Study of the dark fermentative hydrogen production using modified ADM1 models

Accepted Manuscript Title: Study of the dark fermentative hydrogen production using modified ADM1 models Authors: Zineb Guellout, Valentin Clion, Yacine Benguerba, Christine Dumas, Barbara Ernst PII: DOI: Reference:

S1369-703X(17)30357-1 https://doi.org/10.1016/j.bej.2017.12.015 BEJ 6847

To appear in:

Biochemical Engineering Journal

Received date: Revised date: Accepted date:

2-8-2017 27-11-2017 23-12-2017

Please cite this article as: Zineb Guellout, Valentin Clion, Yacine Benguerba, Christine Dumas, Barbara Ernst, Study of the dark fermentative hydrogen production using modified ADM1 models, Biochemical Engineering Journal https://doi.org/10.1016/j.bej.2017.12.015 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Study of the dark fermentative hydrogen production using modified ADM1 models Zineb GUELLOUTa,b,c, Valentin CLIONc,d, Yacine BENGUERBAa,b*, Christine DUMASc, Barbara ERNSTc a b

Department of Processes Engineering, Université Ferhat Abbas, Sétif-1, 19000 Sétif, Algeria. Laboratoire de Génie des Procédés Chimiques, Université Ferhat Abbas, 19000 Sétif, Algérie. c Université de Strasbourg, CNRS, IPHC UMR 7178, Laboratoire de Reconnaissance et Procédés de Séparation Moléculaire (RePSeM), ECPM 25 rue Becquerel F-67000 Strasbourg, France d French Environment and Energy Management Agency, Angers, France

E-mail : [email protected], Mob: +213-661704929.

SC R

Highlights

IP T

*

Modeling of the fermentative production of hydrogen using a modified ADM1 model.



The Modified Model with Aiba Kinetic using Correction Factor (MMAK-CF) was found to be

U



A

High substrate concentrations inhibit bacterial growth.

M



N

the best model.

ED

Abstract

This paper presents a numerical study of the fermentation pathway of a bacterial consortium during production of hydrogen from glucose using three kinetic models. Mathematical

PT

expressions were used to describe glucose consumption, microbial growth and the production of hydrogen and metabolites. The numerical results show a good agreement with the

CC E

experimental data. Microbial growth was described using three different kinetic models. A correction factor (CF) was added to the "Anaerobic Digestion Model number 1" (ADM1) using the Modified Model Aiba Kinetic model (MMAK-CF), allowing to improve the model

A

agreement. Different initial concentrations of substrate were used to study their effect on hydrogen production and bacterial growth. The results show that the cumulative produced hydrogen increases with increasing the substrate concentration to reach a maximum. An optimal value was calculated at the initial substrate concentration of 0.022 mol L-1. Bacterial growth followed the same trend. It was concluded that high substrate concentrations inhibit bacterial growth and hydrogen production, which can distort the metabolism of microorganisms. 1

Keywords: Anaerobic digestion (AD); Kinetics; ADM1; Dark fermentation; Hydrogen;

A

CC E

PT

ED

M

A

N

U

SC R

IP T

Batch bioreactor

2

Nomenclature µ𝒎𝒂𝒙

Maximum specific growth rate (h-1)

𝑭𝑷/𝑺

Products yield coefficient of metabolites (acetate, butyrate, ethanol)

𝑭𝒈/𝑺

Gas yield coefficient (hydrogen and carbon dioxide)

𝑲𝑰

Inhibition constant; corresponding to the substrate concentration, growth rate due to substrate inhibition (mol.L-1) Michaelis constant substrate concentration (mol.L-1)

𝑲𝑺

Substrate constant (mol.L-1)

𝑲𝑺𝑴

Monod-constant (mol.L-1)

𝑲𝑺𝑪

Contois-constant (mol.mol COD-1)

𝑲𝒅

Death rate (h-1)

𝑣

U

Specific growth rate proposed by Michaelis (h-1)

𝑽𝒎𝒂𝒙

N

Maximum rate achieved by the system, at maximum substrate concentrations (h-1) Volume of liquid (L)

𝑆𝑣𝑎𝑟

Substrate, products and biomass concentration (mol.L-1)

𝑄𝑙𝑖𝑞

Output flowrate of liquid (L.h-1)

M

ED

Substrate, products and biomass initial concentration (mol.L-1)

𝑅𝑣𝑎𝑟 µ

CC E

CF

Reaction term (mol.L-1.h-1) Yield coefficient

PT

𝒀𝟏

A

𝑉𝑙𝑖𝑞

𝑆𝑣𝑎𝑟𝑖𝑛

ADM-1 IpH

MMAK-CF

A

SC R

𝑲𝑴

IP T

where bacteria growth is reduced to 50% of the maximum specific

Specific growth rate (h-1) Correction factor Anaerobic digestion model number 1 Inhibitor expression of pH Modified model with Aiba kinetic and with correction factor

MMCK

Modified model with Contois kinetic

MMMK

Modified model with Monod kinetic

P

Product produced (mol)

S

Substrate concentration (mol. L-1)

X

Biomass concentration (mol. L-1)

3

I.

Introduction

Considering the energy security and the global environmental crisis, there is a pressing need to develop non-polluting and renewable energy sources [1]. Hydrogen is considered as a clean and renewable energy carrier, with high energy content (142 kJ/g) [2]. It is an emission-free

IP T

alternative fuel (which does not contribute to the greenhouse effect) that could be produced from diverse renewable sources [3]. Indeed, its transformation into electrical energy in fuel

cells generates only water; in the same way, its transformation into thermal energy by

SC R

combustion produces only easily condensable water vapor. The currently developed technologies for the conversion of biomass into hydrogen can be declined into two categories, thermochemical processes: gasification, pyrolysis, hydrothermal liquefaction and biochemical

U

processes: fermentation, esterification, digestion. These latter, although energy-efficient

N

compared to thermo-chemical processes, produce ethanol, biodiesel and biogas which will have to undergo a reforming treatment, requiring a catalytic thermal step to produce

A

hydrogen. Thus, only the biochemical processes using the photolysis of water using algae or

M

cyanobacteria, photo-fermentation and dark fermentation of organic compounds with bacteria, acting as biocatalysts, allow the direct production of hydrogen at low temperatures (37 °C).

ED

From an economical point of view, the estimated cost of hydrogen produced by electrolysis 4.36-7.36 $/kg is the most expensive compared to other processes which is 2 times the cost of biological process; Hydrogen production from natural gas (steam reforming) is the most

PT

developed process at this moment, and also the less expensive 1.48-2.27 $/kg, but this process has a negative impact on the environment due to the CO2 emission. The cost of energy was

CC E

calculated for various hydrogen generation processes, namely biological or other hydrogen generation methods as compared to conventional fuels. Hydrogen produced from coal gasification and biomass pyrolysis or gasification has an estimated price range of 0.36-1.83

A

$/kg and 1.44-2.83 $/kg, respectively [4-9], making these processes competitive. Although the conversion efficiency of the biological processes was found to be lower than that of conventional fuels [4]. Dark fermentation process, consisting of anaerobic digestion of complex organic substrate in the absence of light, has the main benefit of contributing to waste recycling. Compared to photosynthetic methods [10], dark fermentation is the simplest and the most efficient method, as it usually shows high hydrogen production rates. The cost of hydrogen production by dark 4

fermentation was very recently estimated at 2.57 $/kg by Nikolaidis et al. [9]. In addition, this process can use a diverse range of organic substrates as stillage, sludge, leachate, pomace, stalks, bagasse, etc. [11]. Several studies have recommended using agro-industrial residues from different matrices as substrate in this process [12-16]. The maximum hydrogen yield of hydrogen reported in literature is 4 moles of H2 per mole of hexose [17] if all the substrate is converted to acetate. These values correspond to a theoretical maximum yield of 0.492 L H2/g COD. If all the substrate is converted to butyric

IP T

acid, this value is 2 mol H2/mol hexose. The hydrogen yields of pure cultures have been

reported to range from 0.1 to 3.8 mol H2/mol hexose [18]. Several researchers have begun to

SC R

develop various approaches to increase the yield of hydrogen production [19].

