect can occur through Van der Waals forces
once the particles have been brought into close proximity to each other.
Once the filter begins to ripen the biofilm, or zooglea (Brock & Madigan, 1991), is
formed by bacteria, organic matter and waste produced by the bacteria. This creates
a sticky surface that forms on the surface of the sand grains to which further organic
matter sticks. Here it is either broken down by the bacteria to be assimilated into the
biofilm or is retained on the surface until it is removed when the filter bed is cleaned.
Smaller, colloidal-sized particles are generally not removed through slow sand
filtration, as they are smaller than the void spaces in the sand bed. Such fine silt and
clay requires pre-treatment such as chemical coagulation-flocculation to form flocs
that are large enough to be filtered out by the slow sand filter. This can result in low
flow rates due to increases in headloss as the filter becomes blocked with sediment.
It is therefore recommended to remove such flocs by sedimentation or flotation if
possible prior to filtration.
A mature slow sand filter will produce filtrate of high quality. The raw water should
generally not have a turbidity of over 20 NTU to avoid over frequent bed cleaning,
Optimisation of household Biosand filters Richard Outhaite
- 8 -
unless other processes in the treatment train are used (Tebbut, 1998). Provided
these criteria are met is met the filter will provide a filtrate with:
• Turbidity less than 1 NTU
• Coliform removal rates up to 99%
• Virus and Microorganisms removal, e.g. Cryptosporidium and Giardia cysts,
up to 99%
• 75% removal of colour.
• 10% removal of TOC.
As slow sand filtration was one of the first techniques to be adopted, it has become
seen to be obsolete in the face of more technological and chemical treatment
processes. Currently, few authorities consider it to be appropriate technology,
especially as it requires larger land area in comparison to other treatment processes.
However as recent legislation limit the amount of Trihalomethanes in water, caused
by chlorination of water still containing organic matter, biological treatment methods
such as slow sand filtration are becoming more popular again.
One area where slow sand filtration is still seen as an appropriate technique is in
developing countries. As a standalone process, it has been ranked as the second
most effective technique in improving the physical, chemical and biological quality of
water after desalinisation and evaporation (House & Reed, 1997). The simple
installation, operation and maintenance make it a key technology in improving the
quality of drinking water in the developing world.
Optimisation of household Biosand filters Richard Outhaite
- 9 -
2.2. Biosand filter
The household Biosand filter was first conceptualized in the 1990s as a solution to
the widespread problems of poor water quality and public health in developing
countries. The household scale filter was developed to try and combat the fact that
with a community scale slow sand filtration, a small drop in participation of the
stakeholders would results in complete failure of the system for the entire
community. Conversely, a household scale scheme gives proportional benefits to the
amount of participation. It is often found that 98% of households in a community will
adopt the household Biosand filter (Manz, 2007).
It has been widely adopted throughout parts of the developing world in over 70
countries. Over 200,000 filters have been installed and more are being implemented
every day (Manz, 2007). The health benefits from using the filter have been
documented in two pilot schemes in the Dominican Republic and Kenya. During a six
month trial in the Dominican Republic, households benefited from a 47% reduction in
diarrhoeal disease, representing a significant improvement in health conditions
(Stauber et al, 2006.) Similar improvements were experienced by rural communities
in Kenya, where households recorded a 54% reduction in child diarrhoea cases
(Tiwari et al, 2009).
The main difference between a traditional slow sand filter and the Biosand filter
pioneered by Manz is that it is operated intermittently. Water is loaded into the filter
through a diffuser to avoid damaging the biofilm and schmutzdecke below where it is
allowed to drain freely through the sand bed until the water has reached its standing
Optimisation of household Biosand filters Richard Outhaite
- 10 -
level. The water standing level is recommended to be 50mm above the sand bed.
This is to allow oxygen to diffuse through the water to the biofilm on the surface and
keep the host bacteria alive (Buzunis, 1995). The most efficient depth of standing
water for biofilm development was identified to be 20-30mm (Palmateer et al., 1999),
however to prevent disturbing the biofilm and prevent drying of the filter bed in hot,
dry climates 50mm is generally adopted (Manz, 2007). Thus the entire volume of
water put into the reservoir is filtered and requires storage afterwards. However, this
introduces a new pathway for contamination.
Water has been regularly been recorded to have been re-contaminated after filtering
(CAWST, 2009). This could be avoided if the filter was fitted with a tap, so the user
did not to store filtered water and the filter could be used as a true on-demand based
supply. It has been recommended that the flow of water should not be paused during
its flow (Manz, 2009), however recent studies have shown that residence time plays
a key factor in improving the quality of the filtrate (Baumgartner et al, 2007, Elliot et
al, 2008 & Jenkins et al, 2009).
Previous studies looked at the effect of filtering smaller volumes, therefore keeping
the filtrate within the pore space of the sand bed for as long as possible. If a tap were
provided, the same effect could be achieved, but this would result in depths of raw
water greater than 50mm being retained above sand bed. Whilst it would be higher
than the optimum recommended 20mm-30mm depths, this was not developed
through laboratory tests. Buzunis (1995) recorded good filter performance with a
Optimisation of household Biosand filters Richard Outhaite
- 11 -
standing water level of 12.5cm; therefore there is the possibility of achieving good
filter performance with a flow control device.
As a biological process, the filter requires a certain time for the host bacteria to
develop. During this time it is recommended that the filtrate still requires chlorination
(CAWST, 2009). Many communities in the developing world have a low acceptance
to putting chlorine into their water; therefore this step may be omitted altogether
(Manz, 2007). There is also no clear indicator as to when the filter has achieved
satisfactory either. This is a major disadvantage compared to other household water
filtration systems developed by NGO ʻPotters for Peaceʼ such as the Filtron, which
provides immediate treatment, though at much slower flow rates (Latange &
Ambiental, 2002).
The maturation process can take up to 21 days before acceptable performance is
required. The performance can continue to improve until 53 days into operation
(Elliot et al, 2008). Thus methods to improve and speed up the maturation process
are one area where significant improvements can be made to the filter process. This
could be achieved by increasing the amount of nutrients in the form of additional
bacteria in the water, however in practice this is not feasible. In conventional slow
sand filtration, when filter beds are re-sanded after a period of use, the bed is
comprised of sand from the existing filter, with new, clean sand added on top of that.
This enables some of the bacteria and microorganisms to be retained and thus they
are able to populate the bed much more rapidly (Tebbut, 1998). This process could
Optimisation of household Biosand filters Richard Outhaite
- 12 -
be achieved in Biosand filters using sand from existing filters already established
within a community.
2.3. Pesticide removal
The bacterial and pathogen removal capabilities of the filter are well developed and
documented. However, the filters ability to remove other pollutants such as
pesticides and herbicides are not well known. To assess the filters ability to remove
such chemical compounds, Metaldehyde was used in the test series.
Metaldehyde is a molluscicide commonly used to attract and kill snails and slugs
around the world. It is used in household gardens as well agricultural crops. It takes
the form of pellets, sprays foams and others and is applied to the soil around crops
(WHO, 1996).
Metaldehyde is a cyclic polymer of acetaldehyde, and forms a white powder in its
pure forms. The molecular formula is C8H16O4. The structural formula of the molecule
is shown in Figure 2-2.
Figure 2-2 - Structural formula of Metaldehyde (WHO, 1996)
Metaldehyde is soluble in water up to 190 mg/L and has been classified as having
groundwater contamination potential by the EPA and the EA (Kegley et al, 2010). It
Optimisation of household Biosand filters Richard Outhaite
- 13 -
has been shown that existing removal techniques may not remove the pesticide from
water, often failing to reduce background levels, around 8μg/L, down to the required
limit of 0.1 μg/L required by the Environment Agency (EA, 2010).
The WHO classifies Metaldehyde as a class 2 toxin, which means it can be
moderately hazardous to acute health impacts (Kegley et al, 2010). The effects of
ingesting Metaldehyde are wide ranging depending on the dosage received. A small
dosage of up to 50mg/kg of body mass can cause vomiting, dizziness and nausea.
