You are on page 1of 40

https://www.sciencedirect.

com/science/article/pii/S0009250918306900

Accepted Manuscript

Optimal Reynolds number for liquid-liquid mixing in helical pipes

Michael Mansour, Prafull Khot, Dominique Thévenin, Krishna D.P. Nigam,


Katharina Zähringer

PII: S0009-2509(18)30690-0
DOI: https://doi.org/10.1016/j.ces.2018.09.046
Reference: CES 14522

To appear in: Chemical Engineering Science

Received Date: 15 May 2018


Accepted Date: 26 September 2018

Please cite this article as: M. Mansour, P. Khot, D. Thévenin, K.D.P. Nigam, K. Zähringer, Optimal Reynolds
number for liquid-liquid mixing in helical pipes, Chemical Engineering Science (2018), doi: https://doi.org/10.1016/
j.ces.2018.09.046

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Optimal Reynolds number for liquid-liquid mixing in
helical pipes

Michael Mansoura,b,∗, Prafull Khota , Dominique Thévenina , Krishna D.P.


Nigamc , Katharina Zähringera
a Lab.of Fluid Dynamics & Technical Flows, University of Magdeburg “Otto Von
Guericke”, 39106 Magdeburg, Germany
b Mechanical Power Engineering Department, Faculty of Engineering - Mattaria, Helwan

University, 11718 Cairo, Egypt


c Indian Institute of Technology - Delhi, New Delhi 110016, India

Abstract

A computational study has been performed to determine the mixing behaviour


of two miscible liquids in helical pipes with different geometrical dimensions.
To evaluate the mixing efficiency between the two liquids, the scalar transport
technique has been employed. The main objectives are to investigate the in-
fluence of the geometrical parameters on the mixing behaviour and to find the
optimal values of Reynolds number (Re) corresponding to the highest possi-
ble mixing efficiencies, while keeping acceptable pressure drop. The Reynolds
number has been varied broadly from Re=5 to Re=4,000, corresponding to
Dean numbers from De=1 to De=1,500, covering a wide range of laminar flows
relevant for practical applications. Three main geometrical parameters of the
helical pipe have been considered; the coil pitch, the pipe diameter, and the coil
diameter were varied in the ranges of 16–60 mm, 5–15 mm, and 70–150 mm,
respectively. The results show that an increase in coil pitch results generally in
lower mixing efficiencies, negligibly higher pressure drops, and lower values of
optimal Reynolds numbers. Increasing the pipe diameter leads to lower mixing
efficiencies but significantly lower pressure drops. Finally, as the coil diameter
is increased, the mixing efficiency and the pressure drop are slightly reduced,

∗ Correspondingauthor
Email address: michael.mansour@ovgu.de (Michael Mansour)

Preprint submitted to Chemical Engineering Science August 20, 2018


while the optimal Reynolds number is moderately shifted to higher values. Fur-
thermore, two optimal values of Reynolds number could be obtained for all
configurations, leading systematically to high mixing performance, Re≈40 and
Re≈750. The pressure drop for the lower value of the Reynolds number is con-
siderably lower than that of the second optimal value, while mixing efficiency
is comparable. Therefore, it is recommended to operate liquid-liquid mixing in
helical reactors at Re≈40. This value corresponds to the Dean number range of
7 ≤ De ≤ 15 for all the considered geometries.
Keywords: Liquid-liquid mixing, Optimal Reynolds number, Helical pipes,
Computational fluid dynamics.

Nomenclature

Roman symbols vy Velocity in y axis [m/s]



De Dean number (Re δ)
c̄ Surface-average mass fraction Re Reynolds number (ρU D/µ)
A Cross-sectional area [m2 ] Sc Schmidt number (µ/ρDab )
2
Af Grid face area [m ]
c Local grid face mass fraction Greek Symbols
D Coil diameter [mm]
d Pipe diameter [mm] δ Curvature ratio (d/D)
Dab Mass diffusivity [m2 /s] γ Dimensionless pitch (P/πD)
L Coil length [m] µ Dynamic viscosity [Pa s]
L1 First liquid ρ Fluid density [kg/m3 ]
L2 Second liquid θ Coil angle measured from the
Mc Mixing coefficient inlet surface of the coil [◦ ]
n Number of turns ζ Vorticity vector [1/s]
P Coil pitch [mm] ζz Z-component of the vorticity
U Average velocity [m/s] [1/s]
vx Velocity in x axis [m/s]

2
1. Introduction

Flow mixing is very important for many industrial processes and engineering
applications, including chemical industry, pharmaceutical industry, food pro-
cessing, waste water treatment, paper industry, and heat and mass transfer ap-
5 plications [1–5]. Due to their advantages, curved tubes are widely used as static
mixers in several applications such as power production, electronics, waste heat
recovery, and space applications [6–11]. In helical pipes – a widely employed
type of curved pipes –, very efficient mixing can be achieved with low energy
consumption (low pressure drop) and no maintenance (no moving parts), com-
10 pared to the use of a stirrer or of a static mixer. The main reason for that is the
generation of secondary flows in such pipes, which beneficially boost the radial
mixing [12]. Dean [13, 14] was the first to investigate the secondary flows in
curved pipes, showing a pair of vortices formed in the cross-sectional plane due
to unbalanced centrifugal forces.
15 One commonly technique used to enhance mixing in pipes is by increasing
turbulence level. However, the pressure loss then increases drastically, leading
to a much higher energy consumption [15, 16]. Additionally, in some cases, it is
quite difficult to reach turbulent regimes, such as for highly viscous fluids and
small-diameter tubes [17]. Nevertheless, several studies showed that mixing in
20 curved pipes is much more efficient than in straight pipes [18, 19], due to the
generated secondary flow. Moreover, turbulence onset is delayed in helical tubes
[20–23].
Considering the purely diffusive mixing limit of laminar flows in straight
pipes, many passive mixers have been proposed to support efficient mixing,
25 such as split-and-recombine mixers, advection mixers, etc. [24–29]. Split-and-
recombine mixers [30, 31] are often not very efficient. Advection mixers rely
on the large-scale motion of fluids, involving complex mixing elements and high
pressure drops.
Another configuration called the coiled-flow inverter (CFI) was first intro-
30 duced by Saxena and Nigam [32]. Using CFI, enhanced mixing and heat trans-

3
fer could be achieved [33–39]. The CFI is based on the principle of bending
of helical coils to inverse the flow and the generated vortices, improving mix-
ing. Other geometrical modifications were also used as techniques to improve
mixing, including non-circular cross-sectional pipes, modified flow paths, pipe
35 grooves, and chaotic configurations [40–45]. Nevertheless, for significant mixing
improvement, several flow inversions and/or complicated geometry modifica-
tions are always needed, giving rise to an undesired increase in pressure drop.
Due to the generated Dean vortices, the flow in helical pipes offers a very high
mixing efficiency; however, it also involves higher pressure drops compared to
40 straight pipes to overcome curvature-induced pressure losses [20, 46, 47]. Using
optimal flow conditions in helical pipes, mixing can take place very efficiently in
the laminar flow regime [11], avoiding high pressure drops and eliminating the
need for turbulent flow or complex geometries. Vanka et al. [19] and Mansour
et al. [11] studied the effect of Reynolds number Re and Schmidt number Sc on
45 mixing efficiency in helical pipes, where:

ρU d
Re = (1)
µ

µ
Sc = (2)
ρDab

In these equations, ρ is the fluid density, U is the average flow velocity,


d is the pipe diameter, µ is the fluid dynamic viscosity, and Dab is the mass
diffusivity from the first fluid into the second fluid. It was shown that the mixing
50 efficiency is as expected lower for higher Schmidt numbers, due to the reduced
diffusion. However, the influence of Sc was found to be only important at low
Reynolds numbers, Re < 40. At higher Reynolds numbers, the flow mixing is
dominated by the flow structures and the secondary flow. Mansour et al. [11]
showed also that there are two optimal values of Reynolds number (Re≈ 50 and
55 Re≈ 1, 000) in the range investigated (Re between 5 and 10,000), which lead
to excellent mixing conditions between the fluids. However, this investigation

