Ws

LD

a

ARRAA

KSVCNO

1

bmpww(elHprga

0h

Ecological Engineering 47 (2012) 174– 181

Contents lists available at SciVerse ScienceDirect

Ecological Engineering

j o ur nal homep age : www.elsev ier .com/ locate /eco leng

illow bed fertigated with domestic wastewater to recover nutrients inubarctic climates

ea Rastas Amofah ∗, Jonathan Mattsson, Annelie Hedströmepartment of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, 971 87 Luleå, Sweden

r t i c l e i n f o

rticle history:eceived 13 January 2012eceived in revised form 6 June 2012ccepted 22 June 2012vailable online 20 July 2012

eywords:alixegetation filteronstructed wetlanditrogennsite treatment

a b s t r a c t

Conventional methods for wastewater treatment emphasise protecting human health, receiving watersand the environment. Consequently, they are generally designed to reduce pollutant levels and are notwell-suited for creating resources. This paper describes a new, more sustainable and energy-efficientapproach to wastewater treatment that satisfies health and environmental standards while also facil-itating resource recovery. A full-scale compact willow bed was intensively fertigated with domesticwastewater in a cold climate to examine biomass production, the recovery of nutrients in willow biomass,and wastewater treatment. The performance of the willow bed was assessed for two years, covering threegrowing seasons. The studied frost-tolerant willow clones produced good biomass yields per unit area(6–7 ton dry matter/ha and year) under intensive fertigation with dense planting and continuous har-vesting. The biomass yield of willow species exhibiting vertical growth seemed to be greater than that forlateral growth species in the dense stands studied. In contrast to biomass production, nutrient recoverywas facilitated by intensive fertigation, continuous harvesting and less dense planting with a horizontallygrowing willow clone. The estimated nitrogen accumulation in above-ground biomass was 210 kg/ha and

that of phosphorus was 30 kg/ha. 90% of the accumulated nutrients in the above-ground biomass wereremoved from the site during the experimental period. However, the quantity of nutrients accumulatedin the willow biomass represented only a small fraction of the loaded or removed amount. The willowbed was shown to be an efficient prefilter for reducing the abundance of particulate and organic matter,leaving the bulk of the remaining nutrients in forms that could be recovered in subsequent treatmentsteps.

refNrmt(basie

. Introduction

The purpose of recycling is to conserve resources so that theyecome available for reuse. Current methods for wastewater treat-ent focus on to reduce pollutant levels, and thus, do not conserve

hosphorus (P) and nitrogen (N) resources in municipal/domesticastewater. Phosphorus is removed from sewage by reactionith either iron or aluminium, reducing its availability to plants

Kirkham, 1982; Kyle and McClintock, 1995), while nitrogen lev-ls in sewage are reduced by dispersal into the atmosphere. Theseost nutrients need to be replaced by applying artificial fertilisers.owever, fertiliser production consumes finite global reserves, i.e.hosphate rock (Cordell et al., 2009) and fossil fuels (Smil, 2001),

equires large amounts of energy (Wood and Cowie, 2004), andenerates considerable amounts of pollution (Hettige et al., 1994)nd greenhouse gases (Wood and Cowie, 2004). While systems for

∗ Corresponding author. Tel.: +46 70 531 03 12; fax: +46 920 49 28 18.E-mail address: Lea.Rastas@ltu.se (L. Rastas Amofah).

NmlrIca

925-8574/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.ecoleng.2012.06.030

© 2012 Elsevier B.V. All rights reserved.

ecovering P from wastewater have been studied in detail (Ashleyt al., 2009; Balmér et al., 2002; Valsami-Jones, 2004) and haveound practical applications (Cordell et al., 2011), the recovery of

from wastewater has received less attention. It is possible toecover P and N simultaneously in wastewater treatment plants byeans of struvite formation, but this can only recover a minor frac-

ion of the N found in sewage because the P:N ratio in struvite is 1:1Li and Zhao, 2003). Efficient recycling of P and N can be achievedy source separating systems (Larsen et al., 2009). However, thispproach would require costly reconstructions of existing sewageystems. Source separation systems aside, solutions for recover-ng wastewater P tend to be capital- and energy-intensive (Balmért al., 2002), and are thus sub-optimal in terms of sustainability.atural treatment systems such as constructed wetlands might be aore attractive solution in this respect because in addition to being

ow-cost and low-tech, they provide services by simultaneously

educing and recycling nutrients from wastewater into biomass.n addition to improving the quality of the receiving waters, theonstructed wetlands produce plant biomass, which can be useds a source of biofuel or as a soil enhancer and thus adds value

L. Rastas Amofah et al. / Ecological Engineering 47 (2012) 174– 181 175

F er saturation, N adsorbent is regenerated and the released N could be used in agriculture,w be used in agriculture.

tadahcrn(Iotao(e2waniaitLhatnwttnTsrt

Fig. 2. Monthly average temperatures at a weather station, 19 km south of the site,during the growing periods, along with long-term mean temperatures for Luleå,measured about 19 km south from the experimental site.

F

cris

2

2a

ig. 1. A concept of small scale treatment system designed for nutrient recovery. Afthile P filter material is replaced with fresh P sorbent and the spent P sorbent can

o the treatment process (Polprasert, 2007). Willows (Salix spp.)re widely used as energy crops because of their high biomass pro-uction, efficient nutrient uptake, high rates of evapotranspiration,nd flood tolerance. It has been suggested that in cold regions, theigh frost tolerance of reed canary grass (Landström et al., 1996)ould make it a more suitable crop than willow grown in shortotation coppices (Perttu, 1983). However, willow has some sig-ificant advantages over reed canary grass, e.g. lower ash contentPaulrud et al., 2010) and lower harvest losses (Larsson et al., 2006).n addition, the ratio of the energy input (i.e. the energy expendedn planting, fertilising, harvesting and transport) to the energy con-ent of the crop is lower for willows than for reed canary grass forny given biomass production level (Börjesson, 2007). A varietyf systems based on constructed wetlands planted with willowsBörjesson and Berndes, 2006; Gregersen and Brix, 2001; Larssont al., 2003; Obarska-Pempkowiak and Gajewska, 2005; Wu et al.,011) have been created and used to treat municipal and domesticastewater. However, these systems take up large areas of land

nd much of the N in the feed is lost into the atmosphere viaitrification–denitrification. Similarly, much of the P in the feed

s dispersed into the soil (Dimitriou and Aronsson, 2011; Kadlecnd Wallace, 2008; Larsson et al., 2003); the amount accumulatedn the plants represents only a minor fraction of the total quan-ity removed from the wastewater (Dimitriou and Aronsson, 2011;arsson et al., 2003). In the rural areas of the Nordic countries, manyomes are dependent on onsite systems. As such, there is a need for

robust and compact onsite treatment system that remains func-ional in a cold climate while efficiently removing and recoveringutrients. A suitable system is illustrated in Fig. 1, in which theastewater passes through a highly loaded willow bed and then

hrough N and P filters; the willow bed primarily serves to pre-reat the sewage while also producing biomass, with the bulk ofutrients being retained by (ad)sorption in the subsequent filters.