Larger doses can lead to organ failure and death (Extoxnet, 1993). The differing
health impacts and their corresponding dosages of Metaldehyde are shown in Table
2-1.
Table 2.1 - Health effects of Metaldehyde exposure (INCHEM, 1996)
Dosage Health Impacts Toxicity
Up to ≈ 5 mg/kg
• Salivation
• Facial flushing
• Fever
• Abdominal cramps
• Nausea and vomiting
LOW
Up to ≈ 50 mg/kg
• Drowsiness
• Tachycardia
• Spasms
• Irritability
MEDIUM
Up to ≈ 100-200
mg/kg
• Convulsions
• Tremor
• Hyperreflexia
HIGH
Up to ≈ 400mg/kg • Coma
• Death VERY HIGH
Optimisation of household Biosand filters Richard Outhaite
- 14 -
Whilst the short-term health effects have been identified, the long-term effects of low-
level dosage have not been identified. It is not currently listed as a carcinogenic,
mutagenic or endocrine disrupting substance by a number of authorities, however
the EPA has characterized it of being suggestive of such effects (EPA, 2006; Kegley
et al, 2010).
3. Methodology
This chapter describes the experimental design, and methods and techniques used
throughout each of the tests.
3.1. Experiment Design
Three experiments were designed to study how the filter functioned under different
operating conditions with an aim to improve or assess the filters performance. The
experiments were designed as follows:
• Dye Tests – Operating the filters using organic dye to establish the hydraulic
performance of the filter and the residence time the raw water spends in the
filter.
• Filter maturation study – Seeding of new filter bed using sand from existing
Biosand filter to speed up the maturation process.
• Filter flow control study – Filter paused twice during filter runs using a tap for 1
hour, 2 hours and 3 hours to establish and drop in bacteria removal
performance.
• Pesticide removal study - Filter paused twice for 2 hours during each run
using tap, concentration levels at 180μg/L & 1000 μg/L.
Optimisation of household Biosand filters Richard Outhaite
- 15 -
The raw water volume was derived from a minimum basic water requirement of 20L
per capita per day (WHO, 2000) and the reservoir volume of the filter, 12L. This
represents a basic required amount for drinking, hygiene and cooking, however this
may differ depending on the users, physiological, social and cultural norms (Sphere,
2004). A total volume of 24L ensures that the filtrate being sampled can be
compared directly with the raw water, as verified by dye testing and porosity
calculations. The filtrate was sampled 3 times during the filter run, to establish the
effect of different residence times within the filter on water quality.
The test procedure is shown in Figure 3-1.
3.1.1. Filter maturation study
To establish the effect of adding sand from an existing Biosand filter on the rate at
which a new filter matures, one filter (unseeded) was composed of new sand, whilst
the other (seeded) was composed of 50% new sand and 50% sand from an existing
Biosand filter. The top 20cm of sand within the existing Biosand filter was scraped off
and discarded, and 20cm of new, clean sand was added on top to create the seeded
filter. Following the test procedure shown in Figure 3-1, the effects of residence time
and the effect of seeding the filter to speed up its bacterial removal can be studied.
3.1.2. Tap study
To determine the effect of using a Biosand filter with a tap on the quality of the
filtrate, one filter was operated as per the current practice by allowing the entire
reservoir to freely pass through the filter (Figure 3-1), while the other had a tap. The
volume of raw water used for the tap study was 24L. The tap was added to the filter
that had been seeded in the previous study.
Optimisation of household Biosand filters Richard Outhaite
- 16 -
The flow of raw water was paused twice on the filter with the tap, to simulate water
being used at different intervals throughout a day when it was required, rather than
filtering the entire volume of water in the reservoir each time.
Optimisation of household Biosand filters Richard Outhaite
- 17 -
Figure 3-1 Schematic of test procedure and sampling frequency
T = 0 minutes
Add 1st 12L of raw water to filter
T = 45 minutes
Add 2nd 12L of raw water to filter
T = 90 minutes
Take sample after 24L filtered = sample S1
T = 10 minutes
Take sample after 4L filtered = Sample S2
Step 3 – Day 1
T = 0 minutes
Add 1st 12L of new raw water to filter
T = 45 minutes
Add 2nd 12L of new raw water to filter – Take sample S3
T = 90 minutes
Take sample after 24L filtered = sample S1
Step 2 – Day 1
Step 1 – Day 1 Step 5 – Day 2
Step 6 – Day 2
Step 7 – Day 2
Step 4 – Day 2
Optimisation of household Biosand filters Richard Outhaite
- 18 -
The flow on the filter with a tap was paused each time after eight litres of raw water
had flowed i.e. after 8L and 16L, then the water was allowed to flowed until all 24L
had flowed through and the water reached the 50mm standing water level. The
pause period was varied between one, two and three hours. Thus the effects of
pausing the flow of water through the filter on the bacteria removal capacity of the
filter could be determined. The test procedure for the tap study is shown in Figure 3-
2. For clarity, the steps for one day of testing are shown due to the increased
number of steps within the test.
Optimisation of household Biosand filters Richard Outhaite
- 19 -
Figure 3-2 - Schematic of test procedure and sampling frequency for tap study
T = 20 minutes + P1
Open tap and restart flow from filter
T = 90 minutes + P1 + P2
Take sample after 24L filtered = S1
Take sample after 4L filtered = Sample S2
T = 20 minutes
After 8L filtered, close tap & begin Pause 1 (P1)
T = 45 minutes + P1
Step 3 – Day 1
T = 60 minutes + P1
After 8L filtered, close tap & begin Pause 2 (P2)
Step 5 – Day 1
Step 6 – Day 1
Step 8 – Day 1 Step 4 – Day 1
T = 0 minutes
Add 1st 12L of raw water to filter
T = 10 minutes
Step 2 – Day 1
Step 1 – Day 1
Add 2nd 12L of new raw water to filter – Take sample S3
T = 60 minutes + P1 + P2
Step 7 – Day 1
Open tap and restart flow from filter
Optimisation of household Biosand filters Richard Outhaite
- 20 -
3.1.3. Pesticide removal study
As well as establishing whether the Biosand filter was able to remove the organic
contaminant, the effect of the tap on the filter was also studied. This will help to
establish whether one of the removal processes discussed earlier is more effective.
Thus one of the filters was run without pausing the flow as per Figure 3-1, whilst the
filter with the tap was paused twice during the test for 2 hours at a time, following the
Tap study test procedure (Figure 3-2).
The amount of pesticide was increased from normal background levels to be able to
better detect the amount removed by the filter. Therefore concentrations of 180μg/L
& 1000 μg/L were used. Samples were analyzed to determine the residual
Metaldehyde content in the filtrate. The effect of the pesticide on the filterʼs ability to
remove bacteria was also studied carrying out bacteriological analysis of samples
during the pesticide test.
A schematic of the experiments carried out is shown in Figure 3-3.
Optimisation of household Biosand filters Richard Outhaite
- 21 -
Figure 3-3 - Schematic diagram of experiment variables & sampling frequency
Pesticide study
No Tap,
C=1 mg/L
No Tap, C=180μg/L
No pause
Tap,
C=180μg/L 2-hour pause
Tap,
C=1mg/L
2-hour pause
Filter with
Tap
No Tap
1-hour pause
Filter with
Tap 2-hour pause
Filter with
Tap 3-hour pause
Variable Samples & Approximate residence Time Pause in test
Maturation study
Unseeded Filter
S1 S2 S3 90mins 24hrs 24hrs
Seeded Filter
No pause
No pause
Tap study
No pause
No pause
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3
90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
S1 S2 S3 90mins 24hrs 24hrs
Optimisation of household Biosand filters Richard Outhaite
- 22 -
3.2. Filter Construction
Two Biosand filters were constructed in accordance to the Biosand Filter Manual
(CAWST, 2009). To enable monitoring of the filter during its use, the front of the
filters was constructed from Perspex to allow better inspection of the biofilm and
schmutzdecke development. Coarse and fine gravel were washed and deposited
into the base of the each filter as the under-drain in 50mm layers, before the sand
layer was installed. The 400mm fine sand layer was installed wet into a shallow layer
of water to prevent air bubbles becoming trapped in the filter. The dimensions of
each filter as installed are shown in Figure 3-4.
