- Email: [email protected]
Assessing the variability in biomethane production from the organic fraction of municipal solid waste in batch and continuous operation
Applied Energy 128 (2014) 307–314
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Assessing the variability in biomethane production from the organic fraction of municipal solid waste in batch and continuous operation James D. Browne a, Eoin Allen b,c, Jerry D. Murphy b,c,⇑ a
Agrifood and Biosciences Institute, Hillsborough, Northern Ireland, UK School of Engineering, University College Cork, Ireland c Environmental Research Institute, University College Cork, Ireland b
h i g h l i g h t s Eight organic waste streams were examined for biochemical methane potential (BMP). 1 in continuous trials. day1 led to a reduction in methane yield. 1 The low C:N ratio led to levels of 7000 mg N L at high loading rates. 1 Free ammonia levels of 1000 mg N L were encountered at a pH of 8.
Commercial food waste produced 560 mL CH4 g VS 3
Raising the loading rate to 4 kg VS m
a r t i c l e
i n f o
Article history: Received 28 August 2013 Received in revised form 10 December 2013 Accepted 28 April 2014 Available online 20 May 2014 Keywords: Anaerobic digestion Food waste BMP CSTR
a b s t r a c t This paper examines the variability in biomethane potential from the organic fraction of municipal solid waste depending on source of origin. Eight organic waste streams were examined for biochemical methane potential (BMP). Specific methane yields of between 274 and 368 mL CH4 g VS1 for household waste and 491–535 mL CH4 g VS1 for commercial waste were achieved. Inclusion of garden waste reduced methane yields. A continuous trial on commercial food waste produced an average of 560 ± 29 mL CH4 g VS1 at a moderate organic loading rate (OLR) of 2 kg VS m3 day1 with a hydraulic retention time (HRT) of 30 days. Raising the OLR to 4 kg VS m3 day1 led to a reduction in specific methane yield. The low carbon to nitrogen (C:N) ratio of commercial food waste (14.4) led to process instability due to total ammonia nitrogen levels in excess of 7000 mg L1 towards the end of the trial. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Many municipal organic waste streams are dominated by food waste, particularly catering premises such as restaurants, hotels and office canteens. In the Republic of Ireland, food waste accounts for approximately 25% of domestic household and 40% of commercial waste [1]. The organic fraction of municipal solid waste (OFMSW) is a term often used in Ireland and the UK to describe food and garden waste in household and commercial waste streams. In many EU countries OFMSW is simply referred to as biowaste. National and European legislation places restrictions on the amount of OFMSW which may be sent to landfill [2] while the current EU Waste Framework Directive [3] seeks ⇑ Corresponding author at: School of Engineering, University College Cork, Ireland. Tel.: +353 21 490 2286. E-mail address: [email protected] (J.D. Murphy). http://dx.doi.org/10.1016/j.apenergy.2014.04.097 0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
to encourage waste separation at source and biological treatment of OFMSW. Anaerobic digestion (AD) is a mature biotechnology which can maximise the value of organic waste. The methane component of biogas, produced from the anaerobic process, is a valuable renewable gaseous fuel. The digestate from the biogas process may be used as a mineral rich fertilizer and reduce synthetic fertilizer consumption [4]. One of the objectives of this paper is to outline the variability in methane yields from OFMSW depending on the waste source and type of collection. A selection of organic waste samples from domestic, commercial and food processing waste streams were investigated. The biochemical methane potential (BMP) test was used to assess the methane yield for each substrate. In addition to the BMP tests, a continuous AD trial was carried out for 25 weeks using commercial canteen food waste as substrate to examine the effects of organic loading rate and hydraulic retention time on the specific methane yield.
308
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
2. Materials and methods 2.1. Collection, preparation and characterisation of waste samples Samples were collected in a large centralised facility (Acorn Recycling Ltd.) licensed to treat 45,000 tonnes per annum of OFMSW (referred to as brown bin waste in Ireland). This facility treats a wide range of municipal organic waste streams from across the province of Munster in Ireland (population circa 1.25 million people). As shown in Fig. 1, a total of 8 different waste streams were sampled; 4 household, 2 commercial and 2 food processing streams. All waste streams identified were sampled frequently (samples taken from each incoming waste collection truck) over a 6 week period in autumn/winter to build up a representative sample of each waste stream. Previously only the commercial waste stream had been sampled in the summer time of the same year (2012) as the other waste streams were not available at that time. Samples of each waste stream were labelled and stored in a freezer at 20 °C until required. At the end of the sampling period, samples from each stream were defrosted at room temperature for 24 h and were then thoroughly mixed, screened for nonorganic material and macerated in a Buffalo food mincer to a particle size of less than 5 mm to form a large bulk sample for each waste stream. Approximately 10 kg of material was then sub-sampled from the bulk sample of each waste stream following the German VDI guidelines on sampling solid material [5]. All sub-samples were stored in a freezer at 20 °C until required as previously described by Browne et al. [6]. A proximate and elemental analysis was carried out in triplicate samples from each waste stream as shown in Table 1.
2.2. BMP tests The apparatus used to conduct the BMP tests was the Automatic Methane Potential Test System II (Bioprocess Control Sweden AB). This laboratory instrument is specially designed for determination of the BMP of a substrate. The AMPTS II system consists of three major parts as follows: 1. A temperature controlled water bath with 15 bottle reactors of 500 mL volume, each equipped with a mixer that can be run in either continuous or intermittent mode. 2. A carbon dioxide fixing unit with an alkaline solution (3 N sodium hydroxide) that absorbs the carbon dioxide and hydrogen sulphide produced during the anaerobic digestion process. 3. A gas measuring unit consisting of 15 parallel operating cells, where the gas is measured through water displacement. When approximately 10 mL of gas has been accumulated each cell
opens and releases the gas. For each opening, the time, temperature and pressure are registered and stored locally in an embedded Central Processing Unit (CPU). Based on these measurements, normalised (0 °C, 1 atm and dry gas) accumulated gas production and gas flow rate are calculated. The BMP tests were performed with a working volume of 400 mL. The ratio of inoculum to substrate was chosen to be 2:1 on a volatile solids (VS) basis. The inoculum to substrate ratio is a critical parameter in conducting a BMP test according to the Anaerobic Digestion Specialist Group of the International Water Association [7]. A ratio of 2:1 or greater of inoculum to substrate on a VS basis is recommended for BMP trials by Raposo et al. [8] to limit any inhibitory effects due to the chemical composition of the substrate such as inhibition associated with accumulation of ammonia and volatile fatty acids (VFA) [8]. All samples were tested for BMP in triplicate. A BMP test of the inoculum alone (referred to as a blank) was conducted in triplicate. The average methane yield from the blanks was subtracted from the samples of OFMSW with inoculum to accurately assess the BMP yields from the samples only. A triplicate BMP test was also carried out on cellulose for quality control as the maximum BMP from cellulose is known and can be compared with the BMP yield. The percentage volatile solids destroyed during the batch process was calculated as follows:
%VS destruction ¼ 100 ð1 ðVSf VSfb Þ=ðVSi VSib Þ
ð1Þ
where VSi is the amount of total input VS (g), VSf is the amount of total VS at the end of the BMP test (g), VSib is the amount of VS (g) in the inoculum (blank) at the beginning of the BMP test and VSfb is the amount of VS (g) in the inoculum (blank) at the end of the test. The Buswell equation was used to calculate the theoretical maximum methane potential [9].
n BMPTh ¼
2
þ 8a 4b 22400 ð12n þ a þ 16bÞ
ð2Þ
where n is the number of atoms of carbon; a is the number of atoms of hydrogen; b is the number of atoms of oxygen. The biodegradability index is the ratio of the measured BMP divided by the theoretical methane yield according to the Buswell equation and is used to assess the level of biodegradability of a substrate. 2.3. Source and characteristics of inoculum for BMP tests The inoculum for the BMP tests was obtained from a lab scale 300 L digester treating mostly cattle slurry and a small portion of
Fig. 1. Illustration of samples taken from the organic fraction of municipal solid waste (with and without garden waste).
