Vous êtes sur la page 1sur 18

Agricultural Water Management 98 (2010) 385402

Contents lists available at ScienceDirect


Agricultural Water Management
j our nal homepage: www. el sevi er . com/ l ocat e/ agwat
Decentralised water and wastewater treatment technologies to produce
functional water for irrigation
Adriano Battilani
a,
, Michele Steiner
b
, Martin Andersen
c
, Soren Nohr Back
c
, J. Lorenzen
c
,
Avi Schweitzer
d
, Anders Dalsgaard
e
, Anita Forslund
e
, Secondo Gola
f
, Wolfram Klopmann
g
,
Finn Plauborg
h
, Mathias N. Andersen
h
a
Consorzio di Bonica di Secondo Grado per il Canale Emiliano Romagnolo CER, Via E. Masi, 8 40137 Bologna, Italy
b
Swiss Federal Institute of Environmental Science and Technology, Switzerland
c
Grundfos Biobooster A/S, Denmark
d
Netam, Israel
e
University of Copenhagen, Faculty of Life Science, Department of Veterinary Disease Biology, Denmark
f
Stazione Sperimentale per le Conserve Alimentari (SSICA), Italy
g
BRGM, Service Eau, Orlans, France
h
University of Aarhus, Faculty of Agricultural Sciences, Dept. of Agroecology and Environment, Denmark
a r t i c l e i n f o
Article history:
Available online 20 November 2010
Keywords:
Agricultural water reuse
Decentralised treatments
Faecal contamination
Heavy metals
a b s t r a c t
The EU project SAFIR aimed to help farmers solve problems related to the use of low quality water for
irrigation in a context of increasing scarcity of conventional freshwater resources. New decentralised
water treatment devices (prototypes) were developed to allow a safe direct or indirect reuse of wastew-
ater produced by small communities/industries or the use of polluted surface water. Water treatment
technologies were coupled with irrigation strategies and technologies to obtain a exible, easy to use,
integrated management of the system. The challenge is to apply new strategies and technologies which
allow using the lowest irrigation water quality without harming food safety or yield and fruit or deriva-
tives quality. This study presents the results of prototype testing of a small-scale compact pressurized
membrane bioreactor and of a modular eld treatment system including commercial gravel lters and
heavy-metal specic adsorption materials. Decentralised compact pressurised membrane biobooster
(MBR), was able to remove up to 99.99% of the inlet Escherichia coli and 98.52% of total coliforms. E. coli
was completely removed from irrigation water in 53% of the samples by the last MBR prototype ver-
sion. In 2008, 100% of samples fullled WHO standards (1989) and Global Gap requirement for faecal
contamination. MBR removed from inlet ow in the average 82% of arsenic, 82% of cadmium, 97% of
chromium, 93% of copper and 99% of lead. Boron and manganese were not removed from permeate. The
eld treatment system (FTS) proved to be effective against faecal contamination when applied with its
complete set up including UV treatment. The sole gravel lter and heavy metal removal device (HMR)
cannot provide sufcient and steadily treatment for microbial contamination. Nevertheless, gravel lter
can remove up to 60% of E. coli but the removal process was not stable nor predictable. FTS removed
76% of arsenic, 80% of cadmium and copper, 88% of chromium and lead, and up to 97% of zinc. Like the
MBR, boron and manganese were not removed from the irrigation water. Gravel lter directly fed with
secondary treated wastewater was found able to remove 41% of arsenic, 36% of cadmiumand lead, 48% of
chromium and 46% of copper. The residual heavy metals concentration after the gravel lter was further
reduced by the HMR: 35% for arsenic, 22% for cadmium, 25% for chromium, 33% for copper and 53% for
lead.
2010 Elsevier B.V. All rights reserved.

Corresponding author.
E-mail address: battilani@consorziocer.it (A. Battilani).
1. Introduction
Irrigation allows for crop production where water would other-
wise be a limiting factor. Moreover, under the pressure of market
globalisation and of the changes in dietary habits, agriculture is
evenmore depending onwater-demanding, highincome vegetable
and fruit crops. Therefore, in arid and semi-arid areas of the EU
0378-3774/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.agwat.2010.10.010
386 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
irrigated agriculture is fundamental to the local and national econ-
omy. In humid and temperate areas, irrigation provides a way to
reduce the risk of crop failure during periods of low rainfall or
drought, enhancing the yield and market quality of crops.
Whilst a rise of the agriculture water needs is expected, the
increasing overexploitation of water sources poses a threat to
Europes environment and future water availability and quality. To
mitigate these issues water reuse can be among the suitable solu-
tions, as indicated in the Water Framework Directive (WFD) (EU,
2000). Inland, treated wastewater is normally disposed of in nat-
ural or articial water bodies, where it can be withdrawn as any
other surface water resource to irrigate agricultural crops. Besides,
wastewater treatment plants and sewage collection are often not
functioning or overloaded and thus discharge incompletely treated
efuents into the surface waters. In these cases, it is expected that
the wastewater is sufciently diluted or that the self-depurative
capability of the receiving water body would have enhanced the
water quality before the use. There are several cities in northern
Europe that rely on indirect potable reuse for 70% of their potable
resource during dry summer conditions (UKWIR, 2004) and most
of the agriculture in southern Europe is bound to irrigate crops
with poor quality water resources. This common practice is termed
indirect reuse. Hence, wastewater is indirectly reused to differ-
ent extents without planned schemes. Further, irrigation with raw
or insufciently treated wastewater was and still is reported as a
common practice, e.g. countries of the Mediterranean region not
concerned by the WFD (Angelakis et al., 1999).
Therefore, the high anthropogenic pressure is becoming a criti-
cal factor in the food production chain all over the world. Statistics
andsurveys are reporting eachyear anincreasing number of people
being affected by foodborne diseases, although in fact the majority
of the less harmful cases are not reported (EEA and WHO, 2002).
Foodborne illness is caused by critical and/or toxic levels of micro-
bial pathogens, microbial toxins or heavy metals present in food,
in some cases originating from the irrigation water. Therefore, to
protect consumers and to avoid outbreaks of food borne diseases,
efforts must focus on each point in the eld to fork chain to better
prevent food borne hazards, like apply treatments better adapted
to produce irrigation water with functional characteristics.
A denition of functional water for irrigation might be as fol-
lows: A reclaimed water resource, which compared to the raw
input water, demonstrates measurable and consistent benets to
the crop, the agro-ecosystemandto the whole foodchainfromeld
toforkandwhichhave beenobtainedbyprocessingthe input water
to alter the physical, chemical or biological characteristics.
Thanks to the new technologies now available it is becoming
increasingly attractive to produce functional water from treated
wastewater for direct reuse. In addition, reuse can generate eco-
nomical benets since wastewater, if it is not to be reclaimed, can
require even more costly treatment before disposal.
Nevertheless, a poor irrigation water quality can also have neg-
ative effects on food chain. As an example, as heat treatment can
have a negative effect on the micronutrient or nutraceutic con-
tent of food derivatives and on their organoleptic characteristics as
well, mild treatment technologies are nowadays largely applied to
secure the fresh taste to sauces, pulps, fruit juices and pre-cooked
and industrial prepared dishes. In case of contaminated irrigation
water this could increase the risk of foodborne illnesses because
mild treatment is not sufcient to inactivate, e.g. environmentally
stable stages of the protozoan parasites Giardia and Cryptosporid-
ium. Further, a variety of enteric pathogen strains could reside
within the interiors of fresh fruits and vegetables (Teplitski et al.,
2009; Warriner et al., 2003; Burnett et al., 2000; Gandhi et al., 2001)
and some of them are symbiotic with the plant (Tyler and Triplett,
2008; An et al., 2001; Chelius and Triplett, 2000; Martinez et al.,
2003). Moreover, the edible portions of a plant can become con-
taminated by uptake in the root system and subsequent transport
of the pathogen inside the plant (Solomon et al., 2002; Guo et al.,
2002; Burnett et al., 2000). Rhizosphere colonization and endo-
phytic colonizationare usually highly correlated(Dong et al., 2003).
Therefore, irrigation with bacterially contaminated water can be
the starting point of a watersoilplant contamination pathway
(Chalmers et al., 2000).
Heavy metals are one of the major concerns in the environ-
ment: potentially a large part of the water utilised for irrigation,
not only inindustrialisedcountries, couldbe pollutedby some toxic
heavy metal (Liu et al., 2005; Gupta et al., 2008; Vink et al., 1999).
Heavy metals are a concern as well to human health as they tend to
bioaccumulate in the food chain (Jrup, 2003). Furthermore, water
itself can be geogenically polluted, especially with arsenic, and also
several fertilisers and pesticides contain heavy metals. For their
bioavailability and ecotoxicology, it is crucial whether the heavy
metals are dissolved in soil water, adsorbed onto solids or xed
within the crystal lattice through precipitation of soil minerals. The
latter can be screened out from the water or remain entrapped in
the soil pores, whilst the dissolved fraction can be adsorbed by dis-
solved and colloidal phase organic matter, termed mobile organic
sorbents (MOS), which are important constituents in the ow of
water and transport of solutes through soil macropores (Totsche
and Kgel-Knabner, 2004). As a rule, all surface waters contain col-
loidal matter. Polluted waters also carry colloids of sewage origin
mainly derived fromthe fecal matter, so irrigation can increase the
heavy metals mobility in the root zone resulting in a higher uptake
by roots and bioaccumulation in the food chain.
Furthermore, the use of poor quality water candamage the most
advanced and modern irrigation technologies. Without proper l-
tration, sediment can cause clogging or mechanical corrosion.
Soluble solids like irons or calcium can cause scale into the pipes
or drippers, small amount of organic matter can aggregate form-
ing a slime that can plug emitters (Pitts et al., 2003; Nakayama and
Bucks, 1986; Battilani and Mattarelli, 2000).
Althoughindirect wastewater reuse, evenuntreated, is anunde-
niable reality and amount of treated wastewater suitable for direct
reusehas increasedconsiderablyinrecent years, therehas beenand
still there is minimal pressure toradicallyalter existing water treat-
ment systems or practices. Upgrading technology usually means
mainly to add advanced control systems to improve the ef-
ciency and operation of centralised sewer systems. Decentralised
Wastewater Treatment Systems technologies are raising interest
amongwater stakeholders whoarekeenlyinterestedinnewsingle-
family, onsite and cluster technology and advanced wastewater
treatment, also to reduce the cost of centralised wastewater treat-
ment which are governed by an expensive piping infrastructure.
The high concern about water quality and for specic use and reuse
will force wastewater plant engineers and public ofcials to be
more sensible to novel technologies and to non-conventional solu-
tions. Large-scalesewagetreatment plants areoftenunabletomake
use of the treated water and instead of being recycled and returned
to the aquifers upstream much of it is discharged into rivers. By
treatingthewater onsite, neighbourhoodassets requiringirrigation
can receive the functional water produced. Decentralised solutions
make possible that treated wastewater can be returned to the river
not far from the withdrawal point avoiding that way any exces-
sive depletion of the river water ow and preserving its dilution
capability, thus its water quality. Reducing the sewage overload, as
well the risk of pollution of canal and surface water by the sewage
oodway during storms is reduced.
The purpose of the present study was to develop prototypes
of new technologies and assess the possibilities to reduce micro-
bial and inorganic pollution of irrigation water using decentralised
water treatment techniques. The prototypes developed and tested
were a novel hi-tech device on the one hand (compact pressur-
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 387
Primary Wastewater
Polluted surface and
groundwater
Secondary Treated
Wastewater
Water sources
Membrane
Bioreactor
Field
Treatment
System
Irrigation
Method
SAFIR Treatment
phase 1
SAFIR Treatment
phase 2
SAFIR Treatment
phase 3
Irrigation
Management
CROP
Gravel
Filter
Fig. 1. SAFIR water treatment pathways. The present paper discuss only the SAFIR treatment phase 1.
ized membrane bioreactor technology) and simple, low cost lters
combinedas aninnovative treatment process (gravel lters, heavy-
metal specic lters) on the other hand, the rst adapted to the
economic setting of industrialised countries, the latter also t-
ting the needs of low-cost technology for developing countries.
The developed systems were tested in different hydro-climatic and
socio-economic settings in Serbia, China, Crete and Italy.
2. Materials and methods
Prototypes of a compact pressurised membrane bioreactor
(MBR, Grundfos BioBooster A/S, patent pending) and of a modular
eld treatment system (FTS) were developed to allow a safe use of
wastewater produced by small municipalities/industries or, more
generally, of water sources now polluted by human activities.
The SAFIR project is based on a holistic perception of technolog-
ical development with the aim to signicantly reduce risk of food
quality and hygiene impairments, whilst poor quality water along
with its nutrients content is reclaimed. SAFIR integrated water
treatment consist of three stages (Fig. 1). Treatment in phase 1 pro-
vides water which is treated with MBR or FTS technology. Properly
treated secondary wastewater (SWW), although not ltered and
disinfected, could be treated only with a simple gravel lter. Phase
2 provides a further renement of water by means of a suitable irri-
gation method. Buried drip line (sub-surface drip irrigation, SDI) is
considered the best option, however also drip irrigation, sprinkler
and furrowwere tested and compared with SDI. Phase 3 (irrigation
strategy) is considered as a part of the integrated water treatment:
for example, implementation of resting periods to allow microbial
die-off directly inuence irrigation scheduling. Waste or polluted
water can be treated onsite by the compact pressurised membrane
bioreactor or in centralised sewage plant then discharged into sur-
face water. Polluted surface and groundwater, as well as treated
wastewater, can be improved through FTS treatment. Both proto-
types are designed to deliver treated water directly to the irrigation
system. Only the effectiveness of the MBR and FTS technologies are
discussed in this paper.
Raw water sources ranged from primary untreated wastewater
from small residential or industrial areas (Italy and Beijing, China)
to secondary treated wastewater (Crete and XinXiang, China) or
canal water (Serbia). In Italy primary treated wastewater was
directly fed to the treatment device. In Crete and the two sites in
China, water was transported from the treatment plant and stored
in tanks for several days. Water produced by the MBR prototype
was storedfor a maximumperiodof 24hina 10m
3
tank before use.
Samples were collected before storage from freshly produced per-
meate. The maindifcultyencounteredwas downscaling industrial
devices to the low nominal ow required by the small experimen-
tal areas irrigated with FTS treated water (Table 1). Hence, increase
the nominal owto serve a larger area is easier than further reduce
it. The compact pressurised MBR is designed to be modular, thus it
can be easily expanded with more modules.
The total volume of water treated per year and per site is
reported in Table 2. Variations in the treated volumes are related
to the irrigation requirements and to the number of irrigated
plots.
Table 1
Raw water sources, irrigated area of the experimental plots and prototypes nominal ow.
Site/year 2006 2007 2008 Irrigated area (m
2
) MBR nominal
ow (m
3
h
1
)
FTS nominal ow
(m
3
h
1
)
Italy PWW PWW+HM spiking PWW+HM spiking 2040 0.50
Italy SWW SWW+HM spiking SWW+HM spiking 2040 3.36
Serbia Canal water Canal Water +HM spiking Canal water +HM spiking 421 1.40
Crete SWW
a
SWW
a
+HM spiking SWW
a
+HM spiking 600 1.60
Beijing PWW
a
PWW
a
PWW
a
+HM spiking 600 0.76
XinXiang SWW
a
SWW
a
+HM spiking SWW
a
+HM spiking 432 1.40
Note: SWW, secondary treated wastewater; HM, heavy metals +metalloid; PWW, primary treated wastewater from residential/industrial areas.
a
Stored on site.
388 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Table 2
Volume of water treated with MBR and FTS.
Site Treatment 2006 2007 2008
m
3
year
1
m
3
year
1
m
3
year
1
Italy MBR 398 1044 462
Italy FTS 6243 6601 5331
Serbia FTS 1500 3180 3510
Crete FTS 9044 13,430 14,333
Beijing FTS 3374 9642 5400
XinXiang FTS 7493 2811 4329
2.1. Compact pressurised modular membrane bioreactor (MBR)
technology
Membrane ultraltration performances with respect to remov-
ingorganic, nutrient andmicrobial loads fromrawwastewater or to
rene (tertiary treatment) primary or secondary treated wastewa-
ter are reported by several studies. The U.S. Bureau of Reclamation
approved MBR systems as a municipal water treatment technol-
ogy (Adhamet al., 2004). The innovative compact pressurized MBR,
as investigated in this project, combines biological treatment with
ultraltration in a single process step managed by a unique device.
Thereby, theprototypehas different characteristics thantraditional
at sheet membranes or hollow-bre modules with sidestream or
submerged conguration. In Table 3 the main operating charac-
teristics of the prototype tested in SAFIR are compared with the
design criteria adopted to test the MBR systems approved by the
U.S. Bureau of Reclamation.
Thecompact MBRis developedinto400mmdiameter, 5-mreac-
tors, which are placed in a modular system of 20in. containers.
The prototype developed by Grundfos Biobooster A/S, and tested in
SAFIR, is especially designed to be a turn-key containerised modu-
lar plants, completely constructed at the factory and shipped as a
plug and play device. Containerised plants are delivered pre-wired
and pre-assembled and minimize eld installation labour and they
are easily placed and relocated.
The MBR was fed with primary treated wastewater (PWW).
Untreated wastewater was screened to remove the brous mate-
rial, hair, paper, sanitary products, leaves, straw and greenery
normally found into the sewage water which can damage to the
pump impellers and other components or form a fouling cake on
the membranes. The raw wastewater was then pre-treated over
a 100200m mesh screen, automatically cleaned, to achieve the
primary treatment.
The MBR then treated PWW by active sludge in a pressurised
environment. The active sludge volume stored inside each MBR
reactor is of about 300L. The SAFIR prototype can operate (Table 3)
with a very high mixed liquor suspended solids (MLSS) average
content of 23.7kgm
3
. During the test the average sludge ratio
per gram of inlet COD (F/M) was of 0.36gg
1
with a daily pro-
duction of sludge of 665g in the average. The sludge retention
time of 20.9 days is two to three time higher than in conventional
activated sludge treatment, whilst the average hydraulic retention
time of 2h:02min is more than three time fast. Pressurised air is
Fig. 2. Compact pressurized membrane bioreactor (MBR) conguration.
pumped into the mixed liquor in order to keep the oxygen content
>5.5mgL
1
O
2
, a concentration much higher than the standard.
Temperature inside the reactor ranged from 23.0 to 33.4

