PII: S0043-1354(01)00145-2 Wat. Res. Vol. 35, No. 17, pp. 4137–4149, 2001 # 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter ANAEROBIC TREATMENT OF DOMESTIC WASTEWATER IN TEMPERATE CLIMATES: TREATMENT PLANT MODELLING WITH ECONOMIC CONSIDERATIONS P. D. ZAKKOUR1, M. R. GATERELL1, P. GRIFFIN2, R. J. GOCHIN1 and J. N. LESTER1* 1 Environmental Processes and Water Technology Research Group, T. H. Huxley School of the Environment, Earth Sciences and Engineering, Imperial College of Science, Technology and Medicine, Prince Consort Road, London, SW7 2BP, UK and 2 Severn-Trent Water Limited, Technology and Development, Avon House, St Martins Road, Coventry, Warwickshire, CV3 6PR, UK (First received 3 July 2000; accepted in revised form 7 March 2001) Abstract}Although research suggests that anaerobic treatment of low-strength domestic wastewater is possible in temperate climates, to date, full-scale applications have only been pioneered in hot regions. However, burgeoning environmental legislation in developed countries is giving the impetus to develop anaerobic wastewater treatment systems due to potential economic and environmental benefits they hold over traditional aerobic techniques. In this paper a design rationale for low-temperature, low-strength (COD 51000 mg l
1), two-phase anaerobic wastewater treatment is developed through empirical modelling of data from published research, and from assumptions arising from a literature review. Model calculations are applied to typical domestic sewage characteristics at two different flow rates, based on population equivalents. Results indicate that soluble COD production in the model hydrolytic tank are similar to those achieved in pilot scale plants in the Netherlands. Model anaerobic reactor sludge characteristics are similar to those achieved in pilot and full-scale anaerobic reactors treating low-strength wastewaters. Indicative cost figures for a two-phase anaerobic treatment plant are given, but are incomplete without an assessment of the cost of post-treatment processes. Anaerobic treatment is likely to become more attractive in the future as new legislation relating to sludge disposal and renewable energy generation are introduced. # 2001 Elsevier Science Ltd. All rights reserved Key words}anaerobic treatment, domestic wastewater, low temperature, modelling, biogas, sludge, economics INTRODUCTION Increasingly strict European Union environmental legislation is forcing UK water and sewerage companies (WASCs) to undertake more advanced levels of wastewater treatment, leading to greater energy use and sludge production at sewage treatment works (STWs) (Hassan, 1995; CIWEM, 1995; House of Commons, 1999). At the same time, concerns over greenhouse gas emissions have turned political attention towards the introduction of punitive taxes on large energy consumers (Lord Marshall, 1998), and increasing concern over sewage sludge recycling on UK agricultural land has meant that traditional disposal routes for biosolid wastes are diminishing (Anon., 1998a, b). Marine disposal of sewage sludges was phased out in European Union Member States at the end of 1998 as part of the *Author to whom all correspondence should be addressed. Tel.: +44-207-594-6015; fax: +44-207-594-6464; e-mail: j.lester@ic.ac.uk Urban Waste Water Treatment Directive (91/271/ EEC; UWWTD; European Union Council of Ministers, 1991). Consequently, in the UK, costly sludge incineration or gasification plants have been built by some water companies as a means of mitigating the increasing sludge problem (Anon., 1998c). Energy thrifty wastewater treatment processes, such as membrane technologies, have been cited as means of mitigating rising energy costs (Ofwat, 1998). Alternatively, anaerobic wastewater treatment processes may offer a solution. They produce only small amounts of stabilised (non-putrescible) sludge compared to aerobic processes, while much of the removed organic material is converted to methane, which, depending on delivery rate, may be available for energy recovery. Furthermore, the need to aerate the wastewater is dispensed with, reducing energy and construction costs (Mergaert et al., 1992; Lettinga et al., 1997; van Haandel et al., 1996), although aerobic post-treatment may be necessary to meet typical European STW discharge consents. 4137 4138 P. D. Zakkour et al. Notwithstanding these observations, the mainstream wastewater treatment profession have been slow to recognise anaerobic sewage treatment potential (Switzenbaum, 1995), largely due to the perception of anaerobic digestion as a sensitive biological process (Jewell, 1987). In particular, it is considered that the process is only suitable for high-strength, high-temperature wastewaters, and requires careful pH control and protection from toxic shock. Nevertheless, a review published over a decade ago suggested these opinions were beginning to change (Jewell, 1987). Recent research suggests that anaerobic processes can successfully be applied to lowstrength wastewaters at low temperatures (Lettinga et al., 1983; Alderman et al., 1998; Seghezzo et al., 1998; Barber and Stuckey, 1999), and the recent expansion in environmental legislation is giving fresh impetus to the development of processes suitable for anaerobic domestic wastewater treatment. The aim of this paper is to develop a model that describes low-strength, low-temperature, anaerobic wastewater treatment processes so that assessment of possible plant operation and costs can be made. Quantitative estimates of effluent quality, reactor size, biogas and sludge production are made and capital and operating costs estimated for two different size plants. Results provide a guide for wastewater operations managers to the potential of anaerobic sewage treatment in temperate climates, and highlight where further research may be necessary before implementation of full-scale low-temperature (sub 208C) anaerobic sewage treatment may be possible. not been properly elucidated (Pavlostathis and Giraldo-Gomez, 1991). Attempts at empirical modelling of anaerobic processes have been successful (Young and McCarty, 1969; Oh and Yang, 1986; van Haandel et al., 1996; Wilson et al., 1998), although published regression statistics exhibit poor comparability between different research, creating problems when attempting application beyond the original experiment. Of all the previous models, only the first-order temperature dependent model developed by Alderman et al. (1998) has addressed application at sub 208C, although whether results are applicable beyond the original study is unclear. Sub 208C operation is an essential factor when considering the applicability of anaerobic sewage treatment in temperate