bioengineering

Review
Optimization and Scale-Up of Fermentation Processes Driven
by Models
Yuan-Hang Du 1,† , Min-Yu Wang 2,† , Lin-Hui Yang 1 , Ling-Ling Tong 1 , Dong-Sheng Guo 1, * and Xiao-Jun Ji 2, *

1 School of Food Science and Pharmaceutical Engineering, Nanjing Normal University, Nanjing 210023, China
2 State Key Laboratory of Materials-Oriented Chemical Engineering, College of Biotechnology and
Pharmaceutical Engineering, Nanjing Tech University, Nanjing 211816, China
* Correspondence: [email protected] (D.-S.G.); [email protected] (X.-J.J.)
† These authors contributed equally to this work.

Abstract: In the era of sustainable development, the use of cell factories to produce various com-
pounds by fermentation has attracted extensive attention; however, industrial fermentation requires
not only efficient production strains, but also suitable extracellular conditions and medium compo-
nents, as well as scaling-up. In this regard, the use of biological models has received much attention,
and this review will provide guidance for the rapid selection of biological models. This paper first
introduces two mechanistic modeling methods, kinetic modeling and constraint-based modeling
(CBM), and generalizes their applications in practice. Next, we review data-driven modeling based
on machine learning (ML), and highlight the application scope of different learning algorithms. The
combined use of ML and CBM for constructing hybrid models is further discussed. At the end, we
also discuss the recent strategies for predicting bioreactor scale-up and culture behavior through a
combination of biological models and computational fluid dynamics (CFD) models.

Keywords: mechanistic modeling; data-driven; hybrid modeling; scale-up; computational fluid

Citation: Du, Y.-H.; Wang, M.-Y.; dynamics
Yang, L.-H.; Tong, L.-L.; Guo, D.-S.; Ji,
X.-J. Optimization and Scale-Up of
Fermentation Processes Driven by
Models. Bioengineering 2022, 9, 473. 1. Introduction
https://doi.org/10.3390/
With the increasing consumption of fossil fuels and related environmental issues,
bioengineering9090473
there is an urgent need to find biological substitutes for traditional petrochemical prod-
Academic Editors: Guan Wang ucts through green biological manufacturing. Over the decades, microorganisms have
and Cees Haringa been used as ‘mini-factories’ in biomanufacturing to aid with the diversity of metabolic
Received: 5 August 2022
pathways and their accompanying ability to transform a wide range of renewable raw
Accepted: 9 September 2022
materials into value-added compounds by means of fermentation [1–3]; however, the
Published: 14 September 2022
use of wild-type microorganisms in industrial production is usually affected by many
factors, such as substrate and product toxicity [4,5]. Biologists are provided with many
Publisher’s Note: MDPI stays neutral
modern biotechnologies, such as genetic engineering and synthetic biology, to engineer
with regard to jurisdictional claims in
more powerful mini-factories [6–10]. Successful examples include Escherichia coli, which
published maps and institutional affil-
produce insulin and carotene [11,12], Saccharomyces cerevisiae, which produce geraniol [13],
iations.
Yarrowia lipolytica, which produce N-acetylneuraminic acid [14], Bacillus subtilis, which
produce hyaluronic acid [15], and so on; however, the construction of mini-factories is not
the end goal of biomanufacturing—biological fermentation and industrial production are
Copyright: © 2022 by the authors.
the ultimate goal, which is not a simple task. The engineered strains may not perform
Licensee MDPI, Basel, Switzerland. well in the actual fermentation process due to the lack of suitable fermentation strategies;
This article is an open access article therefore, optimizing fermentation parameters (such as the medium composition and extra-
distributed under the terms and cellular conditions) is a key factor in terms of running the mini-factories efficiently, which
conditions of the Creative Commons is crucial for the process of fermentation [16–18]. In practice, the rich insights generated
Attribution (CC BY) license (https:// by mathematical modeling can assist in the optimization of fermentation processes [19].
creativecommons.org/licenses/by/ Mathematical models, as approximations of reality, can clearly represent fermentation
4.0/). processes whose intrinsic complexity exceeds intuitive understanding, thus providing

Bioengineering 2022, 9, 473. https://doi.org/10.3390/bioengineering9090473 https://www.mdpi.com/journal/bioengineering

Bioengineering 2022, 9, 473 2 of 18

indispensable insight into designing, controlling, and optimizing the process, as well as
minimizing unnecessary experimentation [20,21].
However, it is not easy to model the fermentation process because each cell in the
bioreactor can be viewed as a subsystem of metabolic and signaling networks [22,23].
For fermentation problems, three modeling approaches are generally used: mechanistic
modeling, data-driven modeling, and hybrid modeling [20,24]. The mechanistic modeling
approaches derive the models from prior knowledge using equations that are notable and
acknowledged. Mechanistic models can extract valuable information from the raw data,
and they provide insight into the underlying mechanisms [25]. Kinetics and constraint-
based modeling (CBM) are the two primary mechanistic approaches for analyzing microbial
growth and metabolism [26]. In contrast, the data-driven approaches obtain a model by
analyzing and fitting existing data. Data-driven models are also known as black-box models
because they cannot provide information about the basic mechanisms without considering
their internal structures and phenomena [27]. Machine learning (ML) is a commonly used
data-driven approach. With the advancement of omics technology and various analytical
techniques, the datasets available for fermentation process modeling are rapidly growing,
and academics favor using ML to interpret large-scale datasets for deeper analysis and
optimization [28]. As a result of such circumstances, hybrid modeling has come into
being, which refers to the integration of mechanistic modeling and data-driven approaches.
Recently, comprehensive reviews of hybrid models have been published, indicating that
they are a promising prospect for this field [29,30].
In addition to the complexity of microbial metabolic behavior, fermentation systems
also have complex hierarchical structures. These systems consist of microorganisms and
a fermentation environment, and they are influenced by upstream and downstream op-
erating conditions. Moreover, during industrialization, the expansion of the bioreactor
volume and changes in shape lead to changes in the fermentation environment, which
subsequently leads to the failure of fermentation strategies that are developed during
the laboratory stage [31]; therefore, as biological fermentation shifts from laboratory to
industrial production, we also need to introduce fermentation environmental changes
into biological models to elucidate the effects of mixing and hydrodynamics [32,33]. This
goal can be achieved by coupling biological models with computational fluid dynamics
(CFD) models.
This paper gives an overview of different mathematical modeling methods and their
applications in biological fermentation processes. We first introduce the basic forms of mech-
anistic models that can describe microbial metabolism using kinetic and CBM modeling
methods and their applications in biological fermentation processes. Next, we discuss
different approaches to building data-driven models using ML. The synergistic effect of
CBM and ML is further discussed. In the end, we highlight the coupling of biological
models with CFD models, which prompts the formation of model-based integrated tools
to successfully predict bioreactor scale-up and culture behavior during model-assisted
bioreactor operation design.

2. Methods and Applications of Mechanistic Modeling

Mechanistic modeling describes some, though not all, mechanisms of complex systems,
enabling model parameters to be measured or inferred. Kinetic models are a kind of com-
mon mechanistic model, which usually reveal the dynamic changes between metabolites by
means of kinetic laws expressed by ordinary differential equations (ODEs) [34]; however,
for kinetic models, the mechanism of gene regulation underlying biological phenomena is
usually only explained by kinetic parameters in model equations (e.g., enzyme constants
and metabolite concentrations), rather than a mechanistic description of gene action [35],
and therefore, kinetic modeling is generally applied to fermentation problems in which the
metabolic behavior is well understood. CBM is another mechanistic modeling approach.
CBM studies the behavior of genome-wide systems to elucidate the relationship between
Bioengineering 2022, 9, 473 3 of 18

genotypes, phenotypes, and environmental conditions [27]. Next, we will discuss these two
mechanistic modeling methods and their application to fermentation problems in detail.

2.1. Application of Kinetic Modeling to Fermentation Processes

The derivation of kinetic models depends mainly upon the level of detail in which
microbial growth and metabolic behaviors are described. From a macro perspective, the
fermentation process in a bioreactor is a chemical process affected by physical and chemical
environmental factors (such as temperature, pH, aeration, and substrate concentration).
The transformation of substrate S to product P is catalyzed by bacterial pellets X, which
may be seen as a box containing equally disseminated catalysts and chemicals. In this case,
we do not need to pay attention to the internal structure of bacterial pellets; we only need
to pay attention to the changes in substances and external environmental factors before
and after catalysis. Thus, the kinetic models provide a macro perspective on the bioreac-
tor dynamics without specifying compositional and structural details under reasonable
assumptions. The kinetic models constructed in this case can also be called unstructured ki-
netic models. When the substrate concentration (S) is the only limiting element, the Monod
model is established. Different unstructured kinetic models of fermentation processes
are established with different descriptions of constraints and common kinetic equations, as
shown in Table 1. In practice, biologists construct kinetic models by adjusting and combining
different equations according to the actual situation. For instance, Zhang et al., established
a new kinetic model based on Monod through parameter optimization, which not only
described the relationship between products and by-products in the fermentation process
of 1, 3-propylene glycol, but it also guided the optimization of the fermentation conditions
and improved the yield [36]. Moreover, Garnier et al. developed an analytical solution
that combined the Monod model with the Luedeking–Piret equation which can be used
to simulate the batch fermentation process [37]. Although these models address the least
complex portrayal of cellular behavior, they can prove particularly valuable when quan-
tifying the phenomenon of growth inhibition during early process development. On the
other hand, unstructured kinetic models generally contain few parameters, which, from
a computational perspective, is usually advantageous. At present, a significant portion
of recent studies concern bioethanol production [38–40], biosurfactant production [41,42],
bacterial and fungal biomass production [43,44], as well as microalgal growth [45–47].

Table 1. General macroscopic kinetic equations.

Name Expression Function Refs.

