Sie sind auf Seite 1von 9

Proceedings of the ASME 2017 International Mechanical Engineering Congress and Exposition

IMECE17
November 3-9, 2017, Tampa, Florida, USA
DRAFT

IMECE2017-7228
Correlation of Mixing Efficiency and Entropy Generation Rate in a Square Cross Section Tee Junction
Micromixer

Aric M. Gillispie Evan C. Lemley


Department of Engineering and Physics Department of Engineering and Physics
University of Central Oklahoma University of Central Oklahoma
100 N. Univ. Dr. 100 N. Univ. Dr.
Edmond, Oklahoma 73034 Edmond, Oklahoma 73034
USA USA

NOMENTCLATURE ABSTRACT
𝐷ℎ = Hydraulic Diameter The potential applications of micromixers continues
to expand in the bio-sciences area. In particular, passive
𝑉𝑎𝑣𝑒 = Average Velocity
micromixers that may be used as part of point-of-care
𝑅𝑒 = Reynolds Number (POC) diagnostic testing devices are becoming common-
𝑄 = Volume Flow Rate place and have application in developed, developing, and
relatively undeveloped locales. Characterizing and
𝐿𝑒 = Entry Length
improving mixing efficiency in these devices is an
𝑎, 𝑏 = Duct Width, Height ongoing research effort. Micromixers are used in some
𝑇 = Temperature lab-on-chip (LOC) devices where it is often necessary to
combine two or more fluids into a mixed solution for
𝑥, 𝑦, 𝑧 = Standard Cartesian directions
testing or delivery. The simplest micromixer incorporates
𝑢, 𝑣, 𝑤 = Directional velocity components a tee junction to combine two fluid species in anti-parallel
𝑈, 𝑉, 𝑊 = Non-dimensional velocity components branches, with the mixed fluid leaving in a branch
perpendicular to the incoming branches. Micromixers rely
𝑠̇ ′′′ = Volumetric Entropy Generation Rate
on two modes of mixing: chaotic advection and molecular
𝑣𝐴 = Molal Volume diffusion. In micro-mixers flow is typically laminar,
𝐹 = Association Factor making chaotic advection occur only via induced
secondary flows. Hence, micromixers, unless carefully
𝜌 = Density
designed, rely almost exclusively on molecular diffusion
µ = Viscosity of fluid species. A well designed micromixer should
𝜔, 𝜉, 𝜁 = Arbitrary Constants exhibit significant chaotic advection; which is also a sign
of large strain rates and large entropy generation rates.
𝛼 = Aspect Ratio (b/a)
This paper describes the development of an analytical
𝛾 = Non-dimensionalized weighting factor relationship for the entropy generation rate and the mixing
𝜎 = Standard deviation efficiency as function of the outgoing branch Reynolds
number. Though there has been extensive research on tee
𝛷 = Non-dimensional Concentration junctions, entropy generation, and the mixing efficiencies
Subscripts of a wide variety of micromixers, a functional relationship
𝑜 = Outflow for the mixing efficiency and the entropy generation rate
𝐹 = Viscous Term has not been established. We hypothesize a positive
𝐶 = Concentration Term correlation between the mixing index and the entropy
𝑇 = Temperature Term
generation rate. The worked described here establishes a Where:
method and provides the results for such a relationship.
A basic tee junction with square cross sections has µ 𝜕𝑢 2 𝜕𝑣 2 𝜕𝑤 2
been analyzed using computational fluid dynamics to 𝑠̇𝐹 ′′′ = [2 (( ) + ( ) + ( ) )
𝑇𝑜 𝜕𝑥 𝜕𝑦 𝜕𝑧
determine the entropy generation rate and outgoing
mixing efficiencies for Reynolds numbers ranging from 𝜕𝑢 𝜕𝑣 2 𝜕𝑢 𝜕𝑤 2 𝜕𝑣 𝜕𝑤 2
+( + ) + ( + ) +( + ) ] 4
25-75. The mixing efficiency is determined at a location 𝜕𝑦 𝜕𝑥 𝜕𝑧 𝜕𝑥 𝜕𝑧 𝜕𝑦
in the outgoing branch where the effects of molecular
diffusive mixing is minimized and chaotic advective
mixing is the focus. The entropy generation rate has been
determined for the indicated range of Reynolds number ′′′ 𝐾 𝜕𝑇 2 𝜕𝑇 2 𝜕𝑇 2
𝑠̇ 𝑇 = 2 [( ) + ( ) + ( ) ] 5
and decomposed into its viscous and diffusive entropy 𝑇𝑜 𝜕𝑥 𝜕𝑥 𝜕𝑥
terms. The functional relationships that have been
developed are applicable for micromixer design and serve
as a reference for more complex passive micromixer
designs. 𝑅𝐷 𝜕𝐶 2 𝜕𝐶 2 𝜕𝐶 2
𝑠̇𝐶𝐶 ′′′ = [( ) + ( ) + ( ) ] 6
𝐶𝑜 𝜕𝑥 𝜕𝑦 𝜕𝑧

