Separation and Purification Technology 295 (2022) 121203

Contents lists available at ScienceDirect

Separation and Purification Technology

journal homepage: www.elsevier.com/locate/seppur

Experimental and numerical analysis of Chinese hamster ovary cell viability

loss in mini-hydrocyclones
Liqun He a, b, Li Ji a, b, *, Yujing He c, *, Yao Liu a, b, Songying Chen a, b, Kaiwei Chu d, Shibo Kuang e
a
Key Laboratory of High Efficiency and Clean Mechanical Manufacture, Ministry of Education, School of Mechanical Engineering, Shandong University, Jinan 250061,
China
b
National Demonstration Center for Experimental Mechanical Engineering Education, Shandong University, Jinan 250061, China
c
School of Pharmaceutical Sciences, Shandong First Medical University & Shandong Academy of Medical Sciences, Taian 271000, China
d
School of Qilu Transportation, Shandong University, Jinan 12550, China
e
ARC Research Hub for Computational Particle Technology, Department of Chemical Engineering, Monash University, Clayton, Victoria 3800, Australia

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

Keywords: Mini-hydrocyclones have been increasingly used as retention devices to separate mammalian cells in the
Mini-hydrocyclones continuous cell culture. However, the cell viability loss after running through mini-hydrocyclones varies even for
Cell viability the same cell, e.g., the most widely used Chinese hamster ovary (CHO) cells. The reasons behind this problem are
Shear stress
unclear. This paper presents a quantitative analysis of the CHO cell viability loss in mini-hydrocyclones with
Quantitative
respect to the magnitude of shear stress. The viability loss of CHO cells is measured experimentally. The dis­
Analysis
tributions of shear stress magnitude are revealed using a validated numerical model. The experimental results
show that the cell viability loss increases with the increase of hydrocyclone cone angle or the inlet velocity. This
result can be explained by the increased shear stress magnitude in the mini-hydrocyclones.

retention devices due to their simple geometry design, low operation

costs and high capacity [3,4,12–17].
1. Introduction
Even though reasonably well retention (separation) efficiency (as
high as 99%) can be obtained by mini-hydrocyclones, the cell viability
Mammalian cell culture plays a dominant role in the production of
(the proportion of living cells in the total) after separation is diverse
therapeutic proteins, antibodies, hormones, vaccines, etc. [1,2].
even for the same mammalian cell. For example, Chinese hamster ovary
Continuous culture (perfusion process) has been increasingly used due
(CHO) cells, which take up 70% of the host cells used in the mammalian
to its low cost, high product quality and compatibility with continuous
cell culture [18,19], were found to have different cell viabilities (70%-
processing [3–5]. In the perfusion culture, cell retention devices are
99%) [3,15,20,21] after running through mini-hydrocyclones. For the
employed to separate cells from the exhausted medium, allowing the
convenience of comparison, the viability loss (the difference in cell
metabolites removal (including lactic acid and ammonia, etc.) and thus
viability between the inlet and the underflow outlet) is introduced to
obtaining high productivity [6]. As cell retention devices, they should
characterize the effect of hydrocyclone separation on the cell viability.
present features such as high retention efficiency, low damage to cell
The viability loss of CHO cells is found between 2.9% and 14.4% in the
viability and sufficient capacity.
studies of Pinto et al. [14,22]. A much lower cell viability loss (≤4%)
Different techniques have been used in cell retention, e.g. sedimen­
was reported by Bettinardi et al. [4]. Also, a low cell viability loss (≤5%)
tation [7], filtration [8,9], centrifugation [10] and microfluidic methods
was reported recently by Syed et al. [3]. In addition, Kundu and Hiller
[3,11]. However, they perform with the following limitations.
compared the perfusion process with and without mini-hydrocyclones.
Sedimentation-based apparatuses tend to trigger cell adhesion and have
They found that the former CHO cell viability is 1–8% lower than the
a long operation time. Filtration-based techniques face problems of filter
latter, indicating the cell viability loss as a result of hydrocyclone sep­
fouling even after a relatively short period of operation. Centrifuges
aration may be 1–8% [21]. There are also studies reporting the cell
suffer from cell adhesion, clogging, inactivation and high costs. The
viability instead of the cell viability loss. For example, Elsayed et al. [23]
microfluidics has low throughput (up to a few microliters per minute).
revealed that the CHO cell viability was always maintained at more than
Recently, mini-hydrocyclones have been proposed to perform as

* Corresponding authors.
E-mail addresses: [email protected] (L. Ji), [email protected] (Y. He).

https://doi.org/10.1016/j.seppur.2022.121203
Received 18 February 2022; Received in revised form 27 April 2022; Accepted 30 April 2022
Available online 6 May 2022
1383-5866/© 2022 Elsevier B.V. All rights reserved.
L. He et al. Separation and Purification Technology 295 (2022) 121203

Nomenclature CU Cell concentration in overflow (cells/ml)

CU Cell concentration in underflow (cells/ml)
DC Diameter of the body (mm) CI Cell concentration in inlet (cells/ml)
DU Diameter of spigot (mm) Nviable cell, I Number of viable cells in the inlet (cells)
DO Diameter of vortex finder (mm) Nviable cell, U Number of viable cells in the underflow (cells)
DI Diameter of inlet (mm) Ncell, I Total number of cells in the inlet (cells)
L Length of cylindrical part (mm) Ncell, U Total number of cells in the underflow(cells)
LC Length of cone part (mm) η Kolmogorov length scale (m)
LV Length of vortex finder (mm) v Kinematic viscosity (m2/s)
T Thickness of vortex finder (mm) ε Dissipation rate (m2/s3)
α Cone angle (◦ ) ρ Density (kg/m3)
VI Inlet velocity (m/s) µ Dynamic viscosity (Pa•s)
E Efficiency (%) U Velocity components in the X direction (m/s)
S Split ratio (%) V Velocity components in the Y direction (m/s)
R Amount of water split to underflow (%) W Velocity components in the Z direction (m/s)
VO Volume of suspension obtained in overflow (ml) τ Total shear stress (Pa)
VU Volume of suspension obtained in underflow (ml) τv Laminar viscous shear stress (Pa)
VI Volume of suspension fed through the inlet (ml) τt Turbulent-induced shear stress (Pa)

90% in the bioreactor with hydrocyclone as the retention device.

