The current issue and full text archive of this journal is available on Emerald Insight at:

[Link]

HFF
34,4 Numerical modelling and
experimental validation of dripping,
jetting and whipping modes of gas
1582 dynamic virtual nozzle
Received 25 September 2023 Krištof Kovacic and Jurij Gregorc
Revised 8 December 2023
Accepted 11 December 2023
Laboratory for Fluid Dynamics and Thermodynamics,
Faculty of Mechanical Engineering, University of Ljubljana, Ljubljana, Slovenia, and
Božidar Šarler
Laboratory for Fluid Dynamics and Thermodynamics,
Faculty of Mechanical Engineering, University of Ljubljana, Ljubljana, Slovenia,
and Laboratory for Simulation of Materials and Processes,
Institute of Metals and Technology, Ljubljana, Slovenia

Abstract
Purpose – This study aims to develop an experimentally validated three-dimensional numerical model for
predicting different flow patterns produced with a gas dynamic virtual nozzle (GDVN).
Design/methodology/approach – The physical model is posed in the mixture formulation and copes
with the unsteady, incompressible, isothermal, Newtonian, low turbulent two-phase flow. The computational
fluid dynamics numerical solution is based on the half-space finite volume discretisation. The geo-reconstruct
volume-of-fluid scheme tracks the interphase boundary between the gas and the liquid. To ensure numerical
stability in the transition regime and adequately account for turbulent behaviour, the k-v shear stress
transport turbulence model is used. The model is validated by comparison with the experimental
measurements on a vertical, downward-positioned GDVN configuration. Three different combinations of air
and water volumetric flow rates have been solved numerically in the range of Reynolds numbers for airflow
1,009–2,596 and water 61–133, respectively, at Weber numbers 1.2–6.2.
Findings – The half-space symmetry allows the numerical reconstruction of the dripping, jetting and
indication of the whipping mode. The kinetic energy transfer from the gas to the liquid is analysed, and

© Krištof Kovacic, Jurij Gregorc and Božidar Šarler. Published by Emerald Publishing Limited. This
article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may
reproduce, distribute, translate and create derivative works of this article (for both commercial and
non-commercial purposes), subject to full attribution to the original publication and authors. The full
terms of this licence may be seen at [Link]
The authors wish to thank Dr Saša Bajt and Prof Henry Chapman from DESY (Hamburg, Germany),
a member of the Helmholtz Association HGF, for their support and valuable comments regarding the
manuscript. The authors would also like to thank Dr Rizwan Zahoor (LFDT)for practical advices
regarding the numerical simulation.
International Journal of Numerical
Funding: Funding for this research is provided by the Centre for Free-Electron Laser Science (CFEL)
Methods for Heat & Fluid Flow under the project: Innovative methods for imaging with the use of x-ray free-electron laser(XFEL) and
Vol. 34 No. 4, 2024
pp. 1582-1608 synchrotron sources: simulation of gas-focused micro-jets, and Slovenian Grant and Innovation Agency
Emerald Publishing Limited (ARIS) within core funding P2-0162, project J2-4477 and Young Researcher Program.
0961-5539
DOI 10.1108/HFF-09-2023-0573 Disclosure statement: No potential conflict of interest was reported by the authors.
locations with locally increased gas kinetic energy are observed. The calculated jet shapes reasonably well Numerical
match the experimentally obtained high-speed camera videos.
Practical implications – The model is used for the virtual studies of new GDVN nozzle designs and
modelling and
optimisation of their operation. experimental
Originality/value – To the best of the authors’ knowledge, the developed model numerically reconstructs validation
all three GDVN flow regimes for the first time.
Keywords Flow-focusing, Dripping, Jetting, Whipping, Gas dynamic virtual nozzle, CFD
1583
Paper type Research paper

1. Introduction
Knowledge and understanding of fluids-related phenomena were always vital for humankind.
In recent decades, interest in scientific investigations and technological applications of fluids
has turned from macro to micro-scale, i.e. microfluidics (Nguyen et al., 2019). Knowledge of
microfluidics is a basis for precise manipulation of fluids on a submillimetre level. It enables the
production of different micro-fluid entities, such as jets, drops, bubbles, capsules, emulsions and
sprays (Gañan-Calvo et al., 2013).
One of the possible production techniques is flow focusing (FF), where an external co-
flowing fluid is applied around the internal flow (Gañan-Calvo et al., 2013). This technique
was first introduced by Gañan-Calvo (1998). In the most common FF method, the pressure
gradient induced by an outer gas focuses liquid meniscus, which emits a jet from its tip and
breaks up into droplets further downstream (Gañan-Calvo et al., 2013). However, FF could
also be used in liquid–liquid configuration (Anna et al., 2003).
FF technique is used in many fields such as biotechnology, cytology, drug or sample
delivery, pharmacy, chemistry, industrial engineering and the food and agriculture industry
(Gañan-Calvo et al., 2013) to mention a few.
The FF technology relies on suitable geometrical configuration and proper process
parameters. Two of the most common designs used today are micro-channel and micro-
nozzle configurations. FF based on a micro cross-channel principle has been mainly used to
produce droplets (Chen et al., 2021; Huang and Yao, 2022; Jafari and Shamloo, 2023; Khater
et al., 2021; Pan et al., 2020; Zeng and Fu, 2020), capsules (Aghaei et al., 2021) and emulsions
(Ontiveros et al., 2020) and for the bio-fabrication of hydrogel fibres (Guimarães et al., 2021)
and cell-laden hydrogel microspheres (Nguyen and Seo, 2022).
FF on the micron level was used to produce fibres of polyvinylpyrrolidone down to
the nanometre thickness (Ponce-Torres et al., 2019) and for smooth printing of
viscoelastic microfilms (Ponce-Torres et al., 2017). Dripping and jetting of microfluidic
multiphase flow were applied to particle and fibre synthesis (Nunes et al., 2013). The
dripping regime is widely used in biotechnology to produce droplets (Dewandre et al.,
2020; Duan et al., 2016; Segers et al., 2016), multilayered droplets (Si et al., 2015),
microcapsules (Mu et al., 2020; Si et al., 2016) and emulsions (Evangelio et al., 2016; Wu
et al., 2018; Yu et al., 2019, 2021).
The most common FF micro nozzle geometric configuration is gas dynamic virtual nozzle
(GDVN). It was developed by DePonte et al. (2008) and manufactured by flame polishing, later
fabricated with ceramic moulding (Beyerlein et al., 2015) and finally with three-dimensional
(3D) printing (Knoška et al., 2020; Nelson et al., 2016). GDVNs can produce various types of FF
regimes (Si et al., 2009), including:

dripping – periodic ejection of droplets formed at the nozzle outlet;

jetting – a continuous and stable fluid entity which breaks into a stream of droplets;
HFF
whipping – a continuous and unstable fluid entity with lateral movement perpendicular
34,4 to the jet axis; and

spurting – unstable and non-continuous fluid entity, which is interrupted periodically or
randomly without knowing a period of continuous jet or droplets ejection from the nozzle.

