Journal of Environmental Management 223 (2018) 917–924

Contents lists available at ScienceDirect

Journal of Environmental Management

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

Research article

Anaerobic co-digestion of foodwaste with liquid dairy manure or manure T

digestate: Co-substrate limitation and inhibition
Jonathan Masih-Dasa, Wendong Taob,∗
a
Division of Environmental Science, College of Environmental Science and Forestry, State University of New York, 1 Forestry Drive, Syracuse, NY, 13210, USA
b
Department of Environmental Resources Engineering, College of Environmental Science and Forestry, State University of New York, 1 Forestry Drive, Syracuse, NY,
13210, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Process instability has been a challenge to anaerobic digestion of foodwaste at higher organic loading rates. Co-
Anaerobic digestion digestion is one of the measures to improve stability. This study conducted batch experiments to compare liquid
Dairy manure dairy manure and dairy manure digestate as a co-substrate for anaerobic digestion of foodwaste. The batch co-
Digestate digestion experiments showed a two-stage biogas production process, which could be simulated with a mod-
Food waste
ification of the Gompertz model. The specific biogas yields derived with the two-stage biogas production model
Inhibition
was further simulated against the co-substrate ratios with substrate limitation – inhibition models for identifying
Kinetic model
the optimal co-substrate ratio. The Haldane model was the best to simulate co-substrate limitation – inhibition
kinetics in anaerobic co-digestion of foodwaste. A higher ratio of dairy manure could result in co-substrate
inhibition to biogas production due to recalcitrance of cellulose and toxicity of lignin and lignin derivatives.
Kinetic modeling shows that the optimal volatile solids (VS) ratio of liquid dairy manure is 16.6%, at which the
maximum specific methane yield is 0.54 L/g VS. Semi-continuous co-digestion of 88% foodwaste and 12% liquid
dairy manure at a hydraulic retention time of 14 d attained 94% of the simulated maximum methane yield.
Although co-digestion of foodwaste and manure digestate resulted in lower biogas yields than co-digestion with
liquid dairy manure, manure digestate is still an attractive co-substrate that has several operational advantages
compared with liquid dairy manure.

1. Introduction
Laboratory anaerobic digestion studies have demonstrated high
biochemical methane potential of foodwaste. Process instability, how-
ever, has been a challenge to long-term, continuous anaerobic digestion
of foodwaste, especially at a higher organic loading rate and under
thermophilic conditions (Banks et al., 2012; Braguglia et al., 2018;
Kawai et al., 2014; Komilis et al., 2017; Zhang et al., 2012, 2015). The
process instability and failure has been mostly attributed to fast acid-
ification, ammonia inhibition to acetoclastic methanogens and sub-
sequent accumulation of volatile fatty acids, and inadequate avail-
ability of trace elements. The high contents of non-fiber carbohydrate
and fat of foodwaste may lead to fast acidification. Ammonia accu-
mulation is attributed to its high protein content. Consequently, suc-
cessful long-term mono-digestion of foodwaste has been limited to or-
ganic loading rates typically less than 4.5 g VS/L/d unless an
enhancement measure such as co-digestion is taken (Braguglia et al.,
2018; Komilis et al., 2017).
Animal manure has been the most common co-substrate for food-
waste digestion (Komilis et al., 2017). Earlier studies on co-digestion of
foodwaste and liquid dairy manure (Agyeman and Tao, 2014; Usack
and Angenent, 2015; Zarkadas et al., 2015) reported stable operation
and synergistic effects at organic loading rates as high as 6.2–12 g VS/
L/d, which were attributed to balanced biochemical composition, buf-
fered pH, and supplementation of micronutrients from co-substrates.
Because of the high fiber content of liquid dairy manure (LDM), anae-
robically digested dairy manure (ADDM) from digesters without an
extended hydraulic retention time can have higher residual biomethane
potential. Based on the measured biochemical composition of LDM and
ADDM at a dairy farm (Table 1), it was estimated with an equation
developed by Angelidaki and Sanders (2004) that the ADDM would
have a biomethane potential close to that of the LDM. The higher
biomethane potential of ADDM was mainly attributed to the increased
content of crude protein and non-fiber carbohydrate due to microbial
growth in the digesters. Moreover, the ADDM had higher nutrient
concentrations than LDM (Table S1), which could be attributed to

∗
Corresponding author.
E-mail address: wtao@esf.edu (W. Tao).

https://doi.org/10.1016/j.jenvman.2018.07.016
Received 13 November 2017; Received in revised form 19 May 2018; Accepted 6 July 2018
Available online 10 July 2018
0301-4797/ © 2018 Elsevier Ltd. All rights reserved.
J. Masih-Das, W. Tao Journal of Environmental Management 223 (2018) 917–924

