Available online at www.sciencedirect.com

Bioresource Technology 99 (2008) 3877–3880

Short Communication

Evaluation of inverse anaerobic fluidized bed reactor for treatinghigh strength organic wastewater

R. Sowmeyan a,*, G. Swaminathan b

a Department of Civil Engineering, Periyar Maniammai University, Vallam 613 403, Thanjavur, Tamilnadu, Indiab Department of Civil Engineering, National Institute of Technology, Trichy 620 015, Tamilnadu, India

Received 30 May 2007; received in revised form 14 August 2007; accepted 15 August 2007Available online 27 September 2007

Abstract

The aim of this work is to evaluate the feasibility of an inverse fluidized bed reactor for the anaerobic digestion of distillery effluent,with a carrier material that allows low energy requirements for fluidization, providing also a good surface for biomass attachment anddevelopment. Inverse fluidization particles having specific gravity less than one are carried out in the reactor. The carrier particles chosenfor this study was perlite having specific surface area of 7.010 m2/g and low energy requirements for fluidization. Before starting up thereactor, physical properties of the carrier material were determined. One millimeter diameter perlite particle is found to have a wet spe-cific density of 295 kg/m3. It was used for the treatment of distillery waste and performance studies were carried out for 65 days. Once thedown flow anaerobic fluidized bed system reached the steady state, the organic load was increased step wise by reducing hydraulic reten-tion time (HRT) from 2 days to 0.19 day, while maintaining the constant feed of chemical oxygen demand (COD) concentration. Mostparticles have been covered with a thin biofilm of uniform thickness. This system achieved 84% COD removal at an organic loading rate(OLR) of 35 kg COD/m3/d.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Inverse anaerobic fluidized bed; Start up; Organic loading rate; Hydraulic retention time; Volatile fatty acid

1. Introduction

In the field of anaerobic treatment process, an inverseanaerobic fluidized bed reactor has emerged as a goodalternative for the treatment of wastewater. This anaerobicfluidized bed reactor utilizes small fluidized media particlesto induce extensive cell immobilization thereby achieving ahigh reactor biomass hold-up and a long mean cell resi-dence time (Shieh and Hsu, 1996). The fluidized bed tech-nology presented a series of advantages compared toother kinds of anaerobic processes (Diez-Blanco et al.,1995), like high organic loading rates and short hydraulicretention times. Therefore, a number of design modifica-tions have been tested or adopted in order to improve theperformance of the system. In the classic case of fluidized

0960-8524/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.biortech.2007.08.021

* Corresponding author. Tel.: +91 4362 264600; fax: +91 4362 264660.E-mail address: sowmeyanr@yahoo.co.uk (R. Sowmeyan).

systems, the solid particles have had a higher density thanthe fluid. In an inverse fluidization, the liquid specific den-sity was found to be higher than the specific density of theparticle, and the bed has been expanded downward by theliquid flow.

The down flow fluidized bed reactor or inverse fluidizedbed reactor has been described for application in the anaer-obic treatment of wastewater (Garcia-Calderon et al.,1998a). In their description, of down flow fluidization, par-ticles with a specific gravity smaller than the liquid are flu-idized downward by a concurrent flow of liquid. The paperdescribes the application of the downflow (or inverse) fluid-ization technology for the anaerobic digestion of red winedistillery wastewater. The carrier employed was groundperlite, an expanded volcanic rock.

The biofilm formation and its effect on hydrodynamicsof the reverse fluidized bed reactor have been described(Garcia-Calderon et al., 1998b). The application of inverse

3878 R. Sowmeyan, G. Swaminathan / Bioresource Technology 99 (2008) 3877–3880

fluidization in wastewater treatment from laboratory tofull-scale bioreactors has been described (Karamanev andNikolov, 1996). The biofilm, growing on the surface of sup-port particles, increased the overall bioparticle (supportparticle plus biofilm) diameter. It resulted in bed expansionand very slow movement of the lower bed level downwarduntil the lower bed level reached the lower draft tube open-ing and some of the bioparticles entered the draft tube withthe liquid flow. The inverse fluidized bed biofilm reactordesigned in such a way that the biofilm thickness can becontrolled to avoid the intrabiofilm diffusion limitations.