[S] To describe microbial growth based on the
Monod Kinetics µ = µmax K +[S] [48]
S consumption of one substrate.
[S ][S ] To describe microbial growth based on the
Double Monod Kinetics µ = µmax (K +[S 1])(K2 +[S ]) [49]
s1 1 s2 2 consumption of multiple substrates.
To describe microbial growth in the presence of
Enzyme inhibition Kinetics µ = µmax  [S]  [50]
Ks 1+ KI +[S] competitive substrate inhibition.
I

[S] To describe microbial growth in a high-density

Contois Kinetics µ = µmax K [S ]+[S] [51]
S 0 culture.
To describe microbial growth while considering
Powell Kinetics [S] the basal metabolic consumption of cells (e.g., [52]
µ = (µmax + m) K +[S] − m
S
metabolite turnover).
To describe microbial growth in situations where
Moser Kinetics [S]n cells have multiple pathways to utilize [53]
µ = µmax K +[S]n
S
substrates.
To describe microbial growth without any
[X] biological explanation other than the assumption
Logistic Equation µ = µmax (1 − KX )
[41,42]
that there is a maximum cell growth
concentration.
Bioengineering 2022, 9, 473 4 of 18

Table 1. Cont.

Name Expression Function Refs.

To describe microbial growth while considering
Haldane–Andrew Model [S] that some substrates are toxic to cells and can [39,52]
µ = µmax 2
Ks +[S]+ [SK] inhibit cell growth at high concentrations.
i
To describe microbial growth while considering
[S ]
µ = µm1 K +[1 S ] + that there are two carbon sources, S1 and S2,
Diauxic Growth S1 1 [54,55]
µm2
[ S2 ] during cell growth and that the cell preferentially
[ S ]2
KS2 +[S2 ]+ 2
Ki
uses S1.
To describe the production rate of product P in
Luedeking–Piret Equation dP
dt = α d[dtX ] + β[ X ] the case where product synthesis is related to the [54,55]
growth rate and cell density of microbial cells.
µ: the specific growth rate of a microorganism; µmax : the maximum specific growth rate of a microorganism; KS :
the Monod constant; [S]: substrate concentration; [S0 ]: initial substrate concentration; [X]: biomass concentration;
I: competitive inhibitor; K I : the Monod constant of competitive inhibitor; m: a term of specific maintenance rate;
K I : the inhibition constant equal to the highest substrate concentration [S] when µ = 0.5µmax ; n: number of binding
sites of enzyme to substrate S.

As opposed to unstructured kinetic models, structured kinetic models consider the

physiological state of the cell and treat bacterial pellets as multicomponent entities with in-
ternal structures. Structure, in structured kinetic models, can be introduced by lumping the
metabolites into distinct intracellular pools [25]. Wang et al. constructed a structured kinetic
model that can be used to dynamically reflect changes in microalgal biomass by analyzing
how carbon is allocated to different intracellular compartments under nitrogen-poor and
nitrogen-rich extracellular conditions [56]. In a similar manner, Haringa et al. constructed
a structured kinetic model describing the growth of Penicillium chrysogenum, as well as
product formation, by grouping the most important intracellular metabolites into five pools
and four intracellular enzyme pools [57]. Their research also proved that the structured
models have a higher accuracy and wider application scope than the unstructured models.
We believe that our improved understanding of the behavior of microbial cells, along with
advances in various omics techniques, will facilitate the further development of structured
kinetic modeling.

2.2. Application of Constraint-Based Modeling to Fermentation Processes

CBM is another important mechanistic modeling method that considers the underly-
ing metabolic mechanisms of microorganisms. CBM can be used to analyze the dynamic
characteristics of specific target metabolites, and then implement fermentation strategy opti-
mization from the micro-level, which can enhance the diversity and pertinence of optimiza-
tion target design [58]. To build a CBM model, an in-depth understanding of metabolism is
first required. Metabolism represents all the biochemical reactions in cells. These reactions
form a coordinated set of metabolic pathways that convert substrates into the metabolites
required for the physiological activities of the cells at a rate known as metabolic flux. In
addition to being a basic determinant of a cellular metabolic state and physiology, metabolic
flux is also the key parameter in quantifying metabolic networks [59]. CBM plays an im-
portant role in the metabolic modeling of quantitative metabolic flux [60–62]. Metabolic
flux analysis (MFA) and Flux balance analysis (FBA) are the two primary CBM strategies
for quantifying metabolic flux (Figure 1), which can be implemented by COnstraint-Based
Reconstruction and Analysis (COBRApy) in Python [26,63].
Bioengineering 2022, 9, 473 5 of 18

Figure 1. Schematic of constraint-based modeling (CBM) methods.

2.2.1. Flux Balance Analysis

Flux balance analysis (FBA) is a mathematical metabolic flux technique that analyzes
the flux of metabolites through a detailed metabolic network (e.g., genomic scale) [64]. One
of the earliest applications of FBA was used to calculate the given network model and
substrate to obtain their maximum theoretical yields of metabolic products (Table 2) [64,65].
Papoutsakis et al. first demonstrated this in 1984 by describing the interrelationship
between various products and biomass in butyrate fermentation using FBA, accurately
predicting the theoretical yield of the several fermentation products of butyrate bacteria [66].
The same method can be utilized to work out the maximum biomass yield. For instance,
Varma and Palsson used FBA to quantitatively anticipate the maximum cell growth rate
cell density of wild-type E. coli W3110 [67]. In their study, they also used FBA to predict
the temporal distribution of glucose as well as the concentration of by-products. Another
important application of FBA focuses on the prediction of biomass production under
different substrate conditions, which aims to improve medium formulation and/or medium
feeding strategies [63]. Swayambhu et al. identified amino acids and carbon sources with
significant effects on yield, and they further improved the yield of target compounds
in recombinant E. coli by using medium optimization through FBA [68]. Kaushal et al.
used FBA to study the effect of medium components on the metabolic processes in the
simultaneous production of ethanol and butanol by Clostridium sporogenes NCIM 2918,
which provided ideas for the optimization of butyric acid and ethanol fermentation [69].
Huang et al. also used FBA to compare the catabolism of Chinese hamster ovary (CHO)
cells under different feed conditions in order to optimize medium formulations and increase
antibody production [70]; however, FBA also has disadvantages, such as poor performance
in predicting metabolic fluxes and growth phenotypes of engineered strains. In their study,
Long and Antoniewic et al. demonstrated that FBA is a poor predictor of growth rate and
metabolic flux in knockout strains [71,72]. Several complementary algorithmic approaches
have been created to aid the likelihood of foreseeing the flux of genetically engineered
strains, including regulatory on/off minimization (ROOM) [73], minimization of metabolic
adjustment (MOMA) [74], and the relative change (RELATCH) algorithms [75]; MFA can
compensate for this shortcoming.
Bioengineering 2022, 9, 473 6 of 18

2.2.2. Metabolic Flux Analysis

Metabolic flux analysis (MFA) is an imaging flux omics technique that determines
metabolic flux distributions by analyzing metabolic product production and consump-
tion rates in biological systems [76]. Different from FBA, MFA focuses on studying the
metabolic flux in cells under different environmental conditions without paying attention to
the theoretical optimal solution [76]. For instance, Calik et al. used MFA to study the effect
of pH on the metabolic flux of Bacillus licheniformis, and they proposed a pH manipulation
strategy to improve the yield of β-lactamase [77]. Sarma et al. used MFA to conduct in-depth
research into the enhancing effect of ultrasound-induced biohydrogen production, and they
provided ideas for strategy optimization [78]. Xiong et al. also used MFA to quantitatively
analyze metabolic flux during the production of L-lactate from glucose in Lactobacillus, and
they proposed a temperature control strategy that could maximize the yield of L-lactate [79].
Another common application of MFA is to analyze the production of key cellular co-factors
under different growth conditions. Li et al. used MFA to analyze the uneconomical utiliza-
tion of ATP in soybean oil feeding Acremonium chrysogenum, and they proposed strategies
to improve cephalosporin C (CPC) based on controlling NADPH and ATP production and
consumption [80]. In addition, MFA can predict changes in metabolic flux in metabolically
engineered strains, which provides help when selecting appropriate metabolic engineering
strategies [81]. At present, MFA has been successfully used in the performance optimiza-
tion of many industrial strains. [82–86]. Of course, MFA also has its disadvantages. One
limitation is that the metabolic network used for MFA analysis often needs to be reasonably
simplified; thus, some scholars have proposed optimization algorithms based on MFA,
such as flux variability analysis (FVA), which not only improve the accuracy of MFA, but
also broaden the application range of MFA [87,88].

2.2.3. Dynamic Flux Balance Analysis

It is worth noting that MFA and FBA consider the external environment to be in a
steady state, and they ignore the kinetics of enzymatic reactions during the modeling pro-
cess. Dynamic flux balance analysis (DFBA) was produced by considering the relationship
between the macro-state parameters of the fermentation process and the number of cell
physiological metabolic parameters [89]. Recent studies include predicting shikimic acid
production in E. coli [90], ethanol production in Saccharomyces cerevisiae [91], and the over-
production of secretory proteins in Streptomyces lividans [92]. DFBA can be constructed in
conjunction with macro kinetics; for example, Henson et al. combined the Monod equation
for substrate absorption with FBA to account for missing regulatory mechanisms [93]. In
addition to being combined with kinetics equations, DFBA can also be implemented using
ML in the form of hybrid modeling [35].

Table 2. Overview of CBM applications for the analysis and optimization of the fermentation
parameters demonstrated in this review.

Parameter Approach Case Refs.

The relationship between various products and biomass in the
process of butyric acid fermentation was described, and the
Theoretical maximum FBA [66]
theoretical yield of several fermentation products of butyric acid
bacteria was predicted accurately.
Quantitative prediction of maximum cell growth rate and cell
Theoretical maximum FBA [67]
density of wild-type E. coli W3110 were clarified.
Amino acids and carbon sources that have a significant influence on
Culture medium FBA the yield were identified, and the yield of siderophore compounds [68]
in recombinant E. coli was improved by medium optimization.
The effects of glucose, glycerol, and the mixture of glucose and
glycerol on the distribution of carbon flux in the simultaneous
Culture medium FBA [69]
production of ethanol and butanol by Clostridium sporogenes NCIM
2918 were studied.
Bioengineering 2022, 9, 473 7 of 18

Table 2. Cont.