climates. Therefore, because kinetic models have been unable to successfully describe low-temperature, lowstrength anaerobic applications, this research develops a black-box approach to simulation of the anaerobic degradation process. Following analysis of data from published research (shown in Table 1), a relationship between chemical oxygen demand (COD) removal efficiency, hydraulic retention time (HRT) and temperature was established (Fig. 1). This relationship was then incorporated into a mass-balance model of a two-phase anaerobic reactor, to which different sewage loading rates could be applied (based on population equivalents; pe), and estimates of effluent quality, biogas/sludge production and plant costs could established at different scales. MODEL DEVELOPMENT BACKGROUND TO MODELLING ANAEROBIC PROCESSES To date, a consistent general model of the anaerobic process, able to predict digestion efficiency as a function of operational variables, environmental conditions and influent characteristics, has not been fully developed (van Haandel et al., 1996; Wilson et al., 1998). Monod or Michaelis–Menton type models are of little use at present because meaningful kinetic constants for anaerobic sewage digestion have Because COD removal efficiency data used to construct the model are mainly based on the reduction of total-COD in settled sewage (primary effluent; Table 1), primary separation of the liquid and solid phases in a ‘‘hydrolytic tank’’ (or pre-fermenter) had to be modelled. This led to the development of a sub-model describing hydrolysis of settleable particulate matter in an anaerobic ‘‘hydrolysis’’ tank, simulating either stage one of a two-phase anaerobic plant configuration Table 1. Source data for construction of the empirical model of COD removal Raw waste Substrate COD (mg l
1) Temperature (8C) HRT (h) Treatment efficiency COD removal (%) Settled sewage Non-fat dairy milk Settled sewage Settled sewage Raw sewage Settled sewage 135–218 600 186 288 322–948 391 5–20 5–20 20 20 7.5–18 13–19 1–10 6–24 1–5 24 8–24 1–3.5 35–77 62–95 70–80 73 65–89 16–34 a Conditions Reactor typea and volume References AEBR (2 l) ASBR (6 l) AAFEB (1 l) AF UASB (120 l) EGSB (116 & 205 l) Alderman et al. (1998) Banik et al. (1998) Jewell et al. (1981) Kobayashi et al. (1983) Lettinga et al. (1983) van der Last and Lettinga (1992) AEBR}anaerobic expanded bed reactor; ASBR}anaerobic sequencing batch reactor; AAFEB}anaerobic attached-film expanded bed; AF}anaerobic filter; UASB}upflow anaerobic sludge bed; EGSB}expanded granular sludge bed. Anaerobic treatment modelling and economics 4139 Fig. 1. Relationship between temperature, COD removal efficiency and HRT in experimental anaerobic reactors (data from experiments given in Table 1). (Stronach et al., 1986; van Haandel and Lettinga, 1994; Seghezzo et al., 1998) or the first compartment of an anaerobic baffled reactor (ABR) (Barber and Stuckey, 1999). Design of the hydrolytic tank model is based on data from pre-fermenters used in biological nutrient removal (BNR) plants (Munch and Koch, 1999), and from some limited information on prototype hydrolytic tank designs from the University of Wageningen, The Netherlands (Seghezzo et al., 1998). Anaerobic reactor model design is based on published research using experimental and full-scale reactors treating sewage (described below). Where no consistent data on reactor performance could be found assumptions were made in order to generate a working reactor model. Tables 2 and 3 give the description of rates and transformations occurring within the model. Figure 2 shows the plant layout. The model is based on a daily mass balance which is run consecutively for 467 days starting at the beginning of summer and terminating at the end of the following summer. Performance assessments omit the first summer because the hydrolytic tank and anaerobic reactor have not reached steady-state, although the model assumes that sludge is wasted from day one (although in some operational reactors sludge wasting only commences when concentrations begin to impair reactor performance). Temperature dependent calculations assume a split of 208C, 158C and 108C for summer (90 days), spring/autumn (183 days), and winter (92 days), respectively. Sewage flow rate (dry weather flow; DWF) and load is kept constant at 200 l/pe/day and wastewater characteristics are based on typical parameters (Henze et al., 1997; Table 2), partitioned into different fractions according to Grady et al. (1999; Table 2). Model calculations are undertaken for a 5000 and 50,000 pe works, sized to average flow (equal to DWF*1.2). Hydrolytic tank model The hydrolytic tank model removes 60% of suspended solids (SS) and 30% of the COD from the raw sewage, similar to that achieved in primary clarifiers. HRT in the hydrolytic tank is set at 2 h, again similar to that used in a primary clarifier. Particulate COD settling in the hydrolytic tank is subject to temperature variable hydrolysis rates, while the amount solubilised is dependent on the volume of sludge inside the tank (Table 3, equation (1)). The hydrolysis rate is based on a first order decay coefficient of 0.11 d
1 (at 358C and pH 5.14), taken from research using primary sewage sludge in an anaerobic digester (Pavlostathis and GiraldoGomez, 1991; derived from Eastman and Ferguson, 1981). Based on this decay coefficient, and assuming a 50% reduction in enzymatic activity for every 108C decrease in temperature (Sawyer et al., 1994), hydrolysis rates were estimated for 108C, 158C and 208C (Table 3, equation (1)). A potential drawback of this relationship is that hydrolysis rates may decrease more rapidly below 208C while acidity in the tank may fall below pH 5; however, no consistent rates for low-temperature, low-pH, hydrolysis could be found from a literature survey. Unhydrolysed particulate COD and fixed SS retained in the tank are removed by constant sludge wasting (Table 3, equations (4) and (5)). Sludge wasting is controlled by the solid retention time (SRT) which uses a nominal value equal to the mass of solids in the tank divided by the mass leaving as waste sludge per unit time (days). The nominal SRT 4140 P. D. Zakkour et al. Table 2. Input parameters and definitions of terms used in model calculations Eq. Symbol Definition Dimension Vaule/Comment 1 2 3 4 5 6 7 8 9 10 Liquid flowrate Total suspended solids Fixed suspended solids Volatile suspended solids Total COD Total BOD5 Total particulate COD Soluble COD Particulate inert COD COD to VSS m3 d
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 g g
1 Model inputs Wastewater characteristics Q XTSS XFSS XVSS CCOD CBOD XCOD SCOD XCOD;io fCOD:VSS =0.2 pe 300 90 210 550 250 ( CCOD  2:1)  ¼ XVSS fCOD:VSS ¼ CCOD