he N-saturated adsorbent can be regenerated, but the saturated Porbent has to be replaced with fresh material; in both cases, theecovered nutrients are available for agricultural use. The aim ofhe study described herein was to assess the performance of the

2Mtt

Fig. 3. Layout of the experimental plant with the dimensions of the bed and the b

rom SMHI (1991, 2011).

ompact willow bed with respect to biomass production, nutrientecovery from municipal wastewater and wastewater treatment,n a cold climate over a two-year period covering three growingeasons.

. Materials and methods

An experimental plant was constructed in Luleå (65◦41′N;2◦20′E), approximately 100 km below the Arctic Circle in Swedennd evaluated over three growing seasons, during 2005–2007;005 is referred to as Year 1, 2006 as Year 2, and 2007 as Year 3.

onthly averages temperatures and precipitation at a weather sta-

ion, 19 km south of the studied site, are presented in Fig. 2 duringhe growing seasons along with the long-term means.

ed material grain sizes. The sampling points are indicated with the arrows.

176 L. Rastas Amofah et al. / Ecological En

0.0

0.5

1.0

1.5

2.0

2.5

20.7. 20.11. 20.3. 20.7. 20.11. 20.3. 20.7.

Infl

uen

t fl

ow

, m

3/d

Growing

sea son 2

Growing

sea son 3

Growing

sea son 1

Fp

2

beiasTiwstptiest

i((ltcfb

2

aTpsvulmgaaanoi

(milkwTv3

2

ttltPwSawQA

ueio

2

tbuhstpTalsn

YsoYaudtw

bn

ig. 4. Wastewater flow into the experimental plant during the experimentaleriod, m3/d.

.1. Characteristics of the system

The experimental plant consisted of a dispersion zone and aed planted with two willow clones (see Fig. 3). The total area cov-red by the willow bed, including the dispersion zone, was 32 m2;ts depth was 0.8 m. The willow bed was lined with EPDM rubbernd filled with gravel of a grain size of 4–8 mm, while the disper-ion zone was filled with coarser gravel (8–16 and 16–32 mm).he wastewater entered the dispersion zone through a channelnlet with V-notches placed along the walls. The wastewater flow

as vertical in the dispersion zone and horizontal in the water-aturated willow bed (Fig. 3). A drainage layer was built beneathhe bed to ensure it was not lifted by displacement forces. Twolywood boards with a height of 30 cm were placed vertically athe bottom to improve the mixing of wastewater and precipitationn the willow bed. 10 cm thick insulation boards were installed onach side of the bed. In wintertime, the wells were equipped withubmersible heaters and covered with thick insulation boards, andhe inlet pipe was heated with a heating cable.

Two frost-tolerant willow clones, ((Salix schwerinii × S. vim-nalis) × S. viminalis) × S. burjatica (Karin) and Salix dasycladosGudrun), were planted in a plot of land with an area of 22 m2

see Fig. 3) during Year 1 in May. Growing conditions with respectight exposure and hydraulics were deemed to be rather similar forhe clones. The willow bed was established by planting 20 cm longuttings at a density of 25 units m−2 to enhance the biomass yieldrom Year 1 onwards; dense planting has been reported to increaseiomass yields (Bullard et al., 2002; Willebrand et al., 1993).

.2. Loading of the bed

The primary treatment effluent from a small municipal wastew-ter treatment plant was used as the source of incoming water.he flow of wastewater into the willow bed was dictated by thelant’s pumping cycle, and was maintained at 0.5–1.1 m3/d1 toimulate the wastewater production of a single household. Theolume of wastewater discharged into the system was monitoredsing a flow metre in the pump line, and is shown in Fig. 4. The

oading of the willow bed was calculated separately for each sum-er and winter period to show the conditions during the willows’

rowing and dormant seasons. The growing season was defineds those days for which the average temperature was equal to orbove 5 ◦C (Perttu, 1983). The inflow fluctuated during the autumns

nd winters of Years 1 and 2; the fluctuations were especially pro-ounced during Year 2, with the inflow exceeding 2 m3/d on twoccasions. This high inflow is reflected in the higher surface load-ng during the second winter period compared to other periods

atnl

gineering 47 (2012) 174– 181

36 vs. ∼20 mm/d, see Table 1). The load of nutrients and organicatter was generally higher during the winter periods than dur-

ng the summers. As judged by mass balance, the average dailyoad of the bed was 8 kg tot-N/ha, 6 kg NH4-N/ha and 1.2 tot-Pg/ha; the total nutrient load over the entire evaluation periodas 6270 kg tot-N/ha, 4810 kg NH4-N/ha and 970 kg tot-P/ha (see

able 1). During the experimental period, the daily BOD7 loadingaried between 15 and 27 kg/ha while the COD loading ranged from0 to 70 kg/ha.

.3. Wastewater sampling, analyses and evaluation

Grab samples were taken from the inlet (the pump well duringhe winter and distribution box during summer) and outlet well ofhe willow bed and analysed for total suspended solids (TSS), bio-ogical oxygen demand (BOD7), chemical oxygen demand (COD),otal-N (Tot-N), ammonium-N (NH4-N), nitrate-N (NO3-N), total-

(Tot-P) and phosphate-P (PO4-P). The analytical methods usedere Swedish standards: SS-EN 872 for TSS, SS-EN 1899-1 for BOD7,

S 028131 for Tot-N, and SS 028127 for Tot-P. Further, COD wasnalysed following the Merck spectroquant test cell test 14895,hereas QuAAtro Applications were used for NH4-N/NO3-N (No.-001-04, multitest M9/M10) and for PO4-P (test No. Q-031-04).ll samples were stored in a freezer prior to analysis.