4
was limited to only one single helical pipe geometry and was therefore not
applicable in general, since previous studies have reported the strong influence
of the geometrical parameters on the intensity and shape of the secondary flows
60 generated in helix tubes [13, 48–50]. For instance, it was found that the increase
in coil pitch (P ) decreases the centrifugal force, weakening the developed Dean
vortices [20, 51–53]. However, as coil pitch increases to very large values, the
secondary flow decreases and the flow in helical coils approaches the one in a
straight pipe. Several studies in the literature reported that the curvature ratio
65 δ (the ratio between pipe and coil diameters, δ = d/D) might be even more
important [20, 52–54]; higher curvature ratios create stronger secondary flows,
which result in better mixing [18].
Therefore, the aim of the present work is to identify optimal values of the
Reynolds number corresponding to the highest possible mixing efficiency. At
the difference of the results of [11], the obtained values must be valid in general,
for different helix geometries. For this purpose, starting from the basic geom-
etry investigated experimentally in [55] and numerically in [11], the coil pitch,
the pipe diameter and the coil diameter as the main geometrical parameters
of helical pipes have been systematically varied, leading to 7 different helical
geometries. The ranges of all parameter were selected based on the scientific
literature to ensure significant influence on the secondary flow strength and
structure [49, 50, 56, 57]. Considering the results of [11], the Reynolds number
was varied from 5 to 4,000, covering broadly the previously observed peaks in
terms of mixing efficiency. The corresponding Dean number (De) range for all
the studied configuration is 1 ≤ De ≤ 1, 500, where

De = Re δ (3)

The Schmidt number was kept constant at Sc =1,000, corresponding to water


or aqueous solvents as a working fluid. For all cases, the obtained pressure drop
70 has been monitored. Analyzing the results of this study, values of the Reynolds
number leading to optimal mixing conditions will be derived, that are useable
for a wide range of helical pipes.

5
2. Numerical Modeling

The CFD code Star-CCM+ was used in the present study to simulate the
75 hydrodynamics and flow mixing in the helical pipes. Considering the flow of
two identical miscible liquids, the species transport model was used to calculate
the mixing between them. All details concerning the numerical modeling and
the governing equations can be found in [11] and are not repeated here in the
interest of space.
80 No-slip boundary condition for the velocity and zero-gradient conditions for
the scalars were imposed on the wall. For the studied Re range, a laminar
simulation setup was used, since the flow is laminar in all considered helical
configurations up to a transitional range of approximately 7, 000 ≤ Re ≤ 10, 000
[20–22]. A segregated, steady-state solver was employed, which is suitable for
85 the considered low to moderate Re range (laminar flow), as used previously in
[11, 58]. For all simulations, the numerical solution was considered as converged,
when the normalized residuals were lower than 10−8 for the continuity and
momentum equations, and below 10−6 for the species transport equation.

3. Mixing Efficiency

90 In the present study the area-weighted uniformity index was used to calculate
the uniformity of the distribution of each liquid in a tube cross-section, based
on mass fraction fields. The uniformity index of mass fraction (also called in
short mixing coefficient Mc ) was already described and employed in [11]. To
obtain the mixing coefficient Mc , the mass fraction of either of the fluids can be
95 used in the present two-fluid system. Corresponding results will be compared
and averaged in what follows. The equation defining the mixing coefficient over
a cross-sectional area A is written:

P
f |c − c̄| Af
Mc = 1 − P (4)
|c̄| f Af

6
with

Z
1
c̄ = c dA (5)
A

100 where c is the local mass fraction value of either liquid L1 or L2, c̄ is the
average mass fraction over the tube cross-sectional area A, and Af is the area
of an elementary cell face. The mixing coefficient can vary from 0 to 1, where
0 indicates no mixing at all between the fluids (0% mixing), and 1 indicates
complete mixing (100% mixing).

105 4. Helical Pipe Geometries

Figure 1 shows the main geometrical parameters for an exemplary helical


pipe. The geometrical parameters of the different configurations considered in
the present study are listed in Table 1. The first geometry (G1) represents the
basic geometry, which was previously studied experimentally and numerically
110 in [55] and [11], respectively. The coil pitch (P ) is increased from 16 mm to
30 mm and then 60 mm going from G1 to G2 to G3. The pipe diameter (d)
is changed from 5 mm to 10 mm and then to 15 mm going from G4 to G1,
and then to G5. Similarly, the coil diameter (D) is varied from 70 mm to 118
mm and then 150 mm going from G6 to G1, and then G7. Table 1 shows also
115 the curvature ratio (δ = d/D) and the dimensionless pitch (γ = P/πD) of each
configuration. Here, the pipe and coil diameters have been studied separately,
since they have strongly different effects on the pressure drop [50, 57, 59].
Initially, a number of n = 5 coil turns were considered in the simulations for
all geometries, since it was previously shown that mixing occurs predominantly
120 in the very first turns [11, 58]. However, to provide a fair comparison when
comparing the final mixing efficiency at the outlet surface among different ge-
ometries, the total coil length should be kept constant. Therefore, the number
of turns was changed for G2, G3, G6 and G7 to keep a constant tube length
L = 1.855 m, corresponding to a straight pipe case.

7
Figure 1: Geometrical parameters of a helical pipe.

Geometry no. P , mm d, mm D, mm L, m δ = d/D γ = P/(πD) n


1 16 10 118 1.855 0.0847 0.0431 5
2 30 10 118 1.855 0.0847 0.0809 4.99
3 60 10 118 1.855 0.0847 0.1618 4.94
4 16 5 118 1.855 0.0423 0.0431 5
5 16 15 118 1.855 0.1271 0.0431 5
6 16 10 70 1.855 0.1428 0.0727 8.41
7 16 10 150 1.855 0.0666 0.0339 3.93

Table 1: Geometrical parameters of the different helical pipes considered in this study.

125 The physical properties of the liquids were considered to be those of water,
with a density of ρ = 998.2 kg/m3 and a dynamic viscosity of µ = 1.003 10−3
Pa s. For the considered range of Re (5 ≤ Re ≤ 4, 000), a Hagen-Poiseuille
parabolic velocity profile is used at the pipe inlet, since it was shown in [11]
that a flat velocity field at the inlet can lead to unrealistic mixing results. The
130 inlet surface was divided, by a line parallel to the coil axis, into two identical
half-disks, each half being occupied by only one of the two liquids, as shown in
Figure 1. For all simulations, such a parallel liquid interface was kept at the
inlet, since it was found to lead to the highest mixing compared to perpendicular
or inclined interfaces [11]. Based on the folding action of baker’s transformation
135 [60, 61], the mixing is more efficient, when the vortices appearing as secondary
flow are perpendicular to the liquid-liquid separation interface. For the outlet
surface of the coil, a constant-pressure boundary condition was applied.

8
5. Meshing and Validation

The meshing parameters used in [11] were employed again in the present
140 study. The employed numerical model and discretization have been validated
in [11] against experimental velocity measurements of [62], and were shown to
reproduce very accurately the experimental results. For the basic geometry
(G1), about 2 millions hexahedral cells were used, as shown in Figure 2. For the
other geometries, scale factors were used to always generate cell elements of the
145 same size as in G1, ensuring a constant resolution in space for all configurations.
In this manner, all the results can be directly compared to each other.

Oulet surface

L1 L2
Inlet surface

Figure 2: Grid topology used in the present study for configuration G1.