INTRODUCTION
When considering micromixers, there exist two
important quantities that are often considered: mixing 𝑅𝐷 𝜕𝑇 𝜕𝐶 𝜕𝑇 𝜕𝐶 𝜕𝑇 𝜕𝐶
efficiency and entropy generation rate. The literature is 𝑠̇𝐶𝑇 ′′′ = [( ) ( ) + ( ) ( ) + ( ) ( )] 7
𝑇𝑜 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧
replete with many novel micromixer designs that focus
predominantly on the mixing efficiencies of devices used However, in the isothermal and adiabatic case where
for biomedical, medical, mechanical, heat exchanger and temperature and heat transfer affects are neglected:
chemical applications [1-14]. Hossain, et al. [3] provides
a method for determining the mixing index MI by 𝑠̇ ′′′ = 𝑠̇𝐹 ′′′ + 𝑠̇𝐶𝐶 ′′′ 8
considering the mass fractions of different fluid species
on an outgoing plane. The MI is defined as:
This paper explores a common micromixer geometry for
𝜎2
lab-on-chip (LOC) devices is a rectangular microchannel
𝑀𝐼 = 1 − √ 2 1 with an aspect ratio of:
𝜎 𝑚𝑎𝑥

Where 𝜎 2 is the variance of the mass fraction and 𝜎𝑚𝑎𝑥


2
is 𝑏
𝛼= 9
the maximum variance that would occur if the fluids were 𝑎
completely unmixed. Furthermore, When the channel is defined as in Figure 1.

a
𝑁
1
𝜎 2 = √ ∑(𝐶𝑖 − 𝐶̅ )2 2
𝑁 b
𝑖=1

Where Ci is the local mass fraction and 𝐶̅ is the average


mass fraction of the plane. In addition to the MI, the Figure 1. Rectangular channel geometry with labeled
entropy generation rate is an important design dimensions.
consideration for micromixers, and the available literature
discusses entropy generation in many different A square channel would then be defined by α = 1. The
applications [15-24]. Concisely, the entropy generation simulations presented here are conducted at two different
can be described as the irreversibility of any measured scales 1) a macro scale on the order of centimeters, and 2)
quantity in a system. When considering viscous, thermal, a microscale on the order of tens of microns such that the
and concentration based irreversibility the volumetric results would have the possibility of being validated using
entropy generation is defined as: either macro or micro particle image velocimetry (PIV)
techniques. However, the results are based on flow cases
that share the same dimensionless Reynold number,
𝑠̇ ′′′ = 𝑠̇𝐹 ′′′ + 𝑠̇ 𝑇 ′′′ + 𝑠̇𝐶𝐶 ′′′ + 𝑠̇𝐶𝑇 ′′′ 3 defined as:

2 Copyright © 2017 by ASME


𝜌𝑉𝑎𝑣𝑒 𝐷ℎ are already non-dimensionalized quantities. To non-
𝑅𝑒 = 10
µ dimensionalize Equation 8 the flowing terms are defined:
The microscale simulations are assumed to satisfy the
requirement that any surface roughness does not 𝑢𝜌𝐷ℎ 𝑣𝜌𝐷ℎ 𝑤𝜌𝐷ℎ
𝑈= , 𝑉= , 𝑊=
significantly affect the flow of the fluid [26]. The analysis 𝑅𝑒𝑜 µ 𝑅𝑒𝑜 µ 𝑅𝑒𝑜 µ
is conducted for each scale over the range of target 𝑥 𝑦 𝑧
intermediate Reynolds numbers 20-75. The analyses 𝑋= , 𝑌= , 𝑍=
𝐷ℎ 𝐷ℎ 𝐷ℎ
described here have been conducted using computational
fluid dynamics (CFD). 𝐶 − 𝐶𝑜
𝛷=
𝛥𝐶
Using these factors leads to:
PROBLEM DESCRIPTION
𝑠̇𝑛𝑑 ′′′ = 𝛾1 𝑠̇𝐹,𝑛𝑑 ′′′ + 𝛾2 𝑠̇𝐷,𝑛𝑑 ′′′ 15
The analysis described here seeks a hypothesized
simple analytical expression that relates the MI and the
Where:
entropy generation rate separately to the outgoing
𝑠̇ ′′′ 𝑇𝑜 𝜌2 𝐷ℎ2
Reynolds number with a desired form of: 𝑠̇𝑛𝑑 ′′′ = 16
µ3
𝑓(𝑀𝐼) = 𝑓(𝑅𝑒𝑜 ) + 𝜉 11
And
𝜕𝑈 2 𝜕𝑉 2 𝜕𝑊 2
𝑓(𝑠̇ ) = 𝑓(𝑅𝑒𝑜 ) + 𝜔 12 𝑠̇𝐹,𝑛𝑑 ′′′ = [2 (( ) + ( ) + ( ) )
𝜕𝑋 𝜕𝑌 𝜕𝑍
Where 𝛾 and 𝜉 are constants to be determined. The
results that are obtained should be independent of scale 𝜕𝑈 𝜕𝑣 2 𝜕𝑈 𝜕𝑊 2 𝜕𝑉 𝜕𝑊 2
+( + ) + ( + ) +( + ) ] 17
and applicable for the entire range of target Reynolds 𝜕𝑌 𝜕𝑋 𝜕𝑍 𝜕𝑋 𝜕𝑍 𝜕𝑌
number. These relationships should hold for any
parameters in a simple tee micromixers where the flow
combines from anti-parallel branches with symmetric 𝜕𝛷 2 𝜕𝛷 2 𝜕𝛷 2
flow parameters (Figure 2) such that: 𝑠̇𝐶𝐶,𝑛𝑑 ′′′ = [( ) + ( ) + ( ) ] 18
𝜕𝑋 𝜕𝑌 𝜕𝑍

𝑄1 + 𝑄2 = 𝑄𝑜 = 2𝑄1 13
And 𝛾1 = 𝑅𝑒𝑜2 19
and
𝑅𝑒1 + 𝑅𝑒2 = 𝑅𝑒𝑜 = 2𝑅𝑒1 14
2𝑅𝐷𝑇𝑜 𝜌2 𝐷ℎ2 𝛥C
𝛾2 = 20
µ3
Where 𝛾 is a non-dimensionalized weighting factor. The
results obtained from this analysis should serve as guide
and method for determining similar relationships for more
complex geometries. A relationship of this form would
allow for simple determination of the expected mixing
efficiency and entropy generation rate of a mixer, even for
flow parameters outside those for which a mixer was
initially designed.

METHODOLOGY
The results for this paper were obtained using CFD.
Specifically, the geometries were designed in
SolidWorks®, meshed in ANSYS® ICEM™ CFD,
Figure 2. Alternate channel models. analyzed in ANSYS® Fluent™, and post processed in
VisIt™.
Two geometries were created, both with square
Additionally, in pursuing relationship that are devoid of
channels of uniform side length (Dh). The macroscale
scale, it is recognized that the entropy generation rate be
model was designed with Dh = 2.54 cm, and the
appropriately non-dimensionalized. The MI and Reynolds
microscale model was designed with Dh = 25.4 µm. The