309
J.D. Browne et al. / Applied Energy 128 (2014) 307–314 Table 1 Characterisation of OFMSW samples. Samples
Total solids (%)
Volatile solids (% TS)
Total carbon (% TS)
Total hydrogen (% TS)
Total nitrogen (% TS)
C:N
Household brown bin Rural with garden (RWG) Rural no garden (RNG) Urban with garden (UWG) Urban no garden (UNG)
33.4 (0.4) 30.6 (3.3) 25.66 (0.1) 31.0 (2.4)
82.3 88.4 73.6 93.8
43.3 44.9 41.3 46.5
5.9 6.6 5.2 7.3
2.7 3.1 2.6 3.7
(0.1) (0.2) (0.4) (0.1)
16 14.5 16 12.6
Commercial waste Commercial canteen summer (CCS) Commercial canteen winter (CCW)
32.8 (0.1) 23.8 (0.5)
92.6 (0.3) 90.0 (0.3)
49.0 (0.6) 48.2 (0.2)
7.0 (0.1) 7.0 (0.04)
3.4 (0.2) 3.6 (0.2)
14.4 13.4
Food processing Food processing bakery waste (FPBW) Food processing cheese waste (FPCW)
45.7 (0.4) 15.9 (0.1)
91.9 (0.6) 55.6 (0.3)
52.7 (0.2) 24.9 (0.1)
8.2 (0.05) 4.1 (0.03)
2.8 (0.2) 4.6 (0.03)
18.8 5.4
(0.2) (0.4) (0.4) (0.3)
(0.2) (0.2) (0.2) (0.2)
(0.1) (0.1) (0.1) (0.1)
All values are presented as mean and (standard deviation).
food waste operating at mesophilic temperatures (35 °C). After an incubation period of one week the inoculum had a pH of 7.9, total solids (TS) of 34.2 g VS kg1 and volatile solids (VS) content of 21.4 g VS kg1 after passing through a 2 mm sieve. Inoculum from both rounds was tested using cellulose as a standard control substrate (C12 H20 O10). The maximum theoretical methane yield from cellulose according to the Buswell equation is 415 L CH4 kg VS1. The specific methane yield produced from the cellulose was 371 ± 4 L CH4 kg VS1. This is almost 90% of the theoretical maximum indicating a healthy inoculum. 2.4. Kinetic modelling of BMP tests Two first order kinetic models were used to fit the cumulative methane production data from the BMP tests. Assuming first-order kinetics for the hydrolysis of particulate organic matter, the cumulative methane production can be described by Eq. (3):
YðtÞ ¼ Y m 1 expðkt Þ
ð3Þ
where Y(t) is the cumulative methane yield at digestion time t days (mL CH4 g VS1 added), Ym is methane potential of the substrate (mL CH4 g VS1 added), k is methane production rate constant (first order disintegration rate constant) (day1), t is the time (days). The duration of the lag phase is also an important factor in determining the efficiency of anaerobic digestion. The lag phase (k) can be calculated with the modified Gompertz model as described by [10] in Eq. (4) below:
Rmax e M ¼ P exp exp ðk tÞ þ 1 P
ð4Þ
where M is the cumulative methane yield at a given time (mL CH4 g VS1), P is the max methane potential (L CH4 kg VS1) from the BMP test, Rmax is the maximum methane production rate (L CH4 kg VS1 day1), e = 2.7183, k is the lag phase for methane production to begin (days), t is the time (days). A nonlinear least-square regression analysis was performed using Excel to determine k, Rmax, k, and the predicted methane yield. The predicted methane yield obtained from the regression analysis was plotted with the measured methane yield. The statistical indicators, Correlation coefficient (R2) and root mean square error (RMSE) were calculated to assess the goodness of fit [11]. 2.5. Statistical analysis The significance of differences in the average methane yields was determined by using single factor Analysis of Variance (ANOVA) in Excel software 2007. If the calculated F value was higher than the tabulated F value, the minimum significant difference (MSD) was calculated to judge whether two or more averages
were significantly different or not (Tuckey test). MSD was calculated as P = 0.05 (MSD0.05) [12]. 2.6. Continuous AD trial on canteen food waste The continuous trial was carried out in a continuously stirred tank reactor (CSTR) with a total volume of 5 L (working volume of 4 L) and ran for a period of 25 weeks. The reactor was maintained at a temperature of 37 ± 1 °C and was continuously stirred at a rate of 100 rpm. The reactor was constructed out of thick walled plastic with a vertically mounted stirring mechanism as shown in Fig. 2. The reactor was placed inside a coiled copper pipe frame which was heated by a thermo-circulator. Biogas flow was measured using a tipping bucket mechanism whereby the number of tips was recorded and multiplied by the calibrated gas volume of the tipping bucket (78 mL per tip). Biogas was sampled downstream of the gas flow tipping meter in 1 L Tedlar gas bags and analysed for methane, carbon dioxide and hydrogen sulphide. The continuous trial was operated for 176 days using commercial canteen food waste from the same collection as the commercial canteen summer (CCS) sample in the BMP trials. The continuous system was started at a moderate organic loading rate (OLR) of 2 kg VS m3 day1. The hydraulic retention time (HRT) was initially set at 30 days. This was achieved by adding a portion of digestate back in with the input feed keeping the total solids content of the input feed to 10% which facilitated easy stirring of the digester contents. The reactor was maintained at this OLR for 3 HRTs (period 1). The first HRT incorporated the start up and acclimatisation period. Volatile fatty acids (VFA), total alkalinity (TA), total ammonia nitrogen (TAN) were measured weekly (using methods described in Section 2.7 Analytical methods) and pH measured daily. 2.7. Analytical methods Total solids and volatile solids were determined gravimetrically following the standard methods [13]. The biogas composition in the continuous trial was measured by infra red gas analyser (Status Scientific Control I-R biogas analyser). The instrument was calibrated before the commencement of the trial and showed an accuracy of ±1% when tested weekly on a standard mixture of 65% methane and 35% carbon dioxide provided by BOC specialty gases. All methane yields were adjusted to standard temperature of 273 K and 1 atmosphere (1013 hPa). Volatile organic acids and total alkalinity were measured using the Nordmann titration method (1977) [14] using 0.1 N sulphuric acid and a Titronic Universal Titrator. The pH of the digestate was measure daily using a Jenway 3510 pH meter. Total ammonia was measured using the Hach NH3–N vials and spectrophotometer DR 3900.
310
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
Fig. 2. Continuously stirrer tank reactor (5 L) used for continuous trial.
3. Results and discussion 3.1. Results from the BMP tests Table 2 shows the range of BMP yields from samples tested. The commercial waste samples exhibited a much higher methane yield and greater portion of biodegradable organic matter than the household waste samples. In particular the household waste streams which consisted of mostly garden waste had a much lower biodegradability index than waste streams without garden waste. The BMP result for the cheese waste was much lower than expected. In a previous study by Browne et al. [6] a sample of cheese processing treatment sludge from a different location yielded 461 L CH4 kg VS1. This demonstrates that the type of existing biological waste treatment processes at dairy plants can produce waste sludge with hugely different biomethane potential. A one way Anova analysis showed a large statistical difference between the biomethane potential results depending on the source of OFMSW (F7,16 = 332.6, P < 0.001); where there are 7° of freedom between samples and 16° of freedom within samples. Multiple comparisons testing using the Tuckey test on the mean BMP values of the 8 waste streams tested, resulted in a total of 28 comparative tests which checked if the difference between means was greater than the minimum significant difference (MSD = 34.4 mL g VS1, P < 0.05). The Tuckey test revealed that there was a significant difference in average biomethane potential between 25 out of 28 comparisons. Interestingly, in the household waste stream there was no significant difference between urban and rural samples that came from a similar collection system (P > 0.05). However there was a significant difference in methane potential depending whether garden waste was included or not. For example, samples without garden waste gave higher methane yields than samples which included garden waste. Canteen waste samples gave significantly higher BMP yields than from household waste streams. There was a small but significant difference between canteen waste samples depending on the season. Samples taken from the same waste collection run in summer (2012) gave 9% higher BMP yields than autumn/winter (2012). In the food processing stream
bakery waste samples gave vastly greater methane yields (529 mL CH4 g VS1) than from cheese waste activated sludge (188.5 mL CH4 gVS1). Interestingly the bakery waste sample did not differ significantly from the canteen waste summer. The results from the commercial waste samples are similar to previously reported BMP yields from canteen food waste (480–530 L CH4 kg VS1) by Browne and Murphy [15]. 3.2. Kinetic study results The results of the kinetics analysis using the first order kinetic model and the modified Gompertz model are summarised in Tables 3a and 3b respectively. The first order kinetic model gave k values ranging from 0.12 to 0.17 day1 for household samples, 0.07–0.09 day1 for commercial samples and 0.08–0.13 day1 for food processing samples. The commercial food waste samples had higher percentages of proteins and lipids which take longer to digest than carbohydrates therefore resulting in lower k values Vavilin et al. [16]. The modified Gompertz model showed a lag time of 1.2 and 3 days for all samples tested. The time taken to reach 90% of the maximum BMP value was shown to range from 9 to 15 days indicating that all OFMSW substrates were readily degradable. Both models exhibited a good fit when plotted against the measured data with the coefficient of determination (R2) ranging from 0.93 to 0.95 for the first order model and 0.99 for the Gompertz model. The RMSE ranged from 10.7 to 48.8 mL CH4 g VS1 for the first order model while the Gompertz model gave lower values of over 0.7–9.9 mL CH4 g VS1. Both models can be used to predict the maximum methane potential. The modified Gompertz model gave slightly lower predicted maximum BMP yields than the measured data ranging from 0.8% to 9.3% while the first order model generally gave higher predicted methane yields than measured ranging from 1.8 to +19.2%. In 87.5% of cases the model Gompertz gave a more accurate predicted max biomethane yield than the first order equation. Based on the statistical indicators (RMSE and R2) the modified Gompertz model was found to demonstrate the best fit for the samples tested. The cumulative methane yields of the BMP tests are shown in Fig. 3.