Because of the variety of operating modes, GDVNs have been positioned in many
1584 applications. One of the most common ones is sample delivery in Serial Femtosecond
Crystallography (SFX) (Chapman et al., 2011) and in ultrafast laser-driven sources based on
high-order harmonic generation (Klimešova et al., 2021), where the jetting regime is used.
However, GDVNs were also applied to produce droplets, whose size is a few 100 nanometres
(Mühlig et al., 2019). The jetting and the whipping regimes were combined to fabricate
uniform polymer and composite micro- and nano-fibres (Vasireddi et al., 2019).
Numerical simulations can be used to optimise geometric configurations of GDVNs and
thus reduce production and testing costs compared to experiments (Trebbin et al., 2014). Many
computational fluid dynamics (CFD) approaches to investigate FF jet behaviour have been
analysed recently. Most rely on the axisymmetric description, which has the benefit of a
smaller computational domain in terms of cell number and, therefore, a shorter computational
time than the whole 3D simulation. The drawback of the axisymmetric numerical simulation is
reflected in an inability to treat the entire fluid domain appropriately. Some parts of the FF
devices cannot be described as axisymmetrical, e.g. elbows, structural components and turns in
the supplying part of the geometry. Therefore, most studies investigate only the nozzle tip part
of these devices and assume uniform radial velocity distribution (Chen et al., 2015; Mu et al.,
2018a, 2019, 2018b; Rahimi et al., 2019, 2020; Šarler et al., 2021; Wu et al., 2017; Yu et al., 2021;
Zahoor et al., 2018a, 2018b, 2019, 2020, 2021; Zhao et al., 2020, 2019a, 2019b), fully developed,
radially parabolic flow (Hua et al., 2007; Soroor et al., 2021) or a combination of both (Acero
et al., 2012; Blanco-Trejo et al., 2020; Cabezas et al., 2021; Cruz-Mazo et al., 2017; Dewandre et al.,
2020; Herrada et al., 2008; Jensen et al., 2006; Mamet et al., 2017; Rubio et al., 2021; Vega et al.,
2010, 2014). Another problem of the two-dimensional axisymmetric approach is manifested in
the inability to appropriately deal with the whipping regime (Blanco-Trejo et al., 2020). Therefore,
it is impossible to analyse all three most common FF flow regimes, which are dripping, jetting
and whipping, in axisymmetry. Some researchers successfully implemented jetting and dripping
flow regimes into their numerical simulations at the same nozzle geometry design (Chen et al.,
2015; Dewandre et al., 2020; Hua et al., 2007; Wu et al., 2017). On the other hand, Zhao et al. (2019a)
investigated jetting and turbulent flow regime, which is similar to whipping, with a difference in
chaotic lateral movement from the axis. In contrast, in the whipping flow regime, the lateral
movement of the jet around the axis could be described deterministically.
A 3D description has to be used to overcome the shortcomings of the axisymmetric
approach. So far, the 3D numerical simulations of FF devices are extremely rare. Han et al.
(2019) analysed squeezing, dripping and jetting regimes in a cross-channel configuration.
Khater et al. (2020) implemented a 3D numerical simulation for lab-on-chip technology,
where the dripping type of operating mode is a primary focus of the research. Again, a cross-
channel geometry with uniform velocity inlet boundary condition was investigated.
In addition to cross-channel, also a few 3D numerical simulations of FF nozzles were
investigated. In the first one, Müller et al. (2016) analysed a simplified domain of the 3D
injector system, where the primary jet breakup was investigated using large eddy
simulation (LES) for highly turbulent flow, which is used in the biogas technology field. In
the second one, Belšak et al. (2021) investigated a gas-focused liquid micro-sheet. The third
example contains a rectangular nozzle, where the rounded jets in the dripping and jetting
regime were simulated by Inguva et al. (2019). The last example is a recent study of GDVN,
where Nazari et al. (2023) compared simulated jet lengths with experimental measurements. Numerical
Until now, only the simplified geometry of the nozzle tip was considered, omitting modelling and
geometrically complex supplying part of the nozzle that dictates velocity distribution at the
nozzle tip. The nozzle inlet boundary conditions were implemented as velocity profiles,
experimental
calculated from the flow rates (Müller et al., 2016; Nazari et al., 2023) or with the inlet tubes validation
long enough for a fully developed flow (Belšak et al., 2021; Inguva et al., 2019). However, the
supplying part of the nozzle can cause non-parabolic or non-uniform flow conditions far
from being able to be coped axisymmetrically. 1585
The purpose of this paper is the analysis of all three most common (dripping, jetting and
whipping) GDVN operating modes (patterns), depending on the air and water flow rates at the
same nozzle geometric design, however ten times larger than typically used in DESY for
femtosecond crystallography (Knoška et al., 2020). Ten times scale-up was used due to the limits of
nozzle fabrication and accessible experimental manipulations. The model also copes with the
upper or supplying part of the nozzle, which distributes air to the air–water mixing zone between
the nozzle orifice and feeding capillary. The geometry design with the elbow influences the airflow
conditions. Therefore, we cannot assume that air has a uniform (constant) velocity profile, as it is
usually defined in numerical simulations at the air inlet boundary conditions. Numerical results
are compared with our experimental data. One of the present paper’s goals is to find out if the half-
space 3D model is adequate for simulating all the experimentally observed GDVN phenomena.
The main limitation of the half-space 3D model is the inability to deal with corkscrew whipping
mode. However, to detect a whipping, the proposed model is found to be appropriate.

2. Physical model and numerical solution

2.1 Governing equations
Analysis of incompressible, unsteady, isothermal, Newtonian, two-phase, turbulent flow is
carried out within ANSYS Fluent simulation system, which is based on the finite volume
method (FVM). The volume in which the jet is emanating is at atmospheric pressure. Thus,
we can assume that there is no significant expansion of the air and respective change in its
density, so we assumed incompressible flow and neglected the energy conservation
equation. Owing to the high velocities of a liquid micro-jet and micro-dimensionality of a
nozzle system, one can estimate that acceleration is some hundred (or even thousand) times
higher than the gravitational one. Thus, the gravitational force term in the momentum
equation could be neglected. The governing equations used in our model are mass (1),
momentum (2) and phase fraction (6) conservation equations:
@r
þ r  ð rv Þ ¼ 0 (1)
@t
@
ðrvÞ þ r  ðrvvÞ ¼ rp þ r  s þ f s (2)
@t
Where r denotes mixture density, v(p, t) velocity, p(p, t), pressure and s the Newtonian
viscous stress tensor, defined as s ¼meff ðrv þ rvT ). meff stands for the effective dynamic
viscosity of the mixture that might also include the turbulent correction and fr(p, t) for the
surface tension force, calculated as:
f s ¼ sk n (3)
where s is surface tension, n ¼ ra is normal on the gas-liquid interphase boundary,
pointing from the gas to the liquid, and k is a curvature of the interphase boundary,
calculated based on the continuum surface model (CSF) as:
HFF kðaÞ ¼ r  ðra=jrajÞ (4)
34,4
The volume-of-fluid (VOF) interphase boundary treatment introduces a phase indication
variable a(p, t), which takes values between 0 and 1, as follows:

1 p in liquid
aðp; tÞ ¼ (5)
1586 0 p in gas

The gas–liquid interphase boundary is described by an advection equation with a discontinuity

in a at the interphase boundary:

@
ðraÞ þ r  ðrvaÞ ¼ 0 (6)
@t

The material properties, such as dynamic viscosity and density, are determined by the
phase-weighted average:

# ð a Þ ¼ a# l þ ð 1  aÞ # g (7)

where #l and #g stand for liquid and gas properties, respectively.

Because the predicted air Reynolds number in the nozzle’s orifice area falls for the
experimental conditions observed in the transition regime between turbulent and laminar
flow, we use the k-v shear stress transport (SST) model with two additional transport
equations for k and v, as follows:
  
@ m
ðrkÞ þ r  ðrkvÞ ¼ r  m þ t rk þ Gk  Yk þ Sk þ Gb (8)
@t sk

  
@ m
ðrvÞ þ r  ðrvvÞ ¼ r  m þ t rv þ Gv  Yv þ Sv þ Gvb (9)
@t sv

In these equations, Gk represents the generation of turbulence kinetic energy due to mean
velocity gradients. Gv represents the generation of v. Yk and Yv represent the dissipation of
k and v due to turbulence. Sk and Sv are user-defined source terms. Gb and Gvb account for
buoyancy terms. Detailed information on the k-v SST model, its numerical implementation
and its use are provided in Fluent (2020). The effective viscosity is calculated as
meff ¼ aml þ (1  a)mg þ mt. For the micro geometry used at DESY for SFX, the regime
is laminar and the k  v equations do not need to be used and mt ¼ 0 Pas.

2.2 Discretisation schemes

The described system of equations is solved on the high-performance computing (HPC)
cluster (Intel Xeon E5-2680V3 processors, 2.5 GHz), with ANSYS Fluent solver, release 20.1.
The SIMPLEC scheme is used for the pressure–velocity coupling, and the gradient is least
square cell-based. Discrete values of pressure are calculated with PRESTO! scheme. For the
momentum, k and v equation the second order upwind is used. Two-phase flow is modelled
with VOF model, where the air is set as primary (continuous) and water as secondary
(dispersed) phase. VOF is solved during every iteration with a geo-reconstruct scheme to
track the interface. Transient formulation of the VOF equation is calculated with an explicit
scheme. All other governing equations are calculated with the first-order implicit scheme Numerical
with adaptive timestep. The limiting global Courant number is set to 0.5, and the local modelling and
Courant number associated with the phase transport equation to 0.25. The absolute
convergence criteria of residuals within one timestep is set to 1E-4 for continuity and 1E-5
experimental
for all velocity components, k and v. The maximum number of iterations per one-time step validation
is set to 30. This is enough to guarantee the residuals’ convergence.