Table 1 A screw press was used at the farm to remove coarse particles from the
Characteristics of substrates and inoculum used in this study.a digester effluent. The LDM was collected from the same farm at the
Main substrate Co-substrate Inoculum same time as the ADDM. Each digester was inoculated at 21.345 g VS
with anaerobically digested sludge and ADDM sieved through 0.5-mm
b
Foodwaste LDM ADDM Digested Digested mesh in the same VS ratios as the targeted foodwaste to co-substrate
manurec sludge ratios. Substrate to inoculum VS ratio was set at 0.82. The headspace
pH 4.8 7.7 7.9 7.6 7.5 was flushed with nitrogen gas and sealed immediately. The substrates
Total solids (g/L) 272.5 123.0 45.5 48.5 26.1 and inocula were characterized by Dairy One Forage Laboratory in
VS/TS (%) 92.1 84.6 68.2 67.4 55.4 Ithaca, New York, USA and the results are given in Table 1 and Table
Biochemical composition (% TS) S1. Biochemical composition was determined with the ANKOM Tech-
Crude protein 27.8 16.9 31.9 30.0 31.8
nology and AOAC methods. Elemental content was determined for
Crude fat 22.4 6.1 3.7 5.3 3.3
Non-fiber 35.1 7.5 11.6 7.9 4.9 samples after microwave-accelerated acid digestion, using a Thermo
carbohydrate iCAP 6300 inductively coupled plasma radial spectrometer.
Lignin 2.8 10.6 16.2 9.5 5.6 After 2 d of acclimation, the digesters were fed with substrates and
Cellulose 0.9 23.4 7.9 3.4 1.6
deionized water to a working volume of 1.8 L and set at temperature
Hemicellulose 3.2 21.9 8.0 11.4 14.3
36 °C. Biogas production was recorded every 15 min with Model 1615A
a
Average of two samples. digital mass flow gas meters (Omega Engineering, Norwalk,
b
ADDM after a screw press. Connecticut, USA), displaying biogas flow rate at temperature 25 °C and
c
ADDM passing a 0.5-mm sieve. pressure 1 atm. Based on the dynamics of biogas production rate, the
batch experiments lasted for 25–30 d. Biogas samples (0.1 mL each)
microbial assimilation and mineralization of nutrients. ADDM also has were collected weekly with a gas tight syringe and diluted with air in
greater alkalinity than LDM (Akhiar et al., 2017; Tao et al., 2016). 11.3-mL Restek serum vials for determination of methane content using
Therefore, ADDM can even be a better co-substrate than LDM. a Shimadzu GC-2014 gas chromatograph (Agyeman and Tao, 2014).
Synergistic effects of co-digestion on methane production have been This GC system used a HayeSep-D column to separate N2O, a Porapak-N
reported at different co-substrate ratios, whereas overdosing of a co- pre-column to backflush C2 compounds, a Porapak-N column to sepa-
substrate may result in disappearance of the synergies and even an- rate CO2, and a MS-13× column to separate air/CH4/CO into in-
tagonistic effects. The decreased performance of co-digestion, even dividual gases. Methane is detected with a flame ionization detector.
after extended acclimatization periods, has been attributed to the ef- Helium was used as carrier gas. The temperatures of oven, injector port
fects of a co-substrate that contains inhibitory or toxic compounds such and detector were 100, 140 and 100 °C, respectively. Measurements
as ammonia and other inorganic salts (Regueiro et al., 2015; Usack and were taken in the digesters for pH with a Hach H160 meter connected
Angenent, 2015). The optimum co-substrate ratio is typically de- to an ISFET NMR tube pH probe before feeding and at the end of each
termined through laboratory experiments at discrete substrate combi- batch. Initial and final concentrations of total solids (TS) and VS were
nations or by modeling without consideration of co-substrate inhibition determined for each batch according to Standard Methods 2540 B and E
(Cook et al., 2017; Owamah and Izinyon, 2015; Usack and Angenent, (APHA et al., 2012). Samples collected initially and at the end of each
2015). batch were centrifuged at 2500 g for 20 min and centrate samples were
The first objective of this study was to compare the effects of two co- diluted for determination of total ammonia nitrogen (TAN) concentra-
substrates for foodwaste digestion, i.e., LDM and ADDM, by means of tion with a QuickChem 8500 series automatic flow injection analyzer
batch anaerobic co-digestion experiments. Side-by-side comparison of (LaChat Instruments, Loveland, Colorado, USA), following Standard
the co-substrates could not only discover a new enhancement measure Method flow injection analysis (APHA et al., 2012).
for anaerobic digestion of foodwaste, but also demonstrate a new Both the commonly used first-order and Gompertz models
strategy for identifying a co-substrate. Based on specific biogas yields in (Equations (1) and (2)) were tested to fit the dynamics of cumulative
the batch co-digestion experiments at different co-substrate ratios, this biogas production in the individual batches. The first-order model fitted
study further used co-substrate limitation and inhibition models to si- the biogas production dynamics in the blanks well (R2 = 0.960 and
mulate the kinetics of co-substrate effects on specific biogas yield. By 0.996). The Gompertz model fitted the initial biogas accumulation
curve fitting to typical kinetic models, the second objective was to poorly and sometimes put out negative lag phase times. Finally, the
identify the best kinetic model and subsequently estimate the optimal Gompertz model was modified (Equation (3)) to fit the biogas pro-
co-substrate ratio for anaerobic co-digestion of foodwaste. Finally, the duction dynamics in the anaerobic co-digestion experiments, which
performance of anaerobic co-digestion at the optimal substrate com- showed two stages of biogas production. The kinetic constants were
bination was tested through semi-continuous anaerobic co-digestion of estimated simultaneously for each batch experiment using the Microsoft
foodwaste and liquid dairy manure. Excel 2013 Solver tool.

