Development of a mobile water maker, a sustainable way to

produce safe drinking water in developing countries

L. Groendijka�, H.E. de Vriesb

aDepartment of Environmental and Water Technology, Van Hall Larenstein, Part of Wageningen

University and Research, Post Box 1528, 8901 BV Leeuwarden, The Netherlands

Tel. þ31 58 2846215; Fax þ31 58 2846423

email: leo.groendijk@wur.nlbAquario Water management BV, Smidsstraat 7, 8601 WB Sneek, The Netherlands

Tel. þ31 515 482811; Fax þ31 515 424156

email: info@aquario.nl, www.mobilewatermaker.com

Received 31 January 2008; revised accepted 15 May 2008

Abstract

Moreover, there is a growing demand for a simple, low capacity drinking water treatment used by local people in developing

countries to reduce mortality caused by water born diseases. To solve this problem a small portable water treatment unit with a

production capacity of approximately 500 L/day was developed. The unit can operate without the use of external electricity/

pumps/generators in order to operate completely independent. The mobile water treatment unit uses tubular ceramic membranes

in combination with an anodic oxidation process, powered by a solar power panel. The membranes are cleaned by simple flushing

without the use of chemicals and this enables a sustainable production of safe drinking water on every location in the world without

replacing polluted filter cartridges. Laboratory tests with several kinds of surface water and effluent of a wastewater treatment

plant (WWTP) showed a very stable production and gave very good results in the removal of sediments, colloidal material, bacteria

and viruses.

Keywords: UF membrane filtration; Disinfection; Solar energy; Sustainable Mobile Water Maker

1. Introduction

Unsafe drinking water is the main contribution to

an estimated 4 billion cases of diarrhea each year, caus-

ing 1.8 million deaths mainly among children younger

than 5 years [1]. Even at locations where water is safe

at the source, it is often contaminated during tapping,

collection, storage and use unless it is protected by resi-

dual disinfection.

Safe drinking water is more and scarcer because of

the growing demand and diminishing resources in a lot

of developing countries. Also the growth of population

causes problems in the availability of clean and safe

drinking water.

Besides the amount of water, also the water quality

is a growing problem because the lack of good� Corresponding author.

Presented at the Water and Sanitation in International Development and Disaster Relief (WSIDDR) International

Workshop Edinburgh, Scotland, UK, 28–30 May 2008.

0011-9164/0x/$– See front matter # 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.desal.0000.00.000

Desalination 251 (2010) 106–113

sanitation. Contamination of water sources by feces

and other excrements is a very serious problem in

developing countries.

With the use of ultra filtration (UF) membranes,

combined with disinfection, bacteria and viruses are

removed from the water resulting in safe drinking

water free from water borne diseases.

Main research question is how to design a good

stand-alone membrane filtration system which oper-

ates without the use of pumps, electricity and chemical

cleaning agents for the membranes so that the system

can work independent from the delivery of spare parts

and electrical generators. A lot of systems are available

but most of times you are depending on disposable fil-

ters, cartridges or disinfecting or membrane cleaning

chemicals [6].

The main problem with membrane filtration sys-

tems is fouling of the membranes. Bio-fouling, cake

formation, pore blocking and adsorption of fouling

substances are serious problems which are generally

solved with the use of several techniques like mechan-

ical and chemical cleaning. For mechanical cleaning

periodical backwash combined with cross flow, air

flush and forward flush steps are generally used. Che-

micals for cleaning are often not available so this tech-

nique can cause problems because the lack of these

chemicals in the country. Also the costs of cleaning

agents are very high.

Beside this biological contamination which is fully

removed, there is also the possibility of chemical pol-

lution in the feed water. In this research this aspect is

not considered because this needs treatment with use

of, for example, activated carbon or high energy

demanding reverse osmosis filtration. If the water is

polluted with harmful toxic chemicals the unit will not

be producing safe water.

2. Experimental setup

In the laboratory of Van Hall Larenstein, a mem-

brane filter test unit including an anodic oxidation unit

was built up in sections. The feed water was supplied

via a tank a few meters above the module. The unit was

tested with polluted water from different sources.

Various UF membranes with a pore size between

10 and 100 nm were tested with respect to their fouling

behavior.

At the end a 40 nm tubular ceramic membrane

module (Hyflux Ceparation) was selected for further

research because of the easy way of cleaning this mem-

brane and her mechanical strength.

Hollow fibers, operated outside in appeared very

difficult to clean after only a few hours of production.