Under the fluidized state, each media provided a largesurface area for biofilm formation and growth. It enabledthe attainment of high reactor biomass hold-up and pro-moted the system efficiency and stability. This providedan opportunity for higher organic loading rates and greaterresistance to inhibitors. Fluidized bed technology has beenfound to be more effective than anaerobic filter technologyas it favoured the transport of microbial cells from the bulkto the surface and thus enhanced the contact betweenmicroorganisms and the substrate. Further it enhancedhigher organic loading rate in terms of kg/m3/d over anyother bioreactor.

The purpose of study was to evaluate the COD, totalsuspended solids (TSS) removal efficiencies and volatilefatty acids (VFA) production by varying operating vari-ables OLR and HRT.

2. Methods

2.1. Experimental set-up

The reactor consisted of column with a total volume of5.03 l (0.08 m in diameter, 1 m in height). The flow distrib-utor and the gas outlet have been placed at the removablecap covering the top section. The gas outlet was connectedto gas meter. The pH in the reactor was adjusted to 7 withNaOH during the start up period, and then it was naturallymaintained between 7 and 7.5 without addition of NaOH.Anaerobic conditions were maintained by bubblingnitrogen.

2.2. Reactor inoculation and start-up

The reactor was filled with the solid carrier materialupto 55% of its active volume. This material is a light min-eral granular material mainly composed of silica and alu-mina. The reactor was fed with a synthetic wastewatercontaining glucose as a carbon substrate with rumen asinoculum in a ratio of 2:1. Rumen was one of the by-prod-ucts obtained from slaughterhouses of cattle, sheep andcoat. The most abundant rumen bacterial species are thegeneralists or non-specific such as Prevotella ruminicola

and Butyrivit fibrosolvens, which degrade a wide range ofpolymers, Selenomonas ruminantium that utilizes a varietyof end-products (Stewart et al., 1988). After sampling,the rumen contents were strained through four layers of

1 mm nylon mesh. In a laboratory, it was clarified by cen-trifuging and sterlised by autoclaving. In this work, themain constitution of the synthetic wastewater for the testwas as follows (per liter): 12.5 g glucose, 7.12 g K2HPO4,1.36 g KH2PO4, 1.227 g urea, 0.06 g CaCl2 and 0.008 gFeCl3. The reactor was monitored for temperature, flowrate and pH and TSS and VFA have been routinely ana-lysed. The influent OLR was held initially at 6.11 kg ofCOD/m3/d with a constant HRT of 2.0 days. The OLRwas then increased by additional input flow rate. TheOLR was calculated on the basis of the active volume.There was no sludge recycling.

2.3. Measurements and analysis

During the operation of the inverse fluidized bed reac-tor, temperature, pH, gas production rate were monitoreddaily. Feed and effluents were taken for the analysis ofCOD, VFA, and TSS were carried out once per day. Alldeterminations are performed according to StandardMethods (APHA–AWA–WPCF, 1992). The average bio-film size on particles was estimated from diameter measure-ments. The measurement was performed with an opticalmicroscope. The diameter was determined approximatelyfor 50 particles for each sample.

3. Results and discussion

Perlite was an interesting carrier when compared to oth-ers like cork, polyethylene or polypropylene (Garcia-Cald-eron et al., 1998a). The observed physical properties ofperlite particles were used in this test as follows: meandiameter (U) 1 mm, specific dry density (qd) 205 kg/m3,specific wet density (qw) 295 kg/m3, and specific surfacearea (Sa) 7.030 m2/g.

Specific surface area of particle ðSaÞ

¼ Surface area of particle

Volume of particle

Over the first 15 days, very few changes occurred in thereactor concerning the aspect of the carrier particles. Theadjunction of trace elements and nitrogen source in theinfluent mixture quickly modified the aspect of the reactor.First the colour of the particles obviously turned from al-most white to intense gray. Microscopic observations con-firmed the presence of biomass on the carrier, either aslocal outgrowth colonies, or as uniform covering of theparticles. The biomass coverage on the particle was veryuniform. A large number of particles were covered regu-larly by a thin biofilm of constant thickness. This mightbe due to their perfectly spherical shape and to the frictioneffects enhanced by turbulence.

