You are on page 1of 10

Renewable Energy 163 (2021) 2198e2207

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Kinetic studies of alkaline-pretreated corn stover co-digested with


upset dairy manure under solid-state
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane, S. Rahman*
Department of Agricultural and Biosystems Engineering, North Dakota State University, 1221, Albrecht Boulevard, Fargo, USA

a r t i c l e i n f o a b s t r a c t

Article history: Prolonged lag phase and reactor instability make methane yield description and prediction challenging
Received 23 July 2020 in upset solid-state anaerobic digestion (SSAD). Seven kinetic models (first-order (FOM), modified first-
Received in revised form order (MFOM), Fitzhugh (FzM), logistic (LM), modified logistic (MLM), Cone (CM), and modified Gom-
29 September 2020
pertz (MGM) were employed for treatment evaluation. The treatments were dairy manure blended with
Accepted 24 October 2020
calcium pretreated stover (8% Ca CSDM) or sodium pretreated corn stover (4% Na CSDM) or 2% aqueous
Available online 28 October 2020
ammonia pretreated corn stover (2% NH4 CSDM), or 4% aqueous ammonia pretreated corn stover (4% NH4
CSDM). The LM, CM, MLM with MGM, and MGM best described methane yield for 4% Na CSDM, 2% NH4
Keywords:
Kinetic model
CSDM, 4% NH4 CSDM, and 8% Ca CSDM, respectively. Optimal hydraulic retention time predicted for the
Methane yield least inhibited treatment, 8% Ca CSDM, was 60 days. Therefore, non-hydrolysis rate limiting models (LM,
Solid-state anaerobic digestion MLM, FzM, MGM, and CM) could simulate methane yield, design of reactor’s volume, and develop
Retention time degradation rate models in full-scale SSAD.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction of approximately 31 dry Tg in the US, a potential methane yield of


1.26e3.89  106 Gg is possible with an assumed 0.5 ratio of biogas
Renewable energy has been successfully harnessed through to methane yield [5e10]. In addition to this, pretreatment is often
anaerobic digestion (AD) to mitigate organic waste volume. At employed on lignocellulosic biomass, particularly corn stover, to
present, solid state anaerobic digestion (SSAD) or solid state improve microbial accessibility to the fiber content. Of the present
anaerobic co-digestion (SSCoD) and liquid state anaerobic digestion pretreatment options in AD, alkaline (aqueous ammonia, calcium
(LSAD) or liquid state anaerobic co-digestion (LSCoD) are the two hydroxide, and sodium hydroxide) pretreatment is preferred in
common forms of AD. SSCoD is a good management practice order to maintain suitable pH for anaerobic microbes [11,12].
embedded with resource management compared to LSAD. The Therefore, this study focused on utilizing seven commonly used
SSCoD also supports high solid anaerobic digestion with total solids kinetic models to describe methane yield obtained from the alka-
(TS) > 15% and produces high volumetric methane relative to LSAD. line pretreated corn stover co-digested or blended with upset dairy
Nevertheless, the anaerobic digestion process, especially the manure under SSCoD. The kinetic models, such as first-order,
SSCoD, reactor performance, kinetic modeling, and the overall modified first-order, Fitzhugh, logical growth, modified logical
design of downstream processes of SSCoD in the initially upset growth, Cone, and modified Gompertz were explored in this
digester (VFA > 2.2 g/L [1], have sparsely been studied. Benefits of modeling study.
the kinetic modeling include identification of optimum process Among these models, the first-order kinetic model (FOM) has
variables combination [2], exploration of the properties of the been extensively used to predict methane yield in anaerobic
fermentation process [3], and the estimation of the performance digestion. This straight-line model is governed by a rate-limiting
and kinetic parameters [4]. step known as hydrolysis reaction [13], particularly for lignocellu-
With the yearly potentially available corn stover for biomass losic substrates. However, some non-biodegradable fraction of the
utilization of approximately 100e300 Tg along with dairy manure substrate limits the methane yield prediction with the FOM [14]. In
addition to this limitation, this model ignores both lag-phase and
specific methane production in its estimation [15].
The modified first-order kinetic model (MFOM) addresses the
* Corresponding author.
exclusion of the non-biodegradable fraction of the substrate in
E-mail address: s.rahman@ndsu.edu (S. Rahman).

https://doi.org/10.1016/j.renene.2020.10.110
0960-1481/© 2020 Elsevier Ltd. All rights reserved.
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

FOM. Hence, the predictability of MFOM is better than FOM, and


this model has found application in several studies [14,16]. How-
ever, the MFOM does not follow a sigmoid growth curve often
observed with cumulative methane production though it has a
rate-limiting hydrolysis step like FOM. The rate-limiting step in
hydrolysis of FOM is due to non-pretreatment of the lignocellulosic
substrate [17].
Unlike the FOM and MFOM, the rate-limiting step in logical and
modified logical growth models (MLM) describe acidogenesis (the
second digestion stage). Acidogenesis stage produces volatile fatty
acids and other intermediate products from the hydrolyzed poly-
mers (monomers). However, of these two models, only MLM has a
sigmoid microbial growth and the model fit resembles the con-
ventional shape of the cumulative biogas production kinetics curve
[4]. Besides this, the rate of biogas production in the model is
directly related to the volume of gas produced, peak biogas pro-
duction rate and the capacity [4]. Hence, the high performance of
the MLM (R2 > 0.99) in studies is not surprising [18].
The Gompertz model is another sigmoidal microbial growth
curve model, which is helpful to determine kinetic constants.
However, without modification, the model follows an L-shape
curve [17], limiting its use as a predictive model despite possible
high correlation coefficients [2]. Another challenge the unmodified
Gompertz model could pose is the complexity in distinguishing
kinetic constants for different treatments [2]. Hence, this study
considered the modified Gompertz model (MGM) to predict kinetic
constants. According to Ref. [19]; MGM is suitable for an inhibited
anaerobic digestion process, which has its methane yield propor-
tional to bacterial growth. The MGM has been adopted for most
kinetic studies in solid-state anaerobic digestion [4,20]. Similar to
MGM, the Fitzhugh model (FzM) has also been explored to fit
experimental methane yield in some liquid-state anaerobic diges-
tion studies of lignocellulose substrates [12,21]. Fig. 1. Flow chart illustrating the experimental procedures with kinetics modeling and
Cone model (CM) has also gained attention in most solid state model evaluation with statistical analysis.
anaerobic digestion (SSAD) studies more than FzM, particularly, as
it monitors methane yield and estimates kinetic constant such as
milled corn stover in this study with different alkaline wet-state
maximum cumulative methane yield (CMY) and specific methane
pretreatments (described subsequently). The milled samples were
production rate in substrates comprising of ruminal microbes
stored on air-tight bags under ambient conditions until used for
[15,20].
analyses. The moisture content of the samples was maintained at <
Given the wide applicability of the aforesaid mathematical
4.0% wb. Prior to wet-state pretreatment, the dry matter content,
models with either sigmoid and non-sigmoid curves, the models
volatile solids, total solids, volatile fatty acids concentration and
were utilized in this SSCoD kinetics study to gain a good under-
carbon and nitrogen contents of the milled corn stover were
standing of the kinetic constants and to make an informed decision
93.0 ± 0.2%, 94.1 ± 0.8%, 99.0 ± 0.8%, 0.0 ± 0.0%, 43.7 ± 0.1%, and
for downstream applications. The specific objectives of this study
0.7 ± 0.0%, respectively.
are to (1) describe and predict methane yield with the suitable
The dairy manure was also used as a co-substrate in the study,
kinetic models and (2) predict the optimal retention time for the
which was collected from the NDSU Dairy Research Farm, Fargo,
solid-state digestion study.
ND, USA and used for digestion on the same day of collection. The
dry matter content, volatile solids, total solids, volatile fatty acids
2. Materials and methods
concentration and carbon and nitrogen contents of the dairy
manure were 97.7 ± 0.6%, 86.6 ± 0.0%, 15.6 ± 0.0%, 8.9 ± 0.5 g/L,
2.1. Overall methodology
41.2 ± 0.9%, and 2.6 ± 0.2%, respectively. The inoculum mix used in a
mesophilic digester operated by City of Fargo wastewater treat-
The various processes involved in experimental and modeling
ment facility, Fargo, North Dakota, USA was used in this study. Dry
portions of this study are illustrated in the flowchart (Fig. 1).
matter, volatile solids, total solids, volatile fatty acids, carbon, and
Although details of each process were described subsequently, the
nitrogen contents of the inoculum were 96.7 ± 0.8%, 68.1 ± 0.0%,
flowchart provides an overview for ready reference.
1.5 ± 0.0%, 0.1 ± 0.0 g/L, 33.3 ± 0.4%, and 2.6. ± 0.0%, respectively.
2.1.1. Raw materials e corn stover and dairy manure
Corn stover sample sourced from the Carrington Research 2.1.2. Corn stover pretreatment
Centre, Carrington, North Dakota, USA, and was milled with a
Schuttle Buffalo hammer mill (Model W6H, New York, USA) to pass The corn stover was pretreated with four different alkaline so-
through a mesh sieve of 3 mm diameter (⌀) opening. Our previous lutions, namely 8% calcium hydroxide [Ca(OH)2], 4% sodium hy-
study showed that the corn stover of particle size of 0.42e0.84 mm, droxide [NaOH], 2% and 4% aqueous ammonia [NH4OH], under wet
codigested with dairy manure, produced the maximum methane state (Table 1). The concentrations considered in this study were
yield [22]. Therefore, we used the same particle size range for based on the suitability for methanogens [12,23,24]. A 100 g (db %)
2199
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