2.3 Geometric three dimensional model and spatial discretisation

1587
A half section of the 3D printed nozzle is shown in Figure 1(a). The fluid domain is extracted
from CAD model of FF nozzle with characteristic dimensions shown in Figure 1(b). As the
nozzle geometry is symmetric in one direction, a half-space symmetry is imposed to
extensively economise the simulations. Beyond the nozzle outlet, an external field area is
designed, as shown in Figure 2(a), with a diameter of 6 mm (10D) and a length of 25 mm, to
reduce the influence of the finite discretisation volume and related boundary conditions.
The FVM meshing is set in ANSYS Meshing. It results in an unstructured, fully
conformal mesh, consisting of hexahedral and tetrahedral elements, shown in Figure 2(b). In
the area of higher interest, especially where the interaction between the liquid and gas
occurs (e.g. area between the nozzle capillary and orifice and in the region after the nozzle
outlet), the mesh consists of refined structured hexahedral elements [seen in Figure 2(d)].
Further from that focus area, the mesh element size increases, minding the cells’ growth
factor and equal cell size between the different mesh regions. As the complex wall geometry
in the air-supplying part of the nozzle directly influences the air velocity field and not the
interface reconstruction, the unstructured mesh with tetrahedral elements is prescribed.

2.4 Material properties

The material properties for both fluids at the ambient conditions used in the numerical
simulation are given in Table 1, with the analysed gas (air) and liquid (water) volumetric flow
rates. From now on, the abbreviation GxLy stands for “G” – gas, “x” – volumetric flow rate of
gas (air) measured in [L h1], “L” – liquid and “y” – volumetric flow rate of liquid (water)
measured in [mL min1] of liquid (water), e. g. G2L5 stands for a gas flow rate of 2 L h1 and a
water flow rate of 5 mL min1. Three cases with different air and water flow rate combinations
have been simulated. The ranges of their dimensionless numbers are given in Table 2.
Reynolds number for gas is calculated at the nozzle orifice and minding the area occupied by

Figure 1.
Representation of
nozzle design
HFF
34,4

1588

Figure 2.
Computational mesh
of three-dimensional
FF nozzle

the liquid jet as follows:

4 r g Qg
Reg ¼
(10)
pmg D  Dj
Liquid Reynolds number and Weber number are defined as follows (Herrada et al., 2008):
 1=4
2 r3l Dpg 1=2
Rel ¼ Ql (11)
ml 8 p2
  1=4
1 8 rl Dp3g 1=2
Numerical
We ¼ Ql (12) modelling and
s p2
experimental
where rl and rg denotes the density of a liquid and gas phase, respectively, Dpg pressure validation
drop of a gas phase, Ql and Qg the volumetric flow rate of a liquid and gas phase, ml and mg
dynamic viscosity of a liquid and gas, respectively, D orifice diameter and Dj liquid
jet diameter. Capillary number Ca is defined as a ratio between Weber number We and 1589
Reynolds number of the liquid Rel. In Table 2, non-dimensional numbers are calculated
based on numerical results (Dpg and Dj) and material properties from Table 1. The non-
dimensional numbers of a typical SFX nozzle operation have been added to the Table 2,
according to Zahoor et al. (2018b). As only the jetting is used in SFX, the comparison is
limited to the non-dimensional numbers of the jetting. As one can see from the Table 2, the
results from the ten times scale-up nozzle partially overlap with the actual size of GDVN.

2.5 Boundary, initial and operating conditions

The domain comprises five boundary patches: air and water inlet, outlet, walls and
symmetry. The boundary conditions at the nozzle half-space are of the symmetrical type.
The solid walls are described with no-slip boundary conditions. The inlet boundary
conditions for air and water are set as velocity inlets, with uniform velocity, calculated from
the volumetric flow rates and a cross-sectional area at the inlet location. The pressure outlet
boundary condition around the external field chamber is set to atmospheric pressure, where
the gauge pressure equals zero. All prescribed boundary conditions are shown in Figure 3.
The 3D two-phase flow transient simulation is initialised with a steady–state air velocity
field of the desired air flow rate, calculated separately. For the two-phase simulations, the
entire inner capillary tube is initialised (at t ¼ 0 s) and patched with the water, whereas the

Physical parameter Symbol Unit Value/range

Air density rg kg m-3 1.225

Air viscosity mg Pa s 1.7894  10–5
Water density rl kg m-3 998.2
Water viscosity ml Pa s 1.003  10–3
Surface tension s N m-1 0.072
Air flow rate Qg L h-1 25, 40, 60
Water flow rate Ql mL min-1 500, 1000
Table 1.
Source: Table by authors Physical properties

Case Qg [L h-1] Ql [mL min-1] Drg [mbar] Reg Rel We Ca Table 2.

Cases and
G25L500 25 500 9.8 1009 61 1.2 0.020 dimensionless
G40L1000 40 1000 25.7 1842 109 3.5 0.032
G60L1000 60 1000 55.7 2596 133 6.2 0.047
numbers of ten times
SFX jetting* – – – 17–1223 110–215 2.8–320 0.025–1.49 scale-up model and
actual GDVN used in
Note: *Zahoor et al. (2018b) SFX
HFF
34,4

1590

Figure 3.
Boundary conditions
rest of the fluid domain is with the air. In such a way, the calculation time needed for the Numerical
developed air flow field in a two-phase transient simulation is reduced significantly. modelling and
Simulation data is saved at 2 ms intervals, which is enough to track droplets that break
up from the jet. Results are analysed using ANSYS CFD Post, Python with cv2 and
experimental
Matplotlib library. validation

2.6 Mesh independence study 1591

Four different meshes (A–D) are produced to assess the mesh independence. Their minimum
cell sizes are shown in Table 3. After the equal reduction of cell size in all three dimensions,
the mesh refinement factors, relatively scaled to mesh D, are 0.25, 0.54, 0.74 and 1.0,
resulting in a total of 2,736,084, 5,916,786, 8,224,138 and 11,055,127 cells from the coarse to
the finest mesh (A–D), respectively.
For the mesh independence study, calculations at conditions G40L1000 are made. The
calculations are initialised based on results from mesh D at a flow time equal to 3.746 ms for
all the meshes. The impact of mesh density on results is estimated based on a comparison of
the jet diameters at three different locations: at the nozzle outlet, 0.5 and 1.0 mm from the
nozzle outlet in the downstream direction.
Figure 4 shows changes in the jet diameter at three different locations as a function of the
elementary cell size ratio. The cell size ratio is the ratio between the edge length of the
elementary cell in mesh D (5.0 mm) and different meshes (A. B, C and D see Table 3). As can
be seen, the average jet diameter converges to the final value at all three locations by
refining the elementary cell size. Our jet diameter results are thus independent of the mesh
density and size for meshes B-D. Mesh D (cell size ratio 1.0) is used in all further calculations.

3. Experimental setup
The following measurements are performed for validation of the numerical model. The
laboratory test rig is shown in Figure 5. A simple GDVN (A) is mounted on a triaxial
translational stage (E). The gas is supplied from a pressure tank (J) and controlled via EL-
Flow series thermal mass flow controller (I) from Bronkhorst High-Tech Instruments. A
pressure drop supplied to the GDVN is measured by a differential pressure sensor
Amphenol ELV series. The water flow is metered with a syringe pump Harvard Apparatus
Pump 11 elite (G). In the GDVN (A), water is focused and ejected vertically downward in a
view window of the SpeedCam Visario high-speed camera by Weinberger (C). The image
magnification and focus are achieved using a modular long-distance microscope Optem
Fusion by Edmund optics (B). Back illumination by optical fibre is aligned with the camera’s
optical path and connected to the light source Olympus ILK-5 (D).
The GDVN is printed by the KUDO3D Micro SLA printer. Printing a nozzle with the
0.6 mm orifice diameter takes  4 h.