The percentage reduction in each tested variable was calculatedsing a mass balance equation for each season and for the wholexperimental period based on average pollutant concentrationsn the influent, and the effluent during two sequential samplingccasions.

.4. Willow sampling, analyses and calculations

The performance of the willow bed was assessed for a period ofwo years, covering three growing seasons, as shown in Fig. 5. Theiomass production at the end of the first growing season was eval-ated by harvesting every second willow plant (taking stems thatad yet to bud) during the spring of Year 2, while for growing sea-ons 2 and 3, the weight of stems with leaves was measured duringhe autumn. During the autumn of Year 2, about 70% of the willowopulation was selected at random and harvested by uprooting.he remaining willows were harvested in the autumn of Year 3,nd the distribution of the fresh above-ground-mass between theeaves and the stems was evaluated. Samples of the leaves andtems were collected and analysed with respect to dry matter andutrient content.

The estimation of biomass production, i.e. stems and leaves, forears 2 and 3 was based on the mass ratio between leaves andtems obtained from Year 3, whereas for Year 1, the mass ratio wasnly used for the estimation of leaf biomass production. Further, forears 1 and 2, the estimated biomass production was based on theverage plant shoot biomass, whereas for Year 3, total weight wassed. The mass ratio was deemed to be altered to a minor extenturing the experimental period. For nutrient uptake estimations,he average nutrient concentration in samples collected during thehole experimental period was used.

The dry matter content of the stems and leaves was determinedy drying chipped plant material for 2 d at 107 ◦C. The samples forutrient content analysis were stored in a freezer prior to analysis

nd were determined using Swedish standards SS 028101, ed.1 forot-N and SS-EN 13346 mod/SS11885-1 for tot-P. Analyses of theutrient content of plant material were performed at the accredited

aboratory of ALcontrol AB in Umeå, Sweden.

L. Rastas Amofah et al. / Ecological Engineering 47 (2012) 174– 181 177

Table 1Average wastewater load supplied to the willow bed.

Surface load (mm/d) Tot-N (kg/ha, d) NH4-N (kg/ha, d) Tot-P (kg/ha, d) BOD7 (kg/ha, d) COD (kg/ha, d)

10 May Year 1–19 October Year 1 23 6 5 1.0 27 4020 October Year 1–28 April Year 2 19 9 6 1.3 27 4929 April Year 2–10 October Year 2 19 6 6 0.9 15 –11 October. Year 2–12 April Year 3 36 9 7 1.4 – 6913 April Year 3–1 October Year 3 17 5 4 0.8 – 31Average 23 8 6 1.2 – –

Total 580a 6270b 4810b 970b – –

3

3

wDiDscttob

siTtalb(gtD

3

ir3o

GbifaPTrtbgpgtTae

3

3

lowfrfi

3

a Unit: m3.b Unit: kg/ha.

. Results

.1. Biomass production

The overall biomass production over the three growing seasonsas estimated to 21 ton DM/ha for S. × viminalis (Karin) and 19 tonM/ha for S. dasyclados (Gudrun), see Table 2. During the first grow-

ng season, the estimated productivity of the two clones was 1.4 tonM/ha, whereas during the second and third growing seasons, the

tem biomass production of S. × viminalis exceeded that of S. dasy-lados by 2 and 0.5 ton DM/ha, respectively. During the second year,he stem biomass production increased by a factor of 7–9 relative tohe previous year, reaching 10–12 ton DM/ha. Approximately 70%f the stand was uprooted during the autumn of Year 2 reducingiomass production by ∼30–40%.

During the first growing season, both willow clones had theame estimated leaf biomass production (0.2 ton DM/ha), whichncreased by a factor of 10 or more in the following year (seeable 2). In contrast to the situation with stem biomass produc-ivity, the leaf biomass production of S. × viminalis was generally

few hundred kg DM/ha less than that of S. dasyclados becauseeaves accounted for a smaller proportion of the total above-groundiomass in S. × viminalis (14%) than was the case for S. dasyclados20%). The extensive harvesting of the willow stand after the secondrowing season reduced the production of leaf biomass in Year 3:he third year leaf biomass productivity of S. × viminalis was 2.0 tonM/ha while that for S. dasyclados was 2.4 ton DM/ha (Table 2).

.2. Nutrient accumulation in stems and leaves

The average concentrations of N in the stems and leaves

n S. × viminalis was 4690 ± 1380 and 23,750 ± 5700 mg/g DM,espectively, and the corresponding figures for S. dasyclados were820 ± 560 and 26,480 ± 4060. The corresponding concentrationsf P for Karin were 800 ± 180 and 2330 ± 730 mg/g DM, and for

tsh

Year

Season

Year 0 Y ear 1 Year 2 Autumn Winter Spring Summer Autumn Winter Sp

The plant is constructed.

The bed is filled with wastewater and shortly

after willows are planted.

The plant becomes operational, sampling

of wastewater commences.

Biomass accumulatestimated by cutting

50% of the willows. Sof stem collecte

Fig. 5. Calendar for the evaluation peri

udrun, 810 ± 390 and 2730 ± 400, respectively. The increasediomass production in Year 2 was associated with substantial

ncreases in the uptake of N and P after the first growing season,or both clones. The total uptake of N in stems and leaves increasedt least by a factor of 8, to 110 kg N/ha overall; similarly, overall

uptake increased by a factor of 8, to 19 kg P/ha (see Table 3).he extensive harvesting at the end of the second growing seasoneduced the biomass yield in the following year, and thus reducedhe total nutrient uptake by 20–30%. The leaf N pool was greatery 40–100% than the stem N pool for S. dasyclados during the threerowing seasons, whereas for S. × viminalis, the size of the leaf Nool was similar to that of the stem pool. For P, the stem pool wasreater than the leaf pool for both clones. In total, the N accumula-ion was ∼210 kg/ha for both clones, and that for P was ∼30 kg/ha.he harvesting scheme used in this work recovered 185 kg N/hand 25 kg P/ha that had accumulated in the biomass during thexperimental period.

