Applied Energy 114 (2014) 798–808

Contents lists available at ScienceDirect

Applied Energy

journal homepage: www.elsevier .com/ locate/apenergy

A case study for biogas generation from covered anaerobic pondstreating abattoir wastewater: Investigation of pond performanceand potential biogas production

0306-2619/$ - see front matter Crown Copyright � 2013 Published by Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.apenergy.2013.10.020

⇑ Corresponding author. Tel.: +61 07 46 311 623; fax: +61 07 46 311 530.E-mail addresses: Bernadette.McCabe@usq.edu.au (B.K. McCabe), Ihsan.

Hamawand@usq.edu.au (I. Hamawand), Peter.Harris@usq.edu.au (P. Harris), Craig.Baillie@usq.edu.au (C. Baillie), Talal.Yusaf@usq.edu.au (T. Yusaf).

Bernadette K. McCabe ⇑, Ihsan Hamawand, Peter Harris, Craig Baillie, Talal YusafNational Centre for Engineering in Agriculture, University of Southern Queensland, Toowoomba, QLD, Australia

h i g h l i g h t s

�We report on the performance of a novel covered anaerobic pond system.� Potential biogas production was estimated using BioWin modelling software.� Ponds maintained stable operation; however, accumulation of crust was an issue.� Modelling indicated that biogas yield can be influenced by decomposition efficiency.� Configuration and operation of ponds can also impact potential biogas production.

a r t i c l e i n f o

Article history:Received 10 May 2013Received in revised form 7 October 2013Accepted 9 October 2013Available online 30 October 2013

Keywords:Anaerobic digestionWastewaterBiogasModellingBioWinSlaughterhouse

a b s t r a c t

Covered anaerobic ponds offer significant advantages to the red meat processing industry by capturingmethane rich gas as a fuel source for bioenergy while reducing greenhouse gas emissions (GHG). Thispaper presents the results of a novel-designed anaerobic pond system at an Australian abattoir in relationto pond performance and potential biogas production. Key findings in assessing the effectiveness of thesystem revealed that the covered ponds are capable of efficient wastewater decomposition and biogasproduction. The primary issue with the covered ponds at the abattoir was the build-up of fat/crust thatprevented the accurate measurement of biogas and effective use of the cover. In the absence of field bio-gas data the novel application of the computer modelling software BioWin� was carried out to simulatechemical oxygen demand (COD) removal rates and subsequent biogas yield. The unique parameter usedto fit field data was the fraction of the inlet COD due to a superficial crust which did not follow anaerobicdigestion. Field data effluent COD removal rates were matched to simulated rates predicted by BioWinwhen measured influent COD was reduced to 30%. Biogas modelling results suggest significant variationin the economic benefit of biogas energy, with the quantity of biogas potentially varying tenfold (from328 m3/d to 3284 m3/d) depending on site factors such as pond efficiency, pond configuration and oper-ational practices.

Crown Copyright � 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Anaerobic waste treatment ponds are widely adopted in themeat industry as the first stage of secondary treatment of high-strength abattoir wastewater and are an efficient means wherebythe biochemical oxygen demand (BOD) and chemical oxygen de-mand (COD) are reduced by around 90% during ideal conditions[1]. They are the preferred option for treating agricultural waste-water in Australia due to their relatively low initial cost, negligible

operating costs and simplicity of operation [2]. However, they havea couple of issues including odour emissions and the generation ofmethane, a powerful greenhouse gas (GHG). The Australian redmeat processing industry has a high exposure to carbon pricingdue to wastewater methane emissions and its use of coal for steamgeneration [3]. Consequently, the industry is beginning to installcovered anaerobic pond technology [4]. Despite higher initial infra-structure costs when compared to uncovered anaerobic ponds,covered anaerobic ponds offer significant advantages such as odourcontrol, intensification of the decomposition process and BOD re-moval, an increase in feed rate and the potential for capturingmethane-rich gas as a fuel source for bioenergy and the reductionin GHGs [4–6]. Energy obtained from the biogas can be used in aninternal combustion engine coupled to an electric generator to

Nomenclature

ARE absolute relative errorBOD biochemical oxygen demandCOD chemical oxygen demandDAF dissolved air flotationEC electrical conductivityFOG fats, oils and greasesGHG greenhouse gasHDPE high density polyethyleneHRT hydraulic retention timesML mega litre

OLR organic loading ratesORP oxidation-reduction potentialSRT solid retention timeTA total alkalinitytHSCW tonnes of hot standard carcass weightTKN total Kjeldahl nitrogenTSS total suspended solidsVFA volatile fatty acid

B.K. McCabe et al. / Applied Energy 114 (2014) 798–808 799

produce electrical power, or simply be used in boilers. Also, the li-quid fraction from this process can be used as liquid fertilizer [7].

Knowledge regarding the design of these ponds and the quan-tity and quality of biogas captured remains largely undetermined.Modelling has previously been used to predict biogas productionafter calibration. In a study by Martinez et al. [8] modelling wasused to simulate slaughterhouse effluent waste degradation andmethane generation after it was demonstrated that the modelshowed an accurate reproduction of the behaviour of an anaerobicdigester. Modelling the biogas production process can be used asan indication of the process performance. This may identify actionsfor better control of the process operation that positively impactthe biogas yield [9].

This paper firstly provides contextual background informationand proceeds to report on the performance of a novel designedcovered anaerobic pond system installed at an Australian abattoirin relation to wastewater treatment and biogas production. Thestudy also reports on the novel application of BioWin� computermodelling software to simulate biogas yield in a field-based situa-tion and provides an economic assessment of biogas recovery anduse based on these modelled results.