Flat sheet Kubota submerged membranes clogged very

quickly and were difficult to clean without continuous

air scouring or mechanical brushing.

The UF membrane unit was operated in a dead-end

configuration which means that all the water which

enters the module has to pass the membrane. Particles

bigger than the pore size of the membrane are retained

inside the membrane tubes in the module.

The study was carried out with a feed water storage

tank at a level of 3 m above the membrane filtration unit

which results in a constant Trans Membrane Pressure

(TMP) of 0.3 bar. In the lab experiments, feed water

was supplied from a 6 m3 underground storage tank out-

side the building. The level in the feed water tank was

kept constant by a level controller which could switch

on/off a centrifugal pump connected to the storage tank.

A drawing of the test unit is shown in Fig. 1.

All the valves, including the clean water tap, are

manually operated by hand. The bicycle pump is con-

nected by a quick fit connector to the air pressure con-

tainer. The pressure indicator on this container is not

drawn in the figure. The control box is controlling the

charging of the battery with the solar power panel and

also controls the disinfection level of the filtered water.

2.1. The feed water

The water used in the tests was surface water from

the Prinses Margriet channel near the city of Leeuwar-

den. Besides surface water, tests were also done with

effluent from the WWTP of the city of Leeuwarden.

Before the raw feed water was pumped from the

outdoor underground storage tank into the water supply

tank inside the laboratory it passed a raw filter nylon

screen of 0.13 mm to remove big particles from the

water such as leaves, frogs, fishes, snails and other

material which could clog the membrane inlet tubes.

The membrane unit was connected to the raw water

storage with use of Ø40 mm hoses.

The unit was provided with three valves in order to

be able to clean the membranes with a backwash and

forward flush (see Fig. 1).

L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113 107

2.2. Membrane module

The membranes used in the test were Hyflux

Ceparation ceramic tubular UF membranes [2]. Pore

size is 40 nm and clean water permeability according

to the specification of the supplier 500 l/m2 h bar. The

inner tube diameter is 2.8 m. The membrane module

contained 210 tubes with an effective length of 43

cm resulting in a membrane area of 0.8 m2 [2]. The

membrane insert was built in a normal PVC housing,

built up from general available sewer PVC materials.

If the production tap is closed, the permeate vessel

besides the unit is filled by gravity and in this way

about 10 L of permeate becomes available for a back-

wash flush. A back flush is a high permeate water flow

through the pores of the membrane to the raw water

side in the tubes. Production is inside out and backwash

is outside in.

2.3. The disinfection unit

The permeate of the membrane unit flows to a

2Bsure disinfection unit, working as anodic oxidation

system. This 2Bsure unit was developed by Bright

Spark, Joure, The Netherlands. A couple of elec-

trodes are built in a PVC pipe and this reactor is

responsible for the disinfection of the water. The dis-

infection unit is powered by a 6 V battery. The

2Bsure unit is switched on when the water tap is

opened, so in case no water is produced, no electri-

city is used.

The permeate from the membrane filtration unit

was treated with use of anodic oxidation. During this

process an electrical field (low voltage) between two

electrodes, a cathode (�) and anode (þ) will generate

some chemical reactions. In the water also some chlor-

ide is present.

Membrane filter Tap with switch

Feed water storagebag/ tank 50 – 300 l

Valve

Forward flush valve

Disinfectionunit

Clean water

Pressure vessel air

Bicycle pump

Solar panel

Battery 6 V Control box

One way valve

Push valve

Fig. 1. Experimental laboratory test set up.

108 L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113

At the anodic reaction chloride (Cl�) en hydroxide

(HO�) ions flow towards the anode where they will

oxidize.

At the anode chloride and hydroxyl radicals will be

formed which combine to the disinfectant hypochlor-

ous acid (HOCl), a strong and fast-acting oxidizer.

Microorganisms will be damaged by the HOCl.

The HOCl kills a variety of germs and is therefore

widely used to disinfect drinking water. A further

advantage is that the germs do not become resistant

to the sodium hypochlorite.

The disinfectant returns only to his original form

(NaCl) when the radical oxygen atom is exchanged.

Advantage of disinfection with HOCl is that there

is also a post-disinfection after the water will leave the

Mobile Water Maker. Storage tanks and bottles are dis-

infected with the produced water and recontamination

will be reduced [3,4].

2.4. Energy for disinfection

The energy, needed for the anodic oxidation is gen-

erated by a solar panel of 15 pW. This 12 V solar panel

charges a battery in the system which will store enough

energy for a few days of production. So the system is

self supporting with energy.