Yt = Ym (1 − exp (−kt )) (1)

2. Materials and methods
Rm × e
2.1. Batch experiments on anaerobic co-digestion Yt = Ym exp ⎧ −exp ⎡ (λ − t ) + 1⎤ ⎫
⎨
⎩ ⎢
⎣ Ym ⎥
⎦⎬⎭ (2)
Batch anaerobic digestion experiments were conducted in 2-L con-
tinuously-stirred mesophilic digesters (Agyeman and Tao, 2014). For t R ×e
Yt = Ym1 + Ym2 exp ⎧−exp ⎡ m (λ − t ) + 1⎤ ⎫
each co-substrate (LDM and ADDM), batch co-digestion experiments K+t ⎨
⎩ ⎢
⎣ Ym2 ⎥
⎦⎬⎭ (3)
were conducted at 7 substrate combinations, i.e., foodwaste VS: co-
substrate VS in 95%: 5%, 90%: 10%, 85%: 15%, 80%: 20%, 70%: 30%, where t = time from adding substrates in a batch (d); Yt = specific
60%: 40%, and 50%: 50%, plus two blanks (100% each inoculum). The biogas production at t (L/g VS); Ym = specific biogas yield (L/g VS);
foodwaste was collected over two weeks from buffet leftovers at a ca- Ym1, Ym2 = specific biogas yield at the first and second stages (L/g VS);
sino in central New York, USA, ground through a 2.5-mm-aperture k = first-order digestion rate coefficient (1/d); K = first-stage half sa-
cutting plate, and stored at −21 °C. The ADDM was the effluent of turation time (d); λ = length of lag phase (d); Rm = maximum specific
mesophilic manure digesters at a dairy farm in Skaneateles, New York. biogas production rate (L/d/g VS); and e = Euler's number = 2.718.

918
J. Masih-Das, W. Tao Journal of Environmental Management 223 (2018) 917–924

2.2. Modeling co-substrate limitation – inhibition kinetics Table 2

Solids removal performance and operational conditions of batch anaerobic co-
The variations of specific biogas yield (Ym1 + Ym2 – Ym,inoculum) with digestion of foodwaste and different co-substrates.
co-substrate ratio were fitted with five kinetic models which account for Ratio of co-substrate VS in total substrate VS (%)
substrate limitation and inhibition effects, i.e., the Andrews model in
Equation (4), Edwards model in Equation (5), Haldane model in 5 10 15 20 30 40 50
Equation (6), Aiba model in Equation (7), and an extended Monod
Liquid dairy manure as co-substratea
model in Equation (8) (Andrews, 1968; Edwards, 1970; Sung and Liu, TS reduction 69 67 74 75 72 74 51
2003). The kinetic constants were estimated simultaneously using the (%)
Microsoft Excel 2013 Solver tool. In addition, quadratic function as VS reduction 75 73 77 81 77 77 56
(%)
given in Equation (9) was fitted to the experimental data using the least
Initial pH 7.57 7.60 7.52 7.61 7.58 7.70 7.63
squares method. The variances of the experimental data and model Final pH 7.51 7.42 7.36 7.52 7.48 7.51 7.35
predictions were compared by F-test, giving P-value. Residuals between Initial TAN 958 941 924 907 873 838 804
the experimental data and model predictions were run through a paired (mg/L)
t-test to identify whether the means of the experimental data and model Final TAN (mg/ 1650 1510 1580 1425 1555 1600 1700
L)
predictions were different. The similarity between the experimental a
Anaerobically digested dairy manure as co-substrate
data and model predictions was then assessed with coefficient of de- TS reduction 27 76 55 52 63 50 55
termination, R2. (%)
VS reduction 43 82 68 66 67 59 59
C (%)
q = qmax
C2 Initial pH 7.54 7.53 7.73 7.76 7.75 7.80 7.67
Kc + C + Ki (4) Final pH 7.48 7.41 7.51 7.50 7.45 7.57 7.51
Initial TAN 958 941 924 907 873 838 804
−C −C (mg/L)
q = qmax ⎛e Ki −e Kc ⎞
⎝ ⎠ (5) Final TAN (mg/ 1380 1785 1380 1325 1410 1585 1910
L)
C
q = qmax a
Average of two samples.
(K c + C ) 1 + ( C
Ki ) (6)
digestate. The digestate temperature was set at 36 °C. Hydraulic re-
Ce −C / Ki tention time was set at 14 d by discharging 129 mL of digestate daily
q = qmax
Kc + C (7) and diluting the feed with tap water to 129 mL for each digester.
n Organic loading rate was increased every 14 d from 1.0 to 1.5, and 2.0 g
C⎞ S
q = qmax ⎛1 −
⎜ ⎟ VS/L/d. Unfortunately, biogas production dropped substantially due to
C m
⎝ Ci⎠ S + K 1 −
