Comparison of the treatment performances of blast furnace slag-based and gravel-based vertical flow...

Ecological Engineering 24 (2005) 187–200

Comparison of the treatment performances of blast furnaceslag-based and gravel-based vertical flow wetlands

operated identically for domestic wastewatertreatment in Turkey

E. Asuman Korkusuza, Meryem Beklioglub, Goksel N. Demirerc,∗

a Middle East Technical University (METU), Biotechnology, 06531 Ankara, Turkeyb METU, Department of Biology, 06531 Ankara, Turkey

c METU, Department of Environmental Engineering, Inonu Bulvari, 06531 Ankara, Turkey

Received 13 February 2004; received in revised form 14 October 2004; accepted 29 October 2004

Abstract

In 2001, to foster the practical development of constructed wetlands (CWs) used for domestic wastewater treatment inTurkey, vertical subsurface flow constructed wetlands (30 m2 of each) were implemented on the campus of the METU, Ankara,Turkey. The main objective of the research was to quantify the effect of different filter media on the treatment performanceof vertical flow wetlands in the prevailing climate of Ankara. Thus, a gravel-filled wetland and a blast furnace granulated irons f0 s,a and 59%),ca r than thato neer fori logy coulda©

K g

f

co-ds”

sup-

0

lag-filled wetland were operated identically with primarily treated domestic wastewater (3 m3 d−1) at a hydraulic loading rate o.100 m d−1, intermittently. Both of the wetland cells were planted withPhragmites australis. According to the first year resultverage removal efficiencies for the slag and gravel wetland cells were as follows: total suspended solids (TSS) (63%hemical oxygen demand (COD) (47% and 44%), NH4

+–N (88% and 53%), total nitrogen (TN) (44% and 39%), PO43−-P (44%

nd 1%) and total phosphorus (TP) (45% and 4%). The treatment performances of the slag-filled wetland were bettef the gravel-filled wetland in terms of removal of phosphorus and production of nitrate. Since this study was a pio

mplementation of subsurface constructed wetlands in Turkey using local sources, it has proved that this eco-technolso be used effectively for water quality enhancement in Turkey.2004 Elsevier B.V. All rights reserved.

eywords:Vertical flow constructed wetland; Domestic wastewater treatment; Nutrient removal; Gravel; Blast furnace granulated sla

∗ Corresponding author. Tel.: +90 312 210 58 67;ax: +90 312 210 12 60.E-mail address:[email protected] (G.N. Demirer).

1. Introduction

Being low-cost and low-technology systems, etechnological approaches like “constructed wetlan(CWs) are now standing as potential alternatives or

925-8574/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.ecoleng.2004.10.002

188 E.A. Korkusuz et al. / Ecological Engineering 24 (2005) 187–200

plementary systems for the treatment of municipal, in-dustrial, agricultural wastewater, as well as stormwater(Moshiri, 1993; Kadlec and Brix, 1995; Kadlec andKnight, 1996; Cooper et al., 1996; Vymazal et al.,1998; Haberl, 1999; Kivaisi, 2001). Since the 1950s,throughout the world, constructed wetlands have beenused effectively for several purposes with different con-figurations, scales and designs. This was because oftheir nutrient capturing capacity, simplicity, low con-struction/operation and maintenance cost, low energydemand, process stability, little excess sludge produc-tion, effectiveness and potential for creating biodiver-sity (Haberl, 1999).

Constructed wetland technology is more wide-spread in industrialized countries due to more strin-gent discharge standards, finance availability, change intendency to use on-site technologies instead of central-ized systems and to the existing pool of experience andknowledge based on science and practical work. Eventhough the potential for application of wetland technol-ogy in the developing world is enormous, the rate ofadoption of wetlands technology for wastewater treat-ment in those countries has been slow (Kivaisi, 2001).Recently, as a result of the transfer of the knowledge,technical collaboration and co-operation by the devel-oped countries, a variety of applications for CW tech-nology for water quality improvement has also startedto be implemented in developing countries like China,Kenya, Mexico, Nepal, Nicaragua, Tanzania, Uganda,India, Morocco, Iran, Thailand, and Egypt (Haberl,1

reatn ffec-t s inT

msl so-l teri calec eat-mt on-s reat-m sur-fw ntfi theM a,

Turkey, in 2001. The design and implementation of theconstructed wetlands were based solely on utilizing thelocal resources. The main objective of the research wasto quantify the effect of different filter media (graveland granulated blast furnace slag) on the treatment per-formance of vertical flow wetlands in the prevailingclimate of Ankara.

In this paper, the first year removal performancesof the identically operated gravel-based and blastfurnace slag-based constructed wetlands of METUare presented. Total suspended solids (TSS), chem-ical oxygen demand (COD), total phosphorus (TP),ortho-phosphate phosphorus (PO4

3−-P), ammonium-nitrogen (NH4

+–N), nitrate-nitrogen (NO3−–N), andtotal nitrogen (TN) are compared to each other statis-tically and to the literature.

2. Materials and methods

2.1. Sizing of the constructed wetlands of METU

In 2001, two vertical subsurface flow constructedwetlands with dimensions of 4.5 m× 6.5 m× 0.60 m(W × L × D) and surface areas of 30 m2, were imple-mented at the abandoned wastewater treatment plantat METU. The bottoms of the wetlands were linedwith polyethylene of a thickness of 1 mm. A slopeof 1% was created at the bottom of the wetlands toallow easier water collection. One of these wetlandsw nd3 henw firstfi edb thatw m-p cmo VCp enlyo ver,p thet ctedw wasi

ootsow cam-p den-

999; Kivaisi, 2001).Similar to other developing countries, there is a g

eed for simpler, cheaper, and more reliable, eive and practical wastewater treatment alternativeurkey.