Parameter Approach Case Refs.

The effects of amino acid composition in a culture medium on the
catabolism of Chinese hamster ovary (CHO) cells were analyzed to
Culture medium FBA [70]
optimize culture medium formulation and increase antibody
production.
By analyzing the effect of pH on the intracellular metabolic network
pH MFA of β-lactamase producing Bacillus licheniformis, a pH manipulation [77]
strategy was proposed to improve the yield of β -lactamase.
The effect of ultrasound promoting biological hydrogen production
from glycerol fermentation was understood to a significant extent,
Ultrasound MFA and an optimal strategy of enhancing glycerol uptake and blocking [78]
the butyric acid pathway under the guidance of the MFA model
was proposed.
By quantifying the flux during l-lactic acid production from
Temperature MFA glucose, a temperature control strategy was proposed to maximize [79]
the productivity of L-lactic acid.

3. Methods and Applications of Data-Driven Modeling

To fully describe the nonlinear changes of the fermentation process, more parameters
and state variables need to be used to build the mechanistic models [94,95]. Building such
a model is not always a viable option because parameter determination requires complex
experiments and large calculations [35]. Fortunately, advances in computer technology
have facilitated the emergence of data-driven modeling, which can provide a suitable
solution to this problem [96,97]. Figure 2 illustrates the modeling process for ML, a popular
data-driven modeling technique. ML uses mathematical knowledge such as statistical
theory to analyze datasets and find hidden relationships between existing data to justify
phenomena [96]. Unlike mechanistic models, data-driven models created by ML are built on
mathematical expressions, and model parameters have no physical, chemical, or biological
significance. [28]. Data-driven models reflect mappings between data rather than causal
relationships, so they tend to perform poorly when conditions change. Nonetheless, if
relevant data is readily available, data-driven modeling can save a lot of time compared
with mechanistic modeling [27]. Depending on the algorithm, ML can be categorized into
two groups: supervised ML and unsupervised ML.

Figure 2. Data-driven modeling steps.

3.1. Supervised Machine Learning

The data points for a supervised learning algorithm each have an associated feature
(input) value and a matching target (output) value. The feature value is the original
measurement data (e.g., temperature, pH), and the target value is mainly composed of the
reference data of process variables (e.g., productivity); therefore, each data point includes
Bioengineering 2022, 9, 473 8 of 18

the numerical values or classes for each feature and desired value. The goal of a supervised
learning algorithm is to mine the correlation between the feature and the target values in
the dataset, and then predict the target value by using the new input feature value or find
the optimal feature value by using the required target value [96].
The support vector machines algorithm (SVM) is a common supervised learning al-
gorithm and can be trained using various packages, including Scikit-learn [98] in Python,
Caret [99] in R, and MLJ [100] in Julia. The SVM can realize data classification and regres-
sion, and it can determine the best operating conditions by integrating with orthogonal
experimental design [101]. Dong et al. analyzed the comprehensive effects of corn stalk
weight, ultrasonic duration time, acoustic frequency, and alkali pretreatment time on biogas
production, based on the SVM model [102]. Under the guidance of the model, they further
obtained the optimal operating conditions and improved the gas production and efficiency
of corn straw anaerobic fermentation. Zhang et al. also optimized the fermentation con-
ditions of Rhodotorula glutinis using ethanol wastewater to produce biological lipids by
constructing a SVM model, and they improved the biomass and lipid value yield [103].
In addition, the SVM is often used to classify and annotate the structure and function of
proteins and genes due to its powerful regression and classification functions [104–106].
Artificial neural networks (ANNs) are another supervised learning algorithm that performs
better than SVM when the dataset is large [107,108]. The software programs PyTorch [109]
and Tensorflow [110] are frequently used to train neural networks. At present, ANNs have
been successfully used in many studies on fermentation prediction and optimization. For
example, Melcher et al. built a batch fermentation model of E. coli using ANNs, and they
successfully predicted the concentration of biomass and recombinant protein [111]. In a
recent study, Wang et al. also predicted the bio-hydrogen fermentation conditions and
analyzed the effect of critical parameters during fermentation by combining ANNs and a
response surface methodology [112]. ANNs also show great potential during fermentation
process control. Recently, Tavasoli et al. used an ANN to dynamically regulate the fed-batch
fermentation of recombinant Pichia pastoris, which significantly increased the yield of alpha
1-antitrypsin [113]. In addition, ANNs can also be used to assist processing massive omics
data and parsing microbial metabolic mechanisms [114–116].

3.2. Unsupervised Machine Learning

Unsupervised learning algorithms, which aim to find hidden relationships and clusters,
or to detect outliers, are suitable for datasets consisting of data points with only feature
values [24]. Common unsupervised learning algorithms include dimensionality reduction
algorithms and clustering algorithms [96].
The dimensionality reduction algorithms delete some unimportant data, and they dis-
cover the hidden relationship between data while preserving as much of the relationships
between the data points as possible [96]. Specific methods include principal component
analysis (PCA), principal component regression (PCR), and partial least squares regression
(PLSR) [117]. At present, the algorithms have been widely used in the analysis of fermenta-
tion process. For instance, Gutiérrez-González et al. performed PCA to test whether there
was any relationship between protein expression and favored environmental factors [118].
The results showed that temperature and reaction time had different effects on soluble and
insoluble proteins. As another unsupervised learning algorithm, the clustering algorithm
is widely used in biology. Arian et al. used the k-nearest neighbor algorithm, which is a
clustering algorithm, to cluster protein kinases according to their activity [119]. In addition,
unsupervised learning can also be used as a data preprocessing method during supervised
learning (e.g., dimensionality reduction for a dataset used for supervised learning or the
creation of category features).

4. Hybrid Models and Modeling Methods

Even for the most intensively studied model organisms, developing comprehensive
mechanistic models is still an ambitious goal. It is an ideal solution to use ML’s powerful data
Bioengineering 2022, 9, 473 9 of 18

processing and analysis capabilities to construct CBM models for elucidating metabolic
mechanisms, so hybrid models that combine data-driven and mechanistic models are
expected. Hybrid models have attracted much attention because they combine the ac-
curacy of mechanistic models constructed using CBM with the efficiency of data-driven
models constructed using ML [120,121]. There are three main methods for ML and CBM
integration [122] (Figure 3):

Figure 3. Different forms of CBM and ML integration. (a) Processing omics data and predicting
parameters using ML. (b) Obtaining more biological insights from the metabolic flux date using ML.
(c) Processing omics data using ML, which is then used as input data to construct CBM.

(a) CBM models were constructed using ML by merging and analyzing omics data
from different sources, whereas CBM was trained and reassembled by obtaining
genomic data under specific conditions [123]. This method is suitable for situations
wherein mechanistic models are not accurate enough. Vijayakumar et al. used ML
to analyze RNA sequencing data extracted under 23 different growth conditions,
which was then combined with flux data obtained by using FBA to elucidate the
mechanisms underlying cyanobacterial responses to fluctuations in light intensity
and salinity [124]. The growth rates of yeasts, such as S. cerevisiae, have also been
predicted using this technique. For example, Culley et al. obtained reliable results for
the growth rate prediction of S. cerevisiae by combining CBM-derived flux omics data
with transcriptomics using ANNs [116].
(b) Metabolic flux data obtained from CBM was trained by the ML method to gain
more biological insights into the required system [125]. With this method, potential
phenomena that cannot be mechanistically described can be analyzed. For example,
Sridhara et al. used ML to analyze the metabolic flux data generated by CBM, and
they realized that the retroversion of the culture medium components, which occurred
during bacterial growth, could not be achieved using CBM alone [126].
(c) ML can be used to analyze multi-omics data so as to provide data preprocessing ser-
vices for CBM model construction. In 2016, Wu et al. used the ML method to analyze
and integrate heterotrophic bacterial metabolic data from about 100 papers, finally
constructing MFlux, a Web-based platform that can analyze metabolic fluxes [127].
As stated above, more successful mechanistic models describing metabolic dynamics
can be created with a thorough understanding of microbial metabolism and the devel-
opment of CBM and ML methods [128]; however, in order to accurately describe the
fermentation process and predict the fermentation results, in addition to the metabolic
models of the cells, the fermentation environment in the bioreactor also needs to be consid-
ered [25].
Bioengineering 2022, 9, 473 10 of 18

5. Coupling Biological Models with Computational Fluid Dynamics Models Enabling

Rational Fermentation Scale-Up
The physiology of microorganisms and the flow field in the bioreactor are two im-
portant issues that are highly interrelated throughout the fermentation process, and they
affect the final fermentation yield. The exploration into ideal genetic engineering mod-
ification strategies and fermentation conditions based on mechanistic and data-driven
models enable efficient biological production on a laboratory scale; however, scaling-up
fermentation processes from the laboratory to industrial production level requires more
work. In industrial bioreactors, the spatiotemporal gradients of substrate, temperature,
and pH can affect the metabolism of the resulting strains [39,129]. In addition, strong
agitation in industrial bioreactors may also cause shear damage to cells and reduce their
viability [130]. These factors lead to a loss of performance in scale-up operations, including
a reduction in titer, yield, or productivity, as well as potential pitfalls for process monitoring
and product quality control [131–133]; therefore, to narrow the gap from theory to practice,
and to apply biological models in order to aid scaling-up production, we need to consider
the impact of external environmental changes on microorganisms during modeling. CFD is
a combination of mathematics, fluid mechanics, and computer technology that can be used
to generate detailed flow field information in a bioreactor [134–136]; therefore, coupling
the flow field information obtained by CFD models with the cell metabolism information
provided by biological models can help to clarify how turbulence and environmental gradi-
ents occur, and ultimately, how they affect cell performance, which will provide guidance
for bioreactor design and scaling-up production (Figure 4) [137]. For the purposes of our
analysis, the biomaterial is referred to as the “biotic phase” and the culture environment is
referred to as the “abiotic phase”. In the process of coupling biological models with CFD
models, there are two main ways according to the different treatment methods of biotic
phase: Euler–Lagrange methodology (ELM) and Euler–Euler methodology (EEM) (Table 3).