311
J.D. Browne et al. / Applied Energy 128 (2014) 307–314 Table 2 Spectrum of food waste – samples collected. Source
BMP (30 days) (mL CH4 g VS1)
Theoretical BMPa (mL CH4 g VS1)
Biodegradability index
Volatile solids destruction as measured (% VS)
Rural with garden (RWG) Rural no garden (RNG) Urban with garden (UWG) Urban no garden (UNG) Commercial canteen summer (CCS) Commercial canteen winter (CCW) Food processing bakery waste (FPBW) Food processing cheese waste (FPCW)
274.1 367.8 296.7 343.7 534.5 490.9 529.2 188.5
577 566 625 564 620 620 696 530
0.48 0.65 0.47 0.61 0.86 0.79 0.76 0.36
47 69 51 60 81 80 81 42
(4.6)b (6.2)c (6.1)b (2.7)c (5.0)e (4.8)d (25.4)e (1.2)a
a, b, c, d, e = Levels of Tuckey multiple comparisons testing between sample means. Samples with the same letter are not significantly different (P > 0.05). Results are indicated as a mean with standard deviation is in brackets. a Theoretical BMP calculated from the Buswell equation [9].
Table 3a Results of BMP kinetic analysis using the first order kinetic equation. Sample Household waste Rural with garden Rural no garden Urban with garden Urban no garden Commercial waste Canteen summer Canteen winter Food processing waste Bakery waste Cheese process waste
RMSE (mL CH4 g VS1)
R2
k (day1)
+6.5 +5.5 +1.8 +7.4
21.6 28.7 18.0 27.2
0.93 0.93 0.95 0.93
0.12 0.14 0.17 0.12
603 585
+12.8 +19.2
40.2 36.7
0.94 0.95
0.09 0.07
623 185
+17.7 1.8
48.8 10.7
0.92 0.95
0.08 0.13
BMP measured (mL CH4 g VS1)
BMP predicted (mL CH4 g VS1)
274.1 367.8 296.7 343.7
(4.6) (6.2) (6.1) (2.7)
292 388 302 369
534.5 (5.0) 490.9 (4.8) 529.2 (25.4) 188.5 (1.2)
Difference (%)
Table 3b Results of BMP kinetic analysis using the modified Gompertz equation. BMP measured (mL CH4 g VS1)
BMP predicted (mL CH4 g VS1)
Difference (%)
R2
RMSE (mL CH4 g VS1)
Lag phase (days)
T90 (days)
Household waste Rural with garden Rural no garden Urban with garden Urban no garden
274.1 367.8 296.7 343.7
(4.6) (6.2) (6.1) (2.7)
268 363 288 338
2.2 1.3 2.9 1.7
0.99 0.99 0.99 0.99
2.9 4.1 0.8 3.3
2.2 2.0 1.3 2.2
11 10.2 8.7 11.3
Commercial waste Canteen summer Canteen winter
534.5 (5.0) 490.9 (4.8)
530 484
0.8 1.4
0.99 0.99
3.4 0.7
2.3 2.5
13.4 15.3
529.2 (25.4) 188.5 (1.2)
528 171
0.2 9.3
0.99 0.97
9.9 0.7
3.0 1.2
14.3 10.8
Sample
Food processing waste Bakery waste Cheese process waste
Results are indicated as a mean with standard deviation is in brackets.
3.3. Results from continuous digestion of canteen food waste 3.3.1. Specific methane yields in period 1 The reactor was maintained at an OLR of 2 kg VS m3 day1 for 3 HRTs (period 1). The first HRT incorporated the start up and acclimatisation period. By the end of the first HRT the system had reached a steady state of methane production. Methane yields from the second and third HRT were used to calculate average specific methane yield for period 1 (OLR of 2 kg VS m3 day1) which was 560.1 ± 29.3 mL CH4 g VS1 added. The standard deviation in the second and third HRT was only 5% of the total yield and clearly showed that the reactor was in steady state. The weekly average specific methane yield is shown in Fig. 4(a). The daily percentage methane in the biogas is shown in Fig. 4(b). In the start-up period the percentage methane increased from 40.4% to 60% over the first
30 days with the weighted average methane percentage in the biogas remaining at 60 ± 1.3% for period 1. 3.3.2. Specific methane yields in period 2 After completing 3 HRTs of 30 days at the initial feeding rate, the OLR was increased to 3 kg VS m3 day1 at day 99. By increasing the OLR to 3 kg VS m3 day1 the HRT was reduced to 21 days as the solids content of the input material was kept at 10% TS by recirculation of an increased amount of digestate. The OLR was maintained at 3 kg VS m3 day1 for 2 HRTs (42 days). The average specific methane yield (SMY) for period 2 was 484 ± 72.0 mL CH4 g VS1. This is a reduction of about 13% from the previous SMY in period 1. The standard deviation in period 2 is approximately 15% of the average SMY and shows that there was greater fluctuation in daily gas production at the higher OLR of 3 kg VS m3 day1.
312
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
3.3.4. Conversion of volatile solids to gas To assess the conversion of VS to gas Eq. (5) taken from Kang and Wieland, [17] is used:
MR ¼ LN ðð16 CH4 %Þ þ ð44 CO2 %ÞÞ=22:413
ð5Þ
where MR is the daily average mass of volatile solids removed (g VS); LN is the average daily normalised biogas volume (L) at standard temperature and pressure (STP); CH4 % is the methane content in the biogas; CO2 is the carbon dioxide content in the biogas; There are 22.413 L per mole of gas at STP. According to Eq. (5) the average removal of VS in period 1 was 84% with a HRT of 30 days. This decreased to 72% in period 2 with a HRT of 21 days and further reduced to 54% in period 3 with a HRT of 17 days. The average concentration of total solids in the digestate increased from 5.1 ± 0.5% TS in period 1 to 5.5 ± 0.3% TS in period 2 and 6.7 ± 0.9% TS in the final period. This indicates that reducing the HRT also reduces the degradation of volatile solids. However the large drop in specific methane yield towards the end of the trial may not be entirely as a result of the reduced HRT as signs of process instability emerged towards the end of the trial at an OLR of 4 kg VS m3 day1. 3.4. Monitoring process stability in continuous trial
Fig. 3. BMP cumulative methane yields for (a) household samples and (b) commercial & processing waste samples.
The weighted average methane content in the biogas increased to 61.5 ± 2.8% in period 2. 3.3.3. Specific methane yields in period 3 On day 142 the OLR was further increased to 4 kg VS m3 day1 which resulted in a reduced HRT of 17 days. The trial was completed on day 176. The average SMY in the final period was 381.5 ± 52.0 mL CH4 g VS1 which was a 21% decrease in SMY from period 2 and a 32% decrease from period 1. The average methane content was 60.7 ± 3.6%.
Fig. 4. (a) Weekly average specific methane yield and (b) daily methane percentage.
During the continuous trial the total volatile fatty acids (VFA), total alkalinity (TA), pH and total ammonia nitrogen (TAN) were monitored to assess the stability of the digestion process. The average results from the three time periods are shown in Table 4. In Period 1 (OLR 2 kg VS m3 day1) the concentration of total VFAs was 1128 ± 281 mg Aceq L1. A small increase was observed during Period 2 (OLR of 3 kg VS m3 day1) with an average of 1511 ± 77 mg Aceq L1. The concentration of VFAs rose sharply towards the end of the trial during Period 3 (OLR of 4 kg VS m3 day1) as shown in Fig. 5(a), with an average of 2595 ± 750 mg Aceq L1. Single factor ANOVA showed that there was a significant difference between VFA concentrations between periods (P < 0.001). The sharp increase in VFA concentration in period 3 indicated that the methane producing microbes were stressed and were not fully utilising the VFAs produced by the acidogenic/ acetogenic bacteria. The average total alkalinity for the period 1 was 8093 ± 970 mg CaCO3 L1. This increased to 9830 ± 159 mg CaCO3 L1 in period 2 and 10,230 ± 185 mg CaCO3 L1 in period 3. The ratio of VFA/TA is often used to assess the stability of the AD process. A ratio of 0.4 or less indicates that the process is stable while ratios over 0.8 indicate organic overloading and process instability. During the trial the VFA/TA ratio remained below 0.4, however it is clear that even though the ratio was within stable limits, the specific methane yields (SMY) were in sharp decline in the final period. Linear regression analysis showed that there was a relationship between increasing concentrations of FAN and decreasing SMY. The linear relationship gave the equation; SMY = 0.22 * FAN + 594.5 (r2 = 0.65). Increasing FAN also had a linear relationship with increasing VFA concentrations with VFA = 1.85 * FAN + 690 (r2 = 0.78). This indicates that increasing concentrations of FAN disrupted the uptake of VFAs by the methanogens and resulted in accumulation of VFAs which ultimately lead to a decrease in SMY. During the trial the pH increased from an average of 7.7 ± 0.1 in period 1 to 8.1 ± 0.1 in the final period. The high pH is of concern when combined with high levels of TAN as the relationship between ionised ammonium (NH+4) and unionised (free) ammonia (FAN) is pH and temperature dependent. The fraction of FAN is dependent on TAN, temperature and pH and is described by Eq. (6) taken from Angelidaki and Ahring [18].