XCOD  ¼ XCOD 0:35 1.5 N.B. In the model, concentrations are multiplied by Q to generate a mass balance. Particulate functions Model parametersa 11 12 13 14 ESS ECOD;2 kh XMLSS 15 16 17 18 19 20 FBIOMASS;1 Yobs Y EBIOMASS XVSS;SEED XSLUDGE SS removal efficiency COD removal efficiency Hydrolysis constant Mixed liquor suspended solids (MLSS) Hydrolysed COD to biomass Observed yield Total biomass yield Sludge hold-up Seed sludge Total waste sludge % % d
1 g kg m
3=kg MLSS  V % g g
1 g d
1 % kg m
3 g d
1 0.10 0.15 (acidogens), 0.18 (combined) m3 h d ¼ Q  tþMLSS 2 h, 4–12 h 4–14 days, 40–100 days 0.95 kg=kg m
3 V Volume at 30 kg/m3 (3% d.s) Plant sizes 21 22 23 V Tank volume t HRT yX;NOMINAL Nominal SRT Biogas production and recovery 24 25 26 27 28 29 30 31 32 a SG QG CG EBIOGAS Hi T Mol Pp R Biogas in soluble phase g d
1 Gas flowrate m3 d
1 Total biogas g d
1 Biogas recovery % Henry’s law constant for gas i Temperature 8K Molar weight mol g
1 Partial pressure Atm Ideal gas constant mol l
1 Based on Henze et al. (1997) and Grady et al. (1999). used in the model ignores solids leaving via the effluent stream (equation (5)) and is different to the ‘‘actual’’ SRT, which equals the mass of solids in the tank divided by the mass of solids leaving the tank (as waste sludge and as SS in the effluent) per unit time (equation (6)) (Munch and Koch, 1999). nominal SRT ¼ actual SRT ¼ msolids;TANK ðmsolids; waste sludge Þ=d ð5Þ msolids; TANK ðmsolids; waste sludge þ msolids; effluent Þ=d ð6Þ A nominal value is chosen because, first, the true biological sludge age may be very different to the calculated ‘‘actual’’ SRT within the tank (Munch and Koch, 1999) and second, the actual SRT would become nonsensical because the SS removal efficiency (and therefore effluent SS concentration) of the hydrolytic tank remained constant throughout. Recent research on a range of different pre-fermenter designs in Australia and Canada gave nominal prefermenter SRTs in the range of 4–14 days (Munch and Koch, 1999), and these are used as an upper and lower limit in the hydrolytic tank during model application. In the hydrolytic tank, it is assumed that 10% of the hydrolysed COD is converted to biomass, as it is unlikely that enzymatic hydrolysis could be completely separated from acidogenesis. Biomass production is based on an observed yield (Yobs ) of 0.15, as determined for acidogenic bacteria in laboratory experiments (Henze and Harremoes, 1983; Table 3; equation (2)). It is assumed that no methanogenesis occurs in the hydrolytic tank. Solubilised COD, along with a fraction of the generated biomass, is removed in the effluent and enters the anaerobic reactor (Table 3, equation (3)). Anaerobic treatment modelling and economics 4141 Table 3. Description of transformations modelled in the hydrolytic tank and anaerobic reactor Reactions taking place in the hydrolytic tank (1): 1. Hydrolysis of particulate COD (g O2 d
1): XCOD;1 kh;1 yX;NOMINAL;1 ð1
fBIOMASS;1 Þ 1 þ kh;1 yX;NOMINAL;1 ð1Þ Y1 ¼ SCOD;1 Yobs;1 fBIOMASS;1 ð2Þ CCOD;2 ¼ ðCCOD 2XCOD;1 Þ þ SCOD;1 þ ðY1 ð12EBIOMASS ÞÞ ð3Þ SCOD;1 ¼ kh;1 ¼ 0:019 at 108C; 0.028 at 158C; 0.039 at 208CXCOD;1 ¼ XVSS fCOD:VSS ESS;1 yX;NOMINAL;1 ¼ 4 and 14 days fBIOMASS;1 ¼ 0:10 fCOD:VSS ¼ 1:5 ESS;1 ¼ 0:60 2. Biomass yield (g VSS d
1): Yobs;1 ¼ 0:15 3. Effluent COD (g O2 d
1): EBIOMASS ¼ 0:95 4. MLSS (g d
1): XMLSS;1 ¼ 5. Sludge wasting (g d
1): XSLUDGE;1 ¼  XCOD;1
SCOD;1 fCOD:VSS  þ Y1 EBIOMASS þ ðXFSS ESS;1 Þ ð4Þ X1 yX;NOMINAL;1 Reactions taking place in the anaerobic reactor (2): 6. Effluent COD (g O2 d
1): CCOD;3 ¼ CCOD;2 2ðCCOD;2 ECOD;2 Þ ECOD;2 ¼ 0:202 LN t þ 0:17 at 108C; 0.156LN t þ 0:39 at 158C; 0.113LN t þ 0:56 at 208C 7. Biomass yield (g d
1) Yobs;2 ¼ 0:18 Y2 ¼ ðCCOD;3 2ðXCOD;io ð12ESS;1 ÞESS;2 ÞÞYobs;2 ð7Þ ESS;2 ¼ 0:044 t þ 0:37 8. Hydrolysis of particulate COD (g O2 d
1): SCOD;2 ¼ XCOD;2 kh;2 yX;NOMINAL;2 1 þ kh;2 yX;NOMINAL;2 ð8Þ XCOD;2 ¼ ððXVSS fCOD:VSS Þ2XCOD;1 Þ þ ðY1 ð12EBIOMASS ÞESS;2 ) kh;2 ¼ 0:035 at 108C; 0.050 at 158C; 0.071 at 208C yX;NOMINAL;2 ¼ 40 d at 4 h t; 100 days at 12 h t. 9. MLSS (g d
1) XMLSS;2 ¼ XVSS;SEED þ  XCOD;2
SCOD;2 fCOD:VSS  þ ðXFSS ð1
ESS;1 ÞESS;2 Þ þ ðY2 EBIOMASS Þ ð9Þ EBIOMASS ¼ 0:95 10. Sludge wasting (g d
1) XSLUDGE;2 ¼ 11. Effluent SS (g d
1): X2 yX;NOMINAL;2 XTSS;3 ¼ ðXTSS ð12ESS;1 Þð12ESS;2 ÞÞ þ ðY2 ð12EBIOMASS ÞÞ 12. Biogas production (g d
1) CG ¼ SG ¼ pp 55:6 M Hi ð10Þ ð11Þ ðCCOD;3
ðXCOD;io ð1
ESS;1 ÞESS;2 ÞÞfG Mol 103 Q ð12Þ fG ¼ 0:25 for CH4; 0.6825 for CO2 PP ¼ 0:8 for CH4; 0.06 for CO2 Mol ¼ 16 for CH4; 44 for CO2 HCH4 ¼ 2:97 at 108C; 3.40 at 158C, 3.76 at 208C HCO2 ¼ 0:104 at 108C; 0.123 at 158C; 0.142 at 208C 13. Biogas recovered (l d
1) QG ¼ R ¼ 22:414 T ¼ 283 at 108C, 288 at 158C, 293 at 208C EBIOGAS ¼ 0:90  CG
SG Mol   R TX Q EBIOGAS T273 ð13Þ 4142 P. D. Zakkour et al. Fig. 2. Layout of anaerobic treatment plant. (1) Hydrolytic tank; (2) anaerobic reactor; (3) effluent, P ¼ Pump. Partitioning of produced biomass between the sludge and effluent is controlled by a sludge hold-up rate, which assumes a figure of 0.95 i.e. 95% of produced biomass is retained within the sludge. Anaerobic reactor model Effluent from the hydrolytic tank enters the anaerobic reactor and is subject to variable COD and SS removal rates according to temperature and reactor HRT. Reduction in organic load is measured as total COD removed, determined from regression analysis using 66 individual data points derived from six different research experiments (Table 1). Data were aggregated into three temperature ranges (5138C, 13–188C, 18–208C) so that the effects of seasonal wastewater temperature on organic removal efficiency could be assessed. Data within each range were averaged, plotted against HRT, and a best-fit curve applied using least-squares regression (Fig. 2; Table 3, equation (6)). Relationships exhibited only a moderate level of significance, however, the similarity of the curve profiles suggested that those established for each temperature were likely to be valid. Moderate significance is likely due to use of data from research employing different reactor types; a factor which has been shown to affect process efficiency at temperatures above 208C (van Haandel and Lettinga, 1994). Furthermore, differences in both substrate strength and type is also likely to cause data variation (van Haandel et al., 1996). In all the experiments used; influent substrate concentration did not exceed 1000 mg COD l
1 with an average of 454 mg COD l