.3. Efficiency of pollutant removal

.3.1. Total suspended solidsThe influent concentrations varied greatly, with the minimum

evels below 50 mg/l and peak values above 100 mg/l. Despite theccasionally high loadings, the willow bed removed TSS from theastewater very effectively; more than 90% of the TSS was removed

rom the influent (Table 4). The TSS concentrations in the effluentemained relatively steady throughout the experiment, decliningrom 6 ± 3 mg/l to 2 mg/l over the course of Year 1 and then remain-ng at 2 ± 1 mg/l during Years 2 and 3.

.3.2. Organic matter

The concentrations of oxygen-consuming species, measured in

erms of BOD7 and COD, in the influent were high and occasionallypiked to 368 and 502 mg/l, respectively (see Fig. 6a). Typically,owever, the BOD7 and COD ranged from 100 to 200 mg/l and 190

Year 3

ring Summer Autu mn Winter Spring Summer Autu mn

ion is down amples d.

About 70% of the willows are uprooted with foliage to estimate biomass accumul ation . Sampl es of leaves and stems collected.

The end of the evaluation period; all of the willows are harvested with foliage and

the biomass accumulation is estimated. Samples of leaves and stems collected.

od with key events highlighted.

178 L. Rastas Amofah et al. / Ecological Engineering 47 (2012) 174– 181

Table 2Annual biomass production during the evaluation period.

Willow clone Growing season 1(ton DM/ha)

Growing season 2(ton DM/ha)

Growing season 3(ton DM/ha)

Total (ton DM/ha) Average (tonDM/ha, year)

S. × viminalis (Karin)Stems 1.4 12a 7.7a 21 7.0Leaves 0.19a 2.1a 2.0a 4.3 –b

S. dasyclados (Gudrun)Stems 1.4 10a 7.2a 19 6.2Leaves 0.25a 2.6a 2.4a 5.2 –b

a An estimate based on the mass proportion obtained during the third growing season.b Not given.

Table 3Nutrient uptake by stems and leaves after each growing season during the evaluation period and the total amount of nutrients accumulated in willow stems and leaves after.

Willow clone Year S. × viminalis (Karin) S. dasyclados (Gudrun)

Stem Leaf Total Stem Leaf Total

Nitrogen in willows (kgtot-N/ha)

1 7 5 12 5 7 122 60 50 110 40 70 1103 40 50 90 30 60 90Total 107 105 212 75 137 212

1 1 <1 2 1 <1 25

5

1

tdsrrlov5

3

reiawtwAtcitDm

w<

3

Pior3tmrccPcr

4

4

TR

n

Phosphorus in willows(kg tot-P/ha)

2 10

3 6

Total 17 1

o 300 mg/l, respectively, with exceptions when the influent wasiluted due to precipitation during the winter of Year 2 and bynowmelt during the spring of Year 3. The willow bed efficientlyeduced both BOD7 and COD, by >85% and ∼80% respectively,egardless of the initial values (Fig. 6a; Table 4). The reduced BOD7oading (Table 1) was reflected in the BOD7 values for the effluentver time, which were reduced from 33 to 11 mg/l (Fig. 6a). Con-ersely, the COD values for the effluent remained between 40 and5 mg/l throughout the experimental period (Fig. 6a).

.3.3. Nitrogenous compoundsThe tot-N and NH4-N concentrations in the influent typically

anged between 30–55 and 10–40 mg/l respectively, with somexceptions when the influent was diluted due to precipitation dur-ng the autumns or by snowmelt during the springs (Fig. 6b). Theverage reduction in tot-N and NH4-N after passage through theillow bed did not vary greatly with the seasons; the tot-N concen-

rations was typically reduced by 35–45% after passage through theillow bed while the NH4-N concentration was reduced by 20–30%.n exception occurred at the end of the experimental period (i.e. in

he summer of Year 3), when the average reduction was 5–15 per-entage points greater (Table 4). This reduced the N concentration

n the effluent. Thus, the tot-N concentration declined from 26 ± 5o 15 ± 2 mg/l, while that of NH4-N went from 23 ± 3 to 12 ± 2 mg/l.uring the experimental period, NH4-N typically accounted forore than 80% of the N in the effluent. The NO3-N concentration

(fS

able 4emoval efficiency of the willow bed system (%).

Period Tot-N NH4-N

20 July Year 1–19 October Year 1 39 25

20 October Year 1–28 April Year 2 36 25

29 April Year 2–10 October Year 2 34 20

11 October. Year 2–12 April Year 3 45 29

13 April Year 3–1 October Year 3 50 39

Average 43 32

= 3–6 grab samples.a –, not measured.b Period 29 April–16 May, Year 2.

15 8 6 1411 6 5 1128 15 12 27

as low in both the influent and in the effluent, being <0.7 and0.03 mg/l respectively.

.3.4. Phosphorus compoundsLike the tot-N and NH4-N influent concentrations, the tot-P and

O4-P concentrations in the influent occasionally dropped dur-ng the autumns and springs because of dilution by precipitationr snowmelt. Otherwise, the influent concentrations were usuallyanged between 5–10 and 3–5 mg/l, respectively. Approximately0% of the tot-P in the influent was removed after passage throughhe willow bed, although this declined to ∼10% during the sum-

ers of Years 2 and 3 (Fig. 6c). The studied willow bed did notetain soluble P (PO4-P) to any great extent; typically, the influentoncentrations were reduced by less than 10% and in some cases theoncentration in the effluent was higher than that in the influent.O4-P accounted for more than 90% of the P in the effluent; tot-Poncentrations ranged from 4 to 6 mg/l while PO4-P concentrationsanged from 3 to 5 mg/l.

. Discussion

.1. Production of biomass in cold climate

The average biomass production achieved in this study6.2–7.0 ton DM/ha, year; Table 2) was comparable to that expectedor a well-maintained commercial willow plantation in southernweden (7–10 ton DM/ha, year) or elsewhere in northern Europe

Tot-P BOD7 COD TSS

31 85 79 –a

26 88 80 9410 88b –a –a

32 –a 81 –a

7 –a 79 9623 87 80a 94

L. Rastas Amofah et al. / Ecological En

0

2

4

6

8

10

20.7. 31.8. 20.12. 24.4. 4.7. 26.3. 27.7.