Nr. of Nr. of Min. cell size in x Min. cell size in y Min. cell size in z Nr. of cells across
Mesh cells cells ratio direction [mm] direction [mm] direction [mm] the Dj,avg

A 2,736,084 0.25 8.33 8.33 8.33 11 (10.8)

B 5,916,786 0.54 6.25 6.25 6.25 14 (14.2)
C 8,224,138 0.74 5.56 5.56 5.56 16 (16.0)
Table 3.
D 11,055,127 1.00 5.00 5.00 5.00 18 (18.2) Properties of
different types of
Source: Table by authors meshes
HFF
34,4

1592

Figure 4.
Average jet diameter
at 0.0, 0.5 and 1.0 mm
from the nozzle outlet,
calculated with
different types of
meshes (A–D)

Figure 5.
Experimental setup

3.1 Video analysis

The video samples are recorded in 1536  1024 px resolution at 1,000 Hz. A total of 500
frames are analysed for every volumetric flow rate combination.
We recorded animations and videos of the jet’s behaviour with data from numerical
simulations and experimental measurements, respectively. A purpose-built code, which
analyses jet length and diameter with statistical values, such as average, minimum and Numerical
maximum values, standard deviation and skewness, is developed in Python using the cv2 modelling and
library. The video analysing process is divided into four main steps:
experimental
(1) frames are extracted from the video; validation
(2) frames are subtracted with the background frame and binarised from greyscale
into black-white images (black has a pixel value of 0 and white a value of 255)
using the Otsu threshold method (Otsu, 1979); 1593
(3) the reference object is measured with the code to determine the mm-to-pixel ratio; and
(4) the jet object is recognised from the frame and analysed.

4. Results and discussion

The employed half-space 3D model requires about 30 h of CPU to calculate the steady–state
airflow. Meanwhile, about 20–30 days of CPU are required for two-phase flow transient
calculation. A total of 72 cores on HPC have been booked for the described analyses.

4.1 Validation of numerical results

The numerical results are compared with the experiment under the same process conditions.
Figure 6 shows how the jet diameter, measured experimentally, varies at specific locations
(a, b, c) from the nozzle outlet with time. Table 4 shows values of average jet diameter and
the discrepancy between the numerical and experimental results at different locations
outside the nozzle.
Numerical and experimental results of average jet diameter at different locations match
between 2.6% and 5.8%, whereas the jet length is usually longer in the numerical cases. In the
case of a physical experiment, we measured average jet lengths 1.74 mm. The calculated
average jet length of stabilised numerical solution of 2.93 mm overshoots the experimental data
by 68%. The reason for the discrepancy with great merit originates from the fact that the
experimental conditions do not experience ideal conditions, and that the jet breakup is a
complex non-periodic and multiparametric phenomenon. For an accurate jet length calculation,
all influential parameters have to be taken into account within the representative statistical
sampling in time. From Figure 7, it seems that our numerical model should be calculated over a
longer time period to have a similar density distribution to the experimental jet lengths. We

Figure 6.
Variation of jet
diameter at (a) nozzle
outlet; (b) 0.5 mm and
(c) 1.0 mm from the
nozzle outlet, at
G40L1000 and
1536  1024 px
resolution and
1,000 Hz frequency
rate of a highspeed
camera
HFF expect that with further calculation and additional refinement around the jet breakup area, the
34,4 peak of the calculated jet length distribution might be closer to the experimental one. An
accurate jet length calculation is highly challenging, and therefore, up to now, only studies of
Zahoor et al. (2019) and Nazari et al. (2023) provide a comparison of jet length between
experiments and simulations.
Also, the statistical samples of the numerical calculations and experiment are different,
1594 because of the different timestep size between the frames of numerical animation and
experimental video (2 ms vs 1 ms). However, the number of analysed frames is almost the same
in numerical simulations (513) and experiment (500). Not exactly the same statistical sample of
numerical simulations and experiment have an unneglectable impact on the statistical average
of jet length. If the jet lengths are compared as a histogram, a region exists where higher
experimental and lower numerical jet lengths overlaps, shown in Figure 7 with the hatched
area.
Our numerical model, therefore, correctly predicts the jet length within a range, shown in
Figure 7, but overestimates it on an average by 68%. The achieved matching is better than

Table 4.
Error between the
numerical and
experimental average
jet diameters at
nozzle outlet, 0.5 and
1.0 mm further Qg [L h-1] Ql [mL min-1] Location [mm] Experimental Dj,avg [mm] Numerical Dj,avg [mm] Error [%]
downstream from the
nozzle outlet for 40 1,000 0.0 114.7 108.0 5.8
G40L1000 flow rate 40 1,000 0.5 93.1 90.7 2.6
40 1,000 1.0 88.0 84.0 4.5
combination and
mesh D Source: Table by authors

Figure 7.
Histogram of the jet
length comparison
between
experimental and
numerical results of
jetting (G40L1000)
regime
in the axisymmetric numerical study by Zahoor et al. (2019), where the jet length is Numerical
overestimated by 89% on average at five process conditions A recent study by Nazari et al. modelling and
(2023) predicts jet lengths with a relative error of 10%. It suggests implementing an
adaptive mesh refinement algorithm to calculate jet length accurately. However, in the study
experimental
of Nazari et al. (2023), only the jet length has been validated, but not the jet diameter, as it is validation
also in the present study.
In other investigations of FF jetting regime, scholars mainly validated their numerical model
by comparing only the shapes of the fluid entities (Mu et al., 2018b; Wu et al., 2017; Zahoor et al.,
1595
2018a) or sometimes jet diameter (Rahimi et al., 2020; Zahoor et al., 2018a; Zhao et al., 2019a,
2019b) or jet velocity (Zahoor et al., 2018a). Jet length is a highly transient parameter compared to
the jet diameter, and thus, it is difficult to predict it correctly. However, for the SFX, where the
GDVN is most commonly used, the jet length prediction of the jetting regime is less important
than the jet diameter prediction if the jets are also with the overestimation significantly longer
than a minimum required stable jet length (50 mm or 500 mm on ten times scaled-up model),
where the XFEL hits the stable jet. The predicted variation of local jet diameter indicates
Plateau–Rayleigh instability which potentially causes the jet breakup.
A more accurate comparison could be made with a higher frequency rate of a high-speed
camera and, thus, a direct comparison of experimental and numerical jet length variation for
every numerical timestep. In this case, a jet length development could be analysed
temporally and not just as a global statistical average.
A half-space 3D model and a spatial discretisation of control volumes have a significant
impact on a numerical calculation of jet length. In our fluid domain, cell size grows from a
nozzle outlet in a downstream direction. In the area where the jet breaks up, the control
volumes should be refined even more in the downstream direction to calculate the interface
between air and water accurately. We assume that a complete 3D model should calculate jet
length more precisely.
We have also validated the numerical model by comparing gas pressure drop values in
the GDVN. The pressure drop has been measured between the connecting tube of the GDVN
and the surrounding air [Figure 5(h)]. The numerical values have been calculated by area-
weighted average on six (three in radial and three in axial direction) different planes inside
the GDVN. Experimentally measured versus numerically calculated pressure drops are 10.1
vs 9.8, 25.0 vs 25.7 and 54.3 vs 55.7 mbar for dripping (G25L500), jetting (G40L1000) and
whipping (G60L1000), respectively. Results match within 3% for all three cases.

4.2 Velocity profiles of steady–state simulation

An essential benefit of using half-space 3D numerical simulation compared to axisymmetric
one is a more realistic description of the velocity profile caused by the complex printed 3 D
nozzle geometry. Figure 8(a) and Figure 8(b) show velocity profiles for the gas phase in terms
of velocity magnitude and of u, v and w velocity components inside the nozzle’s mixing zone (a)
and at the nozzle outlet (b), respectively. The velocities are calculated in a steady-state
simulation and used as an initial condition in a two-phase flow transient simulation. It is
clearly seen that the velocity profiles calculated with 3D numerical simulation are not close to
symmetrical in x direction as prescribed in axisymmetric numeric simulation. Also, the
w component of velocity does not take the parabolic profile inside the nozzle. The turn in
the GDVN has an evident influence on airflow, leading to the conclusion that the left part of the
nozzle has a higher average velocity than the right one. This impacts the jet behaviour because
the FF gas acts on one side of the jet with a higher velocity than on the other, and the jet thus
declines from the vertical axis for a certain angle.
HFF 4.3 Numerical results of flow patterns
34,4 The isosurfaces with a volume fraction of 0.5 (air-water interface) at a selected time step for
the cases G25L500 (dripping), G40L1000 (jetting) and G60L1000 (whipping) are shown in
Figure 9. For a better representation, a mirrored solution of a half-sectioned fluid domain is
also displayed. In the dripping regime [Figure 9(a)], a separation of the droplet from the
liquid meniscus occurs inside the nozzle before the orifice. In the jetting regime [Figure 9(b)],
1596 the first part of the liquid jet, emitting from the meniscus, is stable. Further downstream
appear Plateau–Rayleigh (Rayleigh, 1878) instabilities, leading to the jet breaking into
droplets. In the whipping regime [Figure 9(c)], oscillations from the central axis of the jet are
observed, which propagate downstream and lead to a jet breakup.

