Da

AL

a

ARRA

KSBEDW

1

wiimvptd[d(dolo

ettt

f

t

1d

Chemical Engineering Journal 172 (2011) 430– 443

Contents lists available at ScienceDirect

Chemical Engineering Journal

jo u r n al hom epage: www.elsev ier .com/ locate /ce j

ewatering mechanisms in pilot-scale Sludge Drying Reed Beds: Effect of designnd operational parameters

lexandros I. Stefanakis, Vassilios A. Tsihrintzis ∗

aboratory of Ecological Engineering and Technology, Department of Environmental Engineering, School of Engineering, Democritus University of Thrace, 67100 Xanthi, Greece

r t i c l e i n f o

rticle history:eceived 12 January 2011eceived in revised form 30 May 2011ccepted 31 May 2011

a b s t r a c t

Eleven pilot-scale Sludge Drying Reed Beds (SDRBs) have been constructed, and operated treatingmunicipal activated sludge. Quantification of the dewatering mechanisms in these systems (i.e., evap-otranspiration and draining) is attempted here, and the effects of different parameters, i.e., the level ofsludge loading, the presence of reeds, the gradation of substrate materials etc., on these mechanisms

eywords:ludge treatment wetlandsiosolidsvapotranspirationrainageater budget

are presented. Results reveal that evapotranspiration is the major dewatering process that takes placein SDRB systems, accounting for 58–84% of total water losses, depending on unit characteristics, sludgeloading, season and other parameters. Draining follows in significance, with respective values varyingfrom 13% to 41% of total water losses, while 1% to 4% of water remains in the sludge layer. The presence ofreeds was found necessary, since it improved the dewatering efficiency. Precipitation and temperaturewere also found to affect the dewatering process under temperate climate conditions.

. Introduction and background

Sludge treatment and handling are today a major issue inastewater treatment. The basic processes for sludge treatment

nclude stabilisation, thickening, conditioning, dewatering, dry-ng and reduction. Dewatering is a common process and can be

echanical or natural. Mechanical dewatering techniques includeacuum filters, belt filter presses, centrifuges and membrane filterresses [1,2]. Natural dewatering is one of the oldest processes usedo treat sludge. The available methods include sludge lagoons, sandrying beds, wedgewater drying beds and dewatering via freezing1]. Sand drying beds have been the traditional method of sludgeewatering in small to moderate wastewater treatment plantsWWTPs), while the use of constructed wetlands (CW) for sludgeewatering has been rising over the last 15 years. The combinationf traditional sand drying beds with vertical flow constructed wet-ands led to the so-called Sludge Drying Reed Bed (SDRB) Systems,r Sludge Treatment Wetlands (STW).

Although vertical flow constructed wetlands have beenmployed for wastewater treatment for several years, the sys-

ems for sludge dewatering are far less numerous. It is only duringhe recent years that a growing interest emerged regarding SDRBshrough published scientific research [3–9]. Most of the full-scale

∗ Corresponding author. Tel.: +30 25410 79393/6974 993867;ax: +30 25410 79393.

E-mail addresses: astefan@env.duth.gr (A.I. Stefanakis), tsihrin@otenet.gr,sihrin@env.duth.gr (V.A. Tsihrintzis).

385-8947/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.cej.2011.05.111

© 2011 Elsevier B.V. All rights reserved.

experience on SDRBs comes from Denmark [5,10]. SDRBs are lessexpensive than mechanical methods and require far less power,since they rely on the force of gravity, solar radiation or biologi-cal processes as the source of energy for dewatering [8,11,12]. TheSDRB system improves a traditional gravel or sand drying bed byplanting reeds in the sand. The sludge can then be loaded continu-ously for up to 5–10 years, before emptying the beds. The primaryrole of the reeds is to create a rich microflora in the root zone to feedthe organic material in the sludge [1], while it possesses high evapo-transpiration capacity, significant tolerance of different water andsludge levels and drought, and a deep and extended root system[13,14]. The dewatering process results in increased dry mattercontent of the sludge residue, decrease of the sludge volume andorganic matter decomposition [5]. The final treated sludge (oftenreferred to as biosolids) is a useful product which can be furtherused as an organic fertiliser for land application [10].

Several works generally report that the main dewateringmechanisms are evapotranspiration (ET) and draining [10,11,13];however, there is very limited published research on the quan-tification of the evapotranspiration rate and estimation of its roleon the performance of a SDRB. The term evapotranspiration refersto the total amount of water that is converted from the liquid tothe gas phase, and includes the evaporation from a water and soilsurface and transpiration through emergent plants. ET depends onmany parameters, such as the topography and geology of the site,

species and phase of plant growth, climatic conditions (solar radi-ation, temperature, relative humidity, wind speed etc.) and totalprecipitation depth. ET constitutes the main water loss mechanismin wetlands [15]. However, measuring ET in wetlands is a difficult

al Eng

tt[e[

cbsTHq

aivioiteMmrr

2

2

aatvw

hufb(tTtulttdcep

1ifSaWactt

A.I. Stefanakis, V.A. Tsihrintzis / Chemic

ask, due to the variety of shape, surface cover (e.g., water, vegeta-ion, soil, litter), air and water advection, hydrology and topography16]. Therefore, ET still remains one of the most challenging param-ters to quantify, and generally the process is not fully understood17].

Particularly the transition from a pilot to a full-scale CW appli-ation might be more complex than expected in evaluating ET,ecause of edge effects which may lead to increased ET rates inmall-scale units, as described in detail by Kadlec and Wallace [15].hat is to say, wetland size is a parameter that could affect ET.owever, possible effect of wetland size on ET values has not beenuantified yet.

The aim of this study is to quantify ET, other dewatering mech-nisms and the water budget in experimental pilot-scale SDRBs,n an effort to evaluate, through parallel experiments, the effect ofarious design and operational parameters, i.e., the sludge load-ng rate (SLR), vegetation, porous media size and origin, existencef aeration tubes, increased chromium content and meteorolog-cal parameters, on these processes. To the best knowledge ofhe authors, studies on SDRB systems are limited, particularly invaluating the dewatering mechanisms under the specific Northediterranean setting. Therefore, the extrapolation of the experi-ental results of this study to full-scale size can provide a well and

eliable guideline for real SDRBs, since most of these systems areelatively small in size, at least compared to natural wetlands.

. Materials and methods

.1. Pilot-scale unit description

Eleven pilot-scale SDRB units have been constructed and oper-ted in an open-air laboratory (41◦08′47′′N, 24◦55′09′′E), forming

large experiment on the performance optimization of these sys-ems. The units are labelled as S1 to S11. Table 1 summarizes thearious operational and construction characteristics of the units,hile Fig. 1a–d presents different views of the units.

Each unit is a plastic cylindrical tank of diameter 0.82 m andeight 1.5 m (surface 0.57 m2). The substrate thickness in allnits was 40 cm. Ten of the pilot-scale units (S1–S10) includedrom bottom to top a 10-cm thick drainage layer made of cob-les (D50 = 90 mm), and two layers, 15-cm thick each, of mediumD50 = 24.4 mm) and fine gravel (D50 = 6 mm). One unit (S11) con-ained an extended cobbles layer (25 cm thick) and no fine gravel.he materials used were igneous rock obtained from a river bed inhe area and carbonate rock from a quarry. Two plant species weresed, i.e., common reeds (Phragmites australis) and cattails (Typha

atifolia), while one unit was kept unplanted. Ten of the units con-ained aeration tubes. The pilot-scale units received sludge underhree different SLRs in terms of dry matter (dm): 30, 60 and 75 kgm/m2/yr. In three pilot-scale units, which received different SLR,hromium was added. This setup allowed for the evaluation of theffect of various design and operational parameters on SDRB systemerformance.

The units were constructed and planted at an initial density of4 shoots/m2 by transplanting plants from nearby watercources

n early June 2007; they were then filled with tap water, and leftor the plants to grow and adjust to the new environment. Sinceeptember 2007, activated sludge was introduced into the unitsnd experiments started. Activated sludge was produced at theWTP of the municipality of Komotini, Rhodope Province, Greece,

nd was transported to the laboratory at the start of each loadingycle. This WWTP is an extended aeration system with long reten-ion time in the aeration basin. Activated sludge was introduced tohe units in loading cycles: each cycle included one week of loading

ineering Journal 172 (2011) 430– 443 431

at daily equal portions, followed by a resting period of 1 or 3 weeks,depending on the season (longer in winter, shorter in summer).

2.2. Evapotranspiration estimation

For ET quantification, all flows in the pilot-scale units wererecorded on a daily basis for one full year (October 2008–October2009) after the first year of operation, including the sludge loadsand the water volume that drains out from each unit base. Fig. 1epresents a schematic layout of a vertical section of a pilot-scale unit,indicating all water inflows and outflows. The ET rate per monthwas calculated according to the water budget in the unit, includingthe water volume in the sludge layer and in the meso-porous medialayer, as follows:{

P + Qws − Es − Qds = �V1 (1)Qds − T − Qd = �V2 (2)

}→ ET = Es + T = P + Qws − Qd

− �V1 − �V2 (3)

where ET is the estimated evapotranspiration volume (L), Qws is thewater volume contained in the activated sludge (L), Qds is the watervolume that drains to the underneath layer of the meso-porousmedia (L), Qd is the water volume that drains out of the unit base(L), Es is the evaporation volume from the sludge layer surface (L),T is the plant transpiration volume from the meso-porous medialayer (L), P is the precipitation volume (L), V1 is the water volumein the residual sludge at the end of each resting period before addingthe new sludge load (L), V2 is the water volume in the meso-porousmedia layer (L), and � indicates difference.