It is generally known that mammalian cells lack a cell wall and are
very sensitive to shear stress[4,24,25]. However, the explicit reasons for
the cell viability loss still remain unknown. Previous studies reported
that a large magnitude of shear stress could cause cell viability loss as
well as the flocs breakage[3,26] in hydrocyclones. However, it is chal­
lenging to measure the magnitude of shear stress in hydrocyclones. As a
result, very rare studies have been found to report the magnitude of
shear stress and its variation mechanism in hydrocyclones. Computa­
tional methods provide an alternative to gaining an insight into the
shear stress magnitude in hydrocyclones [27] as well in other industrial
machinery [28–30]. For example, Xu et al. numerically investigated the
distribution of laminar viscous shear stress in a 35-mm diameter deoil­
ing hydrocyclones. Note mechanisms explaining cell death in the pres­
ence of shear stress include turbulent eddy-cell interaction, laminar
viscous shear stress and cell–cell collision [27,31]. The cell–cell
collision-induced shear stress could be ignored if the flow is dilute.
However, the turbulent-induced shear stress is supposed to have a
physical effect [32,33] on cells when the scale of the smallest vortices
(Kolmogorov length) approaches the scale of cells [31,32]. Thus,
consideration of the turbulent-induced shear stress is necessary but
lacking for hydrocyclone in which swirling turbulence exists. Instead of
investigating the distribution of shear stress, some researchers briefly
mentioned the variation of shear stress magnitude with specific pa­
rameters in hydrocyclones. For example, in a sentence, Syed et al. and
Elsayed et al. proposed that high inlet flow rates of hydrocyclones in­
crease the magnitude of shear stress [3,15]. Also in a sentence, Liu et al.
mentioned that small cone angles of hydrocyclone reduce the axis ve­
locity gradient and thus decrease the shear stress magnitude [34].
Nevertheless, the aforementioned variations of shear stress in hydro­
cyclones were briefly mentioned in a qualitative way, lacking quanti­
tative and convincing analysis and verification.
In addition to the shear stress magnitude, the exposure time to shear
stress [24] and the operating pressure were reported to be also relevant
to the viability of mammalian cells in some studies [4,13,35]. As the
residence time of mammalian cells in mini-hydrocyclones is very short
(in the range of fractions of seconds), the effect of exposure time to shear
stress has not usually been considered [13,14]. The effect of operating
Fig. 1. Geometric profile of the mini-hydrocyclone.
pressure on the cell viability loss was reported to be minor, provided that
the operating pressure remains within a certain range[21,35,36].
This work presents an attempt to analyze the cell viability loss in mini-hydrocyclones is considered in this work. The cell separation effi­
mini-hydrocyclones with respect to the quantified shear stress magni­ ciency, cell viability loss and the amount of water split of mini-
tude by combining numerical modelling and experiments. In addition to hydrocyclones are measured experimentally. A CFD-LPT (Computa­
the laminar viscous shear stress, the turbulent-induced shear stress in tional Fluid Dynamics-Lagrange Particle Tracking) model is developed

2
L. He et al. Separation and Purification Technology 295 (2022) 121203

Table 1 mean deviation (Ra) of the printing surface roughness is 3.2 μm, ac­
Geometric and operating parameters of the mini-hydrocyclone. cording to the technique report provided by the supplier. To study the
Parameter Symbol Value effect of hydrocyclone cone angle and inlet velocity, mini-hydrocyclones
with different cone angles (6◦ , 8◦ and 10◦ ) are investigated at an inlet
Diameter of the body (mm) DC 10
Diameter of spigot (mm) DU 2 velocity of 6 m/s and those with different inlet velocities (4 m/s, 6 m/s
Diameter of vortex finder (mm) DO 2 and 7.5 m/s) are investigated at a cone angle of 8◦ .
Diameter of inlet (mm) DI 2 CHO.K1 cell line is cultured in a serum-free suspension with com­
Length of cylindrical part (mm) L 8 mercial media (CHO GROW CD2, iCell Bioscience Inc., China). The
Length of cone part (mm) LC 76.3,57.2 and 45.7
Length of vortex finder (mm) LV 5
culture is kept at 37 ◦ C in the atmosphere containing 5% (v/v) CO2. The
Thickness of vortex finder (mm) T 1 PBS buffer solution is used to prepare the cell suspension to the con­
Cone angle (◦ ) α 6, 8 and 10 centration of 4 × 105 cells/ml (0.0067% v/v). The experimental pro­
Inlet velocity (m/s) VI 4, 6 and 7.5 cedure is demonstrated in Fig. 2. Specifically, the cell suspension is fed
into the mini-hydrocyclone by a peristaltic pump (LONGER DG15). The
internal diameter of the tube is 4.8 mm. The feed speed in the tube is
and validated by the measured cell separation efficiency and the amount
0.69, 1.04 and 1.30 m/s, respectively corresponding to the inlet feed
of water split. On the basis of that, the shear stress distribution in the
velocity of 4, 6 and 7.5 m/s. The concentration variation as a result of
mini-hydrocyclones is obtained by the validated numerical model. Last,
the pulsed flow of the peristaltic pump is ignored in this work. The in­
the effect of the hydrocyclone inlet flowrate (inlet velocity) and cone
fluence of peristaltic pumping system, which was reported minor on the
angle on the shear stress magnitude and cell viability loss are carefully
cell viability [14,36], is also ignored in this work. Different hydro­
analyzed.
cyclone inlet velocities (flowrates) are obtained by adjusting the pump
rotation speed. Samples are respectively taken from the underflow
2. Method and material
(spigot) and overflow (vortex finder) after the internal flow field of the
mini-hydrocyclone is stable. The sample volume is measured to calcu­
2.1. Design of experiment
late the split ratio. The cell concentration in the inlet, underflow and
overflow is determined by cell counting under the microscope. Based on
Fig. 1 shows the geometry of the mini-hydrocyclone and Table 1 lists
that, the separation efficiency of CHO cells is calculated. The cell
the corresponding geometric values. The mini-hydrocyclone is manu­
viability is determined by the traditional Trypan blue exclusion test.
factured by a 3D printer (Formlabs Form 3) using high temperature-
Specifically, the dead cells would be colored blue by Trypan blue solu­
resistant resin. The layer thickness of 100 μm is selected for compre­
tion while the viable cells would not. This difference can be observed
hensive consideration of precision and printing time. The arithmetical
under the microscope and the ratio of the number of viable cells (i.e., the

Fig. 2. Experiment procedure of CHO cell separation by the mini-hydrocyclone.

3
L. He et al. Separation and Purification Technology 295 (2022) 121203

Fig. 3. CHO cell separation and viability loss in mini-hydrocyclones.

