Review

Anaerobic reactor design concepts for the treatment of domestic wastewater

Adrianus van Haandel1, Mario T. Kato2,*, Paula F. F. Cavalcanti1 & Lourdinha Florencio21Department of Civil Engineering, Federal University of Campina Grande, Rua Aprıgio Veloso 882, 58109-970, Campina Grande PB, Brazil; 2Department of Civil Engineering, Federal University of Pernambuco,Avenida Academico Helio Ramos s/n, Cidade Universitaria, 50740-530, Recife PE, Brazil (*author forcorrespondence: phone: +55-81-2126-8228/8742; fax: + 55 81 2126-8219/8205; e-mail: kato@ufpe.br)

Key words: anaerobic reactor concepts, classic reactors, combined reactors, design, domestic sewage,modern reactor configurations, performance, UASB

Abstract

Since the earlier anaerobic treatment systems, the design concepts were improved from classic reactors likeseptic tanks and anaerobic ponds, to modern high rate reactor configurations like anaerobic filters, UASB,EGSB, fixed film fluidized bed and expanded bed reactors, and others. In this paper, anaerobic reactors areevaluated considering the historical evolution and types of wastewaters. The emphasis is on the potentialfor application in domestic sewage treatment, particularly in regions with a hot climate. Proper design andoperation can result in a high capacity and efficiency of organic matter removal using single anaerobicreactors. Performance comparison of anaerobic treatment systems is presented based mostly on a single butpractical parameter, the hydraulic retention time. Combined anaerobic reactor systems as well as combinedanaerobic and non-anaerobic systems are also presented.

1. Historical evolution of anaerobic treatment

When the first anaerobic treatment systemswere developed by the end of the 19th century,the design was not really adequate for goodperformance, possibly due to the misconceptionthat the settleable solids were the most impor-tant sewage component to be removed (McCarty1982). Although they are the most visible con-stituent, they account for only about 1/3 of theorganic load in raw sewage, whereas another 1/3 is present in the form of colloids and theremaining 1/3 as dissolved matter. Much of thecolloidal and soluble fractions were not re-moved because the design of the treatment sys-tems was such that there was no possibility ofcontact between the non-settleable fractions andthe bacteria that were active in the biologicalreactor. As a result the organic material re-moval efficiency was low, of the order of 30 to

40%. After the development of the first aerobicsystems in the decade 1910–1920, another mis-conception was that the lower efficiency of theanaerobic systems was attributable to a lowermetabolic capacity of anaerobic bacteria andtherefore a long retention time would be neces-sary. This has confused sanitary engineers for along time and to a large extent persists untiltoday. In reality, the excellent performance ofthe aerobic process like activate sludge wasmainly due to the intense contact of the activebacteria and the influent organic material, guar-anteed by intense mixing by aeration. By con-trast in classic anaerobic systems, mixing wasunintentional.

The high potential of anaerobic systems wasshown by Young and McCarty (1969), whosuccessfully operated an upflow anaerobic filter,treating concentrated mainly soluble rum distillerywastewater. Two very important improvements

Reviews in Environmental Science and Bio/Technology (2006) 5:21–38 � Springer 2006DOI 10.1007/s11157-005-4888-y

were made: (i) the influent introduced at the bot-tom, following an ascending pathway, thus guar-anteeing intense contact between the bacterialmass and the incoming organic material and (ii)the active sludge mass increased by introducing asludge retention device: a bed of macroscopic bod-ies (stones) in the reactor, onto which the bacterialmass could adhere and grow. The improvementsresulted in a digestion capacity of organic mate-rial, in terms of chemical oxygen demand (COD),of more than 10 kg m)3 day)1, which had notbeen possible even with the most advanced aerobicsystems. Thus, the poor performance of the earlysystems was not because of the ‘‘weak’’ anaerobicbiomass, but because of the poor design. Theanaerobic filter today is still widely applied in sew-age treatment and considered as a precursor of thenew modern high rate systems developed in sub-sequent decades. The basis for the anaerobic treat-ment of high capacity with high efficiency wasestablished by the provision of the two essentialrequisites: intense contact and large bacterial massretention.

An important breakthrough was the develop-ment of the upflow anaerobic sludge blanket(UASB) reactor by Lettinga and co-workers(Lettinga et al. 1980). Originally developed forthe treatment of concentrated (agro) industrialwastes, soon its potential for sewage treatmentbecame apparent, especially in regions with a hotclimate. Hundreds of UASB reactors are now inoperation in many countries, especially in regionswith hot climates.

Although by now it has been amply shownthat anaerobic systems, when properly designed,have a high capacity and efficiency, a singleanaerobic reactor alone often cannot produce aneffluent compatible with legal norms and regula-tions. One alternative is to combine units in or-der to have a pre-treatment step mostly toremove solids, and subsequent anaerobic diges-tion of the soluble organic material in a high ratereactor with an active anaerobic sludge. Howeverin many cases, it is still necessary to introduce apost-treatment system in order to adapt theanaerobic effluents to the required dischargestandards, in terms of residual organic matter,pathogenic microorganisms and possibly nutri-ents. Post-treatment of anaerobic effluents withaerobic or physical-chemical options are dis-cussed in a next chapter.

2. Classic anaerobic treatment systems

Classic anaerobic sewage treatment systems arerelated with the earlier digesters developed byMouras in France (1872), Cameron and Travis inEngland (1896 and 1903, respectively), andImhoff in Germany (1906). They conceived tanksthat are known as decanter-digesters, in whichsettleable solids were retained and digested at thebottom by the anaerobic sludge. Typical of theseclassic systems (septic tank and Imhoff tank) isthe horizontal sewage flow through the system inthe upper part, while the anaerobic sludge restsat the bottom of the tank. Consequently, no spe-cific effort is made to ensure and enhance thecontact between influent organic material and thebiological sludge at the bottom. These two classicsystems found wide application until the decadeof 1930, when they started to loose place in fa-vour of aerobic sewage treatment systems, likethe trickling filter and the activated sludge pro-cess. The maximum efficiency in the classicanaerobic systems does not exceed 30 to 50% ofthe biodegradable matter, depending on the nat-ure of the sewage and the settling efficiency.Nevertheless, both systems are still applied todayin many countries, especially for on-site treat-ment of sewage of individual dwellings or forsmall communities. Another flow-through reactorwith horizontal flow is the anaerobic pond,which is amply used for communal sewage treat-ment, normally associated with other units ofwaste stabilization pond (WSP) systems.