Qws was calculated indirectly, by measuring the total solids con-tent (% TS) in the activated sludge. The drained water (Qd) wasmeasured by collecting it from each unit base. Precipitation (P) wasmeasured with an automatic meteorological station, located adja-cent to the SDRBs. The water volume in the sludge layer (V1) wascalculated by combining the measurement of the TS content in theresidual sludge and the thickness of the sludge layer. After mea-suring the volume of each parameter, the ET value was estimatedas a vertical depth (mm), knowing that the sludge layer surface is0.57 m2. The measurement of the net pan evaporation (Ep) wasalso made on a daily basis using a standard Class A evaporationpan, which was placed adjacent to the pilot-scale SDRB units. Panevaporation is used as an indication of the atmospheric evaporativepower in the project geographical location.

Samples from the residual sludge were collected at the end of theresting period of every cycle, using a core sampler. Samples wereanalyzed in the laboratory for total solids (dewatering efficiency),among other quality constituents. Plants from every planted unitwere harvested in January 2009 when dried, and the produced dryplant biomass (above ground part) was measured, after furtherdrying at room temperature for several days.

Meteorological data (air temperature, atmospheric pressure, airhumidity, wind velocity and direction, and precipitation depth)were recorded on site, at a 5-min time interval, using an ELEMM900/950 meteorological station.

2.3. Statistical analyses

A one-way between groups analysis of variance (ANOVA) wasconducted to investigate the effect of various construction andoperation parameters on ET values. Post Hoc Multiple Compar-isons were also performed to test equal variations, using Tukeyhonestly significant difference test. The ANOVA statistical analyses

were performed using SPSS Statistics 17.0 for Windows. The differ-ences were statistically significant when p < 0.05. For the estimationof the effect of different meteorological and operation parameterson SDRB unit dewatering processes (i.e., evapotranspiration and

432 A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443

Table 1Construction and operational characteristics of the pilot-scale SDRB units for sludge dewatering.

Unit Meso-porous media Plant type Aeration tubes SLR (kg dm/m2/yr)

Origin Size

S1 River bed (igneous) Fine-grained Reed Yes 75S2 Quarry (carbonate) Fine-grained Reed Yes 75S3 River bed Fine-grained Cattaila Yes 75S4 River bed Fine-grained Unplanted Yes 75S5 River bed Fine-grained Reed Yes 30S6 River bed Fine-grained Reed Yes 60S7 River bed Fine-grained Reed Yes 75b

S8 River bed Fine-grained Reed Yes 30b

S9 River bed Fine-grained Reed Yes 60b

S10 River bed Fine-grained Reeda No 75

dM

3

3

trytor

burvSI(tau(tciSdStm

rsS(S2Staaicc

S11 River bed Coarse-grained

a Initially planted, then plants dried at the end of Summer 2008.b Chromium added.

raining), the Pearson correlation coefficient was calculated usingicrosoft Excel.

. Results and discussion

.1. Dewatering efficiency

Before calculating the rate of ET in the SDRBs, it is necessaryo present the performance of these systems. Table 2 presents theesults of sludge volume reduction in the SDRB units, after twoears of operation. By subtracting the residual sludge volume fromhe total sludge volume applied after 29 loading cycles (two yearsf operation), one can obtain the amount (percentage) of wateremoved from the total sludge volume.

The dewatering process in the pilot-scale SDRB units proved toe quite effective, since the sludge volume reduction in all plantednits exceeded 90%. In particular, sludge reduction in planted unitseceiving low SLR (30 kg dm/m2/yr; units S5, S8) reached a meanalue of 93.0%, in units with medium SLR (60 kg dm/m2/yr; units6, S9) 94.6%, and in units with high SLR (75 kg dm/m2/yr) 94.9%.t has to be mentioned that plants in units S3 (cattails) and S10reeds; no aeration tubes), both receiving high SLR, did not manageo adjust to the specific environment of these two pilot-scale unitsnd died after summer 2008 operation. Therefore, after plant death,nits S3 and S10 continued to operate as unplanted control unitsreplicates), and sludge volume reduction became rapidly similaro the originally unplanted unit S4 (Table 2). The presence of plantslearly enhanced and accelerated the dewatering process, increas-ng the volume reduction by nearly an additional 10% (95.5% in unit1 and 86.7% in the unit S4; Table 2). This is also expressed by theepth of the residual sludge layer in the three unplanted units (S3,4 and S10 – mean 39 cm; Table 2), which was higher comparedo the residual sludge layer depth in the planted units with high,

edium and low SLR (mean 18.7, 13.4 and 8.6 cm, respectively).An indicator of dewatering efficiency is the TS content in the

esidual sludge. Table 2 contains mean TS values in the activatedludge in all SDRB units. The TS content increased in all pilot-scaleDRB units, compared to the mean TS content in the raw sludge3.19%). At a first glance, the TS content seems to be affected by theLR that each planted unit received; units with high SLR showed2.0 ± 13.4% mean TS, with medium SLR 24.2 ± 13.5% and with lowLR 30.5 ± 19.1%. On the other hand, the unplanted unit (S4) andhe other two units with the dead plants (S3, S10), that also oper-ted as control units, achieved 16.7 ± 13.1% mean TS. These results

lso imply that the presence of the reeds enhances the dewater-ng process mainly through transpiration, resulting in higher TSontent in the residual sludge [9]. It is also interesting that the TSontent in units with high SLR is lower compared to units with low

Reed Yes 75

SLR, although sludge volume reduction in units with high SLR ishigher. The main difference is that in units receiving low SLR, thesludge layer was much thinner compared to units with high SLR.Moreover, all water volume was drained within a few days afterfeeding in the low SLR units. Therefore, almost the entire sludgelayer depth remained exposed (after water removal) to the atmo-sphere for more days during the resting periods, achieving higherfinal solids content, which is also indicated by the higher standarddeviation of the mean value for units with low SLR. On the otherhand, only the upper part of the thicker sludge layer was exposedto the atmosphere in units with high SLR.

3.2. Cumulative evapotranspiration

In ET estimations based on Eq. (3), two observations were made:(a) �V2 ≈ 0: as observed, most water volume for ET was taken fromthe sludge layer, therefore, the water content in the meso-porousmedia layer remains practically steady and close to zero at the endof each loading cycle, due to immediate drainage and transpiration;and (b) a linear water content decline in the sludge layer: in orderto simplify the calculations, and without adding any error, it wasassumed that the decrease of the water content from the initialsaturation point, after each sludge load, proceeds linearly (dailyequal distribution).

Table 3 contains the cumulative ET values for the entire moni-toring period (October 2008–October 2009), while Fig. 2 presentsgraphically the cumulative ET and Ep values for various pilot-scaleunits. Results clearly indicate that there are differentiations in theET process in the SDRB units, depending on the various character-istics of the units.

3.2.1. Effect of sludge loading rate (SLR)Concerning the three SLRs used (low, medium and high; 30,

60 and 75 kg dm/m2/yr, respectively), it seems that the higherthe amount of sludge a unit receives, the higher the ET valuesare. The higher cumulative ET value was found in unit S7 (highSLR; 2481.6 mm/yr) and the smaller value in unit S8 (low SLR;937.6 mm/yr). The mean ET value for all the planted units thatreceive high SLR (S1, S2, S7 and S11) was 2371.1 mm/yr. Then fol-low the units receiving the medium SLR (S6 and S9) with a meanvalue of 1846.0 mm/yr, and last the units with the low SLR (S5and S8) with 1139.3 mm/yr. Thus, for double SLR from 30 to 60 kgdm/m2/yr, the cumulative ET increased on the average by about706.7 mm/yr or 62.0%. For SLR rising up to 75 kg dm/m2/yr (2.5

times up), the cumulative mean ET was 1231.8 mm/yr or 108.0%higher. For a 25% increase of the SLR (from 60 to 75 kg dm/m2/yr),the mean additional cumulative ET value was 525.1 mm/yr or 28.5%higher.

A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443 433

F nd ves

h(tc(uIth

e

ig. 1. Views of the meso-porous media layers: (a) carbonate material; (b) cobbles aection of a pilot-scale SDRB unit indicating water inflows and outflows.

It is also noticeable that the mean ET value for the units withigh SLR (2371.1 mm/yr) is comparable with the respective value2758.0 mm/yr) obtained in another study on pilot-scale Horizon-al Subsurface Flow (HSF) Constructed Wetland units planted withommon reeds and operating as lysimeters under full saturationunit CO-R) [18], while Borin et al. [19] report even higher ET val-es (3899 mm/yr) for vegetated subsurface flow beds in southern

taly (Sicily). The increased rates in HSF CWs (full saturation) imply

hat the maximum SLR for effective ET should be close to or evenigher than 75 kg dm/m2/yr.