1 (Table 1). No consistent data for SS removal in anaerobic reactors could be found, but in general, published removal efficiencies tend to increase with HRT (Seghezzo et al., 1998; Grobicki and Stuckey, 1991). Therefore, in order to generate a working reactor model, a simple linear relationship between the range 50% and 90% for HRTs of 3–12 h is adopted (Table 3, equation (7)). However, due to the tentative nature of this relationship, SS removal rate can be changed by the model operator to any suitable value. Total particulate COD (including biomass washed out of the hydrolytic tank) is taken up into the reactor sludge at a rate determined by the SS removal efficiency. It is then subject to hydrolysis at different rates according to temperature. These were determined from a figure of 0.20 d
1 for total COD at 358C and pH 6.67 (Pavlostathis and Giraldo-Gomez, 1991; derived from Eastman and Ferguson, 1981), which, assuming a 50% decrease in activity for every 108C decrease in temperature, gave rates of 0.035, 0.050 and 0.071 d
1 at 108C, 158C and 208C respectively (Table 3, equation (8)). Inert particulate matter is taken up into sludge at the same rate, and both the unhydrolysed particulate COD and the inert solids are removed from the reactor by continuous sludge wasting, controlled by a nominal SRT in the same way as for the hydrolytic tank (Table 3, equations (9) (10)). The rate of sludge wasting is based on a relationship established between HRT and sludge age for an operational upflow anaerobic sludge blanket (UASB) in Pedregal, Columbia (van Haandel and Lettinga, 1994). The model calculates the actual SRT within the reactor (i.e. sludge wasting and effluent SS) and compares this with the figure for the Pedregal reactor which is offered as a guideline value. As the actual SRT includes solids washed out in the effluent stream in addition to discharged sludge, nominal SRTs used in Anaerobic treatment modelling and economics the model should be more than actual SRTs. This relationship gave nominal SRT values of 40 days at 4 h HRT ( to 22 days actual SRT; Pedregal reactor=22 days) and 100 days at 12 h HRT ( to 85 days actual SRT; Pedregal reactor=86 days) which were used in the model. However; it should be noted that research suggests that actual SRTs in operational reactors are unlikely to represent the true biological sludge age as part of the influent solids will pass through the reactor relatively rapidly, giving a value closer to the HRT rather than the true biological sludge age (van Haandel and Lettinga, 1994). This results in cell residence times in anaerobic reactors being considerably longer than actual SRTs, and that actual SRT is not a reliable parameter to estimate the retention time of bacterial mass in the reactor (van Haandel and Lettinga, 1994). Biomass yield and biogas production is determined by the amount of total COD digested, which is assumed to equate to influent COD minus effluent COD, less the amount of inert COD removed via entrapment into the sludge bed (i.e. removed but not digested) (Table 3, equations (7) and (12)). Biomass growth is based on a observed yield (Yobs ) of 0.18 g VSS/g COD-digested.d (VSS=volatile suspended solids) for combined anaerobic cultures in laboratory conditions (Henze and Harremoes, 1983; Table 3, equation (7)). Observed values of biomass yield in operational reactors are slightly lower (0.05–0.11 g VSS/g COD digested; van Haandel and Lettinga, 1994; Hall, 1992); however, the higher value is adopted in order to depict a worstcase scenario for sludge production. Total yield in an operational two-phase anaerobic reactor may be lower than this because some acidification will have already taken place in the hydrolytic tank, meaning that this fraction of the total combined yield may be lower (combined yield=0.15 for acidogenic bacteria, 0.03 for methanogens). The effect of changing the microbial yield is discussed further in a proceeding section. The fraction of generated biomass retained or washed out of the reactor is partitioned according to the sludge hold-up rate, set at 95% in the model calculations (N.B. This figure is higher than the SS removal efficiency as it is assumed new bacterial growth will occur on biomass flocs already adhered to the sludge bed). The model uses a starting volume of reactor seed sludge (which is assumed to consist of VSS only). Various concentrations of seed sludge have been used in pilot scale reactors ranging from 3.5 to 29 g VSS l
1 (Lettinga et al., 1981; Banik et al., 1998; Elmitwalli et al., 1999) so a mid-range value of 10 kg VSS m
3 is used for model operation. The dynamics of the seed sludge mean it is affected by biomass production, the take-up of inert particulate matter and sludge wasting (Table 3, equations (9) and (10)). No account is made for lysis and decay of anaerobic bacteria because biomass growth in the model is based on observed yields (Yobs ), and, 4143 moreover, it is assumed that there is constant replacement of the biologically active sludge through a cycle of bacterial growth, death, wash-out and wasting which serves to create a sludge mass balance. Effluent SS from the reactor is determined by the biomass production and the associated sludge holdup rate, and the SS removal efficiency for the given HRT (Table 3, equation (11)). Produced biogas is partitioned into liquid and gas phases depending on wastewater temperature (which determines the solubility of the gas) (Table 3, equations (12) and (13)). Biogas production is calculated from the stoichiometric breakdown of 1 mole of organic substrate under anaerobic conditions, which leads to the production of 3 moles of methane (CH4) and 3 moles of carbon dioxide (CO2) (Metcalf and Eddy, 1991). This gives ratios of CH4 and CO2 mass produced per amount COD digested of 0.25 and 0.6875, respectively. Concentrations of each gas entering the liquid phase is determined using Henry’s law, assuming partial pressures of 0.8 for methane and 0.06 for carbon dioxide (van Haandel and Lettinga, 1994). The partial pressure for methane assumes around 80% of the gas inside the reactor is methane, a low value for CO2 is adopted to account for the possible effects of bicarbonate and dissolved ammonia in the wastewater, which will affect the CO2 flux between liquid and gas phases (van Haandel and Lettinga, 1994). Complete recovery of produced biogas is unlikely (van Haandel and Lettinga, 1994) so the model allows the rate to be varied by the operator, although in the results given in this paper, a figure of 90% is used. RESULTS As reactor HRT increases, in-reactor sludge concentrations increase but concentrations of volatile solids in the sludge decrease slightly}the result of greater uptake of inert solids into the sludge as SS removal efficiency improves (at increasing HRTs; Table 4). Longer reactor HRTs (and the subsequent increase in reactor SRT and SS removal efficiency) cause the reactor volume to