Table 1 2.1.6. Biogas volume measurement


Co-digested substrates mix ratios of the treatments by weight.

Treatment Dairy manure (%) Inoculum (%) Pretreated corn stover (%) Biogas volume from these digesters was measured in real-time
8% Ca CSDM 55.7 33.8 7.2
through a tipping bucket incorporated in a gas collector chamber.
4% Na CSDM 55.7 33.8 7.3 The methane concentration of the biogas was analyzed periodically
2% NH4 CSDM 40.0 33.8 8.3 with a gas chromatograph (8610C, SRI Instruments, California,
4% NH4 CSDM 46.9 33.8 7.9 USA). These data were used to evaluate daily methane yield, which
Note: Treatment 8% Ca CSDM represents corn stover pretreated with 8% concen- was further used to fit different kinetic models.
tration of Ca(OH)2, 4% Na CSDM represents corn stover pretreated with 4% of NaOH,
2% NH4 CSDM represents corn stover pretreated with 2% NH4OH, and 4% NH4 CSDM
represents corn stover pretreated with 4% NH4OH, and these four pretreated corn
stover was then blended with dairy manure. Required water was added to achieve a 2.1.7. Kinetic models
TS of 16%. Source [25].
Seven different kinetics models were explored in this study to
analyze CMY (Table 2). These models produced sigmoid curves for
of 0.42e0.84 mm particle size corn stover was pretreated with all the treatments, and they were symmetric; hence they did not
373 mL of respective of alkaline solution (2% NH4OH, 4% NH4OH, require an additional parameter to modify its shape [29]. To model
and 4% NaOH). On the contrary, due to the mild impact of Ca(OH)2 the methane yield obtained for all the treatment at the end of the
on stover, calcium pretreated corn stover was achieved with the experiments, non-linear curve fitting for the kinetic models was
blend of 509 mL of 8% Ca(OH)2 and 100 g of the stover. These performed. Software such as POLYMATH 6.10 with steepest descent
concentrations were considered suitable for methanogens approach using mrgmin () command, and in R with non-linear least
[12,23,24]. The pretreated corn stover samples were oven dried at squares by Levenberg-Marquardt approach using nlsLM() com-
40  C for 24 h to prevent mold formation and ensure a TS of 16% of mand were used for fitting the non-linear regression. However, due
the mix. Detailed properties of these stovers are reported in our to the shape of each model, linear and non-sigmoid models such as
previous study [25]). FOM, MFOM, Fitzhugh, and logical growth models could not be
modeled with the POLYMATH, while R could model them.
The models’ performances were evaluated and compared
2.1.3. Substrate mix preparation
among themselves using various metrics, such as coefficient of
determination (R2), root mean square error (RMSE), normalized
The pretreated corn stover along with dairy manure and inoc-
root mean square error (NRMSE), and Akaike’s information crite-
ulum mix were blended following the procedure described by
rion (AIC) [15,21]; Akaike, 1998 [31]; eqns (1)e(4)). The decision on
Ref. [26] to produce ingestates with TS of 16% and a carbon to ni-
the predictability performance of the models was based on the
trogen (C/N) ratio of 20 (Table 1). Differences in the mass of these
combination of high R2 value (>90%) coupled with low RMSE,
treatments as shown in Table 1 were due to impact of pretreatment
NRMSE, and AIC values.
reagent and the respective concentration on C/N ratio of the pre-
treated corn stover. P
ðYoi  Ypi Þ2
R2 ¼ 1  P i 2
(1)
i ðYoi  Ymn Þ
2.1.4. Anaerobic co-digestion process
Each treatment (Table 1), which contains the blend of dairy sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn 2
manure, inoculum, and the pretreated corn stover, weighing i¼1 ðYoi  Ypi Þ
RMSE ¼ (2)
approximately 1000 g was introduced into a 6 L digester, with 3.5 L nm
working volume, immersed in a water bath maintained at 35  C.
Daily biogas volume was measured from a gas collector chamber  
RMSE
connected to the digester. The gas chamber operates through an in- NRMSE ¼  100 (3)
ðYmax  Ymin Þ
built tipping bucket. To investigate the methane concentration in
the biogas, 5 mL of the biogas was collected from the headspace of  
RSS
the digester with a 20 mL syringe. This was diluted with 500 mL AIC ¼ nm ln þ 2v (4)
nitrogen gas and then introduced into a gas chromatograph (GC, nm
8610C, SRI California, USA). Methane concentration obtained from
where, Yoi is the measured data point for the methane yield (mL/g
this analysis was used to compute the methane yield for each
VS), Ypi is the predicted data point for the methane yield (mL/g VS),
treatment. Digesters in this study were dismantled when either the
Ymn is the mean of the measured data points for the methane yield
daily biogas volume was low (<50 mL) or after 100 days retention
(mL/g VS), nm is the number of the experimented data point for the
time (also referred to as the duration of digestion). Thus, treatment
methane yield, Ymax and Ymin are the maximum and minimum
8% Ca CSDM was dismantled after 79 days retention or digestion
experimental value for the methane yield, respectively, RSS repre-
time while other treatments (2% NH4OH, 4% NH4OH, and 4% NaOH)
sents the residual sum of squares, and v is the number of model
were dismantled after 100 days retention time.
parameters.

2.1.5. Chemical properties measurement


2.2. Statistical analysis
Chemical properties of each treatment such as pH, alkalinity,
volatile fatty acids (VFA), and total ammonia concentration were Data from this study were analyzed with SAS software (Version
estimated following the standard methods (International organi- 9.4, SAS Institute Inc., Cary, NC, USA). The pairwise comparisons of
zation of standardization 1997, [27,28]. Measuring these properties the treatment means were conducted using Duncan multiple range
provides insights on the suitability of each treatment in anaerobic tests (DMRT), preset at a p-value of 5%, to study the statistical
digestion. significance of the treatments.
2200
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

Table 2
Kinetic models used in this study for methane yield prediction.