6. Results and Discussion

6.1. Axial evolution of mixing coefficient


Figure 3 shows (i) the development of the mass fraction and (ii) the corre-
150 sponding mixing coefficients of each liquid individually (L1 and L2) along the
axial length of G1. In general, an oscillating behaviour of the mass fractions
is observed in the first part of the coil due to local accumulation of one liquid.
However, as the liquids become mixed together, the oscillations of the mass
fraction decay along the axial length. When comparing Figure 3a and 3b, it
155 can also be seen that the period of the oscillations is higher at a lower Reynolds
number. In this case the mixing is very slow, since the secondary flow is very
weak for low Reynolds numbers.
The oscillating behaviour of the mass fraction is also observed when look-
ing at the calculated mixing coefficient for each liquid. Finally, the average

9
160 mixing coefficient determined from both individual mixing coefficients shows a
smoothly increasing behaviour. For the considered range of Re, the oscillations
were found to be stable is time. This was checked by performing additional un-
steady simulations; after an initial transient, the obtained mass fractions do not
change any more in time and space. In order to quantify oscillations and track
165 local liquid accumulations, the mixing coefficient of liquid L1 has been used to
quantify the mixing efficiency in the rest of the present study, and will always
be shown and discussed in what follows. As shown in Figure 3, The values of
the two mixing coefficients are different only within the first part (half-length)
of the coil. Therefore, using any of them will not affect the final results of the
170 mixing coefficient in the remaining part of the coil or at the coil outlet.
Number of turns, n Number of turns, n
0 1 2 3 4 5 0 1 2 3 4 5
Mixing coefficient, Mc

1.0 1.0
Mass fraction

0.8 0.8
0.6 0.6
0.4 0.4 L1
L1 L2
0.2 0.2
L2 Average
0.0 0.0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L Dimensionless axial length, X/L
(i) Mass fraction     (ii) Mixing coefficient

(a) Re=5
Number of turns, n Number of turns, n
0 1 2 3 4 5 0 1 2 3 4 5
Mixing coefficient, Mc

1.0 1.0
Mass fraction

0.8 0.8
0.6 0.6
0.4 0.4 L1
L1 L2
0.2 0.2
L2 Average
0.0 0.0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L Dimensionless axial length, X/L
(i) Mass fraction     (ii) Mixing coefficient  

(b) Re=500

Figure 3: Evolution of (i) mass fractions and (ii) corresponding mixing coefficients of liquids
L1 and L2 in G1 for (a) Re=5 and (b) Re=500.

10
6.2. Effect of geometrical parameters on the evolution of mixing

6.2.1. Effect of coil pitch


Figure 4 shows the distribution of mass fraction of liquid L1 at different axial
sections along the coil length in configurations G1 and G3. Here, the influence
of varying the coil pitch between the maximum (G3) and minimum (G1) values
considered are shown for three different values of the Reynolds number. The
first and the second row of the figure present the streamlines and the vorticity
contours of the secondary flow, respectively, for each corresponding condition at
the mid-length of every geometry. The vorticity vector is a measure of the local
rotation rate of fluid elements, which is used here in the analysis as a measure
of secondary flow strength in order to compare differences in mixing behaviour
among different cases. The vorticity is equal to the curl of the velocity field:

~ ×V
ζ=∇ ~ (6)

In the analysis, the z-component of the vorticity vector was used, which
describes the rotation rate in each considered section as shown in Figure 4. The
175 z-component of vorticity is calculated by:

∂vy ∂vx
ζz = − (7)
∂x ∂y
where vx and vy are the velocity components in x and y directions, respectively.
The angle θ is the angle between the local section and the plane of the inlet
section, and is used to denote the axial location of each section. For instance,
θ=180◦ corresponds to a half-turn of the coil, θ=360◦ to one full turn of the
180 coil; and so on.
Generally, it can be seen in Figure 4 that the mixing is quite weak for Re=5,
compared to Re=50 and Re=1,000 (see also Figure 5). The reason for poor
mixing at Re=5 is that the secondary flow is very weak as well at such low
values of the Reynolds number; hence, the flow behaviour is very close to that
185 in a straight pipe. In this case the mixing is dominated by diffusion. At such low
values of Re, a higher coil pitch results in an improved mixing efficiency, due to

11
the higher torsion (flow twisting), which increses the vorticity values. This can
be observed from the given minimum and maximum vorticity values, where the
higher pitch produces vorticity values more than 3 times the values of the lower
190 pitch. It was already shown in [63], that an increase in coil pitch increases flow
vorticity significantly for low Reynolds numbers (Re< 40). However, for a high
Reynolds number (Re=1,000) the flow mixing is dominated by the convection
of the curvature-induced vortices. In this case, a higher pitch deteriorates the
structure of the secondary flow [20, 51] without having a significant effect on
195 the vorticity value, and thus is detrimental concerning mixing efficiency (see last
two columns of Figure 4).
It can be indeed observed from the streamlines in Figure 4 that the sec-
ondary flow structures are worsened for higher pitch values, with shifted vortex
cores and more inclined symmetry lines. Similar observations can be found for
200 instance in [49, 50, 56]. The strong inclination of the vortices reduces even
more the final mixing efficiency. This can be easily explained, since the opti-
mal mixing conditions discussed in [11] and obtained when liquid interface and
symmetry line of the vortices are perpendicular to each other, are not fulfilled
any more.
205 Figure 5 shows the axial evolution of the mixing coefficient for different values
of coil pitch and Reynolds number. Again, a higher pitch is only relevant for
mixing at very low Re values, as shown in Figure 5a. After the development of
strong secondary flows at higher values of Re, a lower pitch is always preferable
to improve mixing, as shown in Figure 5c. In between, at moderate Reynolds
210 numbers (here, Re=50), no significant influence of the coil pitch on mixing
efficiency can be seen (Figure 5b).

12
  Re=5 Re=50 Re=1000
  G1 G3 G1 G3 G1 G3
(P=16mm) (P=60mm) (P=16mm) (P=60mm) (P=16mm) (P=60mm)
Streamlines
(mid-length)

   

Vorticity
contours
   
Vorticity Min= -0.018 Min= -0.066 Min= -0.64 Min= -0.82 Min= -84 Min= -82.4
values (/s) Max= 0.016  Max= 0.059  Max= 0.70  Max= 0.96  Max= 87.6  Max= 99.8 

θ=180°

   

θ=360°

   

θ=540°

   
Distribution
θ=720° of mass
fraction
   

θ=900°

   

θ=1080°

   

θ=1260°

   
 

θ=1440°

   

θ=1620°

   

θ=1800°

   
 
Figure 4: Effect of coil pitch on the axial evolution of the mass fraction of liquid L1 for
different Reynolds numbers.

13
1.0

Mixing coefficient, Mc
0.8

0.6

0.4
G1 (P=16mm)
0.2 G2 (P=30mm)
G3 (P=60mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  
    (a) Re=5
1.0
Mixing coefficient, Mc

0.8

0.6

0.4
G1 (P=16mm)
0.2 G2 (P=30mm)
G3 (P=60mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  
    (b) Re=50
1.0
Mixing coefficient, Mc

0.8

0.6

0.4
G1 (P=16mm)
0.2 G2 (P=30mm)
G3 (P=60mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  
    (c) Re=1,000

Figure 5: Effect of coil pitch on the evolution of the mixing coefficient for (a) Re=5, (b) Re=50
and (c) Re=1,000.