increase, resulting in a bigger in-reactor sludge mass and, consequently, a decrease in specific methane production rates, specific COD removal rates and sludge loading rates (Table 4). Increasing hydrolytic tank SRT to 14 days also increases to amount of soluble COD entering the reactor resulting in higher specific CH4 production and COD removal rates. When the hydrolytic tank SRT is increased from 4 to 14 days, reactor effluent 5-day biochemical oxygen demand (BOD5) and SS concentrations increase (at the same reactor HRT; Table 4). This can be interpreted as a result of the bacterial consortium in the anaerobic reactor being unable to effectively remove the greater mass of incoming soluble COD as a result of increased hydrolysis in the hydrolytic tank 4144 P. D. Zakkour et al. Table 4. Reactor sludge characteristics and final effluent quality at different SRTs and HRTs (same for 5000 and 50,000 pe’s) Hydrolytic tank SRT 4 4 14 14 Reactor HRT 4 12 4 12 Sludge characteristics (mean annual) Reactor solids content (kg/m3) VSS:TSS Specific CH4 prod (g CH4-COD/g VSS.d) Specific COD rem rate (g COD/g VSS.d) 0.15 0.68 0.12 0.14 15.9 0.64 0.05 0.06 15.0 0.68 0.16 0.17 15.9 0.64 0.07 0.08 Sludge load (mean annual) kg COD/m3/d g COD/g TSS/d g COD/ g VSS/d 2.4 0.14 0.28 0.8 0.05 0.09 3.0 0.15 0.27 1.0 0.05 0.09 Effluent BOD5 conc (mg/l) Winter Spring/Autumn Summer Annual average Effluent SS conc (mg/l) Winter Spring/Autumn Summer 100 75 54 76 60 43 30 44 128 95 67 96 76 54 37 56 56 57 57 14 14 15 56 57 58 15 15 16 at the longer SRT (and the same reactor HRT). As expected, seasonal temperatures also have an important effect on reactor effluent quality. Reactor effluent BOD5 concentrations are lower during the warmer periods, reflecting increased microbial activity at higher temperatures, while increased activity also leads to greater biomass production, hence slightly higher effluent SS concentrations during these periods. At a 4-day hydrolytic tank SRT, production of soluble COD i.e. that leaving the hydrolytic tank, is equal to, on-average, a 20% solubilisation of the daily SS trapped in the tank (daily SS load at 5000 pe=216 kg d
1; average hydrolysed COD=64.8 kg O2 d
1  1.5=43.2 kg SS d
1/216 kg O2 d
1=0.2), or 29% of the daily retained total particulate COD (daily XCOD load=226.8 kg
1/64.8 kg O2 d
1=0.29). Increasing the hydrolytic tank SRT to 14 days increases this figure to 58% solubilisation (or 83% of the daily retained total particulate COD). This has implications for the production of sludge in from the hydrolytic tank, where increasing the SRT from 4 to 14 days decreases tank sludge production by 74%, with only a limited effect on tank volume (an increase of 51%; Table 5). Sludge volumes inside the hydrolytic tank range from 18 m3 at 5000 pe with a 4 day SRT to 188 m3 at 50,000 pe with an SRT of 14 days (at a sludge concentration of 3%). In the anaerobic reactor, increasing HRT from 4 h to 12 h leads to an increase in in-reactor sludge production, caused by both increased microbial growth and improved SS removal at longer retention times (and as a result improved effluent quality). The configuration consisting of a 14-day reactor SRT and a 12-h reactor HRT overall produces 22% more sludge than the configuration using only a 4-day tank SRT and a 4-h reactor HRT. Nevertheless, as a result of longer retention times in the tank and reactor, effluent SS and BOD5 are decreased (Table 4). Longer retention times and greater microbial activity also has an important effect on biogas production. Increasing reactor HRT from 4 to 12 h increases biogas production by 48%, while increasing hydrolytic tank SRT from 4 to 14 days produces an additional 8% increase in biogas output, as a result of greater soluble COD input into the reactor. The configuration consisting of a 14-day reactor SRT and a 12-h reactor HRT produces 134% more biogas than a configuration using only a 4-day tank SRT and a 4-h reactor HRT. DISCUSSION Pre-treatment in a hydrolytic tank is a new development in anaerobic technology and consequently, few performance data exist against which comparisons can be made. However, some comparison with the limited research is possible, although because BNR pre-fermenter performance is generally measured by volatile fatty acids (VFA) production, comparison of the model with their performance was not possible. Production of soluble COD from removed SS in the hydrolytic tank represented, on average, 20% of daily influent SS at a 4-day SRT and 58% at a 14days SRT, which compares well to operational figures of over 50% hydrolysis of removed SS achieved in an experimental hydrolysis upflow sludge blanket (HUSB) reactor with a 2-day sludge SRT (at 198C) (Wang, 1994). Production of soluble COD as a percentage of the daily removed particulate COD load equates to 29% at a 4-day SRT and 83% at 14 days. Results from a lab-scale fermenting CSTR 785 1176 76 43.4 7443 703 800 54 43.4 7441 50 187 } 5.6 1148 41 127 } 5.6 1148 79 118 } 5.6 1150 70 80 } 5.6 1149 Sludge production (tds/yr) Average biogas production (m3/d) Power generation capacity (kWe) Power requirement (kW) Total OPEXb (£s/yr) CAPEX includes: Hydrolytic tank costs based on the cost of a circular sludge tank (WRc, 1998); Anaerobic reactor costs based on the cost of circular primary tanks (WRc, 1998) (assuming that an anaerobic reactor would use similar construction material i.e. reinforced concrete); Pump costs from WRc (1998); Gas holder, GBP £364/m3 at 1 day storage capacity}from the cost of £109,600 for a 300 m3 tank Vaughan (1999); Biogas conduits, GBP £82/m for 200 mm stainless steel lined}from McDonalds Engineering (1999); Flarestack on 5000 pe plant, approx. GBP £15,000 for a small unit, from Biogas Technology Limited (1999); CHP on 50,000 pe plant, based on the cost of GBP £800/kW installed for a small gas turbines (30 kWe ; Smith, 2000) (at an electrical conversion efficiency of 0.3; CV of 35,000/m3 for CH4). b OPEX includes: Pumping requirement based on the formula: kW ¼ 0:00272 capacity (m3/h) (head (m)/E). E ¼ 0:6 for wastewater; 0.5 for sludge. Mixer power requirements for hydrolytic tank (0.014 kW/m3). Electricity costs ¼ GBP £0.037 p/kWh, the average cost of electricity to UK manufacturing and process industries (DTI, 1998). Figures for 50,000 pe plant are minus the produced power from biogas generation, assuming the displacement of costs for bought-in electricity. 