Bioresource Technology Reports 29 (2025) 102024

Contents lists available at ScienceDirect

Bioresource Technology Reports

journal homepage: [Link]/journal/bioresource-technology-reports

CFD optimization of an air lift fermenter for Fusarium

venenatum fermentation
Jianyong Zhu a,b, Wuxi Chen a , Yuxiang Chen c , Farrukh Raza Amin a , Yaxiang Li a , Mengjia Lu a ,
Demao Li a,*
a
Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, Tianjin 300308, China
b
School of Bioengineering, Tianjin University of Science and Technology, Tianjin 300457, China
c
Swanson School of Engineering, University Of Pittsburgh, Pittsburgh 15260, USA

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

Keywords: Airlift reactors are widely used in the field of microbial fermentation because of their simple structure and easy
Air lift reactor scale-up. This study investigated the enhancement of gas–liquid two-phase flow and mass transfer in a
CFD laboratory-scale airlift reactor for Fusarium venenatum fermentation through structural modifications. First, CFD
Baffle
modeling of the bubble reactor in the laboratory was performed and the reliability of the CFD model was verified
Fusarium venenatum
by cold model experiments. Secondly, The addition of a baffle in the original reactor transforms it into an
Fermentation
internal-loop split airlift reactor (ALR). The baffle was introduced into the initial reactor, and CFD was employed
to predict the optimal installation position of the baffle. The results indicate that when a gap of 60 mm is
maintained between the upper edge of the baffle and the liquid surface, a 40 mm gap between the lower edge of
the baffle and the bottom of the reactor, and an Ad/Ar ratio of 1, the internal-loop split ALR exhibited a 23.55 %
increase in gas holdup and a 30.07 % improvement in kLa compared to the original reactor., and the speed dead
zone has been significantly improved. Following these specifications, baffe were installed, and the performance
of the modified reactor was assessed against the original design during F. venenatum fermentation. The modified
reactor demonstrated a significant enhancement in internal circulation, gas holdup, and mass transfer efficiency.
Finally, the modified reactor was used for the fermentation experiment of F. venenatum. The fermentation results
of F. venenatum showed that the glucose consumption rate of the modified reactor increased by 0.42 g/L/h, and
the biomass increased by 3.49 g/L. The glucose conversion rate increased by 66.7 %, and the protein content
increased by 3.69 %. These results showed that the internal-loop split ALR substantially improved the efficiency
of the F. venenatum fermentation process.

1. Introduction foods and meat substitutes. Compared to traditional animal protein

production, the synthesis of F. venenatum protein has a lower environ­
Conventional protein provision methods that depend on traditional mental impact and requires fewer resources, contributing to the growing
agriculture and animal husbandry are facing significant challenges as global demand for sustainable food solutions. This protein provides a
the global population continues to grow (Anneli et al., 2017; Stavi et al., high-quality source of essential amino acids and has a positive impact on
2022; Werger, 2022). There has never been a more urgent need for an human health, making it an important candidate to meet the current
innovative approach to protein supply. After comprehensive screening market need for efficient and safe protein sources. Therefore, further
of >3000 soil samples, Fusarium venvenatum was discovered and utilized optimization of the F. venenatum fermentation process is of significant
as a competent cellular platform for the synthesis of Mycobacterium practical relevance, as it holds promise not only for enhanced produc­
protein (Finnigan et al., 2017; Tong et al., 2022b, 2023; Wiebe, 2002). tion efficiency but also for advancing the food industry toward more
F. venenatum protein, due to its rich nutritional profile and exceptional environmentally friendly practices. However, the high viscosity of the
physical properties, has been widely applied in the food industry, fermentation broth during fermentation is caused by the fungal
particularly showing strong potential in the production of plant-based morphology of Fusarium. The quantity and efficiency of F. venvenatum

* Corresponding author.
E-mail address: li_dm@[Link] (D. Li).

[Link]
Received 11 October 2024; Received in revised form 28 November 2024; Accepted 10 January 2025
Available online 11 January 2025
2589-014X/© 2025 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

fermentation are greatly impacted by this phenomenon, which signifi­ reactors that utilize baffe. The reactor geometry and cross-sectional area
cantly hinders mass transfer and disrupts homogeneous mixing in the are the primary factors that determine the impact of baffe on internal
fermentation reactor (Hernandez, 2018; Liu, 2023; Mcintyre et al., flow and mass transfer (Chen et al., 2016; Esperança et al., 2017; Huang
2001). et al., 2010). Therefore, the impact of baffess on the internal flow dy­
Strategic intervention is required to address the challenge of a sig­ namics of cylindrical reactors differs from that observed in rectangular
nificant decrease in dissolved oxygen and nonuniform mixing within the reactors (Ojha and Al Dahhan, 2017). As a result, the impact of baffe on
reactor caused by increasing viscosity during the fermentation of the internal flow dynamics in cylindrical reactors differs from that
F. venvenatum. To reduce the impact of viscosity on the fermentation observed in rectangular reactors. Despite this, in the context of cylin­
process, one technique is to dilute the fermentation broth, but this drical airlift reactors with baffe, the existing body of research has pri­
method can only delay the increase in viscosity and cannot fundamen­ marily explored the overall properties of gas–liquid two-phase flow
tally solve this problem. The meticulous design and optimization of re­ within reactors, with a notable absence of specific and in-depth in­
action devices are the key to a more fundamental resolution. The airlift vestigations into the nuanced impact of baffle structures on gas–liquid
bioreactor (ALR) is particularly suitable for the growth of filamentous flow and mass transfer within reactors (Luo and Al-Dahhan, 2012).
fungi due to its low-shear environment. High shear forces can often The purpose of this study is to enhance the production of Fusarium
disrupt the structure of fungal hyphae, leading to hyphal fragmentation venenatum through structural modifications of the laboratory's airlift
or morphological changes that negatively affect growth and metabolite reactor. Utilizing computational fluid dynamics (CFD), we systemati­
production. In contrast, the ALR relies on gas-driven circulation, which cally analyzed the impact of various baffle structural parameters on
avoids the intense shear forces associated with mechanical agitation. localized flow dynamics within the reactor. This approach enables a
This not only preserves the integrity of the hyphal structure but also precise evaluation of the flow patterns and mass transfer characteristics,
enhances oxygen transfer and nutrient distribution through uniform gas- which traditional empirical methods lack. By determining optimal baffle
liquid interfaces. (Campestrini and Salles-Campos, C., 2021; Ferreira configurations based on established standards for gas holdup and mass
et al., 2020; Neda et al., 2021; Patel et al., 2023). Studies by Papagianni, transfer coefficients, this study offers a novel perspective on reactor
Zhou & Zhong, and Amanullah et al. have shown that moderate shear design. Unlike previous studies that primarily rely on empirical insights,
environments can significantly improve biomass accumulation and this research integrates CFD modeling with experimental validation to
metabolite production in filamentous fungi. These findings support the achieve a more rigorous and effective design process. The proposed
positive impact of the ALR's low-shear environment on fungal improvements were further validated through fermentation experiments
morphology and metabolism in fermentation processes. Despite the with F. venenatum, providing practical and data-supported recommen­
simplicity and ease of manufacturing associated with airlift reactor dations for optimizing and advancing the fermentation process to
structures, the fermentation process often involves elevated aeration, enhance production efficiency.
which complicates the gas–liquid flow. This complexity hinders a
thorough understanding of the local gas and liquid flow dynamics within 2. Materials and methods
the reactor, thus impeding the optimization and refinement of airlift
reactors. To advance the optimization and innovative enhancement of 2.1. Geometric model
airlift reactors in the context of F. venvenatum fermentation, it is
necessary to understand the complexities involved. The fermentation bioreactor initially used was a bubble column
The conventional retrofitting of airlift reactors has largely been reactor in this study, it was composed of a cylindrical glass body
based on empirical insights from fermentation tank designers, which has mounted on a metal base, as shown in the supplementary file (Fig. 1 and
resulted in significant trial-and-error costs (Chen et al., 2020; Li et al., Table 1). The reactor had a total height of 543 mm and an inner diameter
2020), The conventional retrofitting of airlift reactors has largely been of 131.5 mm, corresponding to an overall volume of 7 L and a liquid
based on empirical insights from fermentation tank designers, which has holding capacity of 4.5 L. Internally, the reactor was equipped with a
resulted in significant trial-and-error costs. This empirical approach tubular gas distributor featuring six air outlets, each with a diameter of
often lacks precision, as it relies on generalized assumptions rather than 1 mm, It is installed 95 mm from the bottom of the reactor. The gas
a detailed understanding of the complex fluid dynamics within the volume flow rate of the distributor is controlled at 0.5–1.5 VVM (Li and
reactor. Consequently, it can lead to suboptimal designs that hinder Zhan, 2020).
mass transfer efficiency and uniform mixing, requiring multiple itera­
tions to achieve desired outcomes. These inefficiencies not only increase
operational costs and resource consumption but also limit the ability to 2.2. Method for determining the overall average gas holdup
fully optimize reactor performance for specific fermentation processes.
In recent years, propelled by swift advancements in computer technol­ The overall average gas holdup was determined using the volume
ogy and the continual refinement of foundational theoretical models, the expansion method (Raimundo et al., 2019). The procedure involved
application of computational fluid dynamics (CFD) for the design and opening the outlet valve until the free liquid level reached the premarket
scaling-up of reactors has increased in demand among researchers volume of 4.5 L, allowing it to settle for a period of time. Once the liquid
aiming to enhance filamentous fungal fermentation reactors (Chen et al., reached a stationary state, a straight edge was fixed near the liquid
2023; Esperana et al., 2023; Salehpour et al., 2020). surface, aligning the 0 mark of the straight edge with the interface of the
Indicators such as mass transfer and gas holding capacity play pivotal free liquid level. Subsequently, the air pump was activated, and the gas
roles in characterizing the efficacy of airlift reactors and are key factors flow rate was adjusted to the corresponding volume using a gas flow
that significantly influence filamentous fungal production during the meter. This process inflates the reactor with gas for a certain duration.
latter stages of fermentation. Many design strategies focus on improving After inflation, the position of the free liquid level was captured using a
these performance parameters to enhance the functionality of airlift high-resolution digital camera (Nikon d7200), and the scale H corre­
reactors (Berouaken et al., 2023; Luo et al., 2011). The guide plate is a sponding to the free liquid level on the straight edge was recorded. This
significant component of the reactor structure and has significant im­ value was subsequently substituted into Eq. (1) to calculate the overall
plications for gas holdup and mass transfer (Chai and Zhao, 2012; Choi, average gas holdup.
1995; Luo et al., 2011). By integrating the guide plate into cylindrical /
αg = ΔH (ΔH + H0 ) (1)
and rectangular reactors, the reactor can be divided into two zones: the
ascending zone and the descending zone. Recent research has focused on In this context, H0 represents the stable height of the liquid level
the complex flow field behavior and modifications of rectangular airlift before inflation, m, and △H is the difference in the height of the liquid

