Detection of low solubility volatiles in the liquid phase – H2 & CO2

Get Complete Project Material File(s) Now! »

Anaerobic digestion – a brief history

Scientific records of gas produced by natural decomposition of organic matter was first reported in the 16th century by Robert Boyle and Stephen Hale, who noted that an inflammable gas was released by disturbing the sediment of streams and lakes and associated this “inflammable air” with decomposing organic material in the sediments [86].
However, it was Alessandro Volta who first discovered methane by collecting the gas from marshes in 1776-1777 [85]. One century later (1859), an anaerobic digester was built in Bombay, India, becoming the first record of AD technological use [87]. Then in 1895, a septic tank was used to generate methane for street lighting in Exeter, England, [88].
The classical AD process is widely used to treat organic waste streams as wastewater and organic solids. However, more interest has been placed in the capability of AD to produce renewable energy in the form of biogas rich in methane and CO2, providing an alternative to fossil fuels, and allowing the recovery of nutrients as phosphate fertiliser [89].
Furthermore, since 1997 classical AD has been recognised by the United Nations as one of the most useful decentralised sources of energy supply, as they are less capital intensive than large power plants [70], and in a recent report, AD is considered crucial to fighting against poverty and energy isolation in developing countries [71].

Classic anaerobic digestion process

Anaerobic digestion involves the transformation of organic compounds to various inorganic and organic products. During AD, a portion of an organic compound may be oxidised while another portion is reduced. It is from this oxidation-reduction of organic compounds that anaerobic microorganisms obtain their energy and produce numerous simplistic and soluble organic compounds [90].
The digestion process begins with bacterial hydrolysis of the fed organic materials in order to break down insoluble polymers, making them available for microorganism consumption. Acidogenic bacteria then convert the products of hydrolysis into carbon dioxide, hydrogen, ammonia, alcohols and organic acids. Acetogenic bacteria convert these resulting alcohols and organic acids into acetic acid, along with additional ammonia, hydrogen and carbon dioxide. Finally, methanogenic archaea converts these products to methane and carbon dioxide [90-92]. Figure 2 illustrates this AD structure, which is detailed as follow.
Hydrolysis is the first step required for anaerobic microbial utilisation of complex polymers, which cannot be hydrolysed by methanogenic archaea themselves. Hydrolytic fermentative bacteria facilitate the extracellular enzymatic hydrolysis of the initial complex organic matter formed by polymers such as polysaccharides, proteins and fats. The hydrolases (enzymes) that catalyses these reactions are cellulase, amylase, protease, and lipase, among others. Hydrolases may be secreted to the extracellular environment or be bound to the cell surface [93,94].
Polysaccharides are generally converted into simple monomeric or dimeric sugars as glucose. Hydrolysis of starch and cellulose yields glucose as monomeric sugar, while hemicellulose is degraded to galactose, arabinose, xylose, mannose and glucose. Proteins are broken down into amino acids, small peptide, ammonia and carbon dioxide by proteases. Lipids are hydrolysed into long and short chain fatty acids and glycerol by lipases [90,93,95]. The hydrolysis products then become substrates for the fermentation processes that follow.

Influence of gas-phase composition and pH on acidogenic fermentation

As stated before, the distribution of acidogenic fermentation products is affected by environmental conditions. This is technologically advantageous since it allows controlling and optimising the production of certain compounds giving flexibility to this process. Here, the influence of gas phase composition and pH on acidogenic fermentation product yields is described.