J.D. Browne et al. / Applied Energy 128 (2014) 307–314 Table 4 Summary of results from continuous AD trial of canteen food waste. Period 3
1
OLR (kg VS m day ) HRT (days) SMY (mL CH4 g VS1) % CH4 (weighted average) % VS conversion to gas % TS (digestate) % VS (digestate) pH TAN (mg N L1) FAN (mg N L1) VFAs (mg Aceq L1) Alkalinity (mg CaCO3 L1) VFA/TA
1
2
3
2 30 3 560.1 (29.3) 60.1 (1.3) 84 5.1 (0.5) 3.5 (0.3) 7.7 (0.13) 3543 (525) 237 (50) 1128 (281) 8093 (970) 0.14
3 21 2 483.9 (72.0) 61.5 (2.8) 72 5.5 (0.3) 3.9 (0.2) 7.9 (0.14) 5342 (485) 433 (185) 1511 (77) 9830 (159) 0.15
4 17 2 381.5 (52.0) 60 (3.6) 54 6.7 (0.9) 4.7 (0.7) 8.1 (0.11) 7205 (280) 992 (83) 2595 (750) 10,230 (185) 0.25
Results are indicated as a mean with standard deviation in brackets.
313
pH values. The concentration of TAN increased linearly for the duration of the trial with final concentrations in excess of 7000 mg N L1, as shown in Fig. 5(b). The relatively high concentration of TAN coupled with a pH of about 8, resulted in high concentrations of FAN of approximately 1000 mg N L1. This is a very high concentration of FAN and is considered to be in the toxicity range for methane production by Chen et al., [19]. Several threshold values are reported for TAN and FAN causing inhibition of the biogas process in anaerobic reactors. For example, a threshold value around 1000 mg N L1 FAN is reported by Chen et al. [19], Gerardi [20] and Hansen et al. [21]. Above this concentration, inhibition and an apparent decrease in the specific growth rate in batch cultures at pH 8.0 (reactor pH) is found. Although free ammonia can inhibit anaerobic digestion, relatively high TAN concentrations can be tolerated Chen et al. [19]. 100% Inhibition is reported in the range of 8–13 g N L1 depending on the condition of acclimatisation and the pH of the system. Despite the long delay in methane production, it was reported that methanogens present in digested sewage sludge could tolerate up to 5 g L1 NH3–N (TAN) and the methane producing biomass was able to acclimatise to such high ammonium concentration Chen et al. [19]. Angelidaki and Ahring [18] reported that acetate-utilising bacteria adapted to NH3 at a free ammonia (FAN) concentration of up to 700 mg N L1, while lower free ammonia concentrations (100–150 mg N L1) have been reported for initial inhibition of an unadapted process. Banks and colleagues [22] reported high concentrations of TAN and FAN at high organic loading rates using source separated food waste. They showed that at elevated levels of TAN the acetoclastic methanogens were virtually nonexistent with methane production coming from the hydrogenotrophic route. To overcome the inhibitory effects of high levels of ammonia the addition of trace elements such as iron, cobalt, selenium and molybdenum were successfully shown to improve methane yields at high organic loading rates (e.g. 5 kg VS m3 day1) [22]. 3.6. Comparison of specific methane yields from batch and continuously fed systems
Fig. 5. (a) Monitoring of total alkalinity (TA) and volatile fatty acids (VFA) (b) total ammonia nitrogen (TAN) and free ammonia nitrogen (FAN) and (c) pH in the continuous trial.
Free Ammonia ðFANÞðmg=LÞ
The specific methane yield (SMY) produced during period 1 of the continuous trial was relatively high in comparison to other reported methane yields from food waste. The highest average SMY of 560 ± 29.3 mL CH4 g VS1 was achieved at an OLR of 2 kg VS m3 day1 and HRT of 30 days. This is 90.3% of the Buswell equation value. It is however 7% higher than the average BMP result from the same sample. This indicates that at moderate organic loading rates a continuous AD process may equal or even exceed methane yields from the BMP test. This may be due to acclimatisation of the inoculum with time. Thamsiriroj and Murphy [23] also recorded higher SMYs in continuous digestion than in BMP mode. Zhang and colleagues [24] achieved 425 L CH4 kg VS1 from continuous digestion of source segregated food waste at an OLR of 2 kg VS m3 day1 which was slightly lower than the BMP results of the same material (445–456 L CH4 kg VS1).
¼ TAN ðmg=LÞ ðð1 þ ðð10Þ^ ðpHÞÞ=ð10Þ^ ð0:09018 þ ð2729:92=ðT þ 273ÞÞÞÞÞ1
ð6Þ
As shown in Fig. 5 even slight changes in pH can have a large effect on the proportion of FAN. If pH rose from 7 to 8 according to Eq. (6) this would result in a 10-fold increase in the concentration of FAN. 3.5. The inhibitory effects of high ammonia concentrations Total ammonia nitrogen (TAN) contributes to the buffering capacity of the system but can be toxic to methanogens at higher
4. Conclusions The characteristics of OFMSW can vary largely depending on the source and type of collection system with BMP values of between 274 and 535 mL CH4 g VS1. A continuous trial on commercial food waste produced an average of 560 ± 29 mL CH4 g VS1 at a moderate OLR of 2 kg VS m3 day1 with a HRT of 30 days. At higher OLRs (4 kg VS m3 day1) increasing concentrations of VFAs (2595 mg L1) coupled with high concentrations of free ammonia (952 mg L1) led to a greatly reduced average specific methane yield (344 mL CH4 g VS1).
314
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
Acknowledgements Researchers were funded by the Irish Research Council for Science, Engineering and Technology (IRCSET), Science Foundation Ireland (SFI) and Bord Gais Energy (BGE). Laboratory equipment was funded by Bord Gais Networks (BGN). The authors would like to thank Mr. Ronan Beasley of Acorn Recycling Ltd for providing the samples of OFMSW.
References [1] McCoole F, Kurz I, McDonagh M, Derham J, O’Neill D. National waste report 2010. In: Environmental protection agency I, editor. Environmental Protection Agency, Ireland; 2012. [2] EC. Council Directive 1999/31/EC of 26 April 1999 on the landfill of waste. Official Journal of the European Union; 1999. [3] EC. Directive 2008/98/EC of the European Parliament and of the Council of 19 November 2008 on waste and repealing certain Directives. In: Commission E, editor; 2008. [4] Lukehurst C, Frost P, Al Seadi T. Utilisation of digestate from biogas as biofertiliser, International Energy Agency (IEA) Bioenergy Task 2010;37. Available at: www.iea-biogasnet/technical-brochures.html. [5] VDI. Fermentation of organic materials – characterisation of the substrate, sampling, collection of material data, fermentation tests. Characterization of substrates. Dusseldorf, Germany: Verein Deutscher Ingenieure; 2006. p. 11–9. [6] Browne J, Allen E, Murphy J. Evaluation of the biomethane potential from multiple waste streams for a proposed community scale anaerobic digester. Environ Technol 2013:1–12. [7] Angelidaki I, Alves M, Bolzonella D, Borzacconi L, Campos JL, Guwy AJ, et al. Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays. Water Sci Technol 2009:927–34. [8] Raposo F, Banks CJ, Siegert I, Heaven S, Borja R. Influence of inoculum to substrate ratio on the biochemical methane potential of maize in batch tests. Proc Biochem 2006;41:1444–50.
[9] Buswell AM, Neave SL. Laboratory studies of sludge digestion. Illinois Division of State Water Survey, Bulletin No 30. URBANA, ILLINOIS1930. [10] Kafle GK, Kim SH. Anaerobic treatment of apple waste with swine manure for biogas production: batch and continuous operation. Appl Energy 2013;103: 61–72. [11] Brown AM. A step-by-step guide to non-linear regression analysis of experimental data using a Microsoft Excel spreadsheet. Comput Methods Prog Biomed 2001;65:191–200. [12] Wheater CP, Cook PA. Using statistics to understand the environment. New York: Routledge; 2000. [13] APHA. Standard Methods for the Examination of Water and Wastewater. (21st ed.) American Public Health Association, American Water Works Association, Water Environment Federation (2005). Washington, DC; 2005. [14] Nordmann W. Die Überwachung der Schlammfaulung. Beilage zur Korrespondenz Abwasser; 1977. p. 3/77. [15] Browne JD, Murphy JD. Assessment of the resource associated with biomethane from food waste. Appl Energy 2013;104:170–7. [16] Vavilin VA, Fernandez B, Palatsi J, Flotats X. Hydrolysis kinetics in anaerobic degradation of particulate organic material: an overview. Waste Manage 2008; 28:939–51. [17] Kang H, Weiland P. Ultimate anaerobic biodegradability of some agroindustrial residues. Bioresource Technol 1993;43:107–11. [18] Angelidaki I, Ahring BK. Thermophilic anaerobic digestion of livestock waste: effect of ammonia. Appl Microbiol Biotechnol 1993;38:560–4. [19] Chen Y, Cheng JJ, Creamer KS. Inhibition of anaerobic digestion process: a review. Bioresource Technol 2008;99:4044–64. [20] Gerardi MH. Toxicity. The microbiology of anaerobic digesters. John Wiley & Sons, Inc.; 2003. p. 105–15. [21] Hansen KH, Angelidaki I, Ahring BK. Anaerobic digestion of swine manure: inhibition by ammonia. Water Res 1998;32:5–12. [22] Banks CJ, Zhang Y, Jiang Y, Heaven S. Trace element requirements for stable food waste digestion at elevated ammonia concentrations. Bioresource Technol 2012;104:127–35. [23] Thamsiriroj T, Murphy JD. Modelling mono-digestion of grass silage in a 2stage CSTR anaerobic digester using ADM1. Bioresource Technol 2011;102: 948–59. [24] Zhang Y, Banks CJ, Heaven S. Co-digestion of source segregated domestic food waste to improve process stability. Bioresource Technol 2012;114:168–78.