2.1. Septic tank (ST)

Septic tanks are treatment units where primarytreatment (solids settling) is coupled to anaerobicdigestion of the settled sludge. A septic tank maybe constructed either as a single unit or dividedinto two parts: an upper chamber for settlingand a lower for digestion. The latter configura-tion (Figure 1) is called the Imhoff tank after itsinventor. The advantage of separating the twofunctions is that the generated biogas bubblescannot hinder the settling of the solids. For thatreason the liquid retention time in single cham-ber units (12 to 24 h) is longer than in theImhoff tank (2 h in the settling section). Theshorter retention time is a considerable advan-tage, since the volume is reduced in the same

22

proportion. The treatment efficiency of organicmaterial and suspended solids in raw sewage gen-erally is of the order of 30 to 50%.

Because of their simplicity and low cost, sep-tic tanks are probably the most widely appliedtreatment system in the world for treatment ofsewage from individual households. They can bepre-fabricated in fibre-glass or plastic, or con-structed on-site in masonry or concrete. Nor-mally they are installed under the ground andthe effluent is infiltrated into the soil, ordischarged on the surface (gutters) or receivingwaters, eventually after some form of post-treat-ment, commonly the anaerobic filter. Septictanks for single households usually have onlyone chamber. In Brazil the design of these unitsis regulated by a national norm (NBR 7229),which states that the single chamber septic tanksshould have a volume such, that the reten-tion time is 24 h for flows up to 6000 L day)1,decreasing gradually to a minimum of 12 h forflows of more than 14,000 L day)1. The mini-mum volume of these tanks is 1250 L. In case ofsewage treatment, the biological sludge estab-lishes itself in the septic tank without the need ofinoculation, due to the presence of the requiredbacteria in the influent. With time the non-biodegradable sludge mass increases and eventu-ally will take up much of the useful volume of

the tank, decreasing the liquid retention time andthe settleable solids removal efficiency. The accu-mulated digested sludge must be removed period-ically from the septic tanks. The desludgeingperiod varies considerably from one to severalyears. Along with the low efficiency, the need forperiodical desludgeing is the main draw-back ofthe septic tanks.

2.2. Anaerobic pond or lagoon

In practice, the anaerobic pond is not used as atreatment system alone. In conventional WSPsystems, it receives the raw sewage and treatmentis complemented with facultative and maturationponds, where more organic material as well aspathogens (worm eggs and coliforms) are re-moved. This configuration is amply applied insmall and medium communities, especially indeveloping countries. Ponds are normally con-structed as shallow earthen dams with reinforcedmaterial on the slopes to avoid erosion. Plasticlining or a clay layer is placed on bottom toavoid infiltration.

The organic material removal mechanism inanaerobic ponds is the same as in the septic tank:the liquid flows through the pond and the settle-able influent material accumulates at the bottom,where the biodegradable fraction is digested by the

Figure 1. Schematic drawings of septic tanks with (a) single chamber, left; and (b) two chambers (Imhoff tank), right. Measure-ments are in meters.

23

anaerobic sludge mass. Due to the large flow, therequired area of an anaerobic pond is considerableand it is unpractical to cover it, so that the pro-duced biogas escapes to the atmosphere. Theescaping biogas may frequently cause seriousodour problems that can be perceived at a longdistance from the pond. The odour problem iscaused mainly by hydrogen sulphide (H2S), a vola-tile compound that is formed in the anaerobicenvironment due to decomposition of proteins andreduction of sulphate eventually present in thewastewater. A second problem related to theescaping biogas is the emission of methane intothe atmosphere. It is well known that methane isan important contributor to global warming(green house effect).

The retention time of sewage in anaerobicponds (typically 2 to 5 days) is often longerthan in a primary treatment system (septic tank)and correspondingly the removal efficiency oforganic matter tends to be higher. Mara (1976)reported a biochemical oxygen demand (BOD)removal efficiency of 50 to 70% for raw sewagein anaerobic ponds operated at retention timesof 1 to 5 days. The results of several researchersconcerning the BOD removal in regions with ahot climate are plotted in Figure 2 (log–logscale) and showed that the efficiency tendsto improve with increased liquid retention time.For the set of experimental data (all with

temperatures >19 �C), it was possible to derivethe following empirical relationship, Equation 1:

E ¼ 1� 2:4ðHRTÞ�0:50 ð1Þ

whereE = removal efficiency of organic material(BOD) (fraction)HRT = hydraulic or liquid retention time (h)

Efficient BOD removal (more than 80%) canbe achieved only with a long retention time ofapproximately 6 days. This is only possible dueto the removal of part of the non-settleable or-ganic influent matter, probably as a result of arelatively good contact between the influent andthe sludge. Mixing of the liquid phase (depth 2to 5 m) may occur due to agitation caused byrising biogas bubbles, wind and sunshine(mechanical and thermal mixing, respectively). Itmust be noted that below a load of about1000 kg BOD ha)1 day)1 or 0.1 kg BOD m)2 -day)1, a pond tends to be facultative (i.e. withan aerobic top layer), rather than anaerobic.For a usually employed value of the pond depth(2.5 m) and influent BOD of 250 mg l)1, a loadof 0.1 kg BOD m)2 day)1 is attained for aretention time of 6.25 days. Hence a retentiontime of less than about 6 days is required to as-sure anaerobic conditions in the pond.

Figure 2. Experimental data of the BOD removal efficiency E (%), as a function of the liquid retention time HRT (day) in anaero-bic ponds.

24

3. Modern anaerobic single reactor configurations

A breakthrough in design of anaerobic treatmentsystems came about with the development of‘‘modern’’ or high rate systems in the 1960s and1970s, characterized by reactor configurationsresulting in proper hydraulic mixing and sludgeretention mechanisms. Different types of anaero-bic treatment systems have been applied to agreat variety of industrial wastes, but so far theanaerobic treatment concept is still less employedfor sewage, so that experimental information isscarce. In fact, experimental results of anaerobicsewage treatment in modern systems is restrictedto the use of the anaerobic filter, the conven-tional UASB and some of its modified versions,and to a lesser extent the fluidized and expandedbed reactors.