In 70 days of operation, the first 15 days could be con-sidered as ‘‘lost’’ for the process start-up, because of thenutrient deficiency in the wastewater used and the poorresulting biomass growth. The start-up period was divided

0 10 20 30 40 50 60 700

1

2

3

4

5

6

7

8

VFA(

g/l)

Time (days)

Fig. 2. VFA production with the operation time.

0 10 20 30 40 50 60 700.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

TSS(

kg/m

3 )

Time (days)

Fig. 3. Evaluation of TSS with the operation time.

R. Sowmeyan, G. Swaminathan / Bioresource Technology 99 (2008) 3877–3880 3879

into two parts. From day 1 to 33, the influent wastewaterwas taken from the feed tank without the addition of nutri-ents, but very few microbial growth occurred and inputOLR load had to be kept as 6.11 kg COD/m3/d. Fromday 33, a mixture of trace elements was added to the inlettogether with a nitrogen source. Within 10 days, the reactorhad recovered an excellent COD removal efficiency. Thisincreased from 6.11 to 35.09 kg of COD/m3/d over thetwo months of operation. During this period the CODremoval reached 87%. From day 46 to day 70, the inputloading rate could be increased from 15 to 35 withoutany noticeable change in the measured control parameters.Fig. 1 is the plot of input organic load (kg COD/m3/d) andCOD removal efficiency. On 21st day VFA was droppeddown and continued upto 43rd day and reached less than0.5 g/l and VFA amount slightly rose upto 3.5 g/l. Fig. 2shows the VFA concentration at the outlet of the reactor.Fig. 3 represents the evaluation of the TSS in the exit.

A comparison with previously studied reactors treatingthe same kind of effluents was made. The initial start-upwas similar to what was observed with a classical up-flowfluidized bed working with 384 um pozzolana particle(Buffiere et al., 1995). In the experiment, the OLR wasincreased from 2 to 18 kg/m3/d in 70 days, and the carbonremoval varied between 92% and 75%. In an inverse fluid-ized bed with a downflow liquid fluidization with perlite asbiomass carrier, the OLR was increased from 3 to 15 in 60days, but the reactor was destabilized and the carbonremoval was only 55% at the end of the experiment andthe input load had to be decreased (Garcia-Calderonet al., 1998a). In an inverse turbulent bed with upflow cur-rent of gas using extendosphere as biomass carrier, theOLR was increased from 2 to 23.2 kg/m3/d (Buffiereet al., 2000). The removal of total BOD and COD fromthe wastewater were generally very good with 80–92%COD removal and 90–96% BOD removal. The best perfor-mance was observed with a HRT of 4 days (OLR of2.37 kg/m3/d). When detention time was decreased to 2

0 10 20 30 40 50 600

5

10

15

20

25

30

35

40

OLR COD removal

Time (days)

(OLR

(kg/

m3 /d

))

0

20

40

60

80

100

120

COD

rem

oval

rate

(%)

Fig. 1. Variation of the COD removal efficiency with the OLR throughoutthe operation time.

days the efficiency of the GRABBR (Granular Bed Anaer-obic Baffled Reactor) dropped, but the removal rates werestill comparatively good (Akunna and Clark, 2000). Inanother study, during the start-up period, OLR was main-tained approximately at 1.5 kg TOC/m3/d. When the sys-tem reached the steady state, organic load was increasedby reducing HRT. It attained 85% of carbon removal with4.5 kg TOC/m3/d (approximately 11.3 kg COD/m3/d)without pH regulation (Garcia-Calderon et al., 1998b).Other distillery waste studies have been reported usingother anaerobic process such as anaerobic filers (Perezet al., 1996) and UASB (Upflow Anaerobic Sludge Blan-ket) reactor (Craveiro et al., 1986) with 80% and 83%COD removal at organic loads of 12.0 and 13.2 kg COD/m3/d and HTR of 1.4 and 2.4 days, respectively.