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Assessing the variability in biomethane production from the organic fraction of municipal solid waste in batch and continuous operation James D. Browne a, Eoin Allen b,c, Jerry D. Murphy b,c,⇑ a
Agrifood and Biosciences Institute, Hillsborough, Northern Ireland, UK School of Engineering, University College Cork, Ireland c Environmental Research Institute, University College Cork, Ireland b
h i g h l i g h t s Eight organic waste streams were examined for biochemical methane potential (BMP). 1 in continuous trials. day1 led to a reduction in methane yield. 1 The low C:N ratio led to levels of 7000 mg N L at high loading rates. 1 Free ammonia levels of 1000 mg N L were encountered at a pH of 8.
Commercial food waste produced 560 mL CH4 g VS 3
Raising the loading rate to 4 kg VS m
a r t i c l e
i n f o
Article history: Received 28 August 2013 Received in revised form 10 December 2013 Accepted 28 April 2014 Available online 20 May 2014 Keywords: Anaerobic digestion Food waste BMP CSTR
a b s t r a c t This paper examines the variability in biomethane potential from the organic fraction of municipal solid waste depending on source of origin. Eight organic waste streams were examined for biochemical methane potential (BMP). Specific methane yields of between 274 and 368 mL CH4 g VS1 for household waste and 491–535 mL CH4 g VS1 for commercial waste were achieved. Inclusion of garden waste reduced methane yields. A continuous trial on commercial food waste produced an average of 560 ± 29 mL CH4 g VS1 at a moderate organic loading rate (OLR) of 2 kg VS m3 day1 with a hydraulic retention time (HRT) of 30 days. Raising the OLR to 4 kg VS m3 day1 led to a reduction in specific methane yield. The low carbon to nitrogen (C:N) ratio of commercial food waste (14.4) led to process instability due to total ammonia nitrogen levels in excess of 7000 mg L1 towards the end of the trial. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Many municipal organic waste streams are dominated by food waste, particularly catering premises such as restaurants, hotels and office canteens. In the Republic of Ireland, food waste accounts for approximately 25% of domestic household and 40% of commercial waste [1]. The organic fraction of municipal solid waste (OFMSW) is a term often used in Ireland and the UK to describe food and garden waste in household and commercial waste streams. In many EU countries OFMSW is simply referred to as biowaste. National and European legislation places restrictions on the amount of OFMSW which may be sent to landfill [2] while the current EU Waste Framework Directive [3] seeks ⇑ Corresponding author at: School of Engineering, University College Cork, Ireland. Tel.: +353 21 490 2286. E-mail address: [email protected] (J.D. Murphy). http://dx.doi.org/10.1016/j.apenergy.2014.04.097 0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
to encourage waste separation at source and biological treatment of OFMSW. Anaerobic digestion (AD) is a mature biotechnology which can maximise the value of organic waste. The methane component of biogas, produced from the anaerobic process, is a valuable renewable gaseous fuel. The digestate from the biogas process may be used as a mineral rich fertilizer and reduce synthetic fertilizer consumption [4]. One of the objectives of this paper is to outline the variability in methane yields from OFMSW depending on the waste source and type of collection. A selection of organic waste samples from domestic, commercial and food processing waste streams were investigated. The biochemical methane potential (BMP) test was used to assess the methane yield for each substrate. In addition to the BMP tests, a continuous AD trial was carried out for 25 weeks using commercial canteen food waste as substrate to examine the effects of organic loading rate and hydraulic retention time on the specific methane yield.
308
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
2. Materials and methods 2.1. Collection, preparation and characterisation of waste samples Samples were collected in a large centralised facility (Acorn Recycling Ltd.) licensed to treat 45,000 tonnes per annum of OFMSW (referred to as brown bin waste in Ireland). This facility treats a wide range of municipal organic waste streams from across the province of Munster in Ireland (population circa 1.25 million people). As shown in Fig. 1, a total of 8 different waste streams were sampled; 4 household, 2 commercial and 2 food processing streams. All waste streams identified were sampled frequently (samples taken from each incoming waste collection truck) over a 6 week period in autumn/winter to build up a representative sample of each waste stream. Previously only the commercial waste stream had been sampled in the summer time of the same year (2012) as the other waste streams were not available at that time. Samples of each waste stream were labelled and stored in a freezer at 20 °C until required. At the end of the sampling period, samples from each stream were defrosted at room temperature for 24 h and were then thoroughly mixed, screened for nonorganic material and macerated in a Buffalo food mincer to a particle size of less than 5 mm to form a large bulk sample for each waste stream. Approximately 10 kg of material was then sub-sampled from the bulk sample of each waste stream following the German VDI guidelines on sampling solid material [5]. All sub-samples were stored in a freezer at 20 °C until required as previously described by Browne et al. [6]. A proximate and elemental analysis was carried out in triplicate samples from each waste stream as shown in Table 1.
2.2. BMP tests The apparatus used to conduct the BMP tests was the Automatic Methane Potential Test System II (Bioprocess Control Sweden AB). This laboratory instrument is specially designed for determination of the BMP of a substrate. The AMPTS II system consists of three major parts as follows: 1. A temperature controlled water bath with 15 bottle reactors of 500 mL volume, each equipped with a mixer that can be run in either continuous or intermittent mode. 2. A carbon dioxide fixing unit with an alkaline solution (3 N sodium hydroxide) that absorbs the carbon dioxide and hydrogen sulphide produced during the anaerobic digestion process. 3. A gas measuring unit consisting of 15 parallel operating cells, where the gas is measured through water displacement. When approximately 10 mL of gas has been accumulated each cell
opens and releases the gas. For each opening, the time, temperature and pressure are registered and stored locally in an embedded Central Processing Unit (CPU). Based on these measurements, normalised (0 °C, 1 atm and dry gas) accumulated gas production and gas flow rate are calculated. The BMP tests were performed with a working volume of 400 mL. The ratio of inoculum to substrate was chosen to be 2:1 on a volatile solids (VS) basis. The inoculum to substrate ratio is a critical parameter in conducting a BMP test according to the Anaerobic Digestion Specialist Group of the International Water Association [7]. A ratio of 2:1 or greater of inoculum to substrate on a VS basis is recommended for BMP trials by Raposo et al. [8] to limit any inhibitory effects due to the chemical composition of the substrate such as inhibition associated with accumulation of ammonia and volatile fatty acids (VFA) [8]. All samples were tested for BMP in triplicate. A BMP test of the inoculum alone (referred to as a blank) was conducted in triplicate. The average methane yield from the blanks was subtracted from the samples of OFMSW with inoculum to accurately assess the BMP yields from the samples only. A triplicate BMP test was also carried out on cellulose for quality control as the maximum BMP from cellulose is known and can be compared with the BMP yield. The percentage volatile solids destroyed during the batch process was calculated as follows:
%VS destruction ¼ 100 ð1 ðVSf VSfb Þ=ðVSi VSib Þ
ð1Þ
where VSi is the amount of total input VS (g), VSf is the amount of total VS at the end of the BMP test (g), VSib is the amount of VS (g) in the inoculum (blank) at the beginning of the BMP test and VSfb is the amount of VS (g) in the inoculum (blank) at the end of the test. The Buswell equation was used to calculate the theoretical maximum methane potential [9].
n BMPTh ¼
2
þ 8a 4b 22400 ð12n þ a þ 16bÞ
ð2Þ
where n is the number of atoms of carbon; a is the number of atoms of hydrogen; b is the number of atoms of oxygen. The biodegradability index is the ratio of the measured BMP divided by the theoretical methane yield according to the Buswell equation and is used to assess the level of biodegradability of a substrate. 2.3. Source and characteristics of inoculum for BMP tests The inoculum for the BMP tests was obtained from a lab scale 300 L digester treating mostly cattle slurry and a small portion of
Fig. 1. Illustration of samples taken from the organic fraction of municipal solid waste (with and without garden waste).