2. Overview of operation and performance of anaerobic pondstreating abattoir effluent

Red meat processing produces wastewater with a high pollu-tant load consisting of paunch, manure, fats, oils and greases(FOGs), and uncollected blood. These components contribute to ahigh-strength waste which must be treated to reduce the BOD,COD, FOGs and total suspended solids (TSS) [10]. FOGs are largecontributors to BOD and COD and while FOGs have the potentialto produce large quantities of methane [11], their recalcitrant nat-ure generally results in a number of problems [12]. Some of theproblems include: clogging of pipes; foul odour generation; adhe-sion to the bacterial cell surface and reducing their ability to treatwastewater; and flotation of sludge and loss of active sludge[13,14]. FOGs also tend to accumulate on the surface of ponds toform a recalcitrant scum layer or ‘crust’ [15,16] which can hamperattempts to accurately measure biogas. However, primary treat-ment systems such as dissolved air flotation (DAF) units withchemical treatment are capable of reducing FOGs by up to 89–98% [5].

Anaerobic digestion is said to be working optimally when theacid formation phase (hydrolysis and acidogenesis) and the meth-ane production phase (acetogenesis and methanogenesis) occursimultaneously in dynamic equilibrium [17]. Stability of theanaerobic process is difficult to maintain because a balancefavourable to several microbial populations is necessary and thecomparatively stable nature of the acid formers and the fastidious

nature of the methane formers creates a biosystem that is proneto upset as a result of shock loads or temperature fluctuations[18]. Therefore, for the design of an anaerobic pond to performoptimally, it must be based on the limiting characteristics ofthese microorganisms. Pond oxidation–reduction potential(ORP), temperature, NH3 concentration, pH, volatile fatty acid(VFA) to total alkalinity (TA) ratios (VFA/TA) are all parameterswhich are indicative of pond performance and should be moni-tored [1]. However, criteria for anaerobic pond design are poorlydefined and no widely accepted overall design equation exists[19]. Previously, pond construction criteria for the red meat pro-cessing industry has been derived from other industries, and thishas resulted in pond designs which have not necessarily beensuitable. Design is typically based on organic loading rates(OLR) and hydraulic retention times (HRT) from pilot plants andobservations of existing pond systems [19]. Generally, the desiredgoal is to achieve significant reductions in wastewater organicload with the least HRT possible [15]. Anaerobic ponds are de-signed based on an OLR to promote sedimentation of wastewatersolids and efficient anaerobic digestion to biogas. Compared withanaerobic digesters, anaerobic ponds are designed for relativelylow OLRs [20]. Overloading of ponds has the undesirable effectof accumulating inhibitory substances which inhibit biogas pro-duction and reduce biogas yield. As a general rule, an increasein organic loading must be balanced by an increase in HRT toachieve equivalent treatment efficiency of the wastewater [21].

There is currently a lack of knowledge within the Australianred meat processing industry regarding the design and operationof anaerobic ponds and upgrading these to covered anaerobicponds to minimise GHG from wastewater treatment operations.Also, there is a clear lack of published literature which detailsbiogas production using this technology with the recoverablequantity and quality of biogas remaining largely unclear. In theabsence of meaningful field gas measurements it is difficult to as-sess the feasibility of using covered anaerobic ponds to generatebioenergy. Computer modelling software has become widelyadopted in wastewater engineering over the past two decades.Although evolved principally as a research tool they are now usedmore for design and optimisation of wastewater treatment plants[22]. BioWin is a Windows based computer simulation modelwhich is increasingly used to predict anaerobic digestion pro-cesses and subsequent biogas yield [23]. It is used primarily tosimulate wastewater treatment for domestic sewage and therehas been no application to meat processing waste to date. How-ever, anaerobic ponds that treat abattoir effluent utilise the samecomplex microbiological processes responsible for the anaerobicdecomposition of domestic wastewater. Thus there is great scopeto apply the tool in this situation given the uncertainty surround-ing accurate biogas measurements due to crust and solidaccumulation.

800 B.K. McCabe et al. / Applied Energy 114 (2014) 798–808

3. The case study of Churchill Abattoir: novel pond design andcover construction

Churchill Abattoir Pty Ltd is a medium-sized meat processingfacility located in South East Queensland, Australia. The abattoirslaughters and processes around 3000 head of cattle per weekresulting in around 660 tonnes of hot standard carcass weight(tHSCW) per week.

In 2009 the abattoir started to investigate the use of coveredanaerobic ponds to reduce offensive odours and greenhouse gasemissions with the subsequent capture of methane for bioenergy.This prompted a re-design of the wastewater treatment system.McCabe et al. [24] provides a detailed background of the novelpond design and cover construction. Briefly, 5 smaller anaerobicponds were constructed, each 50 m in length, 20 m in width and5 m in depth, with an effective volume of 2.2 ML each. This designwas selected for two main reasons, namely manageability for des-ludging ponds and ease of removing and applying covers. A newfloating cover design was proposed whereby covers were attachedto a floating raft or truss which holds the cover off the surface ofthe pond. HDPE pipe (100 mm) was used to form the skeleton ofthe raft and these pipes were filled with expansive foam to stiffenthe structure and aid in floatation and HDPE mat was used as thecover material.

4. Methods

4.1. Monitoring schedule and wastewater characterisation

Fig. 1 illustrates a schematic of the 5 pond layout. A total of43 weeks of sampling was performed during two stages on pondsA, B and E. Stage one consisted of 19 weeks and stage two24 weeks. Sampling was conducted twice-weekly at the com-mencement of the sampling campaign for 9 weeks and thenweekly thereafter. The monitoring schedule provided in Table 1lists the parameters that were measured as part of the monitoringprotocol. Both on-site and laboratory analysis was conducted.Wastewater samples for laboratory analysis were collected andanalysed by Australian Laboratory Services (ALS) group (Brisbane,

sample ports

Wastewater from plant

Save-all

Anaerobic pond A Anaerobic pond B

Anaerobic pond C Anaerobic pond D

Anaerobic pond E Facultative anaerobic pond 2

Aerobic pond 3

Irrigated to crops

Fig. 1. Schematic of pond layout indicating sampling points and flow ofwastewater.