14
6.2.2. Effect of pipe diameter
Considering now the pipe diameter influence, the distributions of mass frac-
tion of liquid L1 at different axial sections in configurations G4 (smallest diam-
215 eter) and G5 (largest diameter) are compared for three different values of Re
in Figure 6. The streamlines and the vorticity contours are again shown in the
first two rows of Figure 6. Once again, at very low Re values, there is almost
no secondary flows; thus, mixing is controlled by diffusion and is very weak.
In Figure 6, it can be seen that for the same Reynolds number, a lower pipe
220 diameter leads to a noticeably faster mixing performance. This occurs due to
the higher velocities available in lower diameters, which significantly increase
the intensity of vorticity as shown by the maximum and minimum vorticity
values. At Re=50, the Dean vortices are more noticeable for the larger diameter;
nevertheless, even in this case, the vorticity strength is overall weaker in Case
225 G5, which slightly reduces the mixing speed.
Figure 7 shows the axial evolution of mixing coefficient, considering different
values of pipe diameters and Reynolds number. Similar to the results discussed
previously, the mixing is generally lower for very low Reynolds numbers (Re=5).
In this case, reducing the pipe diameter leads to a significant improvement of
230 the final mixing coefficient at pipe outlet.
For Re=50 and Re=1,000, there are no noticeable differences concerning the
final value of the mixing coefficient. Nevertheless, the mixing process is faster for
lower pipe diameters, increasing so for higher Reynolds numbers. For instance,
for d = 5 mm and Re=1,000, the mixing coefficient reaches a value higher than
235 90% just after the first turn; while, for d = 15 mm and Re=1,000, a value of
90% of the mixing coefficient is first found within the fourth turn of the coil.
Concerning the influence of the pipe diameter, lower values are generally rec-
ommended to improve mixing performance, regardless of the Reynolds number.
However, at Re=50, the mixing behaviour is comparable for all values of pipe
240 diameter. In this case, large diameters are also beneficial to significantly reduce
the pressure drop, as will be shown later.

15
  Re=5 Re=50 Re=1000
  G4 G5 G4 G5 G4 G5
(d=5mm) (d=15mm) (d=5mm) (d=15mm) (d=5mm) (d=15mm)
Streamlines
(mid-length)

   

Vorticity
contours
 
Vorticity Min = -0.07 Min = -0.009 Min = -1.5 Min = -0.446 Min = -214 Min = -52
values Max = 0.066  Max = 0.008  Max = 1.6  Max = 0.475  Max = 224  Max = 54.3 

θ=180°

θ=360°

θ=540°

Distribution
θ=720° of mass
fraction

θ=900°

θ=1080°

θ=1260°

θ=1440°

θ=1620°

θ=1800°

 
Figure 6: Effect of pipe diameter on the axial evolution of the mass fraction of liquid L1 for
different Reynolds numbers.

16
1.0

Mixing coefficient, Mc
0.8

0.6

0.4
G4 (d=5mm)
0.2 G1 (d=10mm)
G5 (d=15mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  
    (a) Re=5
1.0
Mixing coefficient, Mc

0.8

0.6

0.4
G4 (d=5mm)
0.2 G1 (d=10mm)
G5 (d=15mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  
    (b) Re=50
1.0
Mixing coefficient, Mc

0.8

0.6

0.4
G4 (d=5mm)
0.2 G1 (d=10mm)
G5 (d=15mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  

(c) Re=1,000

Figure 7: Effect of pipe diameter on the evolution of the mixing coefficient for (a) Re=5, (b)
Re=50 and (c) Re=1,000.

17
6.2.3. Effect of coil diameter
Figure 8 shows the distribution of mass fraction of liquid L1 at different
axial sections in configurations G6 (lowest coil diameter) and G7 (largest coil
245 diameter) for various values of the Reynolds number. Looking at the structure
of the vortices in the two geometries, no significant change can be observed
concerning vortex shape or vorticity contours. However, the vorticity values in
helical pipes are getting considerably weaker when increasing the coil diameter
as shown in Figure 8. From a theoretical point of view, for very high coil
250 diameters, the behaviour of a straight pipe should be recovered, leading to very
poor mixing.
For Re=5, no strong change of the mixing performance can be observed
in Figure 8, when changing the coil diameter from D = 70 mm to D = 150
mm. For moderate (Re=50) and high Reynolds numbers (Re=1,000), the mass
255 fraction distribution appears to be slightly more homogeneous for the larger coil
diameter. However, this effect is simply due to the longer length available for
mixing within one turn when increasing coil diameter.
A fair comparison of the evolution of axial mixing is shown in Figure 9.
There, the mixing coefficient is plotted as a function of the dimensionless coil
260 length, considering a constant length for all geometries. For Re=5, the compar-
ison shows only slight changes of the final mixing coefficient value at the outlet.
Similarly, for Re=1,000, the behaviour of axial mixing is very similar for the
three coil diameters considered in the study. However, at Re=50, a lower coil
diameter is preferable for moderate Reynolds numbers, since it leads to a more
265 intense secondary flow, and therefore to a significantly faster mixing. Therefore,
lower coil diameter values (higher coil curvatures) are recommended for better
mixing. Similar conclusions, regarding the effect of curvature ratio on mixing
at low Reynolds numbers (Re< 100), were drawn for example in [18].

18
  Re=5 Re=50 Re=1000
  G6 G7 G6 G7 G6 G7
(D=70mm) (D=150mm) (D=70mm) (D=150mm) (D=70mm) (D=150mm)
Streamlines
(mid-length)

   

Vorticity
contours
 
Vorticity Min = -0.273 Min = -0.013 Min = -1.16 Min = -0.55 Min = -126.7 Min = -74
values Max = 0.324  Max = 0.014  Max = 1.012  Max = 0.51  Max = 118.8  Max = 71.5 

θ=180°

   

θ=360°

   

θ=540°

   
Distribution
θ=720° of mass
fraction
   

θ=900°

   

θ=1080°

   

θ=1260°

   
 

θ=1440°

   

θ=1620°

   

θ=1800°

   
 
Figure 8: Effect of coil diameter on the axial evolution of the mass fraction of liquid L1 for
different Reynolds numbers.

19
1.0

Mixing coefficient, Mc
0.8

0.6

0.4
G6 (D=70mm)

0.2 G1 (D=118mm)
G7 (D=150mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L  

(a) Re=5
1.0
Mixing coefficient, Mc

0.8

0.6

0.4
G6 (D=70mm)
0.2 G1 (D=118mm)
G7 (D=150mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L

(b) Re=50
1.0
Mixing coefficient, Mc

0.8

0.6

0.4
G6 (D=70mm)

0.2 G1 (D=118mm)
G7 (D=150mm)
0.0
0 0.2 0.4 0.6 0.8 1
Dimensionless axial length, X/L

(c) Re=1,000

Figure 9: Effect of coil diameter on the evolution of the mixing coefficient for (a) Re=5, (b)
Re=50 and (c) Re=1,000.

