You are on page 1of 8

Sensors and Actuators B 211 (2015) 198–205

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical


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

Convergent–divergent micromixer coupled with pulsatile flow


Arshad Afzal, Kwang-Yong Kim ∗
Department of Mechanical Engineering, Inha University, 253 Younghyun-Dong, Nam-Gu, Incheon 402-751, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: An efficient mixing system for a microfluidic platform that uses periodic sinusoidal characteristics in
Received 16 October 2014 space and time is proposed in this work. A convergent–divergent channel with sinusoidal walls (space-
Received in revised form 1 January 2015 periodic) represented the most effective coupling with pulsatile (time-periodic) flow among the five
Accepted 6 January 2015
tested geometries. The effects of sinusoidal pulsing on mixing performance in the convergent–divergent
Available online 28 January 2015
microchannel were examined using unsteady Navier–Stokes equations. The mixing performance was
analyzed using the Reynolds and Strouhal numbers, and the ratio of pulsing amplitude to steady flow
Keywords:
velocity. The micromixer shows interesting mixing behaviors with these parameters in terms of instan-
Pulsatile flow
Convergent–divergent micromixer
taneous mixing patterns as well as overall mixing performance. The micromixer with pulsing inputs
Mixing index achieved mixing index, 0.92 within two periods of the sinusoidal walls.
Strouhal number © 2015 Elsevier B.V. All rights reserved.
Navier–Stokes equations

1. Introduction effect of flow recirculations on mixing at high Reynolds numbers.


Parsa and Hormozi [8] carried out a study on passive micromix-
Microfluidic systems are widely used in applications involving ers with sinusoidal walls using experiment and CFD. The phase
fast chemical reactions, DNA analysis, cell separation, and sample shift between the side walls turned out to be important parameter
preparation and analysis [1–4]. In microfluidic devices, the laminar affecting the mixing performance via the formation of Dean and
flow condition poses a challenge for the mixing of liquid samples. separation vortices. Nguyen and Wu [9] and Hessel et al. [10] pre-
Mixing is dominated by molecular diffusion instead of turbulence sented brief reviews of the design of different passive micromixers.
at the micro-scale. Therefore, for high performance lab-on-a-chip In another approach, external agitations in the form of mag-
and micro-total analysis systems (␮-TAS), it is essential to find a netic energy, electrical energy, and ultrasonic can be used to stir the
method to achieve fast and compact mixing at the micro-scale. fluids. These active-type micromixers rely on such external fields
The key to effective mixing lies in the ability to mix fluids to enhance the mixing of fluid samples [11–13]. Recently, time-
across the cross-section of the microchannel, i.e., create a large dependent pulsatile flows have been used by many researchers
contact between the fluid samples, thereby, accelerating the dif- for fast and efficient mixing [14–18]. In this approach, the flow is
fusion processes. In the past few years, several innovative methods pulsed from either one or both of the inlets in a regular fashion.
have been proposed to improve micro-scale mixing. Geometrical In some cases, the pulsing can be applied via secondary channels
modifications applied to manipulate flows have been proved to along the path of the main flow. A simple example of pulsing is
be very effective for the development of efficient micromixers. a sinusoidal flow with respect to time. Tabeling et al. [14] stud-
Such passive-type micromixers have advantages of simplicity and ied the mixing in a cross-channel intersection micromixer (a main
easy fabrication. Stroock et al. [5] devised a method that uses bas- channel with two side channels). The main channel has steady
relief structures on the floor of the channel to generate chaotic flow which consists of fluid samples, whereas the side channels
flows that lead to effective mixing. The three-dimensional (3-D) drive time-dependent flows. For large amplitudes of the side flows,
twisted micromixer developed by Liu et al. [6] introduces sec- stretching and folding of the fluid interface cause chaotic advection.
ondary motion at higher Reynolds numbers (Re ∼ 6–70) for efficient Bottauski et al. [15] extended the work [14] and presented a design
mixing. Mengeaud et al. [7] studied the mixing process in a zig- with three cross-stream secondary channels. The results show that
zag microchannel using finite element simulations to highlight the high mixing performance can be achieved by a perturbation of
the main flow via secondary channels using oscillating sinusoidal
jet flows. However, for complete mixing, the secondary channels
∗ Corresponding author. Tel.: +82 32 872 3096; fax: +82 32 868 1716. need to be actuated by a strong sinusoidal flow (high amplitude
E-mail address: kykim@inha.ac.kr (K.-Y. Kim). and frequency). Glasgow and Aubry [16–18] conducted a series of