Australia). Measured parameters included COD, BOD, TSS, FOG,ammonia as nitrogen (NH3–N), total Kjeldahl nitrogen (TKN), alka-linity, and volatile fatty acids (VFA). On-site wastewater analysisinvolved the measurement of wastewater temperature, pH, ECand ORP using a YSI professional plus field logger. Biogas was ana-lysed for methane, carbon dioxide, oxygen, and hydrogen sulphidecontent, as well as the remaining balance using a Geotechnicalinstruments GA2000 landfill gas analyser which is capable of mea-suring methane, carbon dioxide, hydrogen sulphide and oxygen towithin 98%.

Fixed ultrasonic flow meters (Dalian Zerogo RV-100F) were at-tached to the external surface of the inflow pipes to ponds A and B.Flow meter data was logged at a frequency of one minute andstored on a CR1000 data logger. Sampling ports were installed asshown in Fig. 1 and included inlets to pond A and B, and the outletsof ponds A, B and E. Ponds A and B were the primary focus of mon-itoring since these serve as the two primary ponds receiving allincoming wastewater. These two ponds run in parallel and feedinto a further series of three anaerobic ponds; C, D and E. PondsC, D and E function as a single unit with bidirectional flow betweenponds, with flow direction dependent on pond level, although flowis generally unidirectional C ? D ? E ? pond 2. The outflow ofPond E was also monitored to further understand the operationof the novel anaerobic pond system as a whole.

4.2. Biogas simulation

Due to the difficulties encountered in measuring biogas produc-tion at the site, dynamic wastewater treatment modelling usingthe software BioWin (EnviroSim Associated LTD, Canada) wasundertaken to estimate biogas production. BioWin is a MicrosoftWindows-based simulator which is used in the analysis and designof wastewater treatment plants. BioWin uses a general ActivatedSludge/Anaerobic Digestion (ASDM) model which is referred to asthe BioWin General Model. BioWin is interface software which re-quires input data to carry out the simulation. Parameters such asflow rate, total COD, TKN, total P, total N, pH, alkalinity, inorganicS.S., Ca, Mg, and DO are main characteristics of the wastewater re-quired by BioWin. These values (either constant or variable withtime) are presented in a hypothetical setting which then re-createsthe anaerobic digestion process by the software. To enhance theprediction of the software, the wastewater fractions such as readilybiodegradable, non-colloidal slowly biodegradable, unbiodegrad-able soluble and particulate are requested. Moreover, process ki-netic parameters such as hydrolysis rate with stoichiometricparameters are essential for high accuracy predictions. Although,kinetic and stoichiometric parameters are important input datafor the software, BioWin includes default values for these parame-ters which have previously been found reliable in this study [25].BioWin contains two operational modules which include a steadystate module and an interactive dynamic simulator. The steadystate module is used for simulating systems based on constantconditions while the dynamic simulator allows the user to changetime varying inputs or changes in operational strategy which

Table 1Sampling history for ponds A, B and E.

Pond Effluent Numberofsamples

Parameters

A (uncovered) Inflow andoutflow

16 TSS, alkalinity, NH3–N,TKN, FOG, COD, BOD, VFA.pH, EC, ORP, temperatureB (covered) Inflow and

outflow40

E (uncovered) Outflow 17

Fig. 2. Anaerobic pond configuration at Churchill Abattoir.

B.K. McCabe et al. / Applied Energy 114 (2014) 798–808 801

reflect real conditions. Thus, dynamic modelling using BioWin wasdeemed an appropriate tool for simulating the behaviour of thecovered anaerobic ponds in this study.

To simulate the anaerobic ponds at Churchill Abattoir, BioWinwas first calibrated against measured data from the field monitor-ing programme. Data sets used in the calibration process includedeffluent COD concentration and TSS concentration. The calibrationprocess was conducted using a relatively complete and paralleldata set for ponds B and E over 150 days. Data for Pond B was usedto test the skill of BioWin simulating a unit process while data forpond E was used to test the skill in modelling the whole wastewa-ter system. Fig. 2 shows the configuration of the 5 ponds at Chur-chill Abattoir as represented by BioWin interface window.

5. Results and discussion

5.1. Decomposition efficiency

The average flow data into ponds A and B is given in Tables 2and 3. The average OLR for ponds A and B was 2.3 kgCOD m�3 d�1

and 3.4 kgCOD m�3 d�1 respectively with an average HRT of be-tween 2 and 3 days. CSIRO [4] provides a recommended OLR of

Table 2Removal efficiencies of the five pond system during stage 1 sampling.

Parameter Number of samples Average

Pond AFlow rate (m3/d) 79,200 503.29Influent COD (mg/L) 15 7442.00Effluent COD (mg/L) 23 2885.30Influent BOD (mg/L) 15 3402.67Effluent BOD (mg/L) 24 1318.39Influent TSS (mg/L) 15 3235.00Effluent TSS (mg/L) 23 1496.09Influent FOG (mg/L) 15 491.87Effluent FOG (mg/L) 24 111.30

Pond BFlow rate (m3/d)a 142,560 658.44Influent COD (mg/L) 27 7051Effluent COD (mg/L) 27 2696.30Influent BOD (mg/L) 27 3273.04Effluent BOD (mg/L) 27 852.26Influent TSS (mg/L) 27 2990.63Effluent TSS (mg/L) 27 1196.15Influent FOG (mg/L) 27 618.74Effluent FOG (mg/L) 27 95.85

Pond E (Five-pond system)Effluent COD (mg/L) 10 1155.20Effluent BOD (mg/L) 10 188.80Effluent TSS (mg/L) 10 704.10Effluent FOG (mg/L) 10 29.40

a % Total reduction for five-pond system.