The hydrogen production by dark fermentation depends on fermentation conditions (substrate type, temperature, pH, etc.), which may be inhibitors or accelerators of metabolism. The microorganisms found in mixed cultures, such as sewage sludge, are more beneficial than

U

pure cultures because they are more adaptive to environmental conditions, including limited

N

substrates, pH changes and temperature variation [20]. Clostridia are generally used in dark fermentation because of their metabolism. Clostridium is a strict anaerobic heterotroph genus

A

[21]. These bacteria produce hydrogen from ferredoxin with hydrogenase enzymes activities;

M

activated in a medium containing glucose [22].

From the literature, it can be concluded that the use of glucose as substrate gives the highest

ED

performance [23, 24]. The pH is one of the most important factors in dark fermentation, being under typical anaerobic conditions, the only one that directly affects the hydrogenase activity

PT

[1]. Fang and Liu [25] investigated the effect of pH between pH 4.0 and 7.0 and concluded that the optimal pH was 5.5 with a yield of 2.03 mol H2/mol hexose and a specific production rate of 1.35 10-3 mol H2/(mol hexose.h). For degradation of simple substrates, the optimum

CC E

initial pH for Clostridia is between 4.5 and 7.0; a high yield of hydrogen (1.6 mol H2/mol hexose) has been also observed at a pH value of 9.0 by Zhao et al. [26-29]. Different studies proposed and developed models to describe a batch hydrogen production

A

process in order to see the effects of substrate concentration, inhibitor concentration, pH, temperature, and to establish the relationship among the substrate degradation rate, the bacterial growth rate and the product formation rate [30-33]. One of these models is the modified Gompertz equation developed by Zwietering et al. [34] to describe the evolution of hydrogen production in a batch reactor. However, this equation cannot describe the concentrations of substrate and products co-produced or competing with hydrogen. The ADM1 model was published in 2002 by Batstone et al. [35] to describe the process of 5

anaerobic digestion. Some changes have been done on the ADM1 model to describe nonmethanogenic systems, using constant stoichiometry to describe product generation from carbohydrate fermentation. It allows to describe the effects of substrate concentration on substrate consumption rates, hydrogen-producing bacteria growth, and hydrogen production [36, 37]. Recently, several studies have described the inhibition caused by high substrate concentration in dark fermentation. Kan [38] reached 0.9 mol H2/mol glucose, after 72 hours of batch fermentation using as inoculum pretreated anaerobic sludge and an optimum initial

IP T

glucose concentration 0.11 mol/L. Another study showed a maximum yield of 1.52 mol

H2/mol hexose on immobilized Clostridium butyricum using 0.0977 mol of glucose/L [39].

SC R

Zhao et al. [40] obtained even lower yields 0.73-0.83 mol H2/mol glucose using Enterobacter aerogenes mutants, the optimal initial substrate concentration being 0.08 mol of glucose/L.

Several studies have been reported in literature to describe the performance of systems; Lin et al. [41] used the modified ADM1 to describe the glucose metabolism and product distribution

U

(butyrate, acetate and ethanol) by selected Clostridium species in batch cultures.

N

Gadhamshetty et al. [42] used the modified ADM1 for modelling batch hydrogen production in eight dark fermentation systems varying different parameters. Ntaikou et al. [43] developed

A

and applied a modified version of the model to describe and predict batch and continuous

M

fermentative hydrogen production by the fibrolytic bacterium Ruminococcus albus. The description of hydrogen production in semi-continuous systems with gas extraction using

has not been reported.

ED

mixed microbial cultures using a modified ADM1 model including a Correction Factor (CF)

PT

The aim of the present study is to investigate the applicability of a modified ADM1 to the semi-continuous fermentative hydrogen production process to point out the inhibition caused by a high substrate concentration. Our process works with continuous produced gas extraction

CC E

during fermentation using a bacterial consortium from anaerobic seed sludge and glucose as model substrate.

A

II.

Materials and methods II.1. Experimental part

In this study, the bioreactor was inoculated with activated sewage sludge samples coming from the biological lines (anoxic zone) of the waste water treatment plant of the Eurométropole de Strasbourg (France).

6

Fermentation tests were carried out in a 1 L semi-batch reactor (Büchi AG) equipped with a jacket, coupled with a thermostated bath (Bioblock Scientific Polystat 5A). The reactor enclosure was at atmospheric pressure. An activated sewage sludge sample of 800 mL (head space of 200 mL), used as inoculum, was agitated at 220 rpm to maintain sludge in suspension during fermentation and facilitate substrate and gas diffusion. The inoculum was going through a thermal treatment at 70°C during 1 h, to inactivate the hydrogenotrophic and non-hydrogen producing microorganisms. The temperature was then kept at 37°C during

IP T

fermentation. The pH was initially about 6.5 and was not controlled during the fermentation. An initial addition of model substrate (from 3.7 to 73.0 mmol/L of glucose) was performed to

SC R

feed the bacterial consortium. A purging valve was available at the bottom of the reactor to collect biomass samples during fermentation for carbohydrates and metabolites analysis.

A swept gas (N2) was bubbling into the reactor with a flow rate of 50 mL/min regulated by a mass flow meter (5850E, Brooks). A cold trap was installed on the gas extraction pipe to

U

condense liquids (Huber TC40) before on-line analyzers.

N

On-line analysis of produced gas was carried out by gas microchromatography (Agilent M200), equipped with two modules and thermal conductivity detectors (TCD). These

A

modules are composed of a 0.5 nm molecular sieve column (10 m x 0.32 mm, 10 µm) and a

M

PoraPLOT U column (8 m x 0.32 mm, 10 µm) for the separation of N2, CH4, O2, H2 and CO2.

during fermentation tests.

ED

An accurate monitoring of gas production was performed with three analysis every 10 min

Bacterial growth was determined from dry matter variation during fermentation. Dry matter

PT

was obtained by measuring the mass difference before and after dehydration at 105°C of a sample of biomass from the bioreactor. A mean dry matter of 5.9 ± 0.7 g/L was determined for the activated sewage sludge used as inoculum in this study. The microbiological analyze

CC E

by DNA sequencing (Illumina) of the consortium in the bioreactor, at the end of the fermentation experiment, has shown that Clostridium sensu stricto genus represent 54.8% of the bacterial population, followed by the Peptoclostridium genus with 16.1%.

