e.coli over the pathway of the filter
TRANSCRIPT
Household Water Treatment with Gravity Driven Membrane Filters: Field Study Kenya 2011-12
GDM Kenya - an interdisciplinary project of Eawag, the Swiss Federal Institute of Aquatic Science and Technology
Selina N. Derksen-Müller, Maryna Peter-Varbanets, Francis Kage, Joseph Owino, Regula Meierhofer, Wouter Pronk, Richard Johnston
Field Study
In order to test GDM filtration under real conditions, a field study
was launched in Kenya in May 2011. In four different field sites
where people where consuming untreated water, 24 prototypes
were distributed to households. The raw water covers a broad
range of water quality, from tap water in Nairobi to high-turbidity
surface waters in rural areas. After one year of operation, 92%
of the filters are still functioning and none have failed due to
technical reasons. Despite the challenging raw water quality, the
flux has stabilized in all filters.
In collaboration with the Kenya Water for Health Organization
(KWAHO), several water quality parameters and flux were
monitored on a monthly basis.
Gravity-Driven Membrane Filtration
Under certain low water pressure conditions, ultra-filtration
membranes can be sustainably operated for years, even without
backwashing or cleaning1. Flux stabilization occurs due to the
formation of a biofilm on the membrane surface with channels and
cavities that counteract the resistance caused by deposition and
fouling. While the flux of such a gravity-driven membrane (GDM)
filtration would be considered inadequate for industrial applications,
it is ideal for household water treatment. The GDM Kenya project
aims to develop household filters for use in developing countries.
Prefilter
Membrane tank
Membrane Micodyn-Nadir, (Germany)
150 kDa cut-off
(~20nm pore size)
0.69 m2 surface area
Clean water tank
sampling points
2
4
3
1
References
1 Peter-Varbanets et al., Stabilization of
flux during dead-end ultra-low pressure
ultrafiltration. Water Research 2010,
44(12), 3607-3616.
2 Wright et al., Household drinking water
in developing countries: A systematic
review of microbiological contamination
between source and point-of-use.
Tropical Medicine and International
Health. 2004;9(1):106-17
User Behavior
The record from the dataloggers also reveals interesting
information on the user behavior like the volume filtered and the
time of day of filtration.
Filter use declined somewhat in the days following the distribution
but then remained stable throughout the year.
Families with a median of 7 members filled their filter almost 4
times per week with a mean volume of 8.5 liters of water per
filling. The median volume filtered per person almost 4 liters per
week, so it is clear that filters were not used to their full capacity.
Consumption per person was the lowest for tapwater in Nairobi
where most people do not spend their day at home.
02
46
81
0
[fill
s/w
ee
k] [L
/d] [L
]
pond river dug well borehole tapwater
48 datasets of average 30 days in the first 9 months of the field study
volume filtered per day volume per capita per week
mean fills per week mean filling volume
Flux
Water flux was measured using pressure meter dataloggers (Solinst Levelogger Model 3001)
continuously recording the water level in the membrane tank. These allow precise analysis of
the flux and the behavior of the users. All filters reached stable flux within a few days and
there was no systematic decrease observed over the whole year. The flux was calculated for
three different heads, representing the flux shortly after filling (12 mbar), at an intermediate
level (7.5 mbar) and towards the end of a filtration cycle (2.2 mbar). In the worst feedwater
conditions, a filter with 0.5m2 of membrane can still produce 20 liters of water while in more
favorable conditions, 90 liters or more can be filtered per day.
Water Quality
In all raw water sources the fecal indicator bacteria E.Coli could be detected at
times. However, the surface waters and the dug wells showed the highest
contamination with geometric means from 110 to 180 CFU/100ml over all samples.
In 72% of permeate samples no E.Coli was detected in 100ml. The membrane
forms an absolute barrier for bacteria and the integrity of all filters was controlled.
Therefore, the low levels of E.Coli observed in some of the filters are most likely
Recontamination at the tap
of the filter can be observed
in most filters and is also
known to be a problem in
other water treatment
systems2.
due to recontamination
from untreated water or
particles entering the
clean water tank in an
unexpected way. Ex-
amples of E.Coli counts
over the pathway of the
filter are displayed in the
graph. Bacteria growth in
the clean water tank is
also plausible but is not
likely the main
explanation, as
investigation of biofilms
on the clean water tank
walls yielded no E.Coli.
05
10
15
20
Flu
x [L/m
2/h
]
pond river dug well borehole tapwater
data of one exemplary filter for each watersource, mean of 20-54 days continuous monitoring
2.2 mbar 7.5 mbar 12 mbar
Outlook
Based on the experience of the field trials, an improved 'second
generation' GDM filter prototype is being designed in partnership
with Zurich University of Arts and commercial partners.
Next generation filters will be tested in the field in 2013 in a small
number and the first large scale study with the final design of the
filter is planned for 2014.
Further research will be conducted and is already ongoing on the
user acceptance, economic strategies, distribution networks and
design aspects of the filter.
11
01
00
100
0
E.C
oli
[CF
U /
10
0m
l]
1 2 3 4raw water storage vessel membrane tank permeate tap
Pond River Dug well
Borehole tap (feedwater)
data of one exemplary filter of each watersource
E.Coli over the pathway of the filter
Water and Health Conference 2012: Science, Policy and Innovation, Chapel Hill
Contact
Maryna Peter-Varbanets
Eawag, Water and Sanitation in
Developing Countries (Sandec)
Überlandstrasse 133
P.O. Box 611
8600 Dübendorf
Switzerland
Tel. +41 (0)44 823 50 74
Fax +41 (0)44 823 53 99
www.eawag.ch/membranefilter