Co

nc

en

tra

tio

n, m

g/l

Influent tot-P Influent PO4-P

Efflu ent t ot-P Efflue nt PO4-P

Growi ng

sea song 1

Growi ng

seasong 3 Growi ng

sea song 2

0

10

20

30

40

50

60

20.7. 31.8. 20.12. 24.4. 4.7. 26.3. 27.7. Co

ncen

trati

on

, m

g/l

Influ ent tot-N In flue nt NH4-N

Effluent tot-N Effluen t NH4-N

Growi ng

sea song 1

Growi ng

sea song 2

Growi ng

seasong 3

0

20

40

60

80

100

0

100

200

300

400

500

20.7. 31.8. 20.12. 24.4. 4.7. 26.3. 27.7.

Eff

luen

t co

ncen

trati

on

, m

g/l

Infl

uen

t co

ncen

trati

on

, m

g/l

Influent BOD7 Influent COD

Effl uent BOD7 Effl uent COD

Growing sea son 1

Growing season 2

Growing sea son 3

510 a

b

c

FN

(Lcnt2m(twtsttnwaas

t(isgtahfciodt

meidpitnwbaw

utanwchhptmtDwofllha(pvttets

4

caLarsson et al., 2003). However, the willows at the studied site

ig. 6. Influent and effluent concentrations of (a) BOD7 and COD, (b) tot-N and NH4- and (c) tot-P and PO4-P during the experimental period.

Bullard et al., 2002; Ericsson, 1994; Gregersen and Brix, 2001;arsson et al., 2003; Mortensen et al., 1998). This good performancean be attributed to multiple factors. First, the wastewater providedutrients and water at a level that greatly exceeded those requiredo maintain growth during the growing seasons (cf. Larsson et al.,003) as water shortage and insufficient fertilisation are among theost important factors limiting the growth potential of willows

Jonsson, 1997). Second, willow cuttings were planted at a densityhat was 15–25 times greater than is recommended for commercialillow plantations (Gustafsson et al., 2007). This made it possible

o harvest parts of the willow stand at the end of each growingeason, which may have enhanced the biomass yield accordingo Bullard et al. (2002) and Willebrand et al. (1993). In additiono the partial annual harvesting, the dense planting and coarse,utrient-poor bed material made the willow bed inhospitable toeeds, which are known to reduce biomass production (Gregersen

nd Brix, 2001; Tahvanainen and Rytkönen, 1999). Third, the aver-ge monthly temperatures during the three years in which thetudy was conducted were higher than the long-term average

w(t

gineering 47 (2012) 174– 181 179

emperatures for the region, especially during growing season 2see Fig. 2). Also, the biomass production at the site may have beennfluenced by an edge effect, especially during the second growingeason. Finally, the intensive fertigation might have extended therowing season by providing heat (the influent wastewater had aemperature of 6–15 ◦C) and nutrients (von Fircks et al., 2001), cre-ting a more favourable environment for the willows than wouldave been possible otherwise. Notably, while the willows sustained

rost damage in the form of frozen top shoots, plant survival waslose to 100%; for comparative purposes, the survival rate observedn a study conducted in a similar climate without the applicationf wastewater was 40–70% (Söderström, 2010). It thus seems thatense planting and intensive wastewater fertigation throughouthe year promotes biomass production in cold climates.

The stem biomass production of S. × viminalis (Karin) was esti-ated to be slightly higher than that of S. dasyclados (Gudrun),

specially during Year 2 (Table 2). Conversely, S. dasyclados exhib-ted greater leaf biomass production (Table 2). However, theifference in biomass yields decreased after the size of the willowopulation was reduced by 70% at the end of the second grow-

ng season. The difference in yields may be due to differences inhe two clones’ genetic properties; S. × viminalis is taller and thin-er than S. dasyclados (Bullard et al., 2002; Söderström, 2010), andillows with vertical growth are favoured in dense plantations

ecause of competition for light (Bullard et al., 2002). Thus, wheniming to increase biomass yields through dense planting, clonesith vertical growth should be used.

The studied willow bed produced high quantity of biomass pernit area. This suggests that using scaled-up willow beds for simul-aneous wastewater treatment and biomass production would ben attractive way of recovering resources. However, it should beoted that the method may be subject to some limitations. Theillow stand was densely planted and harvested annually which

ontributed to a large number of young, spindly stems with aigh proportion of bark in the stems. Such a bark-rich feedstockas a lower biofuel quality than thicker stems with lower pro-ortion of bark (Adler et al., 2008). Also, if it is necessary to linehe bottom of the plot to prevent groundwater contamination and

aintain control over the inflow to the subsequent (ad)sorption fil-ers, the construction costs of the willow bed increase considerably.uring the experimental period, the system produced sufficientillow stem biomass to generate 240–270 kWh of power (based

n 18.5 MJ/kg, Kenney et al., 1990; Xiong et al., 2008). A horizontal-ow prefiltering willow bed of this kind would thus produce much

ess energy than the amount needed by a typical Swedish house-old for heating and hot water (Swedish Energy Agency, 2011)nd is unlikely to be economically viable for biofuel productionGustafsson et al., 2007). However, Börjesson and Berndes (2006)ointed out that the greatest economic benefit of using willowegetation filters for wastewater treatment comes from the lowerreatment costs compared to conventional methods for N reduc-ion rather than from the biofuel production. In addition to theconomic aspects, willow treatment systems are more sustainablehan conventional alternatives for N reduction because they con-ume less energy (Mulder, 2003).

