- Email: [email protected]
Effects of substrate concentration on methane potential and degradation kinetics in batch anaerobic digestion
Bioresource Technology 194 (2015) 240–246
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Effects of substrate concentration on methane potential and degradation kinetics in batch anaerobic digestion Bing Wang a,⇑, Sten Strömberg a, Chao Li a, Ivo Achu Nges a, Mihaela Nistor b, Liangwei Deng c, Jing Liu a,b a
Department of Biotechnology, Lund University, Getingevägen 60, SE-221 00 Lund, Sweden Bioprocess Control Sweden AB, Scheelevägen 22, SE-223 63 Lund, Sweden c Biogas Institute of Ministry of Agriculture, 610041 Chengdu, China b
h i g h l i g h t s High substrate concentration led to high methane yields. Nutrient/buffer-diluted substrate showed a positive effect on degradation rate. Neither dilution liquid nor substrate concentration had a strong effect on lag time. Dilutions should be avoided at substrate concentration below 10 g VS/L.
a r t i c l e
i n f o
Article history: Received 26 May 2015 Received in revised form 10 July 2015 Accepted 11 July 2015 Available online 17 July 2015 Keywords: Anaerobic digestion Biochemical methane potential Substrate load Substrate concentration Kinetics
a b s t r a c t In this study, two experiments were conducted to evaluate the impact of substrate concentrations on methane potential and degradation kinetics of substrate. The biochemical methane potential (BMP) tests in Experiment I were performed at a constant inoculum to substrate ratio (ISR), whereas, different ISRs were applied in Experiment II. Results obtained from Experiment I revealed that methane potential of substrate increased at a saturating trend with higher substrate concentrations, and could differ by up to 30% between the lowest and highest investigated concentrations. The results of Experiment II verified the results of Experiment I, and further showed that this trend also occurs when the substrate concentration is regulated with ISRs. In contrast, substrate concentration had no significant impact on the degradation kinetics. It was concluded that dilutions should be avoided when the substrate concentration is lower than 10 g VS/L in order to avoid underestimations of methane potential from BMP test. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The methane potential of biomass, which can be determined by a traditional biochemical methane potential (BMP) test, is a key parameter for the design, economy and management in full-scale implementations of anaerobic digestion (AD). In recent years, a high number of scientific articles related to BMP tests of a variety of substrates have been published (Bauer et al., 2009; Møller et al., 2004; Nkemka and Murto, 2013). However, the results from such tests are variable and difficult to compare. The varying results are caused by differences in both instrumentation and protocols, as well as by different experimental conditions. For instance, the pH, headspace volume, mixing options, inoculum to substrate ratio (ISR), inoculum dilutions, but also substrate load (i.e. initial
⇑ Corresponding author. Tel.: +46 462223672; fax: +46 462224713. E-mail address: [email protected] (B. Wang). http://dx.doi.org/10.1016/j.biortech.2015.07.034 0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.
substrate concentration) can differ among different tests (Angelidaki et al., 2009). Particularly, the substrate concentration has been considered as an important factor which influences the efficiency of the anaerobic digestion process (Lianhua et al., 2010; Sánchez et al., 2001). At very low concentrations, there is a risk that microorganisms may exhibit a low metabolic activity and very low quantities of biogas will be produced. In contrast, if the substrate concentration is too high, that might lead to an overload situation in which intermediate compounds may accumulate, resulting in inhibition of the process (Tanimu et al., 2014; Zhang et al., 2014). So far, there has not been so much focus on the substrate concentration in a BMP test, but it is fairly common to suggest diluting the substrate to limit overloading and influence from possible toxic components (Angelidaki et al., 2009). However, the choice of dilution liquid used to adjust the substrate concentration is important. Using a medium containing the necessary nutrients, trace elements and vitamins are recommended to avoid inhibition in microbial
241
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
growth (Angelidaki and Sanders, 2004; Raposo et al., 2012). It has also been shown that additions of these types of medium improve the methane potential of diluted inocula while undiluted ones are not affected (Brulé et al., 2013). Nonetheless, there are few guidelines recommending what substrate concentration to use in a BMP test to reach optimal results. One example is given by the German Standard (VDI 4630, 2006) which recommends using an inoculum concentration between 1.5% and 2.0% in volatile solids (VS) and an inoculum to substrate ratio (ISR) P 2, which translates to a substrate concentration of 610 g VS/L. However, no exact concentration requirement is given and without a lower boundary there is a risk that too low concentrations are used. This uncertainty is also reflected in the conditions used in many BMP assays where the substrate concentration can differ profoundly when different types of inocula and substrates have been used (Raposo et al., 2011). For example, few studies have evaluated the impact on the methane potential using different substrate concentrations at a fixed ISR while several have investigated it at variable ISRs. Maya-Altamira et al. (2008) found that the methane potential of various types of industrial wastewaters increased slightly with increasing ISR (COD based). The highest methane potentials were observed at ISR P 2 and it was assumed that the lower values at the lower ISR were caused by overload situation, in which volatile fatty acid (VFA) or acetate acid was built up, resulting in a slow AD process inhibition. Hashimoto (1989) investigated a large number of ISRs for ball-milled wheat straw and found that the methane potential drastically decreased at an ISR below 0.25. The general conclusion from these studies is that a too low ISR leads to inhibition and that this parameter should be kept above a certain threshold, most commonly ISR P 2 (Raposo et al., 2008; VDI 4630, 2006). However, these studies are almost exclusively focused on identifying points where the ISR is high enough to prevent inhibition, whereas few studies have investigated whether the substrate concentration can be too low. The hypothesis tested in this study is that the substrate concentration can influence the specific methane potential as well as the degradation kinetics. In this light, various experiments were performed to investigate the influences of substrate concentrations on methane production and degradation kinetics. In all, five different substrate concentrations were evaluated at a fixed ISR of 2, as well as the effect from choice of dilution liquid (distilled water and nutrient/buffer solution). Moreover, further tests were also performed at both fixed and varying ISRs.
Table 1 Preparation of substrate/inoculum/diluent mixtures used in Experiment I. Substrate loads (g VS)
Undiluted setup Substrate conc. (g VS/L)
Liquid vol. (mL)
Substrate conc. (g VS/L)
Liquid vol. (mL)
1 2 3 5 6
15.8 15.8 15.8 15.8 15.8
63 127 190 317 381
2.5 5.0 7.5 12.5 15.0
400 400 400 400 400
Diluted setups
in triplicates with an individual set of blanks with no sample. All tests were prepared in 500 mL standard bottles (Schott, Germany) and mixed intermittently (160 rpm, stirrers were on for 5 min and off for 25 min continuously throughout the test) to ensure a good mass transfer. In Experiment II, the following BMP studies with varying substrate concentrations were performed in order to verify the results from Experiment I: (i) cellulose was used as substrate, the concentration was adjusted by varying the ISR instead of diluting the sample, (ii) substrate concentration of soymilk and sugar was adjusted both at a fixed ISR using dilution (diluted with distilled water) and different ISRs. The total liquid volume in each bottle was 400 mL (Table 2). 2.2. Inoculum and substrate The inoculum used to carry out the BMP tests in Experiment I was collected from a mesophilic sewage treatment plant in Källby, Sweden, which was characterized by a VS content of 3.21% (w/w), pH of 8.03 and partial alkalinity (PA) of 6807 mg/L. The anaerobic inoculum (1.07% VS) used to carry out the BMP tests in Experiment II was collected from a mesophilic biogas plant (Ellinge, Sweden), which receives municipal wastewater (20%) and vegetable residue from the food industry (80%). In order to decrease the background biogas production, the inocula were pre-incubated at 37 °C for 5 days prior to the BMP tests (ISO-11734, 1995). Microcrystalline cellulose (Alfa Aesar, Germany, 96.1% VS) was used as a standard substrate in both experiments, and another mixture of soymilk and sugar (1:1 based on VS %) was used as substrate in Experiment II. Soymilk and sugar had a VS content of 9.98%.