A

Chemical analysis (GC-FID, HPLC-UV) were performed on biomass samples after centrifugation (15 min at 15000 rpm), and filtration of the supernatant (pore diameter: 0.22 µm) to limit the presence of solids in samples. Acetate and butyrate analyses were carried out by HPLC-UV (Agilent Technologies 1260 Series). The chromatographic column was a Hi-Plex H (7.7 x 300 mm, 8 µm) specific to the analysis of polar organic compounds. The isocratic mobile phase was composed of 92% of

7

H2SO4 (0.005 mol/L in water) and 8% of acetonitrile. A flow rate of 0.6 mL/min was used at a temperature of 65°C. Ethanol was analyzed by gas chromatography (Agilent 7890A) with flame ionization detector. The chromatographic column was a DB-FFAP (15 m x 0.1 mm, 0.1 µm). Before analysis, the samples were protonated (pH < 2) by addition of a solution of trifluoroacetic acid. The total soluble carbohydrate (including glucose) concentration was analyzed by colorimetric method. The supernatant of centrifuged biomass samples was diluted and mixed

IP T

in test tubes with an anthrone solution (2 g/L in concentrated H2SO4). Test tubes were heated

at 80°C for 20 min, then placed in ice for cooling. Absorbances were measured with a

SC R

spectrophotometer (Varian Cary 3). Results are given in hexose equivalent. The total dissolved carbohydrate concentration in the activated sewage sludge was of 0.23 ± 0.07 mmol/L. An addition of substrate was realized using pure glucose (from 3.7 to 73.0 mmol/L). Calculations are realized on total sugar consumption, taking into account the initial sludge and

U

added glucose concentration of carbohydrate at initial time of fermentation toward final

A

M

1. II.2. Kinetic expressions

N

concentration in the medium.

In 1913, Michaelis and Menten [44] proposed the fundamental kinetic description of microbial growth. This relation can be related to bacterial growth, because the microbial

ED

growth is also an autocatalytic reaction [45]: 𝑣 = 𝜈𝑚𝑎𝑥

𝑆

(1)

𝐾𝑀 +𝑆

PT

where 𝜈𝑚𝑎𝑥 (ℎ−1 ) is the maximum rate achieved by the system, at maximum (saturating) substrate concentrations. The Michaelis constant 𝐾𝑀 (𝑚𝑜𝑙. 𝐿−1 ) is the substrate constant at

CC E

which the reaction rate is half of 𝜈𝑚𝑎𝑥 . 𝑆 (𝑚𝑜𝑙. 𝐿−1 ) is the substrate concentration. In 1949, Monod [46] proposed a model to describe microbial growth: µ = µ𝑚𝑎𝑥

𝑆

(2)

𝐾𝑠𝑀 +𝑆

A

where µ𝑚𝑎𝑥 (ℎ−1 ) is the maximum specific growth rate, 𝐾𝑆𝑀 (𝑚𝑜𝑙. 𝐿−1 ) is the substrate constant and S is the substrate concentration. However, classical Monod’s model doesn't take into consideration the effects of substrate inhibition on the bacterial growth. Therefore, a modified Monod model considering substrate and products inhibition has been developed. The Aiba et al. (1968) model [47], is given as:

8

µ = µ𝑚𝑎𝑥 .

𝑆

𝑆

𝐾𝑠 +𝑆

exp(− )

(3)

𝐾𝐼

where 𝐾𝐼 (𝑚𝑜𝑙. 𝐿−1 ) is the inhibition constant, where bacteria growth is reduced to 50% of the maximum specific growth rate due to substrate inhibition. In 1959, Contois [48] proposed a model for batch processes in steady-state conditions. In the Contois model, the bacterial growth rate depends on the concentration of both substrate and cell-mass. The microbial growth is inhibited at high concentrations of the cell-mass,

µ = µ𝑚𝑎𝑥

IP T

expressed by the following equation: 𝑆

(4)

𝐾𝑆𝐶 𝑋+𝑆

SC R

where 𝐾𝑆𝐶 (𝑚𝑜𝑙. 𝑚𝑜𝑙 𝐶𝑂𝐷 −1 ) is the Contois constant and X the biomass concentration.

For Beba and Atalay [49], the Contois model yields good results both for discontinuous and continuous processes, but its capability to model dynamic processes, is strongly limited.

N

U

2. II.3. Bioreactor model

For the simulation of the experimental data from batch experiments, the modified ADM1

A

model was used to describe the production of hydrogen and other products via dark

M

fermentation by the bacterial consortium. Eqs. (2-4) give the kinetic rate expressions on which the model is based.

ED

To describe the consumption of substrate, the following expression was proposed [50]: 𝑑𝑆

= −𝑌1 µ 𝑋 𝐼𝑝𝐻 + 𝐶𝑠 𝑋

(5)

PT

𝑑𝑡

The first term represents the bacterial glucose consumption; the second term is introduced to take into account (for considering) the hydrolysis of complex substrate contained in the

CC E

biomass to form glucose. 𝑿 is the biomass concentration (𝑚𝑜𝑙 𝐶𝑂𝐷. 𝐿−1 ), 𝒀𝟏 is the substrate yield coefficient,

𝒅𝑺 𝒅𝒕

(𝑚𝑜𝑙. 𝐿−1 . ℎ−1 ) is the change in substrate concentration, µ (ℎ−1) the

specific growth rate and IpH is for the pH inhibition.

A

The microbial growth rate is given by: 𝑑𝑋 𝑑𝑡

= µ 𝑋 𝐼𝑝𝐻 − 𝐾𝑑 𝑋

where 𝐾𝑑 (ℎ−1 ) is the cell death rate;

𝒅𝑿 𝒅𝒕

(6)

(𝑚𝑜𝑙 𝐶𝑂𝐷 . 𝐿−1 . ℎ−1 ) is the change in cell

concentration over time. The growth rate inhibition due to 𝑝𝐻 is given by: 𝑝𝐻−𝑝𝐻𝐿𝐿

𝐼𝑝𝐻 = 𝑒𝑥𝑝 (−3 (𝑝𝐻

𝑈𝐿 −𝑝𝐻𝐿𝐿

9

2

) )

(7)

The given values for upper and lower pH are: 𝑝𝐻𝑈𝐿 = 7.35 and 𝑝𝐻𝐿𝐿 = 4.77, according to experimental data. The widely used eqs. (8) and (9) describe the relationship between substrate consumption rate and acetate, butyrate, ethanol, butanol, carbon dioxide and hydrogen production rates [51]. Acetate, butyrate, ethanol, and butanol (𝑃) production rates are given by the following generic equation: 𝑑𝑡

= −𝐹𝑃/𝑆

𝑑𝑆 𝑑𝑡

𝐶𝐹 + 𝐶𝑃 𝑋

(8)

The production rates of hydrogen and carbon dioxide (g) are given by: 𝑑𝑃 𝑑𝑡

= −𝐹𝑔/𝑆

𝑑𝑆 𝑑𝑡

𝐶𝐹

(9)

SC R

where 𝑃 (𝑚𝑜𝑙) is the cumulative product production.

IP T

𝑑𝑃

𝐹𝑃/𝑆 is the product yield coefficient of metabolites (acetate, butyrate, and ethanol); 𝐹𝑔/𝑆 is the gas (hydrogen and carbon dioxide) yield coefficient.

