Anaerobic Biogas Formation From Crops' Agricultural Residues - Modeling Investigations

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Bioresource Technology 359 (2022) 127497

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Anaerobic biogas formation from crops’ agricultural residues –


Modeling investigations
Marvin Scherzinger a, *, Martin Kaltschmitt a, Amr Y. Elbanhawy b
a
Hamburg University of Technology (TUHH), Institute of Environmental Technology and Energy Economics (IUE), Eissendorfer Strasse 40, 21073 Hamburg, Germany
b
Ain Shams University (ASU), Energy Technology and Climate Change Laboratory, Faculty of Engineering, Elsarayat Street 1, Cairo / Abbasia, Egypt

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Biogas production was modelled based


on kinetic Gompertz-Models.
• The model can be applied to other sub­
strates using relatively simple batch
studies.
• Indications regarding optimal operation
of small-scale biogas plants are supplied.
• A complete feed analysis of various
biogenic residues from Egypt is
presented.

A R T I C L E I N F O A B S T R A C T

Keywords: The cultivation of field crops necessarily produces large quantities of organic residues, which could be used for
Biogas the production of biogas. However, this is only successfully possible if the operators of such biogas plants have
Modeling sufficient expertise in running these plants, e.g. to avoid an overload of the biogas system. Against this back­
Co-digestion
ground, the anaerobic degradability of various Egyptian agricultural residues is determined by laboratory testing;
Gompertz-Models
Lignocellulose
this includes both a determination of degradation behavior in batch tests and a feed analysis. The residues
studied produce biogas yields ranging from 303 to 496 mLN g-1 VS. Co-digestion experiments demonstrate that
hardly any interaction effects occur during the fermentation process for different mixtures of the investigated
residues. Based on these findings, a model is developed to estimate biogas production using the investigated
agricultural residues in continuous operation and to give recommendations for the optimal mode of operation.

1. Introduction agricultural primary production of field crops. While in some places


large parts of such herbaceous residues are already being put to viable
Wherever agricultural activities take place, significant amounts of use (e.g. as construction material or as animal fodder), in many other
organic residues also accrue; among others, this is especially true for areas there is not yet a broader valorization established (Nour et al.,
straw being the most important organic (waste) side stream from 2021). For this reason, farmers in many regions of this world burn large

* Corresponding author.
E-mail address: marvin.scherzinger@tuhh.de (M. Scherzinger).

https://doi.org/10.1016/j.biortech.2022.127497
Received 17 May 2022; Received in revised form 13 June 2022; Accepted 15 June 2022
Available online 22 June 2022
0960-8524/© 2022 Elsevier Ltd. All rights reserved.
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

parts of the occurring straw in the field in order to get rid of it without cases where there are no interaction effects between different substrates,
bearing additional costs; this is especially true for regions where such biogas formation from any substrate mixture can be calculated as the
practices are not prevented and/or controlled by legislation (El-Dewany sum of biogas production from its individual components. This has been
et al., 2018). In this perspective, the most prominent example is the demonstrated for a biogas plant that was continuously fed with a sub­
burning of rice straw. According to current knowledge, the proportion of strate mixture of cow slurry, pig slurry, slaughterhouse waste and corn
rice straw burned in the fields varies from 0.6 to up to 90 % of the total silage (Zepter et al., 2021).
rice straw produced, depending on the country of production (Singh Against this background, in this paper the results of biogas formation
et al., 2021). Fortunately, more and more use options are being devel­ experiments of typical herbaceous residues from agriculture occurring
oped for the use of these agricultural residues, so that in many countries in Egypt are investigated. Based on this, co-fermentations of the inves­
(such as China, which is the main rice-growing country) field burning is tigated substrates are carried out in order to evaluate if interaction ef­
steadily declining. Current estimates indicate that in Egypt, for example, fects occur in the anaerobic digestion process. Based on these
about 25 % of the lignocellulosic residual streams are already being experiments and data from literature, a simple model based on the
used, with an upward trend (the main use is as livestock feed) (Nour degradation kinetics of various agricultural residues within an anaerobic
et al., 2021). However, economically promising and environmentally digestion process is developed and presented. This model is intended to
sound utilization possibilities must also be found for the remaining enable operators of biogas plants to easily calculate the daily input
quantities unused so far. In this way, negative effects, such as unwanted quantity for a freely selectable fermenter size as well as the resulting
air pollution and the release of climate-impacting gases associated with biogas quantity for various substrates or substrate mixtures based on
unsustainable processing strategies, can be reduced and possibly pre­ current knowledge of the framework conditions of anaerobic fermen­
vented in the long term. tation (e.g., taking into account restrictions regarding loading rate per