It is important to note that, in spite of thesludge retention mechanism, the sludge hold-upin the anaerobic reactors is limited. Thus if noaction is taken, the reactor unavoidably will be-come ‘‘full’’ of sludge in the sense that no newlyproduced sludge can be accommodated. Fromthat point on the anaerobic reactor will expelsludge at the same rate it is being produced inthe reactor. Naturally the presence of sludge par-ticles in the effluent will increase its total sus-pended solids (TSS) and COD concentrations.Therefore, sludge must be discharged periodicallyfrom the reactor, or a separation device (forexample a settler) must be introduced to separatewashed-out sludge particles from the effluent. Ifsludge is not discharged, the quality of the efflu-ent will be suboptimal as the sludge particles inthe effluent will lead to relatively high TSS andCOD concentrations. The excess sludge normallyis stable and no specific treatment is required; itcan be processed directly to remove water.

3.1. Anaerobic filter (AF)

Full-scale AF-systems (Figure 3) are operatedto treat various types of industrial wastewaters,but for sewage the system is hardly used atlarge scale. An important concern can be thehigh price of many carrier materials that mayresult in costs of the same order as that of theconstruction costs of the reactor itself (Speece1983). In Brazil, AF reactors have frequently

been used as a unit following the septic tanksfor the treatment of the soluble fraction ofdomestic sewage. Although septic tank-AF sys-tems are used predominantly for individualhouseholds, they have also been used for urbanor rural small communities (200 to 5000 inhabit-ants) and housing projects in urban areas wherethere is lack of service by a central seweragenetwork and treatment plant. In most cases, thecarrier material consists of 5-cm constructionstones. Recent studies showed the feasibility ofusing alternatives materials like bamboo rings,river stones, bricks, and pieces of plastic electroducts. They are relatively easily available in themarket, of lower price, lighter and good specificsuperficial area for bacterial adherence (AndradeNeto 2004). While industrial carrier materialslike Pall rings or other modular media tend toimprove the performance of AF (Young 1990),their price is still very high.

3.2. Upflow anaerobic sludge blanket (UASB):conventional and variations

3.2.1. Conventional UASBThe UASB reactor is by far the most widelyused high rate anaerobic system for anaerobicsewage treatment. Several full-scale plants havebeen put in operation and many more are pres-ently under construction, especially under tropi-cal or subtropical conditions. Some studieshave also been carried out in regions with amoderate climate. Figure 4 is a schematic repre-sentation of the conventional UASB reactor.The most characteristic device of the UASB isthe phase separator, placed in the upper sectionand dividing the reactor in a lower part, thedigestion zone, and an upper part, the settlingzone. The sewage is introduced as uniformlyas possible over the reactor bottom, passesthrough the sludge bed and enters into the set-tling zone via the apertures between the phaseseparator elements and is uniformly dischargedat the surface. The biogas produced in thedigestion section is captured by the separatorso that unhindered settling can take place inthe upper zone. To avoid blocking of the bio-gas outlet and allow separation of biogas bub-ble from sludge particles, a gas chamber isintroduced under the separator element. The

25

settled sludge particles on the separator ele-ments eventually slide back into the digestionzone. Thus, the settler enables the system to

maintain a large sludge mass in the reactor,while an effluent essentially free from the sus-pended solids is discharged.

Figure 4. Schematic representation of a conventional UASB reactor with external hydraulic seal to maintain the required waterlevel in the biogas chambers.

Figure 3. Schematic representation of an upflow anaerobic filter with sludge discharge at the bottom.

26

After the conventional UASB concept hadbeen developed in the seventies by Lettinga andhis group in the Netherlands, some modified ver-sions of anaerobic reactor have been proposed,to improve certain characteristics or to broadenits application. Some versions worth consideringsuch as modified UASB are: the UASB withoutinternal settler (RALF), the improved anaerobicponds, the expanded granular sludge bed (EGSB)reactor, the UASB for individual households andthe compartmented UASB.

3.2.2. UASB without internal settler (RALF)In the (mainly) subtropical state of Parana-Brazil, several dozens of plants have been de-signed and built for the treatment of domesticsewage by the local sanitation company Sanepar(Gomes 1985). These reactors, locally calledRALF, were equipped with a different phase sep-arator. Some have small lateral settlers like theunit represented in Figure 5, instead of having asettling zone. In smaller units the settler wasomitted altogether so that the treatment unittransformed itself effectively in an upflow anaer-obic pond. The phase separators were applied tosimplify construction and to reduce costs. Fig-ure 5 shows a schematic representation of aRALF unit built in the city of Londrina andoperational since 1997.

3.2.3. Improved anaerobic pondsTo improve the COD and TSS removal in classicanaerobic treatment systems, such as anaerobicponds and septic or Imhoff tanks, a variety ofsingle or combined processes has been proposedby introducing an upflow direction through thesludge bed for the influent sewage, as used in theUASB concept. Since the mixing intensity is animportant requisite, especially at lower tempera-tures, the upflow mode, instead of the horizontalflow, can improve significantly the contact nee-ded between the anaerobic sludge and the or-ganic matter. Some traditional existing earthenanaerobic ponds were modified by introducingthe influent in several points at the bottom, inorder to induce an upflow mode. In general, theattempts to improve performance by redirectingthe sewage flow have been successful, allowing atthe same time a reduction of HRT to 0.5–1.0 day(Silva 1989). The RALF units shown previouslycan also be considered as upflow anaerobicponds.

3.2.4. Expanded granular sludge bed (EGSB)An important and interesting feature of theUASB process is that a granular type (1 to 5 mmdiameter) of sludge can develop in these systems.These granules have a high density combinedwith a high settling velocity, an excellent mechan-ical strength, and a high specific methanogenic

Figure 5. UASB without internal settler but with a modified phase separator (RALF). Measurements are in centimeters.

27

activity. Consequently, the use of granular sludgeoffers, at least in principle, important benefits.However if the granular sludge settles very welland the wastewater has a low concentration (andhence gas production), problems such as prefer-ential flows, hydraulic short circuiting and deadzones can occur in the conventional UASB reac-tor with granular sludge. The EGSB was devel-oped to overcome those problems by applying ahigher upflow liquid velocity that can be achievedby using an adequate height/diameter ratio oreffluent recirculation (de Man et al. 1988; van derLast and Lettinga 1991). Improved sludge bedexpansion and bulk mixing can be achieved andthus, promote better biomass-substrate contact(Kato 1994; van Lier et al. 1997).