Gas phase composition effects

Previous works have observed head-space gas composition as an important environmental variable to regulate product spectrum in anaerobic fermentation. Hillman et al. [106] working on anaerobic protozoa batch cultures showed how different initial gas phase compositions resulted in considerable differences in the distribution of produced volatile fatty acids (VFAs) at the end of fermentations. Their main results showed that with an initial 100% N2 gas-phase, the proportion of VFAs produced was 40% acetic, 50% butyric and 10% propionic acid. When the initial gas-phase was composed by a mixture of N2 and H2, the proportion of acetic acid increased, butyric acid decreased and no propionic acid was produced. A mixture of CO2 and N2 in the gas phase resulted in propionic acid increase, butyric acid decrease and acetic acid remaining unchanged.
Similar conclusions were reached for acidogenic fermentations by Tanisho et al. [52] and Karlsson et al. [15]. However, mixed culture fermentation (MCF) experiments with N2 sparging performed by Kim et al. [16] and Mizuno et al. [53], indicated that even when H2 yield was increased, no significant change in liquid end products was observed.
Further research employing hydrogen extraction performed by Karlsson et al. [15] related the increased H2 yield to acetic acid production increase. Experiments on nitrogen-sparged acidogenic reactors by Kraemer and Bagley [17] resulted in both H2 and butyric acid yields increase. Similar observation was found by Kim et al. [16] on their sparging CO2 experiment, where butyric acid yield was increased, acetic and lactic acids were decreased, and propionic acid and ethanol yields not significantly affected by CO2 sparging.
As a consequence, the control of gas-phase composition arose as an attractive method to increase acidogenic produced hydrogen yield close to its theoretical maximum, i.e. 4 molH2·(molhexose)-1 [107].
High H2 partial pressures limit the H2 production by end product inhibition, i.e. making the re-oxidation of reduced ferredoxin and H2-carrying coenzymes less favourable [3,18]. Therefore, a decrease in H2 partial pressure should increase H2 yields [18].
Different techniques have been applied in decreasing the H2 partial pressure in acidogenic bioreactors including gas sparging [16,19,52,53], membrane separation [20,21], and direct suction from reactor headspace [22].

Mathematical Modelling of Acidogenic Fermentation

Mathematical models are crucial tools in the difficult tasks of integrate, analyse and investigate the large quantity of flow-information coming from microbial, chemical and physical phenomena that are taking place within microbiologic mediated processes. Mathematical models can also be used to test scientific hypotheses, to design experiments or to control and optimise an already existing bioprocess.
The key role of mathematical models in biotechnology has become clear and well established over the previous decades. Nowadays, mathematical models are indispensable at every stage of bioprocess development, from the earliest research phase to large-scale industrial implementation. Mathematical models have different levels of complexity that depend on the objective to attain. Model elaboration can be based on a mechanistic approach, empirical knowledge, or a mixture of both. Mechanistic modelling approach accounts for actual mechanisms occurring in the system, while empirical models attempt to fit the observed behaviour using mathematical correlations [4]. In the particular case of acidogenic fermentation, mathematical modelling efforts have been driven by the necessity of process simulation, control and optimisation. Because the natural link between acidogenic fermentation and anaerobic digestion, these modelling efforts were mostly based on previous models designed to simulate anaerobic digestion process. Those models assumed that the systems are kinetically controlled, i.e. without energy limitations as in the aerobic fermentations case.
In this decade, the fact that acidogenic fermentation is an energy-limited process, i.e. being thermodynamically controlled rather than kinetically, has turned the attention to a mechanistic modelling approach based on thermodynamics. With this approach, it is expected to understand and clarify the mechanisms that control acidogenic fermentation.
The next sections present a brief history of anaerobic digestion modelling, followed by the efforts made to modelling acidogenic fermentation processes, and the thermo-kinetic considerations that have been took in this challenge.