Kinetic Model Mathematical expression Reference

First order (FOM) YðtÞ ¼ Ymx  ð1  ekt Þ [24]


Modified first order (MFOM) YðtÞ ¼ Ymx½ð1  YÞ  ð1  YÞ  ekt Þ [14]
Cone (CM) YðtÞ ¼ Ymx=ð 1 þ ðptÞn Þ [30]
 
Modified Gompertz (MGM) mj [30]
YðtÞ ¼ Ymx exp{  exp ðl  th Þ þ 1
Ymx
n
Fitzhugh (FzM) YðtÞ ¼ Ymx  ð1  eðktÞ Þ [12]
Logistic growth (LM) YðtÞ ¼ Ymx = ð1 þ zekt )
Modified logistic growth (MLM) YðtÞ ¼ Ymx = ð1 þ ekðt  lÞ ) [12]

Note: Y(t) represents the predicted cumulative methane yield (CMY) at the cessation of the experiment (mL/g VS); Ymx is the methane potential of the
ingestate (mL/g VS); k is the methane production rate constant or first order rate constant or hydrolysis constant (1/day); t is the digestion time in days; p is
the hydrolysis rate constant for the Cone model (1/day); Y is the non-biodegradable fraction of the treatment; n is the shape factor (dimensionless); m is the
maximum methane production rate (mL/g VS/d), l is the lag-phase time in hours; e Euler’s constant (2.71821828); z is the integrated constant related to the
initial methane yield; and th represents digestion time in hours.

3. Results and discussion followed similar trend with the pH (Table 3). The initial SVFA for the
treatments were greater than 2.3 g/L, this indicates operating an
3.1. Variation in chemical properties among treatments upset digester in this study [1]; Table 3). In addition, treatments 8%
Ca CSDM and 4% Na CSDM both had high initial SVFA (>5.0 g/L,
The chemical properties of each treatment varied with both Table 3), an indication of digester instability at the onset of the
pretreatment alkaline solution and concentration (Table 3). Among experiment which could lead to digester failure [32,33]. Hence, all
these treatments, pH of both NH4 CSDM treatments was between the treatments in this study was at least upset and had inhibition
7.2 and 7.4, a range suitable for methanogens. Reported suitable pH tendency due to potential VFA accumulation during digestion. This
range for methanogens in anaerobic digestion is within 6.5e8.0 unsuitable reactor state is linked to the high VFA concentration of
[26]. Interestingly, other two treatments in this study had higher the precursor (dairy manure, 8.9 ± 0.5 g/L). However, at the
values of pH than the range considered suitable for active meth- cessation of the experiment, all the SVFA concentration for the
anogenic activity (Table 3). Precisely, treatment 8% Ca CSDM and 4% treatments 4% Na CSDM, 2% NH4 CSDM, 4% NH4 CSDM, and 8% Ca
Na CSDM had a pH of 9.4 and 8.6 respectively, an indication that CSDM had reduced by 31%, 66%, 70%, and 80% respectively (data not
methanogens might be possibly subjected to initial stress in these shown). Similarly, the protein fraction of the VFA (PVFA) were
digesters. At the end of the experiment, methane yield for all the completely utilized by microbes for biogas production, except for
treatments were with the threshold suitable for methanogens treatment 4% Na CSDM with PVFA accumulation to about 14-fold
(7.0e8.0). increase (from 0.2 to 2.9 g/L). Consequently, treatment 4% Na
On ammonia nitrogen concentration, all the treatments in this CSDM was possibly unstable during anaerobic digestion, a phe-
study had reasonably close values of initial total ammonia nitrogen nomenon due to initial upset nature of the digester and principally
concentration except for 8% Ca CSDM treatment, which was NaOH impact at enhancing hydrolysis [20].
significantly higher (p < 0.05, Table 3) relative to other treatments. On alkalinity concentration, all the treatments except 4% Na
The reason for this trend is not clear. However, only the ammonia CSDM had suitable initial alkalinity concentration (1.0e8.5 g/L,
concentration in treatment 8% Ca CSDM met the range (1.5e2.8 g/L Table 3 [34], for methanogens at the start of the experiment. High
[20], considered suitable for SSAD. An indication that ammonia initial alkalinity concentration in treatment 4% Na CSDM (>9.0 g/L,
nitrogen will not pose threat to anaerobic microbes in this study. Table 3) probably lead to the disruption of microbial granules,
Interestingly, at the cessation of the experiment, all the treatments prolong reactor stability, and subsequent digester failure [35].
had ammonia nitrogen concentration within this suitable range for Hence, the NH4 CsDMs and 8% Ca CSDM treatments are more
AD (2.1e2.9 g/L). suitable for anaerobic process relative to treatment 4% Na CSDM.
Variation in the starch fraction of VFA (SVFA) in this study
3.1.1. C/N ratio, ORP, and EC regime
On carbon to nitrogen ratio (C/N ratio), as indicated in Table 3,
Table 3 initial C/N ratio for all the treatments was 20. This value is
Properties of the treated samples before digestion with C/N of 20 and 16% TS. considered appropriate in anaerobic digestion [36]. Aside C/N ratio,
Treatment 8% Ca CSDM 4% Na CSDM 2% NH4 CSDM 4% NH4 CSDM methanogens strive well in bioreactor with an oxidation redox
pH 9.4 ± 0.0 8.6 ± 0.0 7.2 ± 0.1 7.4 ± 0.1
potential (ORP) between e 200 and e 400 mV [37,38], which is in
Ammonia-N (g/L) 1.6 ± 0.0 0.9 ± 0.0 0.7 ± 0.0 0.9 ± 0.0 line with the ORP values reported in this study before digestion
Starch VFA (g/L) 7.4 ± 0.0 6.1 ± 0.0 3.0 ± 0.0 3.7 ± 0.0 (270 to e 406 mV, Table 3). On electrical conductivity (EC), the EC
Protein VFA (g/L) 0.2 ± 0.0 0.2 ± 0.0 0.3 ± 0.0 0.2 ± 0.0 values in this study prior to digestion was less than 0.5 uS (Table 3).
Alkalinity (g/L) 5.9 ± 0.4 12.2 ± 0.4 3.4 ± 0.5 3.5 ± 0.0
This is an indication of possibly low salt formation prior to digestion
ORP (mV) 270 ± 0.5 406 ± 0.0 329 ± 11 378 ± 5.7
EC (uS) 0.0 ± 0.0 0.1 ± 0.0 0.3 ± 0.0 0.3 ± 0.0 due to limited reaction between the alkaline and acids fraction of
VS/TS 4.6 ± 0.1 4.8 ± 0.0 6.8 ± 0.3 5.9 ± 0.0 the ingestates.
Note: Treatment 8% Ca CSDM represents corn stover pretreated with 8% concen-
tration of Ca(OH)2, 4% Na CSDM represents corn stover pretreated with 4% of NaOH, 3.2. Volatile solids and total solids ratio
2% NH4 CSDM represents corn stover pretreated with 2% NH4OH, and 4% NH4 CSDM
represents corn stover pretreated with 4% NH4OH, and these four pretreated corn Volatile solids to total solids ratio describe substrate biode-
stover was then blended with dairy manure, VFA represents volatile fatty acids, TS
represents the total solids and VS represents the volatile solids, C/N ratio represents
gradability. In this study, reagent source and respective concen-
carbon to nitrogen ratio, ORP represents oxidation redox potential, and EC repre- tration had impact on the treatment biodegradability. For instance,
sents electrical conductivity. Source for some of the data [25]). biodegradation (VS/TS) in treatments with aqueous ammonia (NH4
2201
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