If 220 V AC is available the unit can be charged and

operated without solar energy. Also an external 12 V

car battery can be used.

3. The tests

3.1. Laboratory tests

The filtration unit is operated during 10 h a day to

simulate the operation in practice (see Table 1). In the

morning the module is first backwashed with a little

amount of permeate from the housing of the membrane

module. The driving force for this backwash is gener-

ated by an air pulse from a compressed air container in

the unit. The compressed air container is pumped up to

a pressure of about 4–6 bar with use of a hand powered

bicycle pump. During the forward flush with feed

water a few pulses compressed air are released to the

permeate side of the membrane. During this backwash

the water tap is closed so the permeate will be pumped

backwards to the feed side of the membrane. After

this hydraulic cleaning step, the unit is switched on for

production. During the day some interruptions in the

production are simulated by closing the tap valve.

The permeate flow out of the unit is measured and

registered automatically with a magnetic flow meter.

This meter is calibrated by hand with a flask and a stop-

watch. The signal of the magnetic flow meter is

recorded in a data logger and stored in an excel file.

Samples of the feed water and the permeate were

analyzed by the Dutch Water Company Vitens. Samples

of the permeate were taken before and after disinfection.

The running time is limited to 10 h a day and the

filtration tests went on for about 52 weeks.

The course of the permeate flow during the filtra-

tion runs gives an indication for the fouling behavior

of the membranes.

The concentrate or brine was drained into the sewer.

3.2. Field tests

In October 2006, a first prototype of the unit was

used for a short field test for 2 weeks at

Restaura&ion, a village at the Dominican Republic

(Fig. 2). A group of field workers of World Servants

Europe, young people in the age of 20–30 years, tested

the unit under field conditions. During this field test a

lot of practical problems were encountered and the

results were used to improve the prototype.

The feed water came from the river beside the camp

and the quality of the feed water was varying a lot

because of heavy rain weather from time to time.

Table 1

Procedure of operating the unit

08.00 am

Pulsed backwash with permeate combined with a feed

water forward flush

08.03 am

Start production of permeate during the day with

continuous filtration or sometimes with an start stop

program that simulates the use by a person several times

a day

17.00 pm

Stop production, closure of the tap water valve

17.00 pm–8.00 am

Relaxation during the night till pulsed backwash at 8.00 in

the morning

L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113 109

In the first water maker, the solar panel was inte-

grated in the housing of the unit. This meant that the

unit must be placed in the sunshine during the day.

Now the panel is separated from the water maker; it

is connected with a 20 m long cable to the unit so the

unit can be installed inside a building.

The electrical unit in the housing caused also pro-

blems because of the inflow of rainwater during stormy

weather and heavy rainfall that was not foreseen. So the

electrical circuit was better insulated in the next

prototype.

Also the transport of a few hundred liters of water to

the vessel at 3 m height appeared a hard job. For that

reason a simple hand pump with a capacity of 30 L/min

is incorporated in the unit.

4. Results

4.1. Filtration tests with effluent water

The membrane and disinfection units were tested in

the laboratory during 4 months with effluent water as

feed water. This effluent was stored in 6 m3 under-

ground storage tank and pumped up to the feed vessel

at 3 m height above the filtration unit.

As can be seen in Figs. 3–5 there is a slow decrease

in permeate production rate during the day. At the end

of the day the unit was stopped and after relaxation dur-

ing the night the production was started again in the

morning with a pulsed backwash and forward flush.

This relaxation and cleaning operation appeared suffi-

cient to increase the production rate again to the initial

0

1

2

3

4

5

6

0

1.5 3

4.5 6

7.5 9

10.5 12

13.5 15

16.5 18

19.5 21

22.5 24

Time (h)

Pro

du

ctio

n (L

/min

)

Fig. 4. The production rate in case the filtration is started

at 8 am without a pulsed backwash on 22/04/2006 and

without a forward flush. Production rate is lower without

the backwash.

Fig. 2. On the left picture the first Mobile Water Maker prototype during the field test in Restaura&ion in the Dominican

Republic in October 2006. On the right picture the second prototype in November 2007 [5].

0

1

2

3

4

5

6

0

1.5 3

4.5 6

7.5 9

10.5 12

13.5 15

16.5 18

19.5 21

22.5 24

Time (h)

Pro

du

ctio

n (L

/min

)

Fig. 3. The production rate during a standard day, 21/11/

2006. Filtration starts at 8 am is started with a pulsed

backwash combined with forward flush.