Anaerobic digestion modelling – a brief historical review

Anaerobic digestion modelling has been an active research area during the last decades. As stated before, its main driving force was the simulation, control and optimisation of AD bioprocess, which resulted on the elaboration of several dynamic models [110], which have mostly been applied to anaerobic wastewater treatment systems [60,62,111].
Mathematical models for AD have several degrees of complexity. These differences are essentially from the hypothesis used for the model structure assumptions, such as metabolic regulation and inhibitions. Simplest models considered organic matter as a global homogeneous whole, enclosing the total substrate composition in the chemical oxygen demand (COD) rather than a detailed specific substrate composition [25]. Therefore, their applications are reduced to substrates with low variability in composition, such as residues from oil, diary [112] and winery industries [110]. The first attempts of AD modelling began in 1974 with Graef and Andrews [113]. Then, based on the previous authors, Hill and Barth [114], Kleinstreuer et Poweigha [115] and Moletta et al. [116] developed new models with Monod-type kinetic equations for simulate process dynamics. Those models describe the last AD step, i.e. methanogenesis, and represent the total volatile fatty acids (VFA) composition as acetic acid equivalent. These models also explain the accumulation of VFAs phenomena by VFA inhibition of methanogenesis, considering Haldane type kinetics equations [117]. The complexity level of these models started to increase. Models proposed by Kalyuzhnyi and Davlyatshina [118] and Kalyuzhnyi [119] consider five steps and five pH dependent microorganism groups.
Angelidaki et al. model [120] was structured into four steps: hydrolysis, acidogenesis, acetogenesis and methanogenesis. The main model hypotheses are: 1) methanogenesis inhibition by ammonia; 2) acetogenesis inhibition by acetic acid; and 3) acidogenesis inhibition by total VFA. Non-competitive inhibition functions were used, where ammonium ionisation is dependent of pH and temperature. Siegriest et al. [121] also considered ammonia into the sludge anaerobic digestion process, predicting the system pH.

ADM1-type based models

Dynamic modelling of acidogenic fermentation has been mostly based on the ADM1 model. However, direct utilisation of ADM1 on acidogenic fermentation does not well simulate the process. Therefore, modifications are needed. These modifications account for a restructuration of model parameters, or by considering product yield variability, as previously discussed by Mosey [122] and Batstone [127,128].
Restructuration of parameters. Aceves-Lara et al. [65] have developed a model based on experimental data to estimate pseudo-stoichiometric coefficients with a constrained nonlinear optimisation. Their results show that two reactions, one being associated with hydrogen production and the other one with acetate production, could explain 89% of the total variance of the experimental data. This accurately predicts the dynamic evolution of H2 production, biomass and VFAs. This model takes into account physicochemical processes kinetics: acid-base reactions and liquid-gas mass transfer. The model has been validated using data from literature and using a second set of dynamic experimental data different from that used for parameter identification. Arudchelvam et al. [133] have developed a model to predict VFA formation. The model was constructed upon experimental data from cattle manure fermentation. The model is based on the assumption that biodegradable components of cattle manure are composed of particulate forms of cellulose and hemicelluloses that are first hydrolysed to their respective forms and then consumed by acidogenesis to produce H2, CO2 and VFAs. Biomass growth was modelled as a single biomass consuming two substrates with different kinetic consumption constants. The acidogenic products are modelled following the ADM1 model, with the exclusion of the methane formation step. Their results conclude that the assumptions used are valid since the model achieved a goodness of fit between the experimentally measured and predicted profiles of COD, acetic acid and butyric acid, and to a lesser extent in the case of propionic acid.
Hafez et al. [57] modelled a CSTR that decouples the SRT from the HRT, incorporating this in the software BioWin (EnviroSim Associates Ltd., Flamborough, Ontario, Canada), which is widely used for modelling wastewater treatment plants. This model was based on two populations: acidogenic and acetogenic microorganisms. The biomass recirculated undergoes decay. The products of decay include unbiodegradable organic, nitrogen and phosphorous components. Hydrolysis is mediated by acidogenic microorganisms. The acidogenic fermentation produces acetic acid, propionic acid, hydrogen and carbon dioxide. Propionic acid is converted to acetic acid by acetogenic bacteria. Hydrogen is switched off at high levels of propionate, using a propionate inhibition expression. The stoichiometry of each of these processes is previously calibrated to achieve the appropriate product mix. The calibration can be done by trial and error to achieve the best match between modelled and measured data.
Variability of product yields. Rodriguez et al. [55] have modified the ADM1 using variable yield coefficients, and assuming that the product transport from the intracellular to extracellular compartments was ATP-dependent. This model considers a variable acetate and butyrate yields dependent on the hydrogen dissolved concentration and reactor pH. The main change predicted is a shift from acetate to butyrate as the main fermentation product at decreasing pH (7-5.5) and/or increasing hydrogen dissolved concentration. A mechanistic explanation of this prediction is related to energetic issues. Lower pH values required higher energy costs for the cells to transport acid molecules outwards the membrane and on the other hand, the maximum concentration of a product is limited by the thermodynamic feasibility of its production where hydrogen plays a key role especially in the acetate production. Based on their simulations, they concluded that acidogenic fermentation is thermodynamic rather than kinetic controlled. Also, their dynamic results assume that the yield changes occur instantaneously because the stoichiometry functions do not incorporate any time related issue. This is not realistic but was used as a starting point to test the effect of such a dynamic yield change would be in the most extreme situations of instantaneous changes of yields. Slow yield changes may occur in reality but similar results are expected maybe with minor time delays or small variations. However, this model badly predict the effect of pH on ethanol production.
Penumathasa et al. [66] have used the variable stoichiometric approach of Rodriguez [55]. They assumed that the biomass and product yields from glucose degradation are dynamically dependent on the total concentration of undissociated acids in the reactor. At each time instant of the modified ADM1 simulation, the total concentration of undissociated acids is calculated from the current acid concentrations and pH. From this value the biomass yield and the acetate, butyrate and hydrogen yields are computed by linear interpolation from values that were obtained from pseudo steady state at total undissociated concentrations. The remaining COD is allocated to lactate to maintain a closed COD balance all times during simulations. The model was validated with mesophilic sucrose fermenting biohydrogen producing reactor. This model achieved good predictions with the implementation of the variable stoichiometry, without any parameter fitting, i.e. using the standard ADM1 parameters values. The model predicts the stationary and dynamic behaviour of hydrogen, acetate and butyrate concentrations.