CSDMs) was at least 18% higher than treatments 8% Ca CSDM and Interestingly, this established that hydrolysis was not a rate limiting
4% Na CSDM (p < 0.05, Table 3). A trend that suggests more step in this study. A similar conclusion was drawn on pretreated
bioconversion of the complex organic fraction of the NH4 CSDMs corn straw digestion with thermophilic microorganisms [40].
treatments during anaerobic digestion. On the contrary, R2 values for models LM, FzM, CM, and MGM
were within the high range of 0.98e0.99 (Table 4) considered
3.3. Daily methane yield suitable for model characterization and evaluation [12], except for
the CM obtained with POLYMATH software. The models fitted using
At the onset of microbial degradation in this study, daily R program produced high R2 and these models were further eval-
methane yield was more than 0.7 L/kg VS for the ammonia treat- uated with RMSE, NRMSE, and AIC values and then ranked from the
ments (2% NH4 CSDM and 4% NH4 CSDM) and then sharply declined highest (low values) to the lowest (high values) with respect to
possibly due to volatile fatty acids production and subsequent treatment. Hence, the best model has the least metric values
accumulation (Fig. 2). This trend suggests earlier favorable condi- assigned per each treatment.
tion for methanogens in the treatments prior to VFA accumulation.
On the contrary, daily methane yield in treatments 4% Na CSDM and 8% Ca CSDM: FzM > MGM > CM > LM > MLM
8% Ca CSDM progressively grew only after the lag phase, an indi- 4% Na CSDM: LM > MLM > CM > MGM > FzM
cation of digester instability linked with high initial VFA concen- 2% NH4 CSDM: CM > MGM > FzM > LM and MLM
tration (VFA > 4.0 g/L). Thus, methanogens might be able to survive 4% NH4 CSDM: FzM and MGM > CM > LM and MLM
in a stressed or upset digester with initial VFA less than 4.0 g/L but
not after this threshold, a trend similarly observed in literature [32]. The best model to fit the CMY data for 4% Na CSDM, 2% NH4
Interestingly, treatment 8% Ca CSDM had the earliest and highest CSDM, 4% NH4 CSDM, and 8% Ca CSDM was LM, CM, either FzM or
methane yield peak after the lag phase. Additionally, the treatment MGM, and FzM, respectively. Nevertheless, all the models were still
had the shortest retention time (79 days, Fig. 2), which suggests the used to fit the CMY data in order to obtain a suitable range, since
treatment had the most favorable process condition in this study, the RMSE, NRMSE, and AIC values for all the suitable models in this
this is irrespective of the initial stressed condition of the digesters. study were reasonably close (Table 4).
For the lag-phase time (l), which is one of the indicators
3.4. Kinetic models performance required for microbial adaptation to the substrate in this study was
between 20 and 77 d, a range dependent on kinetic model explored
The kinetic models (Table 2) were validated based on perfor- (Table 4). This range was significantly higher than the reported
mance measures such as R2, RMSE, NRMSE, and AIC (Table 4). The value (10e21 d, p < 0.05) on a solid-state mono-digestion study
R2, RMSE, NRMSE, and AIC values for the FOM and MFOM models [12]. This huge difference might be due to the presence of Kjeldahl
were <85%, RMSE >15, NRMSE >10, and >700, respectively. These rich dairy manure in the bioreactor, which was slowly hydrolyzed, a
were unlike the LM, FzM, CM, and MGM models whose range of R2, similar condition was also noted previously [41]. However, in both
RMSE, NRMSE, and AIC values were >90%, RMSE <6, NRMSE <4, cases, the lag-phase was >1 d, suggesting biodegradable fraction of
and <600, respectively (Table 4). This is an indication that FOM and the substrates were not readily consumed by anaerobic microbes
MFOM models did not produce a good fit to predict methane yield [17]. Furthermore, prolong lag-phase time (over 20 d) also suggests
and hence could not be used to describe methane production in the methane production could still be significant up to 55 d [33].
study. Specifically, FOM and MFOM models trend in this study Nonetheless, treatment 4% Na CSDM had the highest lag-phase
could be attributed to the poor linear relationship between time (at least 47e75 d, Table 4, p < 0.05) relative to other treat-
methane yield and hydraulic retention time as indicated by the low ments in this study. This was suspected to influence the choice of
R2 values (<0.85, Table 4), which were lower than the range MLM or LM models as the best model to fit the CMY data. The high
(0.90e0.98) considered suitable in SSAD studies [12,39]. lag-phase time generally infers slow startup [20], hence anaerobic
digestion process was exceptionally delayed in treatment with 4%
Na CSDM, which obviously affected the digestion efficiency. Rapid
hydrolysis of complex polymer in treatment 4% Na CSDM due to
NaOH presence could be the possible reason for this outcome.
On the contrary, treatment 8% Ca CSDM had the lowest lag-
phase time (20e37 d), which suggests acidogenesis, the rate
limiting step, was mitigated by the presence of calcium which
probably formed calcium stearate, a compound that prevents acid
accumulation in the bioreactor [37]. Hence, it was not surprising
that the CMY data for the 8% Ca CSDM treatment was best fitted
with the Fitzhugh model in this study. Noticeably, lag-phase time
for treatments 2% NH4 CSDM and 4% NH4 CSDM were relatively
close (Table 4).
Besides lag-phase time, the degradation rate of organic matter
(k), is another indicator often related to CMY data fitting. This was
denoted with k in LM, FzM, and MLM models while denoted with p
Fig. 2. Daily methane yield for the four treatments for retention time of up to 100 days. in the CM model. In this study, treatment 8% Ca CSDM had the
Note: Treatment 8% Ca CSDM represents corn stover pretreated with 8% concentration highest degradation rate for all the model with R2 > 90%, an indi-
of Ca(OH)2, 4% Na CSDM represents corn stover pretreated with 4% of NaOH, 2% NH4 cation of faster degradation rate or high digestion efficiency for all
CSDM represents corn stover pretreated with 2% NH4OH, and 4% NH4 CSDM represents
the models with R2 > 90%. This trend could possibly be attributed to
corn stover pretreated with 4% NH4OH, and these four pretreated corn stover was then
blended with dairy manure (Please use color for Fig. 2 in print). (For interpretation of the readily available organic matter in this treatment [42], obvi-
the references to color in this figure legend, the reader is referred to the Web version of ously ascertained by the short hydraulic retention time (79 d) and
this article.) high methane yield (179 mL/g VS) relative to other treatments.
Nevertheless, high k value might not correspond to high methane
2202
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

Table 4
Kinetic model parameters and their performance measures for treatments with pretreated corn stover blended with dairy manure.