2
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Table 1
Structural parameters of airlift reactor.
Geometry structure Size(mm)

Glass outer wall height 365

Metal base height 178
Internal diameter 131.5
Outside diameter 162
Glass thickness 6
Outlet diameter 1
Air outlet spacing 15
Installation height of gas distributor 95
Gas distributor hole diameter 1
Gas distributor hole spacing 11

level before and after inflation, m. The experiment was repeated three
times, and the average of these 3 experimental results was obtained.

2.3. Determination method for kLa

The experimental method for measuring the gas-liquid mass transfer

coefficient kLa in the reactor usually uses the Na2SO3 oxidation method,
which is divided into a dynamic method and a steady-state method.
However, the gas-liquid flow behavior has a greater impact on the dy­
namic method, and a relatively weaker impact on the steady-state
method. In addition, during the experiment, the dynamic method has
higher requirements for the high sensitivity of the dissolved oxygen
electrode and the change of the saturated solution, while the steady-
state method has relatively weaker requirements. Therefore, the pre­
sent study utilized the steady-state method with Na2SO3 to measure the
mass transfer coefficient in an airlift reactor (Imai et al., 1987). The
chemical principle of this method involves the rapid reaction of Na2SO3
in liquid with oxygen under the catalysis of Co2+ or Cu2+, resulting in
the formation of Na2SO4. The chemical reaction equation is depicted in
Eq. (2).
Co2+
2Na2 SO3 + O2 → 2Na2 SO4 (2)
Initially, water and 10 mL of a 0.001 mol/L CoCl2 solution were
added to the airlift reactor to reach a calibrated volume of 4.5 L. Sub­
sequently, air was introduced into the reactor by controlling the air flow
rate with a gas flow meter. The dissolved oxygen in the water was
monitored using an oxygen electrode (METTLER TOLEDO InPro6800).
When the dissolved oxygen in the water reached saturation, a solution of
0.05 mol/L Na2SO3 was introduced into the reactor. With the addition of
the Na2SO3 solution, the dissolved oxygen in the water begins to
decrease. When the dissolved oxygen in the water stabilizes and no
longer changes with the addition of the Na2SO3 solution, the attainment
of a steady state is indicated. At steady state, the consumption rate of
dissolved oxygen in the reactor represents its overall average mass
transfer coefficient. The measurement of the mass transfer coefficient
using the steady-state method is based on Eq. (3).
dC
= kL a(C* − C) − OUR (3)
dt
In the above equation, C* represents the saturation concentration of
dissolved oxygen in water (mol/L), C is the dissolved oxygen concen­
tration at steady state after adding sodium sulfite (mol⋅L− 1), and OUR is
the oxygen consumption rate (mol⋅m− 3⋅s− 1), calculated by Eq. (4).
OUR = QNa MNa /2V (4)
QNa is the inflow rate of the Na2SO3 solution, which is 0.000137 L/s;
MNa is the concentration of the Na2SO3 solution; and V is the volume of
Fig. 1. The schematic representation of the bioreactor.
water in the airlift reactor, which is 4.5 L. When the dissolved oxygen
concentration reaches a steady state, the concentration of dissolved
oxygen in the water remains constant, so dC/dt is 0. Combining the above
two formulas, the calculation of the volume-averaged mass transfer
coefficient for the airlift reactor is given by Eq. (5).

3
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

kL a = QNa MNa /[2VC* (1 − C/C* ) ] (5) 2.4.4. Determination of protein content

The determination of protein content was carried out using the
The volume-averaged mass transfer coefficients were measured methods of our research group (Tong et al., 2023). The dry mycelia of F.
under three aeration rates: 0.5 VVM, 1 VVM, and 1.5 VVM. Three venatum were ground to powder, and then, a suitable amount of powder
measurements were taken for each aeration rate, and the average value (3–4 mg) was weighed to measure the nitrogen content (%) using a
was calculated. Flashsmart element analyzer (Thermo Ltd., Waltham). The protein
content (%) was calculated as nitrogen content × 6.25 (protein factor),
as the nitrogen content in proteins is approximately 16 %, making 6.25 a
2.4. Fermentation experiment
commonly used protein factor. This experiment was repeated in tripli­
cate, and the error bars represent the standard deviation of three rep­
2.4.1. Microorganisms and cultivation methods
licates. The statistical significance of the data was determined by t-tests,
The strain used in the fermentation experiment was the Fusarium
with significance levels defined as *P < 0.05, **P < 0.01, ***P < 0.001
venvenatum, which was preserved at CGMCC No.20740 and taxonomi­
and ****P < 0.0001. These analyses ensured the reliability of the results
cally named F. venenatum TB01.
obtained in the experiment.
The fermentation methodology employed in this study followed the
procedures established in our previous work conducted by the research
2.5. CFD simulation model
team (Tong et al., 2022a). The cultivation process was divided into two
stages: the seed cultivation stage and the fermentation cultivation stage.
2.5.1. Two-phase flow model
The seed media consisted of glucose (30 g/L) and yeast extract (8 g/L).
There are a large number of bubbles in the liquid phase of the airlift
The fermentation media were composed of glucose (60 g/L), (NH4)2SO4
reactor during operation. The motion of each bubble is calculated by the
(6 g/L), MgSO4 (1.5 g/L), Na2SO4 (0.5 g/L), KH2PO4 (2 g/L), KCl (0.7 g/
Euler–Lagrangian method, which increases the difficulty of solving this
L), CaCO3 (0.5 g/L), and yeast extract (0.5 g/L). All the media
problem. To reduce the amount of calculation while ensuring calcula­
mentioned above were sterilized at 115 ◦ C for 20 min.
tion accuracy, this study uses the Euler–Euler method to calculate
To compare the performances of the two different gas-lift fermenters,
gas–liquid two-phase flows, where the liquid phase is treated as the
the same strain, medium, and operating conditions were used.
continuous phase and the gas phase is treated as the dispersed phase
F. venenatum was inoculated into two triangular flasks, each containing
(Dou and Song, 2010; Xu et al., 2011). The simulation involves solving
50 mL of seed media. After incubating at 28 ◦ C on a shaker rotating at
the continuity equation and the momentum conservation equation. The
180 rpm for 24 h, the seed culture was transferred to two airlift reactors
continuity equation and momentum equation are shown in Eq. (7) and
with a fermentation volume of 4.5 L for continuous fermentation.
Eq. (8), respectively.
Ventilation ratio was 1.5vvm. The fermentation process was carefully
( )
maintained at 28 ◦ C using an automated control system equipped with a ∂
uk = 0
(αk ρk ) + ∇ αk ρk → (7)
temperature electrode. When the temperature fell below the setpoint, ∂t
the system activated a heating belt to warm the fermenter walls, ceasing
( ) ( )
operation once the target temperature was restored. Similarly, the pH ∂ →
u k + ∇ • α k ρk →
αk ρk → u k→
u k = − αk ∇Pk + αk ρk → + ∇αk →
τ k + αk ρk F k
was continuously monitored and precisely regulated at 6 throughout the ∂t g