C, with
an average of 26.8

C. The pH was 6.8logH


+
in the average, with a
standard deviation of 0.28.
The mixed liquor is ultra-ltered by ceramic membranes with a
pore size of 0.06m. The membrane area is of 5.85m
2
per reactor.
Rotating cross-ow impellers between the ltration and aeration
membrane discs (both being ceramic membranes) ensures low
kinematic viscosity in the reactor biomass, control fouling, make it
possible to operate with a four to ve times higher sludge concen-
tration than in conventional MBR systems (Fig. 2) and, ultimately,
with energy consumption compared to a submerged ultraltration
and higher ux.
The MBR concept investigated in the SAFIR project targeted
wastewater quantities in the range between 10 and 200m
3
in
hourly ow. This would correspond to municipalities/communities
between 500 and 10,000 personal equivalents. The MBR prototype
was set up inside a small wastewater treatment plant (<2000 PE)
serving the village of Mezzolara di Budrio (Bologna) in the plain
of the Po valley (44

34

N, 11

32

E). The daily outow ranged from


1095 to 15,237L day
1
. The SAFIR project trailed three iteration of
the product development cycle, and several onsite adjustments for
each of them (Battilani et al., 2009).
2.2. Modular water treatment technology at eld scale (FTS)
A site based small-scale modular and mobile eld treatment
system (FTS) for functional treatment of low quality waters was
developed, adapted and implemented in an advanced lter sta-
tionusing urbanandindustrial wastewater treatment technologies
(Fig. 3).
The FTS do not aim to replace a conventional wastewater
treatment system but merely to allow a safe use of poor quality
water for irrigation purposes on a small scale. The conguration
of the FTS prototype is exible, changeable according to the input
water quality and to the risk of bioaccumulation/contamination
of pollutants and pathogens in the horticultural products. The
Table 3
Sar MBR prototype performances compared with standard design criteria range (MLSS, mixed liquor suspended solids; F/M, feed/mass ratio; HRT, water retention time;
SRT, sludge retention time; DO, dissolved oxygen).
Design criteria/performance Standard MBR-Sar
Range Avg Median Max Min
Flux L/h/m
2
2037 34.90 37.10 61.64 7.80
MLSS kgm
3
812 23.71 23.09 50.98 9.27
F/M g COD/g MLSS 0.13 0.36 0.27 1.59 0.06
HRT h 68 2.02 1.38 6.57 0.83
SRT day 10.00 20.90 16.41 63.78 1.91
DO mgO
2
L
1
1.00 (24) 6.90 6.98 10.09 5.02
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 389
Fig. 3. Modular SAFIR eld treatment system (FTS) with acidication pump (1), gravel lter (3), heavy metal removal device (4), screen lter (5), UV lamp (6) and fertigation
pump (7).
prototypes were small-sized, adapted to the very small size of
the irrigated elds or even of the family farm in developing and
in south Mediterranean countries, which is often less than 0.2ha.
Being the downscaling process from industrial technologies,
designed for large ows, far more difcult than successive upscale
and the daily operation and maintenance of small devices more
complicated and costly, the SAFIR team decided to challenge the
prototypes by worst-case scenarios.
The FTS features a number of components, which can be
switched on/of separately to account for variations of the input
water quality and the need to produce functional irrigation water
for a specic purpose, i.e. to adapt output quality. Only the gravel
lter, the rst barrier, cannot be disconnected. The design of the
FTS is especially suited to reduce loads of microbial pathogens,
specied heavy metals and total suspended solids. The FTS in
its minimum conguration assures a suitable water quality for
subsurface drip irrigation systems. FTS is not intended to remove
salt and/or nitrate fromthe ltered water. The congurations of the
prototype range from a simple gravel lter to a complete set-up
that includes a special lter able to remove the most harmful
heavy metals/metalloids (arsenic, cadmium, chromium), as well
as excess of lead, copper and zinc, and an UV disinfection lamp for
removal or reduction of pathogens.
Gravel lter: the safeguard of the irrigation systems from clog-
ging caused by algae, bacteria or suspended solid is normally
secured in agriculture by ltering surface and ground water with
simple gravel lters. These so-called media lters are cylindrical
tanks which contain gravel with a grain size ranging from 0.6 to
3.5mm (mainly quartz or silicate minerals). Suspended solids and
large particle of materials transported by the entering water are
trapped into the gravel mass and the lters are cleaned by regularly
back ush with water which may be done automatically according
to a dened pressure differential and/or at xed time intervals.
The gravel lter utilised in the SAFIR experiments was a com-
mercial system provided by Netam Ltd. (Tel Aviv, Israel). The
technical specications of the lter are reported in Table 4.
Table 4
Technical specications of the gravel lter tested in the SAFIR experiments.
Filter media size mm 1.21.7
Filter tank volume L 60
Filter net volume
a
L 39
Sieve mesh US n. 1216
Vertical water velocity mh
1
32.4
Flow m
3
h
1
2.3
Water retention time (WRT
b
) min 1.02
Contact surface m
2
44.15
Relative contact surface m
2
L
1
h
1
0.02
a
Filter tank volume minus the media volume.
b
WRT: empty bedwater retentiontime (not calculatedwithlter material poros-
ity).
The main differences between the media lters utilised in agri-
culture andthose installedas tertiary treatment inwater treatment
plant are the short water retention time (WRT), regulated by the
ow rate, and the discontinuous ow. Nevertheless, the ltrations
process and operation is similar.
In 2006 and 2007 backush was managed automatically by the
irrigation controller (Netam Ltd.). The controller was set on an
inlet/outlet pressure difference of 0.8bar. Occasionally the lter
was manually ushed back. In 2008, before the beginning of the
irrigationthe lter media was removed fromthe lter and carefully
washed and disinfected with sodium perchloride (10% active chlo-
rine). The backush protocol was modied so the lter was ushed
back manually before irrigation start and at the end, whilst auto-
matic backush was applied when needed during irrigation time.
Heavy metal removal device (HMR): the device is a downscaling
and adaptation of a well-known industrial wastewater treatment
technology. HMR application is recommended if severe heavy
metal pollution occurs in the irrigation water source, which can-
not be sufciently treated by the gravel lter. The HMR may be
bypassed when necessary.
The HMR device consists in the same body usually utilized for
gravel lters that contains a high porosity adsorber matrix able
to reduce inorganic contaminants. The adsorber matrix consists of
ferric hydroxides, which are well known to adsorb heavy metals.
Granulated ferric hydroxide (GFH), was chosen among the dif-
ferent adsorber matrix commercially due to its high adsorption
capacity and fast adsorption kinetics for arsenic and heavy met-
als usuallyfoundinmunicipal wastewater. Inprinciple all dissolved
substances adsorbingoniron-oxide andhydroxide, canbe removed
with GFH from water. These essential capabilities of GFH are due
to the high internal surface area of 300m
2
per gram and the high
internal grain porosity of 78%. The adsorption occurs at the FeOOH
surface sites, it is specic and irreversible at pH >5.56.0.
The GFH treatment could be below its potential when the raw
water pH is out of the range 5.59.0, oxygen content is less than
0.5mgL
1
or iron, manganese and aluminiumare respectively less
than0.2, 0.05and0.2mgL
1
. The recommendedoperational design
parameters are as follows: (i) lter vertical velocity 520mh
1
,
preferably 10mh
1
; (ii) rinsing vertical velocity 2628mh
1
at
+18