3 Copyright © 2017 by ASME


branches that meet at the origin of the tee junction are Where the striation length, stl(0), is 0.5Dh for this case.
each sufficiently long to guarantee the fluid is fully Non-dimensionalizing Equation 23 leads to:
developed before entering the junction, and allow for
enough length to guarantee that the fluid would be fully ∗
𝑅𝑒 ∗ µ
𝑡𝑑𝑖𝑓𝑓 = 24
developed when exiting the geometry satisfying the 8𝐷𝜌
outflow boundary condition. The branch length was
determined to need to be a minimum length of: Clearly lower Reynolds numbers will achieve complete
mixing in a shorter distance as a result of the higher
residence time of the fluid. Therefore, to account for this
𝐿𝑒 one should consider forcing each flow to have the same
= 0.06𝑅𝑒𝑜,𝑚𝑎𝑥 21
𝐷ℎ residence time to view the relative impact of chaotic
advection alone, as this is most important mode of
mixing. An arbitrary 0.1% of the total diffusion time is
taken for each Reynolds number, or more compactly:
The model as designed is presented in Figure 3.

𝑡 ∗ = 0.001 ∗ 𝑡𝑑𝑖𝑓𝑓 25

With the outgoing length being:

0.001
𝐿𝑜𝑢𝑡 = 𝑄 26
8𝐷 𝑜

Using this method, the entropy generation will be


Figure 3. 3D model of the geometry. calculated for the volume bounded by Lout and the mixing
efficiency is determined for the outgoing face at Lout
(Figure 4).
After designing the models, a few preliminary
decisions had to be conducted prior to conducting the
grid-independence study. First of all, since this work
focuses in large part on the mixing efficiency, it had to be
determined how the fluid species would interact at a
molecular level. The diffusivity of fluid A into fluid B is
given by the Wilke-Chang correlation [25]:

7.4 ∗ 10−8 (𝑭𝑀𝐵 )0.5 𝑇


𝐷𝐴𝐵 = 22
µ𝐷 𝑣𝐴0.6

Additionally, since the mixing efficiency is to be taken on


an outgoing face, it was desired to seek standardized way
of examining the mixing efficiencies for the range of Figure 4. Schematic of the Volume bounded by Lout
intermediate Reynolds numbers. There are two modes of
mixing that are expected to occur in this type of mixer:
chaotic advection and diffusion. It is well known that With these design considerations in place, the
diffusion should not be relied on for mixing, simply geometry of Figure 3 was imported into ICEM CFD™ for
because it is a function of time, and contact area. In a meshing. In pursuit of a sufficient mesh a grid
simple tee mixer as examined here, there is no lamination independence study was conducted for a uniform
for the intermediate Reynolds numbers, as there exists a hexahedral mesh. The study was conducted the
single interface when the fluids combine. For this reason, parameters of entropy generation rate and the standard
lower Reynolds flows would benefit more from diffusion deviation of the mass fraction, as these are the two
over the same lengths. This is made evident when relevant quantities being considered. The grid
considering the diffusion time to obtain a MI=1: independence study was carried out at the maximum
𝑠𝑡𝑙(0)2 Reynolds number. The assumptions made for the
𝑡𝑑𝑖𝑓𝑓 = 23 computational analysis were: Newtonian, laminar, steady
2𝐷 state, pressure driven, isothermal, adiabatic and fully

4 Copyright © 2017 by ASME


developed at the outlet. The walls are considered non-slip
and insulated. The input parameters are shown in Table 1.

Table 1. Fluid properties in Fluent™.


Parameter Value

Range of Reynolds Numbers 0.1-100

Diffusivity (m2s-1) 2.68*10-9

Density (kg*m-3) 998.2

Viscosity (Pa*s) 0.001003 Figure 7. Preliminary Hexahedral Mesh with 270461 Nodes
per Dh3.
Temperature (K) 300

The analysis was conducted using ANSYS® Fluent®, with


second order upwind volume discretization by
interpolating the field variables located at the cell centers
to the adjoining faces of each control volume. Pressure-
velocity coupling using the SIMPLE solution scheme was
used for the solutions. All of the residuals were set to 10 -6,
even during the grid independence study. The first and
final mesh iterations are shown in Figure 5 and 7 with
their corresponding velocity magnitude color plots shown
in Figure 6 and 8. The contour plots were used as a means
of visually determine the quality of the mesh. The grid
independence study is shown in Table 2. Figure 8. Velocity Magnitude Plot for the Mesh in Figure 7.

Table 2. Grid Independence Study.