2. Methods
2.3. Analytical methods
2.1. Experimental setup
2.3.1. Characterization of inoculum and substrate The total solids (TS) and VS of inoculum and substrate were determined according to standard protocols (APHA, 1995). The pH and partial alkalinity (PA) were measured using a TitraLab™
Two experiments were performed to evaluate the influences of substrate concentration on BMP test. In Experiment I, three different ways of varying the substrate load in BMP test were investigated: (i) dilution with a nutrient/buffer solution, (ii) dilution with distilled water and (iii) adjusting the liquid volume without dilution. For the two setups including dilution, after the addition of inoculum and substrate, either distilled water or nutrient/buffer solution was added to reach 400 mL; the corresponding blank samples contained inoculum diluted with distilled water or nutrient/buffer solution respectively. The nutrient/buffer solution was prepared according to Brulé et al. (2013). For each series, five different substrate loads at a fixed ISR of 2 (based on VS) were studied. The substrate concentrations for the diluted and undiluted set-ups can be observed in Table 1. In total 15 unique BMP tests (i.e. 5 different substrate loads with three types of dilution methods) were performed with cellulose as substrate and the inoculum from Källby as seeding sludge. Each of the 15 tests was performed
Table 2 Substrate concentrations and ISR used in the BMP tests in Experiment II. Cellulose
a
Soymilk and sugar
Substrate loads (g VS)
Substrate conc. (g VS/L)
ISRa
Substrate loads (g VS)
Substrate conc. (g VS/L)
ISR
0.52 1.08 2.12 4.20
1.3 2.7 5.3 10.5
1 2 4 8
0.8 1.2 1.6 0.8 1.2 1.6 1.7 2.4
2.0 3.0 4.0 2.0 3.0 4.0 4.2 6.1
2.0 2.0 2.0 6.4 4.2 3.1 3.0 2.0
Inoculum to substrate ratio.
242
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
80 titrator (Radiometer, Copenhagen, Denmark) as reported in another study (Nges and Liu, 2009). 2.3.2. Determination of methane production All the BMP tests were carried out in triplicates with the Automatic Methane Potential Test System II (AMPTS II, Bioprocess Control, Sweden AB) at 37 °C for 30 days. The experimental protocol was performed according to Badshah et al. (2012). At the end of the process, a report presented the normalized (standard temperature and pressure, STP: 273.15 K, 101.325 kPa; compensation of the water vapor content) methane flow rate and accumulative methane volume. 2.3.3. Kinetics A traditional first-order model (Eq. (1)) with a lag-time parameter was used to evaluate the kinetic degradation profile of the substrate (Grieder et al., 2012).
BMPðtÞ ¼ BMPmax ð1 expðk ðt hÞÞÞ
ð1Þ
In Eq. (1) BMP (t) is the methane potential (NmL/g VS) at time t (day), BMPmax is the maximum or ultimate methane potential (NmL/g VS) of the substrate, k is the rate constant or hydrolysis rate constant (1/day) and h is the lag time constant (day). 2.3.4. Statistical calculations Grubbs’ test was used to check and eliminate outliers in BMP tests. One-way and two-way ANOVA was calculated using the MATLAB function anovan at 95% confidence. The uncertainty of the kinetic parameters was estimated in a similar way as presented by Batstone et al. (2003). In order to avoid three-dimensional datasets, that would be difficult to visualize, one of the three model parameters was always kept at its optimal value while the remaining two were varied to determine the two-dimensional surface of the parameter uncertainty. The optimal parameter values were estimated with lsqcurvefit in MATLAB (v.8.1.0.604) and the surface of parameter uncertainty was estimated by using a similar approach as presented by Lobry et al. (1991). 3. Results and discussions 3.1. Results obtained from Experiment I 3.1.1. pH and partial alkalinity The pH and PA of the liquid in the reactor before and after anaerobic digestion for the different scenarios in Experiment I are presented in Table 3. As seen, diluting the liquid mixture changes the pH in the reactor. The undiluted samples showed a high pH at the start which dropped during the test, with slightly stronger effect on the pH at higher substrate loads. This decrease in pH is expected as increasing amounts of carbon dioxide is being dissolved when the equilibrium conditions with the headspace changes from the produced biogas. The slightly stronger effects at higher substrate loads (higher liquid volumes) could also be explained by this theory as the composition of the headspace gas changes faster at smaller headspace volumes. Dilution with both distilled water (pH 8.19) and nutrient/buffer solution (pH 6.40) had a rather substantial effect on both the pH and PA of the evaluated series. Using DW led to a significant decrease in PA for the higher dilutions that also led to lower pH at the end of the tests because of its poor buffer capacity. Interestingly, similar effects were observed for nutrient/buffer solution as well, indicating that the buffering capacity of this nutrient/buffer solution was insufficient. Furthermore, the low starting
pH at a substrate load of 1 g VS indicates that the acidic pH of the nutrient/buffer solution may lead to problems at high dilutions. 3.1.2. Methane potentials of cellulose The obtained methane potentials at different cellulose loads were presented in Fig. 1. In general, analysis of variance (ANOVA) of methane yields demonstrated that for loads of 1 g VS and 2 g VS, that respectively correspond to substrate concentrations of 2.5 g VS/L and 5 g VS/L, dilutions with distilled water or nutrient/buffer solution led to a decrease in methane yields with a load of 1 g VS showing significant (p = 0.05) reduction when diluted. At the lowest substrate load, i.e. 1 g VS, the undiluted sample had the highest methane production compared with the diluted samples. For loads of 3 g VS and 5 g VS (7.5 g VS/L and 12.5 g VS/L, respectively), dilutions did not significantly (p = 0.05) influence the methane yields. However, at a load of 6 g VS (15 g VS/L), dilution with either distilled water or nutrient/buffer solution led to significantly (p = 0.05) higher methane yields. It should be noted that at load 6 g VS (15 g VS/L), the methane yields of the diluted with distilled water and nutrient/buffer solution did not differ significantly (p = 0.05). For undiluted samples, however there is no significant (p = 0.05) difference in methane potential between the substrate loads irrespective of the dilutions. This is expected, as the only difference between the different substrate loads are varying headspace volumes. However, it should be pointed out that the measurement uncertainty from having a high headspace volume in relation to the liquid volume are considerably higher as a consequence of smaller sample volume, lower gas flows and more influence of the initial head space gas (Koch et al., 2015; Strömberg et al., 2014). As seen in Fig. 1, diluting the samples led to increased methane potentials with greater substrate loads and the maximal values of 396 (distilled water) and 394 (nutrient/buffer solution) NmL/g VS are both observed at the highest investigated load of 6 g VS (15 g VS/L). A two-way ANOVA test showed that there was a significant (p = 0.05) difference between methane potentials using different substrate concentrations but not a significant difference between series with different types of dilution liquid. This result implies that the substrate concentration has an apparent effect on the outcome of a BMP test while the choice of dilution liquid is of less importance. However, given the poor buffering capacity of nutrient/buffer solution, this conclusion is limited to this specific medium. The fact that the methane potentials at lower substrate concentrations are rather far from the theoretical one (415 NmL/g VS) suggests that too low substrate concentrations might introduce underestimation in this value. One possible explanation for the lower methane potentials at lower substrate concentrations is the slow growth rates of microorganisms and use cell carbon to satisfy the need for maintenance and metabolism (Seto and Alexander, 1985). Regardless on the mechanism causing this effect, the increase in methane potential with higher substrate concentrations implies that, besides the pH, the substrate concentration should also be considered and kept at similar values for comparable results. 3.1.3. Methane potential as a function of substrate concentration As indicated in Fig. 1, the methane potentials increase with increasing substrate loads, which is in agreement with other study. Raposo et al. (2006) studied on the influence of ISR on BMP of maize at a concentration range of 5–15 g VS/L, and the highest methane production was obtained at concentration of 15 g VS/L. If it is assumed that the methane potential is directly dependent on the substrate concentration and asymptotically reaches a maximal potential at higher concentrations, a saturation kinetics-based equation could be used to describe this effect. In this study a Monod type equation (Eq. (2)) was assessed for fitting the
243
B. Wang et al. / Bioresource Technology 194 (2015) 240–246 Table 3 The pH and partial alkalinity of liquid at different substrate loads. Substrate loads (g VS)
1 2 3 5 6 a b c
Sample diluted with DWa
Undiluted sample pHstart
pHend
PAcend
8.03 8.03 8.03 8.03 8.03
7.97 7.90 7.84 7.76 7.67
5774 6637 6805 6716 6140
(mg/L)
Sample diluted with NBSb
pHstart
pHend
PAend (mg/L)
pHstart
pHend
PAend (mg/L)
8.06 7.91 7.87 7.75 7.75
6.84 7.24 7.31 7.48 7.64
1405 2646 3906 6203 6358
6.71 7.65 7.69 7.77 7.80
7.02 7.05 7.37 7.52 7.63
3323 4180 4894 6509 7818
Distilled water. Nutrient/buffer solution. Partial alkalinity.
No dilution
Distilled water
Nutrient/buffer solution
Methane potential (NmL/gVS)
400
300
200
100
0 1
2
3
5
6
Substrate loads (gVS) Fig. 1. Methane potentials of cellulose at different substrate loads obtained from BMP tests in Experiment I.
substrate concentration-dependent data. The accuracy of the model fit was evaluated based on the root mean squared error (RMSE) and the coefficient of determination (R2) (Kafle and Kim, 2013; Thomsen et al., 2014).