Figure 4. Schematic of metabolic models and computational fluid dynamics coupling.

The ELM considers the biotic phase to be the period wherein discrete particles shuttle
through the continuous abiotic phase, calculating each cell’s time-dependent positions
in the main field in order to track their movement and state [137,138]. The ELM is suit-
able for coupling structured kinetic models in order to explain the differences between cell
metabolisms, which are caused by environmental differences [139]. The ELM can be used
to quantitatively analyze the effects of the concentration gradient on microorganism cul-
tures [133]. For example, Siebler et al. investigated the effect of different concentration
Bioengineering 2022, 9, 473 11 of 18

field residence times on transcriptional changes in Clostridium ljungdahlii cells under stress
conditions during syngas fermentation [140]. Haringa et al. used CFD simulations of in-
dustrial bioreactors in conjunction with structured kinetic models to evaluate the impact of
glucose concentration gradients on penicillin production [57]. Kuschel et al. also predicted
differences between the metabolic statuses of cell populations due to glucose concentration
gradients using the ELM [141]. For microalgae cultures, light intensity is an important
factor. The growth of microalgae can be better predicted by tracking the light changes
experienced by each cell, which can also be achieved using the ELM [142]. Gernigon et al.
used the ELM to simulate the light distribution in two photobioreactors with different
structures to analyze the influence of light on microalgae culture [143]. The results show
that different light distributions can lead to different growth rates in microalgae, even when
using the same light intensity; however, the ELM is computationally expensive and difficult
to be applied on an industrial scale because it requires calculating change in each particle.
Another CFD-based method for modeling bioreactors is the EEM, particularly with
regard to bioreactors that have larger sizes and higher volume fractions of solids. The EEM
treats the biotic phase and abiotic phase as co-existing interpenetrating continua. The EEM
is much less computationally expensive than the ELM because the state of each particle
is not calculated separately [144]. Elqotbi et al. incorporated the Monod equation for
Aspergillus Niger gluconic acid production into CFD using the EEM, and they demonstrated
that an increase in viscosity resulted in a decrease in oxygen mass transfer, which further
affected gluconic acid production [145]. Morchain et al. used the EEM to simulate a
fermentation process in a 70 L bioreactor and an industrial-scale 70 m3 bioreactor to explain
the performance degradation of industrial bioreactors that mix poorly [146]. Du et al. used
the EEM to establish a CFD model coupled with an unstructured kinetic model to explore
the effect of bioreactor size and operating conditions on the fermentation process, and they
were able to scale-up DHA fermentation from a 5 L bioreactor to a 35 m3 bioreactor [49].
Jing et al. also built a coupling model to fine-tune the biological production process to a
reasonable extent, and they were able to scale-up the production of ferulic acid into vanillin,
from shaker to bioreactor, with a conversion rate up to 94% [147].

Table 3. Overview of the applications of different kinetics and computational fluid dynamics coupling
frameworks used during fermentation.

Approach Application Refs.

The transcriptional changes of Clostridium ljungdahlii cells subjected to CO restriction in
ELM a 125 m3 bubble column bioreactor was predicted, which guided the scaling-up of [140]
production.
The decrease in penicillin production when using P. chrysogenum due to glucose
ELM [57]
gradient in a 54 m3 stirred tank reactor was predicted.
ELM The formation of population heterogeneity in E. coli in a 54 m3 bioreactor was predicted. [141]
The difference in microalgae biomass in different photoreactors caused by different light
ELM [143]
distributions was predicted.
The reason for the decrease in the gluconic acid yield during the production of gluconic
EEM acid by Aspergillus Niger was revealed, which was due to the decrease in oxygen mass [145]
transfer due to the increase in medium viscosity during fermentation.
The performance degradation of the industrial bioreactor under poor mixing conditions
EEM was explained by comparing the flow field environment of the laboratory bioreactor (70 [146]
L) with that of the industrial (70 m3 ) bioreactor.
The effects of the size of the bioreactor and the operating conditions on DHA
EEM [49]
fermentation were predicted, and DHA fermentation was scaled up from 5 L to 35 m3 .
The biological production process was fine-tuned by coupling CFD and biokinetics, and
EEM the scale required to turn ferulic acid into vanillin (scaling it up from shaker to [147]
bioreactor) was realized, with a conversion rate up to 94%.

As previously indicated, computational frameworks that combine biological models

and CFD models can provide information on metabolic reactions induced by environmental
Bioengineering 2022, 9, 473 12 of 18

fluctuations in industrial bioreactors; thus, they can indicate what improvements to make
with regard to the design of bioreactors and the fermentation process. One of the main
obstacles in the adoption of coupled frameworks is the computing time required, but
technological advancements, especially in GPU-based computing, may make real-time
simulations feasible [137,148]. In the future, coupled frameworks will be combined with
visualizations, such as virtual reality, to allow operators to ‘see’ what’s going on inside
bioreactors [149–151].

6. Conclusions and Future Perspectives

Biomanufacturing is a research hotspot in the modern manufacturing industry, which
can transform renewable raw materials into value-added compounds through microbial
fermentation. In addition to developing engineered strains that can act as mini-factories,
optimizing the fermentation process and scaling-up production are also essential factors
in improving the productivity of biomanufacturing. Fortunately, today, several modeling
methods are available to build mathematical models of the fermentation process in order
to guide the development of strategies for improving fermentation efficiency. This paper
briefly summarizes two important modeling approaches for building fermentation models:
mechanistic modeling and data-driven modeling.
The earliest and most commonly used mechanistic models are kinetic models, which
can be used to describe the macroscopic dynamic characteristics of fermentation pro-
cess. At present, they still play an important role in biomanufacturing processes such as
bioethanol and surfactants. With the advancement of omics technology, biologists now
have a thorough understanding of how cells function, allowing them to build more micro-
scopically detailed mechanistic models, or so-called CBM models. In comparison to kinetic
models, CBM models provide a more comprehensive representation of complex cellular
factories, providing quantitative and more detailed predictions to aid in the development
of fermentation strategies. CBM can be subdivided into several specific methods, including
MFA, FBA, and DFBA, which can quantify metabolic flux distributions in cellular metabolic
networks. Unlike mechanistic models, data-driven models use mathematical knowledge
to analyze fermentation data in order to guide process optimization. The development
of modern computer technology, especially ML technology, has enriched the application
scenarios of data-driven models, which can now describe the fermentation process more
accurately, and can ignore the internal metabolism of cells. Data-driven models are widely
used in predicting fermentation results, predicting optimal fermentation conditions and
controlling processes. In addition, the development of omics technology and ML technol-
ogy also provides a boost for the development of mechanistic models. Many researchers
tend to use ML to interpret large-scale datasets and analyze metabolic networks, in addition
to constructing a hybrid model of ML and CBM synergism. Finally, an integrated tool
involving the coupling of biological models with CFD models, for model-based culture
behavior prediction and model-aided bioreactor operation design, to be used during the
process of scaling-up bioreactors, was discussed.
In our view, after collaborating with a range of disciplines to develop new mathemat-
ical and computational tools, while enhancing our knowledge of biology, the evolution
of fermentation process modeling could be limitless. In the future, it will be possible to
dynamically reflect fermentation changes using simulation models, and it will be possible
to regulate the fermentation process by digitally controlling it online.

Author Contributions: Conceptualization, X.-J.J. and D.-S.G.; writing—original draft preparation,

Y.-H.D., M.-Y.W., L.-H.Y. and L.-L.T.; writing—review and editing, X.-J.J. and D.-S.G. All authors have
read and agreed to the published version of the manuscript.
Funding: This work was financially supported by the National Key Research and Development
Program of China (2021YFC2101100), the National Science Fund for Excellent Young Scholars of
China (21922806), the Jiangsu Provincial Natural Science Foundation (No. BK20200732).
Institutional Review Board Statement: Not applicable.
Bioengineering 2022, 9, 473 13 of 18

Informed Consent Statement: Not applicable.

Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. European Technology and Innovation Platform: Bioenergy for Industry. Available online: https://etipbioenergy.eu/industry/
production-facilities (accessed on 30 July 2022).
2. The Engineering Chemical Technology Centre: Biodegradable Plastic and World Biopolymers Market 2019–2020. Available online:
https://ect-center.com/blog/biopolymers-market-2019 (accessed on 30 July 2022).
3. Du, J.; Shao, Z.; Zhao, H. Engineering Microbial Factories for Synthesis of Value-Added Products. J. Ind. Microbiol. Biotechnol.
2011, 38, 873–890. [CrossRef] [PubMed]
4. Ling, H.; Teo, W.; Chen, B.; Leong, S.; Chang, M. Microbial Tolerance Engineering toward Biochemical Production: From
Lignocellulose to Products. Curr. Opin. Biotechnol. 2014, 29, 99–106. [CrossRef] [PubMed]
5. Magocha, T.; Zabed, H.; Yang, M.; Yun, J.; Zhang, H.; Qi, X. Improvement of Industrially Important Microbial Strains by Genome
Shuffling: Current Status and Future Prospects. Bioresour. Technol. 2018, 257, 281–289. [CrossRef] [PubMed]
6. Ahmad, M.; Hirz, M.; Pichler, H.; Schwab, H. Protein Expression in Pichia Pastoris: Recent Achievements and Perspectives for
Heterologous Protein Production. Appl. Microbiol. Biotechnol. 2014, 98, 5301–5317. [CrossRef]
7. Sawant, A.M.; Vamkudoth, K.R. Biosynthetic Process and Strain Improvement Approaches for Industrial Penicillin Production.
Biotechnol. Lett. 2022, 44, 179–192. [CrossRef]
8. Favaro, L.; Jansen, T.; van Zyl, W.H. Exploring Industrial and Natural Saccharomyces Cerevisiae Strains for the Bio-Based Economy
from Biomass: The Case of Bioethanol. Crit. Rev. Biotechnol. 2019, 39, 800–816. [CrossRef]
9. Jansen, M.L.A.; Bracher, J.M.; Papapetridis, I.; Verhoeven, M.D.; de Bruijn, H.; de Waal, P.P.; van Maris, A.J.A.; Klaassen, P.;
Pronk, J.T. Saccharomyces Cerevisiae Strains for Second-Generation Ethanol Production: From Academic Exploration to Industrial
Implementation. FEMS Yeast Res. 2017, 17, fox044. [CrossRef]
10. Nielsen, J.; Keasling, J.D. Engineering Cellular Metabolism. Cell 2016, 164, 1185–1197. [CrossRef]
11. Wu, T.; Li, S.; Ye, L.; Zhao, D.; Fan, F.; Li, Q.; Zhang, B.; Bi, C.; Zhang, X. Engineering an Artificial Membrane Vesicle Trafficking
System (AMVTS) for the Excretion of β-Carotene in Escherichia coli. ACS Synth. Biol. 2019, 8, 1037–1046. [CrossRef]
12. Baeshen, N.A.; Baeshen, M.N.; Sheikh, A.; Bora, R.S.; Ahmed, M.M.M.; Ramadan, H.A.I.; Saini, K.S.; Redwan, E.M. Cell Factories
for Insulin Production. Microb. Cell Factories 2014, 13, 141. [CrossRef]
13. Jiang, G.Z.; Yao, M.D.; Wang, Y.; Zhou, L.; Song, T.Q.; Liu, H.; Xiao, W.H.; Yuan, Y.J. Manipulation of GES and ERG20 for Geraniol
Overproduction in Saccharomyces cerevisiae. Metab. Eng. 2017, 41, 57–66. [CrossRef]
14. Zhang, X.; Liu, Y.; Liu, L.; Wang, M.; Li, J.; Du, G.; Chen, J. Modular Pathway Engineering of Key Carbon-precursor Supply-
pathways for Improved N-acetylneuraminic Acid Production in Bacillus subtilis. Biotechnol. Bioeng. 2018, 115, 2217–2231.
[CrossRef]
15. Westbrook, A.W.; Ren, X.; Oh, J.; Moo-Young, M.; Chou, C.P. Metabolic Engineering to Enhance Heterologous Production of
Hyaluronic Acid in Bacillus subtilis. Metab. Eng. 2018, 47, 401–413. [CrossRef]
16. De, T.; Sikder, J.; Narayanan, C.M. Biodiesel Synthesis Using Immobilised Lipase Enzyme in Semi-fluidised Bed Bioreactors-
Bioreactor Design and Performance Analysis. Environ. Prog. Sustain. Energy 2017, 36, 1537–1545. [CrossRef]
17. Brindhadevi, K.; Shanmuganathan, R.; Pugazhendhi, A.; Gunasekar, P.; Manigandan, S. Biohydrogen Production using Horizontal
and Vertical Continuous Stirred Tank Reactor—A Numerical Optimization. Int. J. Hydrog. Energy 2021, 46, 11305–11312. [CrossRef]
18. Chu, R.Y.; Li, S.X.; Zhu, L.D.; Yin, Z.H.; Hu, D.; Liu, C.C.; Mo, F. A review on Co-cultivation of Microalgae with Filamentous
Fungi: Efficient Harvesting, Wastewater Treatment and Biofuel Production. Renew. Sust. Energy Rev. 2021, 139, 110689. [CrossRef]
19. Pistikopoulos, E.N.; Barbosa-Povoa, A.; Lee, J.H.; Misener, R.; Mitsos, A.; Reklaitis, G.V.; Venkatasubramanian, V.; You, F.; Gani, R.
Process Systems Engineering—The Generation Next? Comput. Chem. Eng. 2021, 147, 107252. [CrossRef]
20. Shih, W.; Chai, S. Data-Driven vs. Hypothesis-Driven Research: Making sense of big data. Acad. Manag. J. 2016, 2016, 14843.
[CrossRef]
21. Straathof, A.J.J.; Wahl, S.A.; Benjamin, K.R.; Takors, R.; Wierckx, N.; Noorman, H.J. Grand Research Challenges for Sustainable
Industrial Biotechnology. Trends Biotechnol. 2019, 37, 1042–1050. [CrossRef]
22. Xia, J.Y.; Wang, G.; Lin, J.H.; Wang, Y.H.; Chu, J.; Zhuang, Y.P.; Zhang, S.L. Advances and Practices of Bioprocess Scale-
up. In Bioreactor Engineering Research and Industrial Applications II; Bao, J., Ye, Q., Zhong, J.J., Eds.; Advances in Biochemical
Engineering-Biotechnology; Springer: Berlin/Heidelberg, Germany, 2016; Volume 152, pp. 137–151.
23. Neubauer, P.; Cruz, N.; Glauche, F.; Junne, S.; Knepper, A.; Raven, M. Consistent development of bioprocesses from microliter
cultures to the industrial scale. Eng. Life Sci. 2013, 13, 224–238. [CrossRef]
24. Solle, D.; Hitzmann, B.; Herwig, C.; Pereira Remelhe, M.; Ulonska, S.; Wuerth, L.; Prata, A.; Steckenreiter, T. Between the Poles of
Data-Driven and Mechanistic Modeling for Process Operation. Chem. Ing. Tech. 2017, 89, 542–561. [CrossRef]
25. Flevaris, K.; Chatzidoukas, C. Facilitating the industrial transition to microbial and microalgal factories through mechanistic
modelling within the Industry 4.0 paradigm. Curr. Opin. Chem. Eng. 2021, 33, 100713. [CrossRef]
Bioengineering 2022, 9, 473 14 of 18

26. Yasemi, M.; Jolicoeur, M. Modelling Cell Metabolism: A Review on Constraint-Based Steady-State and Kinetic Approaches.
Processes 2021, 9, 322. [CrossRef]
27. Khaleghi, M.K.; Savizi, I.S.P.; Lewis, N.E.; Shojaosadati, S.A. Synergisms of machine learning and constraint-based modeling of
metabolism for analysis and optimization of fermentation parameters. Biotechnol. J. 2021, 16, 2100212. [CrossRef]
28. Helmy, M.; Smith, D.; Selvarajoo, K. Systems biology approaches integrated with artificial intelligence for optimized metabolic
engineering. Metab. Eng. Commun. 2020, 11, e00149. [CrossRef]
29. Kim, Y.; Kim, G.B.; Lee, S.Y. Machine learning applications in genome-scale metabolic modeling. Curr. Opin. Struct. Biol. 2021, 25,
42–49. [CrossRef]
30. Antonakoudis, A.; Barbosa, R.; Kotidis, P.; Kontoravdi, C. The era of big data: Genome-scale modelling meets machine learning.
Comput. Struct. Biotechnol. J. 2020, 18, 3287–3300. [CrossRef]
31. Nadal-Rey, G.; McClure, D.D.; Kavanagh, J.M.; Cornelissen, S.; Fletcher, D.F.; Gernaey, K.V. Understanding gradients in industrial
bioreactors. Biotechnol. Adv. 2021, 46, 107660. [CrossRef]
32. Wang, G.; Haringa, C.; Tang, W.; Noorman, H.; Chu, J.; Zhuang, Y.; Zhang, S. Coupled metabolic-hydrodynamic modeling
enabling rational scale-up of industrial bioprocesses. Biotechnol. Bioeng. 2020, 117, 844–867. [CrossRef]
33. Joyce, A.R.; Palsson, B. The model organism as a system: Integrating ‘omics’ data sets. Nat. Rev. Mol. Cell Biol. 2006, 7, 198–210.
[CrossRef]
34. Kim, O.D.; Rocha, M.; Maia, P. A review of dynamic modeling approaches and their application in computational strain
optimization for metabolic engineering. Front. Microbiol. 2018, 9, 1690. [CrossRef] [PubMed]
35. Theodoropoulos, C.; Sun, C. 2.45—Bioreactor Models and Modeling Approaches. In Comprehensive Biotechnology, 3rd ed.;
Moo-Young, M., Ed.; Pergamon: Oxford, UK, 2019; pp. 663–680.
36. Zhang, A.H.; Huang, S.Y.; Zhuang, X.Y.; Wang, K.; Yao, C.Y.; Fang, B.S. A novel kinetic model to describe 1,3-propanediol
production fermentation by Clostridium butyricum. AIChE J. 2019, 65, e16587. [CrossRef]
37. Garnier, A.; Gaillet, B. Analytical solution of Luedeking–Piret equation for a batch fermentation obeying Monod growth kinetics.
Biotechnol. Bioeng. 2015, 112, 2468–2474. [CrossRef] [PubMed]
38. Bhowmik, S.K.; Alqahtani, R.T. Mathematical analysis of bioethanol production through continuous reactor with a settling unit.
Comput. Chem. Eng. 2018, 111, 241–251. [CrossRef]
39. Zentou, H.; Zainal Abidin, Z.; Yunus, R.; Awang Biak, D.R.; Zouanti, M.; Hassani, A. Modelling of Molasses Fermentation for
Bioethanol Production: A Comparative Investigation of Monod and Andrews Models Accuracy Assessment. Biomolecules 2019,
9, 308. [CrossRef]
40. Moodley, P.; Gueguim Kana, E.B. Bioethanol production from sugarcane leaf waste: Effect of various optimized pretreatments
and fermentation conditions on process kinetics. Biotechnol. Rep. 2019, 22, e00329. [CrossRef]
41. Ranjbar, S.; Hejazi, P. Modeling and validating Pseudomonas aeruginosa kinetic parameters based on simultaneous effect of bed
temperature and moisture content using lignocellulosic substrate in packed-bed bioreactor. Food Bioprod. Process. 2019, 117, 51–63.
[CrossRef]
42. Ray, M.; Kumar, V.; Banerjee, C. Kinetic modelling, production optimization, functional characterization and phyto-toxicity
evaluation of biosurfactant derived from crude oil biodegrading Pseudomonas sp. IITISM 19. J. Environ. Chem. Eng. 2022, 10,
107190. [CrossRef]
43. Beagan, N.; O’Connor, K.E.; Del Val, I.J. Model-based operational optimisation of a microbial bioprocess converting terephthalic
acid to biomass. Biochem. Eng. J. 2020, 158, 107576. [CrossRef]
44. Finkler, A.T.J.; de Lima Luz, L.F.; Krieger, N.; Mitchell, D.A.; Jorge, L.M. A model-based strategy for scaling-up traditional
packed-bed bioreactors for solid-state fermentation based on measurement of O2 uptake rates. Biochem. Eng. J. 2021, 166, 107854.
[CrossRef]
45. Feng, F.; Li, Y.; Latimer, B.; Zhang, C.; Nair, S.S.; Hu, Z. Prediction of maximum algal productivity in membrane bioreactors with
a light-dependent growth model. Sci. Total Environ. 2021, 753, 141922. [CrossRef]
46. Yu, L.; Li, T.T.; Ma, J.W.; Zhao, Q.B.; Wensel, P.; Lian, J.N.; Chen, S.L. A kinetic model of heterotrophic and mixotrophic cultivation
of the potential biofuel organism microalgae Chlorella sorokiniana. Algal Res. 2022, 64, 102701. [CrossRef]
47. Montenegro-Herrera, C.A.; Portillo, F.V.L.; Hernandez-Chavez, G.T.; Martinez, A. Single-cell protein production potential with the
extremophilic red microalgae Galdieria sulphuraria: Growth and biochemical characterization. J. Appl. Phycol. 2022, 34, 1341–1352.
[CrossRef]
48. Najafpour, G.; Younesi, H. Ethanol fermentation in an immobilized cell reactor using Saccharomyces cerevisiae. Bioresour. Technol.
2004, 92, 251–260. [CrossRef]
49. Du, Y.H.; Tong, L.L.; Wang, Y.; Liu, M.Z.; Yuan, L.; Mu, X.Y.; He, S.J.; Wei, S.X.; Zhang, Y.D.; Chen, Z.L.; et al. Development of a
kinetics-integrated CFD model for the industrial scale-up of DHA fermentation using Schizochytrium sp. AIChE J. 2022, 68, e17750.
[CrossRef]
50. Bezerra, R.M.F.; Dias, A.A. Enzymatic kinetic of cellulose hydrolysis—Inhibition by ethanol and cellobiose. Appl. Biochem.
Biotechnol. 2005, 126, 49–59. [CrossRef]
51. Dong, Y.M.; Yan, X.F.; Lu, F.; Guo, M.J.; Zhuang, Y.P. Development and Optimization of an Unstructured Kinetic Model for
Sodium Gluconate Fermentation Process. Comput. Mater. Contin. 2015, 48, 43–55. [CrossRef]
Bioengineering 2022, 9, 473 15 of 18