s ( )
Ci (8) an unexpected heat shock in the laboratory on Day 29–32 before op-
erating the digesters longer at higher organic loading rates.
q = aC 2 + bC + d (9) Biogas flow rate, pressure and temperature were recorded every
15 min with the digital mass flow meters. Biogas samples (0.1 mL each)
where q = the specific biogas yield observed at a given co-substrate
were collected weekly for determination of methane content using the
ratio (L/g VS); qmax = the theoretical maximum or potential of specific
same methods as for batch experiments. Digester effluent was de-
biogas yield (L/g VS); C = the VS ratio of a co-substrate over the total
termined weekly for pH, temperature, TS, VS and TAN, using the same
substrate VS (%); Kc = the half saturation constant of the co-substrate
methods as described in Section 2.1.
(%); Ki = the half inhibition constant of the co-substrate (%); Ks = the
half saturation constant of the main substrate i.e., foodwaste (%);
Ci = the lethal ratio of the co-substrate (%); m = 0 because of concerns 3. Results and discussion
at high C values; and a, b, d, n = constants. Kc and Ki represent the
lowest and highest co-substrate concentrations, respectively, at which 3.1. Assessment of batch anaerobic co-digestion performance
the biogas yield equals to one half the maximum yield.
The operational conditions in the batch anaerobic co-digestion ex-
2.3. Semi-continuous co-digestion of foodwaste and liquid dairy manure periments are given in Table 2. The measured initial and final pH values
were similar across all the batch experiments. The pH values were in
Batch anaerobic digestion experiments reveal the ultimate biogas the range for stable anaerobic digestion, pH 6.5–8.2 (Cook et al., 2017;
yield, although it may not reach the biomethane potential. The latter is Mao et al., 2015). The slight pH decreases through the batch digestion
determined under optimum conditions while the former could be sub- experiments were attributed to the high pH buffering capacity of LDM
jected to inhibition at some co-substrate ratios. Moreover, the max- and ADDM (Tao et al., 2016). The inoculum sources had TAN con-
imum specific biogas yield may still not be attained in continuous op- centrations at 1170 mg/L in the sludge digestate and 1675 mg/L in the
eration of anaerobic digesters because of limited hydraulic retention sieved ADDM. Correspondingly, the initial TAN concentrations were
time and continuous organic loading. Therefore, semi-continuous estimated for the batch experiments at the specified co-substrate ratios.
anaerobic digestion was performed to evaluate the performance of Through batch anaerobic digestion, TAN concentration increased by
anaerobic digestion at the optimal co-substrate ratio. Based on the re- 44–138%. Both initial and final TAN concentrations, however, were in
sults of the batch co-digestion experiments, four mesophilic digesters the range for stable anaerobic digestion, less than 5000 mg/L (Cook
were operated at 88% foodwaste and 12% ADDM, which was close to et al., 2017), and generally below the inhibition levels (1500–7000 mg/
the finally simulated optimal co-substrate ratio. The four digesters were L) as reviewed by Yenigun and Demirel (2013), Rajagopal et al. (2013),
operated in parallel under the same conditions for 64 days. Inoculum and Chen et al. (2008). Upon feeding, the initial TS and VS con-
(1.8 L each digester) was made of the sludge digestate (88% VS) and centrations were 2.8–3.9% and 1.9–2.7%, respectively. TS and VS
ADDM (12% VS). The headspace was purged with nitrogen gas and concentrations decreased mostly by 50–76% and 56–82%, respectively.
sealed immediately. After 2 d of acclimation, it started to feed daily There were insignificant differences in TS and VS reduction efficiencies
through a chute mounted on the digester cap and submerged in between ADDM and LDM (P = 0.07 and 0.16). The insignificant