U

In this study, when the maximum tolerable substrate concentration is exceeded, the specific

N

growth rate decreases. Therefore, an inhibition term was added as a Correction Factor (CF): 𝑆

𝐶𝐹 = 1 − 𝑒𝑥 𝑝 (− 𝐾 )

(10)

A

𝐼

M

The modelling procedure was divided in two steps. Firstly, the model and kinetic parameters were determined using the Trust-Region-Reflective optimization algorithm in MATLAB

ED

package optimization algorithm to fit equations given in Table 1 coupled with one of the kinetic models given in the same table with the fourth Runge-Kutta for integrating differential equations. The experimental acetate, butyrate and hydrogen time-profiles were used to adjust

PT

the model and kinetic parameters.

Secondly, the modified ADM1 model was used to predict the dark fermentation outputs using

CC E

the previously obtained parameters. It is obvious that the modified ADM1 model simulates with confidence the experimental data. Three models were obtained using the modified ADM1 model and detailed in the Table 1: Results and discussion

A

III.

The metabolic products generation mechanism used for the determination of the mathematical equations that describe the generation of metabolites, is based on the metabolic pathways, illustrated in Fig. 1. By analyzing the experimental profiles obtained, it has been shown in general that the significant products are acetate, butyrate, propionate, ethanol, CO2 and H2. 10

According to our experimental data reported on Fig. 2, corresponding to the production of metabolites during the fermentation (initial glucose content S0 = 11.2 𝑚𝑚𝑜𝑙/𝐿𝑚𝑒𝑑𝑖𝑢𝑚 ), we can observe that butyrate and acetate, which are hydrogen co-products according to eq. (11) and eq. (12), are the main produced metabolites. The concentration of these metabolites is

IP T

almost equivalent with a similar production profile. 𝐶12 𝐻12 𝑂6 → 2𝐶𝐻3 𝐶𝐻2 𝐶𝐻2 𝐶𝑂𝑂𝐻 + 2𝐶𝑂2 + 2𝐻2

(11)

𝐶12 𝐻12 𝑂6 + 2𝐻2 𝑂 → 2𝐶𝐻3 𝐶𝑂𝑂𝐻 + 2𝐶𝑂2 + 4𝐻2

(12)

SC R

A small amount of butanol was produced, and the evolution of the propionate concentration

shows a peak of production at 5 h followed by a rapid consumption of this metabolite. Secondary metabolites such as pyruvate, formate and isovalerate have been produced in small amounts and are consumed during fermentation. It was noticed that the first nine hours of

A

N

U

operation were sufficient to reach the maximum H2 production (0.016 mol).

The modelling procedure was divided in two steps. Firstly, determine model and kinetic

M

parameters using the Trust-Region-Reflective optimization algorithm in MATLAB package optimization algorithm to fit eqs. (5) - (11) coupled with one of the kinetic models given by

ED

eqs. (2) - (4) with the fourth Runge-Kutta for integrating differential equations eqs. (5) - (11). The experimental acetate, butyrate, and hydrogen time-profiles were used to adjust the model

PT

and kinetic parameters.

CC E

After calculation, the estimated parameters obtained are given in Table 2.

Yield factors established in this study are compared in Table 3 with typical values reported in the literature. While most of the parameters established in this study are similar to the reported values, only 𝑭𝑯𝟐 /𝑺 used in this study is higher than that found in the literature. This

A

difference could be due to the use of the continuous extraction of the produced gases by a sweeping gas, which improves the production of biohydrogen [52].

Secondly, the ADM1 model was used to predict the dark fermentation outputs using the previously obtained parameters. It is obvious that the modified ADM1 model simulates with confidence the experimental data. 11

Three models were obtained using the modified ADM1 model: 1. Modified Model with Aiba Kinetic model using a Correction Factor (MMAK-CF); 2. Modified Model with Contois Kinetic model (MMCK); 3. Modified Model with Monod kinetic model (MMMK). According to the results showed in Fig. 3, representing biomass growth, substrate

IP T

consumption, gas and metabolites production as a function of time, the best model, which can be used with confidence, is the MMAK-CF model. This model achieved a better simulation of the experimental data, even though some differences between simulated and experimental

SC R

results remain, especially for the substrate (glucose). To reduce these differences, an

estimation of the parameters in continuous experiments should be used for a better fitting of the experimental data. The MMMK and MMCK model estimates are very poor, especially for biomass growth. These models are not convenient for the description of this type of

N

Parameters sensitivity analysis

A

IV.

U

fermentation process.

By studying the effect of different initial concentrations of glucose on the bacterial growth

M

and hydrogen production, Ntaikou et al. [43] found that the kinetic constants and yield coefficients remained unchanged when the initial glucose concentration was changed. Fig. 4

ED

shows the glucose consumption for different initial substrate concentrations (S0 in mol.L-1) with time according to the MMAK-CF model. For each initial concentration, a maximum

PT

specific growth rate (Table 3) is reached at a certain concentration, above which a decrease of the specific growth rate takes place. During the first hours, it was noted that the sugar

CC E

consumption rate decreases with the increase of the initial quantity of sugar. Beyond a certain amount, bacteria can no longer follow to consume glucose. According to Edwards [53], this could be caused by a high osmotic pressure of the medium or a specific toxicity of the

A

substrate.

a) Effect of initial substrate concentration on bacterial growth

Fig. 5 shows the bacteria growth profiles at different initial substrate concentrations. It has been shown that bacterial growth increases in an exponential manner, reaches a maximum value, and then declines. It is obvious that the biomass concentration increases with increasing glucose initial concentration up to a concentration value of 0.0220 mol. L-1 of substrate, above which it was found that the maximum bacterial concentration decreased. At low substrate 12

concentration, bacterial growth starts quickly, contrarily to the case of high initial substrate concentrations. This could be due to the inhibition of bacterial growth by the substrate. Thus, it is reasonable to say that when we increase the initial substrate concentration, the high substrate concentration will inhibit the activity of bacteria, the specific growth rate (μ) so decreased (exponential term of eq. (3)), and by consequence, this can lead to [54]: a modified chemical potential of substrates or products,

-

an altered permeability of cells,

-

a changed activity of enzymes,

-

a dissociation of enzymes or metabolic aggregates,

-

an affected enzyme synthesis by interaction with the genome or the transcription

SC R

IP T

-

process.

From a biological point of view, it can be said that the high initial substrate concentration

U

changes the direction and nature of enzymatic reactions leading to the production of other products acting as inhibitors [53]. This could be due to the production of high levels of acids

N

(acetate, butyrate, propionate and formate) at high initial substrate concentration. Fig. 6 shows

A

that the final total acid production increases with increasing the initial substrate concentration.

M

The high acid production negatively affects bacterial growth and consequently inhibits the formation of hydrogen [55, 56]. Indeed, researchers have found that the increase in total acid

ED

production leads to a pH variation causing the inhibition of the bacteria under acidogenic

PT

conditions.