309
J.D. Browne et al. / Applied Energy 128 (2014) 307–314 Table 1 Characterisation of OFMSW samples. Samples
Total solids (%)
Volatile solids (% TS)
Total carbon (% TS)
Total hydrogen (% TS)
Total nitrogen (% TS)
C:N
Household brown bin Rural with garden (RWG) Rural no garden (RNG) Urban with garden (UWG) Urban no garden (UNG)
33.4 (0.4) 30.6 (3.3) 25.66 (0.1) 31.0 (2.4)
82.3 88.4 73.6 93.8
43.3 44.9 41.3 46.5
5.9 6.6 5.2 7.3
2.7 3.1 2.6 3.7
(0.1) (0.2) (0.4) (0.1)
16 14.5 16 12.6
Commercial waste Commercial canteen summer (CCS) Commercial canteen winter (CCW)
32.8 (0.1) 23.8 (0.5)
92.6 (0.3) 90.0 (0.3)
49.0 (0.6) 48.2 (0.2)
7.0 (0.1) 7.0 (0.04)
3.4 (0.2) 3.6 (0.2)
14.4 13.4
Food processing Food processing bakery waste (FPBW) Food processing cheese waste (FPCW)
45.7 (0.4) 15.9 (0.1)
91.9 (0.6) 55.6 (0.3)
52.7 (0.2) 24.9 (0.1)
8.2 (0.05) 4.1 (0.03)
2.8 (0.2) 4.6 (0.03)
18.8 5.4
(0.2) (0.4) (0.4) (0.3)
(0.2) (0.2) (0.2) (0.2)
(0.1) (0.1) (0.1) (0.1)
All values are presented as mean and (standard deviation).
food waste operating at mesophilic temperatures (35 °C). After an incubation period of one week the inoculum had a pH of 7.9, total solids (TS) of 34.2 g VS kg1 and volatile solids (VS) content of 21.4 g VS kg1 after passing through a 2 mm sieve. Inoculum from both rounds was tested using cellulose as a standard control substrate (C12 H20 O10). The maximum theoretical methane yield from cellulose according to the Buswell equation is 415 L CH4 kg VS1. The specific methane yield produced from the cellulose was 371 ± 4 L CH4 kg VS1. This is almost 90% of the theoretical maximum indicating a healthy inoculum. 2.4. Kinetic modelling of BMP tests Two first order kinetic models were used to fit the cumulative methane production data from the BMP tests. Assuming first-order kinetics for the hydrolysis of particulate organic matter, the cumulative methane production can be described by Eq. (3):
YðtÞ ¼ Y m 1 expðkt Þ
ð3Þ
where Y(t) is the cumulative methane yield at digestion time t days (mL CH4 g VS1 added), Ym is methane potential of the substrate (mL CH4 g VS1 added), k is methane production rate constant (first order disintegration rate constant) (day1), t is the time (days). The duration of the lag phase is also an important factor in determining the efficiency of anaerobic digestion. The lag phase (k) can be calculated with the modified Gompertz model as described by [10] in Eq. (4) below:
Rmax e M ¼ P exp exp ðk tÞ þ 1 P
ð4Þ
where M is the cumulative methane yield at a given time (mL CH4 g VS1), P is the max methane potential (L CH4 kg VS1) from the BMP test, Rmax is the maximum methane production rate (L CH4 kg VS1 day1), e = 2.7183, k is the lag phase for methane production to begin (days), t is the time (days). A nonlinear least-square regression analysis was performed using Excel to determine k, Rmax, k, and the predicted methane yield. The predicted methane yield obtained from the regression analysis was plotted with the measured methane yield. The statistical indicators, Correlation coefficient (R2) and root mean square error (RMSE) were calculated to assess the goodness of fit [11]. 2.5. Statistical analysis The significance of differences in the average methane yields was determined by using single factor Analysis of Variance (ANOVA) in Excel software 2007. If the calculated F value was higher than the tabulated F value, the minimum significant difference (MSD) was calculated to judge whether two or more averages
were significantly different or not (Tuckey test). MSD was calculated as P = 0.05 (MSD0.05) [12]. 2.6. Continuous AD trial on canteen food waste The continuous trial was carried out in a continuously stirred tank reactor (CSTR) with a total volume of 5 L (working volume of 4 L) and ran for a period of 25 weeks. The reactor was maintained at a temperature of 37 ± 1 °C and was continuously stirred at a rate of 100 rpm. The reactor was constructed out of thick walled plastic with a vertically mounted stirring mechanism as shown in Fig. 2. The reactor was placed inside a coiled copper pipe frame which was heated by a thermo-circulator. Biogas flow was measured using a tipping bucket mechanism whereby the number of tips was recorded and multiplied by the calibrated gas volume of the tipping bucket (78 mL per tip). Biogas was sampled downstream of the gas flow tipping meter in 1 L Tedlar gas bags and analysed for methane, carbon dioxide and hydrogen sulphide. The continuous trial was operated for 176 days using commercial canteen food waste from the same collection as the commercial canteen summer (CCS) sample in the BMP trials. The continuous system was started at a moderate organic loading rate (OLR) of 2 kg VS m3 day1. The hydraulic retention time (HRT) was initially set at 30 days. This was achieved by adding a portion of digestate back in with the input feed keeping the total solids content of the input feed to 10% which facilitated easy stirring of the digester contents. The reactor was maintained at this OLR for 3 HRTs (period 1). The first HRT incorporated the start up and acclimatisation period. Volatile fatty acids (VFA), total alkalinity (TA), total ammonia nitrogen (TAN) were measured weekly (using methods described in Section 2.7 Analytical methods) and pH measured daily. 2.7. Analytical methods Total solids and volatile solids were determined gravimetrically following the standard methods [13]. The biogas composition in the continuous trial was measured by infra red gas analyser (Status Scientific Control I-R biogas analyser). The instrument was calibrated before the commencement of the trial and showed an accuracy of ±1% when tested weekly on a standard mixture of 65% methane and 35% carbon dioxide provided by BOC specialty gases. All methane yields were adjusted to standard temperature of 273 K and 1 atmosphere (1013 hPa). Volatile organic acids and total alkalinity were measured using the Nordmann titration method (1977) [14] using 0.1 N sulphuric acid and a Titronic Universal Titrator. The pH of the digestate was measure daily using a Jenway 3510 pH meter. Total ammonia was measured using the Hach NH3–N vials and spectrophotometer DR 3900.
310
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
Fig. 2. Continuously stirrer tank reactor (5 L) used for continuous trial.
3. Results and discussion 3.1. Results from the BMP tests Table 2 shows the range of BMP yields from samples tested. The commercial waste samples exhibited a much higher methane yield and greater portion of biodegradable organic matter than the household waste samples. In particular the household waste streams which consisted of mostly garden waste had a much lower biodegradability index than waste streams without garden waste. The BMP result for the cheese waste was much lower than expected. In a previous study by Browne et al. [6] a sample of cheese processing treatment sludge from a different location yielded 461 L CH4 kg VS1. This demonstrates that the type of existing biological waste treatment processes at dairy plants can produce waste sludge with hugely different biomethane potential. A one way Anova analysis showed a large statistical difference between the biomethane potential results depending on the source of OFMSW (F7,16 = 332.6, P < 0.001); where there are 7° of freedom between samples and 16° of freedom within samples. Multiple comparisons testing using the Tuckey test on the mean BMP values of the 8 waste streams tested, resulted in a total of 28 comparative tests which checked if the difference between means was greater than the minimum significant difference (MSD = 34.4 mL g VS1, P < 0.05). The Tuckey test revealed that there was a significant difference in average biomethane potential between 25 out of 28 comparisons. Interestingly, in the household waste stream there was no significant difference between urban and rural samples that came from a similar collection system (P > 0.05). However there was a significant difference in methane potential depending whether garden waste was included or not. For example, samples without garden waste gave higher methane yields than samples which included garden waste. Canteen waste samples gave significantly higher BMP yields than from household waste streams. There was a small but significant difference between canteen waste samples depending on the season. Samples taken from the same waste collection run in summer (2012) gave 9% higher BMP yields than autumn/winter (2012). In the food processing stream
bakery waste samples gave vastly greater methane yields (529 mL CH4 g VS1) than from cheese waste activated sludge (188.5 mL CH4 gVS1). Interestingly the bakery waste sample did not differ significantly from the canteen waste summer. The results from the commercial waste samples are similar to previously reported BMP yields from canteen food waste (480–530 L CH4 kg VS1) by Browne and Murphy [15]. 3.2. Kinetic study results The results of the kinetics analysis using the first order kinetic model and the modified Gompertz model are summarised in Tables 3a and 3b respectively. The first order kinetic model gave k values ranging from 0.12 to 0.17 day1 for household samples, 0.07–0.09 day1 for commercial samples and 0.08–0.13 day1 for food processing samples. The commercial food waste samples had higher percentages of proteins and lipids which take longer to digest than carbohydrates therefore resulting in lower k values Vavilin et al. [16]. The modified Gompertz model showed a lag time of 1.2 and 3 days for all samples tested. The time taken to reach 90% of the maximum BMP value was shown to range from 9 to 15 days indicating that all OFMSW substrates were readily degradable. Both models exhibited a good fit when plotted against the measured data with the coefficient of determination (R2) ranging from 0.93 to 0.95 for the first order model and 0.99 for the Gompertz model. The RMSE ranged from 10.7 to 48.8 mL CH4 g VS1 for the first order model while the Gompertz model gave lower values of over 0.7–9.9 mL CH4 g VS1. Both models can be used to predict the maximum methane potential. The modified Gompertz model gave slightly lower predicted maximum BMP yields than the measured data ranging from 0.8% to 9.3% while the first order model generally gave higher predicted methane yields than measured ranging from 1.8 to +19.2%. In 87.5% of cases the model Gompertz gave a more accurate predicted max biomethane yield than the first order equation. Based on the statistical indicators (RMSE and R2) the modified Gompertz model was found to demonstrate the best fit for the samples tested. The cumulative methane yields of the BMP tests are shown in Fig. 3.
311
J.D. Browne et al. / Applied Energy 128 (2014) 307–314 Table 2 Spectrum of food waste – samples collected. Source
BMP (30 days) (mL CH4 g VS1)
Theoretical BMPa (mL CH4 g VS1)
Biodegradability index
Volatile solids destruction as measured (% VS)
Rural with garden (RWG) Rural no garden (RNG) Urban with garden (UWG) Urban no garden (UNG) Commercial canteen summer (CCS) Commercial canteen winter (CCW) Food processing bakery waste (FPBW) Food processing cheese waste (FPCW)
274.1 367.8 296.7 343.7 534.5 490.9 529.2 188.5
577 566 625 564 620 620 696 530
0.48 0.65 0.47 0.61 0.86 0.79 0.76 0.36
47 69 51 60 81 80 81 42
(4.6)b (6.2)c (6.1)b (2.7)c (5.0)e (4.8)d (25.4)e (1.2)a
a, b, c, d, e = Levels of Tuckey multiple comparisons testing between sample means. Samples with the same letter are not significantly different (P > 0.05). Results are indicated as a mean with standard deviation is in brackets. a Theoretical BMP calculated from the Buswell equation [9].