Fig. 2a presents progressive cumulative ET (mm) curves for thentire monitoring period for three pilot-scale SDRB units, with the

ntilated aeration tubes; (c) igneous material; (d) fully grown reeds; and (e) vertical

same characteristics but with different SLRs, and for pan evapora-tion Ep. The slope of each cumulative ET curve represents the ET rate(mm/d). The higher values of the ET rate appeared for the plantedunits receiving the high SLR (4.9–6.0 mm/d), followed by the unitswith medium (3.7–4.8 mm/d) and low SLR (2.0–3.0 mm/d). Theseresults again show that as the sludge load increases, so does thewater volume that the unit receives, because more water is avail-able for evaporation and transpiration. Moreover, statistical tests

[see Supplemental online material (SM); Table SM-1] show thatthe SLR affects the ET process. Specifically, statistical significancewas found among units S1, S5 and S6, which received the high, lowand medium level of SLR (p < 0.05; Table SM-1). Additionally, Post

434 A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443

Table 2Sludge volume reduction, mean TS content, average depth of sludge layer and biomass in the SDRB units for the first two years of operation.

Unit Total volume of appliedsludge (L)

Residual sludgevolume (L)

Sludge volumereduction (%)

TS (%) Average depth of residualsludge layera (cm)

Average plant biomass (start-endof working period) (g/m2)

IN 3.19S1 2411.7 68.4 95.5 23.5 16.3 929.9S2 2411.7 65.6 94.3 25.0 16.0 772.8S3 2411.7 193.8 86.8 17.1 38.6 –S4 2411.7 199.5 86.7 17.2 40.8 –S5 1005.3 22.8 93.8 31.7 7.6 529.0S6 1942.0 65.6 95.0 26.5 13.0 570.2S7 2411.7 88.4 94.1 16.5 25.4 876.3S8 1005.3 25.7 92.1 24.4 9.7 299.1S9 1942.0 57.0 94.1 22.1 13.8 606.2S10 2411.7 193.8 87.7 15.9 37.6 –

22.5

HfauS

uva

F(a

S11 2411.7 94.1 95.8

a For the ET monitoring period (October 2008–October 2009).

oc Multiple Comparison Tests (Table SM-1) revealed that the dif-erence in ET values between the units with high and low SLR (S1nd S5) is statistically significant. As the sludge loads between twonits approach each other (i.e., between units S5 and S6 or S1 and6) the differences are not statistically significant.

Compared to the cumulative Ep (1479.5 mm/yr; Table 3), allnits that received high and medium SLR had higher cumulative ETalues. The cumulative Ep of the operation period is equal to 59.6%nd 80.2% of the mean cumulative ET value for the units receiving

ig. 2. Cumulative evapotranspiration (ET) values for the entire monitoring period in: (a) tEp); (b) the unplanted units (S3, S4 and S10) and the planted S1 unit; (c) the units with dnd S2); (e) the units with (S1, S4) and without (S10) aeration tubes; and (f) the units rec

17.0 802.6

the high and the medium SLR, respectively. These values are com-parable to those given by Allen et al. [20], who suggested that ETfor a reed swamp in a temperate climate could be between 1 and 2times the Ep.

Statistical analysis for comparison between Ep and ET values

of each unit showed that statistical significance exists for mostof the units (p < 0.05; Table SM-1), confirming the measured datathat showed higher ET values compared to Ep. Moreover, thesedifferentiations are presented in Fig. 2a. Cumulative Ep values are

hree pilot-scale SDRB (S1, S5 and S6) units with different SLRs and pan evaporationifferent material size (S1 and S11); (d) the units with different material origin (S1

eiving additional Cr concentrations (S7, S8 and S9) and the S1 unit.

A.I. Stefanakis, V.A. Tsihrintzis / Chemical EngTa

ble

3W

ater

bud

get

in

the

SDR

B

un

its.

Un

it

Infl

uen

t

wat

er

in

feed

ing

slu

dge

(mm

)Pr

ecip

itat

ion

(P, m

m)

Init

ial w

ater

con

ten

t

insl

ud

ge

laye

r(m

m)

Tota

l in

flu

ent

wat

er

(mm

)D

rain

ed

wat

er

(Qd, m

m)

Cu

mu

lati

ve

ET

(mm

)

Fin

al

wat

erco

nte

nt

insl

ud

ge

laye

r(Q

ws,

mm

)

May

toO

ctob

erN

ovem

ber

to

Ap

ril

Tota

l

May

toO

ctob

erN

ovem

ber

to

Ap

ril

Tota

l

May

toO

ctob

erN

ovem

ber

to

Ap

ril

Tota

l

May

toO

ctob

erN

ovem

ber

to

Ap

ril

Tota

l

May

toO

ctob

erN

ovem

ber

to

Ap

rTo

tal

S1

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

116.

5

1571

.1

1330

.5

2901

.6

197.

2

240.

4

437.

5

1373

.9

1059

.6

2433

.5

30.5

S2

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

90.7

1545

.3

1330

.5

2875

.8

234.

4

199.

8

434.

2

1310

.8

1098

.5

2409

.3

32.3

S311

79.5

819.

7

1999

.1

275.

1

510.

9

785.

8

293.

3

1747

.9

1330

.5

3078

.4

768.

2

274.

0

1042

.3

979.

9

931.

5 19

11.2

124.

9S4

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

213.

9

1668

.4

1330

.5

2999

.0

742.

1

260.

5

1002

.6

926.

3

947.

5 18

73.8

122.

5S5

511.

1

288.

6

799.

7

275.

1

510.

9

785.

8

36.0

822.

1

799.

5

1621

.6

91.2

171.

6

262.

8

730.

8

610.

2

1341

.0

17.7

S6

996.

3

603.

0

1599

.3

275.

1

510.

9

785.

8

43.5

1314

.9

1113

.9

2428

.8

261.

1

117.

0

378.

1

1053

.8

973.

3

2027

.1

23.5

S7

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

108.

3

1562

.8

1330

.5

2893

.3

159.

8

216.

0

375.

8

1403

.1

1078

.5

2481

.6

36.0

S8

511.

1

288.

6

799.

7

275.

1

510.

9

785.

8

36.1

822.

3

799.

5

1621

.8

245.

1

415.

6

660.

7

577.

0 36

0.6

937.

6

23.3

S9

996.

3

603.

0

1599

.3

275.

1

510.

9

785.

8

45.4

1316

.8

1113

.9

2430

.7

300.

9

436.

7

737.

5

1016

.1

648.

7

1664

.8

28.4

S10

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

241.

2

1695

.8

1330

.5

3026

.3

735.

4

556.

8

1292

.3

960.

3

662.

3

1622

.6

111.

4S1

1

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

82.6

1537

.2

1330

.5

2867

.7

283.

3

390.

7

674.

0 12

53.8

906.

1

2159

.9

33.7

SLR

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n

Mea

n M

ean

Mea

n

Mea

n

Mea

nH

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

99.5

1554

.1

1330

.5

2884

.6

218.

7

261.

7

480.

4

1335

.4

1035

.7

2371

.1

33.1

M

996.

3

603.

0

1599

.3

275.

1

510.

9

785.

8

44.5

1315

.9

1113

.9

2429

.7

280.

9

276.

8

557.

8

1034

.9

811.

0

1846

.0

26.0

L

511.

1

288.

6

799.

7

275.

1

510.

9

785.

8

36.1

822.

2

799.

5

1621

.7

168.

2

293.

6

461.

8

653.

9

485.

4

1139

.3

20.5

Un

pla

nte

d

1179

.5

819.

7

1999

.1

275.

1

510.

9

785.

8

249.

5

1704

.0

1330

.5

3034

.6

748.

6

363.

8 11

12.4

955.

5

847.

1

1802

.5

119.

6

E p12

32.4

247.

1

1479

.5P

275.

0

510.

8

785.

8

ineering Journal 172 (2011) 430– 443 435

throughout the year always below the respective ET values of thepilot-scale units, as also does the Ep rate (mm/d). A first observationwould be that the differences between the Ep and the ET for the twoSLRs should be attributed to transpiration by the reeds. However,Table 3 also shows that the three unplanted units (S3, S4 and S10)have a mean cumulative ET value of 1802.5 mm/yr (Table 3), whichis 21.8% higher than Ep. It is interesting that the same value (ETfrom unplanted beds equals to 120% of Ep) was also found by Nas-sar et al. [21]. These results indicate that the evaporation processis enhanced in the unplanted units, possibly due to the dark colourof the sludge layer surface (accumulated sludge) which increasesabsorption ability of the sun radiant heat. Moreover, it should benoticed that the sludge layer surface is not flat like an open watersurface, but harsh, thus increasing the available active surface forevaporation.

Concerning the units with the low SLR (S5 and S8), their cumu-lative ET values were both lower compared to the cumulative Ep

(Table 3). This is also obvious in Fig. 2a, where Ep values outreachrespective ET values of the unit receiving low SLR in August. A pos-sible explanation for this could be that the major portion of theavailable water volume in these two systems is rapidly lost withinthe first days of the resting period of each loading cycle. Since thetwo units receive the smallest sludge amounts, compared to theothers, the water volume seems to be smaller than the potentialtranspiration capacity of the reeds. This is also confirmed by thefact that the sludge layer in both units became rather quickly dry(within a few days), thus the water loss through evaporation waspractically limited to zero after 2–3 days. In addition to this, thealbedo of a wetland covered with reeds is greater compared to openwater [20], resulting in less energy flux driving evapotranspiration.