.2. Nutrient recovery

The nutrient accumulation in the willows used in this work wasomparable to that reported for previous similar studies (Dimitriound Aronsson, 2011; Ericsson, 1994; Gregersen and Brix, 2001;

ere densely planted, which tends to cause self-thinning over timeBullard et al., 2002). To counteract this effect, the willow popula-ion was extensively reduced by uprooting 70% of the population

1 ical En

anb1omd2stgomfispTlu

immncisotvsdo

4

spcshesTFtpssttrOo(Tecatfi

p

ucunVirt2iil5mtnfyf(

ss(str(seiiuoa

5

pTwwibipiwvuttsea

f

80 L. Rastas Amofah et al. / Ecolog

t the end of the second growing season. This only reduced theutrient accumulation in the following year by <30% which maye due to high stand maturity with well-developed roots (Ericsson,994). Interestingly, despite the lower stem biomass productionf S. dasyclados, both clones exhibited fairly similar nutrient accu-ulation (see Table 3). This might be due to genetic differences; S.

asyclados grows more laterally than S. × viminalis (Bullard et al.,002; Söderström, 2010), and the lower N accumulation in thetems of S. dasyclados was compensated for by a higher accumula-ion in the leaves. A previous study (Ericsson, 1994) reported a farreater accumulation of N in the leaves compared to the presentne. The higher N accumulation in the study of Ericsson (1994)ay be due to the use of a lower planting density, which would

avour leaf biomass production. At the studied site, annual harvest-ng before abscission facilitated nutrient recovery from the fieldite, i.e. ∼90% of the nutrients accumulated in the above-groundarts were removed from the site during the experimental period.his suggests that nutrient removal could be optimised by usingaterally growing willows with a lower planting density than wassed in this work, and harvesting them annually before leaf fall.

When willow biomass is produced for use in biofuel, the stands harvested after abscission at intervals of 3–4 years. Since 50% or

ore of the above-ground N is contained in the leaves (Table 3),uch of the plant’s N is recycled through leaf fall at sites with high

utrient levels (von Fircks et al., 2001). Furthermore, during theombustion of biofuel, much of the P from the biomass is retainedn the ash, whereas the N is almost completely converted to gaseouspecies, i.e. N2 and NOx (Obernberger et al., 1997). As such, anotherption of utilising nutrients in the produced willow biomass may beo harvest a portion of the willow stand annually, compost the har-ested biomass, and use the compost residue as a soil conditioner. Aystem solution including composting has to be evaluated in moreetail with respect to, e.g. crop most suitable for composting toptimise nutrient recovery.

.3. Wastewater treatment

Septic tanks, commonly used as a first treatment step in on-ite sanitation systems, primarily remove suspended solids andarticulate organic matter. However, the septic tank effluent stillontains considerable amounts of organic matter and particulateolids which with poor septic tank maintenance can cause evenigher effluent concentrations of TSS and organic matter. This couldxplain the sporadic high concentrations of TSS and BOD7 in theeptic tank effluent fed to the studied willow bed; occasionally, theSS levels exceeded 120 mg/l and the COD was above 500 mg/l (seeig. 6a cf. USEPA, 2002). Low concentrations of TSS and organic mat-er in the influent fed to a reactive wastewater filter are essential forroper filter functioning, because these constituents can block theorption sites and clog the filter. As such, reactive filter materialshould be used after removing particles and organic matter ratherhan as stand-alone units (Weber et al., 2007) or incorporating reac-ive filters materials into constructed wetlands to enhance nutrientemoval (e.g. Drizo et al., 1997; Karczmarczyk and Renman, 2011;barska-Pempkowiak and Gajewska, 2005). In addition, the usef separate treatment units makes it easier to replace exhaustedad)sorbents. The studied willow bed significantly reduced theSS and organic matter content of the influent water, to anxtent comparable to that achieved with prefiltering systemsonsidered in previous studies (Jenssen et al., 2010; Mæhlumnd Stålnacke, 1999). The studied willow bed is thus an effec-

ive pretreatment system for use with downstream (ad)sorptionlters.

In the studied highly loaded willow bed, a comparatively smallroportion of the total nutrient content of the influent was taken

pcot

gineering 47 (2012) 174– 181

p into the harvestable biomass of the willows (Tables 1 and 3f. Table 4). This is consistent with previous claims that nutrientptake by plants accounts for only a small proportion of the totalutrient removal in wastewater filtering systems (e.g. Brix, 1994;ymazal, 2007). This is because N entering constructed wetlands

s primarily removed via nitrification–denitrification, whereas P isemoved by reaction with Fe, Al, and Ca in the bed material ashe wastewater flows through the constructed wetland (Vymazal,007). In the horizontal flow constructed wetlands, N reduction

s commonly limited by low nitrification, which results from annsufficient oxygen supply. The NH4-N reduction in the studied wil-ow bed was initially low (30–40%) but increased over time to ca.0% by the end of the third growing season (Fig. 6b; Table 4). Thisight reflect the development of the willows’ (finer) roots during

he first three years after planting (Rytter, 2001), which promotesitrification (Brix, 1994). Without reductions in the N loading, no

urther increases in the N reduction should be expected after threeears, by which point the root system would be fully developedor plants whose growth is not limited by nutrient availabilityEricsson, 1994).

The 40–50% reduction in tot-N concentration achieved in thistudy is typical for constructed wetlands, regardless of the plantpecies used (e.g. Vymazal, 2007). The P retention was low∼10–30%) because of the coarse bed material with low specificurface and low quantities of Fe, Al or Ca. However, after passinghrough the horizontal flow constructed wetland, the bulk of theemaining nutrients were present as soluble species, N as NH4-NFig. 6b) while most of the P was in the form of PO4-P (Fig. 6c). Theseoluble and thus reactive forms of N and P in the willow bed efflu-nt together with an efficient reduction of TSS and organic matterndicate that there is a potential for capturing most of the nutrientsn the willow bed effluent by subsequent sorption filters, e.g. bysing Ca-based materials for P (Rastas Amofah and Hanæus, 2007)r fine grained clinoptilolite filter in the case of NH4-N (Hedströmnd Rastas Amofah, 2008).

. Conclusions

The frost-tolerant willow clones produced good biomass yieldser unit area (20 ton dry matter/ha) over three growing seasons.he intensive fertigation, dense planting, continuous harvestingere more favourable for biomass production with vertical growthillow species rather than those exhibiting lateral growth grown

n cold climate. The results indicated that nutrient recovery cane maximised by using horizontal-growth willow clones under

ntensive fertigation and continuous harvesting with less denselanting than was used in this work. The nitrogen accumulation

n above-ground biomass was 210 kg/ha and that of phosphorusas 30 kg/ha, of which about 90% were be removed under the har-

esting scheme used. Despite this, the quantity of nutrients takenp into the willow biomass represented only a small fraction ofhe loaded or removed amounts. However, the willow bed facili-ates nutrient recovery in subsequent adsorption filters because itubstantially reduced levels of particulate and organic matter andnsured that the bulk of the remaining nutrients in were in solublend thus in reactive forms.