Anaerobic digestion of such residues could be one promising utili­ unit volume). Thus, this study aims to provide a tool with which the
zation option in this regard. During this process, various microorganisms kinetic models widely described in the literature can be usefully applied
decompose the organic material resulting in the release of a biogas as the to predict biogas formation by anaerobic fermentation in both existing
main product; this can then be used as fuel gas to substitute fossil fuels. and new biogas plants to be built. The developed model can easily be
The fermentation residues from biogas production could furthermore be extended to include other substrates for which data regarding degra­
used as valuable fertilizers to close nutrient cycles. However, such her­ dation kinetics are already available.
baceous residues from agricultural production have only suboptimal
degradability characteristics. Amongst others, these include an inap­ 2. Material and methods
propriate Carbon to Nitrogen ratio (C / N-ratio), a shortage of trace el­
ements, and a high proportion of substances that are recalcitrant to 2.1. Investigated substrates
degradation (Paul and Dutta, 2018). A very efficient and frequently used
method to compensate for these limitations is the co-digestion of various Different agricultural residues from different crops and/or gover­
substrates with different characteristics, whereby more favorable norates in Egypt are investigated regarding their potential as substrates
degradation conditions can be artificially adjusted (Siddique and Wahid, for biogas production via anaerobic digestion. The residues (namely rice
2018; Kasinath et al., 2021). straw (Oryza sativa), corn stalks (Zea mays), wheat straw (Triticum aes­
Another significant obstacle preventing the use of such agricultural tivum), sugarcane bagasse (Saccharum officinarum), grape pruning (Vitis
residues for biogas production is their predominantly seasonal occur­ vinifera) and orange pruning (Citrus × sinensis L)) were collected from
rence. Although this can be partially compensated by processes for farms in August 2021 and then sun-dried for a period of three days. In
preservation (e.g. storage of straw bales), it is often not possible to the case of rice straw and maize straw, one sample each from the gov­
guarantee a steady substrate mixture (or substrate quality) throughout ernorates of Behaira and Sharkia were used. Grape and orange pruning
the year, especially on smaller farms without large storage areas. originated from the district of Gharbia, wheat straw was collected in the
Furthermore, at such farms in rural areas of developing countries chal­ governorate Behaira and sugarcane bagasse came from the governorate
lenges due to a lack of technical knowledge occur. Especially the lack of Cairo. For further analyses, the dried samples were sent to Hamburg
knowledge about the size of biodigesters and the volume of organic University of Technology.
substrates often leads to low biogas production rates and digestate
pollution near the respective biogas plants (Sikora, 2021). 2.2. Determination of characteristic values
The development of models to predict biogas production volumes for
specific systems has been used in attempts to counteract these knowl­ 2.2.1. General
edge gaps and to provide valuable tools for plant operators to evaluate Prior to the investigations regarding the biogas formation potential,
the effects of substrate feeding. The models thereby differ greatly in all samples were analyzed with respect to various characteristic prop­
terms of their level of detail. A very detailed model is e.g. ADM1 erties. Both for this and for the subsequent investigations of the biogas
including multiple steps describing biochemical as well as physico- formation potential, a comminution of the samples was necessary. This
chemical processes (Batstone et al., 2002). The original ADM1 model comminution was performed using a laboratory cutting mill (MF10, IKA,
has been adapted several times, e.g. to be able to represent the co- Germany) and the target particle size was < 1 mm. In order to obtain
digestion of agricultural residues. However, the complexity of param­ subsamples that were as homogeneous as possible, the ground samples
eter determination and model calibration was also highlighted in this were further divided using a rotary cone sample divider (Laborette27,
context (Hagos et al., 2017). A much simpler method compared to the Fritsch, Germany).
above-mentioned modeling approach is the empirical determination of
kinetic models. Such models are typically based on enzymatic equations 2.2.2. Water content and volatile solids
or refer to the canonical algebraic form of equations with dimensionless Water content (W) and volatile solids (VS) were determined gravi­
coefficients (Pererva et al., 2020). A large number of studies reporting metrically according to DIN EN ISO 18134 (DIN, 2015) and DIN EN
on kinetic modeling of various substrates exist and a good overview on 15935 (DIN, 2012) in triplicates, respectively. The drying process for
several models based on biogas formation batch test data is given in the water content analysis was performed in a standard drying oven (U80,
literature (Velázquez-Martí and Meneses-Quelal, 2019). Currently, Memmert, Germany) at 105 ◦ C. For the determination of volatile solids,
almost all these kinetic models are derived from laboratory-scale ex­ a muffle furnace (M104, Thermo Scientific Heraeus, USA) was used. The
periments and further use to predict biogas production in continuously ashing took place at a temperature of 550 ◦ C. The respective values for
fed systems is not considered (Siddique and Wahid, 2018). However, in water content and volatile solids were calculated using Equation (1) and