3.2.2. Effect of vegetationDuring the operation of the units, no litter presence was

observed, since the sludge layer surface was almost clear from with-ered leaves or stems, which were often removed manually. Thepresence of the plants seems to be crucial for effective system oper-ation. Fig. 2b presents cumulative ET values for the planted unit S1and the three unplanted units S3, S4 and S10, all receiving high SLR.As mentioned above, cattails in unit S3 dried during the first sum-mer of operation. Koottatep et al. [22] also reported dieback signsfor cattails at the beginning phase of a wetland system operatingunder tropical conditions; under high SLR, cattails were shockedin some beds, although in other beds cattails were used for manyyears.

It seems that vegetation plays a significant role in sludge dewa-tering. In fact, for the whole process to be functional, the use ofreeds is a key component. Since cattails dried in unit S3, their func-tion cannot be evaluated. Results presented in Fig. 2b show that thecumulative ET values in the unplanted units (which receive highSLR) are lower compared to those with plants. The presence of thereeds increases the annual ET by about 631.0 mm/yr or 35.0%, whichconfirms the positive role that plants play. This difference should beattributed to the high transpiration capacity of the common reed[1,13]. The reeds absorb the water content of the sludge throughtheir root system to support their growth, and subsequently, theyrelease it back to the atmosphere.

Another parameter that affects the system performance is plantbiomass. Table 2 also contains data of produced reed biomass(g/m2) for each unit during the study period. Units with high SLR(S1, S2, S7 and S11) have the highest produced biomass (mean845.4 g/m2), followed by the units with medium (S6 and S9;588.2 g/m2) and low SLR (S5 and S8; 414.0 g/m2). Fig. 3a presents

a chart of produced biomass in all units in relation to the ET anddrained water (DW) depth. The trendline of each group is used toshow more clearly the relation between the parameters. It is obvi-ous that as the produced biomass increases, so does the ET depth.

436 A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443

Fig. 3. (a) Plant biomass in each pilot-scale SDRB unit and its correlation to evapotranspiration (ET) and drained water (DW); (b) mean cumulative ET values for the unitsw erage( levels

Taouas

3

dSroc1wpu

ith the three different SLRs (H = high, M = medium, L = low), the unplanted unit avNovember to April) periods; and (c) monthly ET for the SDRB units with the three

his means that increased plant growth results in enhanced reedctivity, and so, increased ET values. Simultaneously, the amountf drained water decreases, since the greatest portion of water issed by the reeds. Therefore, it can be stated that the level of SLRffects significantly the processes taking place in the SDRB units,ince it defines the amount of available water.

.2.3. Effect of porous media size and origin (under same SLR)Fig. 2c contains the cumulative ET values for the two units with

ifferent material size: unit S1 with fine-grained material and unit11 with coarse-grained material, both of the same origin (igneousock, obtained from a river bed). It is clear that higher ET valuesbtained for unit S1 with the fine-grained material (2433.5 mm/yrompared to 2159.9 mm/yr for unit S11; Table 3), a difference of

1.3%. As the material in unit S11 is coarse-grained and the layerith the fine gravel is absent, the substrate possesses a higherorosity, thus larger pore volume, resulting in faster water vol-me draining to the unit base. Therefore, lower water volume

, and pan evaporation (Ep), for the entire, the warm (May to October) and the coldof SLR, monthly precipitation depth, monthly Ep and air temperature variations.

is available for evaporation and reed transpiration. However, nostatistically significant difference was found (p > 0.05; Table SM-1).

On the other hand, the material origin does not affect the ETvalue. As Fig. 2d shows, unit S1 with igneous material from ariver bed and unit S2 with the carbonate material from a quarryof the same size, have more or less the same cumulative ET val-ues (2433.5 mm/yr and 2409.3 mm/yr, respectively). The sameconclusion was confirmed by statistical analysis (p > 0.05, Table SM-1).

3.2.4. Effect of aeration tubesAnother parameter that is investigated is the presence of aera-

tion tubes. Unit S10 has no aeration tubes and, as mentioned above,

is unplanted. The absence of aeration tubes possibly favouredanaerobic conditions in the root zone, and thus, reeds dried dur-ing the first summer of operation. The unplanted unit S4 containedthese tubes. Fig. 2e shows that the cumulative value for unit S4

al Eng

i1baoadtt

3

wlfeai(s

abumaqttw(

atfcrStauTpcd

at[iaiarotftvTwoebat

A.I. Stefanakis, V.A. Tsihrintzis / Chemic

s about 13.5% higher compared to unit S10 (1873.8 mm/yr and622.6 mm/yr, respectively; Table 3). Additionally, the comparisonetween unit S10 and the planted unit S1 shows that the presence oferation tubes seems to induce an air current along the lower layersf the bed, which also assists in better reed growth under a well-erated environment. A cluster of robust reeds succeeds improvedrying rate through their enhanced evaporative activity. Therefore,he simultaneous presence of aeration tubes and reeds seems to behe optimum combination.

.2.5. Effect of chromium contentFig. 2f presents charts of cumulative ET values for the three units

hich received additional chromium (S7, S8 and S9 units with high,ow and medium SLR, respectively) and unit S1 (high SLR). All theseour units were planted with common reeds. Thus, the only differ-nce between units S1 and S7 is the addition of chromium in thectivated sludge feeding unit S7. The concentration of chromiumn the sludge added in these three units was about 6 times higher3.178 mg/g dm) than the original concentration in the activatedludge (0.500 mg/g dm).

It seems that the additional chromium concentration did notffect the operation of unit S7, since the cumulative ET values foroth units are comparable (2433.5 mm/yr and 2481.6 mm/yr fornits S1 and S7, respectively; Table 3). Taking this into account, theain parameter that affects ET among the three units that received

dditional chromium concentration (S7, S8 and S9) seems to be theuantity of loaded sludge (i.e., the SLR). This is also confirmed statis-ically, since statistical significance was found for ET values amonghese three units (p < 0.05; Table SM-1). Same statistical differencesere also found for the other units receiving the three levels of SLR

units S1, S5 and S6).Comparing the curves in Fig. 2a and f, it seems that the slopes

mong the three units in each chart are similar, indicating thathe level of SLR is the dominant parameter responsible for the dif-erences in the cumulative ET values. However, the comparison ofumulative ET values between the units with low and medium SLReveals some differences. Specifically, cumulative ET values of units5 and S8 are 1341.0 mm/yr and 937.6 mm/yr (Table 3), respec-ively, while the only difference between these two units is theddition of chromium in unit S8. The same also occurs betweennits S6 and S9 (2027.1 mm/yr and 1664.8 mm/yr, respectively;able 3). An explanation would attribute these differences to theresence of chromium in units S7 and S8. Although this is a diffi-ult hypothesis to prove, there is some published literature whicheals with this [23,28].

The chromium concentration and accumulation in the reedsnd their – healthy or not – appearance indicate their toleranceo this specific metal, as reported by Mant et al. [23]. These authors23] reported that P. australis, treating tannery wastewater richn Cr (10 mg/L), did not grow normally and never looked healthy,lthough it did not present the classical signs of chromium toxic-ty. In the present study, the chromium load for three units (S7, S8nd S9) was 24.8 mg/L compared to 1.4 mg/L in units S1, S5 and S6,espectively. The reeds in these units did not show any clear signsf toxicity shock (e.g., yellow leaves); however, it was observedhat the reed biomass in unit S8 was not as dense, as in unit S5;urthermore, reeds were shorter. Produced plant biomass confirmshis finding, since unit S8 with additional chromium showed loweralues (299.1 g/m2; Table 2), compared to the unit S5 (529.0 g/m2).his, however, was not true for units S7 and S9 where biomassesere comparable. Apparently, the chromium effects were more

r less compensated by the increased water availability, which

nhanced reed growth. This is implied by the comparable producediomass between units S1 and S7 (without and with chromiumddition; 929.9 and 876.3 g/m2, respectively), which both receivedhe high SLR, as also between S6 and S9, both with medium SLR

ineering Journal 172 (2011) 430– 443 437

(570.2 and 606.2 g/m2, respectively; Table 2). It appears that underlow sludge loadings, the simultaneous lack of water needed byplants for their growth and the presence of high Cr concentration,possibly limited the plant growth in unit S8. Therefore, Cr toxicity islikely to affect plant growth under low sludge loads and respectivelow available water volumes.

Other studies also report that accumulation of heavy metalscaused limited reed growth and reduced length and dry weightof reed roots and shoots [24–27]. The reed aggravation is also indi-rectly indicated by the limited dewatering efficiency of the unitswith chromium. Units S5 and S8 with low SLR had a mean TS con-tent of 31.7% and 24.4% (Table 2), respectively, units S6 and S9 withmedium SLR 26.5% and 22.1%, respectively, and units S1 and S7 withhigh SLR 23.5% and 16.5% TS content, respectively. Furthermore,Zazo et al. [28] reported that the ET rate and chromium concentra-tion of another species (T. latifolia) were reversibly connected; thehighest ET (under highest temperature and light intensity), whichimplies the highest plant activity, corresponded to the minimumchromium concentration in the diurnal cycle of the plant. Finally,Bragato et al. [29] reported a significant increase in chromium accu-mulation in P. australis during December (low temperatures anddecreased ET rate).

In summary, under low sludge loadings the reeds are possiblynot capable of managing water scarcity and chromium presenceat the same time. However, an increased SLR, which increases theavailable water volume for reed growth, seems to alleviate theeffects of chromium.