Overall, our results indicate that frost-tolerant willow clonesertigated with wastewater may be an attractive option as a com-

onent of a nutrient-retaining wastewater treatment system forold climates. However, it is necessary to further assess the impactf potential long-term variations in hydraulic load and ambientemperature on the biomass yield of willow grown in cold climates.

ical En

A

Aanf

R

A

A

B

B

B

B

B

C

C

D

D

E

G

G

H

H

J

J

KK

K

K

K

L

L

L

L

L

M

M

M

O

O

P

P

P

R

R

S

S

S

S

S

T

U

V

W

W

v

W

W

L. Rastas Amofah et al. / Ecolog

cknowledgements

We gratefully acknowledge the Swedish Water and Wastewaterssociation for financing the construction of the experimental plantlong with the municipal office of Luleå which also provided tech-ical assistance. The authors also thank Em. Prof. Jörgen Hanæus

or the ideas he proposed during the initial phase of this study.

eferences

dler, A., Dimitriou, I., Aronsson, P., Verwijst, T., Weih, M., 2008. Wood fuel qualityof two Salix viminalis stands fertilised with sludge, ash and sludge–ash mixtures.Biomass Bioenergy 32 (10), 914–925.

shley, K., Mavinic, D., Koch, F. (Eds.), 2009. International Conference on NutrientRecovery from Wastewater Streams. IWA Publishing, Vancouver, Canada.

almér, P., Book, K., Hultman, B., Jönsson, H., Kärrman, E., Levlin, E., Palm, O., Schön-ning, C., Seger, A., Stark, K., Söderberg, H., Tideström, H., Åberg, H., 2002. Systemsfor Re-use of Phosphorus from Sewage. Report 5221. Swedish EnvironmentalProtection Agency, Stockholm, Sweden.

rix, H., 1994. Functions of macrophytes in constructed wetlands. Water Sci. Tech-nol. 29 (4), 71–78.

ullard, M.J., Mustill, S.J., McMillan, S.D., Nixon, P.M.I., Carver, P., Britt, C.P., 2002.Yield improvements through modification of planting density and harvestfrequency in short rotation coppice Salix spp. – 1. Yield response in two mor-phologically diverse varieties. Biomass Bioenergy 22 (1), 15–25.

örjesson, P., 2007. Bioenergi från jordbruket – en växande resurs. Bilagedel. SOU2007:36. Statens offentliga utredningar, Stockholm, Sweden.

örjesson, P., Berndes, G., 2006. The prospects for willow plantations for wastewatertreatment in Sweden. Biomass Bioenergy 30 (5), 428–438.

ordell, D., Drangert, J.-O., White, S., 2009. The story of phosphorus: global foodsecurity and food for thought. Global Environ. Change 19 (2), 292–305.

ordell, D., Rosemarin, A., Schröder, J.J., Smit, A.L., 2011. Towards global phospho-rus security: a systems framework for phosphorus recovery and reuse options.Chemosphere 84 (6), 747–758.

imitriou, I., Aronsson, P., 2011. Wastewater and sewage sludge application to wil-lows and poplars grown in lysimeters – plant response and treatment efficiency.Biomass Bioenergy 35 (1), 161–170.

rizo, A., Frost, C.A., Smith, K.A., Grace, J., 1997. Phosphate and ammonium removalby constructed wetlands with horizontal subsurface flow, using shale as a sub-strate. Water Sci. Technol. 35 (5), 95–102.

ricsson, T., 1994. Nutrient cycling in energy forest plantations. Biomass Bioenergy6 (1–2), 115–121.

regersen, P., Brix, H., 2001. Zero-discharge of nutrients and water in a willowdominated constructed wetland. Water Sci. Technol. 44 (11–12), 407–412.

ustafsson, J., Larsson, S., Nordh, N.-E., 2007. Manual för Salixodlare. LantmännenAgroenergi, Sweden.

edström, A., Rastas Amofah, L., 2008. Adsorption and desorption of ammonium byclinoptilolite adsorbent in municipal wastewater treatment systems. J. Environ.Eng. Sci. 7 (1), 53–61.

ettige, H., Martin, P., Singh, M., Wheeler, D., 1994. IPPS: The Industrial PollutionProjection System. 1431. World Bank, Washington.

enssen, P.D., Krogstad, T., Paruch, A.M., Mæhlum, T., Adam, K., Arias, C.A., Heis-tad, A., Jonsson, L., Hellström, D., Brix, H., Yli-Halla, M., Vråle, L., Valve,M., 2010. Filter bed systems treating domestic wastewater in the Nordiccountries – performance and reuse of filter media. Ecol. Eng. 36 (12),1651–1659.

onsson, H., 1997. Ramprogram energiskog 1992–1995. 1997:3. National SwedishBoard for Industrial and Technical Development (NUTEK), Stockholm, Sweden.

adlec, R., Wallace, S., 2008. Treatment Wetlands. CRC Press, Boca Raton, FL, USA.arczmarczyk, A., Renman, G., 2011. Phosphorus accumulation pattern in a

subsurface constructed wetland treating residential wastewater. Water 3,146–156.

enney, W.A., Sennerby-Forsse, L., Layton, P., 1990. A review of biomass qualityresearch relevant to the use of poplar and willow for energy conversion. Biomass21 (3), 163–188.

irkham, M.B., 1982. Agricultural use of phosphorus in sewage sludge. In: Brady, N.C.(Ed.), Advances in Agronomy. Academic Press Inc., New York, USA, pp. 129–163.

yle, M.A., McClintock, S.A., 1995. The availability of phosphorus in municipalwastewater sludge as a function of the phosphorus removal process and sludgetreatment method. Water Environ. Res. 67 (3), 282–289.

andström, S., Lomakka, L., Andersson, S., 1996. Harvest in spring improves yield

and quality of reed canary grass as a bioenergy crop. Biomass Bioenergy 11 (4),333–341.

arsen, T.A., Alder, A.C., Eggen, R.I.L., Maurer, M., Lienert, J., 2009. Source separation:will we see a paradigm shift in wastewater handling? Environmental Science &Technology 43 (16), 6121–6125.