The EGSB concept looks particularly usefulat lower temperatures and relatively very lowstrength of wastewaters, when the productionrate of biogas and, consequently, the mixingintensity induced by it, are relatively low. Underthese conditions, the higher kinetic energy con-tent of the influent and the extended height ofthe expanded granular bed contribute to a betterperformance compared to a conventional UASBreactor. Unless a good settler is installed, anEGSB reactor is inadequate for the removal ofparticulate organic matter in the case of employ-ing high upflow liquid velocity. The influent sus-pended solids are "blown" through the granularbed and leave the reactor with the effluent. Onthe other hand colloidal matter is partially elimi-nated as a result of sorption on the sludge flocs.In a first experiment carried out in EGSB reac-tors, the granular sludge bed expanded as a re-sult of the higher upward velocity that wasapplied, in the range of 6–12 m h)1, against lessthan 1–2 m h)1 usually applied in a UASB reac-tor (van der Last & Lettinga 1991). It is usualthat a seed granular sludge from UASB reactoris needed to start-up the EGSB reactor. How-ever, so far the anaerobic sludge developed inexisting full-scale UASB treating municipalwastewater is predominantly flocculent. Never-theless, since excellent BOD and TSS removalefficiencies were achieved, sludge granulation cer-tainly is not necessarily a prerequisite for success-ful anaerobic sewage treatment in a UASBreactor. In the case of very dilute domestic sew-age (COD<250 mg l)1) treated in an EGSB reac-tor with flocculent sludge from a UASB reactor, the

results showed good performance in the COD andsolids removal when a 4-h HRT and upflow veloc-ity values up to 3.75 m h)1 were applied (Katoet al. 2003).

3.2.5. Compartmented UASBThe compartmented UASB version differs fromthe conventional UASB only in the operationalmode. The reactor is divided in 3 parallel cham-bers with vertical walls rising up to the sludgebed height in the bottom part, in order to dividethe incoming flow in 1/3 depending on the periodof the day. When the minimum, medium or max-imum flow arrive at the plant, 1, 2 or 3 chamberswill be used to receive it, by operating an ade-quate influent distribution system. In this wayand in the case where there is great flow varia-tion, which can be very common in many places,the upflow velocity in the operating chamber willbe kept relatively constant and with decreaseddead zones, favouring the mixing and contactand less loss of solids with the effluent.

3.2.6. UASB for individual householdsThe classic septic tank can also be substitutedby the UASB reactor resulting in an advanta-geous UASB-ST reactor (Coelho et al. 2004;Luostarinen & Rintala 2005). Additionally tothe change in flow direction from horizontal toascensional, a liquid-solid separator has to beintroduced. Higher removal efficiency can be ob-tained due to the fact that in the modified ver-sion the sludge mass has now access not only tothe settleable material, but also to the colloidaland dissolved organic matter. In addition to alarger sludge mass that is retained, Coelho et al.(2002) have reported the better performanceof a single family 380-l UASB-ST reactor (Fig-ure 6) compared to a classic septic tank (1500 l).After formation of the sludge bed, the colloidremoval efficiency was higher and the suspendedsolids removal was better than in the classicseptic tank. The better performance occurredeven when settleable solids were present in theeffluent as expelled sludge particles. The sludgemass tended towards a constant value, equiva-lent to an average sludge concentration of 24 gTSS l)1, but it took 6 months to build up thissludge mass. Insofar as construction is con-cerned, there is also a definite advantage for theUASB-ST reactor. The unit has a much smaller

28

‘‘foot print’’ and hence can be accommodatedmore easily in almost any form and size of theavailable space. It is interesting to note that theUASB-ST must not be desludged: the very pres-ence of the sludge mass is the guarantee thatthe anaerobic treatment will be efficient. Thisgives the unit another important advantage overthe classical ST.

3.3. Fixed film fluidized bed (FB) and expandedbed (AAFEB)

In the fixed film fluidised bed (Figure 7) system,introduced by Jeris (1982), the carrier consists of

a granular medium, which is kept fluidized as aresult of the frictional resistance of the wasteflow. As granular medium initially sand wasused, but later on media with a lower densitywere used, like anthracite and plastic materials,in order to reduce the pumping costs. In prac-tical applications considerable difficulties havebeen experienced in controlling the particle sizeand the density of the particles, consequently tomaintain a stable process performance. Theanaerobic attached film expanded bed (AAFEB)reactor differs from the fluidised bed concept, bythe much lower upflow velocities applied; thesludge bed is only expanded by 10 to 20%

Figure 6. Schematic representation of the single family UASB reactor. Measurements are in millimeters.

29

(Jewell et al. 1981). So far, there are no full-scaleinstallations for sewage treatment using suchreactors. Pereira et al. (1999) operated a 15-mheight small-scale (32 m3) AAFEB with granularactivated carbon during 13 months, treating rawsewage. From this work and other experimentalresults from pilot and bench scale studies, the or-ganic matter removal efficiency was obtained as afunction of the retention time. The experimentaldata obtained by several research workers usingthe fluidised or expanded bed configuration arepresented in Figure 9b.

3.4. Other anaerobic reactor configurations

New or modified anaerobic reactors for thetreatment of domestic sewage have been studied,some being strongly based on the UASB conceptand others not. Nevertheless, their resulting con-figurations are meant to obtain intense mixingand contact, and retention of high active bio-

mass. In general, such reactors include devices toimprove the influent distribution, use of effluentand gas recirculation, special devices for separa-tion of solids and gases from the liquid, severalsupport materials, and so on. In this categorymay be included the horizontal compartmentedreactor with baffles, the rotating disks, the anaer-obic sequential batch reactor (ASBR) and thehorizontal anaerobic immobilized sludge (HAIS)reactor. So far these systems have not yet beeninstalled at full-scale. They have been studied atsmall scale as options for sewage treatment aswell as industrial wastewater treatment. In Brazil,studies on such reactors and many others havebeen conducted within a research program(Prosab) on domestic sewage treatment (Campos1999; Chernicharo 2003).

The horizontal compartmented reactor sepa-rated by vertical baffles results in a tank withchambers in series, where the liquid flows succes-sively from one to another chamber in a upflow

Figure 7. Schematic representation of a fluidised bed anaerobic reactor.

30

mode by using vertical pipes (Figure 8a). Theoperation is similar to a UASB-ST in series, butwithout a solid-liquid separator. It is aimed forsmall communities and can be built without cov-er. An HRT of 12 h or higher is recommendedand the first experiments showed good perfor-mance even in low temperature regions (Orozco1988; Povinelli & Campos 2000).