Therefore, implementing low-technology systeike constructed wetlands can also be appropriateutions for treatment of different types of wastewan Turkey. However, there have not been any full-sonstructed wetland applications for wastewater trent in Turkey, until 2001 (Korkusuz et al., 2001). In

his regard, to foster the practical development of ctructed wetlands used for domestic wastewater tent in Turkey, two parallel sets of the vertical sub

ace flow pilot-scale constructed wetlands (30 m2 each)ith identical design configurations but with differelter media, were implemented on the campus ofiddle East Technical University (METU), Ankar

as first filled with gravel (15 cm of 15/30 mm a0 cm of 7/15 mm) from the bottom to the top and tith sand (15 cm of 0–3 mm). The other one waslled with gravel (15 cm 15/30 mm), then with sievlast furnace granulated slag (30 cm of 0–3 mm)as provided from KARDEMIR Iron and Steel Coany, Karabuk, Turkey, and finally with sand (15f 0–3 mm) at the top layer. Several sizes of Pipes were used to distribute the wastewater flow evnto the vertical flow subsurface wetlands. Moreoolyethylene drainflex pipes were used to collect

reated wastewater. The plan view of the construetlands implemented on the campus of METU

llustrated inFig. 1.Constructed wetlands were planted with the sh

f the common reed (Phragmites australis), whichere transferred from the natural reed beds on theus, placed in the soil tubes and transplanted at a

E.A. Korkusuz et al. / Ecological Engineering 24 (2005) 187–200 189

Fig. 1. Plan view of the vertical flow constructed wetlands of METU (P1: submersible pump; CP1: control panel 1; S1: spherical valve 1).

sity of 9 seedlings m−2, in May 2002 (Korkusuz et al.,2002). Even though the plants were not fertilized, theygrew up very fast (1 cm d−1) in the first month as theystarted to receive the domestic wastewater.

2.2. Operation of the constructed wetlands ofMETU

Using a submersible pump, the raw domesticwastewater was diverted from the nearest manhole totwo of the sedimentation tanks. The manhole receivedboth the domestic wastewater and the rainwater. Bothof the sedimentation tanks have effective volumes of3.61 m3 (height: 3.10 m; diameter: 1.35 m). The set-tled domestic wastewater was manually diverted fromthe sedimentation tanks via spherical valves and PVCpipes to the wetlands, once a day for about one hour,

yielding an influent discharge rate of 3 m3 d−1 and ofa hydraulic loading rate of 0.100 m d−1. The operationand monitoring of the wetlands of METU were con-ducted between July 2002 and 2003.

2.3. Monitoring and measurements

The influent and effluent water samples of the pilot-scale constructed wetlands were taken periodically toevaluate the treatment performances of the wetlandsfrom July 2002 through July 2003. Water sampleswere taken and brought to the Chemistry Laboratoryof the Department of Environmental Engineering ofMETU in 15 min. Chemical analyses were performedon the same day according to the Standard Methods(AWWA, 1999). Temperature, conductivity, pH, TSS,COD, NH4

+–N, NO3−–N, TN, PO4

3−-P and TP were


monitored. For each of the parameters, samples wereanalyzed in duplicates. The samples were analyzed forCOD and phosphorus without any filtration. The mete-orological daily average data (air temperature, precipi-tation and evaporation) were provided from the nearestmeteorological station of the General Directory of Ser-vices of Villages, Ankara (2003).

2.4. Statistical analysis

For differentiating the treatment performances ofthe slag system and gravel system, discriminant func-tion analysis has been performed. One-way ANOVA(at a significance level of 0.05) was applied to the re-moval efficiencies for the July 2002–2003 monitoringperiod for each of the water quality parameters. Theseanalyses were conducted by using a sub-programmeof EXCEL XP of MICROSOFT OfficeTM Software.The statistical results were presented in the followingform: (ANOVA; F0.95(d.f.; dN); P) whereF0.95= 95%confidence limit; d.f.: degree of freedom; dN = samplesize;P> 0.05 nonsignificance (accept null hypothesis)in the related sections of results and discussions.

3. Results and discussion

3.1. Characterization of the domestic wastewaterapplied to the wetlands of METU

wa-t re-s tra-t wa-t al-u to byT

er( alBc ,t isa oft r-t es-t bil-i ter.

These differences can be resulting from the differencesof the water usage habits of the inhabitants of METUand also the dilution of the wastewater from the pre-cipitation.

Since the raw domestic wastewater has been kept inthe sedimentation tanks for 2–3 h, TSS concentrationshave been reduced almost by half after primary treat-ment, whereas BOD5 and COD concentrations havebeen reduced by about 15%. There were not signifi-cant differences in the nitrogen and phosphorus con-centrations before and after the primary treatment dueto the prevailing aerobic conditions in the sedimenta-tion tanks.

3.2. The effect of water budget on the pollutantconcentrations of wetlands of METU

For the pilot-scale constructed wetlands of METU,the meteorological data was gathered from the neareststation located at the Research Institute of the Servicesof Villages, Ankara (2003). The monthly average tem-perature, precipitation and evaporation data taken onthe sampling days for the period of July 2002 throughJuly 2003 are illustrated inFig. 2.

The inflow and outflow concentrations of the or-ganics and nutrients to be treated in the constructedwetlands are affected by additional water inputs andoutputs (like precipitation, evapotranspiration, ground-water infiltration, etc.) besides the fluctuations in thewastewater (IWA, 2000). Precipitation dilutes the pol-l them treat-m . Ont con-c ndsd t them e ac-t rgesh f theetr thec flowv dy,t in ar entp haven hese

The characterization of the raw domestic wasteer of METU taken directly from the manhole is pented inTable 1. Generally, the average concenions of the pollutants of the raw domestic wasteer of METU are lower than that of the literature ves given byVymazal et al. (1998); but similar to thaf the typical values of domestic wastewater givenchobanoglous and Burton (1991).

However, the BOD5 value of the METU wastewat65± 30 mg L−1) is comparably lower than the typicOD5 value of raw domestic wastewater (220 mg L−1)ited in Tchobanoglous and Burton, 1991. Moreoverhe COD:BOD5 ratio of the wastewater of METUbout 4.3, which is significantly higher than that

he literature value of 1.14 (Tchobanoglous and Buon, 1991). Thus, it can be concluded that the domic wastewater of the METU has a low biodegradaty and can be classified as “low-strength” wastewa

utant concentrations within the wetland so thateasured effluent values are lower than the actualent performance accomplished by the wetland

he contrary, evaporation and evapotranspirationentrate the pollutant concentrations in the wetlaue to the decrease in the water volume so thaeasured effluent values are higher than that of th

ual ones. Therefore, daily average outflow dischaave been calculated by adding the difference ovaporation and rain values multiplied with 30 m2 tohe daily measured inflow values of 3 m3 d−1. The cor-ection factors have been calculated by dividingalculated outflow values by the measured daily inalues. According to the correction factors of this stuhe measured outflow concentrations can vary withange of−5.9% and 16.5%. However, as the treatmerformances of most of the treatment wetlandsot been presented in the literature considering t


Table 1Characterization of the raw domestic wastewater

Parameter (mg L−1) BOD5 COD TSS PO43−-P TP NH4+–N NO3

−–N TN

Average 65 279.45 102.33 3.34 6.14 23.86 0.60 34.69Standard deviation 30 52.40 9.23 0.74 0.11 6.17 0.34 9.23Minimum 40 221 80 2.60 5.98 14.03 0.13 28.10Maximum 90 348 125 4.34 6.19 29.46 0.91 50.85

Fig. 2. Monthly average meteorological data (July 2002–2003).

correction factors (IWA, 2000), the outflow concentra-tions of this study have also been presented without anycorrections.