fermentation. As the production process naturally led to a decline in pH (8)

due to metabolite accumulation, a pH electrode detected these changes
In the expressions, the subscript k = l corresponds to the liquid
in real-time. Upon sensing a drop below the setpoint, the control system
phase, and k = g corresponds to the gas phase. αk, ρk, and Pk denote the
promptly introduced a 12.5 % ammonia solution to the broth, ensuring
consistent pH stabilization at the desired level. The fermentation gas holdup, density, and pressure of phase k, respectively. →
u k represents
→
stopped when the glucose in the reactor was completely consumed, and the velocity vector, and τ is the shear stress tensor. F signifies the
→
k k
the physiological parameters such as biomass and protein content in the momentum exchange between the liquid and gas phases. The funda­
two reactors were compared. mental cause of momentum exchange between the gas and liquid phases
lies in the interphase forces. Among various interphase forces, drag and
2.4.2. Biomass measurement methods lift have the greatest impact on the results. Therefore, in this study, only
To determine biomass, we employed the method of dry cell weight drag and lift forces are used to close the interphase forces.
(DCW, g/L), which is consistent with the method used in our previously The expression for the drag force is shown in Eq. (9):
published research (Tong et al., 2023). At the conclusion of cultivation, a ⃒ ⃒( )
3 CD ⃒ ⃒
0.03 L sample of the fermentation broth was taken for filtration. Prior to FD = αg αl ρl ⃒⃒→ ug− →u g ⃒⃒ →ug− →
ul (9)
4 db
filtration, the weight of the filter paper was recorded as M0, and the
paper was rinsed twice with distilled water. Subsequently, the rinsed In this equation, CD is the drag coefficient, db is the bubble diameter,
and filtered sample was placed in an oven for drying until a constant αl is the liquid phase volume fraction, αg is the gas phase volume frac­
weight was achieved. The weight recorded at this point is denoted as M1. tion, →
u g and →
u l are the gas and liquid phase velocities. The Tomiyama
Biomass can then be calculated using Eq. (6). drag coefficient model was selected for this study due to its ability to
DCW = (M1 − M0 )/0.03 (6) account for the deformation of bubbles and the effects of the Eötvös
number on drag force(Tomiyama, 1998), and the expressions for the
Tomiyama drag coefficient are shown in Eq. (10) and Eq. (11):
2.4.3. Glucose concentration determination
( ( ))
The determination of glucose concentration was carried out using the 24 8 Eo
method described by Tong et al. (Tong et al., 2023). The sample from the CD = Max 1 + 0.15Reb 0.0687 , (10)
Reb 3 Eo + 4
fermentation broth was centrifuged and then filtered through a 0.22 μm
( )
filter. A 10 μL aliquot of the filtered supernatant was taken and diluted to g ρl − ρg db 2
an appropriate concentration by adding an appropriate amount of Eo = (11)
σ
distilled water. The glucose concentration in the solution was measured
using a biosensor (SBA–40D, Shandong Academy of Sciences). The The expression for the lift force is shown in Eq. (12):
original glucose concentration in the fermentation broth was subse­
quently restored according to the dilution factor.

4
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

( ) ( )
reactors is typically represented by the overall volumetric mass transfer
ug− →
FL = CL αg ρl → ul ul
× ∇×→ (12)
coefficient (kLa). The overall volumetric mass transfer coefficient (kLa) is
the product of the overall mass transfer (kL) and the specific interfacial
In this equation, CL is the lift coefficient, and ρl is the liquid phase
area per unit liquid volume (a). Methods for determining the liquid film
density. The lift coefficient is closed using the Tomiyama model (Liew
mass transfer coefficient, which is indicative of the mass transfer resis­
and Gimbun, 2017), which describes the lateral movement of a single
tance of the liquid film, include Higbie's penetration theory and surface
bubble in shear flow. The expressions for the Tomiyama lift coefficient
renewal theory (Danckwerts, 1981). Among these, the most widely
are shown in Eq. (13) to Eq. (16):
accepted mass transfer model is Higbie's penetration theory. In this
⎧
⎨ Min[0.288tanh0.121Reb , f(Eod) ] Eod < 4 study, we utilize the overall mass transfer model developed by Xue, Yin
CL = f(Eod) 4 ≤ Eod ≤ 10 (13) (Bach et al., 2017) and others based on this theory, as shown in Eqs. (22)
⎩
− 0.29 Eod > 10 and (23).
√̅̅̅̅̅̅̅̅̅̅̅ 1
f(Eod) = 0.00105Eod3 − 0.0159Eod2 − 0.0204Eod + 0.474 (14) kL = 2 DL /π(ερL /μ)4 (22)

( ) In the equation, DL represents the diffusion coefficient of oxygen in

g ρ l − ρ g dh 2 the liquid phase, m2/s; ε is the turbulent energy diffusion rate, and μ
Eod = (15)
σ denotes the viscosity of the liquid phase, Pa⋅s.
/
( )1 a = 6αg db (23)
dh = db 1 + 0.163Eo0.757 3 (16)
In this context, αg represents the gas holdup, and the calculation
2.5.2. Turbulence model method for the average bubble diameter (db) is detailed in Eq. (16).
The gas-liquid two-phase flow is in the form of turbulence in the
airlift reactor. As a widely validated and computationally efficient tur­ 2.5.4. Boundary conditions
bulence model, the standard k-ε model was chosen for this study due to The boundary surface was set to a nonslip wall. The standard wall
its ability to balance computational cost with sufficient accuracy. function was selected. The pressure outlet was adopted for the outlet,
(Laborde-Boutet et al., 2009),. It is computationally efficient and widely and the operating pressure was set to 1 standard atmosphere; the gas
validated, making it suitable for large-scale simulations. While advanced inlet was selected as the mass flow inlet, with the gas volume fraction set
models like RNG k-ε or k-ω may improve accuracy in specific cases, they to 1 at the inlet.
are more computationally intensive. For the flow conditions in this
study, the standard k-ε model provides sufficient precision to capture 2.5.5. Simulation details
key flow characteristics without excessive computational overhead. The The outlet flow rates were set based on the mass flow rates of air
calculation equations are shown in Eq. (17) and Eq. (18). corresponding to 0.5 VVM, 1 VVM, and 1.5 VVM. The diameter of the
[( ) ] bubbles at the inlet was calculated according to the bubble formation
∂αl ρl k ( ) μ
+ ∇ • αl ρl →u lk = ∇ • μ + tl αl ∇k + αl (Pk + Pkb − ρl ε) correlation summarized by Khurana and Kumar (Khurana and Kumar,
∂t σk
1969), and the bubbles were treated as spherical according to the
(17) expression in Eq. (24).
( ) ( )0.75
∂(αl ρl ε)
+∇ • αl ρl →u lε = ∇ 15μQ
∂t Vb = (2π)0.25 (24)
[( ) ] 2ρL g
μtl ε
• μ+ α ∇ε + αl [Cε1 (Pk +Pεb )− Cε2 ρl ε ] The overall average bubble diameter (b) in a gas-lift reactor is ob­
σε l k
tained based on the calculation method proposed by Johansen and
(18)
Boysan (Johansen et al., 1988), among others, for determining the
In the equation, Cε1 = 1.44, Cε2 = 1.92, σk = 1, σε = 1.3, Pkb and Pε average bubble diameter in a gas-lift reactor. The expression is given by
represent the influence of buoyancy, μtl represents the turbulent vis­ Eq. (25):
cosity of the liquid phase, and Pk represents the turbulence term ( / )0.2
generated by the viscosity. db = 0.35 Q2 g (25)
The calculation expression of the effective viscosity of the liquid
phase is: where VB represents the volume of the bubble, m3; Q is the volumetric
flow rate of the gas, m3/s; and the diameter of the bubbles at the inlet is
μeff ,l = μl + μtl + μtg (19) determined by the volumetric formula for a spherical shape, as given in
In the formula, μl represents the molecular viscosity of the contin­ Eq. (15).
uous phase, and μtg is the turbulent viscosity caused by the discrete The simulation model was built using simulation software such as
phase. The calculation formula is: Space Claim based on a 7 L laboratory bubble column. The fluid domain
In the formula, Cμp = 0.6. inside the bubble column was defined, and fluent meshing was subse­
The calculation expression of the effective viscosity of the discrete quently employed to generate the mesh for the fluid domain. This study
phase is: considered four models with different mesh densities, and the calcula­
tion results are shown in Table 2. Among them, the model with a mesh
μeff ，g = μg + μT,g (20) density of 669,492 cells produced results closer to the true values.
Considering both computational cost and accuracy, the model with
In the formula, μTg represents the discrete phase turbulent viscosity,
669,492 cells was selected for subsequent calculations. This study used
and its calculation expression is:
the commercial software Ansys Fluent 2021 (Tianjin Tianhe Computer
μT,g =
ρg
μ (21) Technology Co., Ltd.) for simulation, focusing on an air–water system.
ρl tl The physical properties of the system are detailed in Table 3. The
multiphase Eulerian fluid model was used with tap water as the
2.5.3. Mass transfer model continuous phase and air as the dispersed fluid, and the drag coefficient
The mass transfer capacity of gas–liquid two-phase systems inside model utilized the Tomiyama equation, with a surface tension