0.05–0.08 kgCOD m�3 d�1 with a HRT of 20–40 days. Both BODand COD loading rates are outside these recommended operatingparameters which are expected given the short HRT for each ofthe ponds. If, however, the 5 ponds are considered as an integratedwaste treatment system, ponds C, D and E are operating within de-sign parameters. While the system appears to be operating withindesign parameters there are components which are operating out-side the criteria. This has obvious implications for the maintenanceof the pond system, particularly in regard to solids management.

During the stage one sampling period (Table 2) pond A achieved73% COD removal while pond B achieved a lesser removal of 53%.The total% COD removal of the 5-pond system was 84% based onthe outflow of pond E. The lower COD removal of pond B reflectsthe OLR of this pond which was calculated at an average of2.275 kgCOD m3 d, which was approximately double that of pondA during the same time period (1.03 kgCOD m3 d). The COD re-moval efficiency of pond B was not detrimentally affected whenpond A was taken off line for desludging at the end of stage onemonitoring. The higher OLR of pond B during the second samplingperiod did not result in a corresponding decrease in solids removalefficiency with the% COD removal maintained at 59% (Table 3). Asimilar trend exists for BOD removal for the 3 ponds over the sametwo sampling periods.

Standard deviation Range Av% reduction

11.65 10.04–899.382678.12 2630–12,1002220.68 798–9150 73.221109.87 1410–51501203.00 188–4610 74.951353.16 1370–68301568.08 292–5640 76.25

259.52 73–962816.54 <5–4080 85.26

15.78 15.16–1207.682895.10 1040–12,100

870.97 1680–4710 53.471461.68 163–7020

184.13 575–1500 62.191573.18 457–6870

755.13 567–4020 39.79509.83 5–2110

89.34 21–520 89.25

–a

265.98 672–1660 83.6267.49 78–302 94.23

421.36 138–1700 76.4626.3 8–98 95.25

Table 3Removal efficiencies of the five pond system during stage 2 sampling.

Parameter Number of samples Average Standard deviation range Av% reduction

Pond BFlow rate (m3/d)a 89,280 1019.30 820.04 21.57–3028.04Influent COD (mg/L) 13 9216.15 5978.34 4330–24,200Effluent COD (mg/L) 13 2898.92 1024.00 836–5020 58.89Influent BOD (mg/L) 13 5087.69 6131.70 1060–24,500Effluent BOD (mg/L) 13 714.62 436.91 246–1920 73.49Influent TSS (mg/L) 13 3874.62 1533.58 1760–6130Effluent TSS (mg/L) 13 1988.77 1292.53 824–5360 35.11Influent FOG (mg/L) 13 1388.23 1310.21 136–4570Effluent FOG (mg/L) 13 91.54 38.95 23–167 83.39

Pond E (Five-pond system) –a

Effluent COD (mg/L) 7 900.07 618.74 126–2150 72.94Effluent BOD (mg/L) 7 88.79 33.33 5–130 77.67Effluent TSS (mg/L) 7 413.64 220.47 210–867 77.89Effluent FOG (mg/L) 7 27.71 45.75 49–143 91.98

a % Reduction for five-pond system.

802 B.K. McCabe et al. / Applied Energy 114 (2014) 798–808

It was observed that both COD and BOD removal efficiencies(particularly the latter) decreased over the two sampling periodsfor the 5-pond system owing to the accumulation of crust andsludge over this time. The COD and BOD of outflow samples ofpond E at the end of the second sampling period are 73% and78% respectively. This compares to the earlier efficiencies of CODand BOD removal of 84% and 94% at the end of the first samplingperiod.

The suspended solids removal was more efficient for pond Athen pond B with 76% and 40% recorded respectively over the firststage of monitoring. This probably contributed to the increase insludge build up that occurred in pond A leading to its subsequentdesludging at the end of its first 18 months of operation. The over-all TSS removal of pond B was low over both stage one and stagetwo monitoring periods and may indicate that the short HRT didnot permit adequate sedimentation of wastewater solids.

The removal of FOGS by the 5-pond system is 95% during stageone (pond E data, Table 2) and is generally maintained throughoutstage two at 92% (pond E, Table 3). The increase in OLR for pond Bduring the second stage of monitoring marginally decreased theFOG removal efficiency from 89% to 83% and this would have con-tributed to the slight decrease in FOGs removal efficiency of thewhole system at this time.

Figs. 3a and b shows the fat accumulated on uncovered pond Aand covered pond B respectively. Both ponds A and B accumulatedapproximately 1 m of crust over the 2 year operation since thesetwo ponds received the majority of organic load. This crust accu-mulation meant that a reduction in the effective volume of the

Fig. 3a. Appearance of crust accumulation on uncovered pond A.

Fig. 3b. HDPE cover on pond B. Note the presence of the thick crust.

pond occurred over time which could impact on the bioconversionefficiency of the two ponds.

5.2. Biogas quality

The quality of biogas based on the constituents methane (CH4),carbon dioxide (CO2), oxygen (O2) and hydrogen sulphide (H2S)produced from covered pond B are shown in Figs. 4a and b. It isimportant to note that a spike in OLR occurred during April havingthe effect of lowering CH4 and CO2 at this time. The levels of CH4

and CO2 returned to nominal levels after the shock loading event.

Fig. 4. Pond B biogas major constituents (a) and minor constituents (b) during stage 1 and 2 sampling period.