Mesh 1 2 3 4 %𝐷𝑖𝑓𝑓( 𝑁, 𝑁 − 1)
𝑁𝑜𝑑𝑒𝑠
649 4628 34864 270461 ----- ---- -----
𝐷ℎ3
𝑆𝑓̇ ∗ 109 1.43 1.79 1.93 1.99 19.9 7.41 2.76

Figure 5. Preliminary Hexahedral Mesh with 649 Nodes per 𝜎 0.47011 0.48148 0.484569 0.48241 2.36 0.64 0.45
Dh3.

After meshing, the simulations were conducted for


Reynolds numbers from 20 to 75. The process involved
initializing each case, then exporting the files to compute
in parallel on a supercomputing cluster machine. After
solving each of the flow cases for both scales, the results
were read into Fluent™ and exported for post processing
in VisIt™. The simulations totaled 6886.5 computer core
hours.

Figure 6. Velocity Magnitude Plot for the Mesh in Figure 5.

5 Copyright © 2017 by ASME


RESULTS
To post-process the results, the necessary files were
imported into VisIt™. The entropy generation and the MI and MI* vs Reo
0.040 8
mixing efficiency functions were written for direct
calculation in VisIt™ and the specific Reynolds number 0.035 MI MI*
based values were analyzed. First, consider the mixing 6
efficiency for the macro-scale model as a function of the

MI*
MI
0.030
outgoing Reynolds number (Figure 9). MI* = 0.051Reₒ - 0.0807
4
0.025 R² = 0.998

MI vs Reₒ 0.020 2
40 55 70 85 100 115 130 145 160
0.040 Reo
Figure 10. Linearized mixing efficiency.
0.035
The following relationships emerge:
MI

0.030
0.025
𝑀𝐼 ∗ 𝑃 = 𝑀𝐼 ∗ = 0.051𝑅𝑒𝑜 − 0.087 29
0.020
40 55 70 85 100 115 130 145 160 With
Reₒ
𝑃 = 1.038𝑅𝑒𝑜 + 42.711 30
Figure 9. Mixing Index vs the Outgoing Reynolds number for
the macro scale geometry.
Plotting the macro scale with the micro scale results leads
to Figure 11.
The mixing efficiencies are clearly not very significant,
but the purpose of this paper are not to demonstrate a high
efficiency micromixer, but rather to develop a simple MI* vs Reo
linear expression relating the mixing efficiency and 8
entropy generation to the outgoing Reynolds number. For
this to be possible, a linearizing function should be 6 MI* MI*(µ)
developed to such that a relationship would emerge of the
MI*

form:
4
𝑀𝐼 ∗ 𝑃 = 𝑀𝐼 ∗ = 𝛽 ∗ 𝑅𝑒 + 𝜁 27
2
Where P is some non-dimensional normalizing function 40 55 70 85 100 115 130 145 160
and 𝛽 and 𝜁 are constants to be determined. The desired Reo
form of P would satisfy: Figure 11. Comparting macro and micro mixing efficiencies.

P = 𝑓(𝑅𝑒𝑜 ) 28 Clearly, the mixing efficiencies are devoid of scale so


Equation 29 holds for any scale.
After numerically determining the normalizing function However, looking at the entropy generation as a
the results in Figure 10 are obtained. function of the Reynolds number for both the macro and
micro scales, another relationship emerges. Figures 12
and 13 show the plots for the macro and micro scales
respectively.

6 Copyright © 2017 by ASME


Ṡnd vs Reo Ṡnd,F and Ṡnd,cc vs Reo
8.00E-03
30000000
Sdot'''F(nd) Sdot'''cc(nd)

Ṡnd,F and Ṡnd,cc


25000000 6.00E-03
20000000
4.00E-03
Ṡnd

15000000
10000000 2.00E-03
5000000
0.00E+00
0 40 55 70 85 100 115 130 145 160
40 55 70
100 115 130 145 160 85 Reo
Reo
Figure 12.Macroscale non-dimensionalized entropy generation. Figure 15. Microscale non-dimensionalized entropy generation
constituents.