5
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Table 2 3.2. Distribution and optimization of the flow field inside the fermentation
Grid independence verification. reactor
Minimum Maximum Number Maximum Average
mesh size mesh size of mesh skewness gas hold Numerical simulations of a 7 L filamentous fungus fermentation
(mm) (mm) up reactor operating at an aeration rate of 1.5 VVM revealed several key
Mesh 1 0.05 3 669,492 0.264 0.0248 findings regarding the flow field distribution and its implications for the
Mesh 2 0.03 3 890,890 0.352 0.0233 fermentation process. Fig. 3(a) shows that the distribution of gas holdup
Mesh 3 0.025 3 1,127,871 0.447 0.0236 is asymmetric. Gas out from the gas inlet tended to deviate toward the
Mesh 4 0.025 2 1,326,801 0.306 0.0216
Experiment – – – 0.0258
right wall, leading to a decreased gas–liquid contact area, which could
adversely affect the dispersion of oxygen and the overall fermentation
efficiency. Fig. 3(b) shows that the shear strain rate, which is indicative
coefficient of 0.07. The drag force model was used to describe the of shear forces within the fermentation broth, was highest at the
interaction force between gas and liquid, and the drag force coefficient gas–liquid interface and adjacent to the reactor walls, while significantly
model used the Tomiyama model. Turbulence effects were considered in lower shear forces were present in the other regions. This variation in
the simulation using the standard k-ε equation. In this study, transient shear force can lead to cell agglomeration, particularly in areas of low
calculations were performed under a pressure solver. Gravity was shear, which is detrimental to the growth of filamentous fungi (Cai et al.,
considered during the calculation process (g = 9.81 m/s2). The oper­ 2012). Fig. 3(c) illustrates that the liquid-phase velocity was greater
ating pressure was set to 101,325 Pa. Pressure-velocity coupling was above the gas inlet and near the right wall but that a dead zone with
performed with the phase-coupled SIMPLE algorithm. The time step size minimal liquid-phase circulation was not conducive to effective mixing
in the simulation was set to 0.001 for the first 1000 steps, after which it of gas, liquid, and cells. Dead zones within airlift reactors pose a sig­
was increased to 0.005. Simultaneously, the overall mass transfer co­ nificant challenge to long-term fermentation performance by reducing
efficient was monitored throughout the simulation. The simulation substrate utilization efficiency and creating localized conditions detri­
showed that when the residual reached 10− 3, the overall mass transfer mental to cell viability. The accumulation of substrates and by-products
coefficient exhibited no significant changes. in these stagnant zones can lead to nutrient imbalances or inhibitory
metabolite concentrations, ultimately hindering overall fermentation
3. Results and discussions productivity. Fig. 3(d) depicts the turbulent liquid-phase circulation in
the reactor, characterized by a large vortex in the upper part and a
3.1. Verification and validation of the CFD model marked reduction in the swirling velocity below the gas distributor,
contributing to the formation of a dead zone that impedes mixing be­
In the CFD model simulation of the laboratory bubble column tween the gas and liquid phases. Furthermore, the liquid phase exhibited
reactor, three different aeration rates were examined: 0.5VVM, 1VVM, complex dynamics; as it moved upward on the right side and then
and 1.5VVM. The simulation results revealed an increase in both the horizontally toward the left, it encountered the upward flow induced by
overall gas holdup (αg) and overall mass transfer coefficient (kLa) the gas distributor, causing the upward-moving liquid to deflect toward
concomitant with an increase in gas velocity. As depicted in Fig. 2, there the left wall. This deflection not only affects the liquid phase but also
was a discernible positive correlation between the gas holdup, mass results in smaller bubbles moving toward the left wall, which partially
transfer coefficient, and aeration rate. These findings were the same as explains the higher gas holdup and mass transfer coefficient observed on
the empirical observations reported in the studies conducted by Xu et al. the left side of the reactor in the absence of a baffle plate.
(Xu et al., 2023). Meanwhile, both cold-flow model experiments and Our comprehensive numerical simulations utilizing CFD technology
numerical simulation results indicate that the reactor exhibits optimal revealed significant improvements in the gas–liquid flow dynamics
gas holdup and mass transfer rate at an aeration rate of 1.5 VVM. within the fermentation reactor upon the installation of a baffle plate.
Therefore, we will proceed with a 1.5 VVM aeration rate for subsequent The baffle plate, which was 125 mm in length, 2 mm in width, and 218
experiments. mm in height, was strategically placed at the midpoint of the reactor. As
The comparative analysis of gas holdup and mass transfer coefficient shown in Fig. 3, this modification effectively mitigated the issue of gas
values, as obtained from CFD simulations and experimental observa­ preferentially deviating toward the left wall. Now, the gas is able to
tions at different aeration rates, indicates strong congruence, with one disperse more evenly above the gas inlet, enhancing the distribution on
notable exception. At an aeration rate of 1 VVM, the simulation results both sides of the reactor. The shear strain rate, as indicated in the sim­
for the mass transfer coefficient deviated from the experimental data by ulations, is now predominantly concentrated at the gas–liquid interface,
approximately 15 %. This significant discrepancy could be attributed to above the gas inlet, and around the edges of the baffle plate, consistent
the error associated with the computational estimation of the bubble with the findings reported by Luo et al. (Luo and Al-Dahhan, 2008). The
diameter, which is a critical parameter impacting mass transfer, as liquid-phase velocity distribution within the reactor becomes more
evidenced in prior research (Cerri et al., 2008). While the calculated uniform with the baffle plate, addressing the previously noted low-
bubble diameter introduced a degree of inaccuracy not present in the velocity zone below the gas inlet. A vector map of the liquid-phase ve­
actual physical system, it was important to note that a deviation within locity, shown in Fig. 3(d), reveals that the baffle plate significantly de­
15 % between the simulation and experimental findings was generally creases the size of the vortex below the liquid surface. As the tail of the
considered acceptable. This level of accuracy suggested that the CFD vortex shifts to the left, the baffle plate obstructs the vortex flow of the
model effectively captured the dynamics of gas–liquid two-phase flow liquid phase, directing the liquid downward along the plate. This
within the gas-lift reactor under the conditions tested. alteration prevents the leftward shift of the vortex tail, allowing both the
liquid and gas phases to ascend directly above the distributor, which in
turn rectifies the uneven gas holdup and improves the reactor's gas-
holding capacity. Moreover, the overall liquid-phase flow direction
Table 3
within the reactor now forms a large, effective circulation pattern that is
Physical properties of gas–liquid two-phase flow. beneficial to the process. The descending liquid phase, guided by the
baffle, enters the ascending region from below, enhancing mixing in the
Density (kg⋅m− 3) Viscosity (Pa⋅s) Surface tension (N⋅m− 1)
previously stagnant zone beneath the distributor. This circulation
Air 1.225 0.00001789 – pattern significantly reduces velocity dead zones, thereby enhancing
Water 998.2 0.001003 0.07
substrate utilization and ensuring a more uniform distribution of