B.K. McCabe et al. / Applied Energy 114 (2014) 798–808 803

Average CH4 content was 52%, while CO2 and O2 were 22% and 3%respectively. The levels of O2 should be negligible; however, thecover did become compromised at various stages of the monitoringand did not achieve an air tight seal. Average H2S levels over thesame period were 686 ppm. To compare field results 3 sampleswere sent to analytical labs (SGS, Sydney, Australia). These resultsshow that the CH4 values ranged from 59% to 62% with CO2 and O2

levels averaging 37% and 0.9% respectively. H2S levels ranged be-tween 47 and 196 ppm.

5.3. Wastewater simulation

Wastewater decomposition efficiency was simulated using Bio-Win by implementing a step-wise reduction in influent COD andadjusting this to best match the measured effluent COD. The influ-ent from the covered primary anaerobic pond (pond B) was usedfor modelling purposes. In order to justify the reduction of theCOD input into BioWin, a sensitivity analysis was conducted forthe stoichiometric and the kinetic parameters of the software. Itwas found that altering the parameters in BioWin did not improvethe results between simulated and measured data when the modelwas operating under no COD reduction. Modified input COD dataestablished that a 30% reduction of influent COD was the best fitfor the model. In practice this means only 30% of the influentCOD was taking part in the anaerobic digestion process. Theremaining 70% of the COD can be accounted for through the accu-mulation of fat and other undigested material (such as paunch) atthe top of the pond and undigested sludge at the bottom whichwas consistent with observations at the site.

This first part of the modelling process demonstrated that Bio-Win was able to accurately simulate a single anaerobic pond de-spite the severe fluctuation in both the inflow rate and influentwater composition. Two methods were used to show the agree-ment between the measured and predicted data. Two trends sim-ilar to the predicted data were plotted. They represent a value of20% above and below the predicted data and represents the agree-ment between the predicted and measured data enclosed by ±20%of the predicted data values. Absolute relative error (average)(ARE) was also used to show the agreement between the measured

and predicted data. The equation below was used to estimate theARE [26];

ARE ¼ 1N�XN

i¼1

jðmi � piÞjmi

� 100%

where mi is the measured value of the output variable, pi predictedvalue of the output variable and N number of the observations. Dueto the high complexity of the process, and as stated by otherresearchers [26], an average relative error for the measured andpredicted data of 7–15% is sufficient for indication of correct dy-namic calibration. In Liwarska’s case [26], it is worth mentioningthat monitored data was fitted to the simulated values over a periodof a few hours. In the current study, monitored data was fitted withpredicted BioWin data over a three month period, resulting in abso-lute relative errors of between 14% and 21%. The slightly higher er-ror in comparison to Liwarska’s case was attributed to highfluctuation in the characteristic of the wastewater, the environmentcondition around the ponds, thick crust formation and the long per-iod of sampling. In light of these conditions it is fair to suggest thatan average error of 21% is quite reasonable.

The BioWin prediction of COD effluent from pond B is shown inFig. 5a. Predicted and measured COD results were graphed againsteach other where the absolute relative error was found to be 14%.This was considered to be very good, particularly when consideringthe high fluctuation of the flow rate and varying composition ofinfluent to the pond. BioWin simulations were also able to demon-strate similar skill with data collected at different dates and pondtemperatures as shown in Fig. 5b. In addition to COD BioWin wasable to show skill in simulating the effluent TSS. There was goodagreement between measured TSS and BioWin prediction at twosampling period with an absolute relative error of 18% as shownin Figs. 6a and b.

The next stage of the modelling process focused on the ability ofBioWin to simulate the 5-pond system. The 5-pond system wasconfigured in BioWin and measured effluent COD from Pond E(which represents the final outflow) was plotted against BioWinpredictions of COD. As shown in Fig. 7a the measured COD of theoutlet wastewater from Pond E again correlates very well withthe BioWin predictions with an absolute relative error value of

Fig. 5. Measured and simulated effluent COD from Pond B for (a) stage 1 and (b)stage 2 monitoring periods. Fig. 6. Measured and simulated effluent TSS from Pond B for (a) stage 1 and (b)

stage 2 monitoring periods.

804 B.K. McCabe et al. / Applied Energy 114 (2014) 798–808

16%. In addition to the simulations of COD, TSS measured at theoutlet of pond E was also compared against the data predicted byBioWin and is shown in Fig. 7b. Simulation of TSS gave an absoluterelative error value of 21%. This analysis and interpretation demon-strates clearly the ability of BioWin to simulate both a single ele-ment and the whole system of wastewater treatment at ChurchillAbattoir.

Further validation of the model is provided by comparing mea-sured biogas quality (% methane) obtained from pond B with pre-dicted BioWin results over a period of approximately 3 months.Predicted and measured values were graphed against each otherwhere the absolute relative error was found to be 17% (Fig. 8). Thislevel of correlation gives further support for the validity of the soft-ware in predicting biogas quality and provides additional confi-dence in predicting pond efficiency.

5.4. Biogas production potential

Potential biogas production was estimated by simulating theanaerobic processes within the ponds over 364 days to representannual biogas production. Measured data from the monitoring pro-gram including flow rates and COD were used as inputs into theBioWin simulations. To assess biogas production for the currentsystem and operation practices, two scenarios were modelled. Sce-nario 1 represents a COD reduction efficiency of 85% (default set-tings within BioWin) while scenario 2 represents a CODreduction efficiency of 30%.

The data contained in Table 4 is a summary of the modelled re-sults for potential annual production of biogas under the current

Fig. 7. Measured and simulated effluent COD (a) and TSS (b) from Pond E duringstage 1 monitoring period.

B.K. McCabe et al. / Applied Energy 114 (2014) 798–808 805

configuration (refer to Fig. 2) and management of ponds at theabattoir under an ideal circumstance (i.e. scenario 1) where theefficiency of the pond is high and governed by 85% COD reduction(default within BioWin). The data contained in Table 4 includesminimum, maximum and average biogas production during thisperiod. Total annual biogas production of 431,404 m3 was foundby summing the simulated daily gas production of the 5-pondsystem.