52. Costa, F.; Quintelas, C.; Tavares, T. An approach to the metabolic degradation of diethylketone (DEK) by Streptococcus equisimilis:
Effect of DEK on the growth, biodegradation kinetics and efficiency. Ecol. Eng. 2014, 70, 183–188. [CrossRef]
53. Ghoshal, G.; Banerjee, U.C.; Shivhare, U.S. Xylanase Production by Penicillium citrinum in Laboratory-scale Stirred Tank Reactor.
Chem. Biochem. Eng. Q. 2014, 28, 399–408. [CrossRef]
54. Buendia-Kandia, F.; Rondags, E.; Framboisier, X.; Mauviel, G.; Dufour, A.; Guedon, E. Diauxic growth of Clostridium acetobutylicum
ATCC 824 when grown on mixtures of glucose and cellobiose. AMB Express 2018, 8, 85. [CrossRef]
55. Sun, H.Y.; Zhang, X.Y.; Wang, D.L.; Lin, Z.F. Insights into the role of energy source in hormesis through diauxic growth of bacteria
in mixed cultivation systems. Chemosphere 2020, 261, 127669. [CrossRef]
56. Wang, D.; Lai, Y.C.; Karam, A.L.; de Los Reyes, F.L., 3rd; Ducoste, J.J. Dynamic modeling of microalgae growth and lipid
production under transient light and nitrogen conditions. Environ. Sci. Technol. 2019, 53, 11560–11568. [CrossRef]
57. Haringa, C.; Tang, W.; Wang, G.; Deshmukh, A.T.; van Winden, W.A.; Chu, J.; van Gulik, W.M.; Heijnen, J.J.; Mudde, R.F.;
Noorman, H.J. Computational fluid dynamics simulation of an industrial P. chrysogenum fermentation with a coupled 9-pool
metabolic model: Towards rational scale-down and design optimization. Chem. Eng. Sci. 2018, 175, 12–24. [CrossRef]
58. Cvijovic, M.; Bordel, S.; Nielsen, J. Mathematical models of cell factories: Moving towards the core of industrial biotechnology.
Microb. Biotechnol. 2011, 4, 572–584. [CrossRef]
59. Stephanopoulos, G. Metabolic fluxes and metabolic engineering. Metab. Eng. 1999, 1, 1–11. [CrossRef]
60. Bordbar, A.; Monk, J.M.; King, Z.A.; Palsson, B.O. Constraint-based models predict metabolic and associated cellular functions.
Nat. Rev. Genet. 2014, 15, 107–120. [CrossRef]
61. Ramon, C.; Gollub, M.G.; Stelling, J. Integrating -omics data into genome-scale metabolic network models: Principles and
challenges. Essays Biochem. 2018, 62, 563–574. [CrossRef]
62. Llaneras, F.; Pico, J. Stoichiometric modelling of cell metabolism. J. Biosci. Bioeng. 2008, 105, 1–11. [CrossRef]
63. Orth, J.D.; Thiele, I.; Palsson, B.O. What is flux balance analysis? Nat. Biotechnol. 2010, 28, 245–248. [CrossRef]
64. Varma, A.; Palsson, B.O. Metabolic flux balancing: Basic concepts, scientific and practical use. Nat. Biotechnol. 1994, 12, 994–998.
[CrossRef]
65. Schuetz, R.; Kuepfer, L.; Sauer, U. Systematic evaluation of objective functions for predicting intracellular fluxes in Escherichia coli.
Mol. Syst. Biol. 2007, 3, 119. [CrossRef] [PubMed]
66. Papoutsakis, E.T. Equations and calculations for fermentations of butyric acid bacteria. Biotechnol. Bioeng. 1984, 26, 174–187.
[CrossRef] [PubMed]
67. Varma, A.; Palsson, B.O. Stoichiometric flux balance models quantitatively predict growth and metabolic by-product secretion in
wild-type Escherichia coli W3110. Appl. Environ. Microbiol. 1994, 60, 3724–3731. [CrossRef] [PubMed]
68. Swayambhu, G.; Moscatello, N.; Atilla-Gokcumen, G.E.; Pfeifer, B.A. Flux balance analysis for media optimization and genetic
targets to improve heterologous siderophore production. iScience 2020, 23, 101016. [CrossRef]
69. Kaushal, M.; Chary, K.V.N.; Ahlawat, S.; Palabhanvi, B.; Goswami, G.; Das, D. Understanding regulation in substrate dependent
modulation of growth and production of alcohols in Clostridium sporogenes NCIM 2918 through metabolic network reconstruction
and flux balance analysis. Bioresour. Technol. 2018, 249, 767–776. [CrossRef]
70. Huang, Z.; Xu, J.; Yongky, A.; Morris, C.S.; Polanco, A.L.; Reily, M.; Borys, M.C.; Li, Z.J.; Yoon, S. CHO cell productivity
improvement by genome-scale modeling and pathway analysis: Application to feed supplements. Biochem. Eng. J. 2020,
160, 107638. [CrossRef]
71. Long, C.P.; Gonzalez, J.E.; Sandoval, N.R.; Antoniewicz, M.R. Characterization of physiological responses to 22 gene knockouts in
Escherichia coli central carbon metabolism. Metab. Eng. 2016, 37, 102–113. [CrossRef]
72. Long, C.P.; Antoniewicz, M.R. Metabolic flux responses to deletion of 20 core enzymes reveal flexibility and limits of E. coli
metabolism. Metab. Eng. 2019, 55, 249–257. [CrossRef]
73. Shlomi, T.; Berkman, O.; Ruppin, E. Regulatory on/off minimization of metabolic flux changes after genetic perturbations. Proc.
Natl. Acad. Sci. USA 2005, 102, 7695–7700. [CrossRef]
74. Segrè, D.; Vitkup, D.; Church, G.M. Analysis of optimality in natural and perturbed metabolic networks. Proc. Natl. Acad. Sci.
USA 2002, 99, 15112–15117. [CrossRef]
75. Kim, J.; Reed, J.L. RELATCH: Relative optimality in metabolic networks explains robust metabolic and regulatory responses to
perturbations. Genome Biol. 2012, 13, R78. [CrossRef]
76. Antoniewicz, M.R. A guide to metabolic flux analysis in metabolic engineering: Methods, tools and applications. Metab. Eng.
2021, 63, 2–12. [CrossRef]
77. Calik, P.; Ileri, N. PH influences intracellular reaction network of beta-lactamase producing Bacillus licheniformis. Chem. Eng. Sci.
2007, 62, 5206–5211. [CrossRef]
78. Sarma, S.; Anand, A.; Dubey, V.K.; Moholkar, V.S. Metabolic flux network analysis of hydrogen production from crude glycerol
by Clostridium pasteurianum. Bioresour. Technol. 2017, 242, 169–177. [CrossRef]
79. Xiong, X.J.; Li, Y.F. Study on temperature regulation of lactic acid bacteria using metabolic flux analysis. In Proceedings of the
International Conference on Energy, Environment and Sustainable Development (ICEESD 2011), Shanghai, China, 21–23 October
2011; pp. 127–132.
Bioengineering 2022, 9, 473 16 of 18