The hydrogen yield and hydrogen production rate at different initial concentration of substrate

CC E

have been reported in Table 4. The inhibition constant ranges from 0.0017 mol-1 to 0.0195 mol-1 increasing with the initial substrate concentration; the maximum specific growth rate (Table 4) shows the same variation. A hydrogen yield of 2.24 mol H2/mol glucose and a hydrogen production rate of 0.007 mol/h were obtained at initial substrate concentration of

A

0.022 mol. L-1.

b) Effect of initial concentration of substrate on hydrogen production

Fig. 7 reveals that the substrate concentration was found to significantly affect the final production of hydrogen. Cumulative hydrogen production showed a gradual improvement

13

with increase in initial glucose concentration from 3.7 mmol. L-1 to 73.0 mmol.L-1 to reach a maximum value of 0.06 mol of H2. Indeed, for the cumulative produced hydrogen, the fit between the model and the experiment is good for concentrations up to the optimal initial concentration of sugars (S0 = 0.0220 mol.L-1), whereas beyond this value there are some differences. For the modelled data, after 10 hours of fermentation, the curve is asymptotic (there is no more hydrogen production), whereas the experimental values show a slowing down of the hydrogen production rate after

IP T

this period for values of high initial concentrations of sugars (S0 = 0.0293 mol.L-1 and S0 = 0.073 mol.L-1), to finally tend towards the value of the asymptote. It can be concluded that the

SC R

high concentrations of substrate clearly play a role in the hydrogen production kinetics not

anticipated by the model; when the concentration of substrate is greater than the maximum tolerable, the term of inhibition (correction factor) is no longer suitable for the description of

→

𝑒𝑥 𝑝 (−

𝑆 ) ~0 𝐾𝐼

→

𝐶𝐹 ~ 1

N

𝑆~ +∞

U

high concentrations of substrate:

A

In Fig.8 an improvement in the rate of hydrogen production was observed from 3.7 mmol. L-1

M

to 22 mmol.L-1 and there is a downward trend from 22 mmol.L-1 to 73 mmol.L-1. The glucose was converted into hydrogen, CO2 and acids during dark fermentation and it could be said

ED

that the final acids produced are related to the production of hydrogen, since hydrogen is a coproduct of formation of acids [57]. As a result, the rate of hydrogen production decreases with increasing initial substrate concentration, indicating a change in pH caused by high

CC E

Conclusion

PT

concentrations of acids that inhibit the bacterial activity.

Three models were used to describe the production of dark fermentative hydrogen in semicontinuous gas extraction systems. The MMAK-CF model proved to be the best model with a

A

high reliability. The effects of initial substrate concentration on the biohydrogen production have been well described by this model. In semi-batch tests of dark fermentation, with produced gas extraction, using the bacterial consortium, the maximum rate of hydrogen production is 7 mmol.h-1 at a substrate concentration of 22 mmol.L-1. It was found that working with a higher substrate concentration would reduce the activity of the bacteria and therefore slow down hydrogen production, inhibited by high acid concentration conditions. 14

Enhanced hydrogen production was observed at lower substrate concentration conditions which were found to be more favourable for improving the rate and yield of hydrogen production from anaerobic seed sludge with gas extraction. Thus, it has been confirmed not only that working with higher substrate concentrations would reduce the activity of the bacteria and consequently decrease hydrogen production rate but also that modelling allows to establish an optimum of the substrate concentration for a given system. In addition, based on this modelling work, the hydrogen production profile as a function of time can be reliably

IP T

generated and the period during which the production is maximal (hydrogen flow rate) can be

A

CC E

PT

ED

M

A

N

U

SC R

established; this element is relevant in the dynamics of performing continuous tests.

15

References

[1] S.K. Khanal, W.H. Chen, L. Li, S. Sung, Biological hydrogen production: effects of pH and intermediate products, Int. J. Hydrogen Energ. 29 (2004) 1123–1131. [2] P. Sinha, A. Pandey, An evaluative report and challenges for fermentative biohydrogen production, Int. J. Hydrogen Energ. 36 (2011) 7460-7468.

IP T

[3] F.Wang, Y. Cao, J. Zhou, Thermodynamic analysis of high temperature helium heated fuel reforming for hydrogen production, Int. J. Energ. Res. 39 (3) (2015) 418-432.

SC R

[4] M. Rosen, D. Scott, Comparative efficiency assessments for a range of hydrogen production processes, Int. J. Hydrogen Energ. 23 (1998) 653-659.

[5] JO'M. Bockris, The economics of hydrogen as a fuel, Int. J. Hydrogen Energ. 6 (1981)

U

223-241.

[6] JR. Benemann, Feasibility analysis of photobiological hydrogen production, Int. J.

N

Hydrogen Energ. 22 (1997) 979-987.

A

[7] F. Mueller-Langer, E. Tzimas, M. Kaltschmitt, S. Peteves, Techno-economic assessment

M

of hydrogen production processes for the hydrogen economy for the short and medium term, Int. J. Hydrogen Energ. 32 (2007) 3797-3810. [8] JR. Bartels, MB. Pate, NK. Olson, An economic survey of hydrogen production from

8371-8384.

ED

conventional and alternative energy sources, Int. J. Hydrogen Energ. 35 (16) (2010)

PT

[9] N. Pavlos, P. Andreas, A comparative overview of hydrogen production processes,

CC E

Renew. Sust. Energ. Rev. 67 (2017) 597-611.

[10] J. Benemann, Hydrogen biotechnology: Progress and prospects, Nat. Biotechnol. 14 (9) (1996) 1101-1103.

A

[11] D. Das, T. N. Veziroğlu, Hydrogen production by biological processes: a survey of literature, Int. J. Hydrogen Energ. 26 (2001) 13-28.

[12] X. M. Guo, E. Trably, E. Latrille, H. Carrère, J. Steyer, Hydrogen production from agricultural waste by dark fermentation review, Int. J. Hydrogen Energ. 35 (2010) 10660-10673.

16

[13] T. A. Kotsopoulos, I. A. Fotidis, N. Tsolakis, G. G. Martzopoulos, Biohydrogen production from pig slurry in a CSTR reactor system with mixed cultures under hyperthermophilic temperature (70ºC), Biomass Bioenerg. 33 (2009) 1168-1174. [14] W. Wukovits, A. Drljo, E. Hilby, A. Friedl, Integration of Biohydrogen Production with Heat and Power Generation from Biomass Residues, Chem Eng J 35 (2013) 1003 – 1008.

anaerobic contact filter, Int. J. Hydrogen Energ. 31 (2006) 1284-1291.

IP T

[15] K. Vijayaraghavan, D. Ahmad, Biohydrogen generation from palm oil mill effluent using

[16] Y. Fan, Y. Zhang, S. Zhang, H. Hou, B. Ren, Efficient conversion of wheat straw wastes

SC R

into biohydrogen gas by cow dung compost, Bioresource Technol. 97 (2006) 500-505.

[17] G. Ivanova, G. Rákhely, K.L. Kovacs, Thermophilic biohydrogen production from energy plants by Caldicellulosiruptur saccharolyticus and comparison with related studies, Int. J. Hydrogen Energ. 34 (2009) 3659-3670.

U

[18] O. Elsharnouby, H. Hafez, G. Nakhla, M.H. El Naggar, A critical literature review on

N

biohydrogen production by pure cultures, Int. J. Hydrogen Energ. 38 (12) (2013) 4945– 4966.