Fig. 4. The distribution of Kolmogorov length scale in the mini-hydrocyclone
with a cone angle of 8◦ at an inlet velocity of 6 m/s.
viability) in the inlet and underflow suspension can be calculated, as
shown in Fig. 3. On the basis of that, the viability loss can be determined.
of water–air mixture is modeled using the Reynolds stress model (RSM),
The separation efficiency, split ratio and viability loss are calculated by
facilitated with a standard wall function. The effect of wall roughness is
Equations (1)-(4).
ignored in this study. The interface between the liquid and air is tracked
The split ratio is expressed as,
by the Volume of Fluid (VOF) model. The calculation duration is 10 s
VU which is far more than the residence time of cells in the mini-
S= (1)
VU + VO hydrocyclone. As the feed solid concentration is low (less than4 % by
volume), the trajectories of CHO cells in the mini-hydrocyclone are
where VO and VU represent the volume of suspension obtained in the
tracked by the LPT method, as introduced in detail elsewhere [38–45].
overflow and underflow, respectively.
Note the dense discrete phase model, which combines LPT with the ki­
The water split is expressed as,
netic theory [46], can in principle be applied to the operations with high
VU (1 − CU ) feed solid concentrations. It is well measured that the CHO cell size is
R= (2)
VI (1 − CI ) around 12 μm while the density is around 1 g/cm3 [3,47–51]. In the
simulations, the CHO cells are of a uniform density at 1.19 g/cm3 and a
where VI represents the volume of suspension in the inlet, VI = VU +
uniform size of 10 μm, as done elsewhere [31,45,52].
VO. CU and CI represent the cell concentration in the underflow and inlet,
As the flow in the considered hydrocyclones is dilute, cell–cell
respectively. When the values of CU and CI are small, R is approximate to
collision-induced shear stress is ignored in this work. The Kolmogorov
S.
length scale in the considered hydrocyclone (cone angle of 8◦ at an inlet
The separation efficiency is expressed as,
velocity of 6 m/s), calculated by Equation (5), is shown in Fig. 4. Note
VU C U that other mini-hydrocyclones in this work have the same Kolmogorov
E= (3)
VU CU + VO CO length scale. It can be seen that the Kolmogorov length scale is close to
the diameter of CHO cells. Therefore, in addition to the laminar viscous
where CO represents the cell concentration in overflow.
shear stress [27], the turbulent-induced shear stress (Equation (6)) is
The cell viability loss is expressed as,
supposed to have a physical effect [32,33] on cells and thus be consid­
Viability loss =
Nviablecell,I Nviablecell,U
− (4) ered in this work. The laminar viscous shear stress is calculated by
Ncell,I Ncell,U Equation (7) [30,53–56].
where Nviable cell, I and Nviable cell, U represent the number of viable cells v3
in the inlet and underflow, Ncell, I and Ncell, U represent the total number η = ( )0.25 (5)
ε
of cells in the inlet and underflow, respectively. √̅̅̅̅̅̅̅̅̅̅̅̅̅̅
τt = 0.5ρμε (6)
⎡ ⎤
2.2. Numerical simulation ∂u ∂u ∂u ∂v ∂u ∂w
⎢ ∂x + ∂x +
∂y ∂x
+
∂z ∂x ⎥
⎢ ⎥
The computational domain is divided into hexahedral grids using ⎢
⎢ ∂v ∂u ∂v ∂v ∂v ∂w ⎥
⎥
ICEM CFD 18.2. The grid independence test was carried out in our τ v = μ⎢
⎢ ∂x + ∂y +
∂y ∂y
+ ⎥
∂z ∂y ⎥
⎢ ⎥
previous study [37]. The number of grids is selected to be 200,000 to ⎢ ⎥
⎣ ∂w ∂u ∂w ∂v ∂w ∂w ⎦
ensure both sufficient computation accuracy and low computation costs. + + +
The simulation is carried out on ANSYS Fluent 18.2. The turbulent flow ∂x ∂z ∂y ∂z ∂z ∂z

4
L. He et al. Separation and Purification Technology 295 (2022) 121203

The amount of water split to underflow (%)

The amount of water split to underflow (%)

60 80 60
Efficiency Efficiency
80
Water split Water split
Separation efficiency (%)

Separation efficiency (%)

75
70 58 58

70
60
56 56
50 65

40 54 60 54
6 8 10 4 6 7.5
Cone angle (º) Inlet velocity (m/s)

(a) (b)
Fig. 5. The experimental separation efficiency and amount of water split for mini-hydrocyclone with (a) cone angles of 6◦ , 8◦ and 10◦ at an inlet velocity of 6 m/s
and (b) with a cone angle of 8◦ at inlet velocities of 4, 6 and 7.5 m/s.

15 15

12 12
Cell viability loss (%)
Cell viability loss (%)

9 9

6 6

3 3

0 0
6 8 10 4 6 7.5
Cone angle (º) Inlet velocity (m/s)

(a) (b)
Fig. 6. Cell viability loss after running through the mini-hydrocyclones corresponding to Fig. 5.

⎡ ( ⎤ )0.5 hydrocyclone. For the inlet velocity varying between 4 and 7.5 m/s, as
τxx τxy τxz 1 ∑( )2 ∑ 2
= ⎣ τyx τyy τyz ⎦ = μ τii − τjj + τij (7) shown in Fig. 5 (b), as the inlet velocity increases, the separation effi­
τzx τzy τzz 6 i∕
=j i∕
=j ciency increases, but the water split decreases. This is consistent with the
finding in previous studies on a 5 mm diameter mini-hydrocyclone [59]
τ = τt + τ v (8) and a 75 mm diameter hydrocyclone [60].
Fig. 6 shows the viability loss of the cells after running through the
where η is the Kolmogorov length scale, v is kinematic viscosity, ε is considered mini-hydrocyclones. The maximum standard deviations for
the dissipation rate, ρ is the liquid density, µ is the liquid dynamic vis­ Fig. 6 (a) and (b) are respectively 1.4 % and 0.8 %, indicating the effect
cosity, U, V and W are the velocity components in the X, Y and Z di­ of experiment uncertainty is minor. It can be seen that an increase in the
rections, respectively. hydrocyclone cone angle causes a rise in the cell viability loss, reaching
the maximum value of 11.3% at the cone angle of 10 , as shown in Fig. 6
○