C; (iii) water retention time (WRT) >3min. For practical appli-


cation in adsorbers, the addition of calcite grains showed positive
effect on the mechanical stability of the lter bed. In case of acidic
regenerationof the adsorber, quartz gravel canalsobe used, instead
of calcite. If quartz gravel is used, one should be aware that pH
value in the efuent can achieve values blow pH 3, which is due to
the acidic processes at the FeOOH surface sites when wetted com-
pletely. The same quartz gravel can be used as in the gravel lter.
Thetechnical specications of theHMRlter arereportedinTable5.
The HMR was tested for two years in four experimental stations
in Europe and in China (Serbia, Crete, Beijing and XinXiang) for its
390 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Table 5
Main characteristics of GFH and quartz gravel.
GFH Calcite or quartz gravel
Grain diameter mm 0.30.82.3 1.22.5
GFH lter bed density kgdm
3
1.11.2 1.41.5
Filter bed porosity cm
3
cm
3
0.3 0.30.4
Mixing ratio GFH/quartz gravel kg/kg 2 1
The values in bold signify the grain diameter utilised for the experiments.
capability to remove the target heavy metals (As, Cd, Cr, Cu and
Pb) and in Crete for the microbial load removal (Escherichia coli and
total coliforms).
UV lamp: UV disinfection is a well-known technology, widely
tested and applied in urban wastewater treatment. The needed
water clarity to reach an adequate penetration of UV light through
the entire ow prole, is secured in the FTS by the previous ltra-
tionstage. ThedeviceutilisedintheSAFIRexperiments was tailored
for small irrigated areas. The UV lamp power was 40W, with an
UV-C output at 254nm of 16W and an UV dose of 400J m
2
. The
approximate irradiation volume was 1.5L.
The FTS received a discontinuous ow and, consequently,
between two irrigation the residence time of water in the gravel
lter, HMR and UV lamp ranged from 18h to 7 days. The daily
ow, when the crops were irrigated, ranged from 5600L day
1
to
18,000L day
1
. The FTS was tested in its full system conguration
only in 2007 and 2008.
2.3. Water sampling and analysis
Inlet water was sampled at the same time as the permeate. Due
to the low MBR and FTS retention time no signicant variations
were expected in the inlet water characteristics over short time
periods. The wastewater monitoring during the rst year revealed
that neither the primary nor the secondary input wastewaters con-
tained heavy metals in sufciently high amounts to properly test
the project specic treatment devices and that the input function
was highly variable. As a consequence of this, the inlet water was
spiked in 2007 and 2008 with selected heavy metals and metalloid
(As, Cd, Cr, Cu and Pb). In the aimto stabilise the HMinput entering
into the SAFIR treatment device at a level signicantly higher than
the reference background (tap water), the spiking was realised by
addingaowof aheavymetal containingstocksolutiontotheinput
water. This ow was proportional to the input water ow, so that
the following constant target concentrations entering the project
specic treatment systems (MBR/FTS) shouldbe reached: 20gL
1
in As, 5gL
1
in Cd, 100gL
1
in Cr, 200gL
1
in Cu, 100gL
1
in Pb. The stock solution was prepared by adding solid heavy metal
containing salts to tap water. The stock solution was then added to
the input water ow using a MixRite injection pump (Tefen Ltd.,
Israel) assuring proportional metering. Variations of concentration
for heavy metals or other inorganic compounds over the moni-
toring period were thus due to municipal sewage variability and
to technological improvements of the treatment system. In order
to achieve a representative measurement of the concentration of
metals and metalloids, a continuous sampling systemwas set up to
collect integrated samples over a period corresponding to a dened
plant growth stage. A specically designed pseudo-proportional
sampling device (proportional like device PLD) allowed sampling
a continuous fraction of the waste/irrigation stream at different
steps of the treatment process. Samples were collected after the
heavy metal injection, after the sand lter and after the HMR,
always upstream of the fertigation injector. The PLD capillary ow
was regulated to catch water only during the irrigation time (few
hours a day). Before each integration period, a tank was xed to
the PLD. A sufcient quantity of concentrated suprapure

HNO
3
was added to the tank beforehand to reach pH 2 at the end of
the sampling duration. Acidication was able to stabilise metals
for one month or more whereas other contaminants like arsenic
(redox-sensitive) might havebeenless stable. At theendof theinte-
gration period, sub-samples of 200mL were collected for analysis.
Acidication of the unltered water samples allows to analyse the
total input of contaminants to the plot (suspended and dissolved)
even after a long storage period. To backup integrated sampling,
grab samples were collected three times a season in order to better
assess the variation in the concentrations of the inorganic com-
pounds. The analytical procedures are provided in Surdyk et al.
(2010).
For bacterial water quality analysis, a composite sample consist-
ing of three individual 1L samples was collected over a 4h period.
Samples were collected in 1L sterile glass bottles. For the helminth
egg analysis a 10L composite sample was required. Samples were
collected and kept in clean plastic containers until further pro-
cessing. Samples were stored in a cool box and transported to the
local laboratory for further analysis. Analysis of water samples for
E. coli was always initiated on the day of collection. Samples for
helminth egg analysis were stored at 45