20
6.3. Optimal Reynolds number
270 In this section, the final mixing coefficients obtained at the outlet of the coil
are compared as function of the Reynolds number. The objective is to check
if (an) optimal value(s) of the Reynolds number(s) can be found, which would
correspond to the highest mixing efficiency at coil outlet. Considering first
a purely qualitative analysis, 1) when the Reynolds number is increased, the
275 secondary flow becomes more intense, improving convective mixing; however,
2) increasing Re also decreases the residence time of the liquids within the coil,
decreasing mixing efficiency. Therefore, the influence of the Reynolds number
on mixing is expected to be complex and to reveal optimal conditions, as already
discussed in [11] for a fixed coil geometry.
280 Figure 10 shows the final mixing coefficient of L1 obtained at the outlet as
a function of the Reynolds number, considering all relevant geometrical param-
eters. The effect of varying coil pitch, pipe diameter, and coil diameter are
shown in Figures 10a, 10b, and 10c, respectively. All curves show two peaks in
mixing coefficient. In what follows, the corresponding conditions will systemat-
285 ically be called first, and second optimal value of the Reynolds number, written
respectively Reopt,1 and Reopt,2 .
Looking first at Figure 10a, a higher pitch can only lead to better mixing
performance for Re < 20, when the secondary flow is extremely weak. For Re
> 20, increasing coil pitch systematically reduces mixing efficiency. Comparing
290 all three subfigures in Figure 10, it is clear that the coil pitch plays the most
important role regarding mixing performance, compared to pipe diameter and
coil diameter, apart for Re ≤ 10. When increasing the coil pitch to P = 60 mm,
the mixing is strongly reduced at the second optimal Reynolds number.
Regarding now the influence of the pipe diameter shown in Figure 10b, sig-
295 nificant variations are only visible for either very low (Re < 20) or very high
(Re > 1, 000) Reynolds numbers. In the middle range (20 ≤ Re ≤ 1, 000),
the overall mixing performance is almost independent from the pipe diameter.
Still, the smallest diameter (d = 5 mm) shows always a slightly higher mixing
efficiency, for all values of Re. Accordingly, low pipe diameters can generally be

21
1.00

0.95

Mixing coefficient, Mc
0.90

0.85

0.80
G1 (P=16mm)
0.75 G2 (P=30mm)
G3 (P=60mm)
0.70
1 10 100 1000 10000
Reynolds number, Re  

(a) Coil pitch


1.00

0.95
Mixing coefficient, Mc

0.90

0.85

0.80 G4 (d=5mm)
G1 (d=10mm)
0.75
G5 (d=15mm)
0.70
1 10 100 1000 10000

Reynolds number, Re
   
(b) Pipe diameter
1.00

0.95
Mixing coefficient, Mc

0.90

0.85

0.80 G6 (D=70mm)
G1 (D=118mm)
0.75
G7 (D=150mm)
0.70
1 10 100 1000 10000
Reynolds number, Re
   
(c) Coil diameter

Figure 10: Effect of different geometrical parameters on optimal Reynolds number, considering
(a) Coil pitch, (b) Pipe diameter, and (c) Coil diameter.

22
300 preferable to enhance mixing. However, all pipe diameters show very good mix-
ing near the optimal values of Re. Here, large diameters are still recommended
to significantly reduce pressure drop (as will be shown later in Figure 12b).
Looking finally at Figure 10c, the effect of varying coil diameter on mixing is
obviously significant for low Re vales (Re < 20). There, increasing the coil diam-
305 eter results in very poor mixing, particularly so at Re=10; at this point, leading
to almost no secondary structures, the flow is very close to that in a straight
pipe. Additionally, the secondary flow is stronger for lower coil diameters, shift-
ing the peaks of Re to slightly lower values. At higher Reynolds numbers, the
final mixing coefficient depends only very slightly on the coil diameter.
310 Considering all the different geometries studied, the effect of the Reynolds
number can be globally divided into 6 ranges. For very low Reynolds numbers
(Re < 20), the mixing is very poor due to the the absence of strong secondary
flows, confirming previous observations from the literature [11, 18]. In this
first range, mixing is only controlled by molecular diffusion between liquids,
315 as explained previously. In the second range (20 ≤ Re ≤ 60), the secondary
flow is strong enough to dominate the mixing process, producing finally a very
good mixing efficiency for all cases, culminating at an average value of first
optimal Reynolds number, Reopt,1 ≈ 40. Between Re=60 and Re=200, no im-
portant changes are observed concerning the secondary flow, while the increase
320 of Reynolds number strongly reduces the residence time, which decreases the
mixing efficiency, particularly so for large coil pitches.
In the fourth range, 200 < Re < 500, the mixing efficiency improves again
in spite of the further reduction in residence time. This is due to the significant
increase in the vorticity strength as already shown in Figures 4, 6, and 8. The
325 fifth range, covering the second optimal value of the Reynolds number with an
average value of Reopt,2 ≈ 750, is found from almost Re=500 to slightly higher
than Re=1,000. In this range, a very good mixing behaviour is again observed
for all geometries considered. Finally, increasing further the Reynolds number
beyond 1,000 reduces the residence time significantly without increasing the
330 secondary flow, which decreases continuously the mixing efficiency for all cases.

23
6.4. Optimal conditions regarding pressure drop
For practical purposes, increasing mixing efficiency is obviously an advan-
tage, but is only interesting as long as the associated pressure drop stays in
an acceptable range; operation costs obviously increase directly with increasing
335 pressure losses. In this section, the analysis is extended to take into account
this issue. Figure 11a shows the effect of increasing the pitch on the mixing
coefficient around the two optimal Reynolds numbers. The peaks of mixing effi-
ciency (corresponding to the first optimal Reynolds numbers Reopt,1 in the left
figures, and to the second optimal Reynolds numbers Reopt,2 in the right fig-
340 ures) are marked by circles and connected by dotted lines. As shown previously
in Figure 10a, for P = 16 mm, P = 30 mm, and P = 60 mm, the corresponding
first peaks of Mc occur at Reopt,1 =40, Reopt,1 =40, and Reopt,1 =25, while the
second peaks of Mc are found at Reopt,2 =1,000, Reopt,2 =800, and Reopt,2 =500,
respectively.
345 Confirming previous observations, an increase in coil pitch results in a re-
duction of the peak mixing efficiency. For example, by increasing the pitch from
P = 16 mm to P = 60 mm, the first optimal value Mc decreases from Mc = 0.99
to Mc = 0.97, while the second optimum Mc decreases from Mc = 0.986 to
Mc = 0.95, respectively. Now, Fig. 11b illustrates the pressure drop per unit
350 length at the optimal conditions as a function of the coil pitch. It is obvious
that the pressure drop at the first optimal Reynolds number is much lower than
that of the second optimal Reynolds number (compare scales of left and right
bottom figures); therefore, the first (lowest) optimal Reynolds number Reopt,1
is clearly preferable, since it leads simultaneously to slightly better mixing and
355 much lower pressure drop.
Significant changes concerning pressure drop are not observed when the pitch
is increased at a constant Reynolds number. Indeed, the coil pitch was found
to have negligible effect on the pressure drop in several past studies [9, 20,
59, 64, 65]. Nevertheless, since the peak points of the mixing coefficient are
360 shifted to lower values of Re for increasing pitch, the corresponding pressure
drop decreases slightly.

24
1.00 1.00

Mixing coefficient, Mc
Mixing coefficient, Mc
0.96 0.96

0.92 0.92

0.88 0.88
Re=500
Re=25 Re=800
0.84 Re=40 0.84 Re=1000
1st optimal Re 2nd optimal Re
0.80 0.80
0 20 40 60 80 0 20 40 60 80
Pitch, P (mm) Pitch, P (mm)
(i) 1st optimal Re (ii) 2nd optimal Re
   
(a) Peak mixing coefficient

2.00 80
Pressure drop, ΔP/L (Pa/m)

Pressure drop, ΔP/L (Pa/m)


1.75 70
1.50 60
1.25 50
1.00 40
0.75 30
0.50 20 Re=500
Re=25 Re=800
0.25 Re=40 10 Re=1000
1st optimal Re 2nd optimal Re
0.00 0
0 20 40 60 80 0 20 40 60 80
Pitch, P (mm) Pitch, P (mm)
(i) 1st optimal Re  (ii) 2nd optimal Re 
   
(b) Corresponding pressure drop

Figure 11: Effect of coil pitch on peak mixing coefficient and on the associated pressure drop.
Note the different vertical scales concerning pressure drop.