3. Result and discussion (a). Liu et al.[34] reported a similar finding that a small cone angle could
avoid over-shear force which destroys microorganisms. From Fig. 6 (b),
3.1. Experiment result it can be seen that the cell viability loss decreases with the increase of
inlet velocity, which is consistent with the finding by Syed et al. [3].
Fig. 5 (a) and (b) show the measured separation efficiency and the Note the peristaltic pumping system (including the peristaltic pump and
amount of water split to underflow for the three mini-hydrocyclones the tube) also affect the cell viability. However, less than 1 % cell
with different cone angles and different inlet velocities, respectively. viability drops even for the considered pumping system with the largest
The error bars for the measured separation efficiency and the amount of tube feed speed of 1.30 m/s. This is consistent with the reported finding
water split are also shown in Fig. 5. The maximum standard deviations that the influence of the peristaltic pumping system is minor [14,36].
are 3.9 % and 0.4 %, respectively, indicating the effect of experiment To show the visibility loss more clearly, Fig. 7 shows the observed
uncertainty is minor. It can be seen from Fig. 5 (a) that the separation viable and dead cells under the microscope for the samples at the inlet
efficiency and the water split to underflow decrease with the increase of before the experiment and the underflow after running through mini-
the cone angle. A consistent finding was reported by Saidi et al.[57] and hydrocyclones. To be representative, each figure, e.g., the inlet one in
Fu et al.[58] who studied the effect of cone angle on the separation ef­ Fig. 7 (a), includes four directly observed visions at different locations
ficiency of a 35 mm diameter hydrocyclone and a 68 mm diameter

5
L. He et al. Separation and Purification Technology 295 (2022) 121203

inlet underflow
(a)

inlet underflow
(b)
Fig. 7. CHO cells under the microscope for the samples at the inlet and underflow for the mini-hydrocyclones with (a) a cone angle of 8◦ at an inlet velocity of 7.5 m/
s and (b) a cone angle of 10◦ at an inlet velocity of 6 m/s.
The amount of water split to underflow (%)

The amount of water split to underflow (%)

Experiment efficiency Simulated efficiency Experiment efficiency Simulated efficiency 100

100 100 100
Experiment water split Simulated water split Experiment water split Simulated water split
Separation efficiency (%)
Separation efficiency (%)

80 80 80 80

60 60 60 60

40 40 40 40

20 20 20 20

0 0 0 0
6 8 10 4 6 7.5
Cone angle (º) Inlet velocity (m/s)

(a) (b)
Fig. 8. Comparison of experimental and simulated data for mini-hydrocyclones corresponding to Fig. 5.

under the microscope. As the light may slightly vary with locations, the underflow increases compared to that in the inlet, indicating the cell
background color of the observed visions may be different. As afore­ viability reduces due to the mini-hydrocyclone separation. In addition,
mentioned, the dead cells are dyed blue, as can also be observed in the cell concentration in the underflow is denser than that in the inlet,
Fig. 7. It can be seen that the concentration of dead cells in the indicating the considered mini-hydrocyclones are applicable in the CHO

6
L. He et al. Separation and Purification Technology 295 (2022) 121203

300
400

275

Pressure drop (kPa)

Pressure drop (kPa)
300

250 200

225 100

200 0
6 8 10 4 6 7.5
Cone angle (º) Inlet vlocity (m/s)

(a) (b)
Fig. 9. Simulated pressure drop for mini-hydrocyclones corresponding to Fig. 5.

Fig. 10. Tangential velocity distribution for mini-hydrocyclones corresponding to Fig. 5.

cell separation. quantitatively.

The pressure drop, which reflects the energy consumption of
hydrocyclones, is also an important separation performance indicator in
3.2. Simulation result
addition to the separation efficiency and water split. Note a preliminary
experiment setup is used in this work and the measurement of pressure
Fig. 8 compares the experimental and simulated results for mini-
drop is currently not included. An improved experiment setup with a
hydrocyclones with cone angles of 6◦ , 8◦ and 10◦ at an at inlet veloc­
manometer or pressure sensor would be included in future work.
ity of 6 m/s and those at the inlet velocities of 4, 6 and 7.5 m/s with a
Nevertheless, the simulated pressure drop of those mini-hydrocyclones
cone angle of 8◦ . As shown in Fig. 8, the variation trends of the simulated
is provided and shown in Fig. 9. Obviously, the pressure drop of the
data are consistent with those of experimental ones. Furthermore, the
mini-hydrocyclone increases with the increase of the cone angle, which
relative errors between the experimental and simulated results are
is in line with the finding on a deoiling hydrocyclone by Saidi et al. [57].
within 20%. The consistent variation trend and the relatively acceptable
The increase of the inlet velocity also contributes to a significant rise in
error validate the developed numerical model qualitatively and

7
L. He et al. Separation and Purification Technology 295 (2022) 121203

Fig. 11. Radial velocity distribution for mini-hydrocyclones corresponding to Fig. 5.

the pressure drop. The pressure drop at the inlet velocity of 7.5 m/s is maximum pressure drop (271 kPa) in Fig. 9 (a) is much lower than that
four times that at the inlet velocity of 4 m/s. Previous studies on large (397 kPa) in Fig. 9 (b). However, the corresponding maximum cell
hydrocyclones also reported the consistent variation trend of pressure viability loss (11.3%) in Fig. 6 (a) is larger than that (9.3%) in Fig. 6 (b).
drop with the inlet velocity (inlet flowrate) [60,61]. Note that the This should indicate that the pressure drop is not the sole factor causing

0.93% 1.43% 1.68%

2.41%
3.38% 4.43%
12.74%
18.17%
30.78%
83.92% 77.02% 63.11%

Cone angle 6° Cone angle 8° Cone angle 10°

(a)
0.93% 1.43% 1.68%
2.41%
3.38% 4.43%
12.74%
18.17%
30.78%
83.92% 77.02% 63.11%

Cone angle 6° Cone angle 8° Cone angle 10°

(b)
Fig. 12. Distribution of shear stress magnitude in the whole computational domain of the mini-hydrocyclones corresponding to Fig. 5.