http://dx.doi.org/10.1016/j.snb.2015.01.062
0925-4005/© 2015 Elsevier B.V. All rights reserved.
A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205 199

extensive studies on pulsed flows through “T” channels with differ- with the following transient advection–diffusion model [20] for the
ent configurations and operating parameters using computational species concentration field:
fluid dynamics (CFD). The pulsing was performed using a low fre-
quency sinusoidal flow rate superimposed upon a steady flow rate. ∂Ci  )C = ˛∇ 2 C
+ (V · ∇ i i (3)
In one of these studies [18], they investigated the mixing behaviors ∂t
in ribbed and 3-D twisted micromixers under pulsed flow condi-
tions. The results indicate that mixing can be significantly improved where ˛ is the diffusivity coefficient and Ci is the concentration of
using time pulsing compared to a constant flow rate condition in the species.
all tested micromixers. In the present model, two similar fluids having the same vis-
For a micromixer coupled with pulsatile flow, important dimen- cosity, , and the same fluid density, , were used. The kinematic
sionless parameters that govern the dynamics of the flow through viscosity  and diffusion coefficient ˛, were set to 10−6 m2 s−1
the micromixer are the Reynolds number (Re = Vs D/), Strouhal (the kinematic viscosity of water at room temperature) and
number (St = fD/Vs ), and velocity ratio (Vo /Vs ) where Vs is the steady 10−10 m2 s−1 (a typical diffusivity value for microfluidic applica-
flow velocity and D is the hydraulic diameter of the inlet chan- tions), respectively [16–18]. To study mixing, pure water and a
nel. For the pulsatile flow, Vo is the amplitude of velocity and f is solution of dye in water were used as working fluids with the dye
the pulsing frequency. The Reynolds number is the ratio of iner- mass fraction determined from the solution of Eq. (3) in space and
tial to viscous forces. The Strouhal number relates the flow (D/Vs ) time.
and pulsing (1/f) time scales to characterize the pulsatile flow phe- To introduce the pulsatile flows in the microchannel, the
nomenon. time dependent velocities at Inlet 1 and Inlet 2 are defined as
The objective of the present study was to design a micromixer V = Vs + Vo sin(2f t + ϕ), which represents a superposition of steady
that can utilize the full advantage of pulsatile flows over a short (first term) and sinusoidal (second term) flows. For the sinusoidal
mixing length. A numerical investigation was carried out for a flow, Vo is the amplitude, f is the pulsing frequency, and ϕ is the
convergent–divergent microchannel with sinusoidal walls (space- phase lag between flows at Inlet 1 and Inlet 2. Fig. 1(b) shows the
periodic in the streamwise direction) under pulsed sinusoidal flow mean inlet velocity at Inlet 1 and Inlet 2 for the anti-phase (ϕ = 180◦ )
(time-periodic). The Reynolds number, Strouhal number and the pulsing. T is the time period of the pulse, and is defined as T = 1/f.
ratio of the pulsing amplitude to the steady flow velocity of the The symbols are marked at times t = 0, T/4, T/2, 3 T/4, and T for one
pulsatile flow, were used as the operating parameters in the per- complete cycle. At the outlet, zero static pressure was specified. A
formance evaluation. Also, a comparison was made with other no-slip condition was applied at the walls. The concentrations were
microchannel designs in terms of overall mixing performance with set to 1 and 0 at Inlet 1 and Inlet 2, respectively.
and without pulsing. Hexahedral cells were used to discretize the computational
domain. Along the streamwise direction, the grid lines were well
aligned with the geometry of the channel to capture the sinusoidal
2. Problem formulation
walls. A grid sensitivity test was performed for the mixing index as
the criterion to determine the total number of nodes for spatial res-
A schematic layout of the present problem is shown in Fig. 1(a).
olution. Five different grid systems with 51,219, 98,440, 160,116,
The profile of the convergent–divergent microchannel was gener-
276,861, and 449,072 nodes were tested. The grid with 276,861
ated using a sine function of the form y = A sin(2x/) where A is
nodes was found to be adequate for analysis with a relative error
the amplitude,  is the pitch or wavelength, and x and y are the
of less than 0.5% in the mixing index compared to a finer grid. For
spatial coordinates along the longitudinal and transverse direc-
the unsteady calculations, 40 time steps was used to capture the
tions, respectively. The values of A and  were taken to be 0.15
temporal behavior. This corresponds to time steps of 0.01, 0.005,
and 1.12 mm, respectively. The depth of the channel is 0.125 mm
0.0025, and 0.00125 s for 2.5, 5, 10, and 20 Hz pulsing, respectively.
as measured in the z-direction. The two inlets, Inlet 1 and Inlet 2, are
The present numerical simulations were performed by Intel Core
merged to the main channel with a T-joint as shown in Fig. 1. The
I7 (8 CPUs) with a clock speed of 2.94 GHz. Each calculation was
cross-sections of the inlet channels are rectangular with dimen-
subdivided into eight tasks. The computational time was typically
sions of 0.1 mm × 0.125 mm. The axial lengths of the connecting
12–15 h for each simulation.
channel Li , main channel Lm , and exit channel Le , are 0.2, 2.24, and
A second-order upwind differencing scheme was used to dis-
1.4 mm, respectively.
cretize the advection terms in the governing equations [21]. For
transient calculations, the discretization in time was performed
2.1. Flow and mixing analysis using a second-order backward Euler scheme. The SIMPLEC algo-
rithm [22] was used for pressure–velocity coupling. The linearized
The flow dynamics and mixing inside the microchannel were algebraic system of equations resulting from discretization was
studied by solving the unsteady continuity and Navier–Stokes solved using a multigrid accelerated incomplete lower-upper fac-
equations using ANSYS CFX-12.1 [19], a commercial computational torization procedure. Multigrid techniques significantly improved
fluid dynamics (CFD) software package based on the finite volume the convergence behavior. The criterion for convergence was the
method. The unsteady continuity and Navier–Stokes equations can normalized root mean square residual value of 10−6 . However, the
be expressed as follows: pulsatile flows were computed over enough time to obtain a peri-
 · V = 0 odic solution; namely, a solution that does not change measurably
∇ (1)
from one cycle to the next.
∂V  )V = − 1 ∇ p + ∇ 2 V
A variance-based method was employed to evaluate the mix-
+ (V · ∇ (2) ing efficiency of the micromixer. The variance of the species was
∂t 
determined on a cross-sectional plane perpendicular to the x-axis.
where V represents the fluid velocity, and  and  are the fluid den- Variance is based on the concept of the intensity of segregation,
sity and kinematic viscosity, respectively. For the multi-component which is based on the variance of concentration in relation to the
fluids, the bulk motion of the fluid is modeled by using single veloc- mean concentration. However, Glasgow and Aubry [16] have used
ity and pressure values, and each component has its own equation a weighting factor to account for the differences in the mass flow
for the conservation of mass. The mixing analysis was performed rate from the channel centerline to the walls. The variance of the
200 A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205