Figure 3-4 - Plan and section views of constructed Biosand filters
50mm deep fine gravel
300mm
300m
m
PLAN
1100
m
m
50mm deep coarse gravel
400mm deep fine sand
50mm deep standing water
Section
Optimisation of household Biosand filters Richard Outhaite
- 23 -
3.2.1. Sand media
A sieve analysis was conducted to establish the effective size of the sand grains and
their uniformity coefficients. The test was carried out in accordance to British
Standard 1377-2:1990 – Methods of test for soils. The sand used in the study was
RH45 sand from WBB Mineralsʼ quarry in Redhill, Surrey.
As 50% of the sand to be used in the maturation study was from the existing Biosand
filter and therefore from a different batch of RH45 sand, a sieve analysis was carried
on a sample of that sand as well to determine any differences between the samples.
The sand was washed to remove fines, then oven-dried at 105°C. The sand was
sieved in a mechanical shaker for 15 minutes through the specified sieve sizes.
Based on the results of this test the effective diameter for the unseeded filter was
calculated to be 0.182mm, with a uniformity coefficient of 1.65. The effective
diameter for the seeded sand was calculated to be 0.202mm, with a uniformity
coefficient of 1.80. This is within typical values of grain size between 0.15mm and
0.30mm and uniformity coefficient of less than 4 for use in slow sand filtration
(Huisman & Wood, 1974). The particle size distribution is shown for both the sand
used for the unseeded and seeded filter respectively are shown in Figures 3-5 and 3-
6. Data for the sieve analysis can be found in Appendix A.
Uniformity Co-efficient = D60
D10
Equation 1 - Uniformity Co-efficient
Optimisation of household Biosand filters Richard Outhaite
- 24 -
Figure 3-5 - Particle size distribution - Unseeded sand
Figure 3-6 - Particle size distribution - Seeded sand
D10
D10
D60
D60
Optimisation of household Biosand filters Richard Outhaite
- 25 -
3.2.1.1. Sand bed porosity
The porosity of the sand bed was calculated by weighing a known volume of
saturated and unsaturated samples of the RH45 sand and calculating the weight of
water in the void space, therefore the percentage of voids or porosity. This is
expressed in the equation 2.
Equation 2 - Porosity of sand bed
The sand bed was calculated to have a porosity of 41.5%. Therefore the sand bed
could contain a maximum of 14.5L of water. Including the 50mm of standing water on
top of the sand bed, the total volume of water within the filter was 19.5L.
3.3. Dye tests
Dye tests were carried out to establish the residence time of the raw water in the
filter during test conditions following the test procedure in Figure 3-1 to confirm the
time to take samples during the filter test run. It was also used to establish the
possible presence of any wall effects in the columns (Mehta & Hawley, 1969).
3.3.1. Tracer tests
A volume of 2.5mL of organic green food colouring was simultaneously injected at
three locations along the Perspex face of the filter through a syringe. The plug of dye
was photographed at one-minute intervals throughout the test and the filtrate was
sampled every minute for two hours or until all the dye had run through the filter. The
filtrate samples were analyzed with a Spectrophotometer to determine the amount of
light absorbed, thus the concentration of dye.
Porosity = (Mass saturated sand – Mass dry sand)
(Volume of sand x ρwater)
Optimisation of household Biosand filters Richard Outhaite
- 26 -
The dye was subsequently injected into the middle of the sand bed and the process
repeated. This will determine whether the flow down the edges of the filter is quicker
than through the sand media. This was repeated twice times for each filter.
3.3.2. Solution tests
To further establish the time to concentration of the sample water flowing through the
Biosand filters, dye tests using 12L raw water dyed with 35mL of organic food
colouring were carried out. The dyed water was put into the filter as per the
experimental procedures (Figure 3-1) using two volumes of 12L dyed solutions. The
samples were taken every minute for two hours or until all the influent had run
through the filter, then analyzed with a Spectrophotometer to establish the
concentration of dye within each sample. From this data the time of peak
concentration can be established. This process was repeated twice for each filter.
3.4. Raw water
To simulate the conditions that the Biosand filter will be used in the field as closely
as possible, water from an unprotected source was required. This was preferred to
rainwater or tap water as both could contain chemicals e.g. Chlorine or other
compounds, which may hinder the development of the biofilm. The water was
collected from the lake in Regents Park in two 25L containers. The exact location is
shown in Figure 3-7.
Optimisation of household Biosand filters Richard Outhaite
- 27 -
Figure 3-7 - Raw water sampling location – Regents Park
3.5. Bacteriological analysis & coliform counting
Total coliform bacteria counts were used as the indicator for bacterial improvement
of the filtrate. The membrane filter technique was used to determine the number of
bacteria present in the influent and filtrate. A Nalgene filter and 0.45μm cellulose
membrane filters were used for all bacteriological tests. It is a direct plating method,
therefore as effective as the multiple-tube fermentation method for detecting
coliforms (APHA, 2005). Coliforms were counted using a Stuart SC6 colony counter
and reported as Colony Forming Units (CFU) per 100mL. Raw water was taken in
10mL samples and diluted to make it up to 100mL, as per regulatory requirements
for lake water (APHA, 2005). Filtrate samples were filtered in 100mL quantities. To
Location of sampling from Regents Park
Lake
Optimisation of household Biosand filters Richard Outhaite
- 28 -
allow for clear counting of coliforms, this was separated into four 25mL samples as
the filter was ripening and reduced accordingly as the coliform counts became lower.
3.6. Culture Media
Agar plates were produced using a dehydrated culture media. Fluka Modified-Endo
medium was used for total coliform growth. Culture media were prepared in batches
that were refrigerated and used within one week. The M-Endo media has been noted
to produce uncertain results, due to the non-appearance of metallic sheen on
colonies that form on the membrane due to the presence of non-coliforms
(Burlingame et al., 1983). E-Coli detection could have been included to supplement
the data and improve the robustness of the data by providing another indicator of
bacterial quality.
3.7. Metaldehyde sampling & analysis
Samples containing Metaldehyde were collected in PET bottles and stored below
5°C for up to 21 days prior to analysis. The 50mL samples were extracted using the
J.T Baker Solid Phase Extraction process. The samples were then analysed using
Parkin-Elmer Gas Chromatography-Mass Spectrometer to determine the amount of
Metaldehyde in each sample. The analysis process followed the Environment
Agency guidelines for the determination of Metaldehyde in water using GC-MS.
Detailed information on the extraction and analysis process can be found in
Appendix B.
Optimisation of household Biosand filters Richard Outhaite
- 29 -
3.8. pH meter
A Jenway 4330 conductivity and pH meter was used to measure the pH of the
samples. The pH meter used was calibrated daily using three standard buffer
solutions (pH 4.0, 7.0 and 10.0). Buffers solutions were dated when opened and
changed every week.
3.9. DO meter
A Jenway DO meter was calibrated every two weeks using standard solutions of
0mg/L and 100mg/L. The membrane was kept wet at all times to ensure no air
bubbles become trapped on the surface of the selective membrane.
3.10. Sterilization
All sample collection bottles, membrane filter assembly and pads, dilution and rinse
water used in microbial analysis were autoclaved at 121°C for 15 minutes. All
equipment was washed and rinsed thoroughly before use and stored in foil prior to
autoclaving. Heat indicating tape was applied to equipment to ensure it had reached
the required temperature. Safety cabinets and hoods were cleaned weekly before
and after use and wiped with surgical spirit.
Optimisation of household Biosand filters Richard Outhaite
- 30 -
4. Results and Discussion
4.1. Dye Testing
The tracer dye tests showed that the unseeded filter had a mean time to flow of 66
minutes when the dye was injected in the middle of the filter. When the filter was
injected against the wall of the unseeded filter the mean time to flow was 65 minutes.