BMPðSÞ ¼ BMPmax S=ðS þ K S Þ
ð2Þ
450
A: Distilled water
400
BMP (S) = 411*S/(S+0.9) RMSE = 2.7 NmL/gVS R2 = 0.9422
3.2. Results comparison between Experiment I and II As seen in Fig. 3, where the methane potentials at different substrate concentrations from these two studies in Experiment II are plotted, similar trends with decreasing values at lower concentrations were obtained. Additionally, the difference in methane
Methane yield (NmL/gVS)
Methane yield (NmL/gVS)
In Eq. (2), BMP (S) is the methane potential (NmL/g VS) at initial substrate concentration S (g VS/L), BMPmax is the maximal methane potential (NmL/g VS), and KS is the half saturation constant.
In Fig. 2, the methane potential as a function of the substrate concentration for the two diluted series is presented together with the Monod type model prediction. As seen, the model describes the variation in methane potential at different substrate concentrations for both dilution liquids fairly accurately (i.e. R2 > 0.94). The lower saturation constant for the distilled water series suggests that the methane potential is less influenced by the substrate concentration when distilled water is used as dilution liquid compared to the dedicated nutrient/buffer solution. This can probably be explained by the low pH of nutrient/buffer solution that introduces rather acidic conditions at the start of tests with high dilutions. However, more measurement points at lower substrate concentrations are necessary before any conclusion can be drawn. Furthermore, the BMPmax for the series diluted with nutrient/buffer solution is higher than the theoretically possible one for cellulose (431 vs. 415 NmL/g VS), which further demonstrates that there is a need for more data points before any definitive conclusions can be drawn. The fact that the Monod type model can describe how the methane potential varies as a function of the substrate concentration suggests that this type of model could potentially be applied to estimate a true maximal methane potential or used to investigate how different substrate and inocula perform at different loading conditions.
350
300
450
B: Nutrient/buffer solution
400
BMP (S) = 431*S/(S+1.4) RMSE = 0.2 NmL/gVS R2 = 0.9999
350
300
250
250 0
5
10
15
Substrate concentration (gVS/L)
20
0
5
10
15
20
Substrate concentration (gVS/L)
Fig. 2. Methane potentials at different substrate concentrations for series with distilled water (A) and series with nutrient/buffer solution (B). Included in the graphs is also the Monod type model prediction as a function of the substrate concentration.
244
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
A: Cellulose BMP (S) = 348*S/(S+0.2) RMSE = 1.7 NmL/gVS R2 = 0.9603
Methane potential (NmL/gVS)
Methane potential (NmL/gVS)
400
350
300
250
450
B: Soymilk and sugar
400
BMP (S) = 427*S/(S+0.7) RMSE = 2.2 NmL/gVS R2 = 0.9327
350
300
250 0
4
8
12
Substrate concentration (gVS/L)
0
2
4
6
8
Substrate concentration (gVS/L)
Fig. 3. Methane potentials at different substrate concentrations in Experiment II for series with cellulose (A) and series with soymilk and sugar (B). Included in the graphs is also the Monod type model prediction as a function of the substrate concentration.
potentials were found to be statistically significant (p = 0.05) for both data series. These results provide more evidence that the substrate concentrations influence the methane potential and that higher substrate concentration leads to higher methane potentials. The results further show that this trend also occurs when the substrate concentration is regulated with the inoculum amount (ISR) and not a dilution liquid. Thus, it is mainly a factor of the substrate concentration and not the microorganism concentration. However, the lower achieved methane potentials for cellulose in the validation data show that other experimental factors also have a strong impact on the methane potential. One of these factors is most likely the fact that two different inocula with varying metabolic activity were used in the two studies. The Monod type model could fairly accurately (R2 > 0.93) describe the variation in the two validations series as well. This further confirms that Monod type model could be used to describe the change in methane potential as a function at the substrate concentration. However, as the parameter values for cellulose are rather different between the two studies it is clear the model is not specific towards a substrate.
3.3. Kinetic evaluation In addition to the methane potential, a BMP test may provide useful information with regard to the degradation kinetics of a material (García-Gen et al., 2015; Linke, 2006). The degradation kinetics are often considered to be more sensitive to the experimental conditions compared to the methane potential, as it is highly dependent on the microbial activity and substrate accessibility (Jensen et al., 2011). This is often considered to be even more sensitive towards the experimental conditions compared to the methane potential. Therefore, this study evaluated the influence of the substrate concentration and type of dilution liquid on the calculated model parameters of a first order rate model (Eq. (1)). As seen in Fig. 4A, which presents the parameter uncertainty surfaces (p = 0.05) for k and BMPmax for the three series, the size of the parameter uncertainty for the rate constant (±5–10%) can be argued to be rather substantial. A slightly increasing uncertainty at lower substrate loads is also visible, indicating that the substrate concentration should be kept high in order to reach more confident results. The values for the first order rate constant are in between 0.3 and 0.65 1/day, which are higher compared with the
0.23 ± 0.15 1/day for cellulose reported from the inter-laboratory study by Raposo et al. (2011). This can most likely be explained by different inocula used and the fact that it is unclear how the first order rate constants from the inter-laboratory study were calculated. In particular whether a lag time constant is used or not will have a large influence on the results. By studying how the rate constant varies with the substrate load conditions, it can be seen that there is some separation in the values between the three substrate load series while there is no obvious arrangement when it with regard to the size of the substrate load. This indicates that the choice of dilution liquid or liquid volume influence the degradation rate more compared to the substrate load. Interestingly, dilutions with nutrient/buffer solution led to higher rate constants compared to distilled water. Thus, it seems that the nutrients and buffering components of the nutrient/buffer solution have a positive effect on the degradation rate of the process, while the effects on the methane potential are minimal. The fact that the substrate concentration hardly influenced the degradation rate can be explained by the fixed ISR for all samples. Neither type of dilution liquid nor substrate concentration had a strong effect on the lag time at the beginning of the biodegradation process, which were rather fixed approximately 1–2 days (Fig. 4B). However, the experiments with undiluted samples have substantially larger lag time constants, indicating that the larger headspace in relation to the liquid volume has a negative impact on the gas volume measurement.
3.4. Summarizing discussion and recommendations The results achieved in this study suggest that there is a correlation between substrate concentration and methane potential in BMP tests. It is apparent that the methane potential might be underestimated if really low substrate concentrations are used. The suggested Monod type model is able to describe this variation fairly well and could potentially be used to estimate the maximal methane potential of a substrate. Theoretically it should also be possible to use the model to derive a reasonable minimal concentration to use in order to avoid underestimations in the methane potential. A sensible criterion would be to achieve P90% of the maximal methane potential, which would translate into a substrate concentration P9 KS. Applying this relationship to the three investigated experimental series with cellulose yields minimal concentrations of 8.1, 12.6 and 1.8 g VS/L for dilutions with
245
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
No dilution
400
Distilled water
Nutrient/buffer solution
5
A
B
360 θ (days)
BMPmax (NmL/gVS)
4
3
2
320 1 280 0.25
0.35
0.45
0.55
0.65
0 0.25
0.35
k (1/day)
0.45
0.55
0.65
k (1/day)
Fig. 4. Estimated parameter uncertainty surfaces for rate parameter (k) vs. methane potential parameter (BMPmax) (A) and rate parameter (k) vs. lag time constant (h) (B). The different substrate loads are represented by the thickness of the line with thicker lines for higher substrate loads.