12 h (600 m ) 317.2 4 h (200 m ) 245.0 12 h (600 m ) 291.6 4 h (200 m ) 227.6 Anaerobic HRT (and volume in m ) Total CAPEXa (£’000 s) adjusted Q2 1998 a 500 1873 110 43.3 7436 410 1270 80 43.3 7433 12 h (2000 m3) 1521.8 4 h (6000 m3) 1112.0 12 h (2000 m ) 1233.7 4 h (6000 m ) 911.3 3 3 4 days (1180 m3) 3 3 14 days (119 m3) 3 4 days (118 m3) 3 3 Hydrolytic SRT (and volume in m3) Plant configuration Table 5. Capex, opex, plant size and sludge and biogas production for different anaerobic plant configurations (based on plant layout shown in Fig. 1) 14 days (1188 m3) Anaerobic treatment modelling and economics 4145 reactor operating at 16–208C (Canziani et al., 1996) showed soluble COD production rates to be much lower, with a maximum soluble COD production of around 6% at 14-day SRT. Problematically, it is unclear whether the volumes sludge in the model hydrolytic tank (>188 m3) would be possible in an operational plant. There may be problems over the sludge becoming septic, while the development of methanogens within the tank could also create problems over safety regarding the leakage of methane. Nevertheless, these figures represent upper limits, and it is possible that hydrolytic tank sludge concentrations may be somewhat higher (10–15% as opposed to 3% used in the model), which would result in the in-reactor sludge volume being considerably smaller than that calculated. The extent to which acidification (two-phase separate non-methanogenic and methanogenic digestion) in pre-treatment would be beneficial is subject to some debate (Seghezzo et al., 1998). Some preacidification of the wastewater would be beneficial; however, complete acidification has been shown to be detrimental in some respects (Lettinga and Hulshoff Pol, 1991). Methanogenesis in a phase-one treatment may create safety problems; however, it is likely that it would be severely inhibited by the low pH conceivable in a operational tank (Stronach et al., 1986). Alternatively, recent research suggests that separation of acidogenesis and methanogenesis is beneficial to the formation of a granular sludge on suspended growth reactors (Wentzel et al., 1994). In general, pre-treatment of sewage at low ambient temperatures is likely to be beneficial to the anaerobic process by avoiding excessive solids accumulation in the reactor which would, otherwise, reduce reactor performance. The model successfully achieves a general description of the hydrolysis phase; however, the benefit and extent of the role of acidification in improving the overall process requires further assessment. By drawing on experience with hydrolytic tanks and pre-fermenters it is possible that an optimum pre-treatment consisting of hydrolysis and partial acidification could be successfully developed in the near future. Furthermore, the advent of ABR technology may mean that two-stage treatment may be dispensed with, because separate stages of anaerobic digestion are able to develop inside the reactor independent of operator control, avoiding potentially complex plant operation. Anaerobic reactor performance figures (Table 4) indicates that calculated in-reactor sludge concentrations are lower than levels achieved in some operational UASB reactors treating sewage in tropical climates (van Haandel and Lettinga, 1994), and within upper and lower limits achieved in some UASB reactors treating various wastewaters (Henze et al., 1997; Fig. 3). In the former, reactor sludge concentrations range from 15.6–20.2 kg SS m
3 (2.1–17 h HRT) achieved in Pedregal, Columbia to 4146 a P. D. Zakkour et al. Fig. 3. Modelled in-reactor sludge concentrations in comparison to published values. van Haandel and Lettinga (1994); bHenze et al. (1997); ABR data from Barber and Stuckey (1999). 25–35 kg SS m
3 in plants in Cali and Kanpur, India (van Haandel and Lettinga, 1994). Reactors treating various types of waste have achieved sludge concentrations of up to 40 kg SS m
3 (Henze et al., 1997) but generally these are of the fluidised- or expanded-bed type. No data on biomass concentrations in ABR’s treating domestic wastewater could be found; however, biomass concentrations in baffled reactors treating higher strength wastes range from 4 to 30 kg VSS/l (Barber and Stuckey, 1999; Fig. 3). Generally, in UASB’s treating sewage, volatile solids to total solids ratios are in the range of 0.56–0.70 (van Haandel and Lettinga, 1994), similar to those achieved in the model (Table 4). Sludge loadings in operational reactors are in the order of 1– 3 kg COD m
3.d for sludge blanket and fixed filter plants, 1–4 for fluidized or expanded bed processes (Henze et al., 1997) and 0.85–2.17 kg COD m
3.d in ABR’s treating sewage (Orozco, 1997; Garuti et al., 1992), similar to those employed in the model where sludge loadings are 0.8–2.4 kg COD m
3.d for HRTs of 4–12 h (Fig. 4). One of the problems with the model is that sludge production rates use a high value (0.18) for Yobs , whereas operational reactors show values as low as 0.05 (see above). What is presented here is a worst case scenario-sensitivity analysis using the lower rate suggested sludge production, and associated costs, may be overestimated by as much as 12% (at a 14 day tank SRT, 12 h reactor HRT). It may be that in 2-stage anaerobic treatment sludge production rates may be different from those achieved in combined laboratory cultures. Furthermore, sewage may be significantly acidified in the sewerage network prior to arriving at the works, meaning that on-site sludge production may be significantly lower than this value because only methanogenic cultures may develop in the reactor. Modelled specific methane production rates are of the same order as those achieved in operational reactors treating sewage in hot climates where typical rates are in the range 0.02–0.19 mg CH4COD/mg VSS.d (van Haandel and Lettinga, 1994). At low specific methane production rates (50.25 mg COD/mg VSS.d) research suggests that methanogens may be subject to substrate limitation (van Haandel and Lettinga, 1994), however, in the model, low methanogenic activity may represent lower metabolic activity in response to lower reactor temperatures. From the similarity of modelled reactor sludge characteristics and the range of sludge compositions, concentrations, and microbial activity given in research on pilot and full-scale applications, it appears that model operations employed to control reactor sludge mass are satisfactory. When calculated ranges of sludge loads and methanogenic activity are compared to rates achieved in published research (Figs 3 and 4), the suggestion is that model reactor biomass is neither over- or under-loaded. Therefore, if the levels of total COD reduction employed in the model, assumed suspended solid removal efficiencies and sludge hold-up rate employed, can all be achieved in a working reactor, then it is possible that anaerobic processes can be successfully employed to treat sewage in temperate climates. Anaerobic treatment modelling and economics 4147 Fig. 4. Modelled reactor applied loading compared to published values. Source data same as Fig. 3. ECONOMIC CONSIDERATIONS Anaerobic processes have been widely considered to have as having economic advantages over traditional secondary biological treatments, although, to date examples of research outlining these advantages are limited (Alderman et al., 1998). In order to address this point, directional capital costs (capex) for the type of anaerobic plant modelled in this paper have been calculated (Table 5) using the Water Research Centre’s Technical Report (TR 61, WRc, 1998) and personal communication with experts in industry. Only tentative estimates of operating costs (opex) have been made in this work for several reasons: firstly, given the quality of final effluent modelled for the plant (Table 4), it is unlikely that anaerobic treatment could be used solely in typical western European situations, where discharge consents for STWs are generally less than 45 mg SS l