Fig. 1. Schematic of convergent–divergent microchannel and inlet flow velocities. (a) Geometry and (b) inlet flow velocities (ϕ = 180◦ and St = 0.139).

mass fraction of the mixture on a cross-sectional plane normal to cycle, MIo , was used to measure the overall mixing performance of
the flow direction can be expressed mathematically as follows: the micromixer.

 N
 (C − )2  V  3. Results and discussion
= i i
(4)
N Vmean
i=1 The prediction accuracy of the numerical methods used in the
present study was validated by comparison with the experimen-
where N is the number of sampling points on the plane, Ci is the tal and numerical results of Glasgow and Aubry [16] for the same
mass fraction at sampling point i, and is the optimal mixing mass conditions. In their experiment, the flow in a straight microchan-
fraction (=0.5, the mass fraction in the targeted case of equal mixing nel with two inlets (one in-line with, and the other perpendicular
of the two fluids). Vi is the velocity in the ith cell, and Vmean is the to, the channel) was considered. Fig. 2 shows the mass fraction
area-averaged fluid velocity at the selected cross-sectional plane. distributions for two different cases. In the first case, the inlets are
Finally, the mixing index at any cross-sectional plane perpendicular fed at Re = 0.3 without pulsing (Fig. 2(a)). The other case involves an
to the axial direction is defined as anti-phased pulsing (ϕ = 180◦ ) with Vo /Vs = 7.5, Re = 0.3, and St = 0.13
 (Fig. 2(b)). These two cases exhibit qualitatively similar mass frac-
N
i=1
((Ci − )2 /N)(Vi /Vmean ) tion distributions compared to those obtained by the experiment
MI = 1 − (5) [16].