distilled water, nutrient/buffer solution and different ISR, respectively. These values are unreasonably different and prove that this is not a reliable approach with the current set of data. However, given the few data points available for each data series, there is a high chance that the parameter values are incorrect. Therefore, more tests with a wider range of substrate concentrations should be performed to decrease the uncertainty of the parameters before a final conclusion concerning the applicability of this model can be decided. It is also expected that factors such as type of inoculum, particle size, substrate homogeneousness and hydrophilic properties have a strong influence on the results. Thus it is clear that more research is needed to fully understand this relationship and possibly find a model that more uniformly can describe the relationship between substrate concentration and methane potential. A factor that has not been addressed in the experimental data is what happens when the substrate concentrations are too high leading to inhibition as a consequence of accumulated intermediate products. This might particularly be of consequence for inocula characterized by high solids content due to presence of large amounts of undegraded material beside the microbial biomass. For these particular cases it is recommended to either dilute the inocula slightly or use a higher ISR. The benefit with using a higher ISR is that this leads to a more realistic relationship between the sample and the microbial population, which in reality might be significantly different than the VS content suggests. The downside is that the measurement uncertainty increases since these types of inocula tend to produce large amounts of gas. However, with the modern automatic and accurate systems that are available for these BMP determination today the increased measurement uncertainty is expected to be manageable. The upper concentration limit of the criterion (i.e. 610 g VS) provided by VDI 4630 (2006) would fit fairly well and give reasonable methane potentials with the experimental data obtained in this study. However, as higher methane potentials were observed at even higher substrate concentrations compared to this criterion it could be argued that the lower concentration limit should be adjusted to allow larger contents of the sample. Due to the substantial differences in substrate concentrations between the investigated experimental series in this study, it is difficult to draw a definitive conclusion about what substrate concentration should be used to avoid underestimations in the methane
potential. However, all experimental data indicate that dilution of the inoculum or substrate may introduce underestimations in the methane potential. It is therefore recommended to avoid this, at least if the dilutions lead to a substrate concentration below 10 g VS/L. This limit is proposed as it is similar to the upper limit of the VDI 4630 (2006) criterion and would ensure more than 85% of the maximal methane potential for all the investigated samples in this study. It should be emphasized that substrate concentrations below 10 g VS/L will still be needed for inocula characterized by low VS contents in order to ensure a stable ISR. If dilutions are performed, a dilution liquid with a buffering capacity should preferably be used as distilled water was found to introduce slower degradation kinetics and lower pH. It should be accentuated that these results are limited to the conditions of the BMP tests in this study. Much more research is needed to verify these observations on a wider-scale with more types of samples, inocula and experimental conditions. Furthermore, another dilution liquid with higher alkalinity and pH compared to the one used in this study should be investigated. 4. Conclusions The substrate concentrations in BMP tests have been found to influence the methane potentials of substrates; this effect is most profound at lower concentrations. For most accurate results it is recommended that the substrate concentration should be kept higher, and furthermore dilutions should be avoided if the substrate concentration is below 10 g VS/L. Dilution with nutrient/buffer solution showed positive effect on degradation rate. Monod type model could probably be used to derive a reasonable minimal substrate concentration in order to avoid underestimation in methane yield. Acknowledgements The China Scholarship Council (CSC), Swedish Energy Agency (STEM), Swedish International Development Agency (SIDA), and Open-End Fund of Key Laboratory of Development and Application of Rural Renewable Energy (2014002), Ministry of Agriculture in China are gratefully acknowledged for the financial support.
246
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
References Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, J., Guwy, A., Kalyuzhnyi, S., Jenicek, P., Van Lier, J., 2009. Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays. Water Sci. Technol. 59 (5), 927–934. Angelidaki, I., Sanders, W., 2004. Assessment of the anaerobic biodegradability of macropollutants. Rev. Environ. Sci. Biotechnol. 3 (2), 117–129. APHA, 1995. Standard Methods for the Examination of Water and Wastewater, 19th ed. American Public Health Association American Water Works Association/Environment Federation, Washington D.C.. Badshah, M., Lam, D.M., Liu, J., Mattiasson, B., 2012. Use of an automatic methane potential test system for evaluating the biomethane potential of sugarcane bagasse after different treatments. Bioresour. Technol. 114, 262–269. Batstone, D.J., Pind, P.F., Angelidaki, I., 2003. Kinetics of thermophilic, anaerobic oxidation of straight and branched chain butyrate and valerate. Biotechnol. Bioeng. 84 (2), 195–204. Bauer, A., Bösch, P., Friedl, A., Amon, T., 2009. Analysis of methane potentials of steam-exploded wheat straw and estimation of energy yields of combined ethanol and methane production. J. Biotechnol. 142 (1), 50–55. Brulé, M., Bolduan, R., Seidelt, S., Schlagermann, P., Bott, A., 2013. Modified batch anaerobic digestion assay for testing efficiencies of trace metal additives to enhance methane production of energy crops. Environ. Technol. 34 (13–14), 2047–2058. García-Gen, S., Sousbie, P., Rangaraj, G., Lema, J.M., Rodríguez, J., Steyer, J.-P., Torrijos, M., 2015. Kinetic modelling of anaerobic hydrolysis of solid wastes, including disintegration processes. Waste Manage. 35, 96–104. Grieder, C., Mittweg, G., Dhillon, B.S., Montes, J.M., Orsini, E., Melchinger, A.E., 2012. Kinetics of methane fermentation yield in biogas reactors: genetic variation and association with chemical composition in maize. Biomass Bioenergy 37, 132– 141. Hashimoto, A.G., 1989. Effect of inoculum/substrate ratio on methane yield and production rate from straw. Biol. Wastes 28 (4), 247–255. ISO-11734, 1995. Water Quality – Evaluation of the ‘‘Ultimate’’ Anaerobic Biodegradability of Organic Compounds in Digested Sludge – Method by Measurement of the Biogas Production, first ed. International Organization for Standardization, Geneva. Jensen, P., Ge, H., Batstone, D.J., 2011. Assessing the role of biochemical methane potential tests in determining anaerobic degradability rate and extent. Water Sci. Technol. 64 (4), 880. Kafle, G.K., Kim, S.H., 2013. Anaerobic treatment of apple waste with swine manure for biogas production: batch and continuous operation. Appl. Energy 103, 61– 72. Koch, K., Fernández, Y.B., Drewes, J.E., 2015. Influence of headspace flushing on methane production in Biochemical Methane Potential (BMP) tests. Bioresour. Technol. 186, 173–178. Lianhua, L., Dong, L., Yongming, S., Longlong, M., Zhenhong, Y., Xiaoying, K., 2010. Effect of temperature and solid concentration on anaerobic digestion of rice straw in South China. Int. J. Hydrogen Energy 35 (13), 7261–7266.
Linke, B., 2006. Kinetic study of thermophilic anaerobic digestion of solid wastes from potato processing. Biomass Bioenergy 30 (10), 892–896. Lobry, J., Rosso, L., Flandrois, J., 1991. A FORTRAN subroutine for the determination of parameter confidence limits in non-linear models. Binary 3, 86–93. Maya-Altamira, L., Baun, A., Angelidaki, I., Schmidt, J.E., 2008. Influence of wastewater characteristics on methane potential in food-processing industry wastewaters. Water Res. 42 (8), 2195–2203. Møller, H.B., Sommer, S.G., Ahring, B.K., 2004. Methane productivity of manure, straw and solid fractions of manure. Biomass Bioenergy 26 (5), 485–495. Nges, I.A., Liu, J., 2009. Effects of anaerobic pre-treatment on the degradation of dewatered-sewage sludge. Renew. Energy 34 (7), 1795–1800. Nkemka, V.N., Murto, M., 2013. Biogas production from wheat straw in batch and UASB reactors: the roles of pretreatment and seaweed hydrolysate as a cosubstrate. Bioresour. Technol. 128, 164–172. Raposo, F., Banks, C., Siegert, I., Heaven, S., Borja, R., 2006. Influence of inoculum to substrate ratio on the biochemical methane potential of maize in batch tests. Process Biochem. 41 (6), 1444–1450. Raposo, F., Borja, R., Rincon, B., Jimenez, A.M., 2008. Assessment of process control parameters in the biochemical methane potential of sunflower oil cake. Biomass Bioenergy 32 (12), 1235–1244. Raposo, F., De la Rubia, M., Fernández-Cegrí, V., Borja, R., 2012. Anaerobic digestion of solid organic substrates in batch mode: an overview relating to methane yields and experimental procedures. Renew. Sustain. Energy Rev. 16 (1), 861– 877. Raposo, F., Fernández-Cegrí, V., De la Rubia, M., Borja, R., Béline, F., Cavinato, C., Demirer, G., Fernández, B., Fernández-Polanco, M., Frigon, J., 2011. Biochemical methane potential (BMP) of solid organic substrates: evaluation of anaerobic biodegradability using data from an international interlaboratory study. J. Chem. Technol. Biotechnol. 86 (8), 1088–1098. Sánchez, E., Borja, R., Weiland, P., Travieso, L., Martı´n, A., 2001. Effect of substrate concentration and temperature on the anaerobic digestion of piggery waste in a tropical climate. Process Biochem. 37 (5), 483–489. Seto, M., Alexander, M., 1985. Effect of bacterial density and substrate concentration on yield coefficients. Appl. Environ. Microbiol. 50 (5), 1132–1136. Strömberg, S., Nistor, M., Liu, J., 2014. Towards eliminating systematic errors caused by the experimental conditions in Biochemical Methane Potential (BMP) tests. Waste Manage. 34 (11), 1939–1948. Tanimu, M.I., Ghazi, T.I.M., Harun, M.R., Idris, A., 2014. Effect of feed loading on biogas methane production in batch mesophilic anaerobic digesters treating food waste. Int. J. Chem. Environ. Eng. 5 (1). Thomsen, S.T., Spliid, H., Østergård, H., 2014. Statistical prediction of biomethane potentials based on the composition of lignocellulosic biomass. Bioresour. Technol. 154, 80–86. VDI 4630, 2006. Fermentation of organic materials – characterisation of the substrate, sampling, collection of material data, fermentation tests. Zhang, W., Lang, Q., Wu, S., Li, W., Bah, H., Dong, R., 2014. Anaerobic digestion characteristics of pig manures depending on various growth stages and initial substrate concentrations in a scaled pig farm in Southern China. Bioresour. Technol. 156, 63–69.