1 and 25 mg O2 l
1 for BOD5; second, the application of anaerobic processes in only a pre-treatment capacity would lead to additional energy use and sludge production from any secondary treatment employed. These in turn would be affected by both the required effluent quality, i.e. the STW discharge consent, and the anaerobic configuration used, i.e. the hydrolytic tank SRT and the reactor HRT. What is required is that the anaerobic model presented here be integrated with a model of aerobic post-treatment processes, and the costs compared to those of traditional aerobic treatments such as the biofilter and activated sludge process. Sensitivity assessment could be achieved across different discharge cons- ent from lenient (approximately 45 mg SS l
1, 25 mg O2 l
1 for BOD5) to tight (approximately 15 mg SS l
1; 10 mg O2 l
1 for BOD5; 5 mg N l
1) and the most appropriate anaerobic configuration elucidated, relative to capital costs, effluent quality, biogas and sludge production (and the subsequent operating costs). Notwithstanding these observations, some points can be made about modelled biogas and sludge production as both are likely to have significant impact on water company operations over the coming years. In the UK and across Europe, sludge treatment costs are set to rise in the near future as the full implications of the Safe-Sludge Matrix (Anon., 1998a, b) and the proposed updating of EU Directive on sludge disposal (86/278/EEC) become apparent. Furthermore, renewable power produced from sewage gas is set to form an important component of the UK Government’s renewable energy strategy (DTI, 2000), which may make biogas energy recovery more viable at marginal delivery rates. It can be seen that at 50,000 pe, power requirement is up to 40% less than that available from biogas recovery (43.3 kW compared to 110 kW}Table 5) making nearly twothirds of generated power available for sale off-site. Revenue raised from renewable generation may be higher than that calculated (using market prices for electricity) under the new UK Renewables Obligation (DTI, 2000). Similarly, potential biogas generation at 5000 pe is about twice that of bought in power (not shown) although at the recovery rate in question (about 5–7 kWhe) present economics and technology dictate that energy recovery is not viable. On-site 4148 P. D. Zakkour et al. anaerobic digestion of produced sludge may raise the rate of biogas production to levels that are technically feasible for energy recovery (around 30 kWe), thus also making the small-scale process self-sufficient in energy. CONCLUSIONS * * * * * * * Analysis of previous research using anaerobic processes to treat sewage or low-strength wastewaters at low temperatures showed that empirical relationships exist between COD removal, temperature and HRT. These relationships can be used alongside other anaerobic rate functions to develop a valid model describing the two-stage anaerobic sewage treatment. Modelled low-temperature pre-hydrolysis of the waste stream appeared valid when compared to some research results, but overestimated in terms of performance when compared to others. By drawing on experience with hydrolytic tanks and BNR pre-fermenters, an optimum stage-1 pretreatment could be developed in the near future. Functions controlling sludge characteristics appeared satisfactory when sludge performance is compared to that of operational reactors. Anaerobic processes cannot treat sewage to levels suitable for direct discharge in western Europe meaning some form of post-treatment will be needed in an operational anaerobic STW. Interpretation of cost assessments are likely to be improved by comparison of anaerobic treatment (including post treatment) with traditional aerobic techniques across a range of pe’s and discharge consents. This would allow the full cost implications of reducing aeration requirement and sludge production in a post-treatment aerobic phase to be assessed. New legislation relating to sludge disposal and renewable energy generation are likely to make anaerobic treatment more attractive in future. Acknowledgements}This work is funded as part of a CASE studentship award between the Economic and Social Research Council (ESRC) and Severn-Trent Water Ltd. Views expressed in this paper are those of the authors and not specifically of Severn-Trent Water Ltd. REFERENCES Alderman B. J., Theis T. L. and Collins A. G. (1998) Optimal design for anaerobic pretreatment of municipal wastewater. J. Environ. Eng. ASCE. 124(1), 4–10. Anon. (1998a) Agreement for sludge use on farms. ENDS Report, September 1998. Anon. (1998b) Retailers keep up pressure on spreading of untreated sludge. ENDS Report, March 1998. Anon. (1998c) The road to Bran Sands. Water Magazine (special), 13 November 1998. Banik G. C., Viraraghavan T. and Dague R. R. (1998) Low temperature effects on anaerobic microbial kinetic parameters. Environ. Technol. 19, 503–512. Barber W. P. and Stuckey D. C. (1999) The use of the anaerobic baffled reactor (ABR) for wastewater treatment: a review. Water Res. 33(7), 1559–1578. Biogas Technology Limited (1999) Personal communiction. 6, Brookside Industrial Estate, Sawtree, Cambridgeshire, UK. Tel.: +44-1487-831-701. Canziani R., Pollice A. and Ragazzi M. (1996) Design considerations on primary sludge hydrolysis under psychrophilic conditions. Environ. Technol. 