6
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Fig. 2. Comparison between the experimental results and simulation results for an airlift reactor. (a) Average gas holdup (b) Volumetric oxygen transfer coefficient.

nutrients and metabolites. The improvements are expected to enhance cross-sectional contour maps of the gas holdup and mass transfer coef­
long-term fermentation performance, as a homogeneous environment ficient at different reactor heights, represented by the dimensionless H/
supports sustained cell growth and metabolic activity. Although a D ratio, where H is the height and D is the diameter of the reactor. At H/
smaller vortex persists above the reactor, there is a notable shear force at D = 1, the reactor with a 60 mm gap exhibited the broadest distribution
the vortex tail and along the upper edge of the baffle. This observation of gas holdup and mass transfer coefficient, indicating superior perfor­
was the same as that of Luo et al. (Luo and Al-Dahhan, 2008), who also mance. As the gap widened, this distribution became more confined
identified significant shear forces at the baffle's upper and lower edges. toward the center of the reactor. At a higher reactor level, H/D = 1.5, the
These findings underscore the importance of the baffle plate in ho­ contour maps reveal a significant decrease in the high mass transfer
mogenizing the shear force distribution and improving overall reactor coefficient region, particularly in the lower right corner, as the gap in­
performance. creases. Correspondingly, the mass transfer coefficient at the edge of the
descending region diminishes, contributing to the overall reduction in
gas holdup and mass transfer coefficient observed with larger gaps.
3.3. Optimization of the inner baffle structure in an airlift fermentation
Fig. 5(c) shows the shear strain rate distribution at the upper part of the
reactor
baffle plate in reactors with different gaps. Fig. 4 shows no substantial
difference in the average shear strain rates among the reactors; however,
3.3.1. Influence of the gap between the upper parts of the baffle on the flow
the localized distribution within the reactor shows a decrease in the
field of the reactor
shear strain rate in the ascending region with larger gaps, especially
This section studied the influence of three different distances be­
above the gas inlet. Conversely, there is a marked increase in the shear
tween the upper edge of the baffle and the gas-liquid interface on the
strain rate at the upper edge of the baffle plate, with negligible change at
flow field of the internal-loop split ALR. The results are shown in Fig. 4.
the lower edge. The ascending region is pivotal for mass transfer be­
When the gap is expanded from 60 mm to 100 mm, there is a noticeable
tween the gas and liquid phases, and a pronounced shear force in this
decline in the reactor efficiency, as evidenced by a 6.4 % decrease in the
region enhances the disruption of filamentous fungus aggregates during
overall average gas holdup and a 10.37 % reduction in the volume-
fermentation. Fig. 5(d) illustrates the magnitude and direction of the
averaged mass transfer coefficient. Liu et al., in their investigation of
liquid-phase velocity during the circulation process in reactors with
the overall gas holdup during three-phase mixing in an external loop
different gaps. The data demonstrate that a smaller gap between the
airlift reactor, reached similar conclusions. They attributed the varia­
baffle plate and the liquid surface bolsters the baffle plate's ability to
tions in the upper gap to the turbulent effects induced by solid particles
mitigate vortex interference above the fermentation liquid, thereby
(Liu et al., 2020). Similarly, others reached identical conclusions while
promoting effective liquid circulation. This action, in turn, improved the
studying gas–liquid two-phase flow in a rectangular airlift reactor
gas-holding and mass transfer capacities within the reactor.
(Kilonzo and Margaritis, 2006). These findings suggest that a larger gap
diminishes the reactor's ability to retain gas and hampers the mass
3.3.2. Influence of the gap between the lower edge of the baffle and the
transfer process. A reduction in gas holdup and mass transfer coefficient
bottom of the reactor on reactor flow and mass transfer
(kLa) significantly impacts the fermentation efficiency of Fusarium ven­
This section studies the effects of four different spacings from the
enatum. Lower gas holdup shortens bubble residence time, reducing
lower edge of the baffle to the bottom of the reactor on the flow field of
oxygen transfer, while a 10.37 % decline in kLa indicates reduced oxy­
the internal-loop split ALR. The results are shown in Fig. 6. The data
gen availability. These limitations may lead to oxygen-limiting condi­
indicate a negative correlation between the size of the bottom gap and
tions, impairing F. venenatum's metabolism, biomass production, and
the reactor's gas holdup and mass transfer efficiency. Specifically, as the
product yield. However, the average shear strain rate across different
bottom gap increases from 40 mm to 95 mm, the overall average gas
gap sizes does not exhibit significant variation, indicating that the in­
holdup decreases by 9.14 %, while the mass transfer coefficient de­
fluence of the gap on the reactor's shear forces is minimal. Despite this,
creases by 9.29 %. This trend indicates that a larger bottom gap com­
the reactor with a 100 mm gap exhibited a marginally greater average
promises the reactor's ability to retain gas and facilitate mass transfer
shear strain rate than did the reactor with smaller gaps, which could
between phases. In contrast, the average shear strain rate across the
imply that complex interactions occur between the gap size and local­
reactor appears to be less sensitive to changes in the bottom gap. The
ized shear forces and merits further investigation.
reactor with an 80 mm bottom gap exhibited a marginally greater shear
Fig. 5 provides a detailed visualization of the flow field and mass
strain rate, which was only 1.69 % greater than that of the reactor with
transfer distribution in reactors with different gaps between the upper
the smallest shear strain rate, at a 95 mm bottom gap. This observation
edge of the baffle plate and the liquid surface. Fig. 5a and b present

7
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Fig. 3. Cloud chart of the flow field distribution in the two reactors. (a) Gas holdup cloud map (b) Mass transfer coefficient cloud map. (c) Shear strain rate cloud
map. (d) Liquid phase velocity vector cloud map.

implies that the influence of the bottom gap on the overall shear force of widens. Fig. 7(d) further demonstrates that as the clearance between the
the reactor is relatively minor compared to its significant impact on the bottom of the reactor and the baffle plate increases, the velocity vector
gas holdup and mass transfer capabilities. of the liquid phase entering the upflow zone decreases with decreasing
Fig. 7 shows the local characteristics of the distance between the clearance. This change in fluid dynamics causes the circulating fluid to
lower edge of the baffle and the bottom of the reactor on gas–liquid flow shift laterally toward the wall in the upflow zone, which has a conse­
and mass transfer. Cross-sectional contour maps of gas holdup and mass quential impact on the reactor's overall gas holdup and mass transfer
transfer coefficients at different reactor heights, as displayed in Fig. 7(a capabilities.
& b), reveal a trend where increasing bottom clearance leads to a According to the studies of Peter M. Kilonzo et al. and Chen et al.
contraction of the regions exhibiting high gas holdup and mass transfer (Chen et al., 2023; Kilonzo and Margaritis, 2006), the impact of the
coefficients. This contraction results in these areas migrating closer to reactor bottom gap on the gas holdup and mass transfer coefficients was
the reactor walls, which in turn causes a decrease in the overall average the same as the findings of our current research. Peter M. Kilonzo posits
values of gas holdup and the mass transfer coefficient across all reactors. that a reduction in the bottom gap leads to an increase in gas holdup in
The shear strain rate distribution, as depicted in Fig. 7(c), shows a the riser due to the constraints imposed by the reactor bottom pressure
relatively constant overall value despite variations in the bottom drop, causing the gas phase to predominantly exit through the down­
clearance. However, the contour maps indicate that the area of high comer, as corroborated by the gas holdup distribution cloud chart in
shear strain within the upflow zone diminishes as the bottom clearance Fig. 7(a). Chen et al. suggested that a decrease in the reactor bottom gap

8
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Fig. 4. Average fluid dynamics at different distances from the baffle to the liquid level. (a) Average gas holdup. (b) Average mass transfer coefficient. (c) Average
shear strain rate.