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Effects of substrate concentration on methane potential and degradation kinetics in batch anaerobic digestion Bing Wang a,⇑, Sten Strömberg a, Chao Li a, Ivo Achu Nges a, Mihaela Nistor b, Liangwei Deng c, Jing Liu a,b a
Department of Biotechnology, Lund University, Getingevägen 60, SE-221 00 Lund, Sweden Bioprocess Control Sweden AB, Scheelevägen 22, SE-223 63 Lund, Sweden c Biogas Institute of Ministry of Agriculture, 610041 Chengdu, China b
h i g h l i g h t s High substrate concentration led to high methane yields. Nutrient/buffer-diluted substrate showed a positive effect on degradation rate. Neither dilution liquid nor substrate concentration had a strong effect on lag time. Dilutions should be avoided at substrate concentration below 10 g VS/L.
a r t i c l e
i n f o
Article history: Received 26 May 2015 Received in revised form 10 July 2015 Accepted 11 July 2015 Available online 17 July 2015 Keywords: Anaerobic digestion Biochemical methane potential Substrate load Substrate concentration Kinetics
a b s t r a c t In this study, two experiments were conducted to evaluate the impact of substrate concentrations on methane potential and degradation kinetics of substrate. The biochemical methane potential (BMP) tests in Experiment I were performed at a constant inoculum to substrate ratio (ISR), whereas, different ISRs were applied in Experiment II. Results obtained from Experiment I revealed that methane potential of substrate increased at a saturating trend with higher substrate concentrations, and could differ by up to 30% between the lowest and highest investigated concentrations. The results of Experiment II verified the results of Experiment I, and further showed that this trend also occurs when the substrate concentration is regulated with ISRs. In contrast, substrate concentration had no significant impact on the degradation kinetics. It was concluded that dilutions should be avoided when the substrate concentration is lower than 10 g VS/L in order to avoid underestimations of methane potential from BMP test. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The methane potential of biomass, which can be determined by a traditional biochemical methane potential (BMP) test, is a key parameter for the design, economy and management in full-scale implementations of anaerobic digestion (AD). In recent years, a high number of scientific articles related to BMP tests of a variety of substrates have been published (Bauer et al., 2009; Møller et al., 2004; Nkemka and Murto, 2013). However, the results from such tests are variable and difficult to compare. The varying results are caused by differences in both instrumentation and protocols, as well as by different experimental conditions. For instance, the pH, headspace volume, mixing options, inoculum to substrate ratio (ISR), inoculum dilutions, but also substrate load (i.e. initial
⇑ Corresponding author. Tel.: +46 462223672; fax: +46 462224713. E-mail address: [email protected] (B. Wang). http://dx.doi.org/10.1016/j.biortech.2015.07.034 0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.
substrate concentration) can differ among different tests (Angelidaki et al., 2009). Particularly, the substrate concentration has been considered as an important factor which influences the efficiency of the anaerobic digestion process (Lianhua et al., 2010; Sánchez et al., 2001). At very low concentrations, there is a risk that microorganisms may exhibit a low metabolic activity and very low quantities of biogas will be produced. In contrast, if the substrate concentration is too high, that might lead to an overload situation in which intermediate compounds may accumulate, resulting in inhibition of the process (Tanimu et al., 2014; Zhang et al., 2014). So far, there has not been so much focus on the substrate concentration in a BMP test, but it is fairly common to suggest diluting the substrate to limit overloading and influence from possible toxic components (Angelidaki et al., 2009). However, the choice of dilution liquid used to adjust the substrate concentration is important. Using a medium containing the necessary nutrients, trace elements and vitamins are recommended to avoid inhibition in microbial
241
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
growth (Angelidaki and Sanders, 2004; Raposo et al., 2012). It has also been shown that additions of these types of medium improve the methane potential of diluted inocula while undiluted ones are not affected (Brulé et al., 2013). Nonetheless, there are few guidelines recommending what substrate concentration to use in a BMP test to reach optimal results. One example is given by the German Standard (VDI 4630, 2006) which recommends using an inoculum concentration between 1.5% and 2.0% in volatile solids (VS) and an inoculum to substrate ratio (ISR) P 2, which translates to a substrate concentration of 610 g VS/L. However, no exact concentration requirement is given and without a lower boundary there is a risk that too low concentrations are used. This uncertainty is also reflected in the conditions used in many BMP assays where the substrate concentration can differ profoundly when different types of inocula and substrates have been used (Raposo et al., 2011). For example, few studies have evaluated the impact on the methane potential using different substrate concentrations at a fixed ISR while several have investigated it at variable ISRs. Maya-Altamira et al. (2008) found that the methane potential of various types of industrial wastewaters increased slightly with increasing ISR (COD based). The highest methane potentials were observed at ISR P 2 and it was assumed that the lower values at the lower ISR were caused by overload situation, in which volatile fatty acid (VFA) or acetate acid was built up, resulting in a slow AD process inhibition. Hashimoto (1989) investigated a large number of ISRs for ball-milled wheat straw and found that the methane potential drastically decreased at an ISR below 0.25. The general conclusion from these studies is that a too low ISR leads to inhibition and that this parameter should be kept above a certain threshold, most commonly ISR P 2 (Raposo et al., 2008; VDI 4630, 2006). However, these studies are almost exclusively focused on identifying points where the ISR is high enough to prevent inhibition, whereas few studies have investigated whether the substrate concentration can be too low. The hypothesis tested in this study is that the substrate concentration can influence the specific methane potential as well as the degradation kinetics. In this light, various experiments were performed to investigate the influences of substrate concentrations on methane production and degradation kinetics. In all, five different substrate concentrations were evaluated at a fixed ISR of 2, as well as the effect from choice of dilution liquid (distilled water and nutrient/buffer solution). Moreover, further tests were also performed at both fixed and varying ISRs.
Table 1 Preparation of substrate/inoculum/diluent mixtures used in Experiment I. Substrate loads (g VS)
Undiluted setup Substrate conc. (g VS/L)
Liquid vol. (mL)
Substrate conc. (g VS/L)
Liquid vol. (mL)
1 2 3 5 6
15.8 15.8 15.8 15.8 15.8
63 127 190 317 381
2.5 5.0 7.5 12.5 15.0
400 400 400 400 400
Diluted setups
in triplicates with an individual set of blanks with no sample. All tests were prepared in 500 mL standard bottles (Schott, Germany) and mixed intermittently (160 rpm, stirrers were on for 5 min and off for 25 min continuously throughout the test) to ensure a good mass transfer. In Experiment II, the following BMP studies with varying substrate concentrations were performed in order to verify the results from Experiment I: (i) cellulose was used as substrate, the concentration was adjusted by varying the ISR instead of diluting the sample, (ii) substrate concentration of soymilk and sugar was adjusted both at a fixed ISR using dilution (diluted with distilled water) and different ISRs. The total liquid volume in each bottle was 400 mL (Table 2). 2.2. Inoculum and substrate The inoculum used to carry out the BMP tests in Experiment I was collected from a mesophilic sewage treatment plant in Källby, Sweden, which was characterized by a VS content of 3.21% (w/w), pH of 8.03 and partial alkalinity (PA) of 6807 mg/L. The anaerobic inoculum (1.07% VS) used to carry out the BMP tests in Experiment II was collected from a mesophilic biogas plant (Ellinge, Sweden), which receives municipal wastewater (20%) and vegetable residue from the food industry (80%). In order to decrease the background biogas production, the inocula were pre-incubated at 37 °C for 5 days prior to the BMP tests (ISO-11734, 1995). Microcrystalline cellulose (Alfa Aesar, Germany, 96.1% VS) was used as a standard substrate in both experiments, and another mixture of soymilk and sugar (1:1 based on VS %) was used as substrate in Experiment II. Soymilk and sugar had a VS content of 9.98%.
2. Methods
2.3. Analytical methods
2.1. Experimental setup
2.3.1. Characterization of inoculum and substrate The total solids (TS) and VS of inoculum and substrate were determined according to standard protocols (APHA, 1995). The pH and partial alkalinity (PA) were measured using a TitraLab™
Two experiments were performed to evaluate the influences of substrate concentration on BMP test. In Experiment I, three different ways of varying the substrate load in BMP test were investigated: (i) dilution with a nutrient/buffer solution, (ii) dilution with distilled water and (iii) adjusting the liquid volume without dilution. For the two setups including dilution, after the addition of inoculum and substrate, either distilled water or nutrient/buffer solution was added to reach 400 mL; the corresponding blank samples contained inoculum diluted with distilled water or nutrient/buffer solution respectively. The nutrient/buffer solution was prepared according to Brulé et al. (2013). For each series, five different substrate loads at a fixed ISR of 2 (based on VS) were studied. The substrate concentrations for the diluted and undiluted set-ups can be observed in Table 1. In total 15 unique BMP tests (i.e. 5 different substrate loads with three types of dilution methods) were performed with cellulose as substrate and the inoculum from Källby as seeding sludge. Each of the 15 tests was performed
Table 2 Substrate concentrations and ISR used in the BMP tests in Experiment II. Cellulose
a
Soymilk and sugar
Substrate loads (g VS)
Substrate conc. (g VS/L)
ISRa
Substrate loads (g VS)
Substrate conc. (g VS/L)
ISR
0.52 1.08 2.12 4.20
1.3 2.7 5.3 10.5
1 2 4 8
0.8 1.2 1.6 0.8 1.2 1.6 1.7 2.4
2.0 3.0 4.0 2.0 3.0 4.0 4.2 6.1
2.0 2.0 2.0 6.4 4.2 3.1 3.0 2.0
Inoculum to substrate ratio.