Kinetic models Pretreatments Parameters Model performance measures

Ymx Y K z n p m l R2 RMSE NRMSE AIC

First order (FOM) 8% Ca CSDM 4503.4 0.0006 0.817 20.9 11.7 719
4% Na CSDM 5620.6 0.0001 0.12 16.6 17.3 860
2% NH4 CSDM 5483.6 0.0003 0.745 20.4 11.8 901
4% NH4 CSDM 5673.9 0.0002 0.387 19.2 16.6 890

Modified first order (MFOM) 8% Ca CSDM 10.1 952 0.0002 0.823 20.8 11.7 721
4% Na CSDM 21.5 630 5.068 0.111 16.6 17.3 862
2% NH4 CSDM 13.5 1113 0.0001 0.754 20.2 11.7 902
4% NH4 CSDM 17.9 859 6.922 0.396 19.1 16.5 891

Logistic (LM) 8% Ca CSDM 173.4 0.134 135 0.993 5.5 3.1 508
4% Na CSDM 108.8 0.081 441.1 0.999 1.2 1.3 333
2% NH4 CSDM 174.7 0.088 110.1 0.998 3.0 1.7 517
4% NH4 CSDM 112.6 0.117 978.1 0.993 3.8 3.2 562

Fitzhugh (FzM) 8% Ca CSDM 181.6 0.082 13.2 0.999 2.7 1.5 393
4% Na CSDM 167.6 0.030 10.1 0.997 1.7 1.8 405
2% NH4 CSDM 191.3 0.048 9.4 0.998 2.8 1.6 500
4% NH4 CSDM 117.9 0.075 58.1 0.995 3.2 2.8 531

Cone (CM) 8% Ca CSDM 184.2 4.3 0.027 0.998 3.0 1.7 410
4% Na CSDM 140.4 4.6 0.012 0.998 1.5 1.5 369
2% NH4 CSDM 194.7 3.9 0.018 0.999 2.6 1.5 487
4% NH4 CSDM 117.4 6.4 0.017 0.995 3.4 2.9 540

Cone * (CM*) 8% Ca CSDM 167.3 5.4 0.029 0.922 2.2 1.2 482
4% Na CSDM 86.8 7.6 0.014 0.906 1.0 1.0 469
2% NH4 CSDM 167.0 5.1 0.020 0.927 1.8 1.0 586
4% NH4 CSDM 106.3 8.0 0.018 0.934 1.1 0.9 498

Modified Gompertz (MGM) 8% Ca CSDM 66.5 0.1 32.0 0.998 2.8 1.6 400
4% Na CSDM 57.6 0.0 76.8 0.998 1.6 1.7 394
2% NH4 CSDM 69.5 0.1 47.5 0.999 2.6 1.5 488
4% NH4 CSDM 43.3 0.1 54.1 0.995 3.2 2.8 531

Modified Gompertz* (MGM*) 8% Ca CSDM 180.6 5.7 20.3 0.998 0.3 0.2 175
4% Na CSDM 156.5 2.0 47.5 0.997 0.2 0.2 111
2% NH4 CSDM 189 3.6 28.1 0.998 0.3 0.2 203
4% NH4 CSDM 117.8 3.3 40.9 0.995 0.3 0.3 247

Modified logistic (MLM) 8% Ca CSDM 173.4 0.135 36.4 0.993 5.5 3.1 509
4% Na CSDM 108.8 0.081 75.0 0.999 1.2 1.3 334
2% NH4 CSDM 174.7 0.088 53.5 0.998 3.0 1.7 517
4% NH4 CSDM 112.6 0.117 58.8 0.993 3.8 3.2 562

Modified logistic* (MLM*) 8% Ca CSDM 173.4 0.135 36.4 0.994 0.6 0.3 280
4% Na CSDM 108.8 0.081 75.0 0.999 0.1 0.1 215
2% NH4 CSDM 174.7 0.088 53.5 0.998 0.3 0.2 230
4% NH4 CSDM 112.6 0.117 58.8 0.993 0.4 0.3 276

Note: Ymx is the methane potential of the ingestate (mL/g VS); k is the methane production rate constant or first order rate constant or hydrolysis constant (1/day); p is the
hydrolysis rate constant for the Cone model (1/day); Y is the non-biodegradable fraction of the treatment; n is the shape factor (dimensionless); m is the maximum methane
production rate (mL/g VS/d), l is the lag-phase time in days; and z is the integrated constant related to the initial methane yield. * denotes data obtained with POLYMATH
software. Treatment 8% Ca CSDM represents corn stover pretreated with 8% concentration of Ca(OH)2, 4% Na CSDM represents corn stover pretreated with 4% concentration of
NaOH, 2% NH4 CSDM represents corn stover pretreated with 2% concentration of NH4OH, and 4% NH4 CSDM represents corn stover pretreated with 4% concentration of
NH4OH, and these four pretreated corn stover was then blended with dairy manure.

yield, but only suggests a high degradation rate [43,44]). As ex- models. Hence, the choice of this potential was streamlined to the
pected, treatment 4% Na CSDM had the least methane yield (95 mL/ best model previously noted in this study. Therefore, maximum
g VS) and the lowest degradation rate irrespective of the model methane potential for 4% Na CSDM, 2% NH4 CSDM, 4% NH4 CSDM,
(Table 4). This latter trend was expected, as NaOH pretreatment of and 8% Ca CSDM were 108.8, 194.7, 117.9, and 181.6 mL/g VS,
corn stover enhances the hydrolysis process [20]. Therefore, the respectively. The 2% NH4 CSDM produced the highest methane
poor degradation rate in treatment 4% Na CSDM could be attributed yield potential (194.7 mL/g VS, Table 4), however, the reason for this
to an imbalance among microbes in the digester. high methane yield potential in this treatment was not clear.
The maximum methane production rate (m) in MGM for all Another important indicator is the dimensionless shape factor
pretreated samples were generally small (<0.2 mL/g VS/d, Table 4). (n) in Cone model. This indicator suggests a sigmoidal methane
The low values are an indication of low digestion efficiency; how- yield curve when n > 1 [15], hence a lag-phase existed. As all the n
ever, these values were higher than the ones reported in a previous values in this study were greater than unity for all the treatments
mesophilic solid-state study [5]. (Table 4), which further established that FOM and MFOM models
On maximum methane potential, the values for this indicator were not suitable, since the methane yield for all the treatments
differ significantly for the same treatment under different kinetic followed the sigmoidal curve and not straight line.
models considered in this study, except for the LM and MLM

2203
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

3.5. Calculated and measured methane yield relationship 3.6. Predicted methane yield

The relationship between actual CMY and calculated CMY was The observed (40, 50, 60, 70, 80, 90, and 100 d) and predicted (0,
briefly examined for all the models except LM. All the treatments 40, 50, 60, 70, 80, 90, 100, 110, and 120 d) methane yield was pre-
modeled with MLM, MGM, and Cone models produced R2 > 0.99, dicted under different retention times (Tables 5 and 6). Most of the
except for 4% Na CSDM treatment in Cone model (Fig. 3). This R2 predicted values were very close with marginal variations irre-
range was considered a very good fit for predicted and measured spective of the kinetic models considered except for 4% NH4 CSDM
data [21]. This trend might suggest that MLM, MGM, and Cone at 40 days retention time. The obvious reason for this trend is the
models are suitable to describe process variables combination, long lag phase time in the 4% NH4 CSDM treatment, which extend
explore the properties of the SSCoD process, and equally estimate till 40 days. Precisely, at 40 days retention time, methane produc-
the performance and kinetic parameters in SSCoD. However, tion for 4% NH4 CSDM was at the lag phase relative to other
further validation was needed to establish this result. Coincidingly, treatments (Fig. 2). This might suggest that methane prediction is
MGM, and Cone models are often considered suitable for kinetic ineffective at the lag phase and ought to only be considered at
modeling in SSAD [20,30]). either the growth or stationary phase. Nonetheless, Therefore,

Fig. 3. Predicted and observed methane yield for the four treatments under this study. Shown models are (A) first order, (B) modified first order, (C) modified logistic model, (D)
Fitzhugh model, (E) Cone model, and (F) modified Gompertz. Note: Treatment 8% Ca CSDM represents corn stover pretreated with 8% Ca(OH)2, 4% Na CSDM represents corn stover
pretreated with 4% NaOH, 2% NH4 CSDM represents corn stover pretreated with 4% NH4OH, and 4% NH4 CSDM represents corn stover pretreated with 4%NH4OH, and these four
pretreated corn stover was then blended with dairy manure (Please use color for Fig. 3 in print). (For interpretation of the references to color in this figure legend, the reader is
referred to the Web version of this article.)

2204
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

Table 5
Predicted CMY within the experimental period (40e70 d) using selected kinetic models.