Fig. 5. Flow field cloud diagram of the baffle-to-liquid level at different distances. (a) Gas holdup cloud map. (b) Mass transfer coefficient cloud map. (c) Shear strain
rate cloud map. (d) Liquid phase velocity vector cloud map.

may induce greater turbulence within the reactor, thereby enhancing Al-Busafi et al. (Al, 2016). Specifically, when comparing a reactor with
mass transfer between the gas and liquid phases. an Ad/Ar of 1 to one with an Ad/Ar of 0.24, there is an increase of 13.3 %
in the overall average gas holdup, a 16.9 % increase in the mass transfer
3.3.3. Influence of Ad/Ar on reactor flow and mass transfer coefficient, and an 8.2 % increase in the average shear strain rate. These
This summary studies the effects of three types of descending zone findings indicate that a larger cross-sectional ratio is advantageous,
and ascending zone cross-sectional ratios (Ad/Ar) on the flow field of the leading to improved gas holdup, elevated mass transfer rates, and
internal-loop split ALR. The results are shown in Fig. 8. The data showed increased shear forces within the reactor. Optimal gas–liquid flow and
that an increase in the Ad/Ar ratio was correlated with significant in­ mass transfer conditions are achieved when the baffle plate is positioned
creases in several parameters. These findings are consistent with those of centrally, corresponding to an Ad/Ar of 1, thereby indicating that the

9
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Fig. 6. Average hydrodynamic parameters at different distances from the bottom of the baffle bottom edge. (a) Average gas holdup. (b) Average mass transfer
coefficient. (c) Average shear strain rate.

Fig. 7. Flow field cloud diagram from the lower edge of the baffle to the bottom of the reactor at different distances. (a) Gas holdup cloud map. (b) Mass transfer
coefficient cloud map. (c) Shear strain rate cloud map. (d) Liquid phase velocity vector cloud map.

reactor performance is sensitive to the configuration of the internal flow low shear strain rates, with the downflow areas showing negligible shear
areas. However, placing the baffle at the center of the reactor may lead strain. This suggests that the flow dynamics in these areas are signifi­
to biomass accumulation. To address this challenge, future efforts cantly more quiescent. Fig. 9(d) shows the velocity and direction of
should focus on optimizing the reactor's internal flow dynamics and liquid circulation within the reactor. With a larger downflow area, the
enhancing gas-liquid circulation to mitigate biomass buildup. hindrance posed by the baffle plate to the formation of vortices above
Fig. 9(a & b) illustrates how the gas holdup and mass transfer coef­ the fermentation liquid is reduced. Conversely, as the Ad/Ar ratio in­
ficient vary with changes in the Ad/Ar ratio, with these parameters creases, the baffle plate exerts a stronger inhibitory effect on vortex
predominantly localized near the wall of the outlet. As the Ad/Ar in­ formation. This indicates that the reactor's internal geometry, particu­
creases, both the gas holdup and mass transfer coefficient distributions larly the size of the downflow area, plays a crucial role in influencing the
tend to shift toward the center of the reactor. When the ratio reaches Ad/ flow patterns and missing efficiency.
Ar = 1, these distributions are not only centered directly above the outlet
but also exhibit a wider spread compared to reactors with lower cross-
sectional ratios. Fig. 9(c) highlights that reactors with lower Ad/Ar ra­ 3.4. F. venenatum TB01 fermentation
tios, specifically Ad/Ar = 0.24 and Ad/Ar = 0.53, experience relatively
During the fermentation of F. venenatum, the performance of the

10
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Fig. 8. Average hydrodynamic parameters at the different cross-sectional ratios in the descending and rising zones. (a) Average gas holdup. (b) Average mass transfer
coefficient. (c) Average shear strain rate.

Fig. 9. Flow field distribution cloud map of the different cross-sectional ratios in the descending and ascending regions. (a) Gas holdup cloud map. (b) Mass transfer
coefficient cloud map. (c) Shear strain rate cloud map. (d) Liquid phase velocity vector cloud map.

improved airlift fermentation tank was compared with that of the The improvement in performance was mainly attributed to the intro­
original laboratory-scale fermentation tank. The characteristics of the duction of baffe, which enhanced shear forces and promoted internal
improved fermentation tank included a 60 mm gap between the baffle circulation in the reactor. This improvement improved the efficiency of
and the liquid level, a 40 mm gap between the baffle and the bottom, and mixing gas and liquid phases with fungal cells, resulting in faster sugar
an Ad/Ar ratio of 1. As shown in Fig. 10 (a), 72 h were required for the absorption. In addition, Fig. 10 (b) shows a significant difference in
original reactor to consume all the sugar in the culture medium, but only biomass production between the two fermentation reactors. After
48 h were required for the internal-loop split ALR. This means that the glucose consumption in the culture medium was completed, the glucose
total sugar consumption rate increased from 0.81 g/L/h to 1.25 g/L/h. conversion rate in the original reactor was 0.087 g/g. The glucose

11
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Fig. 10. Fermentation physiological parameters. (a) The sugar consumption rate. (b) The conversion rate. (c) Protein content. (d) Biomass. (e) Glucose content. The
statistical significance of the above three sets of data was determined by t-tests; *P < 0.05, **P < 0.01, ****P < 0.0001.

conversion rate in the internal-loop split ALR was 0.145 g/g, which was
Table 4
66.67 % higher than the conversion rate in the original reactor. The
Research progress on the modification of airlift reactors for optimizing
presence of baffe in the improved fermentation tank increases the gas
fermentation.
content, thereby improving the mass transfer of oxygen to the liquid
References Optimization Results
phase and supporting higher microbial growth rates. As shown in Fig. 10
(c), the protein content produced by fermentation in the two reactors did (Huang Optimizing a circular gas The biomass of Chlorella
not change significantly and increased by only 3.96 %. To address this et al., distributor pyrenoidosa increased by 83.44
2017) %
limitation, future work will focus on optimizing the internal components
(Xue et al., A metal mesh was added to the The factual bioethanol
of airlift reactors, such as gas distributors and innovative backflow de­ 2020). airlift reactor conversion efficiency was 86.8
vices. These modifications aim to improve the fermentation process for %.
filamentous fungi, potentially leading to increased biomass yields and (Okabe The top and bottom of the The production rate of itaconic
et al., drainage tube are covered with acid in the optimized reactor is
more efficient fermentation cycles.
1993) stainless steel clamps twice that of traditional airlift
Compared to the fermentation of F. venenatum TB01 in a mechani­ reactors
cally stirred bioreactor employed by Tong et al. (Tong et al., 2022a), the (Ghosh et al., Reforming the riser of the The improved airlift reactor has
internal-loop split ALR exhibited a noticeable improvement in the sugar 2010) external circulation reactor into a 29 % increase in specific
consumption rate. However, the conversion rate falls below that ach­ a converging-diverging tube growth rate compared to the
original reactor
ieved in a mechanically stirred reactor. Therefore, for airlift reactors to
(Wu and Li, Replacing metal wire drainage This configuration produced BC
entirely replace mechanically stirred reactors for fermentation of 2015) tubes with mesh plates with net that can hold nearly 8-fold more
F. venenatum, further optimization of the airlift reactor is necessary. plates fluid than that produced under
Table 4 lists recent strategies for enhancing microbial fermentation in static fermentation conditions
could.
airlift reactors, focusing on modifications such as the incorporation of
(Ying et al., an airlift loop bioreactor The chlorophyll content has
metal mesh, gas distributors, and alterations to the drainage system. Our 2013) engaged with a fluidic increased by 6–8 times, and the
next steps will involve refining other components of the airlift reactor, oscillator to produce growth rate has increased by 20
with the aim of serving as a viable alternative to mechanically stirred microbubbles % -40 %
reactors, thereby reducing fermentation power consumption and mini­ This study Optimize the installation The sugar consumption rate
position of the baffle increased by 54.32 %, the
mizing costs.
conversion rate increased by
The CFD method has been widely used in reactor optimization, and 66.67 %.
Table 4 lists some of the research results. The results of this study
indicate that optimization has also been significantly improved. The

12
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

different improvement ratios in these literature may be due to differ­ Declaration of competing interest
ences in bacterial strains and reactor types, but overall they are all
improved. The authors declare the following financial interests/personal re­
lationships which may be considered as potential competing interests:
4. Conclusions Demao Li reports equipment, drugs, or supplies was provided by Tianjin
Institute of Industrial Biotechnology Chinese Academy of Sciences.
In this study, CFD technology was used to investigate the role of flow Demao Li reports a relationship with Tianjin Institute of Industrial
deflectors in gas liquid flow and mass transfer in an airlift reactor. The Biotechnology Chinese Academy of Sciences that includes: non-financial
influence of the structure and installation position of flow deflectors on support. There is no conflict of interest.
the inside of the reactor was studied, and the optimal position of the flow
deflectors in the reactor was determined. The optimized reactor was Acknowledgments
compared with the original reactor. The fermentation experiment of
F. venenatum TB01 showed that the sugar consumption rate in the This work was financially supported by the National Key Research
internal-loop split ALR increased by 54.32 %, and the conversion rate and Development Plan of China (2023YFD1201605) and the Belt &
increased by 66.67 %, providing a new perspective for the trans­ Road Fellowship Initiative of Tianjin Synthetic Biotechnology Innova­
formation and optimization of filamentous fungal fermentation reactors. tion Capacity Improvement Project (TSBICIP-BRFI-001).