Figure 12a shows the effect of varying the pipe diameter concerning peak
mixing efficiency, corresponding to the optimal Reynolds number (Reopt,1 in
left figures, Reopt,2 in right figures). As discussed previously together with
365 Figure 10b, when the pipe diameter increases the mixing efficiency is gener-
ally reduced. However, the changes at the optimal conditions are quite small,
since all corresponding values of Mc are higher than 0.97. At the same time,
as expected and already reported in several previous studies [50, 57, 59], an
increase in pipe diameter leads to significantly lower pressure drops, as shown
370 in Figure 12b. Consequently, at both optimal conditions (Reopt,1 and Reopt,2 ),

25
1.00 1.00
Mixing coefficient, Mc

Mixing coefficient, Mc
0.97 0.97

0.94 0.94

0.91 0.91 Re=500


Re=35
Re=1000
0.88 Re=40 0.88 Re=1500
1st optimal Re 2nd optimal Re
0.85 0.85
0 5 10 15 20 0 5 10 15 20
Pipe diameter, d (mm) Pipe diameter, d (mm)
(i) 1st optimal Re  (ii) 2nd optimal Re 

    (a) Peak mixing coefficient

14 1000
Re=35 Re=500
12
Pressure drop, ΔP/L
Pressure drop, ΔP/L

800 Re=1000
10 Re=40
Re=1500
8 1st optimal Re 600 2nd optimal Re
6 400
4
200
2

0 0
0 5 10 15 20 0 5 10 15 20
Pipe diameter, d (mm) Pipe diameter, d (mm)
(i) 1st optimal Re  (ii) 2nd optimal Re 
   
(b) Corresponding pressure drop

Figure 12: Effect of pipe diameter on peak mixing coefficient and on the associated pressure
drop. Note the different vertical scales concerning pressure drop.

a larger pipe diameter can be employed to strongly reduce the total pressure
drop, with only minor reduction in the mixing efficiency at the outlet of the
helix reactor. Once again, the first (lowest) optimal Reynolds number Reopt,1
is always preferable, since it leads simultaneously to slightly better mixing and
375 much lower pressure drop.
Finally, the influence of coil diameter on peak mixing efficiency and on the
corresponding pressure drop are shown in Figure 13. Changing the coil diameter
widely between D = 70 mm and D = 150 mm, no significant variations can be
seen in the peak mixing efficiency within the two optimal ranges of Re, as

26
1.00 1.00
Mixing coefficient, Mc

Mixing coefficient, Mc
0.98 0.98

0.96 0.96

Re=25 Re=500
0.94 Re=40 0.94 Re=800
Re=50 Re=1000
1st optimal Re 2nd optimal Re
0.92 0.92
50 70 90 110 130 150 170 50 70 90 110 130 150 170

Coil diameter, D (mm) Coil diameter, D (mm)


(i) 1st optimal Re  (ii) 2nd optimal Re 

(a) Peak mixing coefficient

4 120
Re=25 Re=500
Re=40 Re=800
Pressure drop, ΔP/L

Pressure drop, ΔP/L


3 90 Re=1000
Re=50 2nd optimal Re
1st optimal Re
2 60

1 30

0 0
50 70 90 110 130 150 170 50 70 90 110 130 150 170
Coil diameter, D (mm) Coil diameter, D (mm)
(i) 1st optimal Re  (ii) 2nd optimal Re 

(b) Corresponding pressure drop

Figure 13: Effect of coil diameter on peak mixing coefficient and on the associated pressure
drop. Note the different vertical scales concerning pressure drop.

380 shown in Figure 13a. The first and second peaks of Mc are almost constant and
identical, at Mc ≈ 0.99 and Mc ≈ 0.98, respectively. Figure 13a shows a slight
reduction in pressure drop with an increase in coil diameter for constant Re
values, as already discussed in previous studies [50, 57, 59]. However, since the
optimal Re values are lower for smaller coil diameters, the minimum pressure
385 drop is observed in both cases for the lowest coil diameter. Hence, if the optimal
Re can be set in an accurate manner, the lowest coil diameter is recommended.
Here also, Reopt,1 is always recommended, leading simultaneously to slightly
better mixing and much lower pressure drop.

27
6.5. Overview concerning optimal ranges of Reynolds number

390 In this last section, the ranges corresponding to the first and the second
optimal Reynolds numbers are discussed. Starting from the values discussed
previously, further CFD computations are carried out in an iterative man-
ner for slightly lower and slightly larger values of Re, in order to identify
the region within which the loss in mixing efficiency is below 1.0%. In other
395 words, the mixing efficiency within each of this region fulfills the condition
1.0 ≥ Mc /Mc,max ≥ 0.99.
Figure 14 shows the obtained first and second optimal ranges of Reynolds
number, for all the geometries considered in the present study. Looking at
Figure 14a, it can be observed that excellent mixing occurs always for a value
400 of Reopt,1 approximately between 25 and 60. For the second optimal range of
Reynolds number Reopt,2 , most peak values are found approximately between
500 and 1,000. Accordingly, to avoid poor mixing, values of the Reynolds num-
ber lower than 25 or beyond 1,000 are not recommended.
For all the studied geometries, two mean values of Reynolds number could be
405 calculated from all points shown in Figure 14, corresponding to the first and the
second optimal Reynolds numbers, which are Re≈40 and Re≈750. Additionally,
based on all studied geometrical dimensions, these two mean values of optimal
Re correspond to first and second optimal ranges of Dean number of 7 ≤ De
≤ 15 and 150 ≤ De ≤ 300, respectively. The mean values of optimal Re can be
410 used as a guideline to ensure excellent mixing for practical cases and for widely
different geometries. In general, the first optimal Reynolds number Reopt,1
should always be preferred, since it has significantly lower pressure drop and
slightly better mixing efficiency.

28
8
G1 (P=16mm)
7
G2 (P=30mm)
6
Geometry number

5 G3 (P=60mm)

4 G4 (d=5mm)

3 G5 (d=15mm)

2
G6 (D=70mm)
1
G7 (D=150mm)
0
0 30 60 90
Reynolds number, Re

(a) Reopt,1

8
G1 (P=16mm)
7
G2 (P=30mm)
6
Geometry number

5 G3 (P=60mm)

4 G4 (d=5mm)

3 G5 (d=15mm)

2
G6 (D=70mm)
1
G7 (D=150mm)
0
0 500 1000 1500 2000
Reynolds number, Re
 
(b) Reopt,2

Figure 14: Range of optimal Reynolds numbers Reopt,1 and Reopt,2

29
7. Conclusions

415 A systematic Computational Fluid Dynamics study was carried out to in-
vestigate optimal mixing in helical pipes, varying geometrical parameters and
flow conditions. Seven different helical geometries were considered, in which the
coil pitch, the pipe diameter, and the coil diameter were changed within the
ranges of 16–60, 5–15, 70–150 mm, respectively. The specific influence of each
420 geometrical parameter on mixing profile, peak mixing efficiency, and associated
pressure drop were studied, changing the Reynolds number broadly in a range
between Re=5 and Re=4,000. Firstly, the axial evolution of mixing along the
coil length was discussed. The results show that, at low Reynolds number, a
high pitch is required to improve mixing in the absence of any noticeable sec-
425 ondary flow. However, low pitch values provide stronger secondary flows and
better mixing efficiency at high Reynolds numbers. Regarding the pipe and
coil diameters, lower values deliver generally faster mixing. The coil pitch and
the coil diameter were found to have insignificant effects on the pressure drop,
compared to pipe diameter. A considerable reduction in pressure drop can be
430 achieved by increasing the pipe diameter, particularly so at high Reynolds num-
bers. It was also shown that, the mixing efficiency is generally poor for Re≤25
or Re≥1,000, due to either very weak secondary flows, or very low residence
times, respectively. Finally, two optimal ranges of Reynolds number could be
obtained, in which the mixing efficiency peaks. The first one corresponds to
435 Reynolds numbers between 25 and 60; the second one is found for Reynolds
numbers between 500 and 1,000. Based on these results, averaged values for the
optimal Reynolds numbers regarding mixing efficiency have been computed,
Reopt,1 ≈ 40 and Reopt,2 ≈ 750 (corresponding to optimal ranges of Dean num-
ber of 7 ≤ De ≤ 15 and 150 ≤ De ≤ 300, respectively), ensuring optimal mixing
440 efficiency for a wide range of geometrical parameters. However, the associated
pressure drop is much lower for the lower Reynolds number. Hence, it is finally
recommended to systematically operate helix reactors at Reopt ≈ 40 in order to
maximize mixing efficiency.