Kinetic model Pretreatment Predicted methane yield (mL/g VS) at different days and their deviation (%)

40 d 50 d 60 d 70 d

Experimental data 8% Ca CSDM 112 140 166 175


4% Na CSDM 5.15 15.1 23.9 43.3
2% NH4 CSDM 46.5 76.5 109 141
4% NH4 CSDM 5.11 27.0 63.7 86.8

Logical model 8% Ca CSDM 108 (3.70%) 150 (6.67%) 167 (0.60%) 172 (1.74%)
4% Na CSDM 5.99 (14.0%) 12.6 (19.8%) 24.8 (3.63%) 43.4 (0.23%)
2% NH4 CSDM 40.9 (13.7%) 74.1 (0.03%) 112 (2.68%) 142 (0.70%)
4% NH4 CSDM 11.2 (54.4%) 29.6 (8.78%) 60.2 (5.81%) 88.6 (2.03%)

Fitzhugh model 8% Ca CSDM 109 (2.75%) 145 (3.45%) 165 (0.61%) 174 (0.58%)
4% Na CSDM 4.40 (17.1%) 12.9 (17.1%) 26.7 (10.49%) 44.4 (2.48%)
2% NH4 CSDM 43.6 (6.65%) 78.9 (3.04%) 112 (2.68%) 138 (0.02%)
4% NH4 CSDM 6.24 (18.1%) 30.1 (10.3%) 62.2 (2.41%) 87.3 (0.57%)

Cone model 8% Ca CSDM 109 (2.75%) 146 (4.11%) 166 (0.00%) 174 (0.58%)
4% Na CSDM 4.84 (6.41%) 12.6 (19.8%) 26.0 (8.08%) 44.2 (2.04%)
2% NH4 CSDM 42.7 (8.90%) 77.9 (1.80%) 112 (2.68%) 138 (0.02%)
4% NH4 CSDM 8.79 (41.9%) 29.8 (9.40%) 61.5 (3.58%) 87.8 (1.14%)

Modified Gompertz model 8% Ca CSDM 109 (2.75%) 146 (4.11%) 165 (0.61%) 174 (0.58%)
4% Na CSDM 4.68 (10.0%) 12.9 (17.1%) 26.5 (9.81%) 44.3 (2.26%)
2% NH4 CSDM 43.2 (7.64%) 78.4 (2.42%) 112 (2.68%) 138 (0.02%)
4% NH4 CSDM 6.35 (19.53%) 30.0 (10.0%) 62.2 (2.41%) 87.3 (0.57%)

Modified logical model 8% Ca CSDM 108 (3.70%) 150 (6.67%) 167 (0.60%) 172 (1.74%)
4% Na CSDM 5.99 (14.0%) 12.6 (19.8%) 24.8 (3.63%) 43.4 (0.23%)
2% NH4 CSDM 40.9 (13.7%) 74.1 (0.03%) 112 (2.68%) 142 (0.70%)
4% NH4 CSDM 11.2 (54.4%) 29.6 (8.78%) 60.2 (5.81%) 88.6 (2.03%)

Table 6
Predicted CMY within and beyond the experimental period (80e120 d) using selected kinetic models.

Kinetic model Pretreatment Predicted methane yield (mL/g VS) at different days and their deviation (%)

80 d 90 d 100 d 110 d 120 d

Experimental data 8% Ca CSDM ND ND ND ND ND


4% Na CSDM 65.5 84.9 97.4 ND ND
2% NH4 CSDM 159 168 174 ND ND
4% NH4 CSDM 100 111 117 ND ND

Logical model 8% Ca CSDM 173 173 173 173 173


4% Na CSDM 65.2 (0.46%) 83.9 (1.19%) 96.2 (1.23%) 103 106
2% NH4 CSDM 159 (0.00%) 168 (0.00%) 172 (1.16%) 174 174
4% NH4 CSDM 104 (3.85%) 110 (0.91%) 112 (4.46%) 112 113

Fitzhugh model 8% Ca CSDM 178 180 181 181 182


4% Na CSDM 63.7 (2.83%) 82.5 (2.91%) 99.6 (2.21%) 114 126
2% NH4 CSDM 156 (1.92%) 169 (0.59%) 177 (1.69%) 182 186
4% NH4 CSDM 102 (1.96%) 110 (0.91%) 114 (2.63%) 116 117

Cone model 8% Ca CSDM 178 181 182 183 183


4% Na CSDM 64.3 (1.87%) 83 (2.29%) 98.4 (2.23%) 110 118
2% NH4 CSDM 157 (1.27%) 169 (0.59%) 177 (1.69%) 182 185
4% NH4 CSDM 103 (2.91%) 110 (0.91%) 114 (2.63%) 115 116

Modified Gompertz model 8% Ca CSDM 178 179 180 180 181


4% Na CSDM 63.8 (2.66%) 82.6 (2.79%) 99.4 (2.01%) 113 124
2% NH4 CSDM 157 (1.27%) 169 (0.59%) 177 (1.69%) 182 185
4% NH4 CSDM 102 (1.96%) 110 (0.91%) 114 (2.63%) 116 117

Modified logical model 8% Ca CSDM 173 173 173 173 173


4% Na CSDM 65.2 (0.46%) 83.9 (1.19%) 96.2 (1.23%) 103 106
2% NH4 CSDM 159 (0.00%) 168 (0.00%) 172 (1.16%) 174 174
4% NH4 CSDM 104 (3.85%) 110 (0.91%) 112 (4.46%) 112 113

Treatment 8% Ca CSDM represents corn stover pretreated with 8% Ca(OH)2, 4% Na CSDM represents corn stover pretreated with 4% NaOH, 2% NH4 CSDM represents corn stover
pretreated with 2% NH4OH, and 4% NH4 CSDM represents corn stover pretreated with 4%NH4OH, and these four pretreated corn stover was then blended with dairy manure.
d represents days. ND denoted no experimental data was available for the hydraulic retention time. All units in mL/g VS.

predicted methane yield was considered applicable under the 2% NH4 CSDM and 4% Na CSDM at < 60 d, a trend better than re-
retention time with standard error less than 10% in this study. For ported earlier [30]. A gradual process stability in treatments 2% NH4
instance, the difference among the observed values after 40 d and CSDM and 4% Na CSDM at < 60 d is the possible explanation for the
the respective predicted values were <10% for all treatments except high standard error values in these treatments. However, most of