3.3. The treatment performances of theconstructed wetlands of METU

At the start of the study, the average height (cm)(±S.D.) of the reed plants was 40± 9 cm (n= 20) forboth of wetlands; whereas at the end of the study, theaverage heights of the plants were 149± 23 cm and120± 18 cm for the slag and gravel wetlands, respec-tively. The annual growth rate of the reed plants in theslag and gravel wetland were 2.72 and 2.00 mm d−1, re-spectively. The plants adapted themselves well into thewetlands and did not wither throughout the monitoringperiod.

The results of the monitoring study (July2002–2003) are presented graphically inFigs. 3–9

Fig. 3. TSS (a and b) time vs. influent and effluent concentrations ofthe slag and gravel wetland systems of METU.


Fig. 4. COD (a and b) time vs. influent and effluent concentrationsof the slag and gravel wetland systems of METU.

for both the gravel and slag systems. In these figures,sampling date versus influent and effluent concentra-tions (with their standard deviations) of TSS, COD,NH4

+–N, NO3−–N, TN, PO4

3−-P and TP have beenillustrated. Throughout the monitoring period, theminimum and maximum concentrations of the waterquality parameters of the primarily treated domesticwastewater were as follows: TSS 20–183 mg L−1

(Fig. 3); COD 85–446 mg L−1 (Fig. 4); NH4+–N

Fig. 5. NH4+–N (a and b) time vs. influent and effluent concentra-

tions of the slag and gravel wetland systems of METU.

Fig. 6. (a and b) time vs. influent NH4+–N and effluent NO3−–N con-centrations of the of the slag and gravel wetland systems of METU.

15–48 mg L−1 (Fig. 5); NO3−–N 0–15 mg L−1

(Fig. 6); TN 19–62 mg L−1 (Fig. 7); PO43−-P

3–8 mg L−1 (Fig. 8); and TP 4–9 mg L−1 (Fig. 9). Thecalculated mean concentrations± standard deviations(mg L−1) of the influent pollutants were as follows:TSS 74.72± 42.91 mg L−1; COD 244± 108 mg L−1;NH4

+–N 24.87± 6.86 mg L−1; NO3−–N 2.00±

3.20 mg L−1; TN 34.24± 10.85 mg L−1; PO43−-P

4.71± 1.54 mg L−1 and TP 6.45± 1.80 mg L−1.

Fig. 7. TN (a and b) time vs. influent and effluent concentrations ofthe slag and gravel wetland systems of METU.


Fig. 8. PO43−-P (a and b) time vs. influent and effluent concentra-

tions of the slag and gravel wetland systems of METU.

Fig. 9. TP (a and b) time vs. influent and effluent concentrations ofthe slag and gravel wetland systems of METU.

The fluctuations in the influent concentrations re-flected the hourly, daily, seasonal and periodical vari-ations of the raw wastewater received by the manhole,from where the wastewater to be treated was diverted.Since the sewerage system of METU also received thesurface runoff, concentrations of some of the wastew-ater parameters changed in rainy seasons. Especiallyduring the heavy rainy days in autumn and snow meltin winter, the influent concentrations of TSS (Fig. 3),

PO43−-P (Fig. 8) and TP (Fig. 9) showed increases

due to the additional inorganics carried by the surfacerunoff; whereas nitrogen concentrations (Figs. 5–7) ofwastewater decreased because of dilution by the rain-water.

The increases in the influent concentrations canalso be attributed to shock loads to the sewer sys-tem. As a result of the start of the new semester(September–October 2002) at METU, academiciansand the students came back to their homes and dor-mitories. They used detergents in large quantities forcleaning purposes, produced more sanitary wastewa-ter as compared to the summer season. This in turnresulted in steep increases in the TSS, PO4

3−-P, TP;as well as increases in NH4

+–N and TN concentrationvalues. Moreover, COD influent concentrations (Fig. 4)showed parallel changes to the changes of suspendedsolids and phosphorus concentrations, since COD val-ues were also affected by the increase of the amount oforganic pollutants and carbons of detergents.

Generally, the influent concentrations of almost allof the water quality parameters monitored in winterare higher than that of the parameters monitored insummer.

For the monitoring period of July 2002–2003, ef-fluent pollutant concentrations of both of the slag andgravel system often fluctuated similarly as the influ-ent concentrations did (Figs. 3–9). For each of the wa-ter quality parameters, the percent removal efficienciesc tew-a s ofM ow-i

3

andg L(ra cieso een4 -t

artso dgea tion

onsidering the influent (pre-settled domestic waster) and effluent concentrations of the wetlandETU, were calculated and presented in the foll

ng sections.

.4. TSS removal

The effluent TSS concentrations of the slagravel systems of METU varied between 4–77 mg−1

26± 17 mg L−1) and 4–82 mg L−1 (32± 22 mg L−1),espectively, during the monitoring period (Fig. 3 (and b). Concentration-based TSS removal efficienf the slag system and gravel system varied betw–77% (63± 22%) and 4–82% (59± 20%), respec

ively (Fig. 4.1.c).The solids in the effluent of the wetlands are the p

f the non-trapped influent solids, the surplus slund plant litter solids in the process of mineraliza


(Borner et al., 1998). Suspended solids effluent con-centrations of the slag and gravel systems of METU(Fig. 3) varied between 4–77 mg L−1 (26± 17 mg L−1)and 4–82 mg L−1 (32± 22 mg L−1), respectively. Dur-ing the start-up period of the wetlands, the effluentTSS concentrations were very low (<10 mg L−1). Asthe time passed, TSS effluent values increased up to amaximum value of 80 mg L−1. Suspended solids in theoutlet of constructed wetlands are a function not only ofthe treatment system, but also the weather conditions.For example after heavy rains or rapid snow melt, theeffluent could be turbid in the wetland systems (Borneret al., 1998). Hence, the increases in the effluent TSSvalues (Fig. 3) could also be related to the flushing ofthe trapped solids from the wetland fill media by theheavy rains in fall and snow melt in winter (Fig. 2).