C until further process-


ing. The sampling and analysis protocol is described by Forslund
et al. (2010).
In Table 6 are reported the average inlet water characteristics.
As expected the water quality was highly variable from site to site,
and in the same site during the cropping season or because of nat-
ural events like storms. The high variability is proved by standard
deviations frequently close or even higher than the average.
3. Results
3.1. Compact pressurised modular membrane bioreactor (MBR)
3.1.1. System process control parameters
The efuent median total suspended solids (TSS) content, mea-
suredwhenthe breakdownandmechanic upgrade of the prototype
were not inuencing the system backush and pipes were enough
clean, was <1.0mgL
1
. Feed TSS ranged from 67 to 600mgL
1
,
thus the system achieved good TSS removal. The level of ammo-
nia in the permeate (3.4mgL
1
) indicate the nitrication process
can be improved, even though nitrate reached in the average
the concentration of 12.1mgL
1
. Feed ammonia, measured in the
45m ltered PWW, was in the range 55.05.7mgL
1
(median
35.9mgL
1
). Nitrication rate, at SRT >10 day, was of 7.48mg
N/g MLSS in the average (median 5.58, max 26.5, min 1.2mg
N/g MLSS). Table 7 show the trends in the residual COD con-
centration after treatment compared with the inlet COD. The last
prototype tested (2008) removed on average 90% of the total COD
and 54% of the soluble COD. The latter residual median concen-
tration were 13.3 and 21.8mgO
2
L
1
, respectively, in 2007 and
2008.
The process performances of the prototypes tested in 2007 and
2008 were consistent with the best performances documented for
traditional at sheet membranes or hollow-bre modules with
sidestreamor submerged conguration (Metcalf &Eddy Inc., 2007;
Alaboud and Magram, 2008).
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 391
Table 6
Inlet water sources characteristics (average 0608).
MBR inlet FTS inlet
Italy Serbia Crete Beijing XinXiang Italy
Avg Std Dev Avg Std Dev Avg Std Dev Avg Std Dev Avg Std Dev Avg Std Dev
DOCtot mgO
2
L
1
338.3 190.0
DOC 45m mgO
2
L
1
66.0 14.6 24.9 28.4 5.8 8.2
TOC mgL
1
19.2 13.6 36.7 36.4 56.9 11.4 7.6 10.7
TSS mgL
1
150.8 119.9
pH logH
+
7.7 0.3
Ntot mgL
1
45.7 18.8 39.3 35.1 40.5 15.8 6.9 10.5
N-NH
4
mgL
1
32.4 14.4 33.2 34.2 18.0 21.3 16.0 18.0 7.3 16.2
N-NO
3
mgL
1
1.6 5.0 0.8 0.5 7.3 9.9 11.4 11.7 10.3 6.4 15.4 26.0
Ptot mgL
1
6.9 10.3 1.3 1.9
SO4 mgL
1
80.3 40.2 50.8 7.4 35.9 19.7 14.4 11.1 103.0 41.1
Ca mgL
1
87.3 26.4 84.1 8.1 60.7 7.9 64.8 45.7 97.7 23.3
Mg mgL
1
20.1 6.1 55.8 3.5 11.5 3.3 25.5 11.5 21.3 4.2
Na mgL
1
105.2 72.7 46.0 5.0 52.3 35.0 50.7 49.0 65.6 52.9
K mgL
1
10.0 4.2 9.4 3.9 10.3 7.6 10.4 9.4 7.2 5.6
Cl mgL
1
172.6 125.1 31.3 1.5 77.8 18.7 27.8 12.3 92.8 95.7
As gL
1
23.8 11.0 3.7 0.7 2.2 1.9 18.9 0.6 3.6 6.3
B gL
1
319.3 203.0 179.4 31.5 209.7 236.1 44.6 27.4 354.0 538.4
Cd gL
1
5.9 2.8 3.0 5.4 1.3 0.8 3.9 3.1 24.5 25.1 1.3 2.0
Co gL
1
1.8 3.4 1.3 0.8 1.3 0.7 1.3 1.8
Cr gL
1
113.4 53.9 6.4 8.6 7.1 10.9 16.8 15.5 176.2 190.9 12.7 32.1
Cu gL
1
257.4 116.1 45.5 93.3 19.3 35.2 73.5 172.2 319.7 334.2 46.2 78.8
Fe gL
1
0.5 0.4 0.1 0.1 0.2 0.2 0.1 0.1 0.2 0.5
Li gL
1
11.8 3.8 6.1 1.5 2.6 1.2 15.0 3.8
Mn gL
1
146.7 85.5 296.0 148.5 16.8 18.1 28.6 43.0 804.4 692.8 70.1 91.0
Ni gL
1
5.7 2.5 3.8 2.2 8.3 15.2 29.1 22.1 5.5 9.8
Pb gL
1
111.0 53.6 6.1 7.6 6.4 11.3 12.5 65.6 14.6 18.2 12.2 33.8
Zn gL
1
217.2 327.0 17.5 24.7 68.3 51.0 57.0 57.7 301.2 251.4 135.9 207.4
E. coli CFUmL
1
14,465 17,164 34 49 1306 2384 370 1302
Total coliform CFUmL
1
523,469 2,586,801 1484 1874 10,047 13,726 18,568 51,281
3.1.2. Reduction of microbial load
The 2008 prototype, the last tested in the SAFIR project before
further improvement were made by the producer (Grundfos Bio-
Booster), was able to remove 99.99% of the inlet E. coli and 98.52%
of total coliforms (Fig. 4). E. coli was completely removed whilst
coliforms strains still were found in the outlet water.
Microbiological analyses showed a complete removal by the
membrane of E. coli in 19% of samples in 2006 (rst prototype),
21% in 2007 (second prototype) and in 53% of samples in 2008
when the nearly full developed prototype was tested (Fig. 5). For
this parameter the highest value found in 2008 was 4CFUmL
1
as
the consequence of a previous mechanic breakdown that caused a
contaminationof the prototype piping. The prototype effectiveness
and robustness increased from 2006 to 2008. In the last year, an
E. coli load higher than 1.0CFUmL
1
was found only in 12 samples
out of 100 (Fig. 6).
Ascaris eggs were never found in inlet PWW. Only 35.5% of the
samples were positive for helmintheggs (average 0.18eggL
1
, max
Fig. 4. 2008 MBR prototype inlet and outlet water average microbial load (E. coli and total coliforms) and calculated removal efciency.
392 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Table 7
MBR inlet and outlet water average, median and range concentrations of COD and calculated removal efciencies.
Inlet (mgO
2
L
1
) Outlet (mgO
2
L
1
) Removal efciency %
COD total COD 45m COD 45m COD total COD 45m
Avg 2007 149.75 41.69 16.83 84.48 61.03
Median 2007 102.00 31.40 13.30 88.40 69.46
Range 2007 46430 9116 645 4197 1190
Avg 2008 338.34 65.98 29.97 90.00 53.60
Median 2008 335.00 73.50 21.80 91.46 61.19
Range 2008 98906 3079 1568 8396 1174
Note: Due to the very low permeate turbidity, total and ltered COD were the same.
Fig. 5. MBR prototypes inlet and outlet water E. coli loads.
Fig. 6. MBR prototypes inlet and outlet water total coliforms loads.
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 393
Table 8
MBR inlet and outlet water average concentration of metals and metalloids (spiked elements in italic).
As B Cd Cr Cu Fe Mn Pb Zn
Inlet (L
1
)
Avg 2007 23.7 199.4 5.8 106.2 215.4 0.4 120.8 103.2 127.0
Median 20.8 133.0 4.9 99.2 183.7 0.1 104.0 89.6 28.0
Range 497 61894 132 23631 47465 10 3025 23431 8535
Avg 2008 23.9 439.1 6.0 121.5 305.4 0.6 176.3 119.9 320.2
Median 24.6 431.0 6.2 129.4 283.9 0.7 175.7 124.3 177.4
Range 3615 614182 94 18277 410245 10 27234 18277 110395
Avg 20072008 23.8 319.3 5.9 113.4 257.4 0.5 146.7 111.0 217.2
Median 22.7 274.5 5.1 111.7 262.8 0.6 143.1 99.9 95.1
Range 497 61894 132 23631 47465 10 3025 23431 11035
Outlet (L
1
)
Avg 2007 7.0 452.2 1.8 2.5 19.1 0.0 116.3 1.1 191.5
Median 5.0 189.5 2.1 1.7 20.1 0.1 132.0 1.0 175.0
Range 171 181983 31 61 2410 00 24011 11 37349
Avg 2008 2.5 942.6 1.1 5.1 26.8 0.4 164.3 2.4 239.2
Median 2.2 959.0 1.1 3.1 23.5 0.1 148.8 1.7 224.3
Range 41 1382351 20 192 518 20 40059 60 53843
Avg 20072008 4.6 716.2 1.5 3.9 23.2 0.2 142.2 1.8 217.2
Median 3.0 866.0 1.2 2.7 22.4 0.1 148.8 1.0 211.0
Range 171 181983 30 191 518 20 40011 60 53843
Table 9
MBR removal efciency of the most harmful heavy metals and metalloids calculated
as means of all individual inputoutput sample pairs.
As Cd Cr Cu Pb
2007 77.2% 85.8% 97.4% 95.0% 99.1%
2008 86.2% 77.7% 96.1% 90.6% 98.5%
Avg 20072008 81.7% 81.8% 96.7% 92.8% 98.8%
0.20eggL
1
). Analyses of permeate showed the complete removal
of helminth eggs, when they were found in the inlet water.
3.1.3. Reduction of heavy metals
The compact pressurised MBR prototype developed and tested
in the SAFIR project proved to be efcient in heavy metals reduc-
tion. The combination of activated sludge, of a high sludge age,
and of ultraltration in a controlled environment ensuring optimal
conditions (inlet PWW pH around 7.58.0; mixed liquor average
temperature 27