4.4 Numerical and experimental flow patterns

The numerical data is recorded every 2 ms for a total physical time of approximately 6 ms, and the
frequency rate of a high-speed camera is 1000 Hz. The numerical values of the jet behaviour can
thus not be compared with the experimental data at each numerical record due to the lack of more
frequent experimental frames. However, the flow regimes from the numerical solution correspond
perfectly with the experimentally observed flow regimes at the same combination of air and water
volumetric flow rates. The same jet shapes can be found in both physical and numerical
experiments, as shown in Figure 10. A part of the nozzle outlet is added to the numerical solution
for a better representation. The experimental and numerical frames are at the same scale.
In the dripping regime [Figure 10(a)], the same distribution of the droplets with similar
distances between them has been found in comparing numerical with experimental results. The
droplet size near the nozzle outlet in simulations also matches the experiment. Further
downstream, the numerically calculated droplet size is smaller than the experimentally observed.
The droplets in the experiment are not falling at the centre of the nozzle axis; therefore, they
seem to be larger because they are falling at a certain angle, declined from the axis and are closer
to the camera window. This can be concluded from the experimental frame, where the droplet at
the nozzle outlet is focused more than the droplet far from the nozzle, since the focus of the high-
speed camera was set to nozzle’s symmetry plane. Our half-space 3D numerical model does not
allow us to simulate a declination perpendicular to the ZX (nozzle’s symmetry) plane. With the
complete 3D model, the droplets’ size far from the nozzle would be more similar to the droplet
size in experimental frames. The appearance of the satellite droplet is not precisely at the same

Figure 8.
Velocity profiles of
magnitude jvj and u,
v, and w components
for mesh D at (a)
1.3 mm inside the
nozzle and (b) nozzle
outlet
 Numerical
modelling and
experimental
validation

1597

Figure 9.
Isosurface of volume
fraction of 0.5 for (a)
dripping; (b) jetting
and (c) whipping
regime

location, but the existence of the satellite droplet from numerical simulation has been confirmed
by experimental recording. In Figure 10(a), arrows are pointing at the satellite droplets.
In the jetting regime [Figure 10(b)], the matching shapes have been found in experimental
and numerical frames. The jet breakup process on those frames is the same as the number of
Plateau–Rayleigh instabilities. The length of the jet is significantly longer in numerical frames
than in the experimental ones. The reasons for the discrepancy are discussed in Section 4.1.
In the whipping regime [Figure 10(c)], the same jet shapes can be recognised in
experimental and numerical frames with the same number of regions where Plateau–
Rayleigh instability occurs. A similar declination from the central axis line is also observed.
It can be concluded that the developed numerical model can predict realistic behaviour
and shapes of jets and droplets for dripping, jetting and whipping regimes.

4.5 Flow pattern map

Figure 11 shows the flow pattern diagram of different flow regimes observed experimentally.
The flow patterns can be sorted into four main groups: jetting, whipping, dripping and
spurting. Different regimes are delineated with subjectively determined borders. On the same
diagram, the corresponding numerical simulations are plotted. For the cases G25L500
HFF
34,4

1598

Figure 10.
A comparison of
experimental vs
numerical jet shapes
for (a) dripping
G25L500; (b) jetting
G40L1000 and (c)
whipping G60L1000
 Numerical
modelling and
experimental
validation

1599

Figure 11.
Flow pattern map

(dripping), G40L1000 (jetting) and G60L1000 (whipping), the numerical flow regimes perfectly
match the experimental ones.

4.6 Velocity
As VOF model calculates only the mixture velocity field of both, liquid and gas phase, we split
the velocity field at the water volume fraction greater or equal to 0.5 to show the velocity of the
liquid jet. The jet surface moves faster than the liquid jet core, as the gas is acting on the jet
surface and transferring momentum to it. Therefore, the impact of the focusing gas is more
significant on the surface since the gas accelerates the fluid at the surface more than at the jet’s
core. The gas flow around the jet changes over time, influencing the jet breakup (see Section
4.7). Both the jet core and surface are accelerating downstream, but the acceleration of the jet
core is delayed. This delay can be seen in Figure 12, where the plots of jet core velocity
magnitude and water volume fraction versus x coordinate at the nozzle outlet, 0.5 and 1.0 mm
further downstream, are plotted. The jet’s surface and core velocities ratio is around 1.7, 1.5 and
1.25 at the nozzle outlet, 0.5 and 1.0 mm downstream, respectively. Obviously, the momentum
from the focusing gas is transferred faster to the jet’s surface than to its core. When the
transferred momentum from the focusing gas reaches the liquid core, the difference between
the jet surface and core velocities decreases in the downstream direction. The plots in Figure 12
show that the jet slightly declined to the left. Also, the jet velocity minimum is slightly moved
in the left direction. The reason is the slightly different velocity distribution of surrounding gas,
which transfers momentum asymmetrically regarding the jet axis, leading to the jet and jet
velocity minimum declination. Owing to the increased jet velocity, the jet diameter decreases
downstream to satisfy the volume (mass) flux conservation condition. An example of
determining the jet diameter from the plot in Figure 12 is shown for the middle timestep.
HFF
34,4

1600

Figure 12.
Velocity field of the
liquid jet core (top)
and surface (middle)
and charts of jet core
velocity magnitude
vs X coordinate for
the jetting regime at
three different
locations during three
different timesteps
4.7 Kinetic energy transfer Numerical
Volume-specific kinetic energy is calculated as modelling and
Ek;a 1 experimental
ek;a ¼ ¼ a ra jvj2 (13) validation
V 2

where a stands for volume fraction, r for density, jvj for velocity magnitude and V for the
cell volume. The volume-specific kinetic energy represents a dynamic pressure of the 1601
individual phase as also of the mixture in the two-phase transitional area (0 < a < 1), where
it is weighted according to the volume fraction value.
In flow-focusing, a given total energy of the gas, expressed as a total pressure of the gas
is first transformed to the dynamic pressure in the nozzle orifice region. Here dynamic
pressure of the focusing gas arises, as also the kinetic energy, which is transferred to the
liquid phase during the downstream flow. Thus, the kinetic energy of the gas decreases, and
the liquid’s kinetic energy increases downstream from the nozzle outlet. The energy transfer
process is continuous and occurs at all fluid entities, also after the breakup.
The volume-specific kinetic energy of air and water at a time instance is shown in Figure 13
for dripping [13(a)], jetting [13(b)] and whipping regimes [13(c)], respectively. In all regimes,
locations with increased values of the specific kinetic energy of gas are observed. In the jetting
regime, these locations are almost symmetrical across the jet axis, whereas in the whipping
regime, the locations are evidently unsymmetrical. These local areas act with locally higher
dynamic pressure on a liquid jet. At these locations, the liquid jet is focused more, resulting in a
locally narrower jet diameter, known as necking or Plateau–Rayleigh instability. Therefore,
focusing gas increases Plateau–Rayleigh instabilities, which are always present in any kind of
the jet, apart from FF. In the whipping regime, locations with increased specific kinetic energy
are distributed unsymmetrically. This results in asymmetric dynamic pressure distribution,
and the jet declines towards the area with locally decreased specific kinetic energy.
The reason of the areas with the locally increased dynamic pressure of the gas could
originate from the decreased cross-section of the airflow due to the locally increased liquid
jet volume due to the increase in jet diameter, which occupies the space of the gas phase
flow. This could be well seen in the dripping regime, where the locally increased areas of
dynamic gas pressure generated at the droplet detaching point. The process is equivalent to
the flow around the cylinder, including the wake region behind the droplet. The occurrence
of the locally increased regions of dynamic gas pressure in the jetting and whipping regime
could be induced by the change in airflow around the PR instabilities of the liquid jet stream.
Where the streamlines of the gas flow are focused together, its cross section is lower and
thus, higher velocity (specific kinetic energy) occurs in this region, seen in Figure 14. These
regions are visible in Figure 13(b) around the thickened region of the liquid jet.