2
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

Equation (2), respectively. Table 1


mw Substrate mixtures in the co-digestion experiments.
W= (1)
(md + mw ) Substrate fractions

( ) Substrate Sugarcane bagasse Grape pruning


ma Mixture 1 (Cairo) (Gharbia)
VS = 1 − (2) (SM1) a) 25 %VS 75 %VS
md
b) 50 %VS 50 %VS
mw is the total mass of analysis sample, md is the mass of analysis c) 75 %VS 25 %VS
sample after drying at 105 ◦ C and ma is the mass of sample after incin­ Substrate Sugarcane bagasse Rice straw
Mixture 2 (Cairo) (Sharkia)
eration at 550 ◦ C.
(SM2) a) 25 %VS 75 %VS
b) 50 %VS 50 %VS
2.2.3. Elemental analysis c) 75 %VS 25 %VS
An elemental analysis quantifying the main elements nitrogen (N), Substrate Corn stalks Wheat straw
carbon (C), hydrogen (H) and sulfur (S) was carried out using a NCHS Mixture 3 (Sharkia) (Behaira)
(SM3) a) 25 %VS 75 %VS
analyzer with internal conductivity detector (vario MACRO cube, Ele­ b) 50 %VS 50 %VS
mentar, Germany). The content of oxygen (O) was determined by the c) 75 %VS 25 %VS
difference method by subtracting the above mentioned elements as well Substrate Rice straw Wheat straw Corn stalks
as the ash content from 100 %. Mixture 4 (Behaira) (Behaira) (Behaira)
(SM4) a) 20 %VS 40 %VS 40 %VS
b) 40 %VS 20 %VS 40 %VS
2.2.4. Weender feed analysis extended by Van Soest method c) 40 %VS 40 %VS 20 %VS
The organic components of the various agricultural residues were
further investigated using the Weender feed analysis extended by Van
Soest method in duplicates (Van Soest, 1967). More precisely, a subdi­ was confirmed by a reference measurement using microcrystalline cel­
vision was made into the following fractions: cellulose, hemicellulose, lulose as substrate (more than 90 % of the theoretically producible
lignin, raw lipids, raw protein, and non-fiber-carbohydrates. The steps biogas (=745 mLN g-1 VS) was formed). In the first test series (i.e. mono-
for determining the individual parameters are briefly described below. digestion), 2.7 gVS substrate was added into each reactor, and only 2.5
gVS in the second series (i.e. co-digestion). This adjustment was neces­
• Cellulose, hemicellulose and lignin. The amounts of the cell wall sary due to minor changes in the inoculum composition; the requirement
components cellulose, hemicellulose and lignin were determined defined within the measurement guideline that the ratio of the organic
according to the consecutive Van Soest dissolution method described components of the substrate to the organic components of the inoculum
in the VDLUFA method manual volume 3 (VDLUFA, 2012). The should be <0.5 was fulfilled in that way. The duration of the studies
method includes three dissolution steps, after each of which the ranged from 35 to 45 days. During that time, the amount of biogas
sample is dried and weighed. In the end, the remainder is incinerated formed was measured by the use of eudiometer tubes connected to the
following the above-mentioned method for volatile solids determi­ sample bottles and filled with a sealing liquid consisting of sodium
nation. The proportions of the substances dissolved out or remaining chloride solution acidified with citric acid. All results were corrected for
in each case can then be determined by calculation (see e.g. biogas production from the inoculum without substrate addition. In
(Scherzinger et al., 2021)). addition, the biogas formed was collected in each case and analyzed for
• Raw lipids. The amount of raw lipids was determined based on the methane content using a portable gas analyzer (Biogas5000, Geotech,
DIN EN ISO 11085 (DIN, 2016). Briefly, around 2 g sample mass was United Kingdom) equipped with a dual-wavelength IR sensor.
subjected to extraction with n-hexane (boiling point 69 ◦ C) in a
Soxleth apparatus for 24 h. The amount of dissolved mass corre­ 2.4. Modeling
sponds to the raw lipid content.
• Raw protein. The raw protein content was calculated by multiplying 2.4.1. General
the amount of nitrogen by the factor 6.25 (a standard protein con­ The modeling approach chosen to simulate biogas production is
tains 16 % nitrogen (N); i.e. the factor 6.25 is always used when there shown in a simplified form in Fig. 1. In the following, the individual
are no other specifications). parts of this modeling concept are discussed and the most important
• Non-fiber-carbohydrates. The amount of non-fiber-carbohydrates equations contained in the model are presented. Wet fermentation is
was calculated by subtracting the above mentioned feed compo­ assumed, with a substrate addition once a day.
nents (i.e. cellulose, hemicellulose, lignin, raw lipids and raw pro­
tein) and the ash content from 100 %. 2.4.2. Choice of substrate mixture
The first part of modeling is related to the choice of a suitable sub­
2.3. Fermentation in batch experiments strate mixture. For this, the previously determined contents of water,
volatile solids as well as the contents of carbon and nitrogen are applied.
The batch tests to investigate the anaerobic degradability of the Equation (3) shows the calculation of the water content of the feed (WF).
agricultural residues were conducted in two runs. First, the biogas for­ Analogous to this procedure, the volatile solid content and the C/N-ratio
mation potential of the individual agricultural residues was investi­ of the substrate mixture is calculated from the sum of the individual
gated, followed by co-digestion studies for selected mixtures of the components multiplied by the respective proportion of the individual
individual substrates. Table 1 provides an overview of the substrate component. With the help of the dry matter content, however, a con­
mixtures used in the second investigation series. version was carried out in these cases in order to relate the results to
All batch experiments were performed in triplicate determinations fresh mass.
according to the VDI guideline 4630 at mesophilic conditions (i.e. 37 ±
WF = (WRM1 × SRM1 ) + ⋯ + (WRMn × SRMn ) (3)
1 ◦ C). As reactors, 500 mL glass bottles were used. As inoculum 400 ± 1
g of digested sludge from a municipal wastewater treatment plant South WF and WRM are the water content of the chosen substrate feed and
of Hamburg was added to each reactor; the sludge was allowed to outgas the raw material, respectively. SRM is the proportion of the respective
for one week prior to substrate addition and all reactors were purged raw material in the substrate mix.
with nitrogen after substrate addition. The suitability of the inoculum The C/N-ratio of the substrate mixture should be between 10 and 30

3
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

Fig. 1. Modeling approach to simulate biogas production.

for anaerobic degradation according to current knowledge (FNR, 2016). Quelal, 2019). To obtain the kinetic parameters of these models, ki­
A corresponding query was implemented in the model to check this. If netic model fits are initially performed. The averaged biogas production
the value is not in the optimal range, an adjustment of the substrate data from the batch experiments of the mono-substrates are used and a
mixture is recommended. Furthermore, a pumpable substrate mixture is nonlinear model fit based on the modified Gompertz equation (Equation
required for the present consideration (wet fermentation). In the model, (7)) is performed using the “curve_fit” function from the Python module
it is assumed that this occurs from a water content of 88 % or higher. If “SciPy”. The kinetic parameters for the substrates studied here are
the water content of the substrate mixture is below this threshold value, shown in the results (Table 3); P is the maximum biogas production
additional water to be added (Wadd) is calculated using Equation (4). potential in mLN g-1VS, Rm is the maximum rate of biogas production in
Parts of the water can be re-used resp. recycled after the fermentation mLN g-1 -1
VS d and λ is the lag phase in d. In addition, the root mean square
process by mixing it with fresh substrate. Therefore, the fermentation error RMSE was calculated to determine the quality of the models (the
residues are pressed; here it is assumed that pressing is possible up to a smaller the value, the more accurate the model). Since a large number of
water content of 50 %. such kinetic studies for different substrates are available in literature,
the model can be extended relatively easy for other substrate types. M(t)
0.88 − WF
Wadd = (4) is the biogas yield in mLN g-1VS with respect to time t in days and e is the
(1 − 0.88)
Euler constant (2.718).
{ ( )}
2.4.3. Operating parameters Rm e
M(t) = P × exp − exp (λ − t) + 1 (7)
The model is designed to freely select the fermenter volume (VF). P
This also applies to the hydraulic retention time (HRT) of the substrate
mixture used, which should, however, be at least 20 days to ensure 2.4.5. Calculation of the expected biogas / methane yield
stable operation. A density (ρ) of 1,000 kg m− 3 is assumed for the sub­ The model can be used to calculate the biogas / methane formation
strate mixture, which either naturally has a water content above 88 % or of a certain substrate mixture over time. Additionally, it is also possible
is brought to this water content by adding additional water. The daily to implement changes in the substrate mixture at certain points in time.
volume flow (V˙) is calculated via Equation (5). Subsequently, the Therefore, the daily biogas production is formed as the sum of the
loading rate per unit volume (BR) is calculated via Equation (6). The specific biogas production rates (M˙(t)) of the individual components of
model also checks whether loading rate per unit volume (BR) is in an the substrate mixture multiplied by their respective shares. The specific
appropriate range (between 1.5 and 3.5). If this is not the case, it pro­ biogas production rates of the individual components are calculated
vides a recommendation for action by adjusting the hydraulic residence using Equation (8) by taking the difference in biogas formation between
time. c is the concentration of organic matter in the substrate mixture. two subsequent time steps.