C and pH 6.57.3; aerobic) resulted in a sharp


reduction in the permeate of nearly all the elements (Table 8).
The removal efciency of the more easily bioaccumulated met-
als and metalloid, arsenic, cadmium, chromium, copper and lead,
was found to be higher than expected (Table 9). More than 80% of
arsenic and cadmium, 93% of copper and 97% and 99% of chromium
and lead were removed by the MBR treatment. As expected, boron
was not removed. Manganese and zinc concentration as well were
not reduced by the treatment.
3.2. FTS
3.2.1. Reduction of microbial load
The complete FTS set-up, including UV disinfection, was tested
only in China and in Crete. In Table 10 are reported the average,
median and range E. coli and total coliform concentrations mea-
sured in inlet and outlet water.
FTS inlet water had low E. coli contamination, ranging from a
maximum of 2000CFUmL
1
to a minimum of 10CFUmL
1
. The
functional irrigation water produced by the FTS had a median
E. coli load of 0CFUmL
1
. Only 2 samples out of 27 tested
positive. The average removal efciency was 100.00, 99.91 and
99.40%, respectively, for inlet concentrations <100, in the range
1001000 and >1000. The median removal was always 100%
(Table 10).
Total coliforms contamination ranged from 22 to
25,900CFUmL
1
. The average and median total coliforms load in
outlet water was found to be correlated to the inlet water load. As
expected, low inlet water load (<100CFUmL
1
) were completely
removed, whilst very low microbial concentrations were still
found at higher inlet microbial load (from 100 to 10000CFUmL
1
or more). The average removal efciency was of 100.00, 98.21,
96.82 and 99.65%, respectively, for the low, medium, medium-high
and high inlet microbial load (Table 10). The total coliforms were
thus efciently removed.
Therefore, all samples, without exception, were efciently
treated for faecal contamination. This study demonstrated that a
Table 10
FTS inlet and outlet E. coli and total coliforms load and treatment effectiveness.
Rank Inlet (CFUmL
1
) Outlet (CFUmL
1
) %Removed
Avg Median Range Avg Median Range Avg Median
E. coli
10<inlet <100 42 40 8210 0 0 00 100.00 100.00
100<inlet <1000 176 180 23010 0 0 20 99.91 100.00
Inlet >1000 2067 2000 20001600 13 0 790 99.40 100.00
Total coliforms
10<inlet <100 27 27 3122 0 0 00 100.00 100.00
100<inlet <1000 491 549 807144 11 3 480 98.21 98.42
1000<inlet <10,000 3067 2410 75601200 40 6 5580 96.82 99.78
Inlet >10,000 19,620 18,000 25,90012,900 63 9 2890 99.65 99.96
394 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Table 11
Gravel lter inlet and outlet E. coli and total coliforms load and treatment effectiveness.
Rank Inlet (CFUmL
1
) Outlet (CFUmL
1
) %Removed
Avg Median Range Avg Median Range Avg Median
E. coli
10<inlet <100 36 33 7016 28 16 901 27.12 23.81
100<inlet <1000 310 205 800100 325 180 10001 25.61 30.72
Inlet >1000 3776 2200 10,7001000 1305 1100 32001 28.88 61.29
Total coliforms
10<inlet <100 26 12 5510 4 1 171 69.09 100.00
100<inlet <1000 276 74 11001 57 4 4000 0.72 66.67
1000<inlet <10,000 4573 3150 98001200 3303 2000 10,0001 19.95 38.13
Inlet >10,000 61,707 30,000 230,00010,820 39,197 14,420 200,0001 17.09 26.71
constant relatively good quality was reached over a range of inlet
contamination.
The gravel lter capability to treat secondary wastewater for
microbial contamination was tested in Italy and Crete during three
irrigation seasons (20062008). The average E. coli concentration
in secondary treated wastewater (SWW), not disinfected or l-
tered, haveatypical content rangingfrom100to100,000CFUmL
1
.
The gravel lter inlet water were frequently found at the lowest
SWW expected range, or even below. The inlet load was less than
10CFUmL
1
in 11 samples out of 40 (27.5%). Only one third of
the samples were higher than 1000CFUmL
1
. The median E. coli
concentration of the SWW utilised in the test was 220CFUmL
1
with a standard deviation of 2584CFUmL
1
, thus faecal contam-
inant concentration in the inlet treated wastewater was highly
variable. In order to assess how the inlet water contamination
level can inuence the removal capability of a simple device as a
gravel lter commonly used in agriculture the results were ranked
following the inlet water E. coli and total coliforms concentra-
tions. Removal efciency of E. coli concentration median value was
found increasing from the lower to the open class >1000CFUmL
1
(Table 11). The standard deviation were high for each class of inlet
water: 30.0, 323.1 and 1027.5CFUmL
1
respectively for the range
10100, 1001000 and >1000CFUmL
1
. The standard deviation
was always higher than the average, thus the outlet concentration
and the removal efciency shown high variability. Total coliforms
removal was more efcient at lower concentration, showing an
inverse trend with respect to E. coli. As for the fecal contamina-
tion the standard deviations were high: 7.5, 430.3, 3238.4 and
58233.9CFUmL
1
, respectively, for the range 10100, 1001000,
100010,000 and >10,000CFUmL
1
. Although the gravel lter is
not designedneither intendedtobeeffectiveagainst faecal contam-
ination, and despite the high variability in the inlet microbial load,
67.5% of samples showed a reduction of E. coli in the ltered water,
whilst for 45.0% of the sample the removal was >50.0%. Similarly,
total coliforms were removed from 75.0% of samples and in 50.0%
the reduction was >50.0%. Interestingly, the frequency of a reduc-
tion does not depend on the inlet contamination. Therefore, even at
low inlet concentrations, a reduction is mostly possible. Gravel l-
ter bacteria removal efciency appears not to be stable and easy to
predict. Notwithstanding, gravel lter reduced the loads of bacteria
entering the food chain via irrigation water.
The hypothesis of the HMR capability to adsorb microbes
was tested in 2007 and 2008. The inuent concentration, from
the gravel lter, was always less than 100CFUmL
1
E. coli and
10,000CFUmL
1
total coliform. HMR removal ranged from 50 to
97% of the E. coli (median 66.7%). The functional water produced
was better than the WHO (1989) safety threshold in six samples
out of nine. Only in one sample the concentration in ltrate was
higher than in inlet water. The sample was collected after a long
period of rain and the consequent stop of irrigation. Changes in the
chemistry of the water, lling the HMR for a long time, were prob-
ably responsible of the observed bacteria desorption and release in
ltrate.
The removal of total coliforms ranged from 22% to 99% (median
55%), but 33% of the samples shown an increase of the microbial
load after the HMR. Similarly to E. coli, intermitting operation of
lter may have increased detachment of biolm and particles at
the beginning of ow. Also changes in input water chemistry (e.g.
pH and conductance), as normally occurs at the end of a rainfall
period, caninterfere withelectrostatic adsorptiononferric hydrox-
ide surfaces and on the bacteria co-adhesion and co-aggregation
thus increasing the microbes desorption and release fromthe lter
media. The higher concentration of total coliforms had made the
phenomena more evident. The inlet and outlet average microbial
concentration and removal efciency are shown in Table 12.
3.2.2. Reduction of heavy metals
The FTS effectiveness reducing heavy metals was tested in
Serbia, ChinaandCrete. Thefunctional irrigationwater producedby
applying the FTS technology showed a sharp reduction of the inor-
ganic pollutants (Table 13). Heavy metals as arsenic, chromiumand
lead were removed in 100% of the samples, and cadmiumand cop-
per inthe92%. Theremoval efciencywas of 68%and69%for copper
and cadmium, 73% for arsenic whilst 87% and 89% of chromiumand
lead were removed from the inlet water.
Among the inorganic pollutants, some of the most harmful
were reduced by the gravel lter (Table 14, Fig. 7). Inlet and out-
let concentrations, consequently removal efciency, were different
among sites and showed a strong variation. Similarly to the micro-
bial load the inorganic pollution was not removed steadily. The
frequency of positive events, when the outlet concentration was
less than the inlet, was around 80%. Arsenic was removed in 86%
of samples, cadmium and lead in 77%, chromium in 82%, and cop-
per in 91%. The outlet concentration of arsenic (As), cadmium (Cd),
chromium (Cr), copper (Cu) and lead (Pb) was found in the aver-
age signicantly less than the measured inlet. Zinc increased after
the gravel lter. This could be only explained by corrosion of zinc
protected lter body components. The resulting removal efciency
Table 12
HMR inlet and outlet E. coli and total coliforms load and treatment effectiveness.
Inlet (CFUmL
1
) Outlet (CFUmL
1
) %Removed
Avg Median Range Avg Median Range Avg Median
E. coli 37 30 1007 23 6 1101 46.23 66.67
Total coliforms 1734 618 6470231 709 500 32501 31.56 54.98
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 395
Table 13
FTS inlet and outlet heavy metals and metalloid load and removal efciency.
Element Inlet (mgL
1
) Outlet (mgL
1
) %Removed
Avg Median Range Avg Median Range Avg Median
As 21.4 20.6 2520 5.9 4.9 15.72 73.2 76.2
Cd 14.7 7.0 61.315.1 2.2 1.3 6.70.49 69.1 80.0
Cr 157.6 101.0 621.5100 17.0 16.5 372 87.1 88.2
Cu 329.4 205.7 942.8202.4 85.3 55.1 320.716.6 67.8 80.4
Pb 107.3 101.2 144.61100.4 11.9 12.0 220 88.6 88.0
Table 14
Gravel lter heavy metals and metalloid removal efciency (%).
As Cd Cr Cu Pb Zn
UB (Serbia) Canal Avg g/100gL
1
75.91 97.79 96.35 96.39 95.46 133.97
Water Max g/100gL
1
89.96 98.04 99.50 99.38 99.01 31.91
Min g/100gL
1
40.00 97.06 87.50 93.77 85.71 415.22
CAAS (China) SWW Avg g/100gL
1
98.07 95.40 73.81 87.44 63.92
Max g/100gL
1
98.39 95.44 82.80 97.60 80.16
Min g/100gL
1
97.75 95.37 64.81 77.28 47.67
CER (Italy) SWW Avg g/100gL
1
32.46 32.14 37.87 51.62 25.84 29.15
Max g/100gL
1
84.42 76.67 90.86 82.94 90.91 34.54
Min g/100gL
1
11.43 0.00 0.00 20.00 60.00 170.14
NAGREF (Crete-GR) SWW Avg g/100gL
1
56.24 63.52 67.77 66.41 52.24 517.78
Max g/100gL
1
72.77 75.00 87.23 86.11 80.04 50.37
Min g/100gL
1
38.19 48.89 41.83 50.38 15.08 1656.37
Overall
a
SWW Avg g/100gL
1
40.95 36.47 48.28 46.17 36.71 179.77
a
Canal water (UB-Serbia) not considered.
varied from element to element and among experimental sites.
Also water source inuence the removal efciency, for that reason
mean values were calculated only for secondary treated wastewa-
ter (Table 14). The very high removal efciency measured in Serbia
is related to the good quality of the canal water. A percentage of
41% of arsenic was removed in the average, with a minimum of
32% in Italy to a max of 56% in Crete. Only 32% of cadmium was
removed in Italy whilst the efciency reached 98% in China, with
an overall average efciency of 36%. A similar trend was measured
for chromiumthat was removed at 38% in Italy and at 95% in China,
with an overall average of 48%. Copper and lead were removed
respectively at 52% and 26% in Italy and at 87% in China. The over-
all removal efciency average of copper and lead were around 46%
and 37%, respectively.
The residual heavy metals concentration after the gravel lter
was reduced by the HMR with further 35% for arsenic, 22% for cad-
mium, 25% for chromium, 33% for copper and 53% for lead. The
removal efciency tends to decrease at low concentration thus the
measuredremoval was inthe expectedrange. Most of the inorganic
pollutants in the inlet water were found at a very low concentra-
tion, nevertheless the HMR device was able to reduce it far below
the no effect limits (Fig. 8).
The adsorber couldas well decrease Zinc. At inlet concentrations
from 48 to 2206gL
1
, the efuent concentrations were from 31
to 123gL
1
with a removal efciency ranging from 72% to 97%.
At the lowest measured inlet concentration the element was not
adsorbed by the GFH matrix.
However, the frequency of positive events, when the outlet con-
centration is less than the inlet, was close to 80%.
3.3. MBR and FTS treated water compliance with reclaimed water
quality standards for safe reuse in agriculture
Although functional water produced for irrigation purposes is
something different than treated wastewater or surface water, till
nowno agreed international or European regulations clearly dene
its minimal quality requirements. This lack of regulation have
prompted inuential international retail organisations to impose
their own rules on the market. Moreover, supermarkets have an
aggressive marketing strategy concerning hygiene and safety of
their own green brand which involve as well the vegetables they
sell. The situation is therefore confused and both producers and
consumers dont have any clear point of reference, and remain
in doubt about the opportunity to reuse treated wastewater, a
situation that may encourage agriculture to overexploit ground-
water and good quality surface water resources. The need of an
European regulation is evident, also to avoid that actions taken to
stimulate wastewater reuse can be frustrated by the consumers