READ  X-ray measurement on ITER with Low Voltage Ionization Chambers 

Biothermodynamics Based Modelling of Acidogenic Fermentation

As already exposed in the previous sections, metabolic variability and difficulties obtaining the same products spectrum is often observed in acidogenic fermentation under similar operational conditions. The mechanisms where environmental factors, such as pH, temperature and concentration affect the product fluxes are not yet fully understood in this system.
There is a general consensus that the laws of physics are well understood today and it is time to apply them to systems and processes with high degrees of complexity such as living systems [101]. Living systems are extremely complex and organised hierarchically. This is clearly a result of the process of evolution.
Biology was the first branch of science that attempted to reconstruct past events from today’s knowledge of the biosphere. This quest started with the discovery of fossils of long-extinct species, such as living creatures with calcium carbonate or silica based skeleton (540 million years ago). Contemporary techniques allow uncovering fossils of much simpler organisms that do not possess a skeleton. The first well preserved petrified micro-stamps of organised living organisms, similar to present cyanobacteria, emerged about 3.5 billion years ago [141].
In 1859, Charles Darwin [142] proposed a method that has had great potential to reconstruct the history of life based on differences in selected features of living animal species and not extinct ones. In modern applications of this methodology, the most fundamental characteristics are the genomes of species. Based on the mathematically well defined distance between genomes, it is possible to reconstruct the history of the species, e.g. human evolution [143].
Contemporary biochemistry and molecular biology provide numerous examples of ‘living fossils’. These are archaic metabolic pathways and more or less conserved domains in enzymes [101,102]. The organisation of existing animate matter reflects the history of its evolution and, conversely, the living structures that we encounter on the Earth today are products of the evolution of life.
The three most important characteristics of life that distinguish it from other physical systems were expressed by Darwin in his theory of evolution. Taking into account the achievements of contemporary genetics and biochemistry, these three characteristics can be defined as: life is a process characterised by continuous (1) reproduction, (2) variability and (3) selection (survival of the fittest) [101].