V̇ =
VF
(5) Ṁ(t) = M(t) − M(t − 1) (8)
HRT
A conversion of the daily biogas production is performed by multi­
V̇ × ρ × c plication with the measured methane contents of the individual com­
BR = (6)
VF ponents in the substrate mixture. This does not take into account that the
methane content of the formed biogas could change over time.
2.4.4. Determination of kinetic parameters
The core of the chosen modeling approach is the prediction of biogas 3. Results and discussion
formation from the selected substrate mixture. This is realized based on
kinetic models of the individual substrate components. More precisely, In the following, the results of the laboratory investigations of the
modified Gompertz models are used (Velázquez-Martí and Meneses- substrates are presented and discussed. Building on this, the results of

4
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

the developed model are then presented using a specific example. Table 2
Main elements (in g kg-1
DM) and C / N – ratio of the investigated agricultural
residues.
3.1. Characteristic values of the investigated agricultural residues C H N S O C / N-ratio

Rice straw (Behaira) 380 45 5.2 < 2.0 379 73


Due to sun drying of agricultural residues, the water content in all Rice straw (Sharkia) 390 46 6.2 < 2.0 352 63
samples ranged from 5.7 ± 0.1 % (for corn stalks from Sharkia and Corn stalks (Behaira) 470 55 6.2 < 2.0 429 76
Behaira) to 7.8 ± 0.2 % (for orange pruning from Gharbia). The water Corn stalks (Sharkia) 440 48 6.9 < 2.0 403 64
Wheat straw (Behaira) 430 49 6.2 2.6 403 69
contents were used to ensure conversion of the results of the analytic-
Sugarcane bagasse (Cairo) 470 58 3.0 < 2.0 456 157
moist samples to dry reference basis. The results of the Weender feed Grape pruning (Gharbia) 480 58 6.1 < 2.0 407 79
analysis extended by the Van Soest method are accordingly shown in Orange pruning (Gharbia) 460 55 6.9 < 2.0 410 67
Fig. 2 in terms of dry matter (the numerical values can be found in the
supplementary material). The total of cell wall components hemicellu­
lose, cellulose, and lignin accounted for the majority of plant biomass for Table 3
all samples studied. Rice straw and sugarcane bagasse samples had the Kinetic parameters of the determined Gompertz-models (RMSE root mean
lowest content of these substances with a total of 70.6 ± 0.3 % for square error) and mean CH4 content of the investigated agricultural residues.
sugarcane bagasse and 70.7 ± 1.4 % and 70.2 ± 0.8 % for rice straw
P Rm λ RMSE CH4 in Mass related
from Behaira and Sharkia, respectively. In the case of the rice straw biogas conversion
samples, this comparatively low total amount of cell wall constituents
mLN mLN d mLN % % of VS
was due to the relatively high ash content of 19.1 ± 0.7 % and 20.6 ± g-1 g-1 -
g-1
VS VS d VS
1.4 % for samples from Behaira and Sharkia, respectively. In the case of 1

sugarcane, a relatively large amount of cell ingredients was found; e.g. Rice straw 450.6 41.4 0.15 16.8 57.3 55.0
the amount of non-fiber-carbohydrates, which mostly are cell in­ (Behaira)
Rice straw 475.9 43.1 0.16 15.6 55.9 59.0
gredients such as simple sugars, starch or pectin, accounted for 25.6 %.
(Sharkia)
During the production of biogas by anaerobic fermentation, various Corn stalks 489.4 41.8 − 0.07 9.1 58.8 58.8
microorganisms can degrade a large part of the organic components of (Behaira)
the biomass. However, due to its complex structure, lignin is an excep­ Corn stalks 392.4 25.5 1.45 8.3 62.3 45.3
(Sharkia)
tion, which cannot be degraded within anaerobic processes. Further­
Wheat straw 470.6 39.1 − 0.39 15.7 58.1 56.9
more, lignin coats the degradable substances and therefore partly (Behaira)
prevents the microorganisms from degrading them. Thus, high lignin Sugarcane 467.7 29.5 − 2.72 21.9 55.8 58.0
contents of the lignocellulosic feedstock hinders an anaerobic fermen­ bagasse
tation. In this case, pretreatment is essential; in most cases, a purely (Cairo)
Grape 298.5 16.4 − 0.52 6.7 60.5 35.2
mechanical shredding of the biomass is realized increasing the surface
pruning
area and thus improving the accessibility of the degradable substances (Gharbia)
(Paul and Dutta, 2018). Here, highest lignin content was found for grape Orange 378.5 11.0 − 5.46 15.3 63.3 43.2
pruning with 19.9 ± 0.2 % and lowest lignin content was found for rice pruning
(Gharbia)
straw form Behaira with 1.5 ± 0.1 %. Typically, severely lignified ma­
terial such as orange pruning is not used for anaerobic fermentation.
However, due to the relatively high proportions of easily degradable Highest amount of carbon was found for grape pruning (480 g kg-1 DM)
substances (e.g. 15.2 % non-fiber-carbohydrates), this material was also while lowest amount of carbon was found for rice straw from Behaira
being further investigated for this purpose. (380 g kg-1
DM). This is well in line with expectations and can be explained
The fat content in all the samples tested was very low, with a by the previously discussed molecular composition. High carbon content
maximum of 2.1 ± 0.1 % for grape pruning. This is typical for such in grape pruning is related to the high amount of lignin, which usually
agricultural residues; in particular, high-fiber biomass such as straw shows the highest carbon content of the investigated molecular sub­
usually has only very low fat contents (Tufail et al., 2018). stances varying from 51 to 67 % (Christ et al., 2017). Lower amounts of
The results of the elemental analysis are presented in Table 2.

Fig. 2. Molecular composition of the investigated agricultural residues.