5. Conclusions
This investigation focused on developing the half-space 3D numerical simulation of the
GDVN, its experimental validation and assessing its capabilities towards the axisymmetric
simulation. Three gas and liquid volumetric flow rate combinations are simulated to
calculate dripping, jetting and whipping regimes, experimentally observed at certain flow
rate combinations. A comparison between experimental and numerical snapshots of jet
shapes and structures are analysed. The same jet shapes are observed in numerical and
experimental approaches for dripping, jetting and whipping flow regimes. For the jetting
regime, it is shown that average jet diameters at different locations agree well with the
experimental results, differing at most 5.8%. A simulated gas pressure drop in dripping,
HFF
34,4

1602

Figure 13.
Specific kinetic
energy of air and
water for (a) dripping
G25L500; (b) jetting
G40L1000 and (c)
whipping G60L1000
regime
 Numerical
modelling and
experimental
validation

1603

Figure 14.
Streamlines of
velocity and volume-
specific kinetic
energy (dynamic
pressure) field for
jetting (G40L1000)

jetting and whipping regimes matches within 3% of the experimental results. However, the
model has difficulties with jet length calculation, which is less important than the accurate
jet diameter calculation in the most widely used application of SFX. For SFX jet diameter
prediction close to the nozzle outlet is a key parameter.
The main advantage of the half-space 3D model over the axisymmetric one is the
possibility of including the nozzle geometry’s influence on the flow and allowing the
simulation of the whipping flow regime with a lateral movement of the jet around the jet
axis. This model allows realistic jet behaviour, including fine jet declinations from the axis
due to the non-axisymmetrical velocity field of the focusing gas.
The numerical model does not provide accurate results of the jet length. Its physics could
be improved by considering the entire 3D model on the expense of more than double
computing time.
With a higher resolution of high-speed camera, the experimental measurements of the jet
diameters and lengths would be more accurate. With a higher frequency rate of frames
HFF filmed with a high-speed camera the numerical results and experimental measurements can
34,4 be compared temporarily, not just with the average jet diameter and length values.
Nevertheless, the presented model is adequate for modelling most typical flow modes of
GDVN.