Table 3a Results of BMP kinetic analysis using the first order kinetic equation. Sample Household waste Rural with garden Rural no garden Urban with garden Urban no garden Commercial waste Canteen summer Canteen winter Food processing waste Bakery waste Cheese process waste
RMSE (mL CH4 g VS1)
R2
k (day1)
+6.5 +5.5 +1.8 +7.4
21.6 28.7 18.0 27.2
0.93 0.93 0.95 0.93
0.12 0.14 0.17 0.12
603 585
+12.8 +19.2
40.2 36.7
0.94 0.95
0.09 0.07
623 185
+17.7 1.8
48.8 10.7
0.92 0.95
0.08 0.13
BMP measured (mL CH4 g VS1)
BMP predicted (mL CH4 g VS1)
274.1 367.8 296.7 343.7
(4.6) (6.2) (6.1) (2.7)
292 388 302 369
534.5 (5.0) 490.9 (4.8) 529.2 (25.4) 188.5 (1.2)
Difference (%)
Table 3b Results of BMP kinetic analysis using the modified Gompertz equation. BMP measured (mL CH4 g VS1)
BMP predicted (mL CH4 g VS1)
Difference (%)
R2
RMSE (mL CH4 g VS1)
Lag phase (days)
T90 (days)
Household waste Rural with garden Rural no garden Urban with garden Urban no garden
274.1 367.8 296.7 343.7
(4.6) (6.2) (6.1) (2.7)
268 363 288 338
2.2 1.3 2.9 1.7
0.99 0.99 0.99 0.99
2.9 4.1 0.8 3.3
2.2 2.0 1.3 2.2
11 10.2 8.7 11.3
Commercial waste Canteen summer Canteen winter
534.5 (5.0) 490.9 (4.8)
530 484
0.8 1.4
0.99 0.99
3.4 0.7
2.3 2.5
13.4 15.3
529.2 (25.4) 188.5 (1.2)
528 171
0.2 9.3
0.99 0.97
9.9 0.7
3.0 1.2
14.3 10.8
Sample
Food processing waste Bakery waste Cheese process waste
Results are indicated as a mean with standard deviation is in brackets.
3.3. Results from continuous digestion of canteen food waste 3.3.1. Specific methane yields in period 1 The reactor was maintained at an OLR of 2 kg VS m3 day1 for 3 HRTs (period 1). The first HRT incorporated the start up and acclimatisation period. By the end of the first HRT the system had reached a steady state of methane production. Methane yields from the second and third HRT were used to calculate average specific methane yield for period 1 (OLR of 2 kg VS m3 day1) which was 560.1 ± 29.3 mL CH4 g VS1 added. The standard deviation in the second and third HRT was only 5% of the total yield and clearly showed that the reactor was in steady state. The weekly average specific methane yield is shown in Fig. 4(a). The daily percentage methane in the biogas is shown in Fig. 4(b). In the start-up period the percentage methane increased from 40.4% to 60% over the first
30 days with the weighted average methane percentage in the biogas remaining at 60 ± 1.3% for period 1. 3.3.2. Specific methane yields in period 2 After completing 3 HRTs of 30 days at the initial feeding rate, the OLR was increased to 3 kg VS m3 day1 at day 99. By increasing the OLR to 3 kg VS m3 day1 the HRT was reduced to 21 days as the solids content of the input material was kept at 10% TS by recirculation of an increased amount of digestate. The OLR was maintained at 3 kg VS m3 day1 for 2 HRTs (42 days). The average specific methane yield (SMY) for period 2 was 484 ± 72.0 mL CH4 g VS1. This is a reduction of about 13% from the previous SMY in period 1. The standard deviation in period 2 is approximately 15% of the average SMY and shows that there was greater fluctuation in daily gas production at the higher OLR of 3 kg VS m3 day1.
312
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
3.3.4. Conversion of volatile solids to gas To assess the conversion of VS to gas Eq. (5) taken from Kang and Wieland, [17] is used:
MR ¼ LN ðð16 CH4 %Þ þ ð44 CO2 %ÞÞ=22:413
ð5Þ
where MR is the daily average mass of volatile solids removed (g VS); LN is the average daily normalised biogas volume (L) at standard temperature and pressure (STP); CH4 % is the methane content in the biogas; CO2 is the carbon dioxide content in the biogas; There are 22.413 L per mole of gas at STP. According to Eq. (5) the average removal of VS in period 1 was 84% with a HRT of 30 days. This decreased to 72% in period 2 with a HRT of 21 days and further reduced to 54% in period 3 with a HRT of 17 days. The average concentration of total solids in the digestate increased from 5.1 ± 0.5% TS in period 1 to 5.5 ± 0.3% TS in period 2 and 6.7 ± 0.9% TS in the final period. This indicates that reducing the HRT also reduces the degradation of volatile solids. However the large drop in specific methane yield towards the end of the trial may not be entirely as a result of the reduced HRT as signs of process instability emerged towards the end of the trial at an OLR of 4 kg VS m3 day1. 3.4. Monitoring process stability in continuous trial
Fig. 3. BMP cumulative methane yields for (a) household samples and (b) commercial & processing waste samples.
The weighted average methane content in the biogas increased to 61.5 ± 2.8% in period 2. 3.3.3. Specific methane yields in period 3 On day 142 the OLR was further increased to 4 kg VS m3 day1 which resulted in a reduced HRT of 17 days. The trial was completed on day 176. The average SMY in the final period was 381.5 ± 52.0 mL CH4 g VS1 which was a 21% decrease in SMY from period 2 and a 32% decrease from period 1. The average methane content was 60.7 ± 3.6%.
Fig. 4. (a) Weekly average specific methane yield and (b) daily methane percentage.
During the continuous trial the total volatile fatty acids (VFA), total alkalinity (TA), pH and total ammonia nitrogen (TAN) were monitored to assess the stability of the digestion process. The average results from the three time periods are shown in Table 4. In Period 1 (OLR 2 kg VS m3 day1) the concentration of total VFAs was 1128 ± 281 mg Aceq L1. A small increase was observed during Period 2 (OLR of 3 kg VS m3 day1) with an average of 1511 ± 77 mg Aceq L1. The concentration of VFAs rose sharply towards the end of the trial during Period 3 (OLR of 4 kg VS m3 day1) as shown in Fig. 5(a), with an average of 2595 ± 750 mg Aceq L1. Single factor ANOVA showed that there was a significant difference between VFA concentrations between periods (P < 0.001). The sharp increase in VFA concentration in period 3 indicated that the methane producing microbes were stressed and were not fully utilising the VFAs produced by the acidogenic/ acetogenic bacteria. The average total alkalinity for the period 1 was 8093 ± 970 mg CaCO3 L1. This increased to 9830 ± 159 mg CaCO3 L1 in period 2 and 10,230 ± 185 mg CaCO3 L1 in period 3. The ratio of VFA/TA is often used to assess the stability of the AD process. A ratio of 0.4 or less indicates that the process is stable while ratios over 0.8 indicate organic overloading and process instability. During the trial the VFA/TA ratio remained below 0.4, however it is clear that even though the ratio was within stable limits, the specific methane yields (SMY) were in sharp decline in the final period. Linear regression analysis showed that there was a relationship between increasing concentrations of FAN and decreasing SMY. The linear relationship gave the equation; SMY = 0.22 * FAN + 594.5 (r2 = 0.65). Increasing FAN also had a linear relationship with increasing VFA concentrations with VFA = 1.85 * FAN + 690 (r2 = 0.78). This indicates that increasing concentrations of FAN disrupted the uptake of VFAs by the methanogens and resulted in accumulation of VFAs which ultimately lead to a decrease in SMY. During the trial the pH increased from an average of 7.7 ± 0.1 in period 1 to 8.1 ± 0.1 in the final period. The high pH is of concern when combined with high levels of TAN as the relationship between ionised ammonium (NH+4) and unionised (free) ammonia (FAN) is pH and temperature dependent. The fraction of FAN is dependent on TAN, temperature and pH and is described by Eq. (6) taken from Angelidaki and Ahring [18].
J.D. Browne et al. / Applied Energy 128 (2014) 307–314 Table 4 Summary of results from continuous AD trial of canteen food waste. Period 3
1
OLR (kg VS m day ) HRT (days) SMY (mL CH4 g VS1) % CH4 (weighted average) % VS conversion to gas % TS (digestate) % VS (digestate) pH TAN (mg N L1) FAN (mg N L1) VFAs (mg Aceq L1) Alkalinity (mg CaCO3 L1) VFA/TA
1
2
3
2 30 3 560.1 (29.3) 60.1 (1.3) 84 5.1 (0.5) 3.5 (0.3) 7.7 (0.13) 3543 (525) 237 (50) 1128 (281) 8093 (970) 0.14
3 21 2 483.9 (72.0) 61.5 (2.8) 72 5.5 (0.3) 3.9 (0.2) 7.9 (0.14) 5342 (485) 433 (185) 1511 (77) 9830 (159) 0.15
4 17 2 381.5 (52.0) 60 (3.6) 54 6.7 (0.9) 4.7 (0.7) 8.1 (0.11) 7205 (280) 992 (83) 2595 (750) 10,230 (185) 0.25
Results are indicated as a mean with standard deviation in brackets.