919
J. Masih-Das, W. Tao
Fig. 1. Biogas production over time in batch anaerobic co-digestion of foodwaste with different VS ratios of a) anaerobically digested dairy manure (ADDM), and b)
liquid dairy manure (LDM).
differences could partially be attributed to no more than 50% co-sub-
strate combined with foodwaste which could be hydrolyzed faster than
undigested and digested dairy manure. Biogas in all the batches had
methane content varied between 58% and 68%, which increased over
the first 10 d. Because biogas production rate had substantial variations,
both hourly and daily as illustrated in Fig. S1, and there were only
weekly biogas sampling and determination of methane content, the
discussion hereinafter is primarily based on the continuously recorded
biogas data instead of methane production.
There were steep initial increases in cumulative biogas production
at all the substrate combinations as shown in Fig. 1. It tended to sta-
bilize in the first 4–5 d, indicating anaerobic oxidation of readily
available organic compounds coming with the substrates. Cumulative
biogas production increased again after a lag phase and tended to sta-
bilize again after approximately 25 d. Although the cumulative biogas
production could be simulated with the Gompertz model (Equation (2))
outputting high R2 values, the Gompertz model did not reflect the two-
stage stabilization process and could generate negative lag phase times
as illustrated in Fig. S2. The two-stage biogas production model
(Equation (3)) strongly fitted the biogas production process in the co-
digestion experiments (R2 = 0.993–0.998), and eliminated the errors in
estimation of lag phase time.
The batches with LDM yielded more biogas than those with ADDM
except for the similar biogas yields at 5% co-substrate (Fig. 2). Using
Equation (10) (Angelidaki and Sanders, 2004), however, biomethane
potential was estimated to be slightly higher for LDM (0.41 L CH4/g VS)
than ADDM (0.38 L CH4/g VS). When excluding the non-biodegradable
lignin from carbohydrate in Equation (10), the biomethane potentials
were still similar between LDM (0.36 L CH4/g VS) and ADDM (0.31 L
CH4/g VS). When lignin is present in a substrate at a high concentra-
tion, biomethane potential can be low, not only because of its non-
biodegradability under anaerobic conditions but also its toxicity
(Koyama et al., 2017; Li et al., 2013; Chen et al., 2008). The mechanism
of lignin inhibition, especially to hydrolytic activity, can be irreversible
adsorption of cellulolytic enzymes on dissolved lignin (Koyama et al.,
2017). Li et al. (2013) found a negative linear relationship between
lignin content and experimental methane yield for fiber-rich substrates
(Equation (11)). Using Equation (11), it was estimated that the LDM
would have a higher methane yield (0.24 L CH4/g VS) than the ADDM
(0.17 L CH4/g VS). Therefore, experimental biogas yield is likely a
function of the contents of crude protein, lipids and non-fiber carbo-
hydrate as well as lignin content. When looking for a potential source of
co-substrate, it is important not only to evaluate biomethane potential
under ideal incubation conditions but also avoid inhibition due to toxic
components in a co-substrate.
Pmethane = 0.496X + 1.014Y + 0.415Z (10)
920
Journal of Environmental Management 223 (2018) 917–924
Ym = 0.374 − 1.249CLignin (11)
where Pmethane = biomethane potential (L CH4/g VS);
Ym = experimental methane yield (L CH4/g VS); and Clignin, X, Y,
Z = fraction of lignin, crude protein, lipid, and carbohydrate in VS.
3.2. Kinetics of anaerobic digestion from co-substrate limitation to
inhibition
At low ratios of a co-substrate, the co-substrate serves as a supple-
mentary source of organic carbon and nutrients for microorganisms. At
a high ratio, the co-substrate may introduce inhibitors to microbial
growth, such as lignin. Inhibitory conditions were found to be the main
cause of microbiome divergence in anaerobic digesters fed at increasing
organic loading rates with dairy manure and different co-substrates
(Regueiro et al., 2015). The values of lag phase length estimated with
the two-stage biogas production model are presented in Fig. 3. The
length of lag phase generally increased with the increasing LDM ratio.
Increased lag phase length is the primary result of inhibition (Andrews,
1968; Li et al., 2016). The longer lag phase at a higher LDM ratio could
be attributed to the higher concentrations of recalcitrant cellulose and
non-digestible, toxic lignin in the combined substrate and inoculum
(Table S2). The lag phase lengths in batch co-digestion with LDM had a
strong correlation with the initial concentrations of lignin and cellulose
(P = 0.009; R2 = 0.89). Although lignin and cellulose concentrations in
co-digestion with ADDM increased as the ADDM ratio increased, there
was no clear relationship of lag phase length and concentrations of
lignin and cellulose, suggesting that ADDM could have other inhibitors
such as lignin derivatives and other toxic phenolic compounds (Chen
et al., 2008).
Besides a longer lag phase, co-substrate inhibition may set off the
effects of enhancers such as micronutrients in the co-substrate. At small
co-substrate ratios, anaerobic digestion is enhanced more at a greater
co-substrate ratio or the co-substrate is a limiting factor for microbial
activity. The kinetics of substrate limitation on anaerobic digestion has
been largely simulated with the Michaelis-Menten equation or its
modifications. Further increase in co-substrate ratio may result in de-
crease in microbial activity due to co-substrate inhibition. Although
inhibition factors have been introduced to assess substrate inhibition to
anaerobic digestion (Owamah and Izinyon, 2015; Zamanzadeh et al.,
2016), substrate inhibition due to over-dosing of a co-substrate in
anaerobic digestion has rarely been simulated except for the extended
Monod model described by Sung and Liu (2003). Inhibition often in-
volves multiple, rather than specific mechanisms, such as reduced ac-
tivity of an enzyme by complexing with excess substrate, modification
of physiochemical variables, and imbalanced metabolism of the cells
(Chen et al., 2008; Edwards, 1970). As Edwards (1970) concluded upon
comparison of kinetic models that significant variations may occur in
J. Masih-Das, W. Tao
Fig. 2. Curve fitting of co-substrate limitation-inhibition kinetics using a) Andrews (Equ. 4); b) Aiba (Equ. 7); c) Haldane (Equ. 6); d) Edwards (Equ. 5); e) extended
Monod (Equ. 8); and f) quadratic (Equ. 9) models for anaerobic co-digestion of foodwaste with ADDM and LDM.
Fig. 3. Variation of lag phase length with co-substrate ratio in anaerobic co-
digestion of foodwaste with anaerobically digested dairy manure (ADDM) and
liquid dairy manure (LDM).
the functional form of substrate inhibition, this study fitted five sub-
strate limitation-inhibition kinetic models to the results of the batch
digestion experiments plus the quadratic function. The Haldane and
Andrews models were derived with the assumption of inactive enzyme-
substrate complex formation. The semi-empirical Edwards model
921
Journal of Environmental Management 223 (2018) 917–924
assumes a protective diffusion-limitation of high and inhibitory sub-
strate concentrations. The Aiba model is an empirical equation. Zhang
et al. (2017) used non-linear regression of the quadratic function to
interpret the effects of crude fat content on anaerobic digestion of
foodwaste, but the quadratic function does not have any kinetic con-
stant.
The kinetic constants are given in Table 3. Haldane model had the
best fit, as displayed in Fig. 2 and indicated in Table S3 by the higher P-
values of F-test for assessing the difference in variances of the experi-
mental data and model simulations, the higher P-values of paired t-test
for assessing closeness of the two means, and the highest R2 values of
non-linear curve fitting. Co-digestion of foodwaste with ADDM was also
fitted well with the Andrews model, while co-digestion of foodwaste
with LDM was also fitted well with the Edwards model. The Aiba model
had relatively lower R2 values and P-values of F-test for both co-sub-
strates. The extended Monod model and quadratic function simulated
the experimental data on co-digestion with LDM poorly, with very small
R2 values and P-values of F-test. Moreover, the constants of a quadratic
function do not have any physical meanings.
Simulation with the best kinetic model (Fig. 2c) showed that the
specific biogas yield increased as LDM VS ratio was increased up to
16.6%, then decreased due to inhibition. Based on the model simulation
results (Fig. 2), the optimal ratio of ADDM as a co-substrate may not
exist or less than 5%. At the optimal LDM ratio, the maximum yield at
standard temperature of 0 °C and pressure of 1 atm (stp) was 0.79 L
J. Masih-Das, W. Tao Journal of Environmental Management 223 (2018) 917–924