3.2.6. Effect of climatic conditionsEvapotranspiration is highly connected to the climatic condi-

tions of the site, such as temperature, solar radiation, precipitationdepth, wind speed, humidity and barometric pressure. The moni-toring period of the units covered a full annual temperature range.Temperature is expected to play a crucial role on ET variations. Themean temperature for the study period (one year) was 16.3 ◦C, withminimum and maximum values −2.5 ◦C and 38.5 ◦C, respectively.In order to estimate the effect of temperature on ET, the work-ing period was divided into two sub-periods based on the annualmean temperature value (about 16 ◦C); the first sub-period had amean temperature above 16 ◦C and the second below 16 ◦C. Otherstudies also report that an important similar temperature value isthat of about 15 ◦C, above which plant growth and microbial activ-ity begins [30–34]. Increased plant activity at temperatures above15–16 ◦C result in consequent increase of ET rates. The first sub-period (November to April) includes mainly winter months witha mean temperature 9.5 ◦C, while the second sub-period (May toOctober) includes mainly summer months with a mean tempera-ture 22.1 ◦C.

Table 3 includes cumulative ET values for each unit for these twosub-periods. Fig. 3b presents the cumulative ET values for the wholeand for the two sub-periods for the units receiving high, mediumand low SLR, the unplanted units (with high SLR) and the cumula-tive pan evaporation (Ep). For the period from May to October, ETvalues are clearly higher compared to the period from Novemberto April. The mean ET value for the warmer period (May to October)for the planted units receiving high SLR (S1, S2, S7 and S11) is 28.9%higher compared to the colder period. Respective figures for unitswith medium (S6 and S9) and low SLR (S5 and S8) are 27.6% and34.7%. These results show that higher temperatures favour evapo-transpiration rate in the SDRB units. Concerning the warmer period,the mean ET in the units with high SLR is 108.4% of the Ep, while for

the other two SLRs (medium and low) respective figures are 84.0%and 53.1%. The lower numbers for the medium and low SLR can beattributed to the limited available water volume, compared to thehigh SLR. This also confirms that the units have the capability to

4 al Eng

tttasBdJmdfE

psucteuca

dmtsavwtttpaEbvaot

Aishuciwt2ttbflauihpwtpbc

38 A.I. Stefanakis, V.A. Tsihrintzis / Chemic

reat even the larger sludge load, reaching a mean ET value equalo 1335.4 mm. It should also be mentioned that months from Mayo October represent the period during which the reeds are grownnd reach their optimum activity, and therefore, reach higher tran-piration levels. Thus, these results imply a seasonal effect on ET.urgoon et al. [3] report an average ET for reed beds receiving 40 kgm/m2/yr of sludge of about 6.4 mm/d for the months May and

une (temperate climate). Results in the current study for the sameonths are comparable with the units receiving the low SLR (30 kg

m/m2/yr) which showed an average ET rate of 5.8 mm/d, whileor the units with the medium SLR (60 kg dm/m2/yr) the averageT rate was 9.1 mm/d.

Moreover, as Fig. 3b shows, the mean ET values for the two sub-eriods for the unplanted units with high SLR (S3, S4 and S10) arelightly different (955.4 mm and 847.1 mm; Table 3), since the val-es for the warmer period is only 12.8% higher compared to theolder period. This indicates that – in the absence of the plants –emperature variations have a smaller impact on the ET rate. Gen-rally, comparing the mean ET values of the planted and unplantednits for the warmer period, it seems that the presence of plantsontributes significantly to the evapotranspiration by increasing itbout 30% at high temperatures.

Fig. 3c presents mean monthly ET values, monthly precipitationepth, monthly Ep and temperature variations during the wholeonitoring period of the SDRB units. The highest ET values for all

he units appeared in early summer months (May and June), pos-ibly as a result of the plant growth and activity (transpiration),s also of the gradual temperature increase. A relatively high ETalue also occurred in January and February. January was the monthith the higher precipitation depth (182.8 mm; Fig. 3c). It seems

hat in winter significant water volume is added to the reed bedshrough precipitation. In fact, precipitation depth was double forhe colder period (November to April) and reached 510.8 mm com-ared to the warmer period (May to October; 275.0 mm). Therefore,ccording to Eq. (3), as precipitation depth increases, so does theT value, since a large portion of the added water volume is usedy the plants. This assumption is also boosted by the lowest Ep

alue appearing in January and February (Fig. 3c), and gener-lly in the winter months, which confirms that the root systemf the reeds remains active and continues to work even at lowemperatures.

An interesting behaviour of the units appeared in April andugust. These two months were characterized by an extended rest-

ng period (about 6 weeks) of the applied loading cycle. Resultshowed that ET values for all planted units were all significantlyigher in April. On the other hand, ET values of the unplantednits were higher in August. A comparison of the meteorologi-al data reveals that different parameters dominate the processn these two months. Precipitation depth in April was 25.6 mm,

hile in August almost no rain was detected (1.2 mm). Moreover,he mean temperature of April was only 14.1 ◦C and that of August6.9 ◦C, while Ep in August (385.6 mm) was 8 times the respec-ive value in April (48.2 mm). These observations indicate that, inhe absence of sludge load, evapotranspiration in summer is ruledy the high temperatures, and subsequently the high solar energyux, which favour intensive water evaporation, thus limiting thevailable water volume for plant transpiration. In the unplantednits, where evaporation from the surface plays the dominant role

n water losses, the higher ET values in August follow the respectiveigh Ep value of the same month. On the contrary, in April (tem-eratures below the mean annual value) the main inflow is rain,hich represents the available water volume for evapotranspira-

ion. Since the temperature values are still relatively low in April,an evaporation is limited. The reeds are in the growth phase andegin their activity, and the beds remain wet for a larger periodompared to August, when they dry within a few days.

ineering Journal 172 (2011) 430– 443

3.3. Draining

Draining (gravitational flow of water) is another dewateringmechanism that appears in SDRBs [5,10,13]. During the monitor-ing period of the pilot-scale units, the water volume that drained ineach unit base was collected and measured on a daily basis (param-eter Qd in Eq. (3); Fig. 1e). Therefore, a simultaneous estimation ofthe drained water was made, in order to derive a final estimation ofthe contribution of each mechanism (evapotranspiration and drain-ing) in the water balance of the SDRB units. Table 3 also containsthe total water depth that drained in each unit during the moni-toring period and respective values for the periods May to Octoberand November to April, as well as mean values for each SLR andthe unplanted units. Additionally, the total water depth containedin sludge loaded during the monitoring period is also included inTable 3. Finally, the mean daily draining rate (Dr, mm/d) of eachunit has been calculated using the known drained water volume(Table 4).

3.3.1. Effect of sludge loading rateResults showed that the units receiving the high and medium

SLR had the highest drained water volumes. Thus, the plantedunits receiving 1999.1 mm (Table 3; high SLR) of water contentin sludge gave a mean value of drained water height of 480.4 mm(Table 3), with respective values for the medium SLR 1599.3 mmand 557.8 mm. Units S6 and S9, which were loaded with themedium SLR, had a significant difference in the drained water(378.1 mm and 737.5 mm, respectively). The only operational alter-ation between these two units was the addition of extra chromiumin unit S9, which might have affected the reed activity in this unit,as mentioned above. In addition, lower ET values (Table 3) wereobserved in unit S9, compared to unit S6. The larger drained watervolume in unit S9 is also explained by the high mean draining rateof this unit (Dr = 2.07 mm/d; Table 4), which is almost double com-pared to that of unit S6 (mean Dr = 1.11 mm/d). Moreover, the unitswith the low SLR (S5 and S8), which received 799.7 mm of waterthrough sludge loads (Table 3), had drained water volumes 262.8and 660.7 mm. As before, there is a difference in the draining ratesbetween these units (Dr = 0.87 and 1.80 mm/month for units S5 andS8, respectively; Table 4), which agrees with the larger drainedwater volume in unit S8. This same finding (i.e., different drain-ing rates) between units S6 and S9 (medium SLR) and units S5 andS8 (low SLR) offers another indication, that the chromium addi-tion may have affected plant growth in units S8 and S9, resultingin decreased transpiration and increased drained water volume.

Generally, it seems that the drained water volume depends onthe sludge load each unit receives and not on the amount of accu-mulated dewatered sludge, as also reported elsewhere [13,35]. Thesmall differences in drained water between the units with low andhigh SLR, coupled with the simultaneous high ET differences, implythat as the added water volume increases and remains availablefor longer time, the process of ET prevails over water drainage. Thisalso explains the fact that the units with medium SLR had the high-est drained water volume. Moreover, in the units with high sludgeloads the sludge layer is thicker compared to the units with lowerloads, which hinders the water drainage to the unit base. This is alsoconfirmed by the fact that the drained water volume in the unitswith low SLR is about 58% (461.8 mm out of 799.7 mm; Table 3) ofthe water added with sludge loads, while for the units with higherSLR this percentage decreases (24% and 35% for the units with highand medium SLR, respectively).

3.3.2. Effect of vegetationThe unplanted units (S3, S4 and S10), which also received

the high SLR, showed higher drained water amount (mean value1112.4 mm; Table 3), compared to the planted units with the same

A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443 439

Table 4Mean draining rates (Dr , mm/d) for each SDRB unit for the three levels of sludge loads and unplanted units for the entire monitoring period and the two sub-periods (Mayto October and November to April).

Month Unit

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11

O-08 1.23 2.23 1.28 1.73 0.00 0.29 0.82 0.53 1.00 2.33 0.04N-o8 2.40 1.42 0.98 0.68 0.10 0.30 0.24 0.64 0.91 2.11 0.06D-08 1.68 0.55 0.00 0.41 0.10 0.02 0.00 1.70 2.24 2.83 0.61J-09 1.98 1.25 0.99 1.84 0.00 0.66 0.57 4.83 4.96 4.96 6.75F09 3.95 3.62 4.08 4.54 3.60 3.20 3.84 3.35 5.01 5.73 4.52M-09 1.88 3.81 4.68 4.79 3.48 3.58 5.27 5.12 5.59 5.52 4.13A-09 0.28 0.24 0.79 0.72 0.20 0.31 0.32 0.24 0.28 1.11 0.23M-09 0.03 0.02 3.03 2.92 0.06 0.01 0.01 0.80 0.16 2.20 0.57J-09 1.88 2.35 4.72 5.52 0.99 1.47 1.84 2.50 3.03 3.20 3.13J-09 0.43 0.51 3.24 3.27 0.46 0.42 0.45 0.95 1.13 2.99 0.69A-09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.33 0.00S-09 1.70 1.83 2.06 1.34 0.53 1.12 0.44 0.82 0.52 2.11 1.26O-09 5.13 4.62 4.32 4.28 1.72 3.07 3.01 1.97 2.05 4.04 2.49

Aver 1.74 1.73 2.32 2.46 0.87 1.11 1.29 1.80 2.07 3.04 1.88May to October 1.49 1.65 2.66 2.72 0.54 0.91 0.94 1.08 1.13 2.46 1.17November to April 2.03 1.82 1.92 2.16 1.25 1.34 1.71 2.65 3.16 3.71 2.72

High SLR Medium SLR Low SLR UnplantedAver 1.66 1.59 1.33 2.61

SinuFwaodo1v

3

wTThueT

3

STarowhrria

3

wp

May to October 1.31 1.02

November to April 2.07 2.25

LR (480.4 mm). Since vertical flow of water is driven by grav-ty and plants uptake water, in the absence of plants, there areo water losses through transpiration. Therefore, more water vol-me is trapped in the bed and drains by gravity to the unit base.ig. 3a confirms that as the produced biomass increases, the drainedater volume decreases. Additionally, the unplanted units possess

higher porosity and hydraulic conductivity, due to the absencef the root system. This is also shown by the comparison of theraining rates of the planted and unplanted units. In the unplantednes, the mean draining rate is 54% higher (2.61 mm/d compared to.70 mm/d; Table 4), which also explains the larger drained waterolume in these units.

.3.3. Effect of aeration tubesAs mentioned earlier, unit S10 contains no aeration tubes and

as unplanted (reeds dried during the first summer operation).his unit showed the highest drained water volume (1292.3 mm;able 3) among all units of the experiment, and consequently, theighest draining rate (3.04 mm/d; Table 4). Compared to the othernplanted units (S3 and S4) with aeration tubes, the main differ-nce is detected during the winter months (November to April), asable 3 shows.

.3.4. Effect of material size and originAnother observation has to do with the substrate thickness. Unit

11 has only coarse-grained material, compared to the other units.his entails a higher hydraulic conductivity of the substrate mediand larger void space volume, resulting in more drained watereaching the unit base, which is confirmed by the high draining ratebserved in this unit (Dr = 1.88 mm/d; Table 4). The total drainedater depth for unit S11 was 674.0 mm (Table 3), which is 40%igher than the mean drained water value for the planted unitseceiving high SLR. Moreover, comparing the two units with mate-ials of different origin (units S1 and S2), the drained water amounts almost the same (437.5 mm and 434.2 mm, respectively; Table 3),s also their Dr (1.74 and 1.73 mm/d, respectively; Table 4).

.3.5. Effect of meteorological conditionsDraining is also affected by season. As Table 3 shows, drained

ater volume in most planted units is higher during the coldereriod (November to April). For the units receiving the high SLR, 55%

0.81 2.611.95 2.60

of the total water drains during this period (e.g., in unit S1 240.4 mmout of total 437.5 mm drain during the period from November toApril; Table 3). Respective values for the units with medium and lowSLR are 50% and 64%. This can be attributed to the fact that at lowtemperatures plant transpiration is limited, and therefore, morewater volume remains in the bed and drains to the unit base. Drain-ing rate calculations (Table 4) also confirm that more water drainsduring the winter months, since for the period November to Aprildraining rates are higher compared to the warmer period. Besidesthat, as mentioned before, the precipitation depth for this period(November to April) is nearly double (510.8 mm) compared to thewarmer period (May to October), which adds more water to thebeds. Reeds create an extensive root system within the substratemedia and the sludge layer, which is continuously growing, andthus, exerts a high suction pressure, so that the sludge/substratelayer always remains permeable for the water that drains in paral-lel with the roots [36]. In the unplanted units, these processes arenot present. Taking also into account that in winter months a largepart of the bed may have frozen temporarily, the lower drainedwater volume observed in the unplanted units can be explained.Draining rates of the unplanted units (S3, S4 and S10) are practi-cally not affected directly by the season. Table 3 also shows that thetotal drained water in unit S10 was the highest observed among allunits (1292.3 mm); similarly, the draining rate 3.04 mm/d was thehighest (Table 4). The main characteristic of this unit is not only theabsence of reeds, but also the absence of aeration tubes.

It is also interesting to estimate the duration drainage takesplace after each sludge load. Fig. 4 presents charts of drained water.Each line presents the percentage of total water that drains in eachloading cycle during the first eleven days after each sludge applica-tion in the SDRB units. Fig. 4a shows that most of the water drainswithin three days after the sludge application in the units receivinghigh SLR. The amount of drained water drops below 5% of the totaldrained water after 6 days. For unit S11 with the coarse-grainedmaterial, the drainage flow proceeded for another 2–3 days, sinceincreased water volume passes through the larger void space ofthe substrate material. The same also holds for the other units with

low (Fig. 4b) and medium (Fig. 4c) SLR. As explained above, the dif-ferences between units S5–S8 (low SLR) and S6–S9 (medium SLR)could be attributed to the chromium addition effect on plants. Itseems that in the units with low and medium SLR the draining flow

440 A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443

F ach s( the cou old an

pochtf

hufbut

ig. 4. Mean percentage values of water drained during the first eleven days after ec) planted units with medium SLR; (d) unplanted units; (e) units with high SLR for

nits with low SLR for the cold and warm periods; and (g) unplanted units for the c

ractically stops after 6–7 days. The maximum drained water wasbserved for all units during the second day after the sludge appli-ation. The comparison of all units indicates that in the units withigh SLR, the unit body remains wet for more days, which explainshe higher ET values, since water is available for evapotranspirationor a longer period.

In the unplanted units (Fig. 4d), the drained water reachesigh values after 2–3 days and then begins to drop more grad-ally compared to the planted units. The drained flow continues

or more than 10 days and only after the 11th day it dropselow 5% of the total water volume, contrariwise to the plantednits. This confirms the positive role of the plants in the sys-em operation, since a major portion of added water is consumed

ludge load for the: (a) planted units with high SLR; (b) planted units with low SLR;ld and warm periods; (f) units with medium SLR for the cold and warm periods; (g)d warm periods.

by the reeds through transpiration and for growth needs. More-over, Fig. 4e–h presents the progress of water drainage during thetwo sub-periods: from May to October (warm period) and fromNovember to April (cold period). Although most of the water drainswithin the first 2–3 days after each sludge load for both periods,the unit body remains wet for 2–3 more days during the coldperiod and continues to drain for 8–9 days till dropping belowthe 5% of the total drained water volume (Figs. 4e–g). During thewarm period, the water volume that drains to the unit base drops

rapidly after the third day. On the other hand, in the unplantedunits there are no significant alterations between the two peri-ods and the units continue to drain water for more than 10 days(Fig. 4h).

al Eng

3

iTiaatttcfia(

ictwari(drcimmarrisAga

EwoaF

gtNloo4Riwt

btEFc

was

A.I. Stefanakis, V.A. Tsihrintzis / Chemic

.4. Water budget

In order to estimate the role of the dewatering processes appear-ng in the SDRB units, the water budget was calculated for each unit.able 3 contains all depths of influent water and water losses. Asnfluent water depths were considered the water content in thepplied sludge of each loading cycle, the precipitation water (P),nd the water content in the sludge layer just before the start ofhe monitoring period. All the influent water depths were addedo present the total influent water depth. Water losses appearedhrough evapotranspiration and drained water, as also the waterontent in the sludge layer at the end of the monitoring period. Allgures are presented as total value (entire monitoring period), aslso for the two sub-periods according to the temperature rangesi.e., above and below 16 ◦C).

As Table 3 presents, the major part of influent water (64%) isntroduced to the units with sludge loads and the rest through pre-ipitation. Results are also graphically presented in Fig. 5, wherehe contribution of each process (evapotranspiration, draining, finalater content in the sludge layer) in the water loss is accounted as

percentage of the total influent water volume. This chart clearlyeveals the major role that ET possesses in the dewatering processn the SDRB units. In the units receiving the high SLR, about 82.2%Table 3) of the influent water is lost through ET and 16.7% throughrainage. As the sludge loading rate falls, so does the percentageelated to ET, while the drained water percentage increases. Thisonfirms – as mentioned above – that the ET capability of the reedsncreases with the sludge load, implying the high treatment perfor-

ance these systems can achieve. As the sludge load drops to theedium and low SLR, the ET percentage drops about 6% every time

nd reaches 76.0% and 70.3% (Table 3) for the medium and low SLR,espectively. On the other hand, drained water volumes increaseespectively, as the sludge load decreases. This means that theres a finite water volume that can pass through the substrate pores,ince the differences in drained water are not so high (Table 3).mong the planted units with high SLR, unit S11 with the coarse-rained material showed the lowest ET percentage (75.3%; Fig. 5)nd the highest drained water percentage (23.5%).

The role of ET is limited in the unplanted units. As Fig. 5 shows,T accounts only for 59.4% (Fig. 5) of the total water losses, whileater drainage through the wetland body is responsible for 36.7%

f water losses. It is also noticeable that in unit S10 with no plantsnd no aeration tubes, the ET process had the lowest effect (53.6%;ig. 5) among all pilot-scale units.

The reported ET effect in SDRB units in the published literatureives a mixed picture. Begg et al. [4] estimated the ET contribution tohe water budget of a full-scale Sludge Drying Reed Bed (495 m2) inew England, USA, planted with Phragmites, to about 63.7% of water

osses, plus 15.4% of evaporation during the winter months, a totalf 79.1% after 6 years of operation. This value is close to the resultsf the current study. Begg et al. [4] measured a water loss of 39 and.7 mm/month during the growing season and winter, respectively.espective values for the present study are 29 and 22 mm/month,

ndicating that the differences in seasonal variations (i.e., heavierinter in England compared to Greece) affect the reed activity and,

hus, the evapotranspiration process.De Maeseneer [13], for a planted system (drained sand/gravel

ed) loaded with 20–30 kg dw/m2/yr in Germany, reported that, ofhe total water input, 46% leaves the system by drainage, 39% byT and 15% remains in the residual sludge. In another system inrance, the reported ET, under comparable sludge loads, was lowerompared to the results of the present study [13].

Heinss and Koottatep [36], for a system in central Europe loadedith 120 kg TS/m2/yr, report comparable percentages between ET

nd drainage (50–50%) and an average ET of 11 mm/d. For anotherystem loaded with 75 kg TS/m2/yr, the given average ET was

ineering Journal 172 (2011) 430– 443 441

12–20 mm/d. In a pilot-scale system in Bangkok, loaded with 250 kgTS/m2/yr and planted with cattails, Koottatep et al. [37] reported awater budget of 45% draining, 50% evapotranspiration and 5% waterin dried sludge. These results differ from those of the present study,but the different climatic conditions (e.g., increased rainfall andhumidity) and the lower TS content (1–2%) in the loaded sludgehave to be considered. However, all these different SDRB systemsshow a satisfying result in sludge dewatering, proving that this spe-cific technology is reliable and applicable in many different regionsof the world; furthermore, the two processes (evapotranspirationand drainage) dominate the dewatering process, with evapotran-spiration holding the key role in the temperate Mediterraneanclimate.

3.5. Statistical tests

In order to correlate measured evapotranspiration from theSDRB units with meteorological and operational data, the Pearsoncorrelation coefficient was used. The measured meteorological dataincluded: air temperature (Ta, ◦C), soil temperature (Ts, ◦C), rela-tive humidity (RH, %), wind speed (U2, m/s), solar radiation (Rs,W/m2), and precipitation depth (P, mm/d). ET (mm/d) was also cor-related with the drained water temperature (Td, ◦C), the measuredpan evaporation (Ep, mm/d), the drained water rate (Dw, mm/d), asalso with the total influent water (TIW, mm/d) and the season. Allcomparisons are made considering the mean parameter value foreach month. Table SM-2 presents Pearson correlation coefficientvalues for evapotranspiration with various parameters.

The parameter with the highest correlation with ET, among allmeteorological parameters tested in the planted units, was the tem-perature of the drained water. Unit S1 presented the highest r value(r = 0.61; Table SM-2) and unit S5 the lowest (r = 0.35) among theplanted units. In all cases, the correlation with the water temper-ature is strong for all planted units (mean r = 0.45), while for theunits receiving the high SLR, the r value raises up to 0.54. TheTotal Inflow Water also has a strong correlation with ET for allplanted units (mean r = 0.48; Table SM-2), while the correlation isenhanced as the SLR increases. Additionally, ET values appear tobe well correlated with solar radiation (mean r = 0.41; Table SM-2), while another good correlation appears between ET and season(mean r = 0.29; Table SM-2) for planted units.

These results confirm the expected rational hypothesis that theamount of water a unit receives plays a significant role in the pro-cesses appearing in the SDRB units, and thus, in the dewateringperformance of the unit. Moreover, comparing the ET correlationswith water temperature, solar radiation and TIW, they present adifferent behaviour relatively to the SLR level. This implies that, inthe units with the low and medium SLR, the need for water dom-inates over the temperature effects, while in the units receivingthe high SLR, the greater available water volume allows for ET toappear as an effect of temperature and solar radiation. This is alsoan indication that almost all planted units, and especially thosewith low and medium SLR, possess a greater evaporative potentialand could possibly handle effectively higher loadings. This probablymeans that the lower the sludge volume addition is, the fastest thedewatering and drying process is, and therefore, more controlled bytemperature. These remarks are confirmed by the r values appear-ing for precipitation. It is noticeable that the correlations betweenET and precipitation for the units receiving the low and mediumSLR (r = 0.36 and 0.32, respectively; Table SM-2) is higher com-pared to the units with high SLR (r = 0.16), which has to do with

the additional water volume entering the system. All these enhancethe aspect obtained from the experimental data that the dewater-ing process through evapotranspiration is a year-round process,although significantly enhanced in the summer.

442 A.I. Stefanakis, V.A. Tsihrintzis / Chemical Engineering Journal 172 (2011) 430– 443

Water balance in SDRB units

0

10

20

30

40

50

60

70

80

90

100

S1 S2 S3 S4 S5 S6 S7 S8 S9S10

S11

H (SLR)

M (SLR)

L (SLR)

nplanted

Wat

er lo

sses

(%

)Final water content in sludge layer Evapotranspiration Percolation water

83.983.8

62.162.5

82.783.5

85.8

57.8

68.5

53.6

75.382.2

76.0

70.3

59.4

15.115.133.933.4 16.215.6 13.0

40.7

30.342.7

23.516.7 23.0 28.5

36.7

d con

Tluattap

blotwnfvtarawtratc

4

wfiliasattaaaf

Fig. 5. Water budget in the SDRB units an

Regarding the unplanted units, some differentiations occurred.he water temperature and solar radiation present a high corre-ation with ET in these systems, with mean r values for the threenplanted units 0.29 and 0.39, respectively (Table SM-2). Addition-lly, humidity showed a strong negative correlation with ET. Sincehe higher humidity values appeared in winter months, it meanshat ET in the unplanted units is mainly controlled by temperaturend radiation. Small correlations between ET and TIW (r = 0.13) andrecipitation (r = 0.22) confirm this remark.

Table SM-2 also presents Pearson correlation coefficientsetween drained water (mm/d) and various parameters (meteoro-

ogical and operational). Results show that drained water dependsn more parameters compared to ET. As before, a strong correla-ion was found with TIW and precipitation, especially for the unitsith low SLR (mean r = 0.70), which implies a possible deficiency ineeded water volume by the reeds. Pearson coefficient r was 0.95

or unit S8, 0.92 for S9 and 0.93 for unit S11 (Table SM-2). The meanalue for all planted units was 0.74. The same also hold for precipita-ion (mean value for planted units r = 0.66). A strong correlation waslso found for humidity and all the temperature expressions, with

values exceeding 0.40 and in some cases 0.50. The negative signgrees with the results presented above, according to which moreater drains as temperature decreases. For the unplanted units,

he most important parameters are also precipitation (mean value = 0.63; Table SM-2), the total inflow water (mean value r = 0.57)nd humidity (mean value r = 0.50). Temperature variations seemo have a greater effect on unit S10, which has no aeration tubes,ompared to the other two unplanted units.

. Summary and conclusions

The efficiency of eleven pilot-scale Sludge Drying Reed Bedsas tested and the main dewatering mechanisms were quanti-ed. Results showed that evapotranspiration is the main water

oss mechanism in SDRBs, and vertical drainage is the secondmportant process. The water loss processes mainly depend on thepplied sludge volume to be treated. Meteorological parameters,uch as temperature, precipitation depth and solar radiation, alsoffect the evapotranspiration process. SDRB design should considerhe following: (1) the presence of plants significantly enhanceshe dewatering process through increased evapotranspiration; (2)

fine-grained material layer enhances system performance; (3)eration tubes contribute to better substrate aeration, preventingnaerobic conditions; (4) the sludge loading should be intermittentollowed by a resting period length dependent on season. For North

U

tribution of each process in water losses.

Mediterranean countries, the recommendation is one week load-ing and three weeks resting in the winter, and one week loadingand one (for low SLR) or two (for higher SLR) weeks resting in thesummer. The increase in sludge loading frequency in the summer isimportant for plant survival; (5) for optimum ET and plant growth,the SLR could exceed the highest of 75 kg dm/m2/yr tested in thisstudy (i.e., reach 80–90 kg dm/m2/yr). Loading rates lower than60 kg dm/m2/yr could lead to plant water starvation; and (6) sludgecontaining high concentrations of heavy metals (only chromiumwas tested here), e.g., industrial sludge, can also be treated at highloading rates with minor effects on the plants and on the ET process;lower loading rates and respective limited available water volumesmight adversely affect plant growth.

Acknowledgement

The study was funded by the General Secretariat of Researchand Technology (GSRT) of Greece, as part of the project “IntegratedManagement of Sludge from Wastewater Treatment Facilities, andWastewater Treatment Using Natural Systems”, Operational Pro-gram of the Region of East Macedonia – Thrace, 2000–2006.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.cej.2011.05.111.

References

[1] B.A. Outwater, Reuse of Sludge and Minor Wastewater Residuals, CRC Press,Florida, 1994.

[2] G. Chen, L.P. Yue, S.A. Mujumdar, Sludge dewatering and drying, Dry. Technol.20 (4-5) (2002) 883–916.

[3] P.S. Burgoon, K.F. Kirkbride, M. Henderson, E. Landon, Reed beds for biosolidsdrying in the arid NW United States, Water Sci. Technol. 35 (5) (1997) 287–292.

[4] S.J. Begg, L.R. Lavigne, L.M.P. Veneman, Reed beds: constructed wetlands formunicipal wastewater treatment plant sludge dewatering, Water Sci. Technol.44 (11) (2001) 393–398.

[5] S. Nielsen, Sludge drying reed beds, Water Sci. Technol. 48 (5) (2003) 101–109.[6] A.I. Stefanakis, C.S. Akratos, P. Melidis, V.A. Tsihrintzis, Surplus activated sludge

dewatering in pilot-scale sludge drying reed beds, J. Hazard. Mater. 172 (2009)1122–1130.

[7] P. Melidis, G.D. Gikas, C.S. Akratos, V.A. Tsihrintzis, Dewatering of primarysettled urban sludge in a vertical flow wetland, Desalination 250 (1) (2010)395–398.

[8] E. Uggetti, I. Ferrer, E. Llorens, J. García, Sludge treatment wetlands: a reviewon the state of the art, Bioresour. Technol. 101 (2010) 2905–2912.

[9] V. Bianchi, E. Peruzzi, G. Msciandaro, B. Ceccanti, S. Mora Ravelo, R. Iannelli,Efficiency assessment of a reed bed pilot plant (Phragmites australis) for sludgestabilisation in Tuscany (Italy), Ecol. Eng. 37 (5) (2011) 779–785.

al Eng

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[Doulaye, M. Strauss, Material fluxes in constructed wetlands treating septage

A.I. Stefanakis, V.A. Tsihrintzis / Chemic

10] S. Nielsen, N. Willoughby, Sludge treatment and drying reed bed systems inDenmark, J. Water Environ. Manage. 19 (4) (2005) 296–305.

11] K.J. Edwards, R.K. Gray, J.D. Cooper, J.A. Biddlestone, N. Willoughby, Reed beddewatering of agricultural sludges and slurries, Water Sci. Technol. 44 (11-12)(2001) 551–558.

12] E. Uggetti, E. Llorens, A. Pedescoll, I. Ferrer, R. Castellnou, J. García, Sludge dewa-tering and stabilization in drying reed beds: characterization of three full-scalesystems in Catalonia, Spain, Bioresour. Technol. 100 (17) (2009) 3882–3890.

13] L.J. De Maeseneer, Constructed wetlands for sludge dewatering, Water Sci.Technol. 35 (5) (1997) 279–285.

14] L.B. Kim, D.E. Smith, Evaluation of sludge dewatering reed beds: a niche forsmall systems, Water Sci. Technol. 35 (6) (1997) 21–28.

15] R.H. Kadlec, S.D. Wallace, Treatment Wetlands, 2nd ed., CRC Press, Bocaraton,USA, 2009.

16] J.Z. Drexler, R.L. Snyder, D. Spano, U.K. Tha Paw, A review of models and microm-eteorological methods used to estimate wetland evapotranspiration, Hydrol.Process. 18 (2004) 2071–2101.

17] C. Souch, C.P. Wolfe, C.S.B. Grimmond, Wetland evaporation and energy parti-tioning: Indiana Dunes National Lakeshore, J. Hydrol. 184 (1996) 189–208.

18] V.A. Papaevangelou, G.D. Gikas, A.I. Stefanakis, V.A. Tsihrintzis, Estimation ofevapotranspiration in pilot-scale horizontal subsurface flow constructed wet-lands, in: Proceedings of the “Protection and Restoration of the EnvironmentX” International Conference, Corfu, Greece, 5–9 July, 2010.

19] M. Borin, M. Milani, M. Salvato, A. Toscano, Evaluation of Phragmites australis(Cav.) Trin. evapotranspiration in Northern and Southern Italy, Ecol. Eng. 37 (5)(2011) 721–728.

20] R.G. Allen, L.S. Pereira, D. Raes, M. Smith, Crop Evapotranspiration (Guidelinesfor computing crop water requirements), FAO Irrigation and Drainage PaperNo. 56, FAO, Rome, 1998.

21] M.A. Nassar, M. Smith, S. Afifi, Sludge dewatering using the reed bed system inthe Gaza Strip, Palestine, J. Water Environ. 20 (2006) 27–34.

22] T. Koottatep, N. Surinkul, C. Polprasert, A.S.M. Kamal, D. Koné, A. Montangero,U. Heinss, M. Strauss, Treatment of septage in constructed wetlands in tropicalclimate: lessons learnt after seven years of operation, Water Sci. Technol. 51(9) (2005) 119–126.

23] C. Mant, S. Costa, J. Williams, E. Tambourgi, Phytoremediation of chromium bymodel constructed wetland, Bioresour. Technol. 97 (2006) 1767–1772.

24] Z.H. Ye, M.H. Wong, A.J.M. Baker, A.J. Willis, Comparison of biomass and metaluptake between two populations of Phragmites australis grown in flooded anddry conditions, Ann. Bot. 82 (1998) 83–87.

25] Z.H. Ye, A.J.M. Baker, M.H. Wong, A.J. Willis, Copper tolerance, uptake and accu-mulation by Phragmites australis, Chemosphere 50 (2003) 795–800.

ineering Journal 172 (2011) 430– 443 443

26] N.A. Ali, M.P. Bernal, M. Ater, Tolerance and bioaccumulation of copper in Phrag-mites australis and Zea mays, Plant Soil 239 (2002) 103–111.

27] A.I. Engloner, Structure, growth dynamics and biomass of reed (Phragmitesaustralis) – a review, Flora 204 (2009) 331–346.

28] J.A. Zazo, J.S. Paul, P.R. Jaffe, Influence of plants on the reduction of hexavalentchromium in wetland sediments, Environ. Pollut. 156 (2008) 29–35.

29] C. Bragato, H. Brix, M. Malagoli, Accumulation of nutrients and heavy metalsin Phragmites australis (Cav.) Trin. ex Steudel and Bolboschoenus maritimus (L.)Palla in a constructed wetland of the Venice lagoon watershed, Environ. Pollut.144 (3) (2006) 967–975.

30] USEPA, Design manual for Constructed Wetlands and Aquatic Plant Systemsfor Municipal Wastewater Treatment, EPA/625/1-88/022, USEPA, Cincinnati,1998.

31] J. Vymazal, Nitrogen removal in constructed wetlands with horizontal sub-surface flow – can we determine the key process? in: J. Vymazal (Ed.), NutrientCycling and Retention in Natural and Constructed Wetlands, Backhuys Publish-ers, Leiden, 1999, pp. 1–17.

32] P. Kuschk, A. Wiebner, U. Kappelmeyer, E. Weibbrodt, M. Kastner, U. Stottmeis-ter, Annual cycle of nitrogen removal by a pilot-scale subsurface horizontalflow in a constructed wetland under moderate climate Water Res. 37 (2003)4236–4242.

33] C.S. Akratos, V.A. Tsihrintzis, Effect of temperature, HRT, vegetation and porousmedia on removal efficiency of pilot-scale horizontal subsurface flow con-structed wetlands, Ecol. Eng. 29 (2007) 173–191.

34] A.I. Stefanakis, V.A. Tsihrintzis, Performance of pilot-scale vertical flow con-structed wetlands treating simulated municipal wastewater: effect of variousparameters, Desalination 248 (2009) 753–770.

35] S. Nielsen, Biological sludge drying in constructed wetlands, in: G.A. Moshiri(Ed.), Constructed Wetlands for Water Quality Improvement, Lewis Publishers,1993, pp. 549–558.

36] U. Heinss, T. Koottatep, Use of Reed Beds for Faecal Sludge Dewatering – ASynopsis of Reviewed Literature and Interim Results of Pilot Investigationswith Septage Treatment in Bangkok, Thailand, Asian Institute of TechnologyBangkok, Thailand – Urban Env. Engineering & Management Program, SwissFederal Institute for Environmental Science & Technology, 1998.

37] T. Koottatep, N. Surinkul, A.S.M. Kamal, C. Polprasert, A. Montangero, K.

and their polishing systems, in: Proceedings, 9th International IWA SpecialistGroup Conference on Wetland Systems for Water Pollution Control and 6thInternational IWA Specialist Group Conference on Waste Stabilization Ponds,Avignon, France, 2004, 27 September–1 October, 2004.