Table of contents :

OBJECTIVES OF THIS WORK
1 STATE OF THE ART
1.1 Acidogenic Fermentation as a Biorefinery Concept
1.1.1 Hydrogen
1.1.2 Acetic acid
1.1.3 Ethanol
1.1.4 Lactic acid
1.1.5 Formic acid
1.2 Acidogenic Fermentation Process
1.2.1 Anaerobic digestion – a brief history
1.2.2 Classic anaerobic digestion process
1.2.3 Acidogenesis fermentation metabolism
1.2.4 Influence of gas-phase composition and pH on acidogenic fermentation
1.3 Mathematical Modelling of Acidogenic Fermentation
1.3.1 Anaerobic digestion modelling – a brief historical review
1.3.2 Acidogenic fermentation modelling
1.3.3 Acidogenic fermentation modelling – thermokinetics considerations
1.4 Biothermodynamics Based Modelling of Acidogenic Fermentation
1.4.1 Fundamental laws of thermodynamic
1.4.2 Modelling of biomass yield
1.4.3 Modelling of metabolic networks
1.4.4 Thermodynamic analysis of acidogenic fermentation
1.5 Membrane Inlet Mass Spectrometry
1.5.1 Membrane – the interface of MIMS
1.5.2 Mass spectrometry – the core of the MIMS
1.5.3 Translation of MIMS signals
2 MATERIALS AND METHODS
2.1 Fermentations Set-Up
2.1.1 Reactor equipments
2.1.2 Media and inoculum
2.2 Analytical Methods
2.2.1 Gas phase
2.2.2 Liquid phase
2.3 MIMS Set-Up
2.4 MIMS Calibration Procedures
2.4.1 Standard calibration
2.4.2 In-process calibration
2.5 Thermodynamic Calculations
2.5.1 Temperature correction
2.5.2 Correction for experimental concentrations
3 RESULTS
3.1 MIMS Signal Translation
3.1.1 Correlation between MIMS signals and measurement
3.1.2 Fermentation kinetics
3.1.3 In-process MIMS signal calibration
3.1.4 Validation of calibration strategies
3.2 Fermentation Results
3.2.1 Effect of head-space composition on product yield spectra
3.2.2 Effect of pH on product yield spectra
3.2.3 Transient states
3.3 Thermodynamic Model
3.3.1 Mass and electron balances
3.3.2 Extracellular thermodynamic model
3.3.3 Catabolic thermodynamic model
3.3.4 Thermodynamics of Anabolism
3.3.5 Equation for the calculation of L
3.4 Thermodynamic Analysis of Product Yield Spectra
3.4.1 Thermodynamic analysis of metabolism
3.4.2 Thermodynamic analysis of anabolism
4 DISCUSSION
4.1 MIMS Signal Translation
4.1.1 Measurement in the gas phase
4.1.2 Detection of low solubility volatiles in the liquid phase – H2 & CO2
4.1.3 Detection of high solubility volatiles in the liquid phase – ethanol
4.1.4 MIMS signal oscillation and noise
4.1.5 Limitations of in-process calibration
4.1.6 Applications of MIMS to fermentation experiments
4.2 Acidogenic Product Spectra Under Environmental Changes
4.2.1 Transient states kinetics under environmental changes
4.2.2 Metabolic shift effects on hydrogen yield
4.2.3 The role of lactate synthesis
4.2.4 The Acetyl-CoA node
4.2.5 H2/NAD+ redox reaction
4.2.6 Homoacetogenesis
4.2.7 Acetate, butyrate and ethanol shifts
4.3 The Thermodynamic Model
4.3.1 Intra-extracellular concentration
4.3.2 Constant intracellular pH
4.3.3 Limitations of this thermodynamic model
5 CONCLUSIONS
6 PERSPECTIVES
6.1 Gas Sparging
6.2 Volatilisation of Fermentation Products by Gas Stripping
6.3 Acidogenic Biofilm
6.4 Gas Controlled Acidogenic Biofilm Reactor
7 REFERENCES LIST 

GET THE COMPLETE PROJECT

Related Posts