V

X

gineering 47 (2012) 174– 181 181

arsson, S., Cuingnet, C., Clause, P., Jacobsson, I., Aronsson, P., Perttu, K., Rosenqvist,H., Dawson, M., Wilson, F., Backlund, A., Mavrogianopoulus, G., Riddel-Black,D., Carlander, A., Stenstrøm, T.A., Hasselgren, K., 2003. Short-Rotation WillowBiomass Plantations Irrigated and Fertilised with Wastewaters Results from a4-Year Multidisciplinary Field Project in Sweden, France, Northern Ireland andGreece. Sustainable Urban Renewal and Wastewater Treatment, No. 37. DanishEnvironmental Protection Agency.

arsson, S., Örberg, H., Kalén, G., Thyrel, M., 2006. Rörflen som energigröda. Erfaren-heter från fullskaleförsök vid Biobränsletekniskt Centrum (BTC) i Umeå underåren 2000-2004. BTK-rapport 2006:11. Unit of Biomass Technology and Chem-istry, Swedish University of Agricultural Sciences, Umeå, Sweden.

i, X.Z., Zhao, Q.L., 2003. Recovery of ammonium–nitrogen from landfill leachate asa multi-nutrient fertilizer. Ecol. Eng. 20 (2), 171–181.

æhlum, T., Stålnacke, P., 1999. Removal efficiency of three cold-climate con-structed wetlands treating domestic wastewater: effects of temperature,seasons, loading rates and input concentrations. Water Sci. Technol. 40 (3),273–281.

ortensen, J., Nielsen, K.H., Jørgensen, U., 1998. Nitrate leaching during establish-ment of willow (Salix viminalis) on two types soil and at two fertilization levels.Biomass Bioenergy 15 (6), 457–466.

ulder, A., 2003. The quest for sustainable nitrogen removal technologies. WaterSci. Technol. 48 (1), 67–75.

barska-Pempkowiak, H., Gajewska, M., 2005. Recent developments in wastewatertreatment in constructed wetlands in Poland. In: Omelchenko, A., Pivovarov, A.,Swindall, W. (Eds.), Modern Tools and Methods of Water Treatment for Improv-ing Living Standards. Springer, Dordrecht, Netherlands, pp. 279–293.

bernberger, I., Biedermann, F., Widmann, W., Riedl, R., 1997. Concentrations ofinorganic elements in biomass fuels and recovery in the different ash fractions.Biomass Bioenergy 12 (3), 211–224.

aulrud, S., Davídsson, K., Holmgren, M.A., Hedman, H., Öhman, R., Leffler, J., 2010.The Conditions for Use of Reed Canary Grass Briquettes and Chopped ReedCanary Grass in Small Heating Plants. SP Report 2010:60. SP Technical ResearchInstitute of Sweden, Borås, Sweden.

erttu, K.L., 1983. Temperature restraints on energy forestry in Sweden. Int. J. Biome-teorol. 27 (3), 189–196.

olprasert, C., 2007. Organic Waste Recycling Technology and Management. IWAPublishing, London, UK.

astas Amofah, L., Hanæus, J., 2007. Nutrient recovery in a small scale wastewatertreatment plant in cold climate. Vatten 62, 355–368.

ytter, R.-M., 2001. Biomass production and allocation, including fine-root turnover,and annual N uptake in lysimeter-grown basket willows. Forest Ecol. Manage.140 (2–3), 177–192.

mil, V., 2001. Enriching the Earth: Fritz Haber, Carl Bosch, and the Transformationof World Food Production. MIT Press, Cambridge, USA.

wedish Energy Agency, 2011. Energy Statistics for One and Two-Dwelling Buildingsin 2009. ES 2011:01. Swedish Energy Agency.

wedish Meteorological and Hydrological Institute (SMHI), 1991. Temperature andPrecipitation in Sweden 1961–1990 Reference Normals. Meteorologi 81. SMHI,Norrköping, Sweden.

MHI, 2011. Klimatdata. Års-och månadsstatistik, http://www.smhi.se/klimatdata/meteorologi/ars-och-manadsstatistik-2.1240 (accessed 7.11.11).

öderström, Y., 2010. Demonstrationsodling av Salix i Norrland med frosttolerans ifokus. Report 1145. VÄRMEFORSK Service AB, Stockholm, Sweden.

ahvanainen, L., Rytkönen, V.M., 1999. Biomass production of Salix viminalis insouthern Finland and the effect of soil properties and climate conditions onits production and survival. Biomass Bioenergy 16 (2), 103–117.

nited States Environmental Protection Agency (USEPA), 2002. Onsite WastewaterTreatment Systems Manual. EPA/625/R-00/008. Office of Water, USEPA.

alsami-Jones, E. (Ed.), 2004. Phosphorus in Environmental Technology Principlesand Applications. IWA Publishing, London, UK.

eber, D., Drizo, A., Twohig, E., Bird, S., Ross, D., 2007. Upgrading constructed wet-lands phosphorus reduction from a dairy effluent using electric arc furnace steelslag filters. Water Sci. Technol. 56 (3), 135–143.

illebrand, E., Ledin, S., Verwijst, T., 1993. Willow coppice systems in short rotationforestry: effects of plant spacing, rotation length and clonal composition onbiomass production. Biomass Bioenergy 4 (5), 323–331.

on Fircks, Y., Ericsson, T., Sennerby-Forsse, L., 2001. Seasonal variation of macronu-trients in leaves, stems and roots of Salix dasyclados Wimm. grown at twonutrient levels. Biomass Bioenergy 21 (5), 321–334.

ood, S., Cowie, A., 2004. A Review of Greenhouse Gas Emission Factors for FertiliserProduction. International Energy Agency, Paris, France.

u, S., Austin, D., Liu, L., Dong, R., 2011. Performance of integrated household con-structed wetland for domestic wastewater treatment in rural areas. Ecol. Eng.37 (6), 948–954.

ymazal, J., 2007. Removal of nutrients in various types of constructed wetlands.Sci. Total Environ. 380 (1–3), 48–65.

iong, S., Zhang, Q.-G., Zhang, D.-Y., Olsson, R., 2008. Influence of harvest time onfuel characteristics of five potential energy crops in northern China. Bioresour.Technol. 99 (3), 479–485.