Similarly to the aerobic version, the anaerobicrotating disks reactor uses plastic porous andlight materials for the development and attach-ment of the microorganisms in submerse paralleldisks on an axis driven by an external motor.The liquid flows horizontally through the reactorand the disks rotating movement results in goodmixing and contact. This concept is aimed toovercome the problems of sludge separation andpreferential flow that occurred in the UASB andin the anaerobic filter, especially at low tempera-tures and low influent COD concentrations. Nev-ertheless, the system shows the weak points offixed film reactors: the biomass and the distribu-tion of the influent are not easily controlled.

After the work of Dague et al. (1966), theASBR has received increased interest due to itspotential use in the practice of wastewater treat-ment, especially when receiving low concentra-tion effluents like domestic sewage and even inthe case of temperate or cold regions. The batchmode operation consists of 4 basic steps, fillingwith influent, reaction, settling and emptying theclarified effluent from a tank. Some attributed

advantages include the good treatment efficiency(BOD and TSS), development of active anddense biomass, and with good settling character-istics, and operational flexibility in the case ofhigh variation of the influent flow. In general thetotal volume required for the tanks is highercompared with that of conventional continuoustanks, but there is no need of a separate settler.

Another fixed film reactor developed byForesti et al. (1995) is the horizontal anaerobicimmobilized sludge reactor (Figure 8b) where theflow is across a long and narrow tube filled withlight polyurethane foam cubes of high specificsuperficial area and of relatively low cost. In thisway, a near piston flux can be achieved and avery high fixed biomass is retained by applyinglow HRT resulting in relatively high organicloading (Zaiat et al. 2000).

3.5. Full scale application of anaerobic treatment

In Brazil, hundreds of anaerobic treatment unitshave been implemented and many more are un-der construction or at the design stage. Thesesystems are scattered over the whole country andhave found particular wide application in thesubtropical south. The vast majority of anaerobicsystems is composed of UASB reactors and itsvariants. The size of the plants varies from smallon site one-family systems (5 inhabitant equiva-lents) to very large communal units (1,000,000

Figure 8. Schematic representation of (a) horizontal compartmented reactor with baffles, left; and (b) horizontal anaerobic immo-bilized sludge reactor (HAIS), right.

31

inhabitant equivalents). Retention times in theplants are usually in the range of 6–10 h. Someanaerobic reactors are followed by post treat-ment units. Aerobic systems (activated sludge orbiological filters) and polishing ponds have beenmost frequently applied.

4. Performance comparison of anaerobic

treatment systems

The most important parameters to compare theperformance of anaerobic treatment systems areefficiency and costs. The efficiency can be conve-niently expressed in terms of the removal effi-ciency of the organic influent material. Inpractice the concentration of the organic materialis usually expressed in terms of COD. Sinceanaerobic treatment systems are simple with littleor no mechanized parts, the investment costs to alarge extent are determined by the volume, whichin turn is given by the flow to be treated and thehydraulic retention time. Thus, in order to com-pare the performance of the different anaerobicsystems, it is important to evaluate the COD re-moval efficiency as a function of the sewageretention time. In Figure 9 the experimental dataof the COD removal efficiency are plotted as afunction of the liquid retention time for 4 differ-ent anaerobic treatment systems. The log–logscale was used because the experimental datapresented by different research workers exhibitvery considerable spread. However, the existingdata do indicate that there is a tendency for anincreasing efficiency with increasing retentiontime. The spread is to be expected since not onlythe treatment units are different but also the ori-gin of the wastewater (degree of separation ofrainwater and industrial contribution, retentiontime in sewerage network) and operational con-ditions such as temperature, weather conditionsand the sludge age. Also it was not always clearif raw or settled effluent was used for the calcula-tions. The available experimental data seem tosuggest a log–log relationship between the re-moval efficiency and the HRT for all of theanaerobic treatment systems for which a reason-able amount of experimental data from full-scaleplants is available. Table 1 shows the empiricrelationships derived from the experimental datafor 5 different types of anaerobic reactors.

Figure 10 shows the tendency of all evaluated anaer-obic treatment systems as a function of the liquidretention time, in accordance to the obtained empiri-cal expressions (Equations 1–5 in Table 1).

It is important to stress that the liquid deten-tion time of a certain anaerobic treatment systemis by no means an unequivocal parameter to esti-mate the treatment efficiency. Cavalcanti (2003)has clearly shown that by introducing a more effi-cient phase separator in a UASB reactor, it waspossible to increase the treatment efficiency, whenthe liquid retention time was kept constant. Thiswas achieved because the improved phase separa-tor allowed to retain a larger sludge mass in thereactor, so that in effect the sludge retention time(SRT) or sludge age (the ratio between the reactorsludge mass and the daily produced excess sludge)was increased. Calvalcanti (2003) also confirmedthat indeed the sludge retention time and not theliquid retention time is the fundamental parameterof anaerobic treatment processes.

Even though the sludge age is at least in prin-ciple a more important parameter than the liquidretention time, the latter has more practicalapplicability. The problem is that the sludge agecannot yet be used to design an anaerobic wastewater treatment plant. Anaerobic digestion the-ory is not yet developed sufficiently to relate apriori the sludge age to reactor size. Apart fromthe complex biological processes that determinethe sludge production rate, other factors such asthe mechanical sludge properties (notably settlea-bility, but also degree of granulation, tendencyto flocculate), liquid currents, intensity of energydissipation from rising biogas bubbles, as well asdesign in features of phase separator and filtermedium, influence the maximum sludge hold-upand hence the sludge age. It is not yet possibleto predict the sludge hold-up in any of the dis-cussed anaerobic treatment systems. Thus, toknow for example, that it is desirable to have asludge age of say 50 days for efficient anaerobicsewage treatment, is not a sufficient basis to de-sign a treatment plant that will effectively exhibitthe required treatment efficiency. Since it is notyet possible to use the sludge age to designanaerobic wastewater plants on a rational basis,a different approach is normally used: from ob-served empiric relationships between efficiencyand liquid retention time in anaerobic plantssimilar to a unit to be designed, the required

32

liquid retention time for some particular treat-ment efficiency is estimated; and the design ofthe plant is made on this ‘‘best estimate’’ of theliquid retention time.