80. Li, J.H.; Yang, Y.M.; Chu, J.; Huang, M.Z.; Li, L.A.; Zhang, X.C.; Wang, Y.H.; Zhuang, Y.P.; Zhang, S.L. Quantitative metabolic
flux analysis revealed uneconomical utilization of ATP and NADPH in Acremonium chrysogenum fed with soybean oil. Bioprocess
Biosyst. Eng. 2010, 33, 1119–1129. [CrossRef]
81. Yang, S.-T.; Liu, X.; Zhang, Y. Chapter 4—Metabolic Engineering—Applications, Methods, and Challenges. In Bioprocessing for
Value-Added Products from Renewable Resources; Yang, S.-T., Ed.; Elsevier: Amsterdam, The Netherlands, 2007; pp. 73–118.
82. Vallino, J.J.; Stephanopoulos, G. Metabolic flux distributions in Corynebacterium glutamicum during growth and lysine overpro-
duction. Biotechnol. Bioeng. 1993, 41, 633–646. [CrossRef]
83. Takaç, S.; Çalık, G.; Mavituna, F.; Dervakos, G. Metabolic flux distribution for the optimized production of l-glutamate. Enzyme
Microb. Technol. 1998, 23, 286–300. [CrossRef]
84. Jørgensen, H.; Nielsen, J.; Villadsen, J.; Møllgaard, H. Metabolic flux distributions in Penicillium chrysogenum during fed-batch
cultivations. Biotechnol. Bioeng. 1995, 46, 117–131. [CrossRef]
85. Shimizu, H.; Shimizu, N.; Shioya, S. Roles of glucose and acetate as carbon sources in l-histidine production with Brevibacterium
flavum FERM1564 revealed by metabolic flux analysis. Biotechnol. Bioprocess Eng. 2002, 7, 171. [CrossRef]
86. Shirai, T.; Nakato, A.; Izutani, N.; Nagahisa, K.; Shioya, S.; Kimura, E.; Kawarabayasi, Y.; Yamagishi, A.; Gojobori, T.; Shimizu, H.
Comparative study of flux redistribution of metabolic pathway in glutamate production by two coryneform bacteria. Metab. Eng.
2005, 7, 59–69. [CrossRef]
87. Mahadevan, R.; Schilling, C.H. The effects of alternate optimal solutions in constraint-based genome-scale metabolic models.
Metab. Eng. 2003, 5, 264–276. [CrossRef]
88. Heirendt, L.; Arreckx, S.; Pfau, T.; Mendoza, S.N.; Richelle, A.; Heinken, A.; Haraldsdóttir, H.S.; Wachowiak, J.; Keating, S.M.;
Vlasov, V.; et al. Creation and analysis of biochemical constraint-based models using the COBRA Toolbox v.3.0. Nat. Protoc. 2019,
14, 639–702. [CrossRef]
89. Mahadevan, R.; Edwards, J.S.; Doyle, F.J. Dynamic Flux Balance Analysis of Diauxic Growth in Escherichia coli. Biophys. J. 2002, 83,
1331–1340. [CrossRef]
90. Kuriya, Y.; Araki, M. Dynamic Flux Balance Analysis to Evaluate the Strain Production Performance on Shikimic Acid Production
in Escherichia coli. Metabolites 2020, 10, 198. [CrossRef]
91. Hanly, T.J.; Henson, M.A. Dynamic model-based analysis of furfural and HMF detoxification by pure and mixed batch cultures of
S. cerevisiae and S. stipitis. Biotechnol. Bioeng. 2014, 111, 272–284. [CrossRef]
92. Valverde, J.R.; Gullón, S.; García-Herrero, C.A.; Campoy, I.; Mellado, R.P. Dynamic metabolic modelling of overproduced protein
secretion in Streptomyces lividans using adaptive DFBA. BMC Microbiol. 2019, 19, 233. [CrossRef]
93. Henson, M.A.; Hanly, T.J. Dynamic flux balance analysis for synthetic microbial communities. IET Syst. Biol. 2014, 8, 214–229.
[CrossRef]
94. Chen, L.; Liu, F. Recursive parameter identification for fermentation processes with the multiple model technique. Appl. Math.
Model. 2012, 36, 2275–2285. [CrossRef]
95. Li, X.; Zhang, S.; Xu, Z.; Feng, E. Parameter Identification Model with the Control Term in Batch Anaerobic Fermentation. In
Proceedings of the 2nd International Conference on Advanced Design and Manufacturing Engineering (ADME 2012), Taiyuan,
China, 16–18 August 2012; pp. 1535–1540.
96. Greener, J.G.; Kandathil, S.M.; Moffat, L.; Jones, D.T. A guide to machine learning for biologists. Nat. Rev. Mol. Cell Biol. 2022, 23,
40–55. [CrossRef]
97. Xu, R.Z.; Cao, J.S.; Wu, Y.; Wang, S.N.; Luo, J.Y.; Chen, X.; Fang, F. An integrated approach based on virtual data augmentation and
deep neural networks modeling for VFA production prediction in anaerobic fermentation process. Water. Res. 2020, 184, 116103.
[CrossRef]
98. de Vazelhes, W.; Carey, C.J.; Tang, Y.; Vauquier, N.; Bellet, A. Metric-learn: Metric Learning Algorithms in Python. J. Mach. Learn.
Res. 2020, 21, 138-1.
99. Kuhn, M. Building Predictive Models in R Using the caret Package. J. Stat. Softw. 2008, 28, 1–26. [CrossRef]
100. Blaom, A.D.; Király, F.J.; Lienart, T.; Simillides, Y.; Arenas, D.; Vollmer, S.J.J.A. MLJ: A Julia package for composable Machine
Learning. Comput. Sci. 2020, 5, 2704–2711. [CrossRef]
101. Noble, W.S. What is a support vector machine? Nat. Biotechnol. 2006, 24, 1565–1567. [CrossRef]
102. Dong, C.; Chen, J. Optimization of process parameters for anaerobic fermentation of corn stalk based on least squares support
vector machine. Bioresour. Technol. 2019, 271, 174–181. [CrossRef]
103. Zhang, L.; Chao, B.; Zhang, X. Modeling and optimization of microbial lipid fermentation from cellulosic ethanol wastewater by
Rhodotorula glutinis based on the support vector machine. Bioresour. Technol. 2020, 301, 122781. [CrossRef]
104. Uslan, V.; Seker, H. Binding affinity prediction of S. cerevisiae 14-3-3 and GYF peptide-recognition domains using support vector
regression. In Proceedings of the 38th Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society
(EMBC), Orlando, FL, USA, 16–20 August 2016; pp. 3445–3448.
105. Lewis, D.P.; Jebara, T.; Noble, W.S. Support vector machine learning from heterogeneous data: An empirical analysis using
protein sequence and structure. Bioinformatics 2006, 22, 2753–2760. [CrossRef]
106. Shah, A.R.; Oehmen, C.S.; Webb-Robertson, B.J. SVM-HUSTLE—An iterative semi-supervised machine learning approach for
pairwise protein remote homology detection. Bioinformatics 2008, 24, 783–790. [CrossRef]
Bioengineering 2022, 9, 473 17 of 18