5
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

carbon in the rice straw samples can be explained by the high amount of stalks. Therefore, for the most accurate prediction of biogas formation, a
inorganics. laboratory investigation of the substrates to be used seems essential.
With regard to anaerobic biogas formation, however, not only the In contrast, both the molecular composition and the biogas forma­
carbon content is important, but also the C/N-ratio. This ratio should be tion potential of the two rice straw samples are very similar. Of the
between 10 and 30 (chapter 2.4.2). Too high C/N-ratios lead to an investigated agricultural residues, grape pruning has the lowest biogas
insufficient metabolism in which the available carbon cannot be formation potential (303 ± 7 mLN g-1 VS). This rather low biogas yield is
completely converted. In the opposite case, excess nitrogen can lead to justified by the strong lignification of this material (see chapter 3.1).
the excessive formation of ammonia (NH3) inhibiting the growth of This is also true for orange pruning, which has a biogas potential of 373
bacteria even in low concentrations; in the worst case, it could even lead ± 15 mlN g-1 VS.
to the complete collapse of the entire microorganism population (FNR, In addition to the maximum biogas potential, however, the biogas
2016). None of the agricultural residues studied had a C/N-ratio within production rate is also of decisive importance in the technical imple­
the optimum range. The lowest C/N-ratio of 64 was measured for corn mentation of an anaerobic fermentation plant. In fact, (too) slow biogas
stalks from Sharkia, while the highest C/N-ratio of 157 was determined formation rates are among the biggest challenges regarding the eco­
for sugarcane bagasse. This also means that no suitable substrate nomic operation of a biogas plant (Shen et al., 2015). Using the Gom­
mixture can be produced by mixing the residual materials investigated pertz functions derived from the measured data, the maximum biogas
here; thus, organic residues and waste streams with a clearly higher formation rates can be directly compared. Following this, the highest
nitrogen content, such as sewage sludge or animal manure, could be maximum biogas formation rate of 43.1 mLN g-1 -1
VS d was found for rice
used to compensate for the lack of nitrogen (Neshat et al., 2017). In the straw from Sharkia while the lowest maximum biogas formation rate of
anaerobic degradation studies conducted, this was implemented by 11.0 mLN g-1 -1
VS d was found for orange pruning (Table 3). Based on the
using sewage sludge as inoculum. biogas formation progressions shown in Fig. 1, no inhibitions of the
process were detected. However, both the two rice straws and corn
3.2. Fermentation of mono-substrates stalks from Sharkia exhibit lag phases; i.e., a delayed biogas production
after substrate addition can been observed. The reason for this behavior
The results of biogas production of the individual agricultural resi­ might be that such substrates contain only very few amounts of readily
dues over time and the corresponding Gompertz models are shown in degradable substances. In this case, hydrolytic microorganisms first
Fig. 3. The highest total yield of 496 ± 6 mLN g-1
VS was measured for corn decompose certain biomass components before the desired biogas for­
stalks from Behaira. Similar high yields were also obtained from both mation can begin. This is typical for a broad variety of lignocellulosic
rice straw samples, wheat straw, and corn stalks from Behaira. Signifi­ substrates, where the hydrolysis phase is usually the rate-determining
cantly less biogas (405 ± 2 mLN g-1 VS) was formed from corn stalks from step during anaerobic fermentation (Shrestha et al., 2017).
Sharkia. This can be explained by the molecular composition. Compared For the other substrates, biogas production starts with a phase of
to the cornstalk sample from Behaira, the sample from Sharkia have had linear or exponential growth and then decreases over time. An exception
lower contents of easily degradable non-fiber carbohydrates and hemi­ to this is the sugarcane bagasse studied; here, the biogas formation de­
cellulose. Additionally, the content of non-degradable lignin was creases after about 5 days, before it increases again in a second increase
somewhat higher. So, although it is the same plant and even the same phase after about 8 days. Such diauxic behavior is rather atypical for
part of the plant, there are significant differences in the case of corn lignocellulosic substrates, but occurs frequently for other substrates such

Fig. 3. Biogas formation of the investigated agricultural residues over time – Measured values and corresponding kinetic Gompertz-models.

6
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

as e.g. gelatin (Gomes et al., 2021). In the case of sugarcane bagasse, it affects the amount of digestate produced in continuous digestion. The
could be that all easily degradable components were metabolized within lower the conversion, the more (solid) biomass is present in the digestate
the first 5 days; the second phase of biogas formation then refers to the at the end.
conversion of the more hardly degradable cell wall components hemi­
cellulose and cellulose. Both before substrate addition and after
completion of the experiments, the pH value was measured. While the 3.3. Fermentation of co-substrates
pH value before substrate addition was about 7.0, it increased slightly in
all reactors (up to about 7.4), but thus remained always in the neutral Fig. 4 shows the results of the co-digestion experiments. In addition,
and optimal range for biogas formation. Inhibitions due to changes in these results are compared with the predicted values, which were
the pH value are therefore considered very unlikely in the investigations calculated on basis of the previously determined Gompertz equations
carried out. (data from Table 3). Here, it becomes obvious that biogas production
In addition to the course of biogas production, the methane content from the different substrate mixtures can be predicted accurately in most
in the biogas formed was also determined. For this purpose, the biogas cases using the Gompertz models derived from the monosubstrates. The
was collected in the eudiometer tubes until enough gas was available for difference between the last measured value after 34 days and the pre­
a measurement (about 300 mL of biogas was needed). Hence, the dicted value is usually <5 %; only for SM1 a) and SM3 c) larger de­
number of measurements carried out during the biogas tests depended viations of 10.8 % and 9.2 % were measured, respectively.
on the total amount of biogas formed. These data are then used to This means that the co-fermentation approach can achieve targeted
calculate the average contents of methane relative to the total biogas changes in both degradation rate and total biogas yield when the kinetic
volume; those values are shown in Table 3. Due to this procedure, no constants of the substrates to be fermented are known. Furthermore, a
exact statements can be made regarding the temporal course of methane controlled change of the lag phase is possible. This is evident, for
formation. However, it can be deduced from the individual measured example, from SM3, where a distinct lag phase was observed in the
values at different times that the methane content of the biogas formed fermentation of cornstalks from Sharkia. By co-fermenting with a sub­
increases over time. This can be explained by the fact that easily strate without such a lag phase (i.e. wheat straw), this lag phase could be
degradable components such as carbohydrates have a lower methane almost completely avoided. In this context, it has to be noted that the
potential (i.e. 415 mL CH4 g-1VS) than components that are more difficult possible changes can only take place within a certain range depending
to degrade, such as proteins or lipids (i.e. 496 and 1,014 mL CH4 g-1 VS, on the properties of the respective substrates.
respectively) (Xue et al., 2020). These findings are in good agreement with insights from other
Based on the amount of biogas formed and the proportions of CH4 studies. For example, it was demonstrated that the lag phase occurring
and CO2 and their density, respectively (it was assumed that biogas during mono-digestion of raw sludge could be significantly reduced by
consists only of these two gases; ρ of CH4 = 0.657 kg m− 3 and ρ of CO2 = the addition of lignocellulosic biomass. This was attributed to the faster
1.997 kg m− 3), the mass-based conversion was calculated. The highest degradation of carbohydrates from the lignocellulosic biomass
mass related conversion was obtained for rice straw from Sharkia with compared to the proteins mainly contained in the raw sludge (Zou et al.,
59 % of VS, while the lowest mass related conversion was achieved for 2018).
Grape pruning with only 35.2 % of VS (see Table 3). Mass conversion Moreover, in the present case, no significant interaction effects were
observed during the co-fermentation of the substrates. This is mainly