242
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
80 titrator (Radiometer, Copenhagen, Denmark) as reported in another study (Nges and Liu, 2009). 2.3.2. Determination of methane production All the BMP tests were carried out in triplicates with the Automatic Methane Potential Test System II (AMPTS II, Bioprocess Control, Sweden AB) at 37 °C for 30 days. The experimental protocol was performed according to Badshah et al. (2012). At the end of the process, a report presented the normalized (standard temperature and pressure, STP: 273.15 K, 101.325 kPa; compensation of the water vapor content) methane flow rate and accumulative methane volume. 2.3.3. Kinetics A traditional first-order model (Eq. (1)) with a lag-time parameter was used to evaluate the kinetic degradation profile of the substrate (Grieder et al., 2012).
BMPðtÞ ¼ BMPmax ð1 expðk ðt hÞÞÞ
ð1Þ
In Eq. (1) BMP (t) is the methane potential (NmL/g VS) at time t (day), BMPmax is the maximum or ultimate methane potential (NmL/g VS) of the substrate, k is the rate constant or hydrolysis rate constant (1/day) and h is the lag time constant (day). 2.3.4. Statistical calculations Grubbs’ test was used to check and eliminate outliers in BMP tests. One-way and two-way ANOVA was calculated using the MATLAB function anovan at 95% confidence. The uncertainty of the kinetic parameters was estimated in a similar way as presented by Batstone et al. (2003). In order to avoid three-dimensional datasets, that would be difficult to visualize, one of the three model parameters was always kept at its optimal value while the remaining two were varied to determine the two-dimensional surface of the parameter uncertainty. The optimal parameter values were estimated with lsqcurvefit in MATLAB (v.8.1.0.604) and the surface of parameter uncertainty was estimated by using a similar approach as presented by Lobry et al. (1991). 3. Results and discussions 3.1. Results obtained from Experiment I 3.1.1. pH and partial alkalinity The pH and PA of the liquid in the reactor before and after anaerobic digestion for the different scenarios in Experiment I are presented in Table 3. As seen, diluting the liquid mixture changes the pH in the reactor. The undiluted samples showed a high pH at the start which dropped during the test, with slightly stronger effect on the pH at higher substrate loads. This decrease in pH is expected as increasing amounts of carbon dioxide is being dissolved when the equilibrium conditions with the headspace changes from the produced biogas. The slightly stronger effects at higher substrate loads (higher liquid volumes) could also be explained by this theory as the composition of the headspace gas changes faster at smaller headspace volumes. Dilution with both distilled water (pH 8.19) and nutrient/buffer solution (pH 6.40) had a rather substantial effect on both the pH and PA of the evaluated series. Using DW led to a significant decrease in PA for the higher dilutions that also led to lower pH at the end of the tests because of its poor buffer capacity. Interestingly, similar effects were observed for nutrient/buffer solution as well, indicating that the buffering capacity of this nutrient/buffer solution was insufficient. Furthermore, the low starting
pH at a substrate load of 1 g VS indicates that the acidic pH of the nutrient/buffer solution may lead to problems at high dilutions. 3.1.2. Methane potentials of cellulose The obtained methane potentials at different cellulose loads were presented in Fig. 1. In general, analysis of variance (ANOVA) of methane yields demonstrated that for loads of 1 g VS and 2 g VS, that respectively correspond to substrate concentrations of 2.5 g VS/L and 5 g VS/L, dilutions with distilled water or nutrient/buffer solution led to a decrease in methane yields with a load of 1 g VS showing significant (p = 0.05) reduction when diluted. At the lowest substrate load, i.e. 1 g VS, the undiluted sample had the highest methane production compared with the diluted samples. For loads of 3 g VS and 5 g VS (7.5 g VS/L and 12.5 g VS/L, respectively), dilutions did not significantly (p = 0.05) influence the methane yields. However, at a load of 6 g VS (15 g VS/L), dilution with either distilled water or nutrient/buffer solution led to significantly (p = 0.05) higher methane yields. It should be noted that at load 6 g VS (15 g VS/L), the methane yields of the diluted with distilled water and nutrient/buffer solution did not differ significantly (p = 0.05). For undiluted samples, however there is no significant (p = 0.05) difference in methane potential between the substrate loads irrespective of the dilutions. This is expected, as the only difference between the different substrate loads are varying headspace volumes. However, it should be pointed out that the measurement uncertainty from having a high headspace volume in relation to the liquid volume are considerably higher as a consequence of smaller sample volume, lower gas flows and more influence of the initial head space gas (Koch et al., 2015; Strömberg et al., 2014). As seen in Fig. 1, diluting the samples led to increased methane potentials with greater substrate loads and the maximal values of 396 (distilled water) and 394 (nutrient/buffer solution) NmL/g VS are both observed at the highest investigated load of 6 g VS (15 g VS/L). A two-way ANOVA test showed that there was a significant (p = 0.05) difference between methane potentials using different substrate concentrations but not a significant difference between series with different types of dilution liquid. This result implies that the substrate concentration has an apparent effect on the outcome of a BMP test while the choice of dilution liquid is of less importance. However, given the poor buffering capacity of nutrient/buffer solution, this conclusion is limited to this specific medium. The fact that the methane potentials at lower substrate concentrations are rather far from the theoretical one (415 NmL/g VS) suggests that too low substrate concentrations might introduce underestimation in this value. One possible explanation for the lower methane potentials at lower substrate concentrations is the slow growth rates of microorganisms and use cell carbon to satisfy the need for maintenance and metabolism (Seto and Alexander, 1985). Regardless on the mechanism causing this effect, the increase in methane potential with higher substrate concentrations implies that, besides the pH, the substrate concentration should also be considered and kept at similar values for comparable results. 3.1.3. Methane potential as a function of substrate concentration As indicated in Fig. 1, the methane potentials increase with increasing substrate loads, which is in agreement with other study. Raposo et al. (2006) studied on the influence of ISR on BMP of maize at a concentration range of 5–15 g VS/L, and the highest methane production was obtained at concentration of 15 g VS/L. If it is assumed that the methane potential is directly dependent on the substrate concentration and asymptotically reaches a maximal potential at higher concentrations, a saturation kinetics-based equation could be used to describe this effect. In this study a Monod type equation (Eq. (2)) was assessed for fitting the
243
B. Wang et al. / Bioresource Technology 194 (2015) 240–246 Table 3 The pH and partial alkalinity of liquid at different substrate loads. Substrate loads (g VS)
1 2 3 5 6 a b c
Sample diluted with DWa
Undiluted sample pHstart
pHend
PAcend
8.03 8.03 8.03 8.03 8.03
7.97 7.90 7.84 7.76 7.67
5774 6637 6805 6716 6140
(mg/L)
Sample diluted with NBSb
pHstart
pHend
PAend (mg/L)
pHstart
pHend
PAend (mg/L)
8.06 7.91 7.87 7.75 7.75
6.84 7.24 7.31 7.48 7.64
1405 2646 3906 6203 6358
6.71 7.65 7.69 7.77 7.80
7.02 7.05 7.37 7.52 7.63
3323 4180 4894 6509 7818
Distilled water. Nutrient/buffer solution. Partial alkalinity.
No dilution
Distilled water
Nutrient/buffer solution
Methane potential (NmL/gVS)
400
300
200
100
0 1
2
3
5
6
Substrate loads (gVS) Fig. 1. Methane potentials of cellulose at different substrate loads obtained from BMP tests in Experiment I.
substrate concentration-dependent data. The accuracy of the model fit was evaluated based on the root mean squared error (RMSE) and the coefficient of determination (R2) (Kafle and Kim, 2013; Thomsen et al., 2014).
