Accepted Manuscript

Computational Fluid Dynamics (CFD) Technique to Study the Effects of Helical

Wire Inserts on Heat Transfer and Pressure Drop in a Double Pipe Heat Ex-
changer

Khashayar Sharifi, Morteza Sabeti, Mehdi Rafiei, Amir H Mohammadi, Laleh

Shirazi

PII: S1359-4311(17)32304-9
DOI: http://dx.doi.org/10.1016/j.applthermaleng.2017.08.146
Reference: ATE 11029

To appear in: Applied Thermal Engineering

Received Date: 6 April 2017

Revised Date: 29 July 2017
Accepted Date: 30 August 2017

Please cite this article as: K. Sharifi, M. Sabeti, M. Rafiei, A.H. Mohammadi, L. Shirazi, Computational Fluid
Dynamics (CFD) Technique to Study the Effects of Helical Wire Inserts on Heat Transfer and Pressure Drop in a
Double Pipe Heat Exchanger, Applied Thermal Engineering (2017), doi: http://dx.doi.org/10.1016/
j.applthermaleng.2017.08.146

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.
Computational Fluid Dynamics (CFD) Technique to Study the
Effects of Helical Wire Inserts on Heat Transfer and Pressure
Drop in a Double Pipe Heat Exchanger

Khashayar Sharifi a, Morteza Sabeti b, Mehdi Rafiei b, Amir H Mohammadi c, d*

, Laleh
Shirazi a

a
Petroleum Refining Division, Research Institute of Petroleum Industry (RIPI), Tehran, Iran
b
Department of Chemical Engineering, Isfahan University of Technology, Isfahan, Iran
c
Institut de Recherche en Génie Chimique et Pétrolier (IRGCP), Paris Cedex, France
d
Discipline of Chemical Engineering, School of Engineering, University of KwaZulu-Natal, Howard College
Campus, King George V Avenue, Durban 4041, South Africa

Abstract – In this communication, the computational fluid dynamics (CFD) technique to

study the influence of coiled wire inserts on the Nusselt number, friction coefficient and
overall efficiency in double pipe heat-exchangers has been employed. For this purpose, some
wire coil inserts fitted inside heat-exchangers were meshed and simulated at various
Reynolds numbers by using the CFD softwares of Gambit and Fluent. To fulfill reliable and
validated results, a big effort was made to generate structural hexahedral meshes over all of
the heat-exchanger geometries. The validated models, then, made clear the object and the
conditions which could exactly facilitate heat transfer throughout double pipe heat-
exchangers. The outcome of this work indicates that taking the advantage of proper wire coils
could improve the Nusselt values to 1.77 times. Following the numerical simulation, proper
friction coefficient and Nusselt number correlations for various coiled wire inserts with
different geometry arrangements under the laminar flow were proposed. Unlike the preceding
relations, the correlations of this work are based on the occupied spaces where helical wires
take up inside tubes; therefore, the two modified correlations can both be used for non-
uniform helical wire insert geometries.

Keywords - Computational fluid dynamics (CFD); Coiled wire insert; Friction factor; Heat
transfer; Double pipe heat exchanger.

*
Corresponding Author: A.H. Mohammadi, E-mail: a.h.m@irgcp.fr & amir_h_mohammadi@yahoo.com

1
1 Introduction

Thermal equipment designers have always attempted to enhance heat transfer to

reduce thermal equipment sizes as well as costs. In industrial processes, heat exchangers have

widely been exploited so as for energy losing to be avoided [1]. To develop a proper heat

exchanger, many investigators are making a big attempt to introduce a type of exchanger to

be not only compact but also to have a high overall heat transfer coefficient [2]. For a long

time, different methods have been used to increase convective heat transfer coefficients [3].

These methods are generally divided into two groups: passive methods and active methods.

Passive methods are much more efficient than active ones due to their simpler performance

and ease of utilization in modern thermal equipment [4]. Active methods, such as generating

a vibrational flow in mobile fluids, require some external power input to enhance heat

transfer. In contrast, passive methods such as coiled wire inserts [5] and micro fin tubes [6]

rely on no external power input. These methods are used to enhance the heat transfer rate by

increasing either the flow path or generating an angular acceleration to the fluid flow. The

importance of swirl flows in industrial applications was discussed by Saqr et al. numerically

[7]. At the end of their numerical study, they derived a relation to bold the effect of swirl

flows on the entropy augmentation in heat-exchangers. Among the passive methods, coiled

wire inserts and twisted tape inserts are more common. The twisted tape inserts inside heat-

exchangers result in vortex flow along the heat-exchangers, without having a proper thermal

contact with the heat-exchanger walls. In contrast, the coiled wire inserts reduce the heat-

exchanger hydraulic diameters and can be used as fin extended surfaces to enhance the heat

transfer rates. They result in less pressure drop than the twisted tape inserts although the

coiled wire inserts increase the heat-exchanger friction coefficients [8]. When selecting either

2
of the two methods, two factors shall be considered: efficiency and cost. Garcia et al. [9]

stated that of all devices used to increase heat transfer rate, the wire coils exhibit