8
L. He et al. Separation and Purification Technology 295 (2022) 121203

Fig. 13. Distribution of (a) laminar viscous shear stress, (b) turbulent-induced shear stress and (c) total shear stress of the mini-hydrocyclone with (I) different cone
angles and (II) different inlet velocities.

the cell viability loss. Such a representation has been used in other studies[62–68] and offers
Fig. 10 shows the variation of tangential velocity distribution with convenience for better showing the flow properties that govern the
the cone angle and inlet velocity. The blank areas represent the air core separation process. Note that the air core is not fully formed in the mini-
whose properties are very different from those of liquid and solid phases. hydrocyclones studied in this work, which is consistent with the

9
L. He et al. Separation and Purification Technology 295 (2022) 121203

the CHO cell trajectory are demonstrated in Figs. 12, 13 and 15,
respectively. The shear stress in Figs. 12, 13 and 15 represent the total
shear stress, which is the sum of the laminar viscous shear stress and the
turbulent-induced shear stress. Fig. 12 describes the percentage for each
shear stress interval in the whole computational domain. Apparently,
the shear stress magnitude between 0 and 30 Pa takes the largest pro­
portion, except the mini-hydrocyclone with an inlet velocity of 7.5 m/s.
Comparatively, the shear stress magnitude above 100 Pa takes the
smallest proportion. Furthermore, as the cone angle or the inlet velocity
increases, the proportion of the shear stress of 0–30 Pa decreases while
the proportion of other levels larger than 30 Pa increases, indicating the
shear stress magnitude is rising in the whole computational domain.
Fig. 13 demonstrates how the laminar viscous shear stress, turbulent-
induced shear stress and total shear stress distribute in the vertical plane
of mini-hydrocyclones. It can be seen from Fig. 13 that the laminar
viscous shear stress is stronger along the hydrocyclone central axis,
especially in the lower conical section. Comparatively, the turbulent-
induced shear stress is more violent under the bottom of the vortex
finder and on the wall of the lower conical section. As a result, the total
Fig. 14. CHO cell trajectory inside the mini-hydrocyclone. shear stress, which is the collective effect of the laminar viscous shear
stress and the turbulent-induced shear stress, is observed stronger under
reported simulation. the bottom of the vortex finder and in the lower conical section.
results of mini-hydrocyclones elsewhere [69–71]. It can be seen from Furthermore, as the cone angle or the inlet velocity increases, both the
Fig. 10 (a) that the tangential velocity increases with the increase of laminar viscous shear stress and turbulent-induced shear stress increase,
cone angle. This was also reported for large hydrocyclones in the liter­ although the increase of laminar viscous shear stress with the increase of
ature[57,58]. A larger tangential velocity is supposed to contribute to cone angle is minor. As a result, the magnitude of total shear stress on
higher separation efficiency. Also, a larger amount of water split to the vertical hydrocyclone plane rises.
underflow could entrain more fine particles to underflow in mini- Fig. 14 shows the trajectory of a CHO cell in the mini-hydrocyclone
hydrocyclones [37]. However, the contribution to separation effi­ with a cone angle of 6◦ at an inlet velocity of 6 m/s. As the size of the
ciency due to the increased tangential velocity may be compromised by CHO cell is considered uniform, the trajectory of a random cell particle
the reduced amount of water split shown in Figs. 5 and 8, leading to a fed from the inlet and discharged from the spigot is taken as an example.
reduction in separation efficiency for mini-hydrocyclones with larger Note the residence time of CHO cells in the considered mini-
cone angles. Fig. 10 (b) demonstrates that the inlet velocity apparently hydrocyclones is within 0.08 s, which is supposed not to cause a sig­
increases the tangential velocity, which is consistent with the finding on nificant cumulative effect on cell viability loss [13,14]. Fig. 15 shows the
the inlet velocity for large hydrocyclones [60,72]. The apparent increase shear stress exerting on the cell particle along the trajectory corre­
in the tangential velocity should contribute to higher separation effi­ sponding to Fig. 14 in the mini-hydrocyclones. The cell is located by its
ciency for the mini-hydrocyclones with larger inlet velocities, which relative axial position (Z position), i.e., the axial position to the total
compensates for the effect brought by the slightly reduced amount of axial length of hydrocyclone. The value of 0 means indicates the inlet,
water split. As a result, the separation efficiency experiences an upward while the value of 1 indicates the spigot. As multiple locations in
trend with the increase of the inlet velocity. Fig. 11 shows the corre­ hydrocyclones may have the same axial position, each Z location in
sponding radial velocity distribution. The negative value indicates radial Fig. 15 may correspond to multiple points of shear stress. It can be seen
velocity points to the center, while the positive value indicates that from Fig. 15 that the shear stress exerting the CHO cells is generally
points to the hydrocyclone wall. It can be seen that the effect of the cone larger for mini-hydrocyclones with larger cone angles or higher inlet
angle and inlet velocity on the distribution of radial velocity is minor. velocities. The consistent finding from Figs. 12, 13 and 15 accounts for
In order to comprehensively study the shear stress inside the mini- the experimental phenomenon in Fig. 6 that an increase in cone angle or
hydrocyclone, the distribution of shear stress magnitude in the whole inlet velocity leads to an increase in the cell viability loss. The findings
computational domain, on the representative vertical plane and along are also consistent with that briefly qualitative statement in the

1750 2000
Cone angle 6º Inlet velocity 4 m/s
1500 Cone angle 8º 1750 Inlet velocity 6 m/s
Cone angle 10º Inlet velocity 7.5 m/s
1500
Shear stress (Pa)

1250
Shear stress (Pa)

1250
1000
1000
750
750
500
500
250 250
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Z Position Z Position
(a) (b)
Fig. 15. Shear stress magnitude exerting a CHO cell along the trajectory corresponding to Fig. 14 in the mini-hydrocyclones with (a) different cone angles and (b)
different inlet velocities.