The calibration of the model however indicated that the pondsare likely converting only 30% of the influent COD via anaerobicdigestion. The annual gas production at Churchill Abattoir is there-fore most likely to be similar to the data presented in Table 5 (i.e.scenario 2) which is based on a 30% reduction in COD resulting in asignificantly lower annual biogas yield of 120,000 m3 (equivalent

to 0.0298 m3/m3/d). This compares with the study conducted bySafley and Westerman [27] which measured 0.03–0.5 m3/m3/dand is illustrative of the large range of biogas quantities whichcan be produced by anaerobic ponds.

Modelling suggested other factors, despite ideal digestion (80–90% COD reduction) are likely to significantly affect the process.These include the HRT, solid retention time (SRT), temperatureand flow rate. The current design of the ponds through the model-ling process was assessed to be operating at 30–40% efficiencywhen combining these factors. It is important to note that this rep-resents the final production yield of the ponds. It is reasonable toexpect that over the lifetime of the ponds the biogas productionyield will initially be much higher due to a greater useable volumeof the pond (i.e. before crust and sludge accumulation). The finalproduction yield of the pond could in fact be enhanced throughthe routine removal of crust and sludge throughout the lifetimeof the ponds.

5.5. Alternative configurations and operational options

Previous lab based studies by Borja et al. [28] suggest that in-creased biogas production can be achieved when the variablesaffecting the performance of the anaerobic process (i.e. ponds)are better controlled. To examine these possibilities and by exploit-ing the functionality of BioWin, alternative configurations andoperational options were simulated to identify the impact on bio-gas yield. As an example, a new configuration is shown in Fig. 9which includes the addition of a clarifier to recycle the activatedsludge leaving the system. Simulation modelling of this systemby BioWin did show a significant improvement in the performanceof the ponds by increasing biogas yield.

Table 6 presents the biogas production from an alternative pondconfiguration (scenario 3) where most of the inlet COD is consid-ered as degradable materials (ideal scenario presented earlier).The potential biogas production is around 3284 m3/d. Even byreducing the amount of degradable COD to 30% (likely scenarioat Churchill presented earlier) the results shown in Table 7 (sce-nario 4) demonstrates significant improvement in biogas produc-tion (i.e. 572 m3/d from 328 m3/d). These figures indicate thepotential to significantly increase biogas production by a relativelyminor change in the configuration of the treatment system. In thisinstance the last pond at Churchill (pond E) could be used as a clar-ifier pond to recycle the activated sludge back into the top of thesystem.

5.6. Cost analysis

Biogas has an energy content of (6.0–7.5) KW h/m3 which iscomparable to coal seam gas (9.9 KW h/m3), making it a veryimportant source of energy [29]. One of the most likely optionsfor biogas capture and use at Churchill Abattoir is via a combinedheat and power generation plant to offset electricity and heatingdemands at the site. Based on the BioWin modelling results theamount of biogas produced from the site is 120,000 m3/year(328 m3/d). Each cubic metre of biogas contains the equivalent of6 kW h or 21.6 MJ of energy. However, when biogas is convertedto electricity, via a biogas powered electric generator, approxi-mately 35% of the total energy is converted to electricity due tothe efficiency of the generator. The remainder of the energy is con-verted into heat, some of which can be recovered for heating appli-cations. It is assumed that 35% of the total energy can also berecovered for low grade heating purposes [30].

5.6.1. Energy offsetsTable 8 presents the amount of useable energy for the site

produced from biogas based on the assumptions described above.

Fig. 8. Measured methane content in biogas vs. BioWin prediction, ARE 17%.

Table 4Scenario 1 – Total and individual Biogas production from the ponds at Churchill plant(Ideal: 85% efficiency).

Pond Biogasproduction(m3/year)

Production (m3/d)

Min Max Average

Pond A 130,639 119 556 362Pond B 136,821 37 742 380Pond C 58,264 61 228 161Pond D 52,604 50 212 146Pond E 48,344 49 184 134Total biogas production m3per year

(Five-pond system)431,404 1183

Table 5Scenario 2 – Total and individual Biogas production from the ponds at the Churchillplant (30% efficiency).

Pond Biogasproduction(m3/year)

Production (m3/d)

Min Max Average

Pond A 48,881 59 103 94Pond B 54,822 24 142 111Pond C 21,200 58 79 49Pond D 19,552 18 71 45Pond E 14,671 4 90 29Total biogas production m3per year

(Five-pond system)120,000 328

Fig. 9. Alternative ponds’ configu

806 B.K. McCabe et al. / Applied Energy 114 (2014) 798–808

Scenario (1) highlights the likely biogas production from the cur-rent operation of the ponds (described earlier) and the opportunityto exploit biogas based energy on the site. Energy savings based onother BioWin modelling scenarios are also presented in Table 8.

Assuming a conservative electricity price of $0.1/kW h, electric-ity costs on site can be offset by $25,200 per annum. Given a coalprice of $88/tonne, the total cost of coal is offset by $2,957 per an-num due to the recoverable heat energy from the biogas powergeneration process. Combined, the total energy costs at Churchillcan be offset by $28,157 (scenario 1) under current operating con-ditions. It is important to note however that the potential is muchgreater depending on the operational configuration and perfor-mance of the ponds. Based on the other BioWin modelling results,energy costs could be offset by $49,040 by changing the operatingconfiguration of the ponds at the current efficiency (scenario 2).