This is shown in Figure 4-1. Tests Centre-2 and Centre-4 showed a time to flow of
71 minutes, slower than the other times. This coincided with a significant drop in the
temperature of the influent water, which was 7°C colder than the mean temperature
of 22°C. This would increase the viscosity of the water and reduce the rate of flow
through the filter, thus explaining the resulting longer time to flow.
Figure 4-1 - Graph of Light absorption against Time – Unseeded filter
Optimisation of household Biosand filters Richard Outhaite
- 31 -
The tracer dye tests for the seeded filter showed a mean time to flow of 60 minutes
when the dye was injected in the middle of the filter. The time to flow when injected
against the wall of the filter was 62 minutes. Using an independent t-test analysis to
compare the time to concentration for when dye was injected in the centre and at the
wall, the significance factor, p, of the injection point is 0.117. Therefore it can be said
that the injection point of the dye within the filter is not significant. This can be seen
in Figure 4-2.
Figure 4-2 - Light absorption against Time - Seeded filter
The unseeded filter would also have had a longer time to flow due to the increased
resistance to the flow from the slightly smaller grain size used in that bed. The
solution dye tests in both filters showed that under test conditions, the time to
concentration of the dye tests occurred at 80 minutes. This was determined when
Optimisation of household Biosand filters Richard Outhaite
- 32 -
the rate of change of light absorbed by the samples showed no further increase. The
total volume of 24L was filtered in 90 minutes. Therefore the samples taken at 90
minutes into the test after 24L is definitely the filtrate of the raw water put in that day.
The time to peak concentration is shown in Figure 4-3 for both filters.
Figure 4-3 - Light absorption against Time - Unseeded & seeded solution tests
Photographs showing the dye tracer tests in progress at time intervals of 5 minutes
and 70 minutes are shown in Figure 4-4.
Optimisation of household Biosand filters Richard Outhaite
- 33 -
(a) Dye Tracer test – Unseeded filter – T = 5 minutes
(b) Dye Tracer test – Unseeded filter – T = 70 minutes
Figure 4-4 - Dye Tracer tests in Unseeded filter at (a) 5 mins & (b) 70 mins
4.2. Filter maturation study
4.2.1. Coliform removal efficiency
The unseeded filter initially produced low coliform removal rates. Over the first five
days to of operation the filter achieved a maximum of 53% coliform bacteria removal.
As the filter and schmutzdecke were still maturing and developing, respectively, in
this period, the main removal mechanisms acting are physico-chemical attachment
of bacteria to the sand and physical straining of bacteria in the sand bed. The filter
matures as the microbiology is established, and the schmutzdecke begins to develop
and the biofilm grows on sand grains.
Optimisation of household Biosand filters Richard Outhaite
- 34 -
The unseeded filter achieved 90% removal after 15 days operation when S2 and S3
were collected with residence time of 24 hours. It took a further 11 days operation to
achieve 99% removal rates at 26 days. This rate of bacteria removal is considered to
be a mature filter (Manz, 2007). The coliform removal rates are shown in Figure 4-5.
Figure 4-5 - Graph of Coliform removal rate against time - Unseeded filter
The seeded filter reached higher coliform removal rates much more rapidly. After 7
days of operation it had achieved 90% coliform removal when samples S2 and S3
were collected. (Figure 4-6). This improved to 99% coliform removal after 16 days of
operation.
Optimisation of household Biosand filters Richard Outhaite
- 35 -
The samples S1 taken at the same day they have been filtered (i.e. residence time
equal 45min) had consistently lower coliform removal rates in both filters and did not
achieve consistently high coliform removal rates. Whilst the removal rate reached
80% removal after 10 days of operation in the seeded filter compared to 18 days in
the unseeded filter, after this it showed no further improvement, fluctuating around
this mark throughout the remainder of the tests. This can be seen in Figure X. It can
also be seen that the seeded filter started with much higher coliform removal rates
(<50%) while the unseeded filter started with lower coliform removal rates (>30%).
Figure 4-6 - Graph of Coliform removal rate against time - Seeded filter
A one-way analysis of variance was carried out to test the null hypotheses that
bacterial removal across with seeded and unseeded filters at each residence time
Optimisation of household Biosand filters Richard Outhaite
- 36 -
was equal. A p value of < 0.05 is considered to be significant. For sample S1,
seeding did not have a significant effect on the bacteria removal of filtrate (p =
0.393). This indicates that after 12L of water have been filtered, there is a decrease
in biological activity as the microorganisms within the filter have consumed as much
of the pathogens as possible, regardless of whether the filter has been seeded or
not. For samples S2 and S3, the seeding had a significant effect on the bacterial
quality of the filtrate, with significance factor p = 0.001 and 0.026 respectively.
Therefore it can be said that when the filter is used as per the standard procedure
i.e. using 12L raw water volumes, seeding has a significant effect on the bacterial
quality of the filtrate. It does not appear to have significant benefits when volumes of
water greater than 12L are filtered.
Table 4.1 – Effect of seeding on coliform removal rates for samples collected at different residence time.
Samples Seeded / Unseeded N Mean Coliform removal
(95% CI) Min–Max removal
Significance, p-value
S1 Unseeded 15 56.8% (41.5%, 72.2%) 0.00% - 83.2% Seeded 15 65.6% (49.9%, 81.6%) 0.00% - 87.3%
0.393
S2 Unseeded 15 65.2% (49.9%, 80.3%) 0.00% - 93.0% Seeded 15 91.7% (86.0%, 92.3%) 77.1% - 99.1%
0.001
S3 Unseeded 15 58.7% (40.0%, 77.4%) 0.00% - 92.6% Seeded 15 84.1% (80.4%, 97.8%) 0.00% - 98.5%
0.026
Note: N = number of samples
The significance of the sampling point within each filter was compared using
independent variable t-tests. For the seeded filter, samples S2 and S3 had
significantly higher removal rates than samples S1 (p = 0.004 and 0.036
respectively). There was no significant difference between samples S2 and S3 (p =
0.280). The unseeded filter showed no significant difference between samples S1,
Optimisation of household Biosand filters Richard Outhaite
- 37 -
S2 or S3 (p = 0.420, 0.869 and 0.576 respectively). This shows that the effect of
increased residence time does not affect performance until the filter is more mature.
As a more mature filter will have higher populations of microorganisms, this suggests
that it is the process of biological predation that is most prevalent during increased
contact time with the filter.
Table 4.2 – Effect of residence time on coliform removal rates for seeded and unseeded filters
Seeded/Unseeded Sample point N Mean Coliform
removal (95% CI) Significance, p-
value Seeded S1 15 65.6% (49.9%, 81.6%)
S2 15 91.7% (86.0%, 92.3%) 0.004
Seeded S1 15 65.6% (49.9%, 81.6%) S3 15 84.1% (80.4%, 97.8%)
0.036
Seeded S2 15 91.7% (86.0%, 92.3%) S3 15 84.1% (80.4%, 97.8%) 0.280
Unseeded U1 15 56.8% (41.5%, 72.2%) U2 15 65.2% (49.9%, 80.3%) 0.420
Unseeded U1 15 56.8% (41.5%, 72.2%) U3 15 58.7% (40.0%, 77.4%) 0.869
Unseeded U2 15 65.2% (49.9%, 80.3%) U3 15 58.7% (40.0%, 77.4%) 0.576
4.2.2. DO Consumption
During the filter maturation, increasing DO consumption levels were observed during
throughout. The consumption rates peaked when the filter reached maturation at 7
days and 15 days for the seeded filter (Figure 4-8) and unseeded filters (Figure 4-7)
respectively. The DO consumption was on average 68.1% for the seeded filter for
samples S2 when mature, while the unseeded filter consumed on average 66%.
Raw water was taken from a source with relatively high DO levels where flora and
fauna are abundant. DO levels in the raw water were always above 6mg/L, the
Optimisation of household Biosand filters Richard Outhaite
- 38 -
lowest levels required to sustain fish and other aquatic life. There was some variation
of DO level depending on the temperature and time of day the water was taken, as
both affect the amount of DO present due to the rate of photosynthesis of plants. If
lower DO levels were present in raw water, which may be found in groundwater from
wells, similar performance may not occur during the maturation process.