Ṡnd vs Reo (µ)


2.00E+02 By breaking the entropy generation into its constituents
1.50E+02
(Equation 17 and 18), the plots now show a very similar
trend with the values differing only by magnitude of the
difference in the scale of the models. This is easily
Ṡnd

1.00E+02
remedied by considering the final relationships:
5.00E+01
0.00E+00 ̇
(𝑆𝐹,𝑛𝑑 )
40 55 70 85 100 115 130 145 160 = (0.0461𝑅𝑒𝑜 + 0.0802) 31
𝑆𝐹
Reo
̇
(𝑆𝑐𝑐,𝑛𝑑 )
Figure 13.Microscale non-dimensionalized entropy generation.
= (0.0203𝑅𝑒𝑜 + 0.0309) 32
𝑆𝐹

Clearly, a single relationship would not be possible that Where SF = 0.01 in standard SI units, and scales down
holds at every scale, so another approach needs to be accordingly as in Table 3.
taken. Of interesting note is the linear nature of the macro
scale model and the quadratic nature of the microscale
model. This could have actually been hypothesized when Table 3. Sample values of SF.
considering the scale based contributions of the 𝛾 Scale of the Model (1 for SI) SF
weightings factors discussed previously. So, as a second
attempt to create a general relation for the non- 𝑘(103 ) 100000
dimensionalized entropy generation as a function of the
Reynolds number, the entropy generation is broken into 1(100 ) 100
its non-dimensionalized constituents as seen in Figures 14
and 15. 𝑐(10−2 ) 1

𝑚(10−3 ) 0.1
Ṡnd,F and Ṡnd,cc vs Reo
8
Sdot'''F(nd) Sdot'''cc(nd) µ(10−6 ) 0.0001
6
Ṡnd,F and Ṡnd,cc

y = 0.0461x + 0.0802
R² = 0.9971
4
CONCLUSIONS
2 Simple analytical relationships for the mixing
y = 0.0203x + 0.0309
R² = 0.9985 efficiency and the entropy generation rate have been
0 determined as a function of the outgoing Reynolds
40 55 70 85
100 115 130 145 160 number in a simple tee mixing geometry for a range of
Reo intermediate Reynolds numbers at two different scales.
Figure 14. Macroscale non-dimensionalized entropy generation This was achieved through a novel means of non-
constituents. dimensionalizing the constituents of the entropy

7 Copyright © 2017 by ASME


generation terms as well as normalizing the mixing [8] Lu, L. H., Ryu, K. S., & Liu, C. (2002). A
efficiency, and taking the results to be bound by a volume magnetic microstirrer and array for microfluidic
that limits the observed impact of diffusive mixing. Using mixing. Journal of microelectromechanical
this method, the results allow for determination of the systems, 11(5), 462-469.
relative impact of the most prominent mixing term in
micromixers, chaotic advection. The work presented here [9] Muradoglu, M., & Stone, H. A. (2005). Mixing in
provides a clear methodology for developing simple a drop moving through a serpentine channel: a
analytical relationships for these important design computational study. Physics of Fluids (1994-
parameters. These and similar relationships provide a present), 17(7), 073305.
method for quickly estimating the theoretical maximum
mixing efficiency for design parameters outside the scope [10] Devahastin, S., & Mujumdar, A. S. (2002). A
of the mixer designed. The benefit of this is evident when numerical study of flow and mixing
a micromixer is to be repurposed outside the scope of the characteristics of laminar confined impinging
original design. streams. Chemical Engineering Journal, 85(2),
215-223.

[11] Wang, S. J., Devahastin, S., & Mujumdar, A. S.


REFERENCES (2005). A numerical investigation of some
[1] Ahmed, D., Mao, X., Juluri, B. K., & Huang, T. J. approaches to improve mixing in laminar
(2009). A fast microfluidic mixer based on confined impinging streams. Applied thermal
acoustically driven sidewall-trapped engineering, 25(2), 253-269.
microbubbles. Microfluidics and nanofluidics,
7(5), 727. [12] Song, H., Bringer, M. R., Tice, J. D., Gerdts, C. J.,
& Ismagilov, R. F. (2003). Experimental test of
[2] Jeong, G. S., Chung, S., Kim, C. B., & Lee, S. H. scaling of mixing by chaotic advection in droplets
(2010). Applications of micromixing technology. moving through microfluidic channels. Applied
Analyst, 135(3), 460-473. Physics Letters, 83(22), 4664-4666.