110 L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113

value of the day before. This indicates that the fouling

which is build up on top and/or inside of the mem-

branes during the continuous filtration is reversible and

easy to remove.

Obviously, a flux of 60 L/m2 h (permeability 250 L/

m2 h bar) is easy to handle for ceramic membranes with

this kind of feed solution. A permeability of 250 L/m2 h

bar on effluent is only a factor 2 lower than the clean

water permeability of the membrane. This is also an

indication of the low fouling behavior under these

process conditions.

Figs. 3–5 show the influence of the relaxation per-

iod during the night and the hydraulic backwash and

forward flush on the production rates. The measure-

ments show the filtration behavior for three subsequent

days with a constant influent water quality.

After the initial start up on the day, the production

rate gradually decreases from 1.5 to 1 L/min.

Comparing Figs. 3 and 4 it is shown that the initial

backwash and forward flushes give raise to higher

production rates during the day.

Intermediate pulsed backwash flushes do not give

rise to higher production rates (see Fig. 5).

Relaxation over a prolonged period of time fol-

lowed by a combined forward and backwash flush seem

the best procedure for long-term sustainable operation.

Optimum time for relaxation and hydraulic clean-

ing are not yet investigated.

Fig. 6 shows the initial (morning) and final (after-

noon) flow rates during 2 months of operation with

surface water.

It can be seen that the initial flow rates are stable

over this period (flow at 30/11/2006: 72 L/h; flow at

30/01/2008: 72 L/h).

At some days lower values are obtained. This is

caused by skipping the pulsed backwash but only doing

a forward flush in the morning. Later this procedure

was improved by adding the pulsed back flush again

and the original initial flow rates were established

again. If a backwash after relaxation is not carried out

the production rate stays low. This is visible in the end

of Fig. 6.

0,00

10,00

20,00

30,00

40,00

50,00

60,00

70,00

80,00

90,00

100,00

30−11

−2007

1−12

−2007

2−12

−2007

03/12

/"07

04/12

/"07

05/12

/"07

06/12

/"07

07/12

/"07

11/12

/"07

12/12

/"07

17/12

/"07

18/12

/"07

14/01

/"08

16/01

/"08

21/01

/"08

28/01

/"08

30/01

/"08 2

Date

Pro

du

ct fl

owra

te in

L/h

Initial flow rate Final flow rate

No backwash

No backwash

Fig. 6. Initial (blue) and final (purple) daily flow rates during two months of production with surface water of the Prinses

Margriet channel.

0

1

2

3

4

5

6

0

1.5 3

4.5 6

7.5 9

10.5 12

13.5 15

16.5 18

19.5 21

22.5 24

Time (h)

Pro

du

ctio

n (L

/min

)

Fig. 5. Filtration is started on 23/04/2006 with a combined

pulsed backwash and forward flush. During the day at

14.00 pm an extra pulsed backwash and forward flush was

carried out but did not have the right effect.

L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113 111

4.2. Chemical and biological analyses of the

laboratory tests

The influent water and purified water is analyzed

periodically by the Vitens water company laboratory.

All samples are taken with a disinfection rate which

resulted in a 0.2 mg/L free chlorine in the produced

water. In Table 2 the results of the test with surface

water are presented.

5. Conclusions

The Mobile Water Maker can be used to produce

drinking water from biological polluted surface water.

Besides membrane filtration a disinfection step with

using in line anodic oxidation for the production of

hypochlorite is used in the treatment process. This dis-

infection also results in a post-disinfection of the water

collection containers.

Rejections of all kind of bacteria are high (>99%).

More accurate removal efficiencies can not be pre-

sented because the initial influent values for the content

microorganisms are not counted exactly. Only results

as >1000 are presented by the laboratory. The turbidity

of the permeate is very low, indicating that the perme-

ate is a clear solution without particles. The CFU of the

permeate sample is <100/mL. The WHO standard is

<100/mL [1]. The mean indicator for faecal pollution

E. coli says that there is a total removal of these bac-

teria. In the permeate the amount of Adsorbed Organic

Halogenated Compounds (AOX) increased through

the use of anodic oxidation. The free chlorine concen-

tration in the permeate is 0.2 mg/L so it will indicate

that the bacteria will be killed [5].

A sufficient relaxation period and a pulsed hydrau-

lic backwash and forward flush are sufficient to

prevent the ceramic membranes against fouling. A

hydraulic cleaning combined with a backwash is not

sufficient if a relaxation time of several hours is left.