As the spread in available experimental datais large, clearly the actual efficiency in any

particular treatment system can deviate signifi-cantly from value predicted by the empiricexpressions. At any rate some important pointscan be made from an analysis of Figures 9 and10: (i) for temperatures above 20 �C, a COD re-moval efficiency exceeding 80% is possible for

Figure 9. Experimental data of the COD removal efficiency E (%) in anaerobic treatment systems of raw sewage, as a function ofthe liquid retention time HRT (h): (a) anaerobic filter; (b) fluidized and expanded bed; (c) RALF; and (d) UASB.

33

the systems considered, but the required reten-tion times differ significantly according to thesystem (Table 1); (ii) in the range of practicalinterest, the performance of a UASB reactor anda fluidised or expanded bed reactor tend to besimilar with the same retention time: (iii) the per-formance of a well designed conventional UASBsystem is much superior to that of the RALFreactor operated at the same retention time; andthe removal efficiency of an anaerobic pond isvery much inferior to all other considered anaer-obic systems, even when BOD instead of COD isconsidered (the BOD removal efficiency is alwayshigher than the COD removal efficiency).

5. Combined anaerobic reactor systems

5.1. Two-stage systems

When treating a wastewater with a large particu-late organic fraction like sewage, it may beadvantageous to apply a two-stage anaerobicprocess. In the first stage the particulate organicmatter can be entrapped and partially convertedinto soluble compounds, which then are digestedin a subsequent stage. The hydrolytic reactor ofthe first stage will typically contain a flocculentsludge and be operated at a relatively low upflowvelocity. Particulate influent matter can be ad-sorbed on the flocs and partially reintroducedback into the liquid phase after hydrolysis, andthen leave the reactor. Almost no methanogenesiswill occur in the hydrolytic reactor, because theenvironmental and operational conditions are notvery adequate. Moreover, the development of acidfermentation may tend to decrease the pH to avalue below the optimal methanogenic range.Also, due to the accumulation of solids in the firstreactor and the fact that only part of the en-trapped matter will be hydrolysed, there is a needto discharge excess sludge from the reactor at arelatively high frequency. Consequently, thesludge age will remain relatively low and the slow

Table 1. Empirical expressions for the COD removal effi-ciency E (fraction), as a function of the liquid retention timeHRT (h) for different types of anaerobic reactors

Reactor type COD Removal Efficiency (E) Equation

Anaerobic pond* E = 1)2.40 (HRT))0.50 (1)

Anaerobic filter E = 1)0.87 (HRT))0.50 (2)

UASB E = 1)0.68 (HRT))0.68 (3)

RALF E = 1)1.53 (HRT))0.64 (4)

Fluidized bed E = 1)0.56 (HRT))0.60 (5)

* BOD removal efficiency.

Figure 10. Tendencies of the COD removal efficiency E (%), as a function of the liquid retention time HRT (h) in accordance tothe empirical expressions (Equations 1 to 5).

34

growing methanogens cannot develop well. Dueto the fact that part of the suspended matter ad-sorbed in the hydrolytic reactor, notably the bio-degradable matter, will return into the liquidphase, the removal efficiency of the suspended sol-ids will be higher than that of the organic matter.Therefore, since the effluent of the first reactorwill contain predominantly dissolved organic mat-ter, its characteristics are appropriate for treat-ment in an EGSB or UASB reactor.

A drawback of the two-stage system can bethe high solids accumulation in the first reactor,which will occur when the hydrolysis rate be-comes low, as is the case at low temperatures.Under such conditions, the sludge retention timemay become too low to achieve a sufficientlyhigh degree of liquefaction to stabilize the excesssludge. Even in that case, however, it is still pos-sible to obtain a satisfactory excess sludge qual-ity by applying sludge stabilisation in a separateheated digester. With this auxiliary hydrolysis di-gester, the stabilised sludge can be separated in aliquid-solid separation step; and the resultingliquid phase, enriched with soluble organic mat-ter can be mixed back with the regular effluent ofthe unheated hydrolytic reactor, and then treatedin the second methanogenic reactor.

The two-stage anaerobic sewage treatmentsystem with an auxiliary sludge digester is shownin Figure 11. On the left hand side the flow sheetof the treatment system is shown. If it is assumedthat the hydrolytic reactor retains a fraction Y of

the influent COD and that a fraction X of thesesolids is hydrolysed, then the COD fraction thatleaves the hydrolytic reactor is 1)[Y(1)X)]. Theremaining fraction [Y(1)X)] is composed of sol-ids that can be discharged into the externaldigestor, which is presumably heated. If thehydrolysis efficiency in this digestor is E, a frac-tion E [Y(1)X)] is solubilised and can be sent tothe methanogenic reactor, whereas the remainingfraction (1)E)[Y(1)X)] is discharged as stabilisedsludge. If there is a methanogenic efficiency Z inthe final reactor, the digested COD fraction willbe a fraction Z of the sum of the solubilisedCOD fractions from the hydrolytic reactor andthe external digester: 1)[(1)E)Y(1)X)]. TheCOD fraction in the final effluent would be(1)Z)[1)(1)E)Y(1)X)]. The numerical values ofthe variables E, X, Y and Z will depend on theoperational conditions of the three reactors (par-ticularly the sludge age) and the influent charac-teristics (particularly the temperature).

Much research will be required to find theoptimal conditions for maximum methane effi-ciency, low residual organic material concentra-tion and low production of stable sludge forminimum costs. From the above it seems that theinfluent COD in treatment system is divided intothree fractions: (i) methanised, (ii) discharged assludge and (iii) remaining in the effluent. Thecurves in Figure 11 (right) indicate the magni-tude of the three fractions as a function of theoperating temperature. At high temperatures

Figure 11. Two-stage anaerobic sewage treatment.

35

most of the biodegradable material is trans-formed into methane, whereas the effluent CODrepresents basically the non-biodegradable andsoluble influent COD, which cannot be removedin the anaerobic reactor. The COD fraction dis-charged as sludge is comprised of biologicalsludge (bacterial mass) and non-biodegradableand particulate influent COD that is flocculatedin the treatment system. As the temperature de-creases, an increasing fraction of biodegradableinfluent COD is not converted into methanebut instead discharged in the effluent as solublebiodegradable material or in the sludge as partic-ulate biodegradable material.