A

[19] R. Kothari, D. P. Singh, V.V. Tyagi, S. K. Tyagi, Fermentative hydrogen production –

M

An alternative clean energy source, Renew. Sust. Energ. Rev. 16 (2002) 2337-2346. [20] Y.M. Wong, T.Y. Wu, J.C. Juan, A review of sustainable hydrogen production using

ED

seed sludge via dark fermentation, Renew. Sust. Energ. Rev. 34 (2014) 471–482. [21] M. Calusinska, T. Happe, B. Joris, A Wilmotte, The surprising diversity of clostridial

1588.

PT

hydrogenases: a comparative genomic perspective, Microbiology+ 156 (2010) 1575-

[22] P.Y. Lin, L.M. Whang, Y.R. Wu, W.J. Ren, C.J. Hsiao, S.L. Li, Chang J.S, Biological

CC E

hydrogen production of the genus Clostridium: Metabolic study and mathematical model simulation, Int. J. Hydrogen Energ. 32 (2007) 1728-1735.

[23] R.K. Thauer, K. Jungermann, K. Decker, Energy conservation in chemotrophic anaerobic

A

bacteria, Bacteriol. Rev. 41 (1977) 100–180.

[24] G. Voordouw, Carbon monoxide cycling by Desulfovibrio vulgaris Hildenborough, J. Bacteriol. 184 (2002) 5903–5911. [25] Q.B. Zhao, H.Q. Yu, Fermentative H2 production in an upflow anaerobic sludge blanket reactor at various pH values, Bioresour Technol. 99 (5) (2008) 1353–1358. [26] B. Dabrock, H. Bahl, G. Gottschalk, Parameters affecting solvent production by Clostridium pasteurianum, Appl. Environ. Microbiol. 58 (4) (1992) 1233–1239. 17

[27] S. Afschar, K. Schaller, K. Schügerl, Continuous production of acetone and butanol with shear-activated Clostridium acetobutylicum, Appl. Microbiol. Biotechnol. 23 (1986) 315–321. [28] Y. Ueno, S. Otsuka, M. Morimoto, Hydrogen production from industrial wastewater by anaerobic microMora in chemostat culture, J. Ferment. Bioeng. 82 (1996) 194–197. [29] R.A. Holt, A.J. Cairns, G.J. Morris, Production of butanol by Clostridium puniceum in batch and chemostat culture, Appl. Microbiol. Biotechnol. 27 (1988) 319–324.

production

by

the

extreme

thermophile,

Caldicellulosiruptor

saccharolyticus,

SC R

Biotechnol. Bioeng. 81(3) (2003) 255–262.

IP T

[30] E.W. Van Niel, P.A. Claassen, A.J. Stams, Substrate and product inhibition of hydrogen

[31] Y. Mu, X.J. Zheng, H.Q. Yu, R.F. Zhu, Biological hydrogen production by anaerobic sludge at various temperatures, Int. J. Hydrogen Energ. 31(6) (2006) 780–785. [32] J.L. Wang, W. Wan, Application of desirability function based on neural network for

U

optimizing biohydrogen production process, Int. J. Hydrogen Energ. 34 (3) (2009)

N

1253-1259.

[33] J.J. Lay, Y.J. Lee, T. Noike, Feasibility of biological hydrogen production from organic

A

fraction of municipal solid waste, Water Res. 33 (1999) 2579-2586.

M

[34] M.H. Zwietering, I. Jongenburger, F.M. Rombouts, K.Van’t Riet, Modeling of the bacterial growth curve, Appl. Environ. Microbiol. 56 (1990) 1875-1881

ED

[35] D.J. Batstone, J. Keller, I. Angelidaki, S.V. Kalyuzhnyi, S.G. Pavlostathis, A. Rozzi, W.T.M Sanders, H. Siegrist, V.A. Vavilin, Anaerobic Digestion Model No.1 (ADM1),

PT

IWA Task Group for Mathematical Modelling of Anaerobic Digestion Processes, IWA Publishing, London, UK. (2002). [36] D.J. Batstone, J. Keller, J.P. Steyer, A review of ADM1 extensions, applications, and

CC E

analysis: 2002-2005, Water. Sci. Technol. 54 (4) (2006) 1-10.

[37] R. Kleerebezem, M.C.M. Van Loosdrecht, Critical analysis of some concepts proposed in ADM1, Water. Sci. Technol. 54(4) (2006) 51-57.

A

[38] E. Kan, Effects of pretreatments of anaerobic sludge and culture conditions on hydrogen productivity in dark anaerobic fermentation, Renew. Energ. 49 (2013) 227-231.

[39] P. Plangklang, A. Reungsang, S. Pattra, Enhanced bio-hydrogen production from sugarcane juice by immobilized Clostridium butyricum on sugarcane bagasse, Int. J. Hydrogen Energ. 37 (2012) 15525-15532.

18

[40] H. Zhao, K. Ma, Y. Lu, C. Zhang, L. Wang, H. Xing, Cloning and knockout of formate hydrogen lyase and H2-uptake hydrogenase genes in Enterobacter aerogenes for enhanced hydrogen production, Int. J. Hydrogen Energ. 34 (2009) 186-194. [41] P.Y. Lin, L.M. Whang, Y.R. Wu, W.J. Ren, C.J. Hsiao, S.L. Li, J.S. Chang, Biological hydrogen production of the genus Clostridium: metabolic study and mathematical model simulation, Int. J. Hydrogen Energ. 32 (12) (2007) 1728-1735. [42] V. Gadhamshetty, Y. Arudchelvam, N. Nirmalakhandan, D.C. Johnson, Modeling dark

IP T

fermentation for biohydrogen production: ADM1-based model vs. Gompertz model, Int. J. Hydrogen Energ. 35 (2010) 479-490.

SC R

[43] I. Ntaikou, H.N Gavala, G. Lyberatos, Application of a modified Anaerobic Digestion

Model 1 version for fermentative hydrogen production from sweet sorghum extract by Ruminococcus albus, Int. J. Hydrogen Energ. 35 (2010) 3423-3432.

[44] L. Michaelis, M.L. Menten, Die Kinetik der Invertinwirkung, Biochem. Z. 49 (1913)

U

333−369.

N

[45] T. Yano, T. Nakahara, S. Kamiyama, K. Yamada, Kinetic studies on microbial activities in concentrated solutions. Part I. effect of excess sugars on oxygen uptake rate of a cell

A

free respiratory system, Agric. Biol. Chem. 30 (1) (1966) 42-48.

M

[46] J. Monod, The Growth of Bacterial Cultures, Annu. Rev. Microbiol. 3 (1949) 371-394. [47] S. Aiba, M. Shoda, M. Nagatani, Kinetics of product inhibition in alcohol fermentation,

ED

Biotechnol. Bioeng. 10 (6) (1968) 845-864. [48] D.E. Contois, Kinetics of bacterial growth: relationship between population density and

PT

specific growth rate of continuous cultures, J. Gen. Microbiol. 21 (1959) 40–50. [49] A. Beba, F.S. Atalay, Mathematical models for methane production in batch Fermenters, Biomass 11(3) (1986) 173-184.

CC E

[50] I.S. Simeonov, B.L. Kalchev, N.D. Christov, Parameter and State Estimation of an Anaerobic Digestion Model in Laboratory and Pilot-Scale Conditions. Preprints of the 18th IFAC World Congress Milano (Italy) (August 28 - September 2 - 2011) 6224-

A

6229.