Symbol table Appendix A. Supplementary data

Symbol Definition Unit Supplementary data to this article can be found online at [Link]
αg Gas holdup / org/10.1016/[Link].2025.102024.
H0 The stable height of the liquid level m
△H The height difference of the liquid level before and after m Data availability
inflation
C* The saturation concentration of dissolved oxygen mol/L
C The dissolved oxygen concentration mol/L No data was used for the research described in the article.
OUR The oxygen consumption rate mol⋅m− 3⋅s− 1

QNa The inflow rate of the Na2SO3 solution L/s References

MNa The concentration of the Na2SO3 solution mol/L
V The volume of water L Al, Ahmed, 2016. Study on Mixing Time and Gas Hold-Up in an Airlift Bioreactor. In: 3rd
kLa The volume-averaged mass transfer coefficients s− 1 International Conference on Civil, Environment and Waste Management (CEWM-
DCW The method of dry cell weight g/L 16).
M0 The weight of the filter paper g Anneli, R., HaKkinen, S.T., Mervi, T., Wiebe, M.G., 2017. Single Cell Protein—State-of-
M1 The dry weight of cells with weighing paper. g the-Art, Industrial Landscape and Patents 2001–2016. Front. Microbiol. 8, 2009..
ρ Density kg/m3 [Link]
P Pressure Pa Bach, C., Yang, J., Larsson, H., Stocks, S.M., Gernaey, K.V., Albaek, M.O., Krühne, U.,
uk
→ The velocity vector m/s 2017. Evaluation of mixing and mass transfer in a stirred pilot scale bioreactor
→ The shear stress tensor N/m2 utilizing CFD. Chem. Eng. Sci. 171, 19–26. [Link]
τk
ces.2017.05.001.
→
Fk Interaction force N
Berouaken, A., Rihani, R., Marra, F.S., 2023. Study of sparger design effects on the
FD Drag force N hydrodynamic and mass transfer characteristics of a D-shape hybrid airlift reactor.
CD Drag coefficient / Chem. Eng. Res. Des. 191, 66–82. [Link]
db The bubble diameter m Cai, M., Sun, X., Zhou, X., Zhang, Y., 2012. Roles of cobalt in biosynthesis stimulation of
Reb Reynolds number for bubble / a cytotoxic compound from marine-derived Aspergillus glaucu. Process Biochem. 47,
EO Eötvös number / 2267–2274. [Link]
g Gravitational Acceleration m/s2 Campestrini, L.H., Salles-Campos, C., 2021. Aspects of mushroom cultivation to obtain
σ Surface Tension N/m polysaccharides in submerged cultivation. AJB. 20, 100–107. [Link]
k Turbulent kinetic energy m2/s2 10.5897/AJB2020.17265.
Cerri, M.O., Futiwaki, L., Jesus, C.D.F., Cruz, A.J.G., Badino, A.C., 2008. Average shear
ϵ Turbulent kinetic energy dissipation rate m2/s3
rate for non-Newtonian fluids in a concentric-tube airlift bioreactor. Biochem. Eng.
μ Dynamic viscosity Pa⋅s
J. 39, 51–57. [Link]
μt Turbulent viscosity Pa⋅s
Chai, X., Zhao, X., 2012. Enhanced removal of carbon dioxide and alleviation of
Pk Production of turbulent kinetic energy due to mean m2/s3 dissolved oxygen accumulation in photobioreactor with bubble tank. Bioresour.
velocity gradients Technol. 116, 360–365. [Link]
Pkb, Peb Production terms due to buoyancy for k and ϵ m2/s3 Chen, Z., Jiang, Z., Zhang, X., Zhang, J., 2016. Numerical and Experimental Study on the
Cϵ1,Cϵ2 Empirical constants / CO2 Gas-Liquid Mass Transfer in Flat-plate Airlift Photobioreactor with Different
∇ Gradient operator m/s3 Baffles. Biochem. Eng. J. 106, 129–138. [Link]
kL liquid phase mass transfer coefficient m/s Chen, Z., Min, D., Hu, D., Zhang, L., Liu, W., 2020. Performance of a novel multiple draft
DL Diffusion coefficient m2/s tubes airlift loop membrane bioreactor to treat ampicillin pharmaceutical
a Specific interfacial area m2/m3 wastewater under different temperatures. Chem. Eng. J. 380, 122521. [Link]
Vb Bubble volume m3 org/10.1016/[Link].2019.122521.
Q The volumetric flow rate of the gas m3/s Chen, Z., Shen, W., Zhang, J., 2023. Computational fluid dynamics study of gas-liquid
mass transfer in concentric tube airlift reactor with helical baffle. Chem. Eng. Res.
Des. 198, 121–128. [Link]
Choi, K.H., 1995. Influence of the Gas-Liquid Separator Design on Hydrodynamic and
Mass transfer Characteristics in Bubble Column and Airlift Reactors. J. Chem.
Technol. Biotechnol. 62, 327–332. [Link]
Danckwerts, P.V., 1981. Significance of Liquid-Film Coefficients in Gas Absorption. Ind.
Eng. Chem. 43, 1460–1467. [Link]
CRediT authorship contribution statement Dou, B., Song, Y., 2010. A CFD approach on simulation of hydrogen production from
steam reforming of glycerol in a fluidized bed reactor. Int. J. Hydrogen Energy 35,
Jianyong Zhu: Writing – review & editing, Writing – original draft. 10271–10284. [Link]
Esperana, M., Cerri, M., Mazziero, V.T., Béttega, R., Badino, A.C., 2023. Hydrodynamic
Wuxi Chen: Writing – review & editing. Yuxiang Chen: Validation, comparison of different geometries of square cross-section airlift bioreactor using
Investigation. Farrukh Raza Amin: Resources, Methodology. Yaxiang computational fluid dynamics. Int. J. Chem. React. Eng. 21, 1291–1303. [Link]
Li: Validation, Investigation. Mengjia Lu: Writing – review & editing. org/10.1515/ijcre-2023-0010.

Demao Li: Writing – review & editing, Project administration, Funding

acquisition.