10
L. He et al. Separation and Purification Technology 295 (2022) 121203

literature[3,34]. Fundamental Research Funds of Shandong University (2019HW041),

The currently calculated shear stress magnitude could account for Key Laboratory of High-efficiency and Clean Mechanical Manufacture at
the cell viability loss for hydrocyclone with the same inlet velocity but Shandong University, Ministry of Education, and Ocean industry leading
different cone angles or with the same hydrocyclone cone angle but talent team of Yantai’s Double Hundred Plan. The authors are grateful
different inlet velocities, i.e., for the variation of single variation in for their kind support.
Fig. 6. However, for the simultaneous variation of two variables, i.e., the
hydrocyclone cone angle and inlet velocity, the currently calculated References
shear stress magnitude is not applicable. For example, in Fig. 6, the cell
viability loss in the mini-hydrocyclone with the cone angle of 10◦ at an [1] J.Y. Kim, Y.-G. Kim, G.M. Lee, CHO cells in biotechnology for production of
recombinant proteins: current state and further potential, Appl. Microbiol.
inlet velocity of 6 m/s is higher than that with the cone angle of 8◦ at an Biotechnol. 93 (3) (2012) 917–930.
inlet velocity of 7.5 m/s. However, the shear stress magnitude shown in [2] R. Kunert, D. Reinhart, Advances in recombinant antibody manufacturing, Appl.
Figs. 12, 13 and 15 is generally lower in the former mini-hydrocyclone. Microbiol. Biotechnol. 100 (2016) 3451–3461.
[3] M.S. Syed, C. Marquis, R. Taylor, M.E. Warkiani, A two-step microengineered
This may be due to the following aspects. First, the shear stress magni­ system for high-density cell retention from bioreactors, Sep. Purif. Technol. 254
tude is not the sole variable causing cell viability loss in the mini- (2021), 117610.
hydrocyclones. The operating pressure and some other undefined vari­ [4] I.W. Bettinardi, A. Castan, R.A. Medronho, L.R. Castilho, Hydrocyclones as cell
retention device for CHO perfusion processes in single-use bioreactors, Biotechnol.
ables could cause cell viability loss, which deserves more dedicated Bioeng. 117 (2020) 1915–1928.
effort in the future. Second, the cell–cell collisions and wall-cell colli­ [5] D.J. Karst, F. Steinebach, M. Morbidelli, Continuous integrated manufacturing of
sions, which are ignored in this work, may have a contribution to the therapeutic proteins, Curr. Opin. Biotechnol. 53 (2018) 76–84.
[6] D. Voisard, F. Meuwly, P.A. Ruffieux, G. Baer, A. Kadouri, Potential of cell
shear stress magnitude in hydrocyclones. More effort is also required in
retention techniques for large-scale high-density perfusion culture of suspended
this aspect in future studies. mammalian cells, Biotechnol. Bioeng. 82 (2003) 751–765.
[7] A. Acrivos, E. Herbolzheimer, Enhanced sedimentation in settling tanks with
4. Conclusions and future work inclined walls, J. Fluid Mech. 92 (2006) 435–457.
[8] S. Kamthan, J. Gomes, P.K. Roychoudhury, Silk screen based dual spin-filter
module for perfusion culture of adherent and non-adherent mammalian cells,
The explicit reasons for the cell viability loss in mini-hydrocyclones Biotechnol. Lett. 36 (2014) 1623–1630.
are unclear. This paper presents the first attempt to quantitatively [9] W. Kelly, J. Scully, D. Zhang, G. Feng, M. Lavengood, J. Condon, J. Knighton,
R. Bhatia, Understanding and modeling alternating tangential flow filtration for
analyze the cell viability loss with respect to the magnitude of shear perfusion cell culture, Biotechnol. Prog. 30 (2014) 1291–1300.
stress by combining CFD-LPT modelling and experiments. The major [10] I.K. Hamamoto K, Tokashiki M, Perfusion culture of hybridoma cells using a
findings from the present study can be summarized as follows: centrifuge to separate cells from culture mixture, J. Biosci. Bioeng., 67 (1989) 190-
194.
[11] M.E. Warkiani, L. Wu, A.K.P. Tay, J. Han, Large-Volume Microfluidic Cell Sorting
(1) The experiment shows that the maximum cell viability loss due to for Biomedical Applications, Annu. Rev. Biomed. Eng. 17 (1) (2015) 1–34.
separation is 11.3%, indicating the considered mini- [12] P.R. Brito-Parada, R.M. Dewes, D. Vega-Garcia, J.J. Cilliers, Influence of Design
Parameters on Biomass Separation in Mini-hydrocyclones, Chem. Eng. Technol. 41
hydrocyclones are applicable to separate CHO cells. As the (2018) 2323–2330.
hydrocyclone cone angle or the inlet velocity increases, the cell [13] E.A. Elsayed, M.A. Wadaan1, The Potential of Hydrocyclone Application for
viability loss increases. Mammalian Cell Separation in Perfusion Cultivation Bioreactors Int. J. Biotech.
Well. Indus., 2 (2013) 153-163.
(2) The distributions of shear stress, including the turbulent-induced
[14] R.C.V. Pinto, R.A. Medronho, L.R. Castilho, Separation of CHO cells using
shear stress, are revealed in the whole hydrocyclone computa­ hydrocyclones, Cytotechnology 56 (1) (2008) 57–67.
tional domain, on the representative vertical plane and along the [15] E.A. Elsayed, R.A. Medronho, R. Wagner, W.-D. Deckwer, Use of Hydrocyclones for
CHO cell trajectory by the validated numerical model. In addi­ Mammalian Cell Retention: Separation Efficiency and Cell Viability (Part 1), Eng.
Life Sci. 6 (4) (2006) 347–354.
tion, it is found that the shear stress is stronger under the bottom [16] L. Ni, J. Tian, T. Song, Y. Jong, J. Zhao, Optimizing Geometric Parameters in
of the vortex finder and the lower hydrocyclone conical section. Hydrocyclones for Enhanced Separations: A Review and Perspective, Sep. Purif.
(3) The quantitative analysis of the shear stress magnitude demon­ Rev. 48 (1) (2019) 30–51.
[17] J. Tian, L. Ni, T. Song, J. Olson, J. Zhao, An overview of operating parameters and
strates that the shear stress magnitude increases with the conditions in hydrocyclones for enhanced separations, Sep. Purif. Technol. 206
hydrocyclone cone angle or the inlet velocity, which could ac­ (2018) 268–285.
count for the variation trend of cell viability loss. [18] M. De Jesus, F.M. Wurm, Manufacturing recombinant proteins in kg-ton quantities
using animal cells in bioreactors, Eur. J. Pharm. Biopharm. 78 (2011) 184–188.
[19] K.F.W.K.P. Jayapal, M.G.S. Yap, W.-S. Hu, Recombinant protein therapeutics from
CHO cells—20 years and counting, Chem. Eng. Prog. 13 (2007) 40–47.
CRediT authorship contribution statement [20] B. Schröder, E.A. Elsayed, J. Olownia, R. Wagner, Advantages of Hydrodynamic
Cell Separation in Industrial Cell Culture Processes, in: T. Noll (Ed.), Cells and
Culture, Springer Netherlands, Dordrecht, 2010, pp. 657–664, https://doi.org/
Liqun He: Investigation, Data curation, Formal analysis, Writing – 10.1007/978-90-481-3419-9_113.
original draft. Li Ji: Conceptualization, Methodology, Software, Fund­ [21] A.M. Kundu, G.W. Hiller, Hydrocyclones as cell retention devices for an N-1
perfusion bioreactor linked to a continuous-flow stirred tank production
ing acquisition, Writing – review & editing, Supervision. Yujing He:
bioreactor, Biotechnol. Bioeng. 118 (2021) 1973–1986.
Resources, Supervision. Yao Liu: Investigation. Songying Chen: Project [22] R.C.V. Pinto, R.A. Medronho, L.R. Castilho, The Effects of Cell Separation with
administration. Kaiwei Chu: Methodology. Shibo Kuang: Supervision, Hydrocyclones on the Viability of CHO Cells, in, in: R. Smith (Ed.), Cell Technology
for Cell Products, Springer Netherlands, Dordrecht, 2007, pp. 745–748, https://
Methodology, Writing – review & editing.
doi.org/10.1007/978-1-4020-5476-1_134.
[23] E. E.A., P. G-W., N. J., M. R.A., D. W-D., W. R., Use of hydrocyclone as an efficient
Declaration of Competing Interest tool for cell retention in perfusion cultures, in: Animal Cell Technology Meets
Genomics, 2003, pp. 679-682.
[24] G. Kretzmer, K. Schiigerl, Response of mammalian cells to shear stress, Appl.
The authors declare that they have no known competing financial Microbiol. Biotechnol. 34 (1991) 613–616.
interests or personal relationships that could have appeared to influence [25] J.T. Leovan der Pol, Shear sensitivity of animal cells from a culture medium
the work reported in this paper. perspective, Trends Biotechnol. 16 (1998) 323–328.
[26] Y. Sun, Y. Liu, Y. Zhang, Y. Huang, L. Wang, L. Dai, J. Xu, H. Wang, Hydrocyclone-
induced pretreatment for sludge solubilization to enhance anaerobic digestion,
Acknowledgments Chem. Eng. J. 374 (2019) 1364–1372.
[27] Y.-X. Xu, Y. Liu, Y.-H. Zhang, X.-J. Yang, H.-L. Wang, Effect of Shear Stress on
Deoiling of Oil-Contaminated Catalysts in a Hydrocyclone, Chem. Eng. Technol. 39
This work is supported by the National Natural Science Foundation (2016) 567–575.
of China (52006125), China Postdoctoral Science Foundation (2020
M682180), Young Scholars Program of Shandong University,