30
Acknowledgments

445 This work is part of the Collaborative Research Center “Integrated Chemical
Processes in Liquid Multiphase Systems” (Subproject B1) coordinated by the
Technische Universitt Berlin. Financial support by the German Research Foun-
dation (Deutsche Forschungsgemeinschaft, DFG) is gratefully acknowledged
(SFB/TR 63). The Ph.D. work of Mr. Mansour is partially supported by a
450 scholarship from the Egyptian government.

References

[1] M. M. Mandal, C. Serra, Y. Hoarau, K. D. P. Nigam, Numerical modeling


of polystyrene synthesis in coiled flow inverter, Microfluid. Nanofluid. 10 (2)
(2011) 415–423.

455 [2] D. Singh, K. D. P. Nigam, Laminar dispersion of polymer solutions in


helical coils, J. Appl. Polym. Sci. 26 (3) (1981) 785–790.

[3] T. J. Rennie, V. G. S. Raghavan, Experimental studies of a double-pipe


helical heat exchanger, Exp. Therm. Fluid Sci. 29 (8) (2005) 919–924.

[4] J. A. Koutsky, R. J. Adler, Minimisation of axial dispersion by use of


460 secondary flow in helical tubes, Can. J. Chem. Eng. 42 (6) (1964) 239–246.

[5] N.-T. Nguyen, Z. Wu, Micromixers: a review, J. Micromech. Microeng.


15 (2) (2004) R1.

[6] S. Agrawal, K. D. P. Nigam, Modelling of a coiled tubular chemical reactor,


Chem. Eng. J. 84 (3) (2001) 437–444.

465 [7] V. Kumar, S. Saini, M. Sharma, K. D. P. Nigam, Pressure drop and heat
transfer study in tube-in-tube helical heat exchanger, Chem. Eng. Sci.
61 (13) (2006) 4403–4416.

[8] V. Kumar, B. Faizee, M. Mridha, K. D. P. Nigam, Numerical studies of a


tube-in-tube helically coiled heat exchanger, Chem. Eng. Process. 47 (12)
470 (2008) 2287–2295.

31
[9] S. Vashisth, V. Kumar, K. D. P. Nigam, A Review on the Potential Appli-
cations of Curved Geometries in Process Industry, Ind. & Eng. Chem. Res.
47 (C) (2008) 3291–3337.

[10] L. Hohmann, R. Gorny, O. Klaas, J. Ahlert, K. Wohlgemuth, N. Kock-


475 mann, Design of a continuous tubular cooling crystallizer for process de-
velopment on lab-scale, Chem. Eng. Technol. 39 (7) (2016) 1268–1280.

[11] M. Mansour, Z. Liu, G. Janiga, K. D. P. Nigam, K. Sundmacher,


D. Thévenin, K. Zähringer, Numerical study of liquid-liquid mixing in he-
lical pipes, Chem. Eng. Sci. 172 (2017) 250 – 261.

480 [12] K. D. P. Nigam, A. K. Saxena, Residence time distribution in straight and


curved tubes, Ency. Fluid Mech. 1 (1986) 675–762.

[13] W. Dean, Note on the motion of fluid in a curved pipe, Lond. Edinb. Dubl.
Phil. Mag. 4 (20) (1927) 208–223.

[14] W. Dean, The stream-line motion of fluid in a curved pipe (Second paper),
485 Lond. Edinb. Dubl. Phil. Mag. 5 (30) (1928) 673–695.

[15] C. Castelain, A. Mokrani, P. Legentilhomme, H. Peerhossaini, Residence


time distribution in twisted pipe flows: helically coiled system and chaotic
system, Exp. Fluids 22 (5) (1997) 359–368.

[16] R. Brodkey, Turbulence in mixing operations: theory and application to


490 mixing and reaction, Elsevier, 2012.

[17] P. Plouffe, R. Anthony, A. Donaldson, D. M. Roberge, N. Kockmann,


A. Macchi, Transport phenomena in two-Phase liquid-liquid micro-reactors,
in: ASME 2016 10th Int. Conf. Nanochannels, Microchannels, Minichan-
nels, no. 44793, ASME, 2012, pp. 611–623.

495 [18] V. Kumar, M. Aggarwal, K. D. P. Nigam, Mixing in curved tubes, Chem.


Eng. Sci. 61 (17) (2006) 5742–5753.

32
[19] S. P. Vanka, G. Luo, C. M. Winkler, Numerical study of scalar mixing in
curved channels at low Reynolds numbers, AIChE J. 50 (10) (2004) 2359–
2368.

500 [20] P. Mishra, S. N. Gupta, Momentum Transfer in Curved Pipes. 1. Newtonian


Fluids, Ind. Eng. Chem. Des. Dev. 18 (1) (1979) 130–137.

[21] E. F. Schmidt, Wärmeübergang und druckverlust in rohrschlangen, Chemie


Ing. Tech. 39 (13) (1967) 781–789.

[22] P. Srinivasan, Pressure drop and heat transfer in coils, Chem. Eng. 218
505 (1968) CE113–CE119.

[23] K. Sreenivasan, P. Strykowski, Stabilization effects in flow through helically


coiled pipes, Exp. Fluids 1 (1) (1983) 31–36.

[24] R. Miyake, T. S. J. Lammerink, M. Elwenspoek, J. H. J. Fluitman, Micro


mixer with fast diffusion, in: Proc. IEEE Micro Electro Mech. Syst., 1993,
510 pp. 248–253.

[25] J. Branebjerg, P. Gravesen, J. P. Krog, C. R. Nielsen, Fast mixing by


lamination, in: Proc. IEEE Micro Electro Mech. Syst., IEEE, 1996, pp.
441–446.

[26] D. M. Hobbs, F. J. Muzzio, The Kenics static mixer: a three-dimensional


515 chaotic flow, Chem. Eng. J. 67 (3) (1997) 153–166.

[27] 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. Microelectromech. Syst. 9 (2) (2000) 190–197.

[28] R. Thakur, C. Vial, K. D. P. Nigam, E. Nauman, G. Djelveh, Static mixers


520 in the process industriesa review, Chem. Eng. Res. Des. 81 (7) (2003) 787–
826.

33
[29] V. Kumar, V. Shirke, K. D. P. Nigam, Performance of kenics static mixer
over a wide range of reynolds number, Chem. Eng. J. 139 (2) (2008) 284–
295.

525 [30] E. A. Mansur, M. Ye, Y. Wang, Y. Dai, A State-of-the-Art Review of


Mixing in Microfluidic Mixers, Chinese J. Chem. Eng. 16 (4) (2008) 503–
516.

[31] F. G. Bessoth, A. J. deMello, A. Manz, Microstructure for efficient contin-


uous flow mixing, Anal. Commun. 36 (1999) 213–215.

530 [32] A. K. Saxena, K. D. P. Nigam, Coiled configuration for flow inversion and
its effect on residence time distribution, AIChE J. 30 (3) (1984) 363–368.

[33] M. Mridha, K. D. P. Nigam, Coiled flow inverter as an inline mixer, Chem.


Eng. Sci. 63 (6) (2008) 1724–1732.