[3] Hossain, S., Husain, A., & Kim, K. Y. (2011). [13] Soleymani, A., Kolehmainen, E., & Turunen, I.
Optimization of micromixer with staggered (2008). Analytical and experimental
herringbone grooves on top and bottom walls. investigations of liquid mixing in T-type
Engineering Applications of Computational Fluid micromixers. Chemical engineering journal, 135,
Mechanics, 5(4), 506-516. S219-S228.

[4] Shi, X., Xiang, Y., Wen, L. X., & Chen, J. F. [14] Mengeaud, V., Josserand, J., & Girault, H. H.
(2012). CFD Analysis of flow patterns and (2002). Mixing processes in a zigzag
micromixing efficiency in a Y-type microchannel microchannel: finite element simulations and
reactor. Industrial & Engineering Chemistry optical study. Analytical chemistry, 74(16), 4279-
Research, 51(43), 13944-13952. 4286.

[5] Stroock, A. D., Dertinger, S. K., Ajdari, A., [15] Gregg, M. C. (1984). Entropy generation in the
Mezić, I., Stone, H. A., & Whitesides, G. M. ocean by small-scale mixing. Journal of physical
(2002). Chaotic mixer for microchannels. Science, oceanography, 14(4), 688-711.
295(5555), 647-651.
[16] McEligot, D. M., Walsh, E. J., Laurien, E., &
[6] Yang, J. T., Huang, K. J., & Lin, Y. C. (2005). Spalart, P. R. (2008). Entropy generation in the
Geometric effects on fluid mixing in passive viscous parts of turbulent boundary layers.
grooved micromixers. Lab on a Chip, 5(10), Journal of Fluids Engineering, 130(6), 061205.
1140-1147.
[17] Bejan, A. (1996). Entropy generation
[7] Shih, T. R., & Chung, C. K. (2008). A high- minimization: The new thermodynamics of finite‐
efficiency planar micromixer with convection and size devices and finite‐time processes. Journal of
diffusion mixing over a wide Reynolds number Applied Physics, 79(3), 1191-1218.
range. Microfluidics and Nanofluidics, 5(2), 175-
183. [18] Magherbi, M., Abbassi, H., Hidouri, N., &
Brahim, A. B. (2006). Second law analysis in

8 Copyright © 2017 by ASME


convective heat and mass transfer. Entropy, 8(1),
1-17.

[19] Sanchez, M., Henderson, A. W., Papavassiliou, D.


V., & Lemley, E. C. (2012, July). Entropy
Generation in Laminar Flow Junctions. In ASME
2012 Fluids Engineering Division Summer
Meeting collocated with the ASME 2012 Heat
Transfer Summer Conference and the ASME 2012
10th International Conference on Nanochannels,
Microchannels, and Minichannels (pp. 325-330).
American Society of Mechanical Engineers.

[20] Del Giudice, S., Nonino, C., & Savino, S. (2007).


Effects of viscous dissipation and temperature
dependent viscosity in thermally and
simultaneously developing laminar flows in
microchannels. International journal of heat and
fluid flow, 28(1), 15-27.

[21] Koo, J., & Kleinstreuer, C. (2004). Viscous


dissipation effects in microtubes and
microchannels. International Journal of Heat and
Mass Transfer, 47(14), 3159-3169.

[22] Schmandt, B., & Herwig, H. (2011). Loss


coefficients in laminar flows: essential for the
design of micro flow systems. PAMM, 11(1), 27-
30.

[23] Schmandt, B., & Herwig, H. (2011). Internal flow


losses: A fresh look at old concepts. Journal of
Fluids Engineering, 133(5), 051201.

[24] Schmandt, B., & Herwig, H. (2011). Diffuser and


nozzle design optimization by entropy generation
minimization. Entropy, 13(7), 1380-1402.

[25] Wilke, C. R., & Chang, P. (1955). Correlation of


diffusion coefficients in dilute solutions. AIChE
Journal, 1(2), 264-270.

[26] Phares, D.J., Smedley, G.T., Zhou, J., 2005,


“Laminar flow resistance in short microtubes”,
Intl. J. of Heat Fluid Flow, 26, 506-512.

9 Copyright © 2017 by ASME

Das könnte Ihnen auch gefallen