11
L. He et al. Separation and Purification Technology 295 (2022) 121203

[28] L. Zhu, B. Song, Z. Wang, Analyzing the suitability of a baffled orbitally shaken [52] Z. Zhu, O. Frey, D.S. Ottoz, F. Rudolf, A. Hierlemann, Microfluidic single-cell
bioreactor for cells cultivation using the computational fluid dynamics approach, cultivation chip with controllable immobilization and selective release of yeast
Biotechnol. Prog. 35 (1) (2019) e2746, https://doi.org/10.1002/btpr.2746. cells, Lab Chip 12 (2012) 906–915.
[29] A. Wyma, L. Martin-Alarcon, T. Walsh, T.A. Schmidt, I.D. Gates, M.S. Kallos, Non- [53] C. Bludszuweit, Three-dimensional numerical prediction of stress loading of blood
Newtonian rheology in suspension cell cultures significantly impacts bioreactor particles in a centrifugal pump, Artif. Organs 19 (1995) 590–596.
shear stress quantification, Biotechnol. Bioeng. 115 (2018) 2101–2113. [54] Z. Chen, S.K. Jena, G.A. Giridharan, M.A. Sobieski, S.C. Koenig, M.S. Slaughter, B.
[30] A. Mitoh, Y. Suebe, T. Kashima, E. Koyabu, E. Sobu, E. Okamoto, Y. Mitamura, P. Griffith, Z.J. Wu, Shear stress and blood trauma under constant and pulse-
I. Nishimura, Shear stress evaluation on blood cells using computational fluid modulated speed CF-VAD operations: CFD analysis of the HVAD, Med. Biol. Eng.
dynamics, Biomed. Mater. Eng. 31 (2020) 169–178. Comput. 57 (4) (2019) 807–818.
[31] B. Vickroy, K. Lorenz, W. Kelly, Modeling Shear Damage to Suspended CHO Cells [55] M.-H. Moosavi, N. Fatouraee, H. Katoozian, Finite element analysis of blood flow
during Cross-Flow Filtration, Biotechnol. Prog. 23 (1) (2007) 194–199. characteristics in a Ventricular Assist Device (VAD), Simul. Model. Pract. Theory
[32] J.-H. Yen, S.-F. Chen, M.-K. Chern, P.-C. Lu, The effect of turbulent viscous shear 17 (2009) 654–663.
stress on red blood cell hemolysis, J. Artif. Organs. 17 (2) (2014) 178–185. [56] T. Yano, K. Sekine, A. Mitoh, Y. Mitamura, E. Okamoto, D.W. Kim, I. Nishimura,
[33] S.A. Jones, A relationship between Reynolds stresses and viscous dissipation: S. Murabayashi, R. Yozu, An estimation method of hemolysis within an axial flow
implications to red cell damage, Ann. Biomed. Eng. 23 (1) (1995) 21–28. blood pump by computational fluid dynamics analysis, Artificial organs 27 (10)
[34] Y.i. Liu, H.-L. Wang, Y.-X. Xu, Y.-Y. Fang, X.-R. Chen, Sludge disintegration using a (2003) 920–925.
hydrocyclone to improve biological nutrient removal and reduce excess sludge, [57] M. Saidi, R. Maddahian, B. Farhanieh, Numerical investigation of cone angle effect
Sep. Purif. Technol. 177 (2017) 192–199. on the flow field and separation efficiency of deoiling hydrocyclones, Heat Mass
[35] B. Bendixen, D. Rickwood, Effects of hydrocyclones on the integrity of animal and Transfer 49 (2012) 247–260.
microbial cells, Bioseparation 4 (1) (1994) 21–27. [58] S. Fu, Y. Qian, H. Yuan, Y. Fang, Effect of cone angles of a hydrocyclone for the
[36] A. Jockwer, R.A. Medronho, R. Wagner, F.B. Anspach, W.-D. Deckwer, in: Animal separation of waste plastics with low value of density difference, Waste Manag. 140
Cell Technology: From Target to Market, Springer Netherlands, Dordrecht, 2001, (2022) 183–192.
pp. 301–306, https://doi.org/10.1007/978-94-010-0369-8_69. [59] G. Zhu, J.-L. Liow, Experimental study of particle separation and the fishhook
[37] L. He, L. Ji, X. Sun, S. Chen, S. Kuang, Investigation of mini-hydrocyclone effect in a mini-hydrocyclone, Chem. Eng. Sci. 111 (2014) 94–105.
performance in removing small-size microplastics, Particuology 71 (2022) 1–10. [60] F. Li, P. Liu, X. Yang, Y. Zhang, Y. Zhao, Effects of inlet concentration on the
[38] J.A. Delgadillo, R.K. Rajamani, A comparative study of three turbulence-closure hydrocyclone separation performance with different inlet velocity, Powder
models for the hydrocyclone problem, Int. J. Miner. Process. 77 (2005) 217–230. Technol. 375 (2020) 337–351.
[39] J.A. Delgadillo, R.K. Rajamani, Exploration of hydrocyclone designs using [61] J.-M. Wang, L.-Z. Wang, Experimental Study of a New Hydrocyclone for Multi-
computational fluid dynamics, Int. J. Miner. Process. 84 (2007) 252–261. Density Particles Separation, Sep. Sci. Technol. 44 (2009) 2915–2927.
[40] Q. Zhao, B. Cui, D. Wei, Y. Feng, Y. He, A.E. Bayly, Linking separation sharpness [62] L. Ji, S. Kuang, Z. Qi, Y. Wang, J. Chen, A. Yu, Computational analysis and