[51] K.R.J. Perera, N.N. Nirmalakhandan, Modeling fermentative hydrogen production from sucrose supplemented with dairy manure, Int. J. Hydrogen Energ. 36 (2011) 2102-2110. [52] V. Clion, C. Dumas, S. Collin, B. Ernst, Key factors for biohydrogen production by dark fermentation, The Can. J. Chem. Eng. 93 (2) (2015) 309-316. [53] V.H. Edwards, The Influence of High Substrate Concentrations on Microbial Kinetics, Biotechnol. Bioeng. 12 (1970) 679-712. 19

[54] N. Kythreotoua, G. Florides, S.A.T. Tassou, A review of simple to scientific models for anaerobic digestion, Renew. Energ. 71 (2014) 701-714. [55] S. Eker, M. Sarp, Hydrogen gas production from waste paper by dark fermentation: Effects of initial substrate and biomass concentrations, Int. J. Hydrogen Energ. 42 (2017) 2562-2568. [56] H. Argun, F. Kargi, I.K. Kapdan, R. Oztekin, Batch dark fermentation of powdered wheat starch to hydrogen gas: effects of the initial substrate and biomass concentrations,

IP T

Int. J. Hydrogen Energ. 33 (2008) 6109-6115.

[57] H. Argun, F. Kargi, Bio-hydrogen production by different operational modes of dark and

A

CC E

PT

ED

M

A

N

U

SC R

photo-fermentation: an overview, Int. J. Hydrogen Energ., 36 (2011) 7443-7459.

20

IP T SC R U N A M ED

A

CC E

PT

Fig. 1 – Assumed metabolic path of glucose consumption (adapted from [42]).

21

acetate

propionate

butyrate

butanol 1E-04 9E-05 8E-05

6E-03

7E-05 6E-05

4E-03

5E-05 4E-05

IP T

3E-05

2E-03

2E-05 1E-05

0E+00

0E+00

0

2

4

6

8

10 Time(h)

12

14

SC R

Metabolite production (mol) - ethanol, acetate, propionate, butyrate

8E-03

Metabolite production (mol)- buthanol

ethanol

16

18

20

Fig. 2 – Evolution of the main metabolites (ethanol, butanol, acetate, propionate, butyrate) production during the fermentation as a function of time.

U

Fermentation test carried out from activated sludge from the anoxic zone in a semi-continuous bioreactor (1,0 L);

N

inoculum volume: 0.8 L, fermentation temperature: 38°C, glucose addition: 11.2 mmol/Lmedium; sweep gas flow

A

CC E

PT

ED

M

A

rate (nitrogen): 65 mL/min/Lmedium

22

(A) 4.23E-02 4.22E-02 4.21E-02 4.20E-02

IP T

Experimental

4.19E-02

MMAK-CF

4.18E-02

MMCK

MMMK

4.17E-02

SC R

Biomass growth (mol COD/L)

4.24E-02

4.16E-02 0

5

10

15

A

Experimental

M

1.0E-02 8.0E-03

MMCK

ED

6.0E-03

MMAK-CF

4.0E-03

MMMK

PT

Substrate concentration (mol/L)

N

(B) 1.2E-02

U

Time (h)

20

2.0E-03

CC E

0.0E+00

2

4

6

8

10

Time (h)

A

0

23

12

14

16

18

20

(C) 7.0E-03 6.0E-03 5.0E-03 Experimental

4.0E-03

MMAK-CF

IP T

3.0E-03

MMCK

2.0E-03

MMMK

1.0E-03 0.0E+00 0

2

4

6

8

10

12

Time (h)

16

18

20

N A

6.0E-03

M

5.0E-03 4.0E-03 3.0E-03

Experimental

ED

MMAK-CF

2.0E-03 1.0E-03 0.0E+00

2

4

6

MMCK MMMK

8

10

Time (h)

A

CC E

0

PT

Butyrate production (mol)

7.0E-03

14

U

(D)

SC R

Acetate production (mol)

8.0E-03

24

12

14

16

18

20

(E)

2.5E-03

2.0E-03

IP T

Experimental MMAK-CF MMCK

1.5E-03

MMMK

SC R

Ethanol production (mol)

3.0E-03

1.0E-03 0

5

10

15

Time (h)

U

(F)

N

2.E-02

M

A

2.E-02

Experimental MMAK-CF

ED

1.E-02

5.E-03

0.E+00

2

4

6

MMCK MMMK

8

10

Time (h)

A

CC E

0

PT

CO2 production (mol)

20

25

12

14

16

18

20

(G) Hydrogen production (mol)

1.6E-02 1.4E-02 1.2E-02 1.0E-02 Experimental

6.0E-03

MMAK-CF

4.0E-03

MMCK

2.0E-03

MMMK

0.0E+00 2

4

6

8

10

12

Time (h)

14

16

18

SC R

0

IP T

8.0E-03

20

Fig. 3 – Experimental data and model simulation profiles of: (a) microbial growth; (b) glucose, (c) acetate, (d) butyrate and (e) ethanol production; (f) carbon dioxide and (g) hydrogen production.

U

Experimental data on tests carried out from activated sludge from the anoxic zone in a semi-continuous bioreactor (1.0 L); inoculum volume: 0.8 L, fermentation temperature: 38°C, glucose addition: 11.2 mmol/L medium; sweep gas flow rate

N

(nitrogen): 65 mL/min/Lmedium

A

CC E

PT

ED

M

uncertainties corresponding to a triplicate of fermentation.

A

The error bars correspond to the measurement uncertainties and for 5, 6 and 9 hours is added to the previous ones, the

26

7E-02 S0=0.0037 mol/lL 6E-02

S0=0.0075 mol/lL

5E-02

S0=0.0110 mol/lL

4E-02

S0=0.0220 mol/lL S0=0.0293 mol/lL

3E-02

S0=0.0730 mol/lL

2E-02 1E-02 0E+00 5

10

Time (h)

15

20

SC R

0

IP T

Substrate concentration (mol.L-1)

8E-02

25

Fig. 4 – Evolution of the substrate concentration as a function of time for different initial substrate

A

CC E

PT

ED

M

A

N

U

concentrations (S0 in mol.L-1) according to the model MMAK-CF

27

S0=0.0037 mol/lL

4.28E-02

S0=0.0075 mol/lL

4.26E-02

S0=0.0220 mol/lL

S0=0.0110 mol/lL S0=0.0293 mol/lL

4.24E-02

S0=0.0730 mol/lL

IP T

4.22E-02 4.20E-02 4.18E-02

SC R

Biomass growth (mol COD/L)

4.30E-02

4.16E-02 0

5

10

15

Time (h)

20

U

Fig. 5- Effect of the initial substrate concentration on microbial growth during batch experiments for

A

CC E

PT

ED

M

A

values).