References
1604 Acero, A.J., Montanero, J.M., Ferrera, C., Herrada, M.A. and Gañan-Calvo, A.M. (2012), “Enhancement of
the stability of the flow focusing technique for low-viscosity liquids”, Journal of Micromechanics
and Microengineering, Vol. 22 No. 11, p. 115039, doi: 10.1088/0960-1317/22/11/115039.
Aghaei, H., Solaimany Nazar, A.R. and Varshosaz, J. (2021), “Double flow focusing microfluidic-assisted
based preparation of methotrexate–loaded liposomal nanoparticles: encapsulation efficacy, drug
release and stability”, Colloids and Surfaces A: Physicochemical and Engineering Aspects,
Vol. 614, p. 126166, doi: 10.1016/[Link].2021.126166.
Anna, S.L., Bontoux, N. and Stone, H.A. (2003), “Formation of dispersions using ‘flow focusing’ in
microchannels”, Applied Physics Letters, Vol. 82 No. 3, pp. 364-366, doi: 10.1063/1.1537519.
Belšak, G., Bajt, S. and Šarler, B. (2021), “Computational modeling and simulation of gas focused liquid
micro-sheets”, International Journal of Multiphase Flow, Vol. 140, p. 103666, doi: 10.1016/j.
ijmultiphaseflow.2021.103666.
Beyerlein, K.R., Adriano, L., Heymann, M., Kirian, R., Knoška, J., Wilde, F., Chapman, H.N. and Bajt,
S. (2015), “Ceramic micro-injection molded nozzles for serial femtosecond crystallography
sample delivery”, Review of Scientific Instruments, Vol. 86 No. 12, p. 125104, doi: 10.1063/
1.4936843.
Blanco-Trejo, S., Herrada, M.A., Gañan-Calvo, A.M., Rubio, A., Cabezas, M.G. and Montanero, J.M.
(2020), “Whipping in gaseous flow focusing”, International Journal of Multiphase Flow, Vol. 130,
p. 103367, doi: 10.1016/[Link]flow.2020.103367.
Cabezas, M.G., Rebollo-Muñoz, N., Rubio, M., Herrada, M.A. and Montanero, J.M. (2021), “Global
stability analysis of axisymmetric liquid–liquid flow focusing”, Journal of Fluid Mechanics,
Vol. 909, p. A10, doi: 10.1017/jfm.2020.953.
Chapman, H.N., Fromme, P., Barty, A., White, T.A., Kirian, R.A., Aquila, A., Hunter, M.S., Schulz, J.,
DePonte, D.P., Weierstall, U. and Doak, R.B. (2011), “Femtosecond X-ray protein
nanocrystallography”, Nature, Vol. 470 No. 7332, pp. 73-77, doi: 10.1038/nature09750.
Chen, Y., Wu, L. and Zhang, L. (2015), “Dynamic behaviors of double emulsion formation in a flow-
focusing device”, International Journal of Heat and Mass Transfer, Vol. 82, pp. 42-50, doi:
10.1016/[Link].2014.11.027.
Chen, Q., Li, J., Song, Y., Chen, B., Christopher, D.M. and Li, X. (2021), “Pressure-driven microfluidic
droplet formation in Newtonian and shear-thinning fluids in glass flow-focusing
microchannels”, International Journal of Multiphase Flow, Vol. 140, p. 103648, doi: 10.1016/j.
ijmultiphaseflow.2021.103648.
Cruz-Mazo, F., Herrada, M.A., Gañan-Calvo, A.M. and Montanero, J.M. (2017), “Global stability of
axisymmetric flow focusing”, Journal of Fluid Mechanics, Vol. 832, pp. 329-344, doi: 10.1017/
jfm.2017.684.
DePonte, D.P., Weierstall, U., Schmidt, K., Warner, J., Starodub, D., Spence, J.C.H. and Doak, R.B. (2008),
“Gas dynamic virtual nozzle for generation of microscopic droplet streams”, Journal of Physics
D: Applied Physics, Vol. 41 No. 19, p. 195505, doi: 10.1088/0022-3727/41/19/195505.
Dewandre, A., Rivero-Rodriguez, J., Vitry, Y., Sobac, B. and Scheid, B. (2020), “Microfluidic droplet
generation based on non-embedded co-flow-focusing using 3D printed nozzle”, Scientific Reports,
Vol. 10 No. 1, p. 21616, doi: 10.1038/s41598-020-77836-y.
Duan, H., Romay, F.J., Li, C., Naqwi, A., Deng, W. and Liu, B.Y.H. (2016), “Generation of monodisperse Numerical
aerosols by combining aerodynamic flow-focusing and mechanical perturbation”, Aerosol
Science and Technology, Vol. 50 No. 1, pp. 17-25, doi: 10.1080/02786826.2015.1123213. modelling and
Evangelio, A., Campo-Cortes, F. and Gordillo, J.M. (2016), “Simple and double microemulsions via the experimental
capillary breakup of highly stretched liquid jets”, Journal of Fluid Mechanics, Vol. 804, pp. 550- validation
577, doi: 10.1017/jfm.2016.498.
Fluent, A. (2020), “Ansys fluent 2020 R1-Theory guide; user’s guide”, Canonsburg, PA, USA.
Gañan-Calvo, A.M. (1998), “Generation of steady liquid microthreads and micron-sized monodisperse
1605
sprays in gas streams”, Physical Review Letters, Vol. 80 No. 2, pp. 285-288, doi: 10.1103/
PhysRevLett.80.285.
Gañan-Calvo, A.M., Montanero, J.M., Martín-Banderas, L. and Flores-Mosquera, M. (2013), “Building
functional materials for health care and pharmacy from microfluidic principles and flow
focusing”, Advanced Drug Delivery Reviews, Vol. 65 Nos 11/12, pp. 1447-1469, doi: 10.1016/j.
addr.2013.08.003.
Guimarães, C.F., Gasperini, L., Marques, A.P. and Reis, R.L. (2021), “3D flow-focusing microfluidic
biofabrication: one-chip-fits-all hydrogel fiber architectures”, Applied Materials Today, Vol. 23,
p. 101013, doi: 10.1016/[Link].2021.101013.
Han, W., Chen, X., Wu, Z. and Zheng, Y. (2019), “Three-dimensional numerical simulation of droplet
formation in a microfluidic flow-focusing device”, Journal of the Brazilian Society of Mechanical
Sciences and Engineering, Vol. 41 No. 6, p. 265, doi: 10.1007/s40430-019-1767-y.
Herrada, M.A., Gañan-Calvo, A.M., Ojeda-Monge, A., Bluth, B. and Riesco-Chueca, P. (2008), “Liquid
flow focused by a gas: jetting, dripping, and recirculation”, Physical Review E, Vol. 78 No. 3,
p. 36323, doi: 10.1103/PhysRevE.78.036323.
Hua, J., Zhang, B. and Lou, J. (2007), “Numerical simulation of microdroplet formation in coflowing
immiscible liquids”, AIChE Journal, Vol. 53 No. 10, pp. 2534-2548, doi: 10.1002/aic.11287.
Huang, J. and Yao, Z. (2022), “Influencing factors and size prediction of bubbles formed by flow
focusing in a cross-channel”, Chemical Engineering Science, Vol. 248, p. 117228, doi: 10.1016/j.
ces.2021.117228.
Inguva, V., Graceffa, R., Schulz, J., Bilsel, O. and Perot, B.J. (2019), “Creating round focused micro-jets
from rectangular nozzles”, Journal of Mechanical Science and Technology, Vol. 33 No. 9, pp. 4281-
4289, doi: 10.1007/s12206-019-0824-x.
Jafari, A. and Shamloo, A. (2023), “CFD study of droplet formation in a cross-junction microfluidic
device: investigating the impact of outflow channel design and viscosity ratio”, Engineering
Applications of Computational Fluid Mechanics, Vol. 17 No. 1, p. 2243091, doi: 10.1080/
19942060.2023.2243091.
Jensen, M.J., Stone, H.A. and Bruus, H. (2006), “A numerical study of two-phase stokes flow in an
axisymmetric flow-focusing device”, Physics of Fluids, Vol. 18 No. 7, doi: 10.1063/1.2214461.
Khater, A., Abdelrehim, O., Mohammadi, M., Mohamad, A. and Sanati-Nezhad, A. (2021), “Thermal
droplet microfluidics: from biology to cooling technology”, TrAC Trends in Analytical
Chemistry, Vol. 138, p. 116234, doi: 10.1016/[Link].2021.116234.
Khater, A., Abdelrehim, O., Mohammadi, M., Azarmanesh, M., Janmaleki, M., Salahandish, R.,
Mohamad, A. and Sanati-Nezhad, A. (2020), “Picoliter agar droplet breakup in microfluidics
meets microbiology application: numerical and experimental approaches”, Lab on a Chip, Vol. 20
No. 12, pp. 2175-2187, doi: 10.1039/D0LC00300J.
Klimešova, E., Kulyk, O., Hoque, Z., Roos, A.H., Khakurel, K.P., Rebarz, M., Jurkovicova, L., Albrecht,
M., Finke, O., Lera, R. and Hort, O. (2021), “A multipurpose end-station for atomic, molecular and
optical sciences and coherent diffractive imaging at ELI beamlines”, The European Physical
Journal Special Topics, Vol. 230 No. 23, pp. 4183-4194, doi: 10.1140/epjs/s11734-021-00192-z.
HFF Knoška, J., Adriano, L., Awel, S., Beyerlein, K.R., Yefanov, O., Oberthuer, D., Peña Murillo, G.E., Roth,
N., Sarrou, I., Villanueva-Perez, P. and Wiedorn, M.O. (2020), “Ultracompact 3D microfluidics for
34,4 time-resolved structural biology”, Nature Communications, Vol. 11 No. 1, p. 657, doi: 10.1038/
s41467-020-14434-6.
Mamet, V., Namy, P. and Dedulle, J.M. (2017), “Numerical modeling of flow focusing: quantitative
characterization of the flow regimes”, Physics of Fluids, Vol. 29 No. 9, doi: 10.1063/1.5003840.
Mu, K., Ding, H. and Si, T. (2018a), “Instability analysis of the cone–jet flow in liquid-driven flow
1606 focusing”, Microfluidics and Nanofluidics, Vol. 22 No. 12, p. 138, doi: 10.1007/s10404-018-2158-x.
Mu, K., Si, T. and Ding, H. (2019), “Nonlinear dynamics and manipulation of dripping in capillary flow
focusing”, Science China Physics, Mechanics and Astronomy, Vol. 62 No. 12, p. 124713, doi:
10.1007/s11433-019-9444-8.
Mu, K., Ding, H. and Si, T. (2020), “Experimental and numerical investigations on interface coupling of
coaxial liquid jets in co-flow focusing”, Physics of Fluids, Vol. 32 No. 4, doi: 10.1063/5.0002102.
Mu, K., Si, T., Li, E., Xu, R.X. and Ding, H. (2018b), “Numerical study on droplet generation in
axisymmetric flow focusing upon actuation”, Physics of Fluids, Vol. 30 No. 1, doi: 10.1063/1.5009601.
Mühlig, K., Gañan-Calvo, A.M., Andreasson, J., Larsson, D.S.D., Hajdu, J. and Svenda, M. (2019),
“Nanometre-sized droplets from a gas dynamic virtual nozzle”, Journal of Applied
Crystallography, Vol. 52 No. 4, pp. 800-808, doi: 10.1107/S1600576719008318.
Müller, T., Sänger, A., Habisreuther, P., Jakobs, T., Trimis, D., Kolb, T. and Zarzalis, N. (2016), “Simulation
of the primary breakup of a high-viscosity liquid jet by a coaxial annular gas flow”, International
Journal of Multiphase Flow, Vol. 87, pp. 212-228, doi: 10.1016/[Link]flow.2016.09.008.
Nazari, R., Ansari, A., Herrmann, M., Adrian, R.J. and Kirian, R.A. (2023), “Numerical and experimental