with the characteristics of axial velocity wave zone in a hydrocyclone, Powder optimization of hydrocyclone size to mitigate adverse effect of particle density,
Technol. 386 (2021) 467–482. Sep. Purif. Technol. 174 (2017) 251–263.
[41] B. Wang, A.B. Yu, Numerical study of particle–fluid flow in hydrocyclones with [63] M. Ghodrat, Z. Qi, S.B. Kuang, L. Ji, A.B. Yu, Computational investigation of the
different body dimensions, Miner. Eng. 19 (2006) 1022–1033. effect of particle density on the multiphase flows and performance of
[42] R. Jin, E. Keshavarzian, K. Dong, S. Dong, B. Wang, K. Kwok, M. Zhao, Numerical hydrocyclone, Miner. Eng. 90 (2016) 55–69.
study on the effect of the supersaturated vapor on the performance of a gas [64] M. Ghodrat, S.B. Kuang, A.B. Yu, A. Vince, G.D. Barnett, P.J. Barnett,
cyclone, Powder Technol. 366 (2020) 324–336. Computational Study of the Multiphase Flow and Performance of Hydrocyclones:
[43] Y. Zhang, R. Jin, S. Dong, Y. Wang, K. Dong, Y. Wei, B. Wang, Heterogeneous Effects of Cyclone Size and Spigot Diameter, Ind. Eng. Chem. Res. 52 (2013)
condensation combined with inner vortex broken cyclone to achieve high 16019–16031.
collection efficiency of fine particles and low energy consumption, Powder [65] M. Ghodrat, S.B. Kuang, A.B. Yu, A. Vince, G.D. Barnett, P.J. Barnett, Numerical
Technol. 382 (2021) 420–430. analysis of hydrocyclones with different vortex finder configurations, Miner. Eng.
[44] M.S. Syed, M. Rafeie, R. Henderson, D. Vandamme, M. Asadnia, M.E. Warkiani, 63 (2014) 125–138.
A 3D-printed mini-hydrocyclone for high throughput particle separation: [66] L. Ji, S. Kuang, A. Yu, Numerical Investigation of Hydrocyclone Feed Inlet
application to primary harvesting of microalgae, Lab on a Chip 17 (14) (2017) Configurations for Mitigating Particle Misplacement, Ind. Eng. Chem. Res. 58
2459–2469. (2019) 16823–16833.
[45] D. Vega-Garcia, P.R. Brito-Parada, J.J. Cilliers, Optimising small hydrocyclone [67] S.B. Kuang, K.W. Chu, A.B. Yu, A. Vince, Numerical study of liquid–gas–solid flow
design using 3D printing and CFD simulations, Chem. Eng. J. 350 (2018) 653–659. in classifying hydrocyclones: Effect of feed solids concentration, Miner. Eng. 31
[46] Q. Zhou, C. Wang, H. Wang, J. Wang, Eulerian-Lagrangian study of dense (2012) 17–31.
liquid–solid flow in an industrial-scale cylindrical hydrocyclone, Int. J. Miner. [68] Z. Qi, S.B. Kuang, A.B. Yu, Numerical investigation of the separation behaviours of
Process. 151 (2016) 40–50. fine particles in large dense medium cyclones, Int. J. Miner. Process. 142 (2015)
[47] T. Seewöster, J. Lehmann, Cell size distribution as a parameter for the 35–45.
predetermination of exponential growth during repeated batch cultivation of CHO [69] G. Zhu, J.-L. Liow, A. Neely, Computational study of the flow characteristics and
cells, Biotechnol. Bioeng. 55 (5) (1997) 793–797. separation efficiency in a mini-hydrocyclone, Chem. Eng. Res. Des. 90 (2012)
[48] C.F. Opel, J. Li, A. Amanullah, Quantitative modeling of viable cell density, cell 2135–2147.
size, intracellular conductivity, and membrane capacitance in batch and fed-batch [70] K.-J. Hwang, Y.-W. Hwang, H. Yoshida, Design of novel hydrocyclone for
CHO processes using dielectric spectroscopy, Biotechnol. Prog. 26 (2010) improving fine particle separation using computational fluid dynamics, Chem. Eng.
1187–1199. Sci. 85 (2013) 62–68.
[49] A.K. Bryan, V.C. Hecht, W. Shen, K. Payer, W.H. Grover, S.R. Manalis, Measuring [71] M.S. Syed, F. Mirakhorli, C. Marquis, R.A. Taylor, M.E. Warkiani, Particle
single cell mass, volume, and density with dual suspended microchannel movement and fluid behavior visualization using an optically transparent 3D-
resonators, Lab Chip 14 (3) (2014) 569–576. printed micro hydrocyclone, Biomicrofluidics 14 (2020), 064106.
[50] E.C. Anderson, D.F. Petersen, R.A. Tobey, Density Invariance of Cultured Chinese [72] B. Cui, C. Zhang, Q. Zhao, D. Hou, D. Wei, T. Song, Y. Feng, Study on interaction
Hamster Cells with Stage of the Mitotic Cycle, Biophys. J. 10 (7) (1970) 630–645. effects between the hydrocyclone feed flow rate and the feed size distribution,
[51] W.H. Grover, A.K. Bryan, M. Diez-Silva, S. Suresh, J.M. Higgins, S.R. Manalis, Powder Technol. 366 (2020) 617–628.
Measuring single-cell density, PNAS 108 (2011) 10992–10996.

12