Table 3 (stp) biogas/g VS or 0.54 L (stp) CH4/g VS. This estimate of specific
Kinetic constants obtained from curve fitting of co-substrate limitation-inhibi- methane yield is higher than those reported by earlier studies on co-
tion models to specific biogas yield against co-substrate VS ratio. digestion of foodwaste and LDM (Table 4) and falls into the upper end
Parameter Co-substrate limitation – inhibition model among mono- and co-digestion of foodwaste as reviewed by Braguglia
et al. (2018) and Komilis et al. (2017). The earlier studies compared
Andrews Edwards Haldane Aiba Extended methane yield in co-digestion of foodwaste and LDM at higher LDM
(Equation (Equation (Equation (Equation Monod
ratios than this study and the highest methane yields were attained at
(4)) (5)) (6)) (7)) (Equation
(8)) their lowest LDM ratios (Table 4).

Anaerobically digested dairy manure as co-substrate

qm (L/g VS) 0.73 0.71 0.73 0.71 0.71 3.3. Digested versus undigested dairy manure as co-substrate
Kc (%) 0.0 0.21 0.0 0.0 –
Ki (%) 79.6 107 79.6 107 – As discussed in Section 3.1, co-digestion of foodwaste with LDM
Ci (%) – – – – 9406
yields more biogas. However, co-digestion with less than approximately
Liquid dairy manure as co-substrate
qm (L/g VS) 1.15 0.93 1.18 1.12 0.89 10% ADDM still has an attractive biogas yield. Digestate is typically low
Kc (%) 2.7 3.2 2.8 2.5 – in residual biomethane potential. As Table 1 shows, however, proteins
Ki (%) 100 225 97.5 135 – are enriched after anaerobic digestion of both municipal sludge and
Ci (%) – – – – 240 dairy manure. The higher protein content of the digestate upon mono-
digestion of LDM renders a relatively higher residual methane potential
–: not applicable.
of ADDM. Besides, ADDM has higher nutrient content than LDM and
foodwaste on volumetric and dry matter bases with few exceptions
Table 4
(Table S1 and Table S4), thus supplementing more macro- and micro-
Comparison of specific methane yields in anaerobic co-digestion of foodwaste
and liquid dairy manure (LDM).
nutrients for co-digestion of foodwaste. Recent studies on anaerobic
digestion of foodwaste have verified that supplementation of trace
Reference Best VS VS ratio of LDM Methane yield at elements such as selenium, nickel, cobalt and iron improves process
ratio of tested (%) optimum LDM ratio (L
LDM (%) (stp)/g VS)
stability and methane production (Banks et al., 2012; Zhang et al.,
2015). The initial concentrations of macro- and micro-nutrients in the
Usack and Angenent 51 51–100 0.30 batch experiments mostly fell into the suggested ranges except for
(2015) calcium and iron. All the macro-nutrients had concentrations in the
El-Mashad and 52 52 and 68 0.31
batch experiments far below the 50% inhibition levels (Table S4).
Zhang (2010)
Zarkadas et al. 29.2 29.2–100 0.41 The lower TS concentration makes ADDM a better co-substrate to
(2015) dilute the thick foodwaste. The ADDM inhibition kinetics (Table 3)
This study 16.6 5–50 0.54 provide insights to recirculating digestate to dilute dry feedstock or
recycling inside a hybrid digestion process (Kusch et al., 2012; Li et al.,
2016; Shi et al., 2014; Wu et al., 2017; Xu and Li, 2012; Zamanzadeh
et al., 2016). Besides, it has less health and odor concerns when
transporting ADDM instead of LDM from a farm for co-digestion at a

Fig. 4. Irreversible impact of a temperature jump during days 29–32 on anaerobic co-digestion of foodwaste and liquid dairy manure. Error bar = 1 standard
deviation of 4 digesters. The results on Day 43–64 are not presented because it did not indicate recovery of biogas production.