Long-term test show no decrease of flow rate caused

by any fouling during 1 year of operation, even with

some breaks without production in summer holiday

(4 weeks) and Christmas holiday (3 weeks). No chemi-

cals and electricity from generators have to be used. A

15 peak Watt solar panel used at 1 day sunshine is big

enough to provide enough back up energy for 3 days

water production. With a 0.8 m2 ceramic tubular UF

module, a water column of 3.5 m is sufficient to

produce continuously 60 L/h. In more than 2 years

no parts have been replaced or repaired so the system

seems to be very reliable. A prototype of the complete

system is produced. The technical features are pre-

sented in Table 3 [5].

The produced water has an excellent quality and, if

no chemical pollutants are present, the water meets the

WHO standard for drinking water [1].

In combination with a built in activated carbon

filter stage the chemical pollutants can be removed

from the produced permeate. To avoid necessary

Table 2

The analytical results of the influent (surface water) and

produced water

Sample Influent

surface

water

Permeate

after

disinfection

Rejection

(%)

pH 8.5 8.5 –

Temperature 17 18 –

HCO3� (mg/L) 213 223 –

Turbidity (FTE) 100 <0.2 >99.8

Color

(mg Pt/Co/L)

51 49 4

Dissolved

oxygen

(mg/L)

9.5 10.1 –

Sulfate (mg/L) 59 64

Coli forms 37�C(#/100 mL)

>275 <1 >99.7

Escherichia coli

(#/100 mL)

>600 <1 >99.8

Clostridia

(#/100 mL)

>300 <1 >99.7

Aeromonas

30�C(#/100 mL)

>600 <1 >99.8

CFU 22�C(#/mL)

>1000 23 >97.7

Bacteriophages

(pve)

<1 <1 –

Na (mg/L) 73 71.2 2.5

Fe (mg/L) 2.96 <0.02 >99.3

Mn (mg/L) 0.148 <0.01 >99.3

AOX (mg/L) 49 77 �57

TOC (mg/L) 19 13 31.5

DOC (mg/L) 26 13 50

Free Cl2 (mg/l) 0 0.2 –

112 L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113

replacement of any carbon filters during operation this

technique is not applied because most water born dis-

eases are caused with using biological polluted water.

Acknowledgements

This research was funded by Technology Centre

North Netherlands in 2006–2008.

The author wishes to thank Prof. Dr. Walter van der

Meer from Vitens Water Company (analyses), Harry

van Dalfsen from Hyflux – Ceparations Liquid (cera-

mic module) Tjerk Kaastra from Bijl en de Jong (mem-

brane housing) and Maurice Tax from Bright Spark

(2Bsure disinfection unit) for the support during this

feasibility research.

Dr. Arie Zwijnenburg from Wetsus Technological

Top Institute Water is thanked for the stimulating dis-

cussions about membrane filtration and his contribu-

tion to this paper.

References

[1] http://www.who.int, March 2008.

[2] http://www.ceparation.com, February 2008.

[3] http://www.gallep.nl/ppt/2B_Sure_UK.ppt, February

2008.

[4] http://www.brightspark.nl, February 2008.

[5] http://www.mobilewatermaker.com, February 2008.

[6] Water purifier essential for hurricane preparedness, J.

Membr. Technol., September 2006.

Table 3

Features of the prototype Mobile Water Maker

Technical specifications:

Production capacity 300–600 L/day (10 h)

Power consumption 2 W

Solar panel capacity 20 pW

Battery capacity 7 Ah

Electrical connectors: Additional external 220 V AC

Additional external 12 V DC

(XLR 3 pin connector male)

Solar panel connector (XLR 3

pin connector male)

Premium filter Ceramic Tubular Filter HC08

Disinfection module Bright Spark Module 2BSafe 1

L/min

Regeneration time Once at every 24 h

Recovery 98%Tube fittings GEKA fast fit connectors

Body of the case Polymer Foamed Hood

Flight case material Alumina and stainless steel

Dimensions H � D � W (75 � 40 � 40) cm

Color case Blue, grey, red, black

Weight 22 kg

Wheels 2 rubber wheels 100 mm

Accessories 40 mm rubber coated flexible

tubes

Water source Surface water, biologically

polluted

Treatment efficiency:

E. coli 37 >99.999%E. coli >99.999%Enterococs >99.99%Total count 22�C >99.99%Total count 37�C >99.99%Chlorine in treated water <0.2 ppm

Turbidity <1 NTU

L. Groendijk, H.E. de Vries / Desalination 251 (2010) 106–113 113