A three-year-old constructed wetland planted withemergent plants and having an extensive root systemcan enhance the TSS removal efficiency by provid-ing a larger surface area, reducing the water velocityand reinforcing settling and filtration in the root net-work (Brix, 1997; Tanner, 2001). Since the wetlands ofMETU have operated for one year, the observed TSSremoval efficiencies of the slag and gravel system couldmostly be related to the processes of sedimentation, fil-tration, bacterial decomposition and adsorption to thewetland media (Stowell et al., 1981). During the op-eration period, there was not any surface overflow ineither of the wetlands of METU, which might be duethe low organic content of the raw wastewater, efficientp nter-i nda

sedT avels( ceso nots e-w ,i om-p slaga ont hichh sus-pe ef-fl ere

below the Wastewater Treatment Plant Discharge Stan-dards of Turkey’s (1991), which is 35 mg L−1 fortreated domestic wastewater.

3.5. COD removal

In wetland systems, settleable organics are rapidlyremoved under quiescent conditions by deposition andfiltration. Organic compounds are degraded both aero-bically and anaerobically by the heterotrophic microor-ganisms in the wetland systems depending on the oxy-gen concentration in the bed (IWA, 2000). Accordingto the wetland design, the oxygen required for aerobicdegradation can be supplied by diffusion, convectionand oxygen leakage from the macrophyte roots intothe rhizosphere. Thus, treatment efficiency of the con-structed wetlands for the removal of organics is, gener-ally, highly dependent on the oxygen concentration inthe bed; the wetland design; treatment conditions; thecharacteristics of the fill medium (Vymazal et al., 1998;IWA, 2000). Uptake of organic matter by the macro-phytes is negligible compared to biological degradation(Watson et al., 1989).

In the METU case, the COD effluent concen-trations of the vertical flow constructed wetlands(Fig. 4) varied between 28–257 mg L−1 for slag sys-tem (120± 64 mg L−1) and 11–290 mg L−1 for gravelsystem (131± 67 mg L−1). The effluent concentrationswere affected by the fluctuations of the influent CODconcentrations, dilution of the wastewater by the rain-w )s ew-a

Dc t-m U.

CODrs t fillm didn int ;P on-t hichp ratesw endo uffi-c uld

retreatment of suspended solids by 50% before eng the wetlands and low plant litter production accumulation.

For the monitoring period, the concentration-baSS removal efficiencies of the slag system and grystem varied between 4–77% (63± 22%) and 4–82%59± 20%), respectively. TSS treatment performanf identically operated slag and gravel system didignificantly differ from each other statistically (onay ANOVA; F0.95(1;96) = 1.09;P> 0.05). Therefore

t could be stated that the differences in the size, cositions and porosities of the substrates of thend gravel system did not show significant effects

he TSS removal performances of the wetlands, wave received pretreated wastewater with lowerended solids concentrations (75± 43 mg L−1). Gen-rally, throughout the monitoring period, averageuent TSS concentration of both of the systems w

ater and seasonal changes. AsVymazal et al. (1998tated, the higher the COD:BODs ratio of the wastter, the less biodegradable the wastewater is.

Therefore, it is more difficult to reduce the COoncentrations below 50 mg L−1 after secondary treaent, which was also valid for the wetlands of METSlag and gravel wetland systems had average

emoval efficiencies of 47± 18% and 44± 21%, re-pectively. Even though the wetlands had differenedia with different particle sizes, both wetlandsot significantly differ from each other statistically

erms of COD (one-way ANOVA;F0.95(1;90) = 0.54> 0.05). It could be explained by the low organic c

ent of the wastewater applied to the wetlands, wrobably has not clogged the pores of the substith settled organics. The similar COD treatment trf both of the wetlands could also be related to sient oxygen diffusion into the wetland cells. It co


be also stated that the COD removal in both of the wet-lands were mainly due to the biological degradationand secondarily due to plant uptake since the one-year-old wetlands of METU might not have established anextensive plant root network during the operation pe-riod. Both of the average effluent COD concentrationswere around a value of 125 mg L−1, which is the limitCOD effluent discharge concentration according to theWastewater Treatment Plant Discharge Standards ofTurkey’s (1991).

3.6. Nitrogen removal

Nitrogen in wastewater exists commonly in the formof organic, ammonia, nitrite, nitrate, and gaseous ni-trogen. All these forms of nitrogen are biochemicallyinterconvertible and are components of nitrogen cycle.Nitrogen compounds, particularly ammonia, can exerta significant oxygen demand through biological nitri-fication; may cause eutrophication in receiving waters;and can be toxic to aquatic organisms. Therefore, theneed for nitrogen control in wastewater effluents has,generally, been recognized and many treatment pro-cesses have been developed to remove nitrogen fromthe wastewater stream (Lee and Lin, 1999).

In recent years, constructed wetlands have alsobeen used extensively for tertiary treatment (IWA,2000). The removal mechanisms for nitrogen in con-structed wetlands are manifold and include volatiliza-tion, ammonification, nitrification/denitrification, plantu havep osto a-t ,n on-s wa-t t toa ew l canb andd ientn s tob ni-t ani ioni m-p hichm ni-

trate produced can subsequently be reduced to nitrogengas by biological denitrification if there is readily avail-able carbon source (Haberl et al., 1995; Cooper et al.,1996; Vymazal et al., 1998).