Fig. 4. Measured and predicted biogas formation of the investigated substrate mixtures (for definition of the substrate mixtures see Table 1).

7
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

due to the fermentation procedure according to the VDI guideline 4630, specific use case.
in which the fermentation takes place under the most optimal conditions
possible. A nutrient deficiency is excluded within this method by the 3.4.2. Choice of fermenter volume, hydraulic retention time and substrate
choice of inoculum and an overload of the fermentation system is mixture
avoided by the relatively small amount of substrate used (the actually In rural Egypt, biogas is mainly produced in so-called “Chinese fixed-
unfavorable C/N-ratios of the substrate mixtures, of which the numeri­ dome household digesters” consisting of a closed cylindrical chamber
cal values are listed in the supplementary material, can be compensated with an immovable gas space and both a digestate in- and outlet
by the sewage sludge (inoculum)). This also means that not all possible (Ioannou-Ttofa et al., 2021). The typical volume of such a simple
benefits of co-digestion can be captured by the studies conducted here. digester is up to about 6 m3. Therefore, a small fermenter volume was
Co-digestion can, among other things, increase the buffer capacity, also assumed for the example case described here (i.e. 5 m3). In general,
change the C/N-ratio and dissolve toxic substances, all of which can however, the fermenter volume can be freely selected in the model. The
have a positive effect on the biogas yield (Xu et al., 2018). Furthermore, case considered here is illustrated in simplified form in Fig. 5.
non-biogenic factors can also influence the process of biogas formation. In the case shown here, a substrate mixture consisting only of agri­
For example, the methane-forming microorganisms require certain trace cultural residues from the governorate Behaira was selected as initial
elements such as iron, cobalt and nickel. During continuous co- substrate mixture (i.e. corn stalks, rice straw and wheat straw, ratio
digestion, an optimal supply of these trace elements can be ensured by 1:1:1 based on fresh mass) and the hydraulic retention time (HRT) was
adjusting the substrate mixture (Li et al., 2014). Moreover, during set to be 40 d. In order to obtain a pumpable substrate mixture, 6.72 kg
continuous co-digestion, adaptations of microorganisms to the substrate of water must be added per kg of substrate under the assumptions made
mixture could also occur, which could improve biogas formation. in chapter 2.4. Of this, 5.87 kg of water can subsequently be recovered
Additionally, substrate-related inhibitions, e.g. due to the formation of by squeezing. A high water recovery rate is particularly important in
floating or sinking layers, could arise. Therefore, the findings obtained water-scarce regions, of which Egypt is one. For the operation of the
here should also be verified in continuously fed biogas reactors in order plant with the selected substrate mixture, this means that for each kg of
to obtain results independent from the used inoculum and under more fresh substrate mass, an additional 0.85 kg of water must be provided.
practice-oriented environments. The total amount of fresh mass to be fed daily is close to 16.2 kg.
Accordingly, an adequate supply of water (i.e. 13.8 kg d–1) is considered
very likely.
3.4. Modeling results The resulting loading rate per unit volume (BR) under these specifi­
cations is 2.66, which is in the optimal range. Although it has already
3.4.1. General been demonstrated that, in particular by co-fermenting different sub­
Despite the mentioned limitations due to the applied method, the strates, higher loading rates per unit volume (BR) can also be realized
collected data can be used in a modeling approach based on the iden­ without process disturbance, the relatively conservative range selected
tified kinetic constants. However, during modeling, care must be taken means that, for example, an accumulation of fatty acids can be largely
that the parameters are selected in such a way that the best possible ruled out (Hegde, 2019). The loading rate per unit volume is in direct
fermentation is ensured, as otherwise effects may occur that cannot be interaction with the hydraulic retention time; the model makes recom­
represented by the model. To investigate such effects (like e.g. acidifi­ mendations to lengthen the hydraulic retention time (HRT) if the
cation of the reactor), a more advanced model would be required. loading rate per unit volume (BR) is too high and to shorten it if it is too
In the following, the developed model is presented based on a

Fig. 5. Simplified representation of the assumptions and individual substrate mixture-dependent parameters in the presented modeling case.