The current anaerobic reactor configurationscan be combined and used for the objectives ofhydrolyse-acidification and methanogenesis in atwo-stage system. The first stage reactor wouldmainly serve for entrapping, hydrolysing andacidifying the suspended solids of the domesticsewage, especially in the case of high COD con-centration. To obtain the desired separation, adifficulty would be the control of the sludgeretention time in order to avoid methane pro-duction in the first reactor. Consequently, excessless stabilised sludge discharge should be morefrequent in order to maintain a low SRT. An-other difficulty would be the possible flotation ofthe accumulated solids. Some alternatives studiedinclude the use of two UASB reactors in series,an AF and a UASB, and a UASB-ST and aUASB.

Halasheh (2002) operated two UASB reactorsfed with raw sewage at HRT of 8–10 h in thefirst stage reactor. The result showed a COD re-moval efficiency of 50 to 60% without much var-iation between summer time (25 �C), and wintertime (18 �C). A confirmed problem was the needfor further stabilization of the sludge produced inthe first reactor, especially during the winter time.A combination of an AF with polyurethane foammedia and a UASB was conducted at 24 �C, withHRT of 4.6 and 23 h, respectively. A COD re-moval efficiency of 71% was obtained in the AFwhich produced excess sludge that also needed fur-ther stabilization, since the ratio of total volatilesolids to total solids was 0.68. The SRT was19 days and good acidification level occurred in theAF reactor. The ratio of volatile fatty acids to sol-uble COD increased from 33% in the influent to62% in the effluent.

In a UASB-ST reactor, an improvement of theconventional septic tank usually installed for on-site treatment, the long SRT improves the reten-tion of viable sludge. This is an important factorin the case of low temperatures, since the removalof solids and conversion of non-acidified organicmatter can be enhanced. Therefore, the resultingdissolved compounds can then be better removedin the second stage reactor. Luostarinen andRintala (2005) evaluated the feasibility of a sys-tem with a UASB-ST and UASB reactors at lowtemperature (10–20 �C) for on-site treatment ofblack water at an HRT of 4.4 days and 1.4 days,respectively. The two-stage reactor system for alltemperature ranges achieved a high COD andTSS removal efficiency of above 90%. Stabilisa-tion of reactor sludges was also incomplete.

5.2. Two-step systems

Some combined anaerobic reactors have beenstudied with the objective of achieving good per-formance for the hydrolytic and methanogenicreactors. Nevertheless, other combinations aremerely sequential anaerobic reactors in which apost-treatment step, to polish off the first reactoreffluent, is left to the second reactor. A classicalexample is the septic tank with anaerobic filter.Some other options of combined anaerobic reac-tor are the UASB and AF, and the UASB andEGSB.

In the classic ST-AF system, settleable solidsare removed in the first tank and the organicfraction will partly be digested at the bottom.The soluble organic fraction will be treated in theAF, since the configuration is not indicated forraw sewage due to possible frequent cloggingproblems in the media. Only organic matter canbe removed in this system, the removal of coli-forms is not high and depending on the AF med-ia helminth eggs can be significantly retained.

The use of a UASB reactor instead of theclassic septic tank in a UASB-AF system hasbeen shown advantageous in several ways. Nev-ertheless, the UASB effluent may contain aconsiderable suspended solids concentration dueto sludge particles being discharged from thereactor. Coelho et al. (2002) have shown that theimprovement by applying post treatment in anAF is insignificant. Experimental results com-pared with single UASB for the same operational

36

(HRT and organic loading rate) conditionsshowed that there is little difference favouring theUASB-AF reactors in terms of COD removal.However, by using a plastic media of small sizeelectro ducts (1–2 cm) in the AF, the number ofhelminth eggs in the effluent of the UASB-AFwas almost nil, which is important in the case ofwastewater reuse (Andrade Neto 2004; Pimentaet al. 2005).

The EGSB can be used as polishing reactor ofUASB or any other anaerobic effluent, since it isindicated for low strength wastewater and lowtemperature, for which conditions a very high in-tense mixing and contact biomass-wastewater is afundamental factor to be achieved. It is impor-tant that effluent enters the EGSB reactor pre-dominantly in dissolved form, either acidified ornon-acidified, due to its characteristics of highupflow velocity that is unfavourable for retainingsuspended solids. Even with flocculent sludge butconsisting of dense biomass, the results in apilot-scale polishing reactor for UASB domesticsewage effluent showed good removal efficiencyin terms of COD and suspended solids (Katoet al. 2003).

5.3. Other combined systems: anaerobicand non-anaerobic systems

Despite the occasional use of combined anaero-bic reactors alone, probably the most appliedalternative for post-treatment is the combinationof a modern anaerobic treatment reactors in ser-ies, in which most of the settleable solids and alarge fraction of the organic material are re-moved, and a high rate aerobic system to treatthe anaerobic effluent substantially free from set-tleable solids and with a low organic materialconcentration (Guimaraes et al. 2003). Other op-tions are application of polishing ponds for resid-ual organic material removal and disinfection(Cavalcanti 2003). Aerobic post-treatment can beused with several reactors. Activated sludge (incontinuous or sequential batch mode), aeratedlagoons, trickling filters, submerged aeratedfilters and biodiscs are some examples. Post-treatment can also be applied for the removal ofsome components that cannot be removed byanaerobic digestion. In this respect the anamoxprocess for the removal of nitrogen is an exampleof a new development, in which autotrophic

ammonium oxidation with nitrite occurs. Theseand other options for post-treatment will be dis-cussed in a following article.

Acknowledgements

We thank S. Gavazza, C. Melquıades and A.G.F.Vieira (Federal University of Pernambuco) forthe goodwill and help during the elaboration ofthis article. We also acknowledge research col-leagues and students for the previous contribu-tion, and the Brazilian agencies Finep, CNPq andCaixa (program Prosab and Pronex) for the fundsover the years.