In METU, the NH4+–N effluent concentrations

of the slag system (Fig. 5) varied between 0 and16.6 mg L−1 with an average of 3.23± 3.28 mg L−1

while the gravel system effluent NH4+–N varied

between 3.2 and 24.5 mg L−1 with an average of12± 5.5 mg L−1. The concentration-based NH4

+–Nremoval efficiencies of the slag system varied be-tween 33 and 100% (88.4± 11.8%), whereas in thegravel system it varied between 15.6 and 78.5%(53.3± 14.6%). Even though the ammonium removalperformance of the slag system was statistically bet-ter than that of gravel system (one-way ANOVA;F0.95(l;100) = 173.85;P< 0.0001), both of the sys-tems indicate better nitrification as compared toother vertical flow wetland systems in other coun-tries. Average NH4+–N reductions of the slag sys-tem of METU display higher removals than the av-erage NH4+–N concentrations (from 27± 8.3 mg L−1

to 3.2± 3.2 mg L−1) of some of the studies citedby Vymazal et al. (1998); whereas the averageNH4

+–N reductions of the gravel system of METU(from 27± 8.3 mg L−1 to 12± 5.5 mg L−1) were stillhigher than that of recorded in France (25 mg L−1 to18 mg L−1) (Lienard et al., 1998).

The higher nitrification capacities of the intermit-tently operated vertical flow wetlands of METU can bea tmo-s rs tew-a bleo ca-t hers o theg er tot en-t didn wedo oft sedn ate.M thee y ofa rea.S ang-

ptake and matrix adsorption. Numerous studiesroven that the major removal mechanism in mf the constructed wetlands is microbial nitrific

ion/denitrification (Vymazal et al., 1998). Howeveritrogen removal performance of subsurface flow ctructed wetlands treating ammonium-rich wasteers is often relatively poor and has proven difficulccurately predict (IWA, 2000). At higher loads to thetlands, only suspended solid and carbon removae obtained, whereas at lower loads nitrificationenitrification can take place, hi order to have efficitrification, most of the biodegradable carbon hae removed first from the wastewater, enabling the

rifying bacteria to convert ammonium to nitrate. Inntermittently loaded vertical flow system, oxygenatn the wetland matrix is increased several fold coared to the horizontal subsurface flow systems, way result in efficient nitrification processes. The

ttributed to enhanced oxygen transfer from the aphere to the beds (Brix, 1997). Moreover, due to lowetrength and biodegradability of the domestic waster (Table 1) applied to the wetlands, the availaxygen in the cells might have been used for nitrifi

ion instead of carbon removal, hi addition, the higurface area of the slag particles as compared travel, might have enhanced more oxygen transf

he biofilm. Due to the low COD and TSS concrations of the wastewater, the smaller slag mediaot clog from the heterotrophic bacteria and shoff its real potential in nutrient removal. In both

he systems, vegetation might have slightly increaitrification through the oxygenation of the substroreover, the root system might have facilitatedstablishment of a rich and productive communitttached nitrifiers by providing higher surface aince both gravel and slag are not cationic exch


ers and their surface area is relatively small comparedwith soil, it is plausible to believe that NH4+–N werenot adsorbed in great quantities in either of the wetlandmatrices.

During the monitoring period, the NO3−–N efflu-ent concentrations of the slag system and gravel sys-tem (Fig. 6) ranged from 1.08 mg L−1 to 32.22 mg L−1

(14.8± 7.6 mg L−1), and from 0 mg L−1 to 36 mg L−1

(8.4± 6.3 mg L−1), respectively. The average pH val-ues of the effluents of both of the systems of METUwere just above 7.5, which indicated that the conditionswere suitable for nitrification (IWA, 2000) within bothof the wetland cells. It is well known that the nitrifi-cation reaction is strongly temperature-dependent andlower temperatures below 15 C can significantly reducenitrification (Reddy and Patrick, 1984). As the temper-ature has decreased in the fall and winter season below15 C, the production of nitrate has also decreased be-cause of the reduction of the nitrification rate in both ofthe systems. Due to the lower carbon and TSS contentof the wastewater treated in the constructed wetlandsof METU, both of the beds had higher nitrate effluentconcentrations as compared to other wetland studiescited byLienard et al., (1998); but had lower concen-trations than that of the nitrate concentrations of someof the constructed wetlands in Austria (Haberl et al.,1998).

Comparing the wetlands of METU with each other,the slag system had higher concentrations of nitrate inthe effluent than the gravel system (One-way ANOVA;F rilym oree tionr here sys-t hasl am-m arges robicn tionp r theg oxy-g r toc

TNe andg L(

(20.90± 9.11 mg L−1), respectively (Fig. 7). The TNremoval efficiencies of the slag system ranged from16% to 81.4% (45.5± 20.4%), whereas it ranged from8.7% to 62.1% (40.2± 13.1%) for the gravel system.Comparing the two wetlands of METU in terms of TNremoval performances, they did not differ from eachother statistically (one-way ANOVA;F0.95(l;100) =2.40;P> 0.05).

Average TN reductions of the slag system(from 34.24± 10.85 mg L−1 to 18.54± 8.36 mg L−1)and gravel system (from 34.24± 10.85 mg L−1 to20.90± 9.11 mg L−1) were similar to that of data ofwetlands operated in France (from 46.0 mg L−1 to23.15 mg L−1; ∼50% removal efficiency); in Austria(from 86.1 to 55.0 mg L−1; ∼36% removal efficiency)and in Germany (from 115.0 to 60 mg L−1; −47%removal efficiency), which have been presented inLienard et al., 1998; Haberl et al., 1998; andBorneret al., 1998, respectively. The average effluent TN con-centrations of both of the wetland systems are higherthan 15 mg L−1, which is the limit value stated in theWastewater Treatment Plant Discharge Standards ofTurkey’s (1991). However, it should be noted that forboth of the systems, these nitrogen removals were ob-tained only from a single wetland bed operated forsecondary treatment, not for tertiary treatment. Betternitrogen reductions could be obtained if a horizontalflow constructed wetland with a higher denitrificationcapacity were operated in series.

3

owc usu-a ndeds n re-m tew-a ithintc re-m ion,p tion( int efflu-e thec rate( ub-

= 23.54;P< 0.0001). However, it did not necessaean that the nitrification in the slag system was mfficient than the gravel system since the denitrificaate was not clearly known. But looking at the higffluent ammonium concentrations of the gravel

em, it could be concluded that the gravel systemess nitrification capacity and could not convert the

onium to nitrate as the slag system could. The lurface area of the slag might have enhanced aeitrification and depressed the anoxic denitrificarocesses. However, this might be an advantage foravel system, since the denitrifiers prefer a lessenated condition in the wetland substrate in ordeonvert the nitrified nitrate to nitrogen gas.