5.6.2. Economic assessment of biogas recovery and useA rudimentary economic analysis was undertaken to assess the

feasibility of biogas energy recovery and use at Churchill and forthe scenarios described above. The economic assessment wasbased on a simple payback period (SPP) approach for a combinedheat and power generation plant. The assessment was based onthe following assumptions:

� The capital cost of the generation equipment plusadditional costs including design, planning and projectmanagement is $1,200/kW.

ration at Churchill Abattoir.

Table 6Scenario 3 – Total and individual Biogas production from the ponds at the Churchillplant (ideal: 85% efficiency; alternate configuration).

Pond Biogasproduction(m3/year)

Production (m3/d)

Min Max Average

Pond A 311,510 198 1516 855Pond B 375,046 544 1630 1030Pond C 267,285 435 1091 734Pond D 242,162 383 1007 665Pond ETotal biogas production m3per year

(Five-pond system)1,209,139 3284

Table 7Scenario 4 – Total and individual biogas production from the ponds at the Churchillplant (30% efficiency; alternate configuration).

Pond Biogasproduction(m3/year)

Production (m3/d)

Min Max Average

Pond A 66,952 216 270 184Pond B 64,169 128 302 176Pond C 40,849 49 278 112Pond D 36,115 42 246 100Pond ETotal biogas production m3per year

(Five-pond system)209,000 572

Table 8Energy saving at Churchill Abattoir plant.

Scenario Biogas(m3/year)

Useableenergyfrombiogas

Energyamt.(GJ/year)

Energyamt.(kW h)

Energysavings($)

Energyoffset

1 431,404 Electricity 3261 905,948 $90,595 Electricityheat 3261 905,948 $10,630 coal

2 120,000 Electricity 907 252,000 $25,200 Electricityheat 907 252,000 $2957 coal

3 1,209,139 Electricity 9141 2,539,192 $253,919 Electricityheat 9141 2,539,192 $29,793 coal

4 209,000 Electricity 1580 438,900 $43,890 Electricityheat 1580 438,900 $5150 coal

B.K. McCabe et al. / Applied Energy 114 (2014) 798–808 807

� The generator required is based on 100 kW per 40 m3/h ofbiogas.

� Other lifetime costs for Operation and Maintenance (O&M)is half of the initial capital cost.

The results from the SPP analysis are presented in Table 9 whichindicates a payback on the investment including an allowance forlife time O&M costs of 2.2 years. As a general guide and for thisexercise an investment with a payback of less than 3 years is con-sidered to be an attractive proposition for the Meat ProcessingIndustry (pers. comm. Mike Spence June 2012). Given the analysisis based on proportional costs and returns relative to the quantityof biogas produced the SPP for both scenarios are the same. The

Table 9Simple payback period (SPP) based on the investment in construction of a combined heat

Scenario Biogas (m3/h) Power generator size (kW) Capital cost

1 49 123 $147,8752 14 34 $41,0003 137 342 $410,5004 24 60 $71,500

financial proposition however will be significantly different overthe lifetime of the investment for each scenario and requires amore detailed analysis.

6. Conclusions

The purpose of this study was to gauge covered anaerobic per-formance in terms of both waste treatment efficiency and subse-quent biogas production. Observations from this work indicatethat the successful design and operation of the covered anaerobicponds is highly sensitive to the inclusion of FOGs in the effluentstream entering the ponds. This problem is not unique to ChurchillAbattoir and is a systemic problem in the Australian red meat pro-cessing industry which hinders the successful uptake of technolo-gies such as covered anaerobic ponds.

This study reports on the novel application of computer model-ling using BioWin software to simulate COD removal rates andsubsequent biogas yield. The application of wastewater modellingusing BioWin in this study has provided some initial insights intohow unbiodegradable portions of COD can affect predicted waste-water treatment and subsequent biogas yield. Due to the highstrength nature of abattoir wastewater, the accumulation of crustson anaerobic ponds can result in limited ability to accurately ob-tain biogas measurements. The simulated results provide an initialindication that BioWin may be a useful tool in determining biogasyield in complex systems where it is difficult to obtain accuratedata. Once calibrated, BioWin was found to closely predict mea-sured data, despite the severe fluctuation in both inlet flow andwater quality parameters. In this instance BioWin was able to sim-ulate the behaviour of the anaerobic ponds and simulated an aver-age biogas yield of 328 m3/d. The modelling has shown that thetotal energy cost at Churchill Abattoir can be offset by $28,157 un-der current operating conditions. This includes electricity costs of$25,200 and cost of coal of $2,957 per annum. Modelling also sug-gests this can be significantly increased (by a factor of ten) withrelatively minor changes to the system configuration andoperation.

In terms of industry benefits BioWin has the ability to be a use-ful tool in the analysis of pond performance, that is, varying the de-fault parameters in BioWin has the potential to determine howefficient the anaerobic pond is operating. For example, reducingthe hydrolysis rate’s default value in order to match the measureddata with BioWin prediction is an indication of low efficiency ofthe anaerobic pond. This can be related to the pond design, lowdegradability of the influent waste, and/or to microbiological as-pects. The actual cause can then be determined via further simula-tion through altering other related parameters to aid in optimisingthe amount of potential biogas produced.

Acknowledgements

The work described in this article was fully supported by theAustralian Meat Processor Corporation (AMPC) and Meat and Live-stock Australia (MLA). Support given by Mike Spence (CompanyEngineer, Churchill Abattoir) is gratefully acknowledged.

and power generation plant.

($) O&M ($) Total costs ($) Offset ($) SPP (years)

$73,938 $221,813 $101,225 2.2$20,500 $61,500 $28,157 2.2

$205,250 $615,750 $283,712 2.2$35,750 $107,250 $49,040 2.2

808 B.K. McCabe et al. / Applied Energy 114 (2014) 798–808

References

[1] Mittal GS. Treatment of wastewater from abattoirs before land application—areview. Bioresource Technol 2006;97(9):1119–35.