As a preliminary step, the effect of pulsing the flow on mix-
The variance was taken to be the maximum for completely ing performance in the convergent–divergent microchannel was
unmixed fluids and the minimum for completely mixed fluids. A analyzed with different flow settings at the inlets. Also, in case
higher value of MI (close to 1) indicates a better mixing quality. of pulsing from both inlets, the phase difference, ϕ, varies from
To account for the time dependency of Eqs. (4) and (5), the mix- 0 to 180◦ . The various inlet flow settings and the corresponding
ing index was averaged over four values at quarter-cycle intervals. mixing indexes are listed in Table 1. It is found that the mixing per-
The mixing index 0.2 mm downstream of the end of the 2nd mixing formance of the convergent–divergent microchannel is maximum
A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205 201

Fig. 2. Comparison of dye mass fraction distribution on central x–y plane between (left) present computation and (right) experiment of Glasgow and Aubry [15] for two
cases: (a) without pulsing at both inlets (Re = 0.3), and (b) anti-phased pulsing (St = 0.13, Re = 0.3, and Vo /Vs = 7.5).

Table 1 Inlet 1 and Inlet 2, with dimensions of 0.1 mm × 0.125 mm as


Mixing performance of the convergent–divergent microchannel with pulsatile flow
shown in Fig. 1. Since the fluids in the serpentine channels (square-
(Re = 0.5, St = 0.278, and Vo /Vs = 1.88) with inflow conditions.
wave, zig-zag, and sinusoidal) travel different paths, the mixing
Inlet flow setting Mixing index length was fixed for the purpose of comparison. The mixing length
(a) No pulsing 0.14 was measured along the streamwise direction in each channel
(b) Pulsing from single inlet 0.40 and was kept constant at 2.4 mm. The anti-phased (180◦ phase
Pulsing from both inlets difference between inlets) pulsing from the inlets was used to
(c) In phase (ϕ = 0) 0.14 input fluids into the microchannels. The Reynolds number, Strouhal
(d) 90◦ phase difference (ϕ = 90◦ ) 0.35 number, and Vo /Vs were maintained at the following reference
(e) 180◦ phase difference (ϕ = 180◦ ) 0.83
values: Re = 0.5, St = 0.278 and Vo /Vs = 1.88, respectively. Table 2
lists the microchannel designs tested and the corresponding mix-
(MIo = 0.83) in the case of pulsing from both inlets with phase dif- ing indices calculated in this work, with and without pulsing. As
ference of 180◦ . Fig. 3 shows mass fraction contours on the x–y previously mentioned, the mixing index was calculated by aver-
plane at one-half of the channel depth at time instant, t = 0/T (data aging over four values at quarter-cycle intervals. It is noted that
were collected after the pulsing was fully established) for various the proposed convergent–divergent microchannel shows remark-
inlet conditions listed in Table 1. For constant flow at the inlets ably high mixing performance compared to other designs. In the
(no pulsing) (Fig. 3(a)), the flow velocity at each inlet corresponds case without pulsing, due to the low Reynolds number (Re = 0.5),
to Re = 0.5. The interface is straight and mixing is governed only all of the micromixer designs show low mixing indexes that are
by molecular diffusion. With pulsing, the Strouhal number, St and not much different from each other in a range from 0.14 to 0.18.
velocity ratio, Vo /Vs were fixed at 0.278, and 1.88, respectively. The convergent–divergent geometry shows the lowest mixing
With pulsing the flow at single inlet (Fig. 3(b)), the mixing per- index, 0.14, in the case without pulsing, but shows the highest
formance is found to improve as a result of interface perturbation mixing index, 0.83, in the case with pulsing among the tested
(MIo = 0.40). However, pulsing at both the inlets in phase (ϕ = 0◦ ) designs. In the case with pulsing, the convergent–divergent geom-
lead to the mixing index value (MIo = 0.14) same as that without etry improves the mixing index by 66.7% compared to the straight
pulsing as shown in Table 1. Fig. 3(c) shows that no stretching of channel (T-channel). This indicates that the effect of pulsing flow
the interface is observed and the interface remains straight along strongly depends on the microchannel geometry. Another impor-
the microchannel, which is same as the pattern shown in Fig. 3(a) tant parameter in pulsed flow mixing is the variation in mixing
for no-pulsing case. For ϕ = 90◦ (Fig. 3(d)), interface perturbation is with each pulse cycle; i.e., the difference between maximum and
observed with an improvement in mixing performance (MIo = 0.35). minimum mixings in each cycle [16]. During each cycle, the varia-
In case of anti-phase pulsing (ϕ = 180◦ ), a remarkable increase in the tions in the mixing index (MIo ) were found to be ±10%, ±9%, ±2%,
mixing performance is shown in Table 1 (MIo = 0.83), and a signif- ±5% and ±0.1% for straight, square-wave, zig-zag, sinusoidal, and
icant increase in the interfacial area between the co-flowing fluids convergent–divergent microchannels, respectively. This indicates
manifested by fine puffs in the microchannel as shown in Fig. 3(e). that the convergent–divergent microchannel also resulted in the
Also, from a practical viewpoint, the inlet flow condition with phase most stable mixing. Therefore, the convergent–divergent channel
difference of 180◦ is desirable because it provides a nearly constant represents the most effective coupling with pulsatile flow among
flow rate at the outlet. the tested cases.
In Table 2, the proposed convergent–divergent microchannel The effect of varying the Strouhal number on mixing perfor-
is compared with different micromixer designs, i.e., straight [16], mance is important for pulsatile flows. The Reynolds number
square-wave [6], zig-zag [7], and sinusoidal [23] found in the and velocity ratio were fixed at 0.5 and 1.88, respectively. For
literature, with and without pulsing the flow. All of the tested pulsed flow, the Strouhal number was varied over a range of
micromixer designs have two inlets connected by a T-junction, 0.070 ≤ St ≤ 0.556. A quantitative variation of the mixing index, MIo ,
202 A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205