Figure 4-7 - Dissolved oxygen consumption against time (Unseeded filter)
Optimisation of household Biosand filters Richard Outhaite
- 39 -
Figure 4-8 - Dissolved oxygen consumption against time (Seeded filter)
The rate of consumption of DO seems to change as the filter is maturing, as the
unseeded filter shows higher consumption in the early stages of the maturation, then
plateaus. This can be seen in the seeded filter, as the DO consumption is almost
constant.
4.2.3. Headloss development
Headloss was monitored through the experiment and increased for both filters
throughout the maturation period and through the other tests. The unseeded filter
experienced much higher headlosses throughout the maturation period. The
headloss increased from an initial value of 6cm to 8.5cm after 19 days of filtration.
This continued to increase until the final test 54 days later, reaching a maximum of
12.4cm, an increase of 31%. This can be seen in Figure 4-9.
Optimisation of household Biosand filters Richard Outhaite
- 40 -
The seeded filter experienced lower headlosses through the tests. Initial losses were
4.3cm, increasing to 5.9cm after the maturation period. This is an increase of 28%, a
similar increase to the unseeded filter. However, by the final test headlosses had
increased to only 6.8cm, a further 14% increase, as shown in Figure 4-9.
Figure 4-9 - Headloss development for seeded and unseeded filters
This suggests that microbial dynamics therefore play an important role in headloss
development in slow sand filtration and the consumption of organic material reduces
filter clogging.
It can be seen that the headloss increases as the filters are operated and
schmutzdecke is developed. As the seeded filter experienced significantly less
headloss after maturity has been reached, this suggests that the biofilm and bacterial
Optimisation of household Biosand filters Richard Outhaite
- 41 -
population had fully developed by this stage. Headloss increase was observed in the
unseeded filter throughout the tests after the filter was considered mature, even
though the turbidity of the raw water was the same for both filters. This suggests that
deeper biological activity was still being established and takes much longer to
develop than the initial biofilm. This seems to be in contrast to typical headloss
profiles of uncovered filters, where the entire headloss occurs in the schmutzdecke
layer and the sand layers make no significant contribution (Barrett & Silverstein,
1988).
4.2.4. Turbidity reduction
The ability of the filter to reduce the turbidity of the filtrate was present from the initial
test in both the unseeded and seeded filter. Even when the biofilm wasnʼt considered
to be mature, the turbidity was reduced from a mean value of 8.85 NTU in the raw
water to below 1 NTU at sample points 2 and 3 for both filters. At sample point 1, the
recorded turbidity was generally higher than 1 NTU, as shown in Figure 4-11 and
Figure 4-12 for the unseeded and seeded filter respectively. This indicates that as
well as the physical straining of the sand bed; the increased residence time of
samples taken at 4L and 12L enabled greater physico-chemical removal of
suspended particles from the raw water. As the biofilm matured further, greater
turbidity reduction was observed. The biofilm and schmutzdecke thus increases
physical removal of particles as they adhere to the sticky surface or zooglea it
creates or are trapped within the porous mat of the schmutzdecke.
Optimisation of household Biosand filters Richard Outhaite
- 42 -
The unseeded filter showed higher reduction in turbidity levels (Figure 4-12). This
could be attributed to the lower effective diameter (0.182mm) of the grains used in
that filter, as the grains from the existing Biosand filter were 10% larger in their
effective diameter (0.202mm). This indicates that the effective size of the filter may
be more important in the reduction of turbidity levels than the maturity of the biofilm.
Figure 4-10 - Turbidity of raw water
Optimisation of household Biosand filters Richard Outhaite
- 43 -
Figure 4-11 - Turbidity of Filtrate - Unseeded filter
Figure 4-12 - Turbidity of Filtrate - Seeded Filter
Optimisation of household Biosand filters Richard Outhaite
- 44 -
The seeded filter clearly displays improved performance and speeds up filter
maturation i.e. 99% bacteria removal, 16 days before the unseeded filter (Figure 4-
8). However in the field it may not be possible to obtain sand 50% of the volume of
sand required from existing filters. Further studies should be carried out with varying
depths of seeded sand to measure the impact. The position of the seeded sand
within the filer should also be studied. As the biological activity within the filter
occurs) and the biofilm at the surface develops first near the surface, where food and
oxygen are more abundant for bacteria, it may be that putting the seeded sand at the
bottom of the sand bed enables the deeper bacteria community, up to depths of
0.4m within the filter (ASCE, 1991), to develop more quickly as they move towards
the food and oxygen source above.
Figure 4-13 – Possible positions of seeded sand within a filter
1. Bacteria in seeded sand move up through the bed closer to oxygen and food source.
2. Bacteria in seeded sand remain near surface to oxygen and food source
Scenario 1 Scenario 2
Optimisation of household Biosand filters Richard Outhaite
- 45 -
In addition to increasing the amount of bacteria within the filter by using sand from an
existing Biosand filter, it may be possible to add nutrients to the raw water to allow
the biofilm and host microorganisms to multiply more quickly if the raw water is low in
nutrients and bacteria. A simple packet of glucose could be added to the raw water
prior to being filtered. This can be a cheap, organic way to increase the nutrients for
the bacteria within the filter.
4.3. Tap study
4.3.1. Coliform removal efficiency
The mean coliform removal rate for the filter with tap was above 99% for samples S2
and S3 collected with residence time of 24h. Even at overflow conditions, samples
S1 achieved 97% mean coliform removal rates. This is contrary to recommendations
that the flow should not be stopped during the filtration (Manz, 2007). The mean
coliform removal rates for filters with and without tap are shown in Table 4-3.
Table 4.3 - Coliform removal rates and significance factors for filters with and without Tap
Sampling Point
Tap / No Tap N Mean Coliform removal
(95% CI) Min–Max removal
Significance, p-value
1 No Tap 15 84.2% (82.5%, 85.9%) 80.0% - 88.0% Tap 15 97.1% (94.2%, 99.0%) 91.7% - 99.9% <0.001
2 No Tap 15 98.9% (97.8%, 99.9%) 95.4% - 99.9% Tap 15 99.9% (99.7%, 100%) 99.1% - 100% 0.042
3 No Tap 15 98.3% (97.3%, 99.4%) 95.0% - 99.8% Tap 15 99.6% (99.2%, 99.9%) 98.1% - 100% 0.023
A one-way analysis of variance (ANOVA) was carried out to assess the effect of the
tap on the coliform removal rate at different sampling points. The tap had the most
significant effect on sample point 1. The filtrate from the filter with the tap was
significantly higher (p < 0.001%) than the filter without. At sample points 2 and 3, the
Optimisation of household Biosand filters Richard Outhaite
- 46 -
tap still provided significantly higher removal rates (p = 0.042 and 0.023
respectively).
This shows that at the filter with tap consistently provided higher coliform removal
rates. The tap increases the residence time and allows smaller volumes of water to
be filtered without storage. This follows work by Baumgartner et al. (2007), which
identified that smaller dosing volumes of water filtered had lower coliform counts.
Table 4.4 – Coliform removal rates on filters with tap and without tap
Tap / No Tap Sampling Point N Mean Coliform removal
(95% CI) Significance,
p-value Tap 1 15 97.1% (94.2%, 99.0%)
2 15 99.9% (99.7%, 100%) 0.004
Tap 1 15 97.1% (94.2%, 99.0%) 3 15 99.6% (99.2%, 99.9%) 0.036
Tap 2 15 99.9% (99.7%, 100%) 3 15 99.6% (99.2%, 99.9%) 0.280
No Tap 1 15 84.2% (82.5%, 85.9%) 2 15 98.9% (97.8%, 99.9%) 0.005
No Tap 1 15 84.2% (82.5%, 85.9%) 3 15 98.3% (97.3%, 99.4%) 0.012
No Tap 2 15 98.9% (97.8%, 99.9%) 3 15 98.3% (97.3%, 99.4%) 0.088
There were also significantly higher removal rates between samples at different
residence time for filters with and without the tap. Independent t-tests showed that
samples S1 had statistically significantly lower coliform removal rates over both
samples S2 and S3 for both filters, p = .0.004 and 0.036 for the filter with tap and
0.005 and 0.012 for the filter without tap. This shows that even with the improved
performance of the tap, the effect of residence time within the filter is significant.