[2] Laginestra M, van-Oorschot. Wastewater treatment pond systems – anAustralian experience. Ozwater Paper 082. In: Proceedings of AustralianWater Association Conference. Melbourne, Australia; March 2009.

[3] MLA. Industry environmental sustainability review. Meat and LivestockAustralia. Sydney, Australia; 2010.

[4] Meat Technology Update. Covered anaerobic ponds. Meat Technology Update4/10; October 2010. <www.meatupdate.csiro.au>.

[5] Johns M. Developments in wastewater treatment in the meat processingindustry: A review. Bioresource Technol 1995;54(3):203–16.

[6] Pöschl M, Ward S, Owende P. Evaluation of energy efficiency of various biogasproduction and utilization pathways. Appl Energy 2010;87:3305–21.

[7] Marcos A, Al-Kassir A, López F, Cuadros F, Brito P. Environmental treatment ofslaughterhouse wastes in a continuously stirred anaerobic reactor: Effect offlow rate variation on biogas production. Fuel Process Technol2012;103:178–82.

[8] Martinez E, Marcos A, Al-Kassir A, Jaramillo MA, Mohamad AA. Mathematicalmodel of a laboratory-scale plant for slaughterhouse effluents biodigestion forbiogas production. Appl Energy 2012;95:210–9.

[9] Thorin E, Lindmark J, Nordlander E, Odlare M, Dahlquist E, Kastensson J,Leksell N, Pettersson C-M. Performance optimization of the Växtkraft biogasproduction plant. Appl Energy 2012;97:503–8.

[10] Nakhla G, Al-Sabawi M, Bassi A, Liu V. Anaerobic treatability of high oil andgrease rendering wastewater. J Hazard Mater 2003;102:243–55.

[11] Luostarinen S, Luste S, Sillanpää M. Increased biogas production at wastewatertreatment plants through co-digestion of sewage sludge with grease trapsludge from a meat processing plant. Bioresource Technol 2009;100:79–85.

[12] Petrury R, Lettinga G. Digestion of a milk-fat emulsion. Bioresource Technol1997;61:141–9.

[13] Cammarota M, Freire D. A review on hydrolytic enzymes in the treatment ofwastewater with high oil and grease content. Bioresource Technol2006;97:2195–210.

[14] Cirne D, Paloumet X, Bjornson L, Alves M, Mattiasson B. Anaerobic digestion oflipid-rich waste – effects of lipid concentration. Renewable Energy2007;32:965–75.

[15] UNSW CRC for waste management and pollution control. Treatment ofabattoir wastewater using a covered anaerobic lagoon. Meat and LivestockAustralia. Sydney, Australia; 1998.

[16] Wan C, Zhou Q, Fu G, Li Y. Semi-continuous anaerobic co-digestion ofthickened waste activated sludge and fat, oil and grease. Waste Manage2011;31(8):1752–8.

[17] Gujer W, Zehnder A. Conversion processes in anaerobic digestion. Water SciTechnol 1983;15(8/9):127–67.

[18] Chen Y, Cheng J, Creamer K. Inhibition of anaerobic digestion process: Areview. Bioresource Technol 2008;99(10):4044–64.

[19] EPA. Development document for the proposed effluent limitations guidelinesand standards for the meat and poultry products industry point sourcecategory (40 CFR 432). Office of Water, U.S. Environmental protection agency,Washington, DC; 2002.

[20] John Green. Effluent treatment ponds, CSIRO Meat Research Laboratory; 1990<http://www.meatupdate.csiro.au/data/Waste_water_and_odour_05.pdf>.

[21] Long JH, Aziz TN, De los Reyes III FL, Ducoste JJ. Anaerobic co-digestion of fat,oil, and grease (FOG): A review of gas production and process limitations.Process Saf Environ 2012;90(3):231–45.

[22] Langergraber G, Reiger L, Winkler S, Alex J, Wiese J, Owerdieck C, Ahnert M,Simon J, Maurer M. A guide for simulation studies of wastewater treatmentplants. Water Sci Technol 2004;50:131–8.

[23] Jones R, Parker W, Zhu H, Houweling D, Murthy M. Predicting the degradabilityof waste activated sludge. WEFTEC.08; 81st Annual Technical Exhibition andConference; 2008.

[24] McCabe B, Hamawand I, Baillie C, Pittaway P, Yusaf T. Assessing a newapproach to covered anaerobic pond design in the treatment of abattoirwastewater. Australian Journal of Multidisciplinary Engineering2013;10(1):81–93.

[25] McCabe B, Hamawand I, Baillie C. Investigating wastewater modelling as a toolto predict anaerobic decompoisition and biogas yield of abattoir effluent.Journal of Environmental Chemical Engineering. 2013. http://dx.doi.org/10.1016/j.jece.2013.07.015.

[26] Liwarska-Bizukojc E, Olejnik D, Biernacki R, Ledakowicz S. Calibration of acomplex activated sludge model for the full-scale wastewater treatment plant.Bioprocess Biosyst Eng 2011;34:659–70.

[27] Safley L, Westerman P. Biogas production from anaerobic lagoons. Biol Waste2013;23:181–93.

[28] Borja R, Banks C, Wang Z, Mancha A. Anaerobic digestion of slaughterhousewastewater using a combination sludge blanket and filter arrangement in asingle reactor. Bioresource Technol 1998;65(1–2):125–33.

[29] Hamawand I, Yusaf T, Hamawand SG. Coal Seam Gas and Associated Water: Areview paper. Renew Sust Energy Rev 2013;22:550–60.

[30] ADIAC, Anaerobic Digestion Initiateve Advisory Committee of BC, <http://www.bcfarmbiogas.ca/Feedstockenergy/Energy> [accessed 29.07.13].