distrust. Furthermore, climate changes will oblige Mediterranean
countries to increase water reuse for irrigation of food crops: this
could be a pretext for retailer organisations to prefer, for market-
ing reasons, agricultural products coming from areas where direct
wastewater reuse is less neededor pollutionof water resources less
evident.
Table 15
MBR, FTS and gravel lter frequency of compliance to the water reuse Italian Standards and to the WHO (1989) guidelines.
FTS Gravel lter MBR
2006 2007 2008
0.1CFUmL
1
E. coli (IT-Dlgs 152/06) 92.5% 24.0% 18.8% 21.4% 52.9%
1.0CFUmL
1
E. coli (IT-Dlgs 152/06) 95.0% 24.0% 18.8% 28.6% 88.2%
10.0CFUmL
1
E. coli
a
WHO (1989) 97.5% 30.0% 25.0% 85.7% 100.0%
a
WHO standard refers to faecal coliforms.
396 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Fig. 7. Gravel lter inlet and outlet heavy metals and metalloid measured concentration.
The quality of the water produced by the Sar FTS and MBR
treatment compares favourably to quality guidelines set out by
Italian Law and WHO (Tables 15 and 16). In Table 16 the column
VV.AA synthesise the guideline provided by the following insti-
tutions: Australian Gov., 2000; Canadian Council of Ministers of
the Environment, 1999; South African Dept. of Water Affaires and
Forestry, 1996; WHO, 2006.
As for faecal contamination, all the samples of the water pro-
duced by the last MBR prototype fullled the WHO(1989) standard
for safe reuse of water and excreta in agriculture for irriga-
tion of crops likely to be eaten uncooked, sports elds, public
parks (Table 15, in italic). The strict Italian guidelines for direct
wastewater reuse in agriculture were fullled only by 53% of the
samples. Nevertheless, 88.2% of samples were found below the
critical threshold, which obliges to stop irrigation. Applying the
FTS complete setup, which included UV disinfection, 92.5 and
97.5% of samples met the Italian and WHO standards respec-
tively. The simple, widespread gravel lter, as used in agriculture,
reduced the E. coli load below the thresholds only in 2430% of the
samples.
Regarding heavy metals and metalloid contamination, FTS and
MBR permeate characteristics were suitable for agricultural reuse.
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 397
Fig. 8. Heavy metal removal (HMR) device inlet and outlet heavy metals and metalloid measured concentration.
Table 16
Sar FTS, MBR and gravel lter measured treated water quality (in italic bold) compared with international guideline for irrigation water quality.
Parameter Units FAO 1992 IT Dlgs 152/06 US-EPA VV.AA FTS Gravel Filter
a
MBR
Arsenic gL
1
100 20 100 100 6 6 2
Cadmium gL
1
10 5 10 10 3 2 1
Chromium
b
gL
1
100 5 100 100 17 17 5
Lead gL
1
5000 100 5000 5000 12 68 2
Copper gL
1
200 1000 200 200 85 26 27
Zinc gL
1
2000 500 2000 2000 63 400 239
a
Direct reuse of SWW.
b
Chromium (III) unless for the IT Dlgs 185/03 where Cr(VI) is considered.
398 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Only FTS and Gravel lter chromiumcontent do not met the Italian
Law for direct reuse in agriculture (Table 16).
However, some regulations impose a total nitrogen threshold
ranging from 15 to 50mgL
1
as well as a phosphorus limit of
2.0mgL
1
for the Italian law. The average total nitrogen concen-
trations were of 28.9 and 38.3mgL
1
in FTS and MBR treated
water, respectively. Phosphorus was found at a concentration of
0.5mgL
1
in the FTS treated water and of 5.6 after the MBR treat-
ment. The macro-nutrients concentration in Sar treated water
fullled most of the guidelines unless the Italian law Dlgs 152/06
(ex 185/03) for direct water reuse in agriculture.
Forslund et al. (2010) reports the results of a quantitative
microbial risk assessment (QMRA) model combined with Monte
Carlo simulations, which was used to assess whether the different
functional water produced and irrigation practices complied with
guidelines set by the World Health Organization (WHO). Further
investigation on the transfer of heavy metals fromlowquality sur-
face water to soil and potato plants in a Serbian eld study are
discussed in Surdyk et al. (2010).
4. Discussion
The economic interest in decentralised treatment as integration
or substitute of traditional activated sludge sewage plant is dis-
cussed by Maton et al. (2010). Additionally, the compact design
of the MBR containerised solution as well as a relatively sim-
ple functionality, contribute signicantly to a minimisation of the
space required to implement it. Compact containerised MBR water
treatment infrastructure requires 5070% less space than a tradi-
tional treatment plant, with a corresponding reductions in cost of
4050%. For small and rural communities these reductions rep-
resent opportunities to preserve water quality and to stimulate
economic development preserving job and property values. Essen-
tially, thanks to new exible technologies, like MBR, it is now
possible to shift from large sewage collection systems and cen-
tralised treatment plants to small and decentralised management
systems. Of course this is not an alternative to centralised sewerage
systems. Rather, it is a complimentary technology helping to pre-
serve the effectiveness of the existing infrastructures, which are
frequently overloaded.
Onthe other hand, decentralisedtreatment at eldscale, like the
proposed FTS, involves mainly a renement of the characteristics
of poor quality irrigation water. It cannot be considered, neither is
intended, as a substitute todecentralisedor centralisedwastewater
treatment.
Guglielmi et al. (2007) found that the submerged MBR perme-
ate quality was compatible with the strict limitations for treated
wastewater direct reuse inagriculture inItaly (D.Lgs. 152/06) for all
the investigatedmacro-pollutants andmicrobiological parameters,
thus suitable to irrigate apple orchards. Tertiary treatments done
with membranes, as part of a traditional wastewater plant, have
been proven to achieve good removal efciency for microbiologi-
cal parameters also by Ueda and Noran (2000), Adhamet al. (2004)
and Shang et al. (2005). The innovative, compact pressurised MBR
device tested in Sar, showed process control parameters compa-
rable with the best performances found in literature.
The microbial analysis done in 2008 on the permeate produced
by the most advanced prototype showed a nearly complete reduc-
tion of the E. coli load. The same efciency was not reached in
the removal of total coliforms strains. Using membranes, a sys-
tem applying a mechanical barrier, the efciency is expected to
be the same for all the microbial contaminants, which have simi-
lar cell size. The total coliform colonies found in the outlet water
could have been due to contamination of the prototype pipes
that occurred whilst mechanic problems were xed and during
component upgrading. During these extraordinary maintenance
operations small amounts of sludge infected the outlet pipes. Being
a prototype, a disinfection of all the pipes, pumps and instruments
connected to the MBRoutlet was impossible so that, even after sev-
eral days after the substitution of mechanical parts, total coliforms
strains were found in the outlet samples. Of course, the ordinary
and extra-ordinary maintenance program, as well as the design of
the commercial release of the prototype, are intended to avoid any
permeate contamination risk as far is possible.
When found in the inlet ow the parasites eggs, as Ascaris and
other Helminths, were totally removed from the permeate, as the
ultraltrationmembraneporesizewas aninsurmountablephysical
barrier for the enteric parasites.
The fate of heavy metals in activated sludge processes is widely
discussed. A large literature reports that most of the heavy metals
are bioadsorbed or bounded by the sludge or precipitate as partic-
ulate matter inside the sludge mass. Stephenson and Lester (1987)
foundthat cadmiumandcopper are predominantly insoluble inthe
settled sewage and have a high percentage of removal efciencies
due to the interaction of particle-associated metals with the set-
tleable biological solids. Nickel is mostly soluble in the inuent and
is poorly removed whilst lead is mainly removed by precipitation.
Several research ndings have revealed that metal biosorption by
activated sludge is rapid; about 70% of the soluble metals in solu-
tion are removed during the rst 30min. The sludge adsorption
capacity for chromium and copper was found of 3368% and of
8174%, respectively, as a function of a sludge age from 5 days to
15 days (Alkan and Eleren, 2008). Although no reliable indication
about arsenic removal by sludge was found in literature, arsenic is
oxidised to As(V) in well oxygenated mixed licor, conditions that
prevail in compact pressurised MBR. The oxidized form then tends
to precipitate.
With respect to the MBR membrane pore size even ultraltra-
tion is not foreseen to remove metal ions. Nevertheless, in a study
recently carried out, Malamis et al. (2009) found that ultraltra-
tionmembranes withno mineral additioncould remove signicant
Cr(III) from the nal efuent, with removal efciencies ranging
between 43.2 and 69.2%.
Our results obtained applying a compact pressurised MBR sys-
tem were in agreement or even better than those reported in
literature.
Boronis of great concernas pollutant inagriculture. The element
tends to accumulate in soils and several crops are boron sensitive.
Reduction of soil fertility or crop yield related to boron concentra-
tion in soil and water are widely documented (Ayers and Westcot,
1985; Maas, 1987). Boron is contained in wastewater due to perbo-
rate additives in washing powders leading to concentrations of up
to >1mg/L (Vengosh et al., 1994). It is extremely mobile in dis-
solved form (borate and undissociated boric acid depending on
pH). Sar MBR was found not to be effective for removing boron.
In some samples permeate boron concentration were higher than
in the inuent ow. However, no standard wastewater treatment,
including all membrane types fromultra-ltration to conventional
reverse osmosis is able to remove boron effectively (Kloppmann
et al., 2008a), and only RO with specically designed membranes
at high pH (borate ions prevail) will reduce boron concentration
in wastewater or salt water streams (Kloppmann et al., 2008b).
Nonetheless, boron can be, to some extent, and in a high pH range,
adsorbed onto clay minerals and organic matter (Karahan et al.,
2006; Goldberg et al., 2000; Keren and Communar, 2009). The
increase of boron concentrations in some permeates, compared to
input water, might be to due to temporary storage of a fraction
of boron in the sludges and subsequent release, triggered by pH
variations in the input water and the process.
The main difference between boron and manganese is the redox
dependency of the mobility of the latter. Linnik and Nabivanels
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 399
(1977) found that the main amount of the dissolved forms of man-
ganese in surface water was present as complex cation and anion
compounds. These complex compounds originated from chelating
reactions withdissolved organic compounds infreshwater (Rashid,
1974), thus greatly increasing manganese solubility and stability.
Soluble manganese chelates are difcult to remove from water by
ltration or adsorption. Due to the high organic matter content in
the wastewater stream, reducing conditions are expected in the
FTS system. Under such conditions the mobility of manganese is
enhanced so that even manganese hydroxides initially present in
the lter material can be dissolved and manganese released to the
ltrate (Oren et al., 2007). Manganese as well could be toxic for
crops growninwet soils richinorganic matter, or grownunder high
temperature or high light intensity. Manganese toxicity is likely
with plants that are fertilized with acid-forming fertilizers, high
rates of superphosphate, or nitrate (El-Jaoual and Cox, 1998). Fur-
thermore, manganese upon oxidation can precipitate clogging the
emitters in drip irrigation systems. MBR treatment had no effect
on the permeate manganese content, which was found below the
200gL
1
critical threshold.
ThemainproblemencounteredwhentestingtheMBRprototype
for direct wastewater reuse was the prototype breakdown occur-
rence. Such incidents can cause mainly microbial contamination
of irrigated crops. Heavy metals concentration is normally too low
to cause acute plant toxicity or bioaccumulation in edible parts in
short time. Also ordinary or extraordinary maintenance can cause
a temporary increase of faecal contaminant passing through the
MBR. Aneffectivecontingencyplanmust beenvisagedtoavoidcrop
hygiene impairments. Farmers must be warnedingoodtime tostop
irrigation until breakdown is xed and their residual effects would
end. Implementation of SAFIR treatment phase 2 and 3 can con-
siderably help reducing contamination risk (Forslund et al., 2010).
Moreover, thecommercial compact pressurisedMBRplant is highly