Table 2
Mixing performance of various micromixers with pulsatile flow (Re = 0.5, St = 0.278, and Vo /Vs = 1.88).

Microchannel design Type MIo (no pulsing) MIo (pulsing) MIo variation over a cycle (%)

Straight channel 0.15 0.51 10

Square-wave 0.17 0.70 9

Zig-zag 0.16 0.68 2

Sinusoidal 0.18 0.66 5

Convergent–divergent 0.14 0.83 0.1

with the Strouhal number, St is depicted in Fig. 4. The mixing index performance at St = 0.556, as shown in Fig. 4. However, in the core
was calculated by averaging over four values at quarter-cycle inter- region, the diffusion occurs very rapidly due to the thinning of the
vals for a particular Strouhal number. It was observed that the puffs at this Strouhal number.
overall mixing performance was sensitive to the Strouhal number, The ratio of pulsing velocity Vo to the steady flow velocity Vs
and thus to the frequency of the pulsatile flow. At St = 0.278, the is also an important factor that affects the mixing characteristics.
mixing performance of the convergent–divergent microchannel Fig. 6 shows the variation in mixing index MIo with the velocity
was found to be maximum (83%). To explain the mixing behavior ratio Vo /Vs . Here, the Reynolds and Strouhal numbers were fixed at
with varying Strouhal number, mass fraction contours on the x–y Re = 0.5 and St = 0.278, respectively. It is noted that the mixing per-
plane at one-half of the channel depth are plotted at time instants formance increases rapidly with the increase in the velocity ratio;
t = 0/T (data were collected after the pulsing was fully established) i.e., the magnitude of pulsing relative to the steady flow velocity
for four different Strouhal numbers, as shown in Fig. 5. It can be seen for Vo /Vs less than 1.88. Beyond this velocity ratio, the increasing
that alternate puffs of liquids 1 and 2 from the two inlets are shed rate slows down, and the mixing index was found to be 90% at
along the streamwise direction at all Strouhal numbers. The shape Vo /Vs = 3.76. With a further increase in pulsing velocity, Vo /Vs = 7.52
and thickness of the puffs depend on the Strouhal number, with shows only a 2.0% increase in the mixing index. Fig. 7 shows the
the centerline of the channel being the line of symmetry. At low dye mass fraction distributions in the x–y plane at one-half of the
Strouhal numbers (St = 0.070 and 0.139), the impact of pulsing the channel depth at time t = 0/T (data were collected after the pulsing
flow is weak, and is manifested by large-sized puffs and low mix- was fully established). For Vo /Vs = 0.47, the distribution represents
ing performance. With an increase in Strouhal number (St = 0.278), a weak oscillatory behavior of the interface between two fluids.
the large puffs are broken down to produce finer striations along As the Vo /Vs ratio increases, the dominance of the pulsed flow
the streamwise direction. The puffs grow in size, starting with an increases, which results in a rapid distortion of the interface, and
arc-shaped design following the divergent slope, which increas- thus an enlargement of the interfacial area. This phenomenon was
ingly converges at the center as dictated by the convergent slope observed up to a Vo /Vs ratio of 0.94. However, a quite different pat-
of the sinusoidal wall. Moving farther downstream, the sizes of the tern was observed at Vo /Vs = 1.88: alternate puffs of the two mixing
puffs decrease as they enter the next divergent part of the chan- fluids were shed downstream, resulting in remarkable increase in
nel, with the shapes dictated by the convergent–divergent slope as the interfacial area. This is the reason why the mixing is enhanced
previously described. However, the puffs appear to be more elon- significantly at this velocity ratio, as shown in Fig. 7. With a further
gated with the diffusion, causing the puffs to dissipate in the bulk increase in Vo /Vs , the alternate puffs in the main flow remained
fluid. At this optimum Strouhal number (Fig. 5), the enhanced mix- unaffected, and a relatively small increase in the mixing index was
ing is due to enlargement of the interfacial area between the two observed as shown in Fig. 7.
fluids, which results from the reduced thickness of the puffs. Inter- The effect of Reynolds number on the mixing performance of the
estingly, a further increase in Strouhal number (St = 0.556) results convergent–divergent micromixer with pulsing inputs was ana-
in slightly different mixing characteristics. Even though the thick- lyzed as shown in Fig. 8. The Strouhal number and velocity ratio
ness of the puffs is further reduced, the reduction in the width were fixed at St = 0.278 and Vo /Vs = 1.88, respectively. The symbols
of the puffs overrides the effect of the thickness. This results in correspond to the Reynolds numbers, Re = 0.25, 0.5, 1, 2, and 4,
a reduction of the interfacial area, and the reduced puffs fail to respectively. It is found that the mixing index is almost constant
interact with the fluids in the near-wall region, leaving zones of with the Reynolds number. Therefore, the dynamics of mixing in the
unmixed fluids. These are the reasons for the decrease in the mixing convergent–divergent microchannel is invariant in this relatively
A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205 203

Fig. 5. Dye mass fraction distributions on the x–y plane at one-half of the channel
depth (z = 0.0625 mm) at time t = 0/T in a cycle with Strouhal number at Re = 0.5 and
Vo /Vs = 1.88.

short range of the Reynolds number based on the steady base flow,
and is solely governed by the dimensionless parameters, Strouhal
number, St, and velocity ration, Vo /Vs . The dye mass fraction distri-
butions on the x–y plane at one-half of the channel depth at time
t = 0/T were similar for different Reynolds numbers, as shown in
Fig. 5 for Re = 0.5, Vo /Vs = 1.88, and St = 0.278.
Fig. 3. Dye mass fraction distributions on the x–y plane at one-half of the chan- Since the mixing performance also depends on the geometry of
nel depth (z = 0.0625 mm) at time t = 0/T in a cycle for different settings of inlet the convergent–divergent sinusoidal walls, an exercise was per-
conditions marked in Table 1 with Vo /Vs = 1.88 and St = 0.278. formed to reduce the overall length of the microchannel. Fig. 9

Fig. 4. Variation of mixing index with Strouhal number at Re = 0.5 and Vo /Vs = 1.88.
Fig. 6. Variation of mixing index with velocity ratio Vo /Vs at Re = 0.5 and St = 0.278.
204 A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205

Fig. 9. Variation of mixing index with wavelength of the sinusoidal walls at Re = 0.5,
St = 0.278, and Vo /Vs = 1.88.

Fig. 7. Dye mass fraction distributions on the x–y plane at one-half of the channel
depth (z = 0.0625 mm) at time t = 0/T in a cycle with Vo /Vs at St = 0.278 and Re = 0.5.