N

different initial substrate concentrations (S0 in mol L-1) according to the MMAK-CF model (simulation

28

0.05 0.04 0.03 0.02 0.01 0 0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

SC R

Initial substrate concentration (mol/L)

IP T

Final total acid production (mol)

0.06

0.08

A

CC E

PT

ED

M

A

N

U

Fig. 6- Variation of final total acid production with initial substrate concentration (simulation values)

29

(a) 7E-02 S0=0.0037 mol/lL S0=0.0075 mol/lL S0=0.0110 mol/lL S0=0.0220 mol/lL

5E-02

S0=0.0293 mol/lL S0=0.0730 mol/lL

4E-02 3E-02

IP T

Hydrogen production (mol)

6E-02

2E-02

0E+00 0

5

10

15

(b)

7E-02

S0=0.0293 mol/lL S0=0.0730 mol/L S0=0.058 mol

2E-02 1E-02 0E+00

25

ED

3E-02

M

S0=0.0220 mol/lL 4E-02

20

A

S0=0.0110 mol/lL

5E-02

25

N

S0=0.0075 mol/lL

PT

Hydrogen production (mol)

S0=0.0037 mol/lL 6E-02

20

U

Time (h)

SC R

1E-02

CC E

0

5

10

15

Time (h)

Fig. 7- Effect of initial concentration of substrate on hydrogen production (a) modelling according to the

A

model MMAK-CF (b) experimental, for different initial substrate concentrations (S0 in mol L-1)

30

Hydrogen production rate (mol/h)

(a)

S0=0.0037 mol/lL 7.0E-03

S0=0.0075 mol/lL

6.0E-03

S0=0.0110 mol/lL S0=0.0220 mol/lL

5.0E-03

S0=0.0293 mol/lL 4.0E-03

S0=0.0730 mol/lL

3.0E-03

1.0E-03 0.0E+00 0

5

10

15

(b)

S0=0.0037 mol/lL

U

7.0E-03

N

6.0E-03

3.0E-03 2.0E-03 1.0E-03

0

5

PT

0.0E+00

10

S0=0.0110 mol/lL

S0=0.0293 mol/lL S0=0.0730 mol/lL

M

4.0E-03

S0=0.0075 mol/lL

S0=0.0220 mol/lL

A

5.0E-03

ED

Hydrogen production rate (mol/h)

20

SC R

Time (h)

IP T

2.0E-03

15

20

Time (h)

CC E

Fig. 8- Effect of initial concentration of substrate on hydrogen production rate (a) modelling according to

A

the MMAK-CF model (b) experimental, for different initial substrate concentrations (S0 in mol/L)

31

Table 1 – Differential equations and kinetics of each model

𝑑𝑆𝑣𝑎𝑟 𝑄𝑙𝑖𝑞 = (𝑆 − 𝑆𝑣𝑎𝑟 ) + 𝑅𝑣𝑎𝑟 𝑑𝑡 𝑉𝑙𝑖𝑞 𝑣𝑎𝑟𝑖𝑛

ADM 1 Model [35] Modified Model with Aiba Kinetic model

µ = µ𝑚𝑎𝑥 .

using a Correction Factor (MMAK-CF)

𝑆 𝑆 exp(− ) 𝐾𝑠 + 𝑆 𝐾𝐼

SC R

IP T

𝑑𝑆 = −𝑌1 µ 𝑋 𝐼𝑝𝐻 + 𝐶𝑠 𝑋 𝑑𝑡 𝑑𝑋 = µ 𝑋 𝐼𝑝𝐻 − 𝐾𝑑 𝑋 𝑑𝑡 𝑑𝑃 𝑑𝑆 = −𝐹𝑃/𝑆 𝐶𝐹 + 𝐶𝑃 𝑋 𝑑𝑡 𝑑𝑡 𝑑𝑃 𝑑𝑆 = −𝐹𝑔/𝑆 𝐶𝐹 𝑑𝑡 𝑑𝑡 𝑆 µ = µ𝑚𝑎𝑥 𝐾𝑆𝐶 𝑋 + 𝑆 𝑑𝑆 = −𝑌1 µ 𝑋 𝐼𝑝𝐻 + 𝐶𝑠 𝑋 𝑑𝑡 𝑑𝑋 = µ 𝑋 𝐼𝑝𝐻 − 𝐾𝑑 𝑋 𝑑𝑡 𝑑𝑃 𝑑𝑆 = −𝐹𝑃/𝑆 + 𝐶𝑃 𝑋 𝑑𝑡 𝑑𝑡 𝑑𝑃 𝑑𝑆 = −𝐹𝑔/𝑆 𝑑𝑡 𝑑𝑡

Modified Model with Contois Kinetic model

ED

M

A

N

U

(MMCK)

Modified Model with Monod kinetic model

PT

𝑆 𝐾𝑠𝑀 + 𝑆

𝑑𝑆 = −𝑌1 µ 𝑋 𝐼𝑝𝐻 + 𝐶𝑠 𝑋 𝑑𝑡 𝑑𝑋 = µ 𝑋 𝐼𝑝𝐻 − 𝐾𝑑 𝑋 𝑑𝑡 𝑑𝑃 𝑑𝑆 = −𝐹𝑃/𝑆 + 𝐶𝑃 𝑋 𝑑𝑡 𝑑𝑡 𝑑𝑃 𝑑𝑆 = −𝐹𝑔/𝑆 𝑑𝑡 𝑑𝑡

A

CC E

(MMMK)

µ = µ𝑚𝑎𝑥

32

0.0456 (mol. L-1)

𝒀𝟏

16.005

𝑪𝒆𝒕𝒉𝒂𝒏𝒐𝒍

15.10-5 (h-1)

𝑲𝒅

0.0041 (h-1)

µ𝒎𝒂𝒙 𝑪𝒂𝒄𝒆𝒕𝒂𝒕𝒆

0.693 (h-1) 0.0014 (h-1)

𝑪𝒃𝒖𝒕𝒚𝒓𝒂𝒕𝒆

2.2.10-6 (h-1)

𝑪𝑺 𝑲𝑰

0.0707(h-1) 0.0037 (mol.L-1)

A

CC E

PT

ED

M

A

N

U

SC R

𝑲𝑺

IP T

Table 2 – Estimated parameters

33

Table 3 – Acetate, butyrate, ethanol, CO2 and H2 yield factors

Found in this study (g/g)

Found in literature (g/g)

𝑭𝒂𝒄𝒆𝒕𝒂𝒕𝒆/𝑺

0.3227

0.32 [53] 0.41 [42] 0.11 [35]

𝑭𝒃𝒖𝒕𝒚𝒓𝒂𝒕𝒆/𝑺

0.1300

0.52 [53] 0.13 [42] 0.54 [35]

𝑭𝒆𝒕𝒉𝒂𝒏𝒐𝒍/𝑺

0.0017

-

𝑭𝑪𝑶𝟐 /𝑺

0.2325

-

𝑭𝑯𝟐 /𝑺

0.0237

0.015 [53] 0.014 [35, 42]

A

CC E

PT

ED

M

A

N

U

SC R

IP T

Yield factor

34

Table 4- Inhibition constant, maximum specific growth rate, hydrogen yield and hydrogen production rate for selected initial substrate concentration (in silico data)

Substrate initial concentration

KI (mol

L-1)

µmax

Hydrogen yield

Hydrogen

(h-1)

(mol H2/mol glucose)

production rate (mol.h-1)

(S0) (mol. L-1) 0.0017

0.735

2.39

0.0035

0.0075

0.0028

0.701

2.05

0.0048

0.0110

0.0037

0.694

1.98

0.0047

0.0220

0.0064

0.689

2.24

0.0070

0.0293

0.0082

0.652

2.20

0.0064

0.0730

0.0195

0.601

1.10

0.0058

A

CC E

PT

ED

M

A

N

U

SC R

IP T

0.0037

35