Optimisation of household Biosand filters Richard Outhaite
- 47 -
Figure 4-14 - Coliform removal rate against time – Filter without Tap
This is in contradiction to current recommendations that pauses in filter flow are
detrimental to the filterʼs performance (Manz, 2007). Clear improvement can be seen
in the filtrate of the filter, which was paused using the tap. Even a 1-hour pause
increases the performance of the filter. Thus using tap for on-demand supply may be
a possible solution to increase performance of the filter and avoid one of the
pathways to recontamination.
However, the filter was always run down to 50mm pause level after 24-litres had
been filtered during the experiment. The effect of different pause levels on filter
performance should be investigated to assess any unfavourable combinations of
standing water level and pause time on filter performance.
Optimisation of household Biosand filters Richard Outhaite
- 48 -
Figure 4-15 - Coliform removal rate against time – Filter with Tap
4.3.2. DO consumption
The higher bacteria removal rates correlated with higher DO consumption when
using the tap. During the pause introduced by the tap, it is proposed that during the
increased residence time more of the organic matter begins to be broken down in the
raw water into more assimilable elements while still at the supernatant stage
(Huisman & Wood, 1974) and the bacteria consume more of the pathogens through
predation. Therefore as both of these processes require the consumption of more
oxygen, a reduction in DO of the filtrate occurs and the water quality is improved.
The DO consumption of each filter over the course of the Tap study is shown in
Figure 4-17. It can be seen from the graph that higher bacteria removal rates
occurred when using the tap.
1hr pause 2hr pause 3hr pause
Optimisation of household Biosand filters Richard Outhaite
- 49 -
Figure 4-16 - Coliform removal rate against DO consumption – No Tap
Figure 4-17 - Coliform removal rate against DO consumption - Tap
Optimisation of household Biosand filters Richard Outhaite
- 50 -
From the above graphs it can be seen that the amount of DO concentrations in the
raw water is critical. The raw water collected for this study was sufficient to sustain
fish and other wildlife, i.e. it contained a minimum DO concentration of 6 mg/l, in its
natural environment. These high concentrations of DO may have sustained the
bacteria within the biofilm and sand bed for longer during pause times. In addition to
this, the water inside the filter is not exposed to sunlight, as a normal slow sand filter
would be. As such it is not subject to the diurnal variations in DO levels caused by
the photosynthesis of algae (Tebbut, 1998). In this situation, water with higher
amounts of algae present will consume more DO during the paused periods when
using the tap. Despite this, it is recommended that the use of a tap be considered in
the design of the Biosand filter.
There are no current recommendations for the minimum DO levels for raw water to
be used with Biosand filters, as in developing countries it is difficult to be continually
monitored. However, it could be assessed during the pilot scheme or prior to
implementation of the filters in a community.
4.3.3. Turbidity Reduction
The effect of the tap on turbidity reduction was minimal, with no improvement gained
from increasing the residence time within the filter. The mean turbidity values for the
filter with and without a tap are shown in Table 4-5.
Optimisation of household Biosand filters Richard Outhaite
- 51 -
Table 4.5 - Mean turbidity for filtrate from filters with & without tap Mean Turbidity (NTU)
No pause No pause No pause No Tap-S1 1.44 1.51 1.44 No Tap-S2 0.82 0.62 0.90 No Tap-S3 0.82 0.80 0.99 1hr pause 2hr pause 3hr pause Tap-S1 1.44 1.56 1.23 Tap-S2 1.03 1.26 1.01 Tap-S3 1.19 1.33 1.32
The main factor affecting turbidity was again the sampling point. Samples S2 and S3
had lower turbidity than the sample S1. This can be seen in Figure 4-18 and Figure
4-19, as the samples S1 clearly shows the lowest turbidity. This suggests that the
processes involved in reducing turbidity during residence time in the filter require
more than 3 hours (the longest pause period using the tap) to be effective.
Figure 4-18 - Turbidity of filtrate - Filter without tap
Optimisation of household Biosand filters Richard Outhaite
- 52 -
Figure 4-19 - Turbidity of filtrate - Filter with tap
4.3.4. pH reduction
The filtrate of both filters showed a reduction in pH. This reduction in pH shows that
the filter is degrading the DOC within the raw water, thus producing organic acids
(Zheng et al, 2007). The mean pH values for the filtrate are shown in Table 4-6.
From this it can be seen that the tap does not increase the removal of DOC, however
similar improvement is seen in S2 and S3 that have had longer residence time, S2
showing greatest reduction in pH.
Table 4.6 - Mean filtrate pH for filter with & without tap Mean pH
No pause No pause No pause No Tap-S1 8.44 8.64 8.44 No Tap-S2 8.30 8.51 8.33 No Tap-S3 8.33 8.57 8.36 1hr pause 2hr pause 3hr pause Tap-S1 8.39 8.46 8.37 Tap-S2 8.30 8.59 8.32 Tap-S3 8.33 8.50 8.34
1hr pause 2hr pause 3hr pause
Optimisation of household Biosand filters Richard Outhaite
- 53 -
5. Pesticide removal
The first tests used lower concentrations of Metaldehyde within the raw water. The
filter did not consistently or effectively remove the Metaldehyde from the water. A
paired t-test showed that the mean amount of Metaldehyde removed was not
significant, compared to the influent concentration (p = 0.78). A large variation was
recorded in the maximum and minimum amount of Metaldehyde removed (Table 5-
1).
Table 5.1 - Reduction in Metaldehyde concentration at 180µg/L & 1000µg/L
Concentration Tap / No Tap N
Mean reduction in Metaldehyde
concentration (95% CI)
Min–Max removal
Significance, p-value
180µg/L No Tap 12 6.1% (1.0%, 10.9%) 0.00% - 70.5% Tap 12 2.9% (0.5%, 6.3%) 0.00% - 93.3% 0.893
1000µg/L No Tap 12 11.7% (18.7%, 4.9%) 0.00% - 70.1% Tap 12 10.2% (16.6%, 3.8%) 0.00% - 73.4% 0.731
There was no significant reduction in the amount of Metaldehyde removed when
using the filter with a tap at the high and low concentrations, (p = 0.731 and 0.893
respectively). Some of the results even showed higher concentrations than those in
the influent. This suggests that the concentration of the pesticides added to the filter
did not remain evenly distributed throughout the process, giving rise to large
variations in concentration. This variation is shown for the filter with tap in Figures 5-
1 and 5-2 and for the filter without tap in Figures 5-3 and 5-4.
Optimisation of household Biosand filters Richard Outhaite
- 54 -
Figure 5-1 – Concentration of Metaldehyde in filtrate at 180µg/L – Filter without Tap
Figure 5-2 Concentration of Metaldehyde in filtrate at 1000µg/L – Filter without Tap
Optimisation of household Biosand filters Richard Outhaite
- 55 -
Figure 5-3 Concentration of Metaldehyde in filtrate at 180µg/L – Filter with Tap
Figure 5-4 Concentration of Metaldehyde in filtrate at 1000µg/L – Filter with Tap
Optimisation of household Biosand filters Richard Outhaite
- 56 -
Despite the filter being mature when the tests were carried out, the biofilm and
schmutzdecke do not seem able to breakdown the Metaldehyde. During this period,
the filters still recorded high levels of bacteria removal (mean coliform removal of
95.2% for the unseeded filter, 99.3% for the seeded filter). This shows that the
presence of pesticides within the water does not affect the activity of the bacteria
within the filter, as coliform removal rates are still high.
There was also no significant difference between the samples 1,2 and 3, which was
previously identified during the bacteriological testing. This indicates that increase in
residence time does not affect the processes involved in breaking down the
pesticide, or that the residence time of 24hrs is not sufficient for the process to have
an effect.