BMPðSÞ ¼ BMPmax S=ðS þ K S Þ
ð2Þ
450
A: Distilled water
400
BMP (S) = 411*S/(S+0.9) RMSE = 2.7 NmL/gVS R2 = 0.9422
3.2. Results comparison between Experiment I and II As seen in Fig. 3, where the methane potentials at different substrate concentrations from these two studies in Experiment II are plotted, similar trends with decreasing values at lower concentrations were obtained. Additionally, the difference in methane
Methane yield (NmL/gVS)
Methane yield (NmL/gVS)
In Eq. (2), BMP (S) is the methane potential (NmL/g VS) at initial substrate concentration S (g VS/L), BMPmax is the maximal methane potential (NmL/g VS), and KS is the half saturation constant.
In Fig. 2, the methane potential as a function of the substrate concentration for the two diluted series is presented together with the Monod type model prediction. As seen, the model describes the variation in methane potential at different substrate concentrations for both dilution liquids fairly accurately (i.e. R2 > 0.94). The lower saturation constant for the distilled water series suggests that the methane potential is less influenced by the substrate concentration when distilled water is used as dilution liquid compared to the dedicated nutrient/buffer solution. This can probably be explained by the low pH of nutrient/buffer solution that introduces rather acidic conditions at the start of tests with high dilutions. However, more measurement points at lower substrate concentrations are necessary before any conclusion can be drawn. Furthermore, the BMPmax for the series diluted with nutrient/buffer solution is higher than the theoretically possible one for cellulose (431 vs. 415 NmL/g VS), which further demonstrates that there is a need for more data points before any definitive conclusions can be drawn. The fact that the Monod type model can describe how the methane potential varies as a function of the substrate concentration suggests that this type of model could potentially be applied to estimate a true maximal methane potential or used to investigate how different substrate and inocula perform at different loading conditions.
350
300
450
B: Nutrient/buffer solution
400
BMP (S) = 431*S/(S+1.4) RMSE = 0.2 NmL/gVS R2 = 0.9999
350
300
250
250 0
5
10
15
Substrate concentration (gVS/L)
20
0
5
10
15
20
Substrate concentration (gVS/L)
Fig. 2. Methane potentials at different substrate concentrations for series with distilled water (A) and series with nutrient/buffer solution (B). Included in the graphs is also the Monod type model prediction as a function of the substrate concentration.
244
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
A: Cellulose BMP (S) = 348*S/(S+0.2) RMSE = 1.7 NmL/gVS R2 = 0.9603
Methane potential (NmL/gVS)
Methane potential (NmL/gVS)
400
350
300
250
450
B: Soymilk and sugar
400
BMP (S) = 427*S/(S+0.7) RMSE = 2.2 NmL/gVS R2 = 0.9327
350
300
250 0
4
8
12
Substrate concentration (gVS/L)
0
2
4
6
8
Substrate concentration (gVS/L)
Fig. 3. Methane potentials at different substrate concentrations in Experiment II for series with cellulose (A) and series with soymilk and sugar (B). Included in the graphs is also the Monod type model prediction as a function of the substrate concentration.
potentials were found to be statistically significant (p = 0.05) for both data series. These results provide more evidence that the substrate concentrations influence the methane potential and that higher substrate concentration leads to higher methane potentials. The results further show that this trend also occurs when the substrate concentration is regulated with the inoculum amount (ISR) and not a dilution liquid. Thus, it is mainly a factor of the substrate concentration and not the microorganism concentration. However, the lower achieved methane potentials for cellulose in the validation data show that other experimental factors also have a strong impact on the methane potential. One of these factors is most likely the fact that two different inocula with varying metabolic activity were used in the two studies. The Monod type model could fairly accurately (R2 > 0.93) describe the variation in the two validations series as well. This further confirms that Monod type model could be used to describe the change in methane potential as a function at the substrate concentration. However, as the parameter values for cellulose are rather different between the two studies it is clear the model is not specific towards a substrate.
3.3. Kinetic evaluation In addition to the methane potential, a BMP test may provide useful information with regard to the degradation kinetics of a material (García-Gen et al., 2015; Linke, 2006). The degradation kinetics are often considered to be more sensitive to the experimental conditions compared to the methane potential, as it is highly dependent on the microbial activity and substrate accessibility (Jensen et al., 2011). This is often considered to be even more sensitive towards the experimental conditions compared to the methane potential. Therefore, this study evaluated the influence of the substrate concentration and type of dilution liquid on the calculated model parameters of a first order rate model (Eq. (1)). As seen in Fig. 4A, which presents the parameter uncertainty surfaces (p = 0.05) for k and BMPmax for the three series, the size of the parameter uncertainty for the rate constant (±5–10%) can be argued to be rather substantial. A slightly increasing uncertainty at lower substrate loads is also visible, indicating that the substrate concentration should be kept high in order to reach more confident results. The values for the first order rate constant are in between 0.3 and 0.65 1/day, which are higher compared with the
0.23 ± 0.15 1/day for cellulose reported from the inter-laboratory study by Raposo et al. (2011). This can most likely be explained by different inocula used and the fact that it is unclear how the first order rate constants from the inter-laboratory study were calculated. In particular whether a lag time constant is used or not will have a large influence on the results. By studying how the rate constant varies with the substrate load conditions, it can be seen that there is some separation in the values between the three substrate load series while there is no obvious arrangement when it with regard to the size of the substrate load. This indicates that the choice of dilution liquid or liquid volume influence the degradation rate more compared to the substrate load. Interestingly, dilutions with nutrient/buffer solution led to higher rate constants compared to distilled water. Thus, it seems that the nutrients and buffering components of the nutrient/buffer solution have a positive effect on the degradation rate of the process, while the effects on the methane potential are minimal. The fact that the substrate concentration hardly influenced the degradation rate can be explained by the fixed ISR for all samples. Neither type of dilution liquid nor substrate concentration had a strong effect on the lag time at the beginning of the biodegradation process, which were rather fixed approximately 1–2 days (Fig. 4B). However, the experiments with undiluted samples have substantially larger lag time constants, indicating that the larger headspace in relation to the liquid volume has a negative impact on the gas volume measurement.
3.4. Summarizing discussion and recommendations The results achieved in this study suggest that there is a correlation between substrate concentration and methane potential in BMP tests. It is apparent that the methane potential might be underestimated if really low substrate concentrations are used. The suggested Monod type model is able to describe this variation fairly well and could potentially be used to estimate the maximal methane potential of a substrate. Theoretically it should also be possible to use the model to derive a reasonable minimal concentration to use in order to avoid underestimations in the methane potential. A sensible criterion would be to achieve P90% of the maximal methane potential, which would translate into a substrate concentration P9 KS. Applying this relationship to the three investigated experimental series with cellulose yields minimal concentrations of 8.1, 12.6 and 1.8 g VS/L for dilutions with
245
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
No dilution
400
Distilled water
Nutrient/buffer solution
5
A
B
360 θ (days)
BMPmax (NmL/gVS)
4
3
2
320 1 280 0.25
0.35
0.45
0.55
0.65
0 0.25
0.35
k (1/day)
0.45
0.55
0.65
k (1/day)
Fig. 4. Estimated parameter uncertainty surfaces for rate parameter (k) vs. methane potential parameter (BMPmax) (A) and rate parameter (k) vs. lag time constant (h) (B). The different substrate loads are represented by the thickness of the line with thicker lines for higher substrate loads.