313
pH values. The concentration of TAN increased linearly for the duration of the trial with final concentrations in excess of 7000 mg N L1, as shown in Fig. 5(b). The relatively high concentration of TAN coupled with a pH of about 8, resulted in high concentrations of FAN of approximately 1000 mg N L1. This is a very high concentration of FAN and is considered to be in the toxicity range for methane production by Chen et al., [19]. Several threshold values are reported for TAN and FAN causing inhibition of the biogas process in anaerobic reactors. For example, a threshold value around 1000 mg N L1 FAN is reported by Chen et al. [19], Gerardi [20] and Hansen et al. [21]. Above this concentration, inhibition and an apparent decrease in the specific growth rate in batch cultures at pH 8.0 (reactor pH) is found. Although free ammonia can inhibit anaerobic digestion, relatively high TAN concentrations can be tolerated Chen et al. [19]. 100% Inhibition is reported in the range of 8–13 g N L1 depending on the condition of acclimatisation and the pH of the system. Despite the long delay in methane production, it was reported that methanogens present in digested sewage sludge could tolerate up to 5 g L1 NH3–N (TAN) and the methane producing biomass was able to acclimatise to such high ammonium concentration Chen et al. [19]. Angelidaki and Ahring [18] reported that acetate-utilising bacteria adapted to NH3 at a free ammonia (FAN) concentration of up to 700 mg N L1, while lower free ammonia concentrations (100–150 mg N L1) have been reported for initial inhibition of an unadapted process. Banks and colleagues [22] reported high concentrations of TAN and FAN at high organic loading rates using source separated food waste. They showed that at elevated levels of TAN the acetoclastic methanogens were virtually nonexistent with methane production coming from the hydrogenotrophic route. To overcome the inhibitory effects of high levels of ammonia the addition of trace elements such as iron, cobalt, selenium and molybdenum were successfully shown to improve methane yields at high organic loading rates (e.g. 5 kg VS m3 day1) [22]. 3.6. Comparison of specific methane yields from batch and continuously fed systems
Fig. 5. (a) Monitoring of total alkalinity (TA) and volatile fatty acids (VFA) (b) total ammonia nitrogen (TAN) and free ammonia nitrogen (FAN) and (c) pH in the continuous trial.
Free Ammonia ðFANÞðmg=LÞ
The specific methane yield (SMY) produced during period 1 of the continuous trial was relatively high in comparison to other reported methane yields from food waste. The highest average SMY of 560 ± 29.3 mL CH4 g VS1 was achieved at an OLR of 2 kg VS m3 day1 and HRT of 30 days. This is 90.3% of the Buswell equation value. It is however 7% higher than the average BMP result from the same sample. This indicates that at moderate organic loading rates a continuous AD process may equal or even exceed methane yields from the BMP test. This may be due to acclimatisation of the inoculum with time. Thamsiriroj and Murphy [23] also recorded higher SMYs in continuous digestion than in BMP mode. Zhang and colleagues [24] achieved 425 L CH4 kg VS1 from continuous digestion of source segregated food waste at an OLR of 2 kg VS m3 day1 which was slightly lower than the BMP results of the same material (445–456 L CH4 kg VS1).
¼ TAN ðmg=LÞ ðð1 þ ðð10Þ^ ðpHÞÞ=ð10Þ^ ð0:09018 þ ð2729:92=ðT þ 273ÞÞÞÞÞ1
ð6Þ
As shown in Fig. 5 even slight changes in pH can have a large effect on the proportion of FAN. If pH rose from 7 to 8 according to Eq. (6) this would result in a 10-fold increase in the concentration of FAN. 3.5. The inhibitory effects of high ammonia concentrations Total ammonia nitrogen (TAN) contributes to the buffering capacity of the system but can be toxic to methanogens at higher
4. Conclusions The characteristics of OFMSW can vary largely depending on the source and type of collection system with BMP values of between 274 and 535 mL CH4 g VS1. A continuous trial on commercial food waste produced an average of 560 ± 29 mL CH4 g VS1 at a moderate OLR of 2 kg VS m3 day1 with a HRT of 30 days. At higher OLRs (4 kg VS m3 day1) increasing concentrations of VFAs (2595 mg L1) coupled with high concentrations of free ammonia (952 mg L1) led to a greatly reduced average specific methane yield (344 mL CH4 g VS1).
314
J.D. Browne et al. / Applied Energy 128 (2014) 307–314
Acknowledgements Researchers were funded by the Irish Research Council for Science, Engineering and Technology (IRCSET), Science Foundation Ireland (SFI) and Bord Gais Energy (BGE). Laboratory equipment was funded by Bord Gais Networks (BGN). The authors would like to thank Mr. Ronan Beasley of Acorn Recycling Ltd for providing the samples of OFMSW.
References [1] McCoole F, Kurz I, McDonagh M, Derham J, O’Neill D. National waste report 2010. In: Environmental protection agency I, editor. Environmental Protection Agency, Ireland; 2012. [2] EC. Council Directive 1999/31/EC of 26 April 1999 on the landfill of waste. Official Journal of the European Union; 1999. [3] EC. Directive 2008/98/EC of the European Parliament and of the Council of 19 November 2008 on waste and repealing certain Directives. In: Commission E, editor; 2008. [4] Lukehurst C, Frost P, Al Seadi T. Utilisation of digestate from biogas as biofertiliser, International Energy Agency (IEA) Bioenergy Task 2010;37. Available at: www.iea-biogasnet/technical-brochures.html. [5] VDI. Fermentation of organic materials – characterisation of the substrate, sampling, collection of material data, fermentation tests. Characterization of substrates. Dusseldorf, Germany: Verein Deutscher Ingenieure; 2006. p. 11–9. [6] Browne J, Allen E, Murphy J. Evaluation of the biomethane potential from multiple waste streams for a proposed community scale anaerobic digester. Environ Technol 2013:1–12. [7] Angelidaki I, Alves M, Bolzonella D, Borzacconi L, Campos JL, Guwy AJ, et al. Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays. Water Sci Technol 2009:927–34. [8] Raposo F, Banks CJ, Siegert I, Heaven S, Borja R. Influence of inoculum to substrate ratio on the biochemical methane potential of maize in batch tests. Proc Biochem 2006;41:1444–50.
[9] Buswell AM, Neave SL. Laboratory studies of sludge digestion. Illinois Division of State Water Survey, Bulletin No 30. URBANA, ILLINOIS1930. [10] Kafle GK, Kim SH. Anaerobic treatment of apple waste with swine manure for biogas production: batch and continuous operation. Appl Energy 2013;103: 61–72. [11] Brown AM. A step-by-step guide to non-linear regression analysis of experimental data using a Microsoft Excel spreadsheet. Comput Methods Prog Biomed 2001;65:191–200. [12] Wheater CP, Cook PA. Using statistics to understand the environment. New York: Routledge; 2000. [13] APHA. Standard Methods for the Examination of Water and Wastewater. (21st ed.) American Public Health Association, American Water Works Association, Water Environment Federation (2005). Washington, DC; 2005. [14] Nordmann W. Die Überwachung der Schlammfaulung. Beilage zur Korrespondenz Abwasser; 1977. p. 3/77. [15] Browne JD, Murphy JD. Assessment of the resource associated with biomethane from food waste. Appl Energy 2013;104:170–7. [16] Vavilin VA, Fernandez B, Palatsi J, Flotats X. Hydrolysis kinetics in anaerobic degradation of particulate organic material: an overview. Waste Manage 2008; 28:939–51. [17] Kang H, Weiland P. Ultimate anaerobic biodegradability of some agroindustrial residues. Bioresource Technol 1993;43:107–11. [18] Angelidaki I, Ahring BK. Thermophilic anaerobic digestion of livestock waste: effect of ammonia. Appl Microbiol Biotechnol 1993;38:560–4. [19] Chen Y, Cheng JJ, Creamer KS. Inhibition of anaerobic digestion process: a review. Bioresource Technol 2008;99:4044–64. [20] Gerardi MH. Toxicity. The microbiology of anaerobic digesters. John Wiley & Sons, Inc.; 2003. p. 105–15. [21] Hansen KH, Angelidaki I, Ahring BK. Anaerobic digestion of swine manure: inhibition by ammonia. Water Res 1998;32:5–12. [22] Banks CJ, Zhang Y, Jiang Y, Heaven S. Trace element requirements for stable food waste digestion at elevated ammonia concentrations. Bioresource Technol 2012;104:127–35. [23] Thamsiriroj T, Murphy JD. Modelling mono-digestion of grass silage in a 2stage CSTR anaerobic digester using ADM1. Bioresource Technol 2011;102: 948–59. [24] Zhang Y, Banks CJ, Heaven S. Co-digestion of source segregated domestic food waste to improve process stability. Bioresource Technol 2012;114:168–78.