Further investigations could be done on such kind ofreactor. For instance, the amount of solid to be used inthe reactor has been fixed here at 55% of the working vol-ume. Other workers reported that a correct fluidized statecould be obtained for filling rates as high as 70–80%, all

3880 R. Sowmeyan, G. Swaminathan / Bioresource Technology 99 (2008) 3877–3880

the more, since the minimal fluidization velocity was lowerfor high solids amount (Comte et al., 1997).

4. Conclusion

This work was intended to test a new kind of anaerobicwith attached biomass, the inverse turbulent bed. Theapplication to the treatment of distillery wastewater pre-sented satisfactory results compared to other reactors.The input organic loading rate could be increased from6.11 to 35.09 kg COD/m3/d within less than 75 days,despite 15 days delay due to a nutrient deficiency problem.The COD removal was kept around 84% and the hydraulicretention time could be reduced down to 0.19 day. Theinverse anaerobic fluidized bed reactor appeared to be agood option for anaerobic treatment of distillery wastewa-ter. The system attained high OLR with good CODremoval rates and exhibited a good stability to the varia-tions in OLR and HRT.

The carrier material was found to be a very importantparameter, because biomass accumulation has broughtabout changes in particle volume and density, affectingthe whole system. Perlite has been found to be good carrierfor the anaerobic digestion of distillery wastewater ininverse fluidized bed.

References

Akunna, J.C., Clark, M., 2000. Performance of a granular-bed anaerobicbaffled reactor (GRABBR) treating whisky distillery wastewater.Bioresource Technology 74, 257–261.

APHA, AWA, WPCF, 1992. Standards Methods for the examination ofwater and wastewater, sixteenth ed. American Public Health Associ-ation, American Water Works Association, Water Pollution ControlFederation, Washington, DC.

Buffiere, P., Fonade, C., Moletta, R., 1995. Continuous operation of afluidized bed reactor for anaerobic digestion: residence time influ-ence on degradation kinetics. Biotechnology Letters 17 (8),833–838.

Buffiere, P., Jean-Pierre, Bergeon, Moletta, R., 2000. The inverse turbulentbed: a novel bioreactor for anaerobic treatment. Water Research 34,673–677.

Comte, M.P., Bastoul, D., Hebrand, G., Roustan, M., Lazarova, V., 1997.Hydrodynamics of a three-phase fluidized bed – the inverse turbulentbed. Chemical Engineering Journal 52, 13971–13977.

Craveiro, A.M., Rocha, B.B.M., Schmidell, W., 1986. In: Water Treat-ment Conference, Aquatech, vol. 86. Amsterdam, pp. 307–319.

Diez-Blanco, V., Garcia-Encina, P., Fernandez-Polanco, F., 1995.Effect of biofilm growth, gas, liquid velocities on the expansionof an anaerobic fluidized bed reactor. Water Research 29,1649–1654.

Garcia-Calderon, D., Buffiere, P., Moletta, R., Elmaleh, S., 1998a.Influence of biomass accumulation on bed expansion characteristicsof a down-flow anaerobic fluidized-bed reactor. Biotechnology &Bioengineering 57, 136–144.

Garcia-Calderon, D., Buffiere, P., Moletta, R., Elmaleh, S., 1998b.Anaerobic digestion of wine distillery wastewater in down-flowfluidized bed. Water Research 32, 3593–3600.

Karamanev, D.G., Nikolov, 1996. Inverse fluidization in wastewatertreatment. Environmental Progress 15, 194–196.

Perez, M., Romero, L.I., Sales, D., 1996. Degradacion anaerobiatermofilica de vinazas de vino: efecto de 1 pH de la alimentacion.Technologia del Agua 158 (II), 41–45.

Shieh, W., Hsu, Y., 1996. Biomass loss from an anaerobic fluidized bedreactor. Water Research 30, 1253–1257.

Stewart, C.S., Flint, H.J., Bryant, M.P., 1988. The rumen bacteria. TheRumen Microbial Ecosystem, 10–72.