2205
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

the prediction showed that extending the retention time, particu- appeared to influence the work reported in this paper.
larly for treatment 8% Ca CSDM after 60 days does not result in a
considerable increase in the methane yield (<10%, Tables 5 and 6). Acknowledgement
On the contrary, optimal retention time for NH4 CSDM treatments
was 80 d (Methane yields < 10%, Tables 5 and 6). A trend strongly The authors gratefully acknowledge the funding from the
linked with reactor inhibition (l > 27 d, 3.2 < m < 3.7 mL/g VS/d, Development Foundational Grant, North Dakota State University
Table 4). On 4% Na CSDM, increasing the retention time from 100 to (NDSU), USA and support from NDSU Graduate School and Dean of
120 d for treatment 4% Na CSDM led to a marginal increase in College of Agriculture, Food Systems, and Natural Resources, NDSU,
methane yield (<11%) using the best fit models (LM and MLM). USA.
Nevertheless, with other kinetic models (FzH, CM, and MGM), 4%
Na CSDM treatment showed a promising increase in CMY (20e27%,
References
Tables 5 and 6) by extending the retention time for an additional
20 d (from 100 to 120 d). This long optimal retention time in [1] Y. Li, J. Zhu, C. Wan, S.Y. Park, Solid-state anaerobic digestion of corn stover for
treatment 4% Na CSDM could be linked to the late recovery of the biogas production, Trans. ASABE (Am. Soc. Agric. Biol. Eng.) 54 (4) (2011)
reactor after possibly VFA accumulation (Fig. 2). Therefore, 1415e1421, https://doi.org/10.13031/2013.39010.
[2] V. Gadhamshetty, Y. Arudchelvam, N. Nirmalakhandan, D.C. Johnson,
extending retention time beyond these stated optimum values with Modeling dark fermentation for biohydrogen production: ADM1-based model
respect to treatments considered in this study is possibly a waste of vs. Gompertz model, Int. J. Hydrogen Energy 35 (2) (2010) 479e490, https://
resources such as heating energy. This information will be helpful doi.org/10.1016/j.ijhydene.2009.11.007.
[3] G. Yang, J. Wang, Improving mechanisms of biohydrogen production from
in decision making both at the downstream and for operational grass using zero-valent iron nanoparticles, Bioresour. Technol. 266 (2018)
purposes. 413e420, https://doi.org/10.1016/j.biortech.2018.07.004.
Hence, aside ascertaining findings from the literature that Cone [4] A. Donoso-Bravo, S.I. Pe rez-Elvira, F. Fdz-Polanco, Application of simplified
models for anaerobic biodegradability tests. Evaluation of pre-treatment
and modified Gompertz models are suitable models for in solid- processes, Chem. Eng. J. 160 (2) (2010) 607e614, https://doi.org/10.1016/
state anaerobic co-digestion studies [20,30], this study has j.cej.2010.03.082.
further suggested that partial pretreatment of substrates with the [5] Y. Li, R. Zhang, Y. He, X. Liu, C. Chen, G. Liu, Thermophilic solid-state anaerobic
digestion of alkaline-pretreated corn stover, Energy Fuels 28 (6) (2014b)
alkaline solution could influence kinetic model that best fit the 3759e3765, https://doi.org/10.1021/ef5005495.
CMY data. In addition, the high predictability and good fitness from [6] A. Prochnow, M. Heiermann, M. Plo € chl, B. Linke, C. Idler, T. Amon, P.J. Hobbs,
the logical and Fitzhugh models in this study equally indicate that Bioenergy from permanent grasslandeA review: 1. Biogas, Bioresour. Technol.
100 (21) (2009) 4931e4944, https://doi.org/10.1016/j.biortech.2009.05.070.
these models can be harnessed to describe and predict methane
[7] C.W. Murphy, A. Kendall, Life cycle inventory development for corn and stover
yield in solid-state anaerobic digestion. Further studies could be production systems under different allocation methods, Biomass Bioenergy 58
conducted to investigate downstream and operational applications (2013) 67e75, https://doi.org/10.1016/j.biombioe.2013.08.008.
with the logical and Fitzhugh models. [8] D.L. Karlen, S.J. Birrell, J.M. Johnson, S.L. Osborne, T.E. Schumacher, G.E. Varvel,
et al., Multilocation corn stover harvest effects on crop yields and nutrient
removal, BioEnergy Research 7 (2) (2014) 528e539. https://link.springer.com/
4. Conclusions article/10.1007/s12155-014-9419-7.
[9] H. Nam, A.L. Maglinao Jr., S.C. Capareda, D.A. Rodriguez-Alejandro, Enriched-
air fluidized bed gasification using bench and pilot scale reactors of dairy
Seven kinetic models were employed to describe and predict manure with sand bedding based on response surface methods, Energy 95
methane yield in this upset solid-state anaerobic digestion study. (2016) 187e199, https://doi.org/10.1016/j.energy.2015.11.065.
Non-hydrolysis rate limiting models such as Cone, logistic, Fitz- [10] Y. Wei, X. Li, L. Yu, D. Zou, H. Yuan, Mesophilic anaerobic co-digestion of cattle
manure and corn stover with biological and chemical pretreatment, Bio-
hugh, modified logistic, and modified Gompertz were ideal to resour. Technol. 198 (2015) 431e436, https://doi.org/10.1016/
describe and predict methane yield. Prolonged microbial adapta- j.biortech.2015.09.035.
tion and thus lengthy lag phase time (>19 d) were attributed to the [11] M. Hassan, W. Ding, J. Bi, E. Mehryar, Z.A.A. Talha, H. Huang, Methane
enhancement through oxidative cleavage and alkali solubilization pre-
initial high VFA of substrate in the reactor (>3.0 g/L) and subse-
treatments for corn stover with anaerobic activated sludge, Bioresour. Tech-
quent inhibition in the reactor. Lowest lag phase time (20 d) in nol. 200 (2016) 405e412, https://doi.org/10.1016/j.biortech.2015.09.115.
treatment 8% Ca CSDM suggests that acidogenesis, the rate limiting [12] H. Yuan, Y. Lan, J. Zhu, A.C. Wachemo, X. Li, L. Yu, Effect on anaerobic digestion
performance of corn stover by freezingethawing with ammonia pretreat-
step during digestion, was mitigated with possibly calcium stearate
ment, Chin. J. Chem. Eng. 27 (1) (2019) 200e207, https://doi.org/10.1016/
formation. Despite the low digestion efficiency (<0.2 mL/g VS/d) j.cjche.2018.04.021.
observed in this study, 60 d hydraulic retention time was predicted [13] C. Mao, J. Xi, Y. Feng, X. Wang, G. Ren, Biogas production and synergistic
optimal for the 8% Ca CSDM treatments. As for the remaining correlations of systematic parameters during batch anaerobic digestion of
corn straw, Renew. Energy 132 (2019) 1271e1279, https://doi.org/10.1016/
treatments (4% Na CSDM, 2% NH4 CSDM, and 4% NH4 CSDM), long j.renene.2018.09.009.
lag phase time resulting from reactor inhibition influenced the [14] Y. Li, R. Zhang, C. Chen, G. Liu, Y. He, X. Liu, Biogas production from co-
optimal retention time. The findings from this study could be used digestion of corn stover and chicken manure under anaerobic wet, hemi-
solid, and solid-state conditions, Bioresour. Technol. 149 (2013) 406e412,
to establish a significant relationship among processes, operations, https://doi.org/10.1016/j.biortech.2013.09.091.
and kinetic parameters in batch and full-scale solid-state anaerobic [15] D.R.S. Lima, O.F.H. Adarme, B.E.L. Bae ^ta, L.V.A. Gurgel, S.F. de Aquino, Influence
digestion. of different thermal pretreatments and inoculum selection on the bio-
methanation of sugarcane bagasse by solid-state anaerobic digestion: a ki-
netic analysis, Ind. Crop. Prod. 111 (2018) 684e693, https://doi.org/10.1016/
Author contribution j.indcrop.2017.11.048.
[16] V.A. Vavilin, B. Fernandez, J. Palatsi, X. Flotats, Hydrolysis kinetics in anaerobic
degradation of particulate organic material: an overview, Waste Manag. 28 (6)
Ademola Ajayi-Banji: Conceptualization, Methodology, Data (2008) 939e951, https://doi.org/10.1016/j.wasman.2007.03.028.
curation, Investigation, Writing - original draft. S. Sunoj Software, [17] J. Kainthola, M. Shariq, A.S. Kalamdhad, V.V. Goud, Enhanced methane po-
Validation. C. Igathinathane: Software, Validation. Writing - review tential of rice straw with microwave assisted pretreatment and its kinetic
analysis, J. Environ. Manag. 232 (2019) 188e196, https://doi.org/10.1016/
& editing, S. Rahman: Supervision. Writing - review & editing.
j.jenvman.2018.11.052.
[18] B. Deepanraj, V. Sivasubramanian, S. Jayaraj, Effect of substrate pretreatment
Declaration of competing interest on biogas production through anaerobic digestion of food waste, Int. J.
Hydrogen Energy 42 (42) (2017) 26522e26528, https://doi.org/10.1016/
j.ijhydene.2017.06.178.
The authors declare that they have no known competing [19] G. Zhen, X. Lu, T. Kobayashi, Y.Y. Li, K. Xu, Y. Zhao, Mesophilic anaerobic co-
financial interests or personal relationships that could have digestion of waste activated sludge and Egeria densa: performance