922
J. Masih-Das, W. Tao
foodwaste processing site (Risberg et al., 2017). Furthermore, farms are
often more willing to provide ADDM than LDM as LDM has traditionally
been land-applied as a liquid fertilizer.
3.4. Semi-continuous co-digestion of foodwaste and liquid dairy manure
When organic loading rate was increased stepwise from 1.0 to 2.0 g
VS/L/d in the semi-continuous co-digestion at a substrate VS combi-
nation of 88% foodwaste and 12% LDM, biogas production rate in-
creased steadily to 1.78 L/Lreactor/d (Fig. 4a) with 68.0% methane in the
biogas. Unfortunately, biogas production dropped to approximately
0.48 L/Lreactor/d due to an unexpected room temperature increase and a
subsequent biogas temperature jump of approximately 13 °C on days
29–32 (Fig. 4b). In the last 6 d before this mishap, nevertheless, specific
biogas yield reached 0.75 L (stp)/g VS and specific methane yield 0.51 L
(stp)/g VS, which were already 94% of the simulated maximum specific
biogas yield despite the shorter hydraulic retention time. As reviewed
by Komilis et al. (2017), the averages of specific methane yield and
hydraulic retention time for continuous anaerobic mono- and co-di-
gestion of foodwaste were 0.42 L/g VS and 36.7 d, respectively. The
higher methane yield achieved by the present study confirmed the
optimized, low LDM ratio needed for co-digestion of foodwaste.
This 3-d temperature shock resulted in a sharp decrease of digestate
pH (Fig. 4c). On day 32 sodium bicarbonate was added to adjust pH to
above 7 and the co-substrate ratio was increased from 12% to 30% for
increased pH buffering by LDM. Weekly measurements of digester ef-
fluent showed a mean digestate temperature of 35.4 °C with a small
standard deviation (0.6 °C). Digestate TAN concentration was between
629 and 1189 mg/L, which were probably non-inhibitory (Chen et al.,
2008; Rajagopal et al., 2013; Yenigun and Demirel, 2013). Although pH
was adjusted to between 6.68 and 7.02 in the 14 days following the
biogas temperature jump, the biogas production rate was never re-
covered and TS and VS concentrations kept increasing (Fig. 4d). This
type of incidents rarely occurs and has not been addressed yet, but it
indicated the influence of headspace temperature fluctuation on di-
gestion stability. Practically, similar headspace temperature jumps may
occur in anaerobic digesters operated outdoor in temperate climates.
The microbial response to a temperature shock from headspace needs to
be investigated in the future.
4. Conclusions
Batch anaerobic co-digestion of foodwaste and dairy manure pro-
duces biogas in two stages, which could be better simulated with a
modified Gompertz model than the original Gompertz function.
A lower ratio of a co-substrate improves biogas production from
anaerobic co-digestion of foodwaste. Overdosing of a co-substrate,
however, results in inhibition due to inhibitors such as lignin in LDM
and ADDM.
The Haldane model can highly simulate the combined limitation
and inhibition effects of a co-substrate on specific biogas yield. Unlike
earlier studies that compared co-digestion of foodwaste at higher ratios
of LDM, this study found that biogas yield reached the highest when co-
digesting foodwaste with only 16.6% LDM. Meanwhile, the small VS
ratio of LDM needed for efficient digestion of foodwaste suggests that
the ratio of foodwaste as a co-substrate is less concerned when dairy
manure is the main substrate for co-digestion at dairy farms.
Although co-digestion of foodwaste with ADDM yielded less biogas
than co-digestion with LDM, manure digestate can still be an attractive
co-substrate for anaerobic co-digestion of foodwaste, considering less
health and odor concerns in handling ADDM and its greater potential
for diluting main substrate, buffering pH and supplementing trace
elements.
923
Journal of Environmental Management 223 (2018) 917–924
Acknowledgements
This study was supported by the U.S. Environmental Protection
Agency [SU835937]. The views expressed in this document are solely
those of the authors and do not necessarily reflect those of the Agency.
Appendix A. Supplementary data
Supplementary data related to this article can be found at https://
doi.org/10.1016/j.jenvman.2018.07.016.
References
Agyeman, F., Tao, W., 2014. Anaerobic co-digestion of food waste and dairy manure:
effects of food waste particle size and organic loading rate. J. Environ. Manag. 133,
268–274.
Akhiar, A., Battimelli, A., Torrijos, M., Carrere, H., 2017. Comprehensive characterization
of the liquid fraction of digestates from full-scale anaerobic co-digestion. Waste
Manag. 59, 118–128.
Andrews, J.F., 1968. A mathematical model for the continuous culture of microorganisms
utilizing inhibitory substrates. Biotechnol. Bioeng. 10, 707–723.
Angelidaki, I., Sanders, W., 2004. Assessment of the anaerobic biodegradability of mac-
ropollutants. Rev. Environ. Sci. Biotechnol. 3, 117–129.
APHA, AWWA, WEF, 2012. Standard Methods for the Examination of Water and
Wastewater, twenty-second ed. American Public Health Association/American Water
Works Association/Water Environment Federation, Washington, DC.
Banks, C.J., Zhang, Y., Jiang, Y., Heaven, S., 2012. Trace element requirements for stable
food waste digestion at elevated ammonia concentrations. Bioresour. Technol. 104,
127–135.
Braguglia, C.M., Gallipoli, A., Gianico, A., Pagliaccia, P., 2018. Anaerobic bioconversion