investigation of gas flow field variations in three-dimensional printed gas-dynamic virtual
nozzles”, Frontiers in Mechanical Engineering, Vol. 8, p. 958963, doi: 10.3389/fmech.2022.958963.
Nelson, G., Kirian, R.A., Weierstall, U., Zatsepin, N.A., Farago, T., Baumbach, T., Wilde, F., Niesler, F.
B., Zimmer, B., Ishigami, I. and Hikita, M. (2016), “Three-dimensional-printed gas dynamic
virtual nozzles for x-ray laser sample delivery”, Optics Express, Vol. 24 No. 11, pp. 11515-11530,
doi: 10.1364/OE.24.011515.
Nguyen, H.Q. and Seo, T.S. (2022), “A 3D printed size-tunable flow-focusing droplet microdevice to
produce cell-laden hydrogel microspheres”, Analytica Chimica Acta, Vol. 1192, p. 339344, doi:
10.1016/[Link].2021.339344.
Nguyen, N.T., Wereley, S. and Shaegh, S.A.M. (2019), Fundamentals and Applications of Microfluidics,
3rd ed., Artech, available at: [Link]
Nunes, J.K., Tsai, S.S.H., Wan, J. and Stone, H.A. (2013), “Dripping and jetting in microfluidic
multiphase flows applied to particle and fibre synthesis”, Journal of Physics D: Applied Physics,
Vol. 46 No. 11, p. 114002, doi: 10.1088/0022-3727/46/11/114002.
Ontiveros, J.F., Company, R., Vaz, M.O. and Nardello-Rataj, V. (2020), “Microfluidic emulsification:
process and formulation variables effects in flow behavior pattern on a flow-focusing device”,
Colloids and Surfaces A: Physicochemical and Engineering Aspects, Vol. 601, p. 125038, doi:
10.1016/[Link].2020.125038.
Otsu, N. (1979), “A threshold selection method from gray-level histograms”, IEEE Transactions on
Systems, Man, and Cybernetics, Vol. 9 No. 1, pp. 62-66, doi: 10.1109/TSMC.1979.4310076.
Pan, D., Chen, Q., Zeng, Y. and Li, B. (2020), “Droplets containing large solid particle inside formation
and breakup dynamics in a flow-focusing microfluidic device”, Experimental Thermal and Fluid
Science, Vol. 115, p. 110103, doi: 10.1016/[Link]flusci.2020.110103.
Ponce-Torres, A., Vega, E.J., Castrejon-Pita, A.A. and Montanero, J.M. (2017), “Smooth printing of
viscoelastic microfilms with a flow focusing ejector”, Journal of Non-Newtonian Fluid Mechanics,
Vol. 249, pp. 1-7, doi: 10.1016/[Link].2017.09.004.
Ponce-Torres, A., Ortega, E., Rubio, M., Rubio, A., Vega, E.J. and Montanero, J.M. (2019), “Gaseous flow Numerical
focusing for spinning micro and nanofibers”, Polymer, Vol. 178, p. 121623, doi: 10.1016/j.
polymer.2019.121623.
modelling and
Rahimi, M., Shams Khorrami, A. and Rezai, P. (2019), “Effect of device geometry on droplet size in co-axial
experimental
flow-focusing microfluidic droplet generation devices”, Colloids and Surfaces A: Physicochemical validation
and Engineering Aspects, Vol. 570, pp. 510-517, doi: 10.1016/[Link].2019.03.067.
Rahimi, M., Yazdanparast, S. and Rezai, P. (2020), “Parametric study of droplet size in an axisymmetric
flow-focusing capillary device”, Chinese Journal of Chemical Engineering, Vol. 28 No. 4, pp. 1016- 1607
1022, doi: 10.1016/[Link].2019.12.026.
Rayleigh, L. (1878), “On the instability of jets”, Proceedings of the London Mathematical Society, Vol. s1-
10 No. 1, pp. 4-13, doi: 10.1112/plms/s1-10.1.4.
Rubio, M., Rubio, A., Cabezas, M.G., Herrada, M.A., Gañan-Calvo, A.M. and Montanero, J.M. (2021),
“Transonic flow focusing: stability analysis and jet diameter”, International Journal of
Multiphase Flow, Vol. 142, p. 103720, doi: 10.1016/[Link]flow.2021.103720.
Šarler, B., Zahoor, R. and Bajt, S. (2021), “Alternative geometric arrangements of the nozzle outlet orifice
for liquid micro-jet focusing in gas dynamic virtual nozzles”, Materials, Vol. 14 No. 6, p. 1572,
doi: 10.3390/ma14061572.
Segers, T., de Rond, L., de Jong, N., Borden, M. and Versluis, M. (2016), “Stability of monodisperse
phospholipid-coated microbubbles formed by flow-focusing at high production rates”, Langmuir,
American Chemical Society, Vol. 32 No. 16, pp. 3937-3944, doi: 10.1021/[Link].6b00616.
Si, T., Li, F., Yin, X.Y. and Yin, X.Z. (2009), “Modes in flow focusing and instability of coaxial liquid–
gas jets”, Journal of Fluid Mechanics, Vol. 629, pp. 1-23, doi: 10.1017/S0022112009006211.
Si, T., Feng, H., Luo, X. and Xu, R.X. (2015), “Formation of steady compound cone-jet modes and
multilayered droplets in a tri-axial capillary flow focusing device”, Microfluidics and
Nanofluidics, Vol. 18 Nos 5/6, pp. 967-977, doi: 10.1007/s10404-014-1486-8.
Si, T., Yin, C., Gao, P., Li, G., Ding, H., He, X., Xie, B. and Xu, R.X. (2016), “Steady cone-jet mode in
compound-fluidic electro-flow focusing for fabricating multicompartment microcapsules”,
Applied Physics Letters, Vol. 108 No. 2, doi: 10.1063/1.4939632.
Soroor, M., Zabetian Targhi, M. and Tabatabaei, S.A. (2021), “Numerical and experimental
investigation of a flow focusing droplet-based microfluidic device”, European Journal of
Mechanics - B/Fluids, Vol. 89, pp. 289-300, doi: 10.1016/[Link]flu.2021.06.013.
Trebbin, M., Krüger, K., DePonte, D.P., Roth, S.V., Chapman, H.N. and Förster, S. (2014), “Microfluidic
liquid jet system with compatibility for atmospheric and high-vacuum conditions”, Lab Chip,
Vol. 14 No. 10, pp. 1733-1745.
Vasireddi, R., Kruse, J., Vakili, M., Kulkarni, S., Keller, T.F., Monteiro, D.C.F. and Trebbin, M. (2019),
“Solution blow spinning of polymer/nanocomposite micro-/nanofibers with tunable diameters
and morphologies using a gas dynamic virtual nozzle”, Scientific Reports, Vol. 9 No. 1, p. 14297,
doi: 10.1038/s41598-019-50477-6.
Vega, E.J., Montanero, J.M., Herrada, M.A. and Gañan-Calvo, A.M. (2010), “Global and local instability
of flow focusing: the influence of the geometry”, Physics of Fluids, Vol. 22 No. 6, doi: 10.1063/
1.3450321.
Vega, E.J., Acero, A.J., Montanero, J.M., Herrada, M.A. and Gañan-Calvo, A.M. (2014), “Production of
microbubbles from axisymmetric flow focusing in the jetting regime for moderate Reynolds
numbers”, Physical Review E, Vol. 89 No. 6, p. 63012, doi: 10.1103/PhysRevE.89.063012.
Wu, L., Liu, X., Zhao, Y. and Chen, Y. (2017), “Role of local geometry on droplet formation in axisymmetric
microfluidics”, Chemical Engineering Science, Vol. 163, pp. 56-67, doi: 10.1016/[Link].2017.01.022.
Wu, Q., Yang, C., Yang, J., Huang, F., Liu, G., Zhu, Z., Si, T., et al. (2018), “Photopolymerization of
complex emulsions with irregular shapes fabricated by multiplex coaxial flow focusing”,
Applied Physics Letters, Vol. 112 No. 7, doi: 10.1063/1.5018207.
HFF Yu, C., Wu, L., Li, L. and Liu, M. (2019), “Experimental study of double emulsion formation behaviors in
a one-step axisymmetric flow-focusing device”, Experimental Thermal and Fluid Science,
34,4 Vol. 103, pp. 18-28, doi: 10.1016/[Link]flusci.2018.12.032.
Yu, W., Li, B., Liu, X. and Chen, Y. (2021), “Hydrodynamics of triple emulsion droplet generation in a
flow-focusing microfluidic device”, Chemical Engineering Science, Vol. 243, p. 116648, doi:
10.1016/[Link].2021.116648.
Zahoor, R., Bajt, S. and Šarler, B. (2018a), “Influence of gas dynamic virtual nozzle geometry on micro-
1608 jet characteristics”, International Journal of Multiphase Flow, Vol. 104, pp. 152-165, doi: 10.1016/j.
ijmultiphaseflow.2018.03.003.
Zahoor, R., Belšak, G., Bajt, S. and Šarler, B. (2018b), “Simulation of liquid micro-jet in free expanding
high-speed co-flowing gas streams”, Microfluidics and Nanofluidics, Vol. 22 No. 8, p. 87, doi:
10.1007/s10404-018-2110-0.
Zahoor, R., Regvar, R., Bajt, S. and Šarler, B. (2020), “A numerical study on the influence of liquid
properties on gas-focused micro-jets”, Progress in Computational Fluid Dynamics, an
International Journal, Vol. 20 No. 2, pp. 71-83, doi: 10.1504/PCFD.2020.106407.
Zahoor, R., Knoška, J., Bajt, S. and Šarler, B. (2021), “Experimental and numerical investigation of gas-
focused liquid Micro-Jet velocity”, International Journal of Multiphase Flow, Vol. 135, p. 103530,
doi: 10.1016/[Link]flow.2020.103530.
Zahoor, R., Knoska, J., Gregorc, J., Bajt, S. and Sarler, B. (2019), “Validation of a computational model for
a coupled liquid and gas flow in Micro-Nozzles”, VIII International Conference on Computational
Methods for Coupled Problems in Science and Engineering, 3 Jun 2019 - 5 Jun 2019, Barcellona
(Spain), p. 7, doi: 10.3204/PUBDB-2019-01908.
Zeng, W. and Fu, H. (2020), “Precise monodisperse droplet production in a flow-focusing microdroplet
generator”, Chemical Engineering Research and Design, Vol. 160, pp. 321-325, doi: 10.1016/j.
cherd.2020.06.002.
Zhao, J., Ning, Z. and Lü, M. (2019a), “Large eddy simulation of Two-Phase flow pattern and
transformation characteristics of flow mixing nozzle”, Journal of Mechanics, Vol. 35 No. 5,
pp. 693-704, doi: 10.1017/jmech.2018.51.
Zhao, J., Ning, Z. and Lv, M. (2020), “Experimental and numerical simulation of flow pattern evolution
in a flow mixing nozzle under a moderate inlet flow rate”, ACS Omega, Vol. 5 No. 35, pp. 22335-
22347, doi: 10.1021/acsomega.0c02702.
Zhao, J., Ning, Z., Lü, M. and Wang, G. (2019b), “Numerical simulation of flow focusing pattern and
morphological changes in two-phase flow inside nozzle”, Chinese Journal of Chemical
Engineering, Vol. 27 No. 1, pp. 63-71, doi: 10.1016/[Link].2018.08.022.

Corresponding author
Božidar Šarler can be contacted at: [Link]@[Link]

For instructions on how to order reprints of this article, please visit our website:
[Link]/licensing/[Link]
Or contact us for further details: permissions@[Link]