2206
A.A. Ajayi-Banji, S. Sunoj, C. Igathinathane et al. Renewable Energy 163 (2021) 2198e2207

assessment and kinetic analysis, Appl. Energy 148 (2015) 78e86, https:// digestion of microalgae residues resulting from the biodiesel production
doi.org/10.1016/j.apenergy.2015.03.038. process, Appl. Energy 88 (10) (2011) 3454e3463, https://doi.org/10.1016/
[20] J. Feng, Y. Li, E. Zhang, J. Zhang, W. Wang, Y. He, C. Chen, Solid-state co- j.apenergy.2010.10.020.
digestion of NaOH-pretreated corn straw and chicken manure under meso- [33] Y. Zhou, C. Li, I.A. Nges, J. Liu, The effects of pre-aeration and inoculation on
philic condition, Waste Biomass Valoriz 9 (6) (2018) 1027e1035. https://link. solid-state anaerobic digestion of rice straw, Bioresour. Technol. 224 (2017)
springer.com/article/10.1007/s12649-017-9834-z. 78e86, https://doi.org/10.1016/j.biortech.2016.11.104.
[21] H.M. El-Mashad, Kinetics of methane production from the codigestion of [34] Y. Li, Y. Li, D. Zhang, G. Li, J. Lu, S. Li, Solid state anaerobic co-digestion of
switchgrass and Spirulina platensis algae, Bioresour. Technol. 132 (2013) tomato residues with dairy manure and corn stover for biogas production,
305e312, https://doi.org/10.1016/j.biortech.2012.12.183. Bioresour. Technol. 217 (2016) 50e55, https://doi.org/10.1016/
[22] A.A. Ajayi-Banji, S. Rahman, S. Sunoj, C. Igathinathane, Impact of corn stover j.biortech.2016.01.111.
particle size and C/N ratio on reactor performance in solid-state anaerobic Co- [35] A.J. Ward, P.J. Hobbs, P.J. Holliman, D.L. Jones, Optimization of the anaerobic
digestion with dairy manure, Air Waste Manag. Ass. J. 70 (4) (2020a) digestion of agricultural resources, Bioresour. Technol. 99 (17) (2008)
436e454, https://doi.org/10.1080/10962247.2020.1729277. 7928e7940, https://doi.org/10.1016/j.biortech.2008.02.044.
[23] A.M. Mustafa, H. Li, A.A. Radwan, K. Sheng, X. Chen, Effect of hydrothermal [36] X. Wang, X. Lu, F. Li, G. Yang, Effects of temperature and carbonenitrogen
and Ca (OH)2 pretreatments on anaerobic digestion of sugarcane bagasse for (CeN) ratio on the performance of anaerobic co-digestion of dairy manure,
biogas production, Bioresour. Technol. 259 (2018) 54e60, https://doi.org/ chicken manure and rice straw focusing on ammonia inhibition, PloS One 9
10.1016/j.biortech.2018.03.028. (5) (2014) 1e7, https://doi.org/10.1371/journal.pone.0097265.
[24] J. Li, L. Wei, Q. Duan, G. Hu, G. Zhang, Semi-continuous anaerobic co-digestion [37] S. Chen, J. Zhang, X. Wang, Effects of alkalinity sources on the stability of
of dairy manure with three crop residues for biogas production, Bioresour. anaerobic digestion from food waste, Waste Manag. Res. 33 (11) (2015)
Technol. 156 (2014) 307e313, https://doi.org/10.1016/j.biortech.2014.01.064. 1033e1040, https://doi.org/10.1177/2F0734242X15602965.
[25] A.A. Ajayi-Banji, S. Rahman, L. Cihacek, N. Nahar, Comparison of the reactor [38] L. Wang, Q. Zhou, F.T. Li, Avoiding propionic acid accumulation in the
performance of alkaline-pretreated corn stover co-digested with dairy anaerobic process for biohydrogen production, Biomass Bioenergy 30 (2)
manure under solid-state, Waste Biomass Valoriz (2020b) 1e12. https://link. (2006) 177e182, https://doi.org/10.1016/j.biombioe.2005.11.010.
springer.com/article/10.1007/s12649-020-01116-z. [39] D. Brown, J. Shi, Y. Li, Comparison of solid-state to liquid anaerobic digestion
[26] K. Wang, J. Yin, D. Shen, N. Li, Anaerobic digestion of food waste for volatile of lignocellulosic feedstocks for biogas production, Bioresour. Technol. 124
fatty acids (VFAs) production with different types of inoculum: effect of pH, (2012) 379e386, https://doi.org/10.1016/j.biortech.2012.08.051.
Bioresour. Technol. 161 (2014) 395e401, https://doi.org/10.1016/ [40] S.F. Fu, F. Wang, X.Z. Yuan, Z.M. Yang, S.J. Luo, C.S. Wang, R.B. Guo, The
j.biortech.2014.03.088. thermophilic (55 C) microaerobic pretreatment of corn straw for anaerobic
[27] APHA, Standard Methods for the Examination of Water and Wastewater, digestion, Bioresour. Technol. 175 (2015) 203e208, https://doi.org/10.1016/
eighteenth ed., American Public Health Association, Washington, DC, 1992, j.biortech.2014.10.072.
2320B. [41] E. Elbeshbishy, G. Nakhla, Batch anaerobic co-digestion of proteins and car-
[28] B.R. Baumgardt, Practical Observations on the Quantitative Analysis of Free bohydrates, Bioresour. Technol. 116 (2012) 170e178, https://doi.org/10.1016/
Volatile Fatty Acids (VFA) in Aqueous Solutions by Gas-Liquid Chromatog- j.biortech.2012.04.052.
raphy, Department of Dairy Science, University of Wisconsin, 1964. [42] P. Parameswaran, B.E. Rittmann, Feasibility of anaerobic co-digestion of pig
[29] M. Peleg, M.G. Corradini, Microbial growth curves: what the models tell us waste and paper sludge, Bioresour. Technol. 124 (2012) 163e168, https://
and what they cannot, Crit. Rev. Food Sci. Nutr. 51 (10) (2011) 917e945, doi.org/10.1016/j.biortech.2012.07.116.
https://doi.org/10.1080/10408398.2011.570463. [43] K.K. Prajapati, N. Pareek, V. Vivekanand, Pretreatment and multi-feed anaer-
[30] I. Syaichurrozi, Biogas production from co-digestion Salvinia molesta and rice obic co-digestion of agro-industrial residual biomass for improved bio-
straw and kinetics, Renew. Energy 115 (2018) 76e86, https://doi.org/10.1016/ methanation and kinetic analysis, Frontiers in Energy Research 6 (111) (2018)
j.renene.2017.08.023. 1e18, https://doi.org/10.3389/fenrg.2018.00111.
[31] H. Motulsky, A. Christopoulos, Fitting Models to Biological Data Using Linear [44] C. Mao, T. Zhang, X. Wang, Y. Feng, G. Ren, G. Yang, Process performance and
and Nonlinear Regression: a Practical Guide to Curve Fitting, Oxford Univer- methane production optimizing of anaerobic co-digestion of swine manure
sity Press, 2004. and corn straw, Sci. Rep. 7 (1) (2017) 9379. https://www.nature.com/articles/
[32] E.A. Ehimen, Z.F. Sun, C.G. Carrington, E.J. Birch, J.J. Eaton-Rye, Anaerobic s41598-017-09977-6.

2207

You might also like