References

Andrade Neto CO (2004) Anaerobic filter applied for thetreatment of domestic sewage. Ph.D. Thesis, Federal Uni-versity of Campina Grande, Brazil. In Portuguese

Campos JR (1999) Treatment of Domestic Sewage by Anaer-obic Process and Controlled Disposal on Land. Finep, CNPq& Caixa (Program Prosab), ABES, Rio de Janeiro, Brazil. InPortuguese

Cavalcanti PFF (2003) Integrated Application of the UASBReactor and Ponds for Domestic Sewage Treatment inTropical Regions. Ph.D. Thesis. Wageningen University, TheNetherlands

Chernicharo CAL (2003) Post-treatment of Anaerobic ReactorEffluents. Finep, CNPq & Caixa (Program Prosab), BeloHorizonte, Brazil. In Portuguese

Coelho ALSS, Nascimento MBH, Cavalcanti PFF & vanHaandel AC (2004) The UASB reactor as an alternative forthe septic tank for on-site sewage treatment. Water Scienceand Technology 48: 221–226

Dague RR, McKinney RE & Pfeffer JT (1966) Anaerobicactivated sludge. J. Water Poll. Control Federation 38: 220–225

de Man AWA, van der Last ARM & Lettinga G (1988) TheUse of EGSB and UASB anaerobic systems for low strengthsoluble and complex wastewaters at temperatures rangingfrom 8 to 30 �C In: Hall ER & Hobson PN (Eds) Proceedingsof the Fifth International Symposium on Anaerobic Diges-tion (pp 735–738). Bologna, Italy

Foresti E, Zaiat M, Cabral AKA & Del Nery V (1995)Horizontal-flow anaerobic immobilized sludge (HAIS) reac-tor for paper industry wastewater treatment. BrazilianJ. Chem. Eng. 12: 235–239

Gomes CS (1985) Research at SANEPAR, State of ParanaBrazil, with anaerobic treatment of domestic sewage in full-scale and pilot plants In: Schwitzenbaum MS (Ed) Proceed-ings of Seminar-Workshop on Anaerobic Treatment ofSewage, University of Massachusetts, Amherst, USA

Guimaraes P, Melo HNS, Cavalcanti PFF & van Haandel AC(2003) Anaerobic-aerobic sewage treatment using the com-

37

bination UASB-SBR activated sludge. J. Environ. Sci.Health A 38(11): 2633–2642

Halasheh MM (2002) Anaerobic pre-treatment of strong sew-age. Ph.D. Thesis. Wageningen University, The Netherlands

Jeris JS (1982) Industrial waste water treatment using anaerobicfluidised bed reactors. In: Proceedings of the IAWPRCSeminar on Anaerobic Treatment, Copenhagen, Denmark

Jewell WJ, Schwitzenbaum MS & Morris JW (1981) Municipalwaste water treatment with the anaerobic attached microbialfilm expanded bed process. J. Water Poll. Control Federation53: 482–490

Kato MT (1994) The Anaerobic Treatment of Low StrengthSoluble Wastewaters. Ph.D. Thesis. Wageningen University,The Netherlands

Kato MT, Florencio L & Arantes RFM (2003) Post-treatmentof UASB effluent in an expanded granular sludge bed reactortype using flocculent sludge. Water Sci. Technol. 48: 279–284

Lettinga G, van Velsen AFM, Hobma SW, de Zeeuw W &Klapwijk A (1980) Use of the upflow sludge blanket (USB)reactor concept for biological wastewater treatment, especiallyfor anaerobic treatment. Biotechnol. Bioeng. 22: 699–734

Luostarinen SA & Rintala JA (2005) Anaerobic on-sitetreatment of black water and dairy parlour wastewater inuasb-septic tank at low temperatures. Water Res. 39: 436–448

Mara DD (1976) Sewage Treatment in Hot Climates. JohnWiley & Sons, Chichester, UK

McCarty PL (1982) One hundred years of anaerobic treatmentIn: Hughes et al. (Eds) Anaerobic Digestion 1981 (pp 3–22).Elsevier Biomedical, Amsterdam, The Netherlands

Orozco A (1988) Anaerobic wastewater treatment using anopen plug flow baffled reactor at low temperature In: HallER & Hobson PN (Eds) Proceedings of the Fifth Interna-tional Symposium on Anaerobic Digestion (pp 22–26).Bologna, Italy

Pereira JAR, Campos JR, Mendonca NM, Niciura NM & SilvaMA (1999) Operation of a full-scale anaerobic expanded bedreactor treating raw domestic sewage. In: Proceedings of the20th Brazilian Congress of Sanitary and EnvironmentalEngineering, Rio de Janeiro Brazil (pp 1–9). In Portuguese

Pimenta M, Kato MT, Gavazza S & Florencio L. (2005)Performance of Pilot UASB and Hybrid Reactors for theTreatment of Domestic Sewage. In: Proceedings of the 23rdBrazilian Congress of Sanitary and Environmental Engineer-ing, Campo Grande, Brazil (pp 1–5). In Portuguese

Povinelli SCS & Campos JR (2000) Compartmented anaerobicreactor for the treatment of domestic sewage: granulescharacteristics In: Campos JR (Ed) Treatment of DomesticSewage by Anaerobic Process and Controlled Disposal onLand. Collection of Technical Articles (pp 272–279). Finep,CNPq & Caixa (Program Prosab), Sao Carlos, Brazil. InPortuguese

Speece RE (1983) Anaerobic biotechnology for industrialwastewaters. Environ. Sci. Technol. 17: 416A–427A

Silva SA (1989) Evaluation of UASB Efficiency in the Treat-ment of Domestic Sewage in Northeast Brazil, in Proceedingsof the 15th Brazilian Congress of Sanitary and Environmen-tal Engineering, Belem, Brazil, (pp 94–100). In Portuguese

van der Last ARM & Lettinga G (1991) Anaerobic Treatmentof Domestic Sewage Under Moderate Climatic (Dutch)Conditions Using Upflow Reactors at Increased SuperficialVelocities, in Proceedings of the Sixth International Sympo-sium on Anaerobic Digestion, Sao Paulo, Brazil (pp 153–154)

van Lier J, Rebac S & Lettinga G (1997) High-rate anaerobicwastewater treatment under psychrophilic and thermophilicconditions. Water Sci. Technol. 36: 199–206

Young JC (1990) Summary of Design and Operating Factorsfor Upflow Anaerobic Filters, in Proceedings of the Interna-tional Workshop on Anaerobic Treatment Technology forMunicipal and Industrial Wastewater, Valladolid, Spain

Young JC & McCarty PL (1969) The anaerobic filter for wastetreatment. J. Water Poll. Control Federation 41: 160–165

Zaiat M, Passig FH & Foresti E (2000) Horizontal-flowfixed bed anaerobic reactor applied for the treatment ofdomestic sewage In: Campos JR (Ed) Treatment ofDomestic Sewage by Anaerobic Process and ControlledDisposal on Land. Collection of Technical Articles (pp 280–295). Finep, CNPq & Caixa (Program Prosab), Sao Carlos,Brazil. In Portuguese

38