8
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

low. Until the substrate change on day 25, the biogas production conse­
The C/N-ratio for the selected substrate mixture is 72.8, which is quently remains largely constant. Due to the slightly lower biogas for­
significantly higher than the optimal C/N-ratio. However, an optimal mation potential of the new substrate mixture and the changed course of
ratio cannot be achieved by mixing the agricultural residues investi­ degradation (e.g. lower maximum daily biogas formation), the biogas
gated here, since all investigated residues show a too low nitrogen yield then initially decreases before it slightly increases again over time
content. A warning is therefore displayed in the model. It could be that and settles at a level comparable to that immediately before the sub­
due to the unfavorable C/N-ratio, lower biogas yields are achieved in strate change.
practice than calculated in the model. But, nitrogen is concentrated by This developed model is intended to enable operators of biogas
the decomposition of carbon compounds within the digestate. This is plants to easily obtain common process data regarding the anaerobic
due to the fact that relatively more carbon compounds are converted, digestion of various agricultural residues. That this can be achieved via
while large proportions of nitrogen compounds pass into the liquid such an approach has already been demonstrated in another study
phase (Dong et al., 2020). Hence, by returning the process liquid to the (Zepter et al., 2021). In contrast to the aforementioned study, however,
fermenter, the C/N-ratio could be lowered through the recirculation of an adjustment of substrate properties (i.e., change of substrate mixture)
nitrogen, bringing it closer to the optimum range in theory. was included here. This is considered to be highly relevant for plant
Due to the varying accumulation of agricultural residues in the operators due to varying substrate availability throughout the year.
course of the year, the influence of a substrate change was also inves­ A major open issue is that the developed model has not yet been
tigated in the model. Therefore, it was assumed that the substrate validated on real measured data. It is therefore not clear to which extent
mixture described above would only be used in the initial phase (up to the model can provide reliable data. In the optimal range of anaerobic
day 25). digestion, it is highly probable that reliable data can be provided (this
From this point on, a substrate mixture of cornstalks, rice straw and was confirmed by the co-digestion experiments). However, under un­
sugarcane bagasse (ratio 1:1:3 based on fresh mass) is used. This also favorable conditions, this is rather not the case as the model cannot
changes the amount of water to be added to obtain a pumpable substrate include inhibitory effects. Also, no trace element deficiency was
mixture (i.e. 6.75 kg water need to be added per kg substrate mixture, of considered in the model. Due to these clear limitations, the model is only
which 5.89 kg can be recovered by squeezing). With this new substrate recommended for the applications mentioned above. Here, it is mainly
mixture, the loading rate per unit volume changes slightly to 2.84 and characterized by its simplicity, which should facilitate a broad use.
still lies in the optimum range. The total amount of fresh mass to be fed
daily is 16.1 kg from that day on. 4. Conclusion

3.4.3. Feeding management and biogas generation A user-oriented model that predicts the most optimal process pa­
The developed model is based on a daily one-time feeding of the rameters as well as the expected biogas production for small-scale biogas
biogas plant, with the same amount of substrate being fed each day. plants was developed. The model is both based on kinetic Gompertz
Accordingly, the biogas formation is always related to the substrate models and specific substrate properties. For the investigated substrates,
addition on a given day; this is illustrated in Fig. 6. Consequently, the it was demonstrated that under optimum conditions almost no interac­
total biogas production is then the sum of the biogas production of the tion effects occur during co-digestion. The deviation of predicted results
respective fed substrate mixture at the respective time. It also follows from obtained results from co-fermentation experiments was <5 % in 10
from this presumption that under steady state conditions the same out of 12 cases. Therefore, it can be assumed that the model provides a
amount of biogas (with the same methane content), is formed every day. good approximation of the actual biogas production also in a quasi-
A steady state is formed only after at least one hydraulic residence time continuous operation.
has elapsed, although in practice rather longer periods are needed (i.e. 3
to 5 residence times) for this (VDI, 2016). Accordingly, in the modeling
results shown in Fig. 6, a steady state is not achieved. Rather, a start-up CRediT authorship contribution statement
phase is shown from day 0, which then slowly approximates to a steady
state phase after about 20 days. The figure indicates also that the biogas Marvin Scherzinger: Conceptualization, Methodology, Software,
production of one specific substrate addition using the first substrate Investigation, Writing – original draft, Writing – review & editing,
mixture spans approx. over a time period of 30 days, whereby already Visualization. Martin Kaltschmitt: Supervision, Project administration,
after approx. 20 days only small biogas amounts are still formed. Funding acquisition. Amr Y. Elbanhawy: Resources, Supervision,
Project administration.

Fig. 6. Modeling results for the selected use case; the gray curves show the daily biogas production with respect to the individual feeds (left axis), the solid black line
shows the total daily biogas production (right axis) and the dotted black line is the time of substrate change.