distilled water, nutrient/buffer solution and different ISR, respectively. These values are unreasonably different and prove that this is not a reliable approach with the current set of data. However, given the few data points available for each data series, there is a high chance that the parameter values are incorrect. Therefore, more tests with a wider range of substrate concentrations should be performed to decrease the uncertainty of the parameters before a final conclusion concerning the applicability of this model can be decided. It is also expected that factors such as type of inoculum, particle size, substrate homogeneousness and hydrophilic properties have a strong influence on the results. Thus it is clear that more research is needed to fully understand this relationship and possibly find a model that more uniformly can describe the relationship between substrate concentration and methane potential. A factor that has not been addressed in the experimental data is what happens when the substrate concentrations are too high leading to inhibition as a consequence of accumulated intermediate products. This might particularly be of consequence for inocula characterized by high solids content due to presence of large amounts of undegraded material beside the microbial biomass. For these particular cases it is recommended to either dilute the inocula slightly or use a higher ISR. The benefit with using a higher ISR is that this leads to a more realistic relationship between the sample and the microbial population, which in reality might be significantly different than the VS content suggests. The downside is that the measurement uncertainty increases since these types of inocula tend to produce large amounts of gas. However, with the modern automatic and accurate systems that are available for these BMP determination today the increased measurement uncertainty is expected to be manageable. The upper concentration limit of the criterion (i.e. 610 g VS) provided by VDI 4630 (2006) would fit fairly well and give reasonable methane potentials with the experimental data obtained in this study. However, as higher methane potentials were observed at even higher substrate concentrations compared to this criterion it could be argued that the lower concentration limit should be adjusted to allow larger contents of the sample. Due to the substantial differences in substrate concentrations between the investigated experimental series in this study, it is difficult to draw a definitive conclusion about what substrate concentration should be used to avoid underestimations in the methane
potential. However, all experimental data indicate that dilution of the inoculum or substrate may introduce underestimations in the methane potential. It is therefore recommended to avoid this, at least if the dilutions lead to a substrate concentration below 10 g VS/L. This limit is proposed as it is similar to the upper limit of the VDI 4630 (2006) criterion and would ensure more than 85% of the maximal methane potential for all the investigated samples in this study. It should be emphasized that substrate concentrations below 10 g VS/L will still be needed for inocula characterized by low VS contents in order to ensure a stable ISR. If dilutions are performed, a dilution liquid with a buffering capacity should preferably be used as distilled water was found to introduce slower degradation kinetics and lower pH. It should be accentuated that these results are limited to the conditions of the BMP tests in this study. Much more research is needed to verify these observations on a wider-scale with more types of samples, inocula and experimental conditions. Furthermore, another dilution liquid with higher alkalinity and pH compared to the one used in this study should be investigated. 4. Conclusions The substrate concentrations in BMP tests have been found to influence the methane potentials of substrates; this effect is most profound at lower concentrations. For most accurate results it is recommended that the substrate concentration should be kept higher, and furthermore dilutions should be avoided if the substrate concentration is below 10 g VS/L. Dilution with nutrient/buffer solution showed positive effect on degradation rate. Monod type model could probably be used to derive a reasonable minimal substrate concentration in order to avoid underestimation in methane yield. Acknowledgements The China Scholarship Council (CSC), Swedish Energy Agency (STEM), Swedish International Development Agency (SIDA), and Open-End Fund of Key Laboratory of Development and Application of Rural Renewable Energy (2014002), Ministry of Agriculture in China are gratefully acknowledged for the financial support.
246
B. Wang et al. / Bioresource Technology 194 (2015) 240–246
References Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, J., Guwy, A., Kalyuzhnyi, S., Jenicek, P., Van Lier, J., 2009. Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays. Water Sci. Technol. 59 (5), 927–934. Angelidaki, I., Sanders, W., 2004. Assessment of the anaerobic biodegradability of macropollutants. Rev. Environ. Sci. Biotechnol. 3 (2), 117–129. APHA, 1995. Standard Methods for the Examination of Water and Wastewater, 19th ed. American Public Health Association American Water Works Association/Environment Federation, Washington D.C.. Badshah, M., Lam, D.M., Liu, J., Mattiasson, B., 2012. Use of an automatic methane potential test system for evaluating the biomethane potential of sugarcane bagasse after different treatments. Bioresour. Technol. 114, 262–269. Batstone, D.J., Pind, P.F., Angelidaki, I., 2003. Kinetics of thermophilic, anaerobic oxidation of straight and branched chain butyrate and valerate. Biotechnol. Bioeng. 84 (2), 195–204. Bauer, A., Bösch, P., Friedl, A., Amon, T., 2009. Analysis of methane potentials of steam-exploded wheat straw and estimation of energy yields of combined ethanol and methane production. J. Biotechnol. 142 (1), 50–55. Brulé, M., Bolduan, R., Seidelt, S., Schlagermann, P., Bott, A., 2013. Modified batch anaerobic digestion assay for testing efficiencies of trace metal additives to enhance methane production of energy crops. Environ. Technol. 34 (13–14), 2047–2058. García-Gen, S., Sousbie, P., Rangaraj, G., Lema, J.M., Rodríguez, J., Steyer, J.-P., Torrijos, M., 2015. Kinetic modelling of anaerobic hydrolysis of solid wastes, including disintegration processes. Waste Manage. 35, 96–104. Grieder, C., Mittweg, G., Dhillon, B.S., Montes, J.M., Orsini, E., Melchinger, A.E., 2012. Kinetics of methane fermentation yield in biogas reactors: genetic variation and association with chemical composition in maize. Biomass Bioenergy 37, 132– 141. Hashimoto, A.G., 1989. Effect of inoculum/substrate ratio on methane yield and production rate from straw. Biol. Wastes 28 (4), 247–255. ISO-11734, 1995. Water Quality – Evaluation of the ‘‘Ultimate’’ Anaerobic Biodegradability of Organic Compounds in Digested Sludge – Method by Measurement of the Biogas Production, first ed. International Organization for Standardization, Geneva. Jensen, P., Ge, H., Batstone, D.J., 2011. Assessing the role of biochemical methane potential tests in determining anaerobic degradability rate and extent. Water Sci. Technol. 64 (4), 880. Kafle, G.K., Kim, S.H., 2013. Anaerobic treatment of apple waste with swine manure for biogas production: batch and continuous operation. Appl. Energy 103, 61– 72. Koch, K., Fernández, Y.B., Drewes, J.E., 2015. Influence of headspace flushing on methane production in Biochemical Methane Potential (BMP) tests. Bioresour. Technol. 186, 173–178. Lianhua, L., Dong, L., Yongming, S., Longlong, M., Zhenhong, Y., Xiaoying, K., 2010. Effect of temperature and solid concentration on anaerobic digestion of rice straw in South China. Int. J. Hydrogen Energy 35 (13), 7261–7266.
Linke, B., 2006. Kinetic study of thermophilic anaerobic digestion of solid wastes from potato processing. Biomass Bioenergy 30 (10), 892–896. Lobry, J., Rosso, L., Flandrois, J., 1991. A FORTRAN subroutine for the determination of parameter confidence limits in non-linear models. Binary 3, 86–93. Maya-Altamira, L., Baun, A., Angelidaki, I., Schmidt, J.E., 2008. Influence of wastewater characteristics on methane potential in food-processing industry wastewaters. Water Res. 42 (8), 2195–2203. Møller, H.B., Sommer, S.G., Ahring, B.K., 2004. Methane productivity of manure, straw and solid fractions of manure. Biomass Bioenergy 26 (5), 485–495. Nges, I.A., Liu, J., 2009. Effects of anaerobic pre-treatment on the degradation of dewatered-sewage sludge. Renew. Energy 34 (7), 1795–1800. Nkemka, V.N., Murto, M., 2013. Biogas production from wheat straw in batch and UASB reactors: the roles of pretreatment and seaweed hydrolysate as a cosubstrate. Bioresour. Technol. 128, 164–172. Raposo, F., Banks, C., Siegert, I., Heaven, S., Borja, R., 2006. Influence of inoculum to substrate ratio on the biochemical methane potential of maize in batch tests. Process Biochem. 41 (6), 1444–1450. Raposo, F., Borja, R., Rincon, B., Jimenez, A.M., 2008. Assessment of process control parameters in the biochemical methane potential of sunflower oil cake. Biomass Bioenergy 32 (12), 1235–1244. Raposo, F., De la Rubia, M., Fernández-Cegrí, V., Borja, R., 2012. Anaerobic digestion of solid organic substrates in batch mode: an overview relating to methane yields and experimental procedures. Renew. Sustain. Energy Rev. 16 (1), 861– 877. Raposo, F., Fernández-Cegrí, V., De la Rubia, M., Borja, R., Béline, F., Cavinato, C., Demirer, G., Fernández, B., Fernández-Polanco, M., Frigon, J., 2011. Biochemical methane potential (BMP) of solid organic substrates: evaluation of anaerobic biodegradability using data from an international interlaboratory study. J. Chem. Technol. Biotechnol. 86 (8), 1088–1098. Sánchez, E., Borja, R., Weiland, P., Travieso, L., Martı´n, A., 2001. Effect of substrate concentration and temperature on the anaerobic digestion of piggery waste in a tropical climate. Process Biochem. 37 (5), 483–489. Seto, M., Alexander, M., 1985. Effect of bacterial density and substrate concentration on yield coefficients. Appl. Environ. Microbiol. 50 (5), 1132–1136. Strömberg, S., Nistor, M., Liu, J., 2014. Towards eliminating systematic errors caused by the experimental conditions in Biochemical Methane Potential (BMP) tests. Waste Manage. 34 (11), 1939–1948. Tanimu, M.I., Ghazi, T.I.M., Harun, M.R., Idris, A., 2014. Effect of feed loading on biogas methane production in batch mesophilic anaerobic digesters treating food waste. Int. J. Chem. Environ. Eng. 5 (1). Thomsen, S.T., Spliid, H., Østergård, H., 2014. Statistical prediction of biomethane potentials based on the composition of lignocellulosic biomass. Bioresour. Technol. 154, 80–86. VDI 4630, 2006. Fermentation of organic materials – characterisation of the substrate, sampling, collection of material data, fermentation tests. Zhang, W., Lang, Q., Wu, S., Li, W., Bah, H., Dong, R., 2014. Anaerobic digestion characteristics of pig manures depending on various growth stages and initial substrate concentrations in a scaled pig farm in Southern China. Bioresour. Technol. 156, 63–69.