[34] V. Kumar, M. Mridha, A. K. Gupta, K. D. P. Nigam, Coiled flow inverter


535 as a heat exchanger, Chem. Eng. Sci. 62 (9) (2007) 2386–2396.

[35] M. M. Mandal, P. Aggarwal, K. D. P. Nigam, Liquid-liquid mixing in coiled


flow inverter, Ind. Eng. Chem. Res. 50 (23) (2011) 13230–13235.

[36] M. G. Gelhausen, S. K. Kurt, N. Kockmann, Mixing and heat transfer in


helical capillary flow reactors with alternating bends, in: ASME 2014 12th
540 Int. Conf. Nanochannels, Microchannels, Minichannels, ASME, 2014, pp.
V001T13A001–V001T13A001.

[37] S. K. Kurt, M. Akhtar, K. D. P. Nigam, N. Kockmann, Continuous reactive


precipitation in a coiled flow inverter: Inert particle tracking, modular
design, and production of uniform caco3 particles, Ind. Eng. Chem. Res.
545 56 (39) (2017) 11320–11335.

[38] S. K. Kurt, I. V. Gürsel, V. Hessel, K. D. P. Nigam, N. Kockmann, Liquid–


liquid extraction system with microstructured coiled flow inverter and other

34
capillary setups for single-stage extraction applications, Chem. Eng. J. 284
(2016) 764–777.

550 [39] L. Sharma, K. D. P. Nigam, S. Roy, Single phase mixing in coiled tubes and
coiled flow inverters in different flow regimes, Chem. Eng. Sci. 160 (2017)
227–235.

[40] A. K. Saxena, K. D. P. Nigam, Laminar dispersion in helically coiled tubes


of square cross-section, Can. J. Chem. Eng. 61 (1) (1983) 53–57.

555 [41] F. Jiang, K. S. Drese, S. Hardt, M. Küpper, F. Schönfeld, Helical flows and
chaotic mixing in curved micro channels, AIChE J. 50 (9) (2004) 2297–2305.

[42] Y. Lasbet, B. Auvity, C. Castelain, H. Peerhossaini, Thermal and hydro-


dynamic performances of chaotic mini-channel: application to the fuel cell
cooling, Heat Transf. Eng. 28 (8-9) (2007) 795–803.

560 [43] A. Tohidi, H. Ghaffari, H. Nasibi, A. S. Mujumdar, Heat transfer enhance-


ment by combination of chaotic advection and nanofluids flow in helically
coiled tube, Appl. Therm. Eng. 86 (2015) 91–105.

[44] A. Alam, K.-Y. Kim, Analysis of mixing in a curved microchannel with


rectangular grooves, Chem. Eng. J. 181 (2012) 708–716.

565 [45] A. Yamagishi, T. Inaba, Y. Yamaguchi, Chaotic analysis of mixing en-


hancement in steady laminar flows through multiple pipe bends, Int. J.
Heat Mass Transf. 50 (7) (2007) 1238–1247.

[46] H. Ju, Z. Huang, Y. Xu, B. Duan, Y. Yu, Hydraulic Performance of Small


Bending Radius Helical Coil-Pipe, J. Nucl. Sci. Technol. 38 (10) (2001)
570 826–831.

[47] H. Ito, Friction factors for turbulent flow in curved pipes, J. Basic Eng
81 (2) (1959) 123–134.

[48] T. M. Liou, Flow visualization and LDV measurement of fully developed


laminar flow in helically coiled tubes, Exp. Fluids 13 (5) (1992) 332–338.

35
575 [49] L. Tang, Y. Tang, S. Parameswaran, A numerical study of flow character-
istics in a helical pipe, Adv. Mech. Eng. 8 (7) (2016) 1687814016660242.

[50] L. Tang, S. Yuan, M. Malin, S. Parameswaran, Secondary vortex-based


analysis of flow characteristics and pressure drop in helically coiled pipe,
Adv. Mech. Eng. 9 (4) (2017) 1687814017700059.

580 [51] T. J. Hüttl, R. Friedrich, Direct numerical simulation of turbulent flows in


curved and helically coiled pipes, Comput. Fluids 30 (5) (2001) 591–605.

[52] K. Yamamoto, S. Yanase, T. Yoshida, Torsion effect on the flow in a helical


pipe, Fluid Dyn. Res. 14 (5) (1994) 259–273.

[53] T. J. Hüttl, R. Friedrich, Influence of curvature and torsion on turbulent


585 flow in helically coiled pipes, Int. J. Heat Fluid Flow 21 (3) (2000) 345–353.

[54] A. Saxena, K. D. P. Nigam, On rtd for laminar flow in helical coils, Chem.
Eng. Sci. 34 (3) (1979) 425–426.

[55] M. Jokiel, L.-M. Wagner, M. Mansour, N. M. Kaiser, K. Zähringer,


G. Janiga, K. D. Nigam, D. Thévenin, K. Sundmacher, Measurement and
590 simulation of mass transfer and backmixing behavior in a gas-liquid heli-
cally coiled tubular reactor, Chem. Eng. Sci. 170 (2017) 410–421.

[56] S. Litster, J. Pharoah, N. Djilali, Convective mass transfer in helical pipes:


effect of curvature and torsion, Heat Mass Transfer. 42 (5) (2006) 387–397.

[57] A. Cioncolini, L. Santini, An experimental investigation regarding the lami-


595 nar to turbulent flow transition in helically coiled pipes, Exp. Therm. Fluid
Sci. 30 (4) (2006) 367–380.

[58] M. Mansour, G. Janiga, K. D. P. Nigam, D. Thévenin, K. Zähringer, Nu-


merical study of heat transfer and thermal homogenization in a helical
reactor, Chem. Eng. Sci. 177 (2018) 369 – 379.

36
600 [59] H. A. Mohammed, K. Narrein, Thermal and hydraulic characteristics of
nanofluid flow in a helically coiled tube heat exchanger, Int. Commun.
Heat Mass Transf. 39 (9) (2012) 1375–1383.

[60] J. M. Ottino, The kinematics of mixing: stretching, chaos, and transport,


Vol. 3, Cambridge University Press, 1989.

605 [61] S. Wiggins, J. M. Ottino, Foundations of chaotic mixing, Philos. Trans. R.


Soc. London A Math. Phys. Eng. Sci. 362 (1818) (2004) 937–970.

[62] Y. Mori, W. Nakayama, Study on forced convective heat transfer in curved


pipes. i- laminar region, Int. J. Heat Mass Transf. 8 (1965) 67–82.

[63] C. Wang, On the low-reynolds-number flow in a helical pipe, J. Fluid Mech.


610 108 (1981) 185–194.

[64] A. M. Fsadni, J. P. M. Whitty, A review on the two-phase pressure drop


characteristics in helically coiled tubes, Appl. Therm. Eng. 103 (2016) 616–
638.

[65] M. R. Salem, K. M. Elshazly, R. Y. Sakr, R. K. Ali, Effect of Coil Torsion


615 on Heat Transfer and Pressure Drop Characteristics of Shell and Coil Heat
Exchanger, J. Therm. Sci. Eng. Appl. 8 (1) (2015) 11015–11017.

37
Graphical Abstract

Secondary flow Re ≈ 40 Excellent mixing


streamlines
Mass fraction
of Liquid 1

Vorticity contours

Liquid 2
Liquid 1
*Highlights

Highlights

 Optimal liquid-liquid mixing conditions are investigated in helical pipes

 Considering a wide range of relevant geometrical dimensions and Reynolds numbers Re

 Two general values of Re are found, leading to excellent mixing (Re ≃ 40, Re ≃ 750

 From those two, Re ≃ 40 leads to a lower pressure drop, and is thus recommended

You might also like