The failure of the filter to significantly reduce the concentration of Metaldehyde could
be because the concentration was too great for the filter, and the removal rates were
not noticeable. The high concentrations that were higher than the influent
concentration could also have been due to errors during the extraction and analysis
process. The solvent used to dissolve the Metaldehyde (Dichloromethane) into the
vials used for analysis is extremely reactive. As such any exposure to air will reduce
the volume of solvent very quickly. Any reduction in volume of solvent will
concentrate the sample further giving rise to errors within the sample. However the
distribution in concentration is too large to be attributed to this, therefore there may
be another mechanism, which concentrates the Metaldehyde into certain parts of the
filtrate.
Optimisation of household Biosand filters Richard Outhaite
- 57 -
6. Conclusions and Recommendations
6.1. Maturation Study
The filter seeded with sand from an existing filter showed greatly improved
performance, and identified that the process is greatly speeded up by adding 50% of
the sand from another existing filter. The effect of putting smaller volumes of seeded
sand should be studied, as this would increase the number of filters that could be
seeded using the same volume of sand. The position of where the seeded sand was
installed in the new filter could be important an important consideration, as it could
affect how the biofilm develops. The effect of adding nutrients to the water was
identified as being another possible route to speeding the rate of maturation.
To confirm the performance of the filter on removing pathogens more specifically, the
samples should be assessed for E-Coli bacteria as well, rather than the effect on
broad indicator bacteria that total coliform counts give.
The characteristics of the raw water being filtered will also have an impact on how
the biofilm develops. DO was shown to be consumed much more quickly in the initial
stages of maturation, therefore variations in DO and nutrients in the water can have
an effect on the maturation process, possibly making the effect of seeding even
more prevalent.
As the filters were not required to be cleaned during the process due to low turbidity
of raw water, the performance of how the filter performs after it is cleaned should
Optimisation of household Biosand filters Richard Outhaite
- 58 -
also be studied, as the biofilm and schmutzdecke will be disturbed during this
process.
6.2. Tap study
The tests conducted on the filters with the tap proved that the tap did not adversely
affect the performance of the filter to which it was fitted, and the pause in the flow
that it introduced did not reduce the bacterial quality of the filtrate. The filter that was
seeded in the previous study was the one fitted with a tap; therefore the filter may
have been performing better than the other one before the study. It is therefore not
possible to definitively say that the performance of the filter with a tap was better
than the one without a tap. However, it can be shown that the tap did not have a
detrimental effect on the filter it was fitted to. This therefore should help to eliminate a
pathway to recontamination of the filtrate, as storage of the water after it has been
filtered is no longer required.
The provision of the tap may also have an impact on the maturation of the filter, so
should be incorporated into a further study of how the filter mature when provided
with a tap.
Again, incorporating the removal of E-coli to confirm the removal rate of pathogenic
bacteria as well as total coliforms would help to confirm the performance of the filter.
Optimisation of household Biosand filters Richard Outhaite
- 59 -
6.3. Pesticide Study
The study into pesticide removal did not confirm whether the filter could significantly
remove Metaldehyde from the raw water due to very wide ranging concentration of
Metaldehyde in the filtrate. Despite this, certain results from the test did show
marked reductions in the concentration of the pesticide.
It is therefore possible that at lower concentrations, different levels of performance
may be noticed and a better understanding of the filter to remove such contaminants
can be gained.
Different methods to analyse the samples should also be tested. There are currently
five approved methods for use by the environment agency. Further research into
other procedures using GC-MS should be carried, as well as analysis by High
performance Liquid Chromatography (HPLC).
Optimisation of household Biosand filters Richard Outhaite
- 60 -
7. References
APHA (2005). Standard Methods for the Examination of Water & Wastewater: Contennial Edition 21st ed., American Technical Publishers, New York, USA. ASCE (1991) Slow sand filtration. Logsdon, G.S. (Ed). American Society of Civil Engineers, New York, USA. Barrett, J.M., and Silverstein, J. (1988) The effects of high carbon and high coliform feed waters on the performance of slow sand filters under tropical conditions, in Slow sand filtration: recent developments in water treatment technology. Edited by N.J.D. Graham, Ellis Horwood Ltd., Chichester, UK pp 231–252. Baumgartner, J., Murcott, S. & Ezzati, M. (2007) Reconsidering ʻappropriate technologyʼ: the effects of operating conditions on the bacterial removal performance of two household drinking-water filter systems. Environmental Research Letters, 2 (2), pp 24-33. Brock, T.D. & Madigan, M.T. (1991) Biology of Microorganisms. 6th Edition. Prentice Hall, New Jersey.
Burlingame, G. A., Mcelhaney, J., Bennett, M. & Pipes, W. (1983) Bacterial Interference with Coliform Colony Sheen Production on Membrane Filters, Applied And Environmental Microbiology, Vol 47(1), pp 56-60. Buzunis, B.J. (1995) Intermittently Operated Slow Sand Filtration: A New Water Treatment Process. University of Calgary, Canada. Centre for Alternative Water and Sanitation (2009) Biosand Filter Manual Design, Construction, Installation, Operation And Maintenance, Calgary, Canada.
Optimisation of household Biosand filters Richard Outhaite
- 61 -
Elliott, M.A., Stauber, M.A., Koksal, F., Ortiz, G.M., 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), pp 2662-2670. Environment Agency (2010) The Environment Agencyʼs position on Metaldehyde, Accessed at: http://www.environmentagency.gov.uk/static/documents/Business/Metaldehyde_position_statement.doc-27.1.10.pdf on 7 April 2010.
Extension Toxicology Network (1996) Pesticide Information Profiles: Metaldehyde. Accessed at http://extoxnet.orst.edu/pips/metaldeh.htm on 4 April 2010. Haarhoff, J. & Cleasby, J. (1991) Biological and Physical Mechanisms in Slow Sand Filtration. In Slow Sand Filtration. American Society of Civil Engineers, New York, USA, pp. 19-68. House, S. & Reed, R. (1997) Emergency Water Sources. WEDC, Loughborough, UK, p.171. Huisman, L. & Wood, W.E., (1974) Slow Sand Filtration, WHO, Geneva, Switzerland. International Program on Chemical Safety, (1990) Metaldehyde. Accessed at http://www.inchem.org/documents/pims/chemical/pim332.htm#SectionTitle:3.3%20%20Physical%20properties on 5 April 2010. Kegley, S.E., Hill, B.R., Orme S., & Choi A.H., PAN (2010) Pesticide Database, Pesticide Action Network, San Francisco, North America, http://www.pesticideinfo.org.
Latagne, S.D. & Ambiental, A., Research on clay filters impregnated with colloidal silver promoted by Potters for Peace, USAID, New York.
Optimisation of household Biosand filters Richard Outhaite
- 62 -
Manz, D. (2007) BioSand Water Filter Technology: Household Concrete Design, University of Calgary, Canada. Mehta, D. & Hawley, M.C. (1969) Wall Effect in Packed Columns, Industrial & Engineering Chemistry Process Design and Development, Vol 8(2), pp 280-282. Palmateer, G., Manz, D., Jurkovic, A., McInnis, R., Unger, S., Kwan, K.K. & Dutka, B.J. (1999) Toxicant and Parasite Challenge of Manz Intermittent Slow Sand Filter. Environmental Toxicology, Vol. 14, pp 217- 225.
Potters for Peace (2008) www.pottersforpeace.org, accessed 19 March 2010 Prüss-Üstün A, Bos R, Gore F, Bartram J. (2008) Safer water, better health: costs, benefits and sustainability of interventions to protect and promote health, World Health Organization, Geneva. Stauber, C.E., Elliot, M.A., Koksal, F., Ortiz, G.M., Digiano, F.A., Sobsey, M.D. (2009) A randomized controlled trial of the concrete biosand filter and its impact on diarrheal disease in Bonao, Dominican Republic. The American Journal of Tropical Medicine and Hygiene, 80(2), pp 286-293. Tebbutt, T.H.Y., (1997). Principles of Water Quality control, 5th edition., Butterworth-Heinemann, Oxford, UK. Tiwari, S.K., Schmidt, W.P., Darby, J., Kariuki, Z.G & Jenkins, M.W. (2009) Intermittent slow sand filtration for preventing diarrhoea among children in Kenyan households using unimproved water sources: randomized controlled trial, Tropical Medicine & International Health: TM & IH, 14(11), pp 374-1382.