13
J. Zhu et al. Bioresource Technology Reports 29 (2025) 102024

Esperança, M.N., Béttega, R., Badino, A.C., 2017. Effect of geometric design on Neda, R., Coralie, H., Steven, W., Lukitawesa, L., Swarnima, A., Kamran, R., 2021.
performance of square cross-section concentric-duct and split airlift bioreactors. The. Filamentous fungus aspergillus oryzae for food: from submerged cultivation to
Can. J. Chem. Eng. 95, 2324–2332. [Link] fungal burgers and their sensory evaluation—a pilot study. Foods 10, 2774. https://
Ferreira, J.A., Varjani, S., Taherzadeh, M.J., 2020. A critical review on the ubiquitous [Link]/10.3390/foods10112774.
role of filamentous fungi in pollution mitigation. Curr. Pollut. Rep. 6, 295–309. Ojha, A., Al Dahhan, M., 2017. Local gas holdup and bubble dynamics investigation
[Link] during microalgae culturing in a split airlift photobioreactor. Chem. Eng. Sci. 157,
Finnigan, T., Needham, L., Abbott, C., 2017. Mycoprotein: a healthy new protein with a 185–198. [Link]
low environmental impact. Sustain. Protein Sources 305–325. [Link] Okabe, M., Ohta, N., Park, Y.S., 1993. Itaconic acid production in an air-lift bioreactor
10.1016/B978-0-12-802778-3.00019-6. using a modified draft tube. J. Ferment. Bioeng. 76, 117–122. [Link]
Ghosh, T.K., Banerjee, S., Majumder, C.B., Kim, T., 2010. Comparison of modified airlift 10.1016/0922-338X(93)90067-I.
reactor with conventional airlift reactor. Int. J. Control Autom. 3, 25–40. Patel, A.K., Agrawal, R., Dong, C.D., Chen, C.W., Singhania, R.R., Pandey, A., 2023. 7 -
Hernandez, D.A.Q., 2018. Influence of Fungal Morphology on the Performance of Filamentous fungal morphology in industrial aspects. Curr. Dev. Biotechnol. Bioeng.
Industrial Fermentation Processes for Enzyme Production. Technical University of 197–217. [Link]
Denmark, Kgs. Lyngby. Raimundo, P.M., Cloupet, A., Cartellier, A., Beneventi, D., Augier, F., 2019.
Huang, Q., Yang, C., Yu, G., Mao, Z.S., 2010. CFD simulation of hydrodynamics and mass Hydrodynamics and scale-up of bubble columns in the heterogeneous regime:
transfer in an internal airlift loop reactor using a steady two-fluid model. Chem. Eng. Comparison of bubble size, gas holdup and liquid velocity measured in 4 bubble
Sci. 65, 5527–5536. [Link] columns from 0.15 m to 3 m in diameter. Chem. Eng. Sci. 198, 52–61. [Link]
Huang, Y, Zhao, S., Ding, Y.D., Liao, Q., Huang, Y, Zhu, X., 2017. Optimizing the gas org/10.1016/[Link].2018.12.043.
distributor based on CO2 bubble dynamic behaviors to improve microalgal biomass Salehpour, R., Jalilnejad, E., Nalband, M., Ghasemzadeh, K., 2020. Hydrodynamic
production in an air-lift photo-bioreactor. Bioresour. Technol. 233, 84–91. https:// behavior of an airlift reactor with net draft tube with different configurations:
[Link]/10.1016/[Link].2017.02.071. Numerical evaluation using CFD technique. Particuology. 51, 91–108. [Link]
Imai, Y., Takei, H., Matsumura, M., 1987. A Simple Na2SO3 Feeding Method for KLa org/10.1016/[Link].2019.09.005.
Measurement in Large-Scale Fermentors. Biotechnol. Bioeng. 29, 982–993. https:// Stavi, I., Pinho, J.R.D., Paschalidou, A.K., Adamo, Galvin, K., Sherbinin, A.D., Even, T.,
[Link]/10.1002/bit.260290810. Heaviside, C., Geest, K.V.D., 2022. Food security among dryland pastoralists and
Johansen, S.T., Robertson, D.G.C., Woje, K., Engh, T.A., 1988. Fluid dynamics in bubble agropastoralists: the climate, land-use change, and population dynamics nexus.
stirred ladles: Part II. Mathematical modeling. Metall. Trans. B 19, 745–754. https:// Anthropocene Rev. 9, 299–323. [Link]
[Link]/10.1007/BF02650195. Tomiyama, A., 1998. Struggle with computational bubble dynamics. Multiph. Sci.
Khurana, A.K., Kumar, R., 1969. Studies in bubble formation — III. Chem. Eng. Sci. 24, Technol. 10, 369–405. [Link]
1711–1723. [Link] Tong, S., An, K., Chen, W., Zhou, W., Sun, Y., Wang, Q., Li, D., 2022a. Evasion of Cas9
Kilonzo, P, Margaritis, A, 2006. Influence of the baffle clearance design on toxicity to develop an efficient genome editing system and its application to increase
hydrodynamics of a two riser rectangular airlift reactor with inverse internal loop ethanol yield in Fusarium venenatum TB01. Appl. Microbiol. Biotechnol. 106,
and expanded gas–liquid separator. Chem. Eng. J. 121, 17–26. [Link] 6583–6593. [Link]
10.1016/[Link].2006.05.003. Tong, S., An, K., Zhou, W., Chen, W., Sun, Y., Wang, Q., Li, D., 2022b. Establishment of
Laborde-Boutet, C., Larachi, F., Dromard, N., Delsart, O., Schweich, D., 2009. CFD High-Efficiency Screening System for Gene Deletion in Fusarium venenatum TB01.
simulation of bubble column flows: Investigations on turbulence models in RANS JoF 8, 169. [Link]
approach. Chem. Eng. Sci. 64, 4399–4413. [Link] Tong, S., An, K., Chen, W., Chai, M., Sun, Y., Wang, Q., Li, D., 2023. Identification of
ces.2009.07.009. neutral genome integration sites with high expression and high integration efficiency
Li, X., Zhan, X., 2020. Application of airlift reactor with helical sieve plate in fungal in Fusarium venenatum TB01. Synth. Syst. Biotechnol. 8, 141–147. [Link]
fermentation. Food Ferment. Ind. 46, 93–99. [Link] 10.1016/[Link].2022.12.006.
1802/ts.023704. Tong, S., Chen, W., Hong, R., Chai, M., Sun, Y., Wang, Q., Li, D., 2023. Efficient
Li, X., Chen, Y., Zheng, Z., Gao, M., Wang, Z., Zhang, K., Liu, H., Zhan, X., 2020. Power- mycoprotein production with low CO2 emissions through metabolic engineering and
saving airlift bioreactor with helical sieve plates: Developmental and performance fermentation optimization of Fusarium venenatum. J. Agr. Food Chem. 72, 604–612.
studies. Chem. Eng. Res. Des. 158, 1–11. [Link] [Link]
cherd.2020.03.014. Werger, C., 2022. European Food Security; Calamities to the Supply of Soy.
Liew, S.Y., Gimbun, J., 2017. CFD simulation on the hydrodynamics in gas-liquid airlift Wiebe, M., 2002. Myco-protein from Fusarium venenatum: a well-established product for
reactor. Chem. Prod. Process. Model. 12, 20170030. [Link] human consumption. Appl. Microbiol. Biotechnol. 58, 421–427. [Link]
2017-0030. 10.1007/s00253-002-0931-x.
Liu, J., 2023. Recent advances in Chitin Biosynthesis Associated with the Morphology Wu, S.C., Li, M.H., 2015. Production of bacterial cellulose membranes in a modified
and secondary Metabolite Synthesis of Filamentous Fungi in Submerged airlift bioreactor by Gluconacetobacter xylinus. J. Biosci. Bioeng. 120, 444–449.
Fermentation. J. Fungi 9, 205. [Link] [Link]
Liu, H., Li, Z., Geng, S., Gao, F., He, T., Huang, Q., 2020. Influences of top clearance and Xu, Y., Luo, L., Yuan, J., 2011. CFD simulations to portray the bubble distribution and the
liquid throughput on the performances of an external loop airlift slurry reactor hydrodynamics in an annulus sparged air-lift bioreactor. Can. J. Chem. Eng. 89,
integrated mixing and separation. Chin. J. Chem. Eng. 28, 1514–1521. [Link] 360–368. [Link]
org/10.1016/[Link].2020.03.027. Xu, C., Liu, X., Gu, X., 2023. Computational studies of airlift and stirred airlift bioreactors
Luo, H.P., Al-Dahhan, M.H., 2008. Local characteristics of hydrodynamics in draft tube with non-Newtonian fluid: a comparison of hydrodynamics and gas-liquid mass
airlift bioreactor. Chem. Eng. Sci. 63, 3057–3068. [Link] transfer. J. Environ. Chem. Eng. 11, 110800. [Link]
ces.2008.03.015. jece.2023.110800.
Luo, H.P., Al-Dahhan, M.H., 2012. Airlift column photobioreactors for Porphyridium sp. Xue, D., Yao, D., Sukumaran, R.K., You, X., Gong, C., 2020. Tandem integration of
culturing: Part II. Verification of dynamic growth rate model for reactor performance aerobic fungal cellulase production, lignocellulose substrate saccharification and
evaluation. Biotechnol. Bioeng. 109, 942–949. [Link] anaerobic ethanol fermentation by a modified gas lift bioreactor. Bioresour. Technol.
Luo, L., Liu, F., Xu, Y., Yuan, J., 2011. Hydrodynamics and mass transfer characteristics 302, 122902. [Link]
in an internal loop airlift reactor with different spargers - ScienceDirect. Chem. Eng. Ying, K., Gilmour, D.J., Shi, Y., Zimmerman, W.B., 2013. Growth Enhancement of
J. 175, 494–504. [Link] Dunaliella salina by Microbubble Induced Airlift Loop Bioreactor (ALB)—the
Mcintyre, M., Müller, C., Dynesen, J., Nielsen, J., 2001. Metabolic engineering of the Relation between Mass transfer and Growth Rate. J. Biomater. Nanobiotechnol. 4,
morphology of Aspergillus. Adv. Biochem. Eng. Biotechnol. 73, 103–128. [Link] 1–9. [Link]
org/10.1007/3-540-45300-8_6.

14