The minimum, maximum and averageffluent concentrations of the slag systemravel system were as follows: 7.96–38.78 mg−1

18.54± 8.36 mg L−1) and 11.46–53.63 mg L−1

.7. Phosphorus removal

Contrary to the horizontal flow subsurface flonstructed wetlands, which usually have beenlly designed to remove organic matter and suspeolids, the subsurface flow constructed wetlands caove higher amounts of phosphorus from the waster when the suitable conditions are created w

heir substrates (Vymazal, 2002). In subsurface flowonstructed wetlands; the main mechanisms for Poval are adsorption, complexation and precipitatlant absorption (plant uptake), and biotic assimilaWatson et al., 1989). Generally, phosphate retentionhe constructed wetlands is dependent upon thent quality, loading rate and type of root media andalcium, aluminum, and iron content of the substPant et al., 2001). Since the materials used as a s


strate (e.g. pea gravel, crushed stones, sand, etc.) in thesubsurface constructed wetlands usually do not containhigh concentrations of these elements, the removal ofphosphate is, generally, low and varies widely amongsystems due to different materials used. The removalmay be enhanced by the use of industrial by-productsand natural materials like limestone, gravel, wollas-tonite, spoil from mining, and sand with higher con-centrations of calcium, aluminum, and iron (Vymazalet al., 1998; Brooks et al., 2000).

To have high phosphorus retention, an adequate sub-strate was sought to be used in the wetlands of METU.The candidate substrates have been chosen among thenatural or industrially produced materials available inTurkey at a reasonable price. The phosphorus sorptioncapacity investigation experiments were performed asdescribed byZhu et al. (1997, 2003)and Johansson(1997)in order to choose the suitable substratum to beused in some of the constructed wetlands for tertiarytreatment of domestic wastewater. In this context, theblast furnace granulated slag provided by KARDEMIRIron and Steel Industry, Karabiik, Turkey gave the bestresult in terms of P-sorption capacity (9,150 mg P ad-sorbed/kg slag at a concentration of 320 mg P L−1)among other local candidates. Thus, it has been used asfill medium in one of the wetlands of METU (Korkusuzet al., 2002). In the literature, the adsorption capaci-ties varied between 420 and 44.200 mg P adsorbed/kgslag (Mann and Bavor, 1993; Sakadevan and Bavor,1998; Gruneberg and Kern, 2001). The CaO, A12O3a d inK %,r

s-p ngesi Gen-e avedi s-p -m -t ereaaP velsa of4 ari-a to

the PO43−-P concentrations, since TP is a func-

tion of PO43−-P. The TP effluent concentrations of

the slag system and gravel system have changedbetween 0.81–7.49 mg L−1 (3.29± 1.75 mg L−1) and3.49–9.95 mg L−1 (6.03± 1.50 mg L−1) (Fig. 9). TheTP removal efficiencies of the slag and gravel systemranged from−35.20% to 85.73% (44.85± 28.35%)and from−50.11% to 40.40% (4.33± 16.58%), re-spectively. The slag and gravel vertical flow wetlandsconstructed at METU differed significantly from eachother in terms of phosphorus removal. For PO4

3−-P andTP, one-way ANOVA results wereF0.95(l;100) = 77.05;P< 0.001 andF0.95(l;100) = 84.6;P< 0.0001, respec-tively.

The effluent phosphorus concentrations of both sys-tems (Figs. 8 and 9) were almost constant and very lowfor the gravel wetland only at the beginning of the op-eration period (independent of the influent concentra-tions). However, they rose with the increase in phos-phorus loading rates and operation time. Additionally,the effluent phosphorus concentrations have respondedto precipitation events as the influent phosphorus con-centrations did. The increase in effluent concentrationsof both systems after rain could be explained as a com-bination of dilution and flushing effect on the wetlandbeds. In such cases, it is believed that the already ad-sorbed phosphorus might have been washed down withthe rainwater to the outlet pipes. The already adsorbedphosphorus in the matrix would only be “flushed out”when the phosphorus concentration in the bulk solutionw ion.S n oft thed oor-g ed totn ndso tionsc tionb cha-n rt he pHv fillm ando pHv ntionn ntsc pro-

nd Fe content of the blast furnace slag produceARDEMIR Company was about 34%, 13% and 1

espectively.During the monitoring period, the influent pho

horus concentrations fluctuated reflecting the chan seasons, water and detergent usage habits.rally, effluent phosphorus concentrations beh

n a similar fashion to that of influent phohorus concentrations (Fig. 8). The minimumaximum and average PC4

3−-P effluent concenrations of the slag and gravel system ws follows: 0.34–4.50 mg L−1, 2.36± 1.25 mg L−1

nd 2.32–7.20 mg L−1, 4.54± 1.18 mg L−1; whereasO4

3−-P removal efficiencies of the slag and graystem ranged from−42.86% to 89.03% (44.31± 31%nd from −50.11 to 40.40% with an average.33± 16.58%, respectively. As expected, the vtions in the TP concentrations were similar

ere low, causing an equilibrium shift and desorpto the rainwater could have diluted the bulk solutio

he wetland and resulted in desorption. Also, fromead parts of the plants (common reed) and micranisms, additional phosphorus might have leach

he wetland cells (Vymazal et al., 1998) resulting inegative removal efficiencies. In vertical flow wetlalder than one year, outflow phosphorus concentraan increase with the change of the oxidation-reducehavior and reverse adsorption-precipitation meisms (Lantzke et al., 1999). Another explanation fo

he phosphorus releases could be the decrease in talues of the wetland beds. Since the affinity of theedium for phosphorus depends on its contentsn the pH of the environment, the change in thealues might have affected the phosphorus reteegatively. It is well known that the emergent plaan change the pH values of the fill mediums by


ton release during nutrition, exudation of organic acids,production of carbohydrates, and CO2 release by roots(Olsen et al., 1981).