automated. To enable testing of the efuent quality before andafter
the plant, two ow proportional auto samplers with refrigerators
are installed as part of the system. For controlling the biological
process in the MBR, online probes are installed for NH
4
N, NO
3
N,
PO
4
Pandturbidity. Furthermore probes for measuringpH, oxygen
content and temperature are installed. On top, a complete SCADA
system (Supervisory Control And Data Acquisition) operates the
compact pressurisedMBRplant. Access tothe SCADAcantake place
from any internet connected PC or directly from the control panel
in the container. The interface can communicate with the exter-
nal control systems. This can be used to manually or automatically
start and stop the plant and transfer relevant alarms and status
messages.
The gravel lter effectiveness with respect to reducing soluble
contaminants or microbes is directly related to input water sus-
pended solid particle size and nature, as well as it relates to the
hydrodynamic design of the lter. Both have great inuence on the
collision frequency between suspended particles and lter media
(Adin and Elimelech, 1989). The process is expected to be greatly
variable, due to changes in hydrodynamics that Stoodley et al.
(1999, 2001) found have signicant inuence on biolm develop-
ment, thus on lter maturation. A gravel lter must be biologically
mature to be effective in ltration. A biolm must have developed
on the lter media surface to which particle aggregates, organic
matter or inorganic suspended solids may adhere once a collision
occurs. Attachment to surfaces has been proved to stimulate bac-
terial exopolysaccharide synthesis, further enhancing aggregation
(Vandevivere and Kirchman, 1993) and laying the basis for further
growth of the biolm coating the lter media. Biolm growth and
health can be reached and maintained only when irrigation inter-
vals are not prolonged, thus avoiding lack of oxygen, starvation
of bacteria and changes in the chemistry of the water lling the
lter. Irrigation is a stop and go practice, regulated by rain, plant
evapotranspiration and by the available soil water storage. As a
consequence, the irrigation systemwould remain in stand by, lled
with irrigation water, overnight or for several days (e.g. during the
crop early growth stages or after a rainfall period). Thereby, the
interval between irrigations in some period can be long enough to
allow particle aggregates decomposition and biolm detachment,
resulting inthe moderate andvariable microbial removal efciency
observed in the present experiments.
Also contaminant concentration in inlet water inuence. E. coli
removal by gravel lters and was found to be enhanced at concen-
tration higher than 1000CFUmL
1
. This agree with Bengtsson and
Lindquist (1995) and Wollum and Cassel (1978), who found the
rate of bacteria adsorption on lter media increased with bacte-
ria cell concentration in water. Fletchers (1977) studied a porous
media (polystyrene) and observed that increasing bacterial con-
centration led to an increase in the number of bacterial collisions
with the media surface, hence to an increase of adhesion opportu-
nities. On the contrary, when hydrodynamic forces into the lter
mass prevails, small particles and microbes were not efciently
screened out by the standard media (1.21.7mm), irrespectively
to the vertical water velocity. However, this cannot explain why
total coliforms were removed more efciently at the lowest con-
centration (Table 11).
The HMR was not designed to remove microbial pollution, par-
ticularly when operated a very low microbes concentrations. The
GFH grain average size in the heavy metal removal device was
nearly the same of the gravel lter standard media. Otherwise,
the surface chemistry and the porosity of the ferric hydroxide
matrix were very different fromthe crushed silica or quartz media.
In porous media the dominant mechanism for retention of bac-
teria is surface adsorption (Gerba et al., 1975; McDowell-Boyer
et al., 1986). Particularly, the GFHs positively charged surfaces and
roughness was expected to promote bacteria attachment. Further-
more, Roberts (2004) found that iron oxide surfaces were heavily
colonised by bacteria, and colonization patterns in a pHrange from
2.0 to 8.0 were dominated by electrostatic interactions between
the mineral surface and microbial cells. This might explain why
HMR was found to be moderately effective in removing faecal
contamination. The amount of bacteria removed was often not
enough to reduce the contamination below the safety threshold
and, as for the gravel lter, the removal process was not stable nor
predictable.
Most of the heavy metal in secondary treated wastewater are
present as soluble molecules, in forms adsorbed and/or occluded
in suspended particles aggregates, or as metalhumic complexes.
The latter fraction can be signicantly high due to the abundance of
humic substances in the treated wastewater dissolved organic car-
bon fraction (Imai et al., 2002). In SWW, approximately 6070%
of dissolved copper and 20% of dissolved lead can be found as
humic complexes and rather strong interaction are reported for
chromium(III) and humic substances. Thereby, heavy metals canbe
removed from SWW or polluted surface water (direct or indirect
reuse) removing suspended solids by mechanic ltration (gravel
lter) or being adsorbed on a specic matrix (HMR). Although the
gravel lter removal efciencycanvaryfollowingthebiolmhealth
and efciency or the dissolved organic carbon content in inuent
water, a prudential removal efciency of 35% can be ascribed to the
gravel lter.
As reported for the gravel lter also the HMR cannot remove
steadily the inorganic pollution, mainly at low concentrations.
This might be related to metalhumic complex in inuent water.
Metalhumic complex prerequisites for adsorption onto GFH are
both a minimum number of carboxyl and hydroxyl functional
groups and a molecular size large enough to promote the con-
tact with the GFH rugged surface. The larger fulvic acids are well
adsorbed whilst the smaller molecular fractions are poorly or not
400 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
adsorbed (Genz et al., 2008). HMR was found able to further reduce
35% of arsenic, 22% of cadmium, 25% of chromium, 33% of copper,
53% of lead and up to 97% of zinc. As a rule of thumb, HMR can be
added after the gravel lter when the concentration in the water
source of one or more of the heavy metals exceed the treatment
goal or the thresholds xed by local regulations more than 3540%.
The FTS system treatment chain is an innovative process never
applied before to produce functional water to irrigate food crops.
The effectiveness of each treatment step was already widely tested
for their specic purposes and against parameters identied for
different uses and scale than those tested in SAFIR. As an exam-
ple, nor the standard gravel lter effectiveness removing faecal
contamination neither the possibility to reduce the heavy metal
concentration in ltrate were tested before for devices utilised
in agriculture. Exiting literature refers to media lters utilised in
wastewater treatment plant, which have different media size, a
much higher lter volume and mass and hydraulic retention time.
Further, the combined effect of gravel lter and adsorber matrix to
produce functional water for irrigation at small scale was no tested
previously.
The application of UV-C disinfection against bacteria, fungi and
viruses in irrigation water recycling systems in greenhouse is a
well-known technology (Newman, 2004). Poncet et al. reported
that UV disinfection have no bactericidal effect for greenhouse
rose recycled irrigation water at 715J m
2
, even with several
passes across the UV lamp. In general, commercial UV irrigation
water disinfection systems are designed for exposures from800 to
2500J m
2
whilst with the FTS system 400J m
2
, with a very short
exposure time, were sufcient to completely remove the residual
bacterial load fromthe HMR outlet. This is primarily in relationship
to the good clarity of the double ltered water. The cost for the UV
lamp, if calculated per cubic meter of treated water depends on the
radiation intensity (Battilani et al., 2009), thus the sharp reduction
of UV exposure required by the FTS system involves a substantial
reduction of the operating cost as well.
The SAFIR project was intended rst of all to demonstrate the
feasibility of downscaling of industrial technologies and of their
effective combination in a simple water renement process appli-
cable at eld level. Some aspect of the process, as the impact of
detachment and release of biolmfromthe gravel lter on the sub-
sequent HMR effectiveness, or changes in water chemistry during
prolonged FTS stand by periods, were unknown before thus no spe-
cic measurements were envisaged. Although in SAFIR not all the
process kinetics were fully explored, or evenidentied, the FTS sys-
temwas foundable to substantially reduce the microbial andheavy
metals and metalloid load of a range of treated wastewater sources,
also at a very low concentration. As a rule of thumb, the results
obtained indicated that a daily irrigation frequency allows a better
lter maturation and a more constant removal activity. Therefore,
it would be preferable to combine the FTS treatment with high fre-
quency, low volume, drip irrigation than with highly intermittent
irrigation methods/schedule.
5. Conclusion
The application of decentralised water treatment technologies
which allows to remove efciently faecal and inorganic contam-
inants from irrigation water, keeping the microbial, metals and
metalloids loads below the safety thresholds, is central in areas
were indirect reuse is a common practice or direct reuse can be
stimulated. Decentralised technologies, like compact pressurized
MBR (Membrane Bio Reactor, Grundfos BioBooster A/S), may as
well reduce the risk of water contamination caused by sewage
spillage and overow. A less technologically advanced water treat-
ment at eld scale, like the FTS, may help to cope with the
widespread contamination of surface and groundwater, dealing as
well with diffuse indirect treated wastewater reuse.
Decentralised compact pressurised membrane biobooster
(MBR), was able to remove up to 99.99% of the inlet E. coli and
98.52% of total coliforms. E. coli was completely removed from
irrigation water in 53% of the samples by the last MBR prototype
version. In 2008 100% of samples fullled WHO standards (1989)
and Global Gap requirement for faecal contamination.
MBR removed from inlet ow in the average 82% of arsenic,
82% of cadmium, 97% of chromium, 93% of copper and 99% of lead.
Boron and manganese were not removed from permeate. In some
samples an increase of boron concentration was observed as a con-
sequence of adsorption/desorption dynamics of soluble boron on
active sludge.
The eld treatment system (FTS) proved to be effective against
faecal contamination when applied with its complete set up. The
sole gravel lter and heavy metal removal device (HMR) cannot
provide sufcient and steadily treatment for microbial contamina-
tion. Nevertheless, gravel lter can remove up to 60% of E. coli but
the removal process was not stable nor predictable.
FTS removed 76% of arsenic, 80% of cadmiumand copper, 88% of
chromium and lead, and up to 97% of zinc. As for MBR, boron and
manganese were not removed from the irrigation water. Gravel
lter directly fed with secondary treated wastewater was found
able to remove 41% of arsenic, 36% of cadmium and lead, 48% of
chromium and 46% of copper. The residual heavy metals concen-
tration after the gravel lter was reduced by the HMR with further
35%for arsenic, 22%for cadmium, 25%for chromium, 33%for copper
and 53% for lead.
The FTS and MBR treatment developed and tested in this
study resulted in water qualities suitable for microirrigation, as
demonstrated by the bench tests on the dripline after three
years of irrigation (Battilani et al., 2009). No damages to the
emitters or clogging were found after three year of intense use.
Thus the functional water produced by FTS or MBR can be used
for irrigation and fertigation in gardening and agriculture, with-
out damages even to the most advanced and fragile irrigation
technologies.
The quality of the water produced by the SAFIR FTS and
MBR treatment fullled the irrigation water quality standards
required by the most reliable international guidelines, only the
most restrictive regulation as the Italian Law D.Lgs. 152/06 were
not always fullled. Nevertheless, the technologies developed and
tested in SAFIR proved to be an effective barrier against those
pollutants which can threaten the food quality and safety. The
MBR and FTS technologies allowed a safe direct and indirect water
reuse. The correct application of these technologies could be con-
sidered as a quality label guaranteeing the food safety thus
fullling quality standards worldwide required by the retailers