9
M. Scherzinger et al. Bioresource Technology 359 (2022) 127497

Declaration of Competing Interest Ioannou-Ttofa, L., Foteinis, S., Seifelnasr Moustafa, A., Abdelsalam, E., Samer, M., Fatta-
Kassinos, D., 2021. Life cycle assessment of household biogas production in Egypt:
Influence of digester volume, biogas leakages, and digestate valorization as
The authors declare that they have no known competing financial biofertilizer. J. Cleaner Prod. 286, 125468.
interests or personal relationships that could have appeared to influence Li, J., Wei, L., Duan, Q., Hu, G., Zhang, G., 2014. Semi-continuous anaerobic co-digestion
the work reported in this paper. of dairy manure with three crop residues for biogas production. Bioresour. Technol.
156, 307–313.
Kasinath, A., Fudala-Ksiazek, S., Szopinska, M., Bylinski, H., Artichowicz, W.,
Data availability Remiszewska-Skwarek, A., Luczkiewicz, A., 2021. Biomass in biogas production:
Pretreatment and codigestion. Renew. Sustain. Energy Rev. 150, 111509.
Neshat, S.A., Mohammadi, M., Najafpour, G.D., Lahijani, P., 2017. Anaerobic co-
Data will be made available on request. digestion of animal manures and lignocellulosic residues as a potent approach for
sustainable biogas production. Renew. Sustain. Energy Rev. 79, 308–322.
Acknowledgements Nour, M., Amer, M., Elwardany, A., Attia, A., Li, X., Nada, S., 2021. Pyrolysis, kinetics,
and structural analyses of agricultural residues in Egypt: For future assessment of
their energy potential. Cleaner Engineering and Technology 2, 100080.
The results presented in this publication were obtained as part of the Paul, S., Dutta, A., 2018. Challenges and opportunities of lignocellulosic biomass for
“Co-Agrow+” project (grant number: 01DH21002), which was funded anaerobic digestion. Resour. Conserv. Recycl. 130, 164–174.
Pererva, Y., Miller, C.D., Sims, R.C., 2020. Existing empirical kinetic models in
by the German Federal Ministry of Education and Research. biochemical methane potential (BMP) testing, their selection and numerical
solution. Water 12, 1831.
Appendix A. Supplementary data Scherzinger, M., Kaltschmitt, M., Thoma, M., 2021. Effects of vapothermal pretreatment
on anaerobic degradability of common reed. Energy Technol. 9, 2001046.
Shen, Y., Linville, J.L., Urgun-Demirtas, M., Mintz, M.M., Snyder, S.W., 2015. An
Supplementary data to this article can be found online at https://doi. overview of biogas production and utilization at full-scale wastewater treatment
org/10.1016/j.biortech.2022.127497. plants (WWTPs) in the United States: Challenges and opportunities towards energy-
neutral WWTPs. Renew. Sustain. Energy Rev. 50, 346–362.
Shrestha, S., Fonoll, X., Khanal, S.K., Raskin, L., 2017. Biological strategies for enhanced
References hydrolysis of lignocellulosic biomass during anaerobic digestion: Current status and
future perspectives. Bioresour. Technol. 245, 1245–1257.
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., Siddique, M.N.I., Wahid, Z.A., 2018. Achievements and perspectives of anaerobic co-
Sanders, W.T.M., Siegrist, H., Vavilin, V.A., 2002. The IWA Anaerobic Digestion digestion: A review. J. Cleaner Prod. 194, 359–371.
Model No 1 (ADM1). Water Sci. Technol. 45 (10), 65–73. Sikora, A., 2021. Anaerobic Digestion in Built Environments.
Christ, D., Scherzinger, M., Neuling, U., Kaltschmitt, M., 2017. Thermochemical Singh, G., Gupta, M.K., Chaurasiya, S., Sharma, V.S., Pimenov, D.Y., 2021. Rice straw
conversion of solid biofuels: processes and techniques. In: Meyers, R.A. (Ed.), burning: a review on its global prevalence and the sustainable alternatives for its
Encyclopedia of Sustainability Science and Technology. Springer, New York, New effective mitigation. Environ. Sci. Pollut. Res. Int. 32125–32155.
York, NY, pp. 1–22. Tufail, T., Saeed, F., Imran, M., Arshad, M.U., Anjum, F.M., Afzaal, M., Bader Ul Ain, H.,
DIN, 2012. DIN EN 15935: Sludge, treated biowaste, soil and waste - Determination of Shahbaz, M., Gondal, T.A., Hussain, S., 2018. Biochemical characterization of wheat
loss on ignition. DIN Deutsches Institut für Normung e. V. straw cell wall with special reference to bioactive profile. International Journal of
DIN, 2015. DIN EN ISO 18134: Solid biofuels – Determination of moisture content. DIN Food Properties 21, 1303–1310.
Deutsches Institut für Normung e. V. Van Soest, P.J., 1967. Development of a comprehensive system of feed analyses and its
DIN, 2016. DIN EN ISO 11085: Cereals, cereal-based products and animal feeding stuffs – application to forages. J. Anim. Sci. 26, 119–128.
Determination of crude fat and total fat content by the Randall extraction method. VDI, 2016. VDI 4630 - Fermentation of organic materials: Characterisation of the
DIN Deutsches Institut für Normung e. V. substrate, sampling, collection of material data, fermentation tests, 132 pp.
Dong, D., Choi, O.K., Lee, K., Lee, J.W., 2020. Pilot-scale demonstration of nitrogen Velázquez-Martí, B., W. Meneses-Quelal, O., Gaibor-Chavez, J., Niño-Ruiz, Z., 2019.
recovery in the form of ammonium phosphate (AP) from anaerobic digestate. Review of mathematical models for the anaerobic digestion process. In: Rajesh Banu,
Bioresour. Technol. 297, 122392. J. (Ed.) Anaerobic Digestion. IntechOpen, pp. 1–20.
El-Dewany, C., Awad, F., Zaghloul, A.M., 2018. Utilization of rice straw as a low-cost VDLUFA, 2012. The chemical analysis of feedstuffs, 8th. VDLUFA-Verlag, Darmstadt.
natural by-product in agriculture. Int. J. of Environmental Pollution & Xu, R., Zhang, K., Liu, P., Khan, A., Xiong, J., Tian, F., Li, X., 2018. A critical review on
Environmental Modelling 1, 91–102. the interaction of substrate nutrient balance and microbial community structure and
FNR, 2016. Guide to biogas: from production to use. Fachagentur Nachwachsende function in anaerobic co-digestion. Bioresour. Technol. 247, 1119–1127.
Rohstoffe, Gülzow, p. 232. Xue, S., Wang, Y., Lyu, X., Zhao, N., Song, J., Wang, X., Yang, G., 2020. Interactive effects
Gomes, C.S., Strangfeld, M., Meyer, M., 2021. Diauxie studies in biogas production from of carbohydrate, lipid, protein composition and carbon/nitrogen ratio on biogas
gelatin and adaptation of the modified Gompertz model: Two-phase Gompertz production of different food wastes. Bioresour. Technol. 312, 123566.
model. Applied Sciences 11, 1067. Zepter, J.M., Engelhardt, J., Gabderakhmanova, T., Marinelli, M., 2021. Empirical
Hagos, K., Zong, J., Li, D., Liu, C., Lu, X., 2017. Anaerobic co-digestion process for biogas validation of a biogas plant simulation model and analysis of biogas upgrading
production: progress, challenges and perspectives. Renew. Sustain. Energy Rev. 76, potentials. Energies 14, 2424.
1485–1496. Zou, H., Chen, Y., Shi, J., Zhao, T., Yu, Q., Yu, S., Shi, D., Chai, H., Li, G.u., He, Q., Ai, H.,
Hegde, T., 2019. Anaerobic digestion of food waste with unconventional co-substrates 2018. Mesophilic anaerobic co-digestion of residual sludge with different
for stable biogas production at high organic loading rates. Sustainability 11, 3875. lignocellulosic wastes in the batch digester. Bioresour. Technol. 268, 371–381.

10

You might also like