The Sphere Project (2004) Humanitarian Charter and Minimum Standards in Disaster Response, The Sphere Project: Geneva, Switzerland.
Optimisation of household Biosand filters Richard Outhaite
- 63 -
United Nations Children's Fund (2006) Annual Report, New York, US. Accessed at http://www.unicef.org.uk/publications/pdf/ar2006.pdf 4 February 2010.
United Nations Development Program (2006) Beyond scarcity: Power, poverty and the global water crisis, Palgrave Macmillan, Hampshire UK.
United Nations Water (2008) UN Water Annual report 2008. Accessed at http://www.unwater.org/downloads/annualreport2008.pdf, 17 March 2010.
Visscher, J.T. (1990) Slow Sand Filtration: Design, Operation and Maintenance,
Journal of American Water works Association, Vol. 82 (6), pp 67-71.
Weber-Shirk, M. & Dick, R. (1997) Biological mechanisms in slow sand filters. Journal American Water Works Association, Vol 89 (1), pp 87-100.
Wheeler, D., Bartman, J. & Loyd, B.J. (1988) The removal of viruses by filtration through slow sand filtration : recent developments in water treatment technology. Graham, N.J.D. (Ed.) John Wiley & Sons, New York, USA.
World Health Organisation (2009) How much water is needed in emergencies: Technical Note for Emergencies 9, Geneva, Switzerland.
World Health Organisation (1996) Data sheets on pesticides No. 93: Metaldehyde, Accessed at http://www.inchem.org/documents/pds/pds/pest93_e.htm#1.4 on 3 April 2010.
Zheng, X., Liang., Z. & Jekel, M. (2009) Effect of slow sand filtration of treated wastewater as pre-treatment to UF, Desalination, Vol. 249 (2), pp 591-595.
Optimisation of household Biosand filters Richard Outhaite
- 64 -
8. Appendices
Optimisation of household Biosand filters Richard Outhaite
- 65 -
Appendix A
Sieve Analysis Data
Optimisation of household Biosand filters Richard Outhaite
- 66 -
Sieve Analysis to BS 1377-2:1990 - Methods of test for soils for civil engineering purposes
Sand type RH 45 - Unseeded
Sand source Redhill quarry, Surrey
Mechaical shaking time 15 minutes > 10 minutes = OK
Weight of sample 200g > 150g = OK
Sieve Size (micron)
Weight in sieve (g)
Weight retained (g)
% Retained % Retained (cumulative)
% Passing sieve
1000 0 0 0.00% 0.00% 100%
710 0.2 0.2 0.10% 0.10% 99.90%
500 0.12 0.32 0.06% 0.16% 99.84%
355 50 50.32 25.00% 25.16% 74.84%
250 73.8 124.12 36.90% 62.06% 37.94%
180 55.28 179.4 27.64% 89.70% 10.30%
125 17.8 197.2 8.90% 98.60% 1.40%
106 1.2 198.4 0.60% 99.20% 0.80%
63 1.46 199.86 0.73% 99.93% 0.07%
<63 (Pan) 0.12 199.98 0.06% 99.99% 0.01%
Red denotes amount in sieve more than 50g. Sieve in two parts
Total Sieved weight within 0.1% of Total = OK
Optimisation of household Biosand filters Richard Outhaite
- 67 -
Sieve Analysis to BS 1377-2:1990 - Methods of test for soils for civil engineering purposes
Sand type RH 45 - Seeded
Sand source Redhill quarry, Surrey
Mechaical shaking time 15 minutes > 10 minutes = OK (Clause 9.3.4.5)
Weight of sample 200g > 150g = OK (Clause 9.33 & Table 3)
Sieve Size (micron)
Weight in sieve (g)
Weight retained (g)
% Retained % Retained (cumulative)
% Passing sieve
1000 0 0 0.00% 0.00% 100%
710 0.36 0.36 0.18% 0.18% 99.82%
500 4.1 4.46 2.05% 2.23% 97.77%
355 74.8 79.26 37.40% 39.63% 60.37%
250 77.9 157.16 38.95% 78.58% 21.42%
180 28.48 185.64 14.24% 92.82% 7.18%
125 12.2 197.84 6.10% 98.92% 1.08%
106 0.9 198.74 0.45% 99.37% 0.63%
63 1.16 199.9 0.58% 99.95% 0.05%
<63 (Pan) 0.1 200 0.05% 100.00% 0.00%
Red denotes amount in sieve more than 50g. Sieve in two parts
Total Sieved weight within 0.1% of Total = OK
Optimisation of household Biosand filters Richard Outhaite
- 68 -
APPENDIX B
Determination of Metaldehyde in water using gas
chromatography and mass spectrometry.
Detailed Methodology
Optimisation of household Biosand filters Richard Outhaite
- 69 -
Introduction
To detect and analyse the amount of Metaldehyde remaining in the filtrate from the
Biosand filters, the Environment Agency - Method A (EA, 2009) was followed. Prior
to analysis, the samples were processed using J.T baker solid phase extraction. This
section will detail the methods used throughout.
Methodology
Solid Phase Extraction
1. Condition the solid phase cartridge using 10mL of Ethanol. Elute the cartridge
with the Ethanol and discard the Eluate. Do not allow the meniscus to drop
below the level of the cartridge.
2. Elute the cartridge with 2mL of de-ionised water. Discard the eluate. The
meniscus of the water should still remain above the level of the cartridge.
3. Elute the cartridge with the sample to be analyzed. Typically the sample size
is 250mL, depending on the concentration of Metaldehyde in the water. As the
concentration in this experiment is higher than typical background levels,
50mL was used. Discard the Eluate.
4. Dry the sample thoroughly by passing air through the cartridge. This takes up
to 45 minutes.
5. Elute the cartridge with 2mL of Dichloromethane. Collect the eluate.
6. Evaporate the combined Dichloromethane sample. Nitrogen gas was used in
the blow-down process, as it is the most inert gas. The sample should be
Optimisation of household Biosand filters Richard Outhaite
- 70 -
evaporated to the required volume, typically 0.5mL. To enable proper
sampling by the GC-MS auto-sampler 1mL was used.
7. Add 5.0μL of working internal standard for every 0.5mL of sample volume. As
the sample prepared was 1.0mL, 10μL of or working internal standard was
added. The sample is now ready for GC-MS determination.
GC-MS Determination & Analysis
1. Set up the GC-MS with the following parameters
• Injection temperature – 260°C.
• Oven Temperature – Initial temp at 50°C, increase at a rate of 20°C per
minute to 260°C. Hold at 260°C for 1 minute.
The other parameters used in the GC method are shown below in Figure X.
Figure 8-1 - Gas Chromatography parameters
Optimisation of household Biosand filters Richard Outhaite
- 71 -
• The Mass spectrometer was run in Selective Ion Monitoring mode to give
clean peaks during analysis. The monitoring was carried out for 12 minutes
(Target time for Metaldehyde is around 7 minutes) and the monitoring was
subject to a three minute delay to avoid detecting the solvent peak from the
Dichloromethane and allow clearer analysis of the Metaldehyde peak.
• Once the chromatogram has been drawn, confirm the presence of
Metaldehyde. Target ions no 45 and qualifier 89 confirm the presence of
Metaldehyde in the sample.
A sample of the chromatogram containing Metaldehyde is shown below in
Figure X.
Figure 8-2 - Chromatogram for Metaldehyde sample with internal standard
Optimisation of household Biosand filters Richard Outhaite
- 72 -
2. The area under the peak of the chromatogram for each sample is then
compared to a chromatogram of a standard solution containing 100µg/mL of
Metaldehyde. An example of the integrated areas, with the corresponding
parameters, is shown in Figure X. The ratio of the two areas can then be used
to calculate the amount of Metaldehyde in the sample, therefore the
concentration in the filtrate.
Figure 8-3 - Integral of chromatogram to determine quantity of Metaldehyde