Fig. 10. Variations of mixing index with Reynolds number, Strouhal number, veloc-
ity ratio, and wavelength of the sinusoidal walls (Re = 0.5, St = 0.278, and Vo /Vs = 1.88).

affecting the overall mixing performance of the micromixer


(MIo = 83%). To summarize, Fig. 10 shows the mixing characteris-
tics of the convergent–divergent microchannel under pulsatile flow
with Reynolds number (St = 0.278 and Vo /Vs = 1.88), Strouhal num-
ber (Re = 0.5 and Vo /Vs = 1.88), velocity ratio (Re = 0.5 and St = 0.278),
and ratio of wavelength to amplitude, /A (Re = 0.5, St = 0.278, and
Fig. 8. Variation of mixing index with Reynolds number at St = 0.278 and Vo /Vs = 1.88. Vo /Vs = 1.88). This figure confirms that the best mixing index in the
tested ranges of the parameters is 0.92 at Vo /Vs = 7.52.
shows the variation of the mixing index with the wavelength of the
microchannel for anti-phased (180◦ phase difference) input flows 4. Conclusions
for St = 0.278, Re = 0.5, and Vo /Vs = 1.88. It is found that the wave-
length of the walls (i.e., the axial length of the micromixer) can A convergent–divergent microchannel with sinusoidal channel
be changed in the range of /A from 6.0 to at least 7.5 without walls is proposed as the geometry that is best coupled with pulsatile
A. Afzal, K.-Y. Kim / Sensors and Actuators B 211 (2015) 198–205 205

inlet flows. Using Navier–Stokes analysis, the mixing character- [4] D.J. Harrison, K. Fluri, N. Chiem, T. Tang, Z. Fan, Micromachining chemical and
istics were analyzed with respect to the Reynolds and Strouhal biochemical analysis and reaction systems on glass substrates, Sens. Actuators
B 33 (1996) 105.
numbers, and the ratio of pulsing amplitude to steady flow velocity. [5] A.D. Stroock, S.K. Dertinger, A. Ajdari, I. Mezic, H.A. Stone, G.M. Whitesides,
The numerical results were validated for the mass fraction dis- Chaotic mixer for microchannels, Science 295 (2002) 647–651.
tributions with and without pulsing in a straight microchannel [6] R.H. Liu, M.A. Stremler, K.V. Sharp, M.G. Olsen, J.G. Santiago, R.J. Adrian, H. Aref,
D.J. Beebe, Passive mixing in a three-dimensional serpentine microchannel, J.
through comparison with experimental data. The mixing charac- Microelectromech. Syst. 9 (2000) 190–197.
teristics were tested for anti-phased pulsing and two periods of the [7] V. Mengeaud, J. Josserand, H.H. Girault, Mixing processes in a zigzag microchan-
sinusoidal walls of the convergent–divergent microchannel. The nel: finite element simulations and optical study, Anal. Chem. 74 (2002)
4279–4286.
reference conditions are Re = 0.5, St = 0.278, and Vo /Vs = 1.88. The
[8] M.K. Parsa, F. Hormozi, Experimental and CFD modeling of fluid mixing in
proposed convergent–divergent microchannel shows at least a 19% sinusoidal microchannels with different phase shift between side walls, J.
higher overall mixing index with the most stable mixing with time Micromech. Microeng. 24 (2014) 1–7.
[9] N.T. Nguyen, Z. Wu, Micromixers – a review, J. Micromech. Microeng. 15 (2005)
at the exit of micromixer compared to the other designs (such as
R1–R16.