organisations.
Acknowledgements
The authors gratefully acknowledge the EU project SAFIR Team.
The research was funded under the 6th European Framework Pro-
gram by the EU-DG XII (CT Food-023168).
References
Adham, S., DeCarolis, J.F., Pearce, W., 2004. Optimisation of Various MBR
Systems for Water Reclamation Phase III. Final Report No. 103. Desali-
nation and Water Purication Research and Development Program. U.S.
Department of the Interior Bureau of Reclamation Technical Ser-
vice Center Environmental Resources Team Water Treatment Engineer-
ing and Research Group Denver, Colorado (accessed 28 august 2010)
http://www.usbr.gov/pmts/water/media/pdfs/report103.pdf.
Adin, A., Elimelech, M., 1989. Particle ltration for wastewater irrigation. J. Irrig.
Drain. Eng. 115 (3), 474487.
A. Battilani et al. / Agricultural Water Management 98 (2010) 385402 401
Alkan, U., Eleren, S.C., 2008. Int. J. Environ. Pollut. 34 (1/2/3/4), 400411.
Alaboud, T.M., Magram, S.H., 2008. Adiscourse onfeasibility of the membrane biore-
actor technology for wastewater reuse in Saudi Arabia. Res. J. Environ. Sci. 2 (6),
445455.
An, Q.L., Yang, X.J., Dong, Y.M., Feng, L.J., Kuang, B.J., et al., 2001. Using confo-
cal laser scanning microscope to visualize the infection of rice roots by GFP
labelled Klebsiella oxytoca SA2, an endophytic diazotroph. Acta Bot. Sin. 43, 558
564.
Angelakis, A.N., Marecos Do Monte, M.H.F., Bontoux, L., Asano, T., 1999. The status
of wastewater reuse practice in the Mediterranean basin: need for guidelines.
Water Res. 33, 22012217.
Australian Gov., 2000. Australia and New Zealand Guidelines for Fresh and Marine
Water Quality. Volume 3 Primary Industries, Water Quality for Irrigation and
General Use.
Ayers, R.S., Westcot, D.W., 1985. Water Quality for Agriculture. FAO Irrigation and
Drainage Paper No. 29. FAO, Rome.
Battilani, A., Steiner, M., Neumann Andersen, M., Schweitzer, A., Lorenzen, J., Bak,
S.N., Plauborg, F., 2009. User Manual for Decentralised Functional Water Pro-
duction for Irrigation Purposes. SAFIR Public Report (accessed on 8 April 2010)
http://www.sar4eu.org.
Battilani, A., Mattarelli, P., 2000. Quality problems of water in irrigation and compat-
ibility with re-use of treated urban efuents. Atti del convegno internazionale
H2Obiettivo 2000, Torino 2-5 maggio 2000/43-53.
Bengtsson, G., Lindquist, R., 1995. Transport of soil bacteria controlled by density-
dependent sorption kinetics. Water Res. 31, 12471256.
Burnett, S.L., Chen, J.R., Beuchat, L.R., 2000. Attachment of Escherichia coli O157:H7to
the surfaces and internal structures of apples as detected by confocal scanning
laser microscopy. Appl. Environ. Microbiol., 46794687.
Canadian Council of Ministers of the Environment, 1999. Canadian Water Quality
Guidelines. Protocols for Deriving Water Quality Guidelines for the Protection
of Agricultural Water Uses (Irrigation and Livestock Water).
Chalmers, R.M., Aird, H., Bolton, F.J., 2000. Waterborne Escherichia coli O157. J. Appl.
Microbiol. 88, 124S132S.
Chelius, M.K., Triplett, E.W., 2000. Immunolocalization of dinitrogenase reductase
produced by Klebsiella pneumoniae in association with Zea mays L. Appl. Environ.
Microbiol. 66, 783787.
Dong, Y., Iniguez, A.L., Ahmer, B.M.M., Triplett, E.W., 2003. Kinetics and strain
specicity of rhizosphere and endophytic colonization by enteric bacteria on
seedlings of Medicago sativa and Medicago truncatula. Appl. Environ. Microbiol.
69 (3), 17831790.
EEAandWHO, 2002. Water andhealthinEurope. In: Bartram, J., Thyssen, N., Gowers,
A., Pond, K., Lack, T. (Eds.), WHO Reg. Publications. European Series, No. 93, p.
240.
El-Jaoual, T., Cox, D.A., 1998. Manganese toxicity in plants. J. Plant Nutr. 21 (2),
353386.
EU Council Directive of 23 October, 2000. Establishing a framework for community
actionintheeldof water policy(WFD), 2000/60/EC. Ofcial Journal of European
Communities, L327, Luxembourg, 22 December 2000.
Fletcher, M., 1977. Effects of culture concentration and age, time, and temperature
on bacterial attachment to polystyrene. Can. J. Microbiol. 23, 16.
Forslund, A., Ensink, J.H.J., Battilani, A., Kljujev, I., Gola, S., Raicevic, V., Jovanovic, Z.,
Stikic, R., Sandei, L., Fletcher, T., Dalsgaard, A., 2010. Faecal contamination and
hygiene aspect associated with the use of treated wastewater and canal water
for irrigation of potatoes. Agric. Water Manage. 98, 440450.
Gandhi, M., Golding, S., Yaron, S., Matthews, K.R., 2001. Use of green uorescent
protein expressing Salmonella Stanley to investigate survival, spatial location,
and control on alfalfa sprouts. J. Food Prot. 64, 18911898.
Genz, A., Baumgarten, B., Goernitz, M., Jekela, M., 2008. NOM removal by adsorp-
tionontogranular ferric hydroxide: equilibrium, kinetics, lter andregeneration
studies. Water Res. 42 (1-2), 238248.
Gerba, C.P., Wallis, C., Melnick, J.L., 1975. Fate of wastewater bacteria and viruses in
soil. Proceedings of the American Society of Civil Engineers. J. Irrig. Drain. Div.
101, 157174.
Goldberg, S., Lesch, S.M., Suarez, D.L., 2000. Predicting boron adsorption by soils
using soil chemical parameters in the constant capacitance model. Soil Sci. Soc.
Am. J. 64, 13561363.
Guglielmi, G., Chiarani, D., Andreottola, G., Ziglio, G., 2007. Membrane bioreac-
tor technology for agricultural reuse of municipal wastewater: a comparative
study. In: 2nd IWA National Young Water Professionals Conference, Germany
Membrane Technologies for Wastewater Treatment and Reuse, Kompe-
tenzZentrum Wasser Berlin Publication Series, Berlin (D), 45 June, 1.2004 ff.,
Vol. 7/18.
Guo, X., vanIersel, R.W., Chen, J., Brackett, R.E., Beauchat, L.R., 2002. Evidence of asso-
ciation of salmonellae with tomato plants grown hydroponically in inoculated
nutrient solution. Appl. Environ. Microbiol..
Gupta, N., Khan, D.K., Santra, S.C., 2008. Anassessment of heavymetal contamination
in vegetables grown in wastewater-irrigated areas of Titagarh, West Bengal.
India Bull. Environ. Contam. Toxicol. 80 (2), 115118.
Imai, A., Fukushima, T., Matsushigea, K., Kim, Y.-H., Choid, K., 2002. Characteriza-
tion of dissolved organic matter in efuents fromwastewater treatment plants.
Water Res. 36 (4), 859870.
Jrup, L., 2003. Hazards of heavy metal contamination. Br. Med. Bull. 68, 167182.
Karahan, S., Yurdakoc, M., Seki, Y., Yurdakoc, K., 2006. Removal of boron fromaque-
ous solution by clays and modied clays. J. Colloid Interface Sci. 293, 3642.
Keren, R., Communar, G., 2009. Boron sorption on wastewater dissolved organic
matter: pH effect. Soil Sci. Soc. Am. J. 73 (6), 20212025.
Kloppmann, W., Van Houtte, E., Picot, G., Vandenbohede, A., Lebbe, L., Guerrot, C.,
Millot, R., Gaus, I., Wintgens, T., 2008b. Monitoring reverse osmosis treated
wastewater recharge into a coastal aquifer by environmental isotopes (B, Li,
O, H). Environ. Sci. Technol. 42, 87598765.
Kloppmann, W., Vengosh, A., Guerrot, C., Millot, R., Pankratov, I., 2008a. Isotope and
ion selectivity in reverse osmosis desalination: geochemical tracers for man-
made freshwater. Environ. Sci. Technol. 42, 47234731.
Linnik, P.N., Nabivanels, B.I., 1977. Determination of various forms of metal ions in
natural waters. Hydrobiologia 31, 9198.
Liu, W., Zhao, J., Ouyanga, Z., Sderlund, L., Liu, G., 2005. Impacts of sewage irrigation
on heavy metal distribution and contamination in Beijing, China. Environ. Int.
31 (6), 805812.
Maas, E.V., 1987. Salt tolerance of plants. In: Christie, B.R. (Ed.), Handbook of Plant
Science in Agriculture, 2. CRC Press, Boca Raton, FL, USA, p. 57.
Malamis, S., Katsou, E., Chazilias, D., Loizidou, M., 2009. Investigation of Cr(III)
removal from wastewater with the use of MBR combined with low-cost addi-
tives. J. Membr. Sci. 333 (12), 1219.
Martinez, L., Caballero-Mellaod, J., Orozco, J., Martinez-Romero, E., 2003. Dia-
zotrophic bacteria associated with banana (Musa spp.). Plant Soil 257,
3547.
Maton, L., Psarras, G., Kasapakis, G., Lorenzen, J.R., Andersen, M., Boesend, M., Back,
S.N., Chartzoulakis, K., Pedersen, S.M., 2010. Assessing the net benets of using
wastewater treated with a membrane bioreactor for irrigating vegetables in
Crete. Agric. Water Manage. 98, 458464.
McDowell-Boyer, L.M., Hunt, J.R., Sitar, N., 1986. Particle transport through porous
media. Water Resour. Res. 22, 19011921.
Metcalf & Eddy Inc., 2007. Water Reuse: Issues, Technologies and Applications.
McGraw Hill, New York.
Nakayama, F.S., Bucks, D.A., 1986. Trickle Irrigation for Crop Production. Design,
OperationandManagement. Elsevier Publisher B.V. Development inAgricultural
Engineering, no. 9.
Newman, S.E., 2004. Disinfecting irrigation water for disease management. In: Acta
of the 20th Annual Conference on Pest Management on Ornamentals. Society of
American Florists San Jose, California, pp. 110.
Oren, O., Gavrieli, I., Burg, A., Guttman, J., Lazar, B., 2007. Manganese mobilization
and enrichment during soil aquifer treatment (SAT) of efuents, the Dan Region
Sewage Reclamation Project (Shafdan), Israel. Environ. Sci. Technol. 41, 766
772.
Pitts, D.J., Haman, D.Z., Smajstrla, A.G., 2003. Causes and Prevention of Emitter
Plugging in Microirrigation Systems. Agricultural and Biological Engineering
Department, Florida Cooperative Extension Service, IFAS BUL258.
Poncet, C., Offroy, M., Bonnet, G., 2001. Disinfection of recycling water in rose cul-
tures. Acta Horticult. 547, 121126.
Rashid, M.A., 1974. Adsorption of metals on sedimentary and peat humic acids.
Chem. Geol. 13, 115123.
Roberts, J.A., 2004. Inhibition and enhancement of microbial surface colonization:
the role of silicate composition. Chem. Geol. 212, 313327.
Shang, C., Wong, H.M., Chen, G., 2005. Bacteriophage MS-2 removal by submerged
membrane bioreactor. Water Res. 39, 42114219.
Solomon, E.B., Yaron, S., Matthews, K.R., 2002. Transmission of Escherichia coli
O157:H7 from contaminated manure and irrigation water to lettuce plant tis-
sue and its subsequent internalization. Appl. Environ. Microbiol. 68 (1), 397
400.
South African Dept. of Water Affaires and Forestry, 1996. Water Quality Guidelines
Volume 4 Agricultural Use: Irrigation.
Stephenson, T., Lester, J.N., 1987. Heavy metal behaviour during the activated sludge
process. I. Extent of soluble and insoluble metal removal. Sci. Total Environ. 63,
199214.
Stoodley, P., Dodds, I., Boyle, J.D., Lappin-Scott, H.M., 1999. Inuence of hydrody-
namics and nutrients on biolm structure. J. Appl. Microbiol. 85, 1928.
Stoodley, P., Wilson, S., Caro, R., Piscitteli, C., Rupp, C.J., 2001. Detachment and Other
Dynamic Processes in Bacterial Biolms in Surfaces in Biomaterials 2001 Sym-
posiumProceedings. Surfaces inBiomaterials Foundation, Minneapolis, USA, pp.
189192 (abstract 01-038).
Surdyk, N., Cary, L., Blagojevic, S., Jovanovic, Z., Stikic, R., Vucelic-Radovic, B.,
Zarkovic, B., Sandei, L., Pettenati, M., Kloppmann, W., 2010. Impact of irriga-
tion with treated low quality water on the heavy metal contents of a soil-crop
system in Serbia. Agric. Water Manage. 98, 451457.
Teplitski, M., Barak, J.D., Schneider, K.R., 2009. Human enteric pathogens in produce:
un-answered ecological questions with direct implications for food safety. Curr.
Opin. Biotechnol. 20 (2), 166171.
Totsche, K.U., Kgel-Knabner, I., 2004. Mobile organic sorbent affected contaminant
transport in soil. Vadose Zone J. 3, 352367.
Tyler, H.L., Triplett, E.W., 2008. Plants as a habitat for benecial and/or human
pathogenic bacteria. Ann. Rev. Phytopathol. Vol.46, 5373.
Ueda, T., Noran, N.J., 2000. Fate of indigenous bacteriophage in a membrane biore-
actor. Water Res. 34, 21512159.
UKWIR, WATEREUSE Foundation and AWWA Research Foundation, 2004. Frame-
work for Developing Water Reuse Criteria with Reference to Drinking Water
Supplies. Draft Report Ref. No. 05/WR/29/1.
Vandevivere, P., Kirchman, D.L., 1993. Attachment stimulates exopolysac-
charide synthesis by a bacterium. Appl. Environ. Microbiol. 59, 3280
3286.
Vengosh, A., Heumann, K.G., Juraske, S., Kasher, R., 1994. Boron isotope application
for tracing sources of contamination in groundwater. Environ. Sci. Technol. 28,
19681974.
402 A. Battilani et al. / Agricultural Water Management 98 (2010) 385402
Vink, R., Beherendt, H., Salomons, W., 1999. Development of the heavy metal pollu-
tion trends in several European rivers: an analysis of point and diffuse sources.
Water Sci. Technol. 30 (12), 215223.
Warriner, K., Ibrahim, F., Dickinson, M., Wright, C., Waites, W.M., 2003. Interac-
tion of Escherichia coli with growing salad spinach plants. J. Food Prot. 66 (10),
17901797.
WHO, 1989. Health guidelines for the use of wastewater in agriculture and aquacul-
ture. In: WHO Technical Report Series No. 77. WHO, Geneva, Switzerland.
WHO, 2006. Guidelines for the Safe Use of Wastewater. In: Excreta and Grey Water.
Volume 2. Wastewater Use in Agriculture. WHO edt Geneve (CH).
Wollum, A.G., Cassel, D.K., 1978. Transport of microorganisms in sand columns. Soil
Sci. Soc. Am. J. 42, 7276.

Vous aimerez peut-être aussi