of food waste into energy: a critical review. Bioresour. Technol. 248, 37–56.
Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: a
review. Bioresour. Technol. 99, 4044–4064.
Cook, S.M., Skerlos, S.J., Raskin, L., Love, N.G., 2017. A stability assessment tool for
anaerobic codigestion. Water Res. 112, 19–28.
Edwards, V.H., 1970. The influence of high substrate concentration on microbial kinetics.
Biotechnol. Bioeng. 12, 679–712.
El-Mashad, H.M., Zhang, R., 2010. Biogas production from co-digestion of dairy manure
and food waste. Bioresour. Technol. 101, 4021–4028.
Kawai, M., Nagao, N., Tajima, N., Niwa, C., Matsuyama, T., Toda, T., 2014. The effect of
the labile organic fraction in food waste and the substrate/inoculum ratio on anae-
robic digestion for a reliable methane yield. Bioresour. Technol. 157, 174–180.
Komilis, D., Barrena, R., Grando, R.L., Vogiatzi, V., Sánchez, A., Font, X., 2017. A state of
the art literature review on anaerobic digestion of food waste: influential operating
parameters on methane yield. Rev. Environ. Sci. Biotechnol. 16, 347–360.
Koyama, M., Yamamoto, S., Ishikawa, K., Ban, S., Toda, T., 2017. Inhibition of anaerobic
digestion by dissolved lignin derived from alkaline pre-treatment of an aquatic
macrophyte. Chem. Eng. J. 311, 55–62.
Kusch, S., Oechsner, H., Jungbluth, T., 2012. Effect of various leachate recirculation
strategies on batch anaerobic digestion of solid substrates. Int. J. Environ. Waste
Manag. 9, 69–88.
Li, Y.-F., Shi, J., Nelson, M.C., Chen, P.H., Graf, J., Li, Y., Yu, Z., 2016. Impact of different
ratios of feedstock to liquid anaerobic digestion effluent on the performance and
microbiome of solid-state anaerobic digesters digesting corn stover. Bioresour.
Technol. 200, 744–752.
Li, Y., Zhang, R., Liu, G., Chen, C., He, Y., Liu, X., 2013. Comparison of methane pro-
duction potential, biodegradability, and kinetics of different organic substrates.
Bioresour. Technol. 149, 565–569.
Mao, C., Feng, Y., Wang, X., Ren, G., 2015. Review on research achievements of biogas
from anaerobic digestion. Renew. Sustain. Energy Rev. 45, 540–555.
Owamah, H.I., Izinyon, O.C., 2015. Development of simple-to-apply biogas kinetic models
for the co-digestion of food waste and maize husk. Bioresour. Technol. 194, 81–90.
Rajagopal, R., Masse, D.I., Singh, G., 2013. A critical review on inhibition of anaerobic
digestion process by excess ammonia. Bioresour. Technol. 143, 632–641.
Regueiro, L., Spirito, C.M., Usack, J.G., Hospodsky, D., Werner, J.J., Angenent, L.T., 2015.
Comparing the inhibitory thresholds of dairy manure co-digesters after prolonged
acclimation periods: Part 2 – correlations between microbiomes and environment.
Water Res. 87, 458–466.
Risberg, K., Cederlund, H., Pell, M., Arthurson, V., Schnürer, A., 2017. Comparative
characterization of digestate versus pig slurry and cow manure – chemical compo-
sition and effects on soil microbial activity. Waste Manag. 61, 529–538.
Shi, J., Xu, F., Wang, Z., Stiverson, J.A., Yu, Z., Li, Y., 2014. Effects of microbial and non-
microbial factors of liquid anaerobic digestion effluent as inoculum on solid-state
anaerobic digestion of corn stover. Bioresour. Technol. 157, 188–196.
Sung, S., Liu, T., 2003. Ammonia inhibition on thermophilic anaerobic digestion.
Chemosphere 53, 43–52.
Tao, W., Fattah, K.P., Huchzermeier, M.P., 2016. Struvite recovery from anaerobically
digested dairy manure: a review of application potential and hindrances. J. Environ.
Manag. 169, 46–57.
Usack, J.G., Angenent, L.T., 2015. Comparing the inhibitory thresholds of dairy manure
co-digesters after prolonged acclimation periods: Part 1 – performance and operating
limits. Water Res. 87, 446–457.
J. Masih-Das, W. Tao
Wu, D., Lü, F., Shao, L., He, P., 2017. Effect of cycle digestion time and solid-liquid
separation on digestate recirculated one-stage dry anaerobic digestion: use of intact
polar lipid analysis for microbes monitoring to enhance process evaluation. Renew.
Energy 103, 38–48.
Xu, F., Li, Y., 2012. Solid-state co-digestion of expired dog food and corn stover for
methane production. Bioresour. Technol. 118, 219–226.
Yenigun, O., Demirel, B., 2013. Ammonia inhibition in anaerobic digestion: a review.
Process Biochem. 48, 901–911.
Zamanzadeh, M., Hagen, L.H., Svensson, K., Linjordet, R., Horn, S.J., 2016. Anaerobic
digestion of food waste – effect of recirculation and temperature on performance and
microbiology. Water Res. 96, 246–254.
924
Journal of Environmental Management 223 (2018) 917–924
Zarkadas, I.S., Sofikiti, A.S., Voudrias, E.A., Pilidis, G.A., 2015. Thermophilic anaerobic
digestion of pasteurised food wastes and dairy cattle manure in batch and large vo-
lume laboratory digesters: focussing on mixing ratios. Renew. Energy 80, 432–440.
Zhang, Y., Banks, C.J., Heaven, S., 2012. Co-digestion of source segregated domestic food
waste to improve process stability. Bioresour. Technol. 114, 168–178.
Zhang, W., Lang, Q., Fang, M., Li, X., Bah, H., Dong, H., Dong, R., 2017. Combined effect
of crude fat content and initial substrate concentration on batch anaerobic digestion
characteristics of food waste. Bioresour. Technol. 232, 304–312.
Zhang, W., Wu, S., Guo, J., Zhou, J., Dong, R., 2015. Performance and kinetic evaluation
of semi-continuously fed anaerobic digesters treating food waste: role of trace ele-
ments. Bioresour. Technol. 178, 297–305.