straight, square-wave, zig-zag, and sinusoidal microchannels) for a [10] V. Hessel, H. Lowe, F. Schönfeld, Micromixers – a review on passive and active
fixed mixing length. The mixing index showed a strong dependence mixing principles, Chem. Eng. Sci. 60 (2005) 2479–2501.
on the Strouhal number of the pulsatile flow, with a maximum at [11] R.H. Liu, R. Lenigk, R.L. Druyor-Sanchez, J. Yang, P. Grodzinski, Hybridiza-
tion enhancement using cavitation microstreaming, Anal. Chem. 75 (2003)
St = 0.278 (0.070 ≤ St ≤ 0.556). The enhanced mixing performance at 1911.
the optimum Strouhal number (St = 0.278) was due to an enlarge- [12] H.H. Bau, J. Zhong, M. Yi, A minute magneto hydrodynamic (MHD) mixer, Sens.
ment of the interfacial area between the two fluids, resulting from Actuators B 79 (2001) 207–215.
[13] A.O.E. Moctar, N. Aubry, J. Batton, Electro-hydrodynamic micro-fluidic mixer,
the reduced thickness of the fluid puffs. The mixing performance Lab Chip 3 (2003) 273–280.
increases rapidly with an increase in the velocity ratio for Vo /Vs less [14] P. Tabeling, M. Chabert, A. Dodge, C. Jullien, F. Okkles, Chaotic mixing in cross-
than 1.88. However, beyond this value, the mixing index increases channel micromixers, Philos. Trans. R. Soc. Lond. A 362 (2004) 987–1000.
[15] F. Bottauski, C. Cardonne, C. Meinhart, I. Mezić, An ultrashort mixing length
slowly and reaches 0.92 at Vo /Vs = 7.52. The mass fraction distribu- micromixer: the shear superposition micromixer, Lab Chip 7 (2007) 396–398.
tions show that the interfacial area increases rapidly as the velocity [16] I. Glasgow, N. Aubry, Enhancement of microfluidic mixing using time pulsing,
ratio increases until Vo /Vs = 1.88, where the interface separates to Lab Chip 3 (2003) 114–120.
[17] I. Glasgow, S. Lieber, N. Aubry, Parameters influencing pulsed flow mixing in
produce discrete puffs of fluids. From the variation of the mix-
microchannels, Anal. Chem. 76 (2004) 4825–4832.
ing index with Reynolds number at a fixed Strouhal number and [18] I. Glasgow, N. Aubry, October, California, USA, 7th International Conference on
velocity ratio of the pulsatile flow (St = 0.278 and Vo /Vs = 1.88), the Miniaturized Chemical and Biochemical Analysis Systems, 2003.
[19] CFX-12.1, Solver Theory, ANSYS Inc., Canonsburg, PA, 2006.
mixing performance was found to be independent of Re in a range
[20] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomenon, Wiley, New York,
of Reynolds number, 0.25 ≤ Re ≤ 4. It was found that the mixing per- 1960.
formance does not change with the axial length of the micromixer [21] M. Liu, Computational study of convective–diffusive mixing in a microchannel
in the range of /A from 6.0 to at least 7.5. With a simple in- mixer, Chem. Eng. Sci. 66 (2011) 2211–2223.
[22] J.P. Van Doormaal, G.D. Raithby, Enhancement of the SIMPLE method for pre-
plane structure, the proposed convergent–divergent micromixer dicting incompressible fluid flows, Numer. Heat Transf. 7 (1985) 147–163.
combined with pulsing inputs provides efficient mixing in a short [23] A. Afzal, K.-Y. Kim, Mixing performance of passive micromixer with sinusoidal
length. channel walls, J. Chem. Eng. Japan 46 (3) (2013) 230–238.