17, 747–754. CIWEM (1995) Sewage Sludge Utilization and Disposal, pp. 176. Chartered Institute of Water and Environmental Management, London. DTI (1998) Digest of UK energy statistics. Department of Trade and Industry, London. DTI (2000) New and Renewable Energy: Prospects for the 21st Century. The Renewable Obligation Preliminary Consultation. Department of Trade and Industry, London. Eastman J. A. and Ferguson J. F. (1981) Solubilisation of particulate organic carbon during the acid phase of anaerobic digestion. J. Water Pollut. Control Fed. 53, 352. Elmitwalli T. A., Zandvoort M. H., Zeeman G., Bruning H. and Lettinga G. (1999) Low temperature treatment of domestic sewage in upflow anaerobic sludge blanket and anaerobic hybrid reactors. Water Sci. Technol. 39(5), 177–185. European Union Council of Ministers (1991) Council Directive of 21 May 1991 concerning urban waste water treatment. Official J. European Communities, 135/40. Garuti G., Dohanyos M. and Tilche A. (1992) Anaerobic– aerobic combined process for the treatment of sewage with nutrient removal: the Ananox1 process. Water Sci. Technol. 25(7), 383–394. Grady C. P. L. J., Daigger G. T. and Lim H. C. (1999) Biological Wastewater Treatment. 2nd ed., pp. 1017. Marcel Dekker Inc., New York. Grobicki A. M. W. and Stuckey D. C. (1991) Performance of an anaerobic baffled reactor under steady state and shock loading conditions. Biotechnol. Bioeng. 37, 344–355. Hall E. R. (1992) Anaerobic treatment of wastewaters in suspended growth and fixed film processes. In Design of Anaerobic Processes for the Treatment of Industrial and Municipal Wastes, eds J. F. J. Malina and F. G. Pohland, pp. 41–118. Technomics Publishing, Lancaster, Pennsylvannia. Hassan J. (1995) The impact of EU environmental policy on water industry reform. European Environ. 5(2), 45–51. Henze M. and Harremoes P. (1983) Anaerobic treatment of waste water in fixed film reactors}a literature review. Water Sci. Technol. 15(8/9), 1–101. Henze M., Harremoes P., la Cour Jansen J. and Arvin E. (1997) Wastewater Treatment: Biological and Chemical Processes. 2nd ed., pp. 383. Springer-Verlag, Berlin. House of Commons (1999) Impact on Industry of the Climate Change Levy: Volume II. Annex 36, Submission by Water UK. Ninth Report, House of Commons Select Committee on Trade and Industry. London, The Stationary Office. Jewell W. J. (1987) Anaerobic sewage treatment. Environ. Sci. Technol. 121(1), 14–21. Jewell W. J., Switzenbaum M. S. and Morris J. W. (1981) Municipal wastewater treatment with the anaerobic attached microbial film expanded bed process. J. Water Pollut. Control Fed. 53(4), 483–490. Anaerobic treatment modelling and economics Kobayashi H. A., Stenstrom M. K. and Mah R. A. (1983) Treatment of low strength wastewater using the anaerobic filter. Water Res. 17(8), 903–909. Lettinga G., Field J., van Lier J., Zeeman G. and Hulshoff Pol L. W. (1997) Advanced anaerobic wastewater treatment in the near future. Water Sci. Technol. 35(10), 5–12. Lettinga G. and Hulshoff Pol L. W. (1991) UASB-process design for various types of wastewaters. Water Sci. Technol. 24(8), 87–107. Lettinga G., Roersma R. and Grin P. (1983) Anaerobic treatment of raw domestic sewage at ambient temperatures using a granular bed UASB reactor. Biotechnol. Bioeng. 25, 1701–1723. Lettinga G., Roersma R., Grin P., De Zeeuw W., Hulshof Pol L., Van Velson L., Hobma S. and Zeeman G. (1981) Anaerobic treatment of sewage and low strength wastewaters. Second International Symposium on Anaerobic Digestion, Travemunde, Germany. Lord Marshall. (1998) Economic Instruments and the Business Use of Energy. Her Majesty’s Treasury, London. Mergaert K., Vanderhaegen B. and Verstraete W. (1992) Applicability and trends of anaerobic pre-treatment of municipal wastewater. Water Res. 26(8), 1025–1033. Metcalf and Eddy, Inc. (1991) Wastewater Engineering: Treatment, Disposal, Reuse. 3rd ed., pp. 1334. McGraw Hill, Inc., New York. McDonalds Engineering Limited. (1999) Personal Communication. Derby, Derbyshire, UK. Tel.: +44-1332-524100. Munch E. v. and Koch F. A. (1999) A survey of prefermenter design, operation and performance in Australia and Canada. Water Sci. Technol. 39(6), 105–112. Ofwat (1998) 1997–98 Report on water and sewerage service operating costs and efficiency. Office of Water Services, Birmingham. Oh Y. M. and Yang B. C. (1986) Anaerobic wastewater treatment using floating media. Water Sci. Technol. 24, 17. Orozco A. (1997) Pilot and full-scale anaerobic treatment of low-strength wastewater at sub-optimal temperature (158C) with a hybrid plug-flow reactor. Eighth International Conference on Anaerobic Digestion, Sendai, Japan. Pavlostathis S. G. and Giraldo-Gomez E. (1991) Kinetics of anaerobic treatment: a critical review. Crit. Rev. Environ. Control 21(5,6), 411–490. 4149 Sawyer C. N., McCarty P. L. and Parkin G. F. (1994) Chemistry for Environmental Engineering. 4th ed., pp. 643. McGraw Hill, Inc., New York. Seghezzo L., Zeeman G., van Lier J. B., Hamelers H. V. M. and Lettinga G. (1998) A review: the anaerobic treatment of sewage in UASB and EGSB reactors. Bioresource Technol. 65, 175–190. Smith R. (2000) Personal Communication. Sales Manager, Energy and Environment, BG Technology, Loughborough, England. Tel. +44-1509-282-000. Stronach S. M., Rudd T. and Lester J. N. (1986) Anaerobic Digestion Processes in Industrial Wastewater Treatment, pp. 184. Springer, Berlin. Switzenbaum M. S. (1995) Obstacles in the implementation of anaerobic treatment technology. Bioresource Technol. 53, 255–262. van der Last A. R. M. and Lettinga G. (1992) Anaerobic treatment of domestic sewage under moderate (Dutch) conditions using upflow reactors at increased superficial velocities. Water Sci. Technol. 25(7), 167–178. van Haandel A. C. and Lettinga G. (1994) Anaerobic Sewage Treatment: A Practical Guide for Regions with a Hot Climate. Wiley, Chichester. van Haandel A., Frassinetti C. C. and Lettinga G. (1996) The UASB reactor for industrial and domestic waste water treatment. In Environmental Waste Management: A European Perspective, ed. N. Horan, pp. 1–31. Wiley, Chichester. Vaughan A. (1999) Personal Communication. Severn-Trent Water Limited, The Grange, Fowke Street, Rothley, Leicestershire, LE7 7JP, UK. Tel.: +44-1283-512-555. Wang K. (1994) Integrated anaerobic and aerobic treatment of sewage. Ph.D. Thesis. Agricultural University, Wageningen, The Netherlands. Wentzel M. C., Moosbrugger R. E., Sam-Soon P. A. L. N. S., Ekama G. A. and Marais G. v. R. (1994) Tentative guidelines for waste selection, process design, operation and control of upflow anaerobic sludge bed reactors. Water Sci. Technol. 30(12), 31–42. Wilson F., Yu H., Tay J. -H. and Gu G. (1998) An empirical model for predicting the organic concentration of anaerobic filter effluents. Water Environ. Res. 70(3), 299–305. WRc (1998) Water Industry Construction Cost Estimating Manua}TR61 5th ed. TR61, Restricted circulation. Water Research Centre, UK, Swindon. Young J. C. and McCarty P. L. (1969) The anaerobic filter for waste treatment. J. Water Pollut. Control Fed. 41(5), 160.