107. Shahid, N.; Rappon, T.; Berta, W. Applications of artificial neural networks in health care organizational decision-making: A
scoping review. PLoS ONE 2019, 14, e0212356. [CrossRef]
108. Paszke, A.; Gross, S.; Massa, F.; Lerer, A.; Bradbury, J.; Chanan, G.; Killeen, T.; Lin, Z.M.; Gimelshein, N.; Antiga, L.; et al. PyTorch:
An Imperative Style, High-Performance Deep Learning Library. In Proceedings of the 33rd Conference on Neural Information
Processing Systems (NeurIPS), Vancouver, BC, Canada, 8–14 December 2019.
109. Abadi, M.; Barham, P.; Chen, J.; Chen, Z.; Davis, A.; Dean, J.; Devin, M.; Ghemawat, S.; Irving, G.; Isard, M.; et al. TensorFlow:
A system for large-scale machine learning. In Proceedings of the 12th USENIX conference on Operating Systems Design and
Implementation, Savannah, GA, USA, 2–4 November 2016; pp. 265–283.
110. Darsey, J.A.; Griffn, W.O.; Joginipelli, S.; Melapu, V.K. Architecture and biological applications of artifcial neural networks: A
tuberculosis perspective. Methods Mol. Biol. 2015, 1260, 269–283. [CrossRef]
111. Melcher, M.; Scharl, T.; Spangl, B.; Luchner, M.; Cserjan, M.; Bayer, K.; Leisch, F.; Striedner, G. The potential of random forest and
neural networks for biomass and recombinant protein modeling in Escherichia coli fed-batch fermentations. Biotechnol. J. 2015, 10,
1770–1782. [CrossRef]
112. Wang, Y.; Yang, G.; Sage, V.; Xu, J.; Sun, G.; He, J.; Sun, Y. Optimization of dark fermentation for biohydrogen production using a
hybrid artificial neural network (ANN) and response surface methodology (RSM) approach. Environ. Prog. Sustain. Energy 2021,
40, e13485. [CrossRef]
113. Tavasoli, T.; Arjmand, S.; Ranaei Siadat, S.O.; Shojaosadati, S.A.; Sahebghadam Lotfi, A. A robust feeding control strategy adjusted
and optimized by a neural network for enhancing of alpha 1-antitrypsin production in Pichia pastoris. Biochem. Eng. J. 2019, 144,
18–27. [CrossRef]
114. Schinn, S.M.; Morrison, C.; Wei, W.; Zhang, L.; Lewis, N.E. A genome-scale metabolic network model and machine learning
predict amino acid concentrations in Chinese Hamster Ovary cell cultures. Biotechnol. Bioeng. 2021, 118, 2118–2123. [CrossRef]
115. Oyetunde, T.; Liu, D.; Martin, H.G.; Tang, Y.J. Machine learning framework for assessment of microbial factory performance.
PLoS ONE 2019, 14, e0210558. [CrossRef]
116. Culley, C.; Vijayakumar, S.; Zampieri, G.; Angione, C. A mechanism-aware and multiomic machine-learning pipeline characterizes
yeast cell growth. Proc. Natl. Acad. Sci. USA 2020, 117, 18869–18879. [CrossRef]
117. Luo, Y.; Kurian, V.; Ogunnaike, B.A. Bioprocess systems analysis, modeling, estimation, and control. Curr. Opin. Chem. Eng. 2021,
33, 100705. [CrossRef]
118. Jolliffe, I.T.; Cadima, J. Principal component analysis: A review and recent developments. Philos. Trans. R. Soc. A 2016, 374,
20150202. [CrossRef]
119. Arian, R.; Hariri, A.; Mehridehnavi, A.; Fassihi, A.; Ghasemi, F. Protein kinase inhibitors’ classification using K-Nearest neighbor
algorithm. Comput. Biol. Chem. 2020, 86, 107269. [CrossRef]
120. Thompson, M.L.; Kramer, M.A. Modeling chemical processes using prior knowledge and neural networks. AIChE J. 1994, 40,
1328–1340. [CrossRef]
121. Xu, C.M.; Jackson, S.A. Machine learning and complex biological data. Genome Biol. 2019, 20, 76. [CrossRef]
122. von Stosch, M.; Oliveira, R.; Peres, J.; Feyo de Azevedo, S. Hybrid semi-parametric modeling in process systems engineering:
Past, present and future. Comput. Chem. Eng. 2014, 60, 86–101. [CrossRef]
123. Reel, P.S.; Reel, S.; Pearson, E.; Trucco, E.; Jefferson, E. Using machine learning approaches for multi-omics data analysis: A
review. Biotechnol. Adv. 2021, 49, 107739. [CrossRef]
124. Vijayakumar, S.; Rahman, P.K.S.M.; Angione, C. A Hybrid Flux Balance Analysis and Machine Learning Pipeline Elucidates
Metabolic Adaptation in Cyanobacteria. iScience 2020, 23, 101818. [CrossRef]
125. Vernon, I.; Liu, J.; Goldstein, M.; Rowe, J.; Topping, J.; Lindsey, K. Bayesian uncertainty analysis for complex systems biology
models: Emulation, global parameter searches and evaluation of gene functions. BMC Syst. Biol. 2018, 12, 1. [CrossRef]
126. Sridhara, V.; Meyer, A.G.; Rai, P.; Barrick, J.E.; Ravikumar, P.; Segrè, D.; Wilke, C.O. Predicting growth conditions from internal
metabolic fluxes in an in-silico model of E. coli. PLoS ONE 2014, 9, e114608. [CrossRef]
127. Wu, S.G.; Wang, Y.; Jiang, W.; Oyetunde, T.; Yao, R.; Zhang, X.; Shimizu, K.; Tang, Y.J.; Bao, F.S. Rapid Prediction of Bacterial
Heterotrophic Fluxomics Using Machine Learning and Constraint Programming. PLoS Comput. Biol. 2016, 12, e1004838.
[CrossRef]
128. Noor, E.; Cherkaoui, S.; Sauer, U. Biological insights through omics data integration. Curr. Opin. Syst. Biol. 2019, 15, 39–47.
[CrossRef]
129. Hewitt, C.J.; Nienow, A.W. The scale-up of microbial batch and fed-batch fermentation processes. Adv. Appl. Microbiol. 2007, 62,
105–135. [CrossRef] [PubMed]
130. Liu, Y.; Wang, Z.-J.; Xia, J.-Y.; Haringa, C.; Liu, Y.-P.; Chu, J.; Zhuang, Y.-P.; Zhang, S.-L. Application of Euler-Lagrange CFD for
quantitative evaluating the effect of shear force on Carthamus tinctorius L. cell in a stirred tank bioreactor. Biochem. Eng. J. 2016,
114, 212–220. [CrossRef]
131. Lara, A.R.; Galindo, E.; Ramírez, O.T.; Palomares, L.A. Living with heterogeneities in bioreactors: Understanding the effects of
environmental gradients on cells. Mol. Biotechnol. 2006, 34, 355–381. [CrossRef]
132. Heins, A.L.; Weuster-Botz, D. Population heterogeneity in microbial bioprocesses: Origin, analysis, mechanisms, and future
perspectives. Bioprocess Biosyst. Eng. 2018, 41, 889–916. [CrossRef]
Bioengineering 2022, 9, 473 18 of 18

133. Wang, G.; Zhao, J.; Haringa, C.; Tang, W.; Xia, J.; Chu, J.; Zhuang, Y.; Zhang, S.; Deshmukh, A.T.; van Gulik, W.; et al. Comparative
performance of different scale-down simulators of substrate gradients in Penicillium chrysogenum cultures: The need of a biological
systems response analysis. Microb. Biotechnol. 2018, 11, 486–497. [CrossRef]
134. Bolic, A.; Larsson, H.; Hugelier, S.; Lantz, A.E.; Kruhne, U.; Gernaey, K.V. A flexible well-mixed milliliter-scale reactor with high
oxygen transfer rate for microbial cultivations. Chem. Eng. J. 2016, 303, 655–666. [CrossRef]
135. Mishra, S.; Kumar, V.; Sarkar, J.; Rathore, A.S. CFD based mass transfer modeling of a single use bioreactor for production of
monoclonal antibody biotherapeutics. Chem. Eng. J. 2021, 412, 128592. [CrossRef]
136. Villiger, T.K.; Neunstoecklin, B.; Karst, D.J.; Lucas, E.; Stettler, M.; Broly, H.; Morbidelli, M.; Soos, M. Experimental and CFD
physical characterization of animal cell bioreactors: From micro- to production scale. Biochem. Eng. J. 2018, 131, 84–94. [CrossRef]
137. Wang, G.; Haringa, C.; Noorman, H.; Chu, J.; Zhuang, Y. Developing a Computational Framework to Advance Bioprocess
Scale-Up. Trends Biotechnol. 2020, 38, 846–856. [CrossRef]
138. Lapin, A.; Muller, D.; Reuss, M. Dynamic behavior of microbial populations in stirred bioreactors simulated with Euler-Lagrange
methods: Traveling along the lifelines of single cells. Ind. Eng. Chem. Res. 2004, 43, 4647–4656. [CrossRef]
139. Haringa, C.; Mudde, R.F.; Noorman, H.J. From industrial fermenter to CFD-guided downscaling: What have we learned? Biochem.
Eng. J. 2018, 140, 57–71. [CrossRef]
140. Siebler, F.; Lapin, A.; Hermann, M.; Takors, R. The impact of CO gradients on C. ljungdahlii in a 125 m3 bubble column: Mass
transfer, circulation time and lifeline analysis. Chem. Eng. Sci. 2019, 207, 410–423. [CrossRef]
141. Kuschel, M.; Siebler, F.; Takors, R. Lagrangian Trajectories to Predict the Formation of Population Heterogeneity in Large-Scale
Bioreactors. Bioengineering 2017, 4, 27. [CrossRef]
142. Gao, X.; Kong, B.; Vigil, R.D. Simulation of algal photobioreactors: Recent developments and challenges. Biotechnol. Lett. 2018, 40,
1311–1327. [CrossRef]
143. Gernigon, V.; Chekroun, M.A.; Cockx, A.; Guiraud, P.; Morchain, J. How Mixing and Light Heterogeneity Impact the Overall
Growth Rate in Photobioreactors. Chem. Eng. Technol. 2019, 42, 1663–1669. [CrossRef]
144. Guha, D.; Ramachandran, P.A.; Dudukovic, M.P.; Derksen, J.J. Evaluation of large eddy simulation and Euler-Euler CFD models
for solids flow dynamics in a stirred tank reactor. AIChE J. 2008, 54, 766–778. [CrossRef]
145. Elqotbi, M.; Vlaev, S.D.; Montastruc, L.; Nikov, I. CFD modelling of two-phase stirred bioreaction systems by segregated solution
of the Euler–Euler model. Comput. Chem. Eng. 2013, 48, 113–120. [CrossRef]
146. Morchain, J.; Gabelle, J.-C.; Cockx, A. A coupled population balance model and CFD approach for the simulation of mixing issues
in lab-scale and industrial bioreactors. AIChE J. 2014, 60, 27–40. [CrossRef]
147. Yeoh, J.W.; Jayaraman, S.S.O.; Tan, S.G.-D.; Jayaraman, P.; Holowko, M.B.; Zhang, J.; Kang, C.-W.; Leo, H.L.; Poh, C.L. A
model-driven approach towards rational microbial bioprocess optimization. Biotechnol. Bioeng. 2021, 118, 305–318. [CrossRef]
148. Tajsoleiman, T.; Spann, R.; Bach, C.; Gernaey, K.V.; Huusom, J.K.; Krühne, U. A CFD based automatic method for compartment
model development. Comput. Chem. Eng. 2019, 123, 236–245. [CrossRef]
149. Venkatasubramanian, V. The promise of artificial intelligence in chemical engineering: Is it here, finally? AIChE J. 2019, 65,
466–478. [CrossRef]
150. del Rio-Chanona, E.A.; Wagner, J.L.; Ali, H.; Fiorelli, F.; Zhang, D.; Hellgardt, K. Deep learning-based surrogate modeling and
optimization for microalgal biofuel production and photobioreactor design. AIChE J. 2019, 65, 915–923. [CrossRef]
151. Zhang, D.; Del Rio-Chanona, E.A.; Petsagkourakis, P.; Wagner, J. Hybrid physics-based and data-driven modeling for bioprocess
online simulation and optimization. Biotechnol. Bioeng. 2019, 116, 2919–2930. [CrossRef] [PubMed]