Acknowledgments Biographies

This work was supported by the National Research Foundation A. Afzal received B.Tech (Mechanical Engineering) and M.Tech (Thermal Engineer-
of Korea (NRF) grant (No. 20090083510) funded by the Korean ing) degrees from Aligarh Muslim University (AMU) in India, in 2006 and 2009,
respectively, and Ph.D. degree from Inha University, South Korea in 2015. His main
government (MSIP) through Multi-phenomena CFD Engineering
research interests are design and optimization of micromixers using CFD, surrogate
Research Center. The authors gratefully acknowledge this support. modeling and optimization algorithms.

Kwang-Yong Kim received the B.S. degree from Seoul National University in 1978,
References and the M.S. and Ph.D. degrees from the Korea Advanced Institute of Science and
Technology (KAIST), Korea in 1981 and 1987, respectively. He is currently an Inha
[1] Y. Li, F. Xu, C. Liu, Y. Xu, X. Feng, B.-F. Liu, A novel microfluidic mixer based Fellow Professor in the School of Mechanical Engineering of Inha University, Incheon,
on dual-hydrodynamic focusing for interrogating the kinetics of DNA–protein Korea. Professor Kim is also the current Editor-in-Chief of the International Journal
interaction, Analyst 138 (2013) 4475–4482. of Fluid Machinery and Systems (IJFMS), associate editor of ASME Journal of Fluids
[2] H.A. Stone, S. Kim, Microfluidics: basic issues, applications and challenges, Engineering, and the chairman of Asian Fluid Machinery Committee (AFMC). He is
AIChE J. 47 (2001) 1250–1254. a Fellow of the American Society of Mechanical Engineers (ASME) and an Associate
[3] G.S. Jeong, S. Chung, C.-B. Kim, S.-H. Lee, Applications of micromixing technol- Fellow of the American Institute of Aeronautics and Astronautics (AIAA).
ogy, Analyst 135 (2010) 460–473.

You might also like