The constant effluent concentrations of the slagsystem at the beginning might also be explainedmainly by the higher adsorption capacity, where theslag substrate was fresh and the adsorption sites werefree of phosphorus. Since the slag has higher calciumcontent and throughout the monitoring period thepH of the slag system ranged from 7.11 to 8.56 withan average of 7.67± 0.37, it could be claimed thatinsoluble calcium-P compounds have precipitated andslowly converted to apatite, hi acid soils, P is mainlysorbed to Fe- and Al-oxide minerals. Even though slaghad in its composition 12.47% of Al and 0.64% of Fe,since the pH of the slag system did not fall below 7.11,it has been believed that the Al-P and Fe-P oxideswere the least available form (Vymazal et al., 1998).The increase in the effluent phosphorus concentrationsof the slag system could be, therefore, related to a pos-sible decrease in the free calcium ion concentrations,hi addition, the slag system with its smaller particlesize might have given slightly better solids-associatedP removal (reduction of organic phosphorus), whichalso might have contributed to better P removalperformance (ANOVA,P< 0.001). However, sinceboth of the systems did not differ from each other interms of TSS removal (ANOVA,P> 0.05), the largephosphorus removal performance difference betweengravel and slag systems indicated that the organicP deds theo dualo ed toa vedb ver,t eredt s thea sys-t ed tob ke.E cov-e tionc ibeda ca-p itieso bep

The TP results of the slag system were compa-rable especially with the data summarized from theTP performance results of some of the vertical flowwetlands in Belgium (influent 6.8 mg L−1, effluent3.6 mg L−1); in Austria (influent 10.7 mg L−1, effluent4.0 mg L−1); and in Germany (influent 15.9 mg L−1,effluent 4.8 mg L−1) (Vymazal et al., 1998). Thosestudies were supposed to include the wetland systemswith gravel substrate, not slag substrate. Therefore, av-erage TP reductions of the slag system displayed higherefficiencies than those studies, whereas the removal ef-ficiencies of the gravel system closely agreed with theabove-mentioned studies.

For the conventional treatment systems, the al-lowable discharge limit for TP is 2 mg L−1 accord-ing to the Turkish Regulations (1991). In this re-gard, the average effluent TP concentration of theslag system of METU was comparable to the Turk-ish Regulations. Even though it is very difficult to re-duce the TP effluent concentrations below 8 mg L−1

after the secondary units of the conventional treat-ment systems (Lee and Lin, 1999), the slag wet-land system used as a secondary treatment stage inMETU has a very promising TP removal performance.But, asKadlec and Brix (1995)stated, once adsorp-tion sites will become saturated, there will be in-creases in the TP effluent concentrations of thosewetlands.

4

thec nt inT tedw puso a-t ntifyt re-m per-a hed ndsw f thec en-t ndsi ingo cesu

fraction was not only associated with suspenolids. This suggests that a significant fraction ofrganic P in the domestic wastewater was resirganic detergents with phosphate groups attachn alkyl carbon chain, which might be mainly remoy adsorption mechanism in the slag bed. Howe

he carbon chain of the detergents could have covhe adsorption sites so that the P could not accesvailable adsorption sites anymore. In the gravelem, the main P removal mechanisms were suppose filtration, biological assimilation and plant uptaven though the top layer of the gravel bed wasred with sand, which might have also a P-adsorpapacity, the gravel system could not be descrs a wetland cell with noticeable P-adsorptionacity. To investigate the P-adsorption capacf sand and gravel, further investigations shoulderformed.

. Conclusion

In order to foster the practical development ofonstructed wetlands for water quality enhancemeurkey, two of the vertical subsurface flow construcetlands were implemented in 2001 on the camf METU to treat primarily treated domestic wastew

er. The main objective of the research was to quahe effect of different fill mediums on the nutrientoval performance of the constructed wetlands oted identically in the prevailing climate of Ankara. Tesign and implementation of the constructed wetlaere based on the local resources and priorities oountry. This study has been the pioneer for implemation of the outdoors pilot-scale constructed wetlan Turkey. Hence, it will contribute to the understandf how wetland systems function with local resournder the prevailing climate in Turkey.


In one of the wetland beds (gravel system), the filtra-tion media (total depth of 60 cm) chosen was sand andgravel, whereas in the other one (slag system), blastfurnace granulated iron slag was used in addition tosand and gravel. Both of the wetlands (30 m2 each)were planted withP. australis; operated identically ata flow rate of 3 m3 d−1 and a hydraulic loading rate of0.1 m d−1, intermittently. The domestic wastewater ap-plied to the constructed wetlands of METU had lowerbiodegradability when compared with literature values.The influent and effluent water samples of both of thewetlands have been taken periodically and analyzedin parallel for some water quality parameters accord-ing to the procedures described in Standard Methods(AWWA, 1999).

According to the first year monitoring resultsand the statistical analysis, the slag-based wetlandcell with thinner biofilm formation, more surfacearea for adsorption and higher CaO, Al2O3 and Fecontent, was more efficient than the gravel-basedwetland in terms of phosphorus (PO4

3−-P and TP)removal and NO3−–N production. However, bothwetlands had similar treatment performances in termsof TSS and COD. Both of the systems had higherammonification and nitrification capacities whencompared to other wetland applications in othercountries. The differences in the removal perfor-mances have resulted from the physical structuresand the chemical compositions of the fill mediums,as well as the differences of the aerobic and anaerobice ultsi atedc darya ess-f

A

e-s Thea dert icalU thes redo peri-e pprec

References

AWWA, 1999. Standard Methods for the Examination of Water andWastewater. AWWA, Washington, USA.

Borner, T., Von Felde, K., Gschlossl, E., Kunst, S., Wissing, F.W.,1998. Germany. In: Vymazal, J., Brix, H., Cooper, P.F., Green,M.B., Haberl, R. (Eds.), Constructed Wetlands for WastewaterTreatment in Europe. Backhuys Publishers, Leiden.

Brix, H., 1997. Do macrophytes play a role in constructed wetlands?Water Sci. Technol. 35 (5), 11–17.

Brooks, A.S., Rozenwald, M.N., Geohring, L.D., Lion, L.W., Steen-huis, T.S., 2000. Phosphorus Removal by Wollastonite: A Con-structed Wetland Substrate. Ecological Engineering, vol. 15. El-sevier Science, pp. 121–132.

Cooper, P., Smith, M., Maynard, H., 1996. The design and perfor-mance of a nitrifying vertical flow reed bed treatment system.Water Sci. Technol. (Oxford) 35 (5), 215–221.

Gruneberg, B., Kern, J., 2001. Phosphorus retention capacity of iron-ore and blast furnace slag in subsurface flow constructed wet-lands. Water Sci. Technol. 44 (11), 69–75.

Haberl, R., Perfler, R., Mayer, H., 1995. Constructed wetlands inEurope. Water Sci. Technol. (Oxford) 32 (3), 305–315.

Haberl, R., Perfler, R., Laber, J., Grabher, D., 1998. Austria. In:Vymazal, J., Brix, H., Cooper, P.F., Green, M.B., Haberl, R.(Eds.), Constructed Wetlands for Wastewater Treatment in Eu-rope. Backhuys Publishers, Leiden.

Haberl, R., 1999. Constructed wetlands: a chance to solve wastewaterproblems in developing countries. Water Sci. Technol. (Oxford)40 (3), 11–17.

IWA Specialist Group on Use of Macrophytes in Water Pol-lution Control, 2000. Constructed Wetlands for PollutionControl—Processes, Performance, Design and Operation, Sci-entific and Technical Report No.: 8. In: Kadlec, R.H., Knight,R.L., Vymazal, J., Brix, H., Cooper, P., Haberl, R. (Eds.), Inter-national Water Association, London.

J ates)ech-

K wa-1–

K lish-

K.

K -esticrna-on-

K fora re-

L 9. Anted

nvironments within the wetland cells. These resndicated that the properly designed and operonstructed wetlands could also be used for seconnd tertiary wastewater treatment in Turkey, succ

ully.

cknowledgements

This work was partially funded by the METU Rearch Coordination and Industrial Liaison Office.uthors thank all of the directors of the offices un

he General Secretariat of the Middle East Technniversity for their assistance and help duringtudy. All of the wetland researchers, who answeur questions related to the project, shared their exnces and opinions and also encouraged us, are aiated.

-

ohansson, L., 1997. Use of LECA (light expanded clay aggregfor the removal of phosphorus from wastewater. Water Sci. Tnol. 35 (5), 87–94.

adlec, R.H., Brix, H. (Eds.) 1995. Wetland systems forter pollution control. Water Sci. Technol. (Oxford) 32 (3),376.

adlec, R., Knight, R.L., 1996. Treatment Wetlands. Lewis Pubers, Boca Raton, FL.

orkusuz, E.A., Demirer, N.G., Beklioglu, M., 2001. AtiksularmAntiminda Ekilmis. Sulakalanlar, C¸ evre Muhendisleri Odasi IVUlusal Cevre Muhendisligi Kongresi, Mersin, Turkiye.

orkusuz, E.A., Demirer, N.G., Beklioglu, M., 2002. First pilotscale constructed wetland implemented in Turkey for domwastewater treatmant. In: Proceedings of the IWA 8th Intetional Conference on Wetland Systems for Water Pollution Ctrol, Arusha, Tanzania.

ivaisi, A.K., 2001. The Potential for constructed wetlandswastewater treatment and reuse in developing countries:view. Ecol. Eng. 16 (4), 545–560.

antzke, I.R., Mitchell, D.S., Heritage, A.D., Sharma, K.P., 199model of factors controlling orthophosphate removal in plavertical flow wetlands. Ecol. Eng. 12 (1–2), 93–105.


Lee, C.C., Lin, S.D., 1999. Handbook of Environmental EngineeringCalculations, vol. I. McGraw-Hill, New York.

Lienard, A., Coutin, C., Esser, D., 1998. France. In: Vymazal, J., Brix,H., Cooper, P.F., Green, M.B., Haberl, R. (Eds.), ConstructedWetlands for Wastewater Treatment in Europe. BackhuysPub-lisher, Leiden.

Mann, R.A., Bavor, H.J., 1993. Phosphorus removal in constructedwetlands using gravel and industrial waste substrata. Water Sci.Technol. 27, 107–113.

Tchobanoglous, G., Burton, F.L., 1991. Wastewater Engineering:Treatment, Disposal, and Reuse. Metcalf and Eddy, McGraw-Hill Series in Water Resources and Environmental Engineering.New York.

Moshiri, G.A., 1993. Constructed Wetlands for Water Quality Im-provement. Lewis Publishers, Boca Raton, Florida, USA.

Olsen, R.A., Clark, R.B., Bennett, J.H., 1981. The enhance-ment of soil fertility by plant roots. American Scientists 69(July–August), 378–384.

Pant, H.K., Reddy, K.R., Lemon, E., 2001. Phosphorus retention ca-pacity of root bed media of subsurface flow constructed wetlands.Ecol. Eng. 17 (4), 345–355.

Reddy, K.R., Patrick Jr., W.H., 1984. Nitrogen transformations andloss in flooded soils and sediment. CRC Crit. Rev. Environ. Con-trol 13, 273–309.

Sakadevan, K., Bavor, H.J., 1998. Phosphate adsorption charac-teristics of soils, slags and zeolite to be used as substrates in

constructed wetland systems. Water Resources 32 (2), 391–399.

Stowell, R., Ludwig, R., Colt, J., Tchobanoglous, G., 1981. Con-cepts in aquatic treatment design. J. Env. Eng. Div. ASCE 107,919–940.

Tanner, C., 2001. Plant as ecosystem engineers in subsurfaceflow treatment wetlands. Water Sci. Technol. 44 (11–12), 9–17.

Turkey’s Wastewater Discharge Standards, 1991.Vymazal, J., Brix, H., Cooper, P.F., Green, M.B., Haberl, R. (Eds.),

1998. Constructed Wetlands for Wastewater Treatment in Eu-rope. Backhuys Publishers, Leiden.

Vymazal, J., 2002. The use of sub-surface constructed wetlands forwastewater treatment in the Czech Republic: 10 years experience.Ecol. Eng. 18, 633–646.

Watson, J.T., Reed, S.C., Kadlec, R.H., Knight, R.L., Whitehouse,A.E., 1989. Performance expactations and loading rates for con-structed wetlands. In: Hammer, D.A. (Ed.), Constructed Wet-lands for Watewater Treatment, Michigan.

Zhu, T., Jenssen, P.D., Maehlum, T., Krogstad, T., 1997. Phosphoroussorption and chemical characteristics of lightweight aggregates(LWA)-potential filter media in treatment wetlands. Water Sci.Technol. (Oxford) 35 (5), 103–108.

Zhu, T., Maehlum, T., Jenssen, P.D., Krogstad, T., 2003. Phosphorussorption characteristics of a light-weight aggregate. Water Sci.Technol. 48 (5), 93–100.

Comparison of the treatment performances of blast furnace slag-based and gravel-based vertical flow...

Documents

Transcript of Comparison of the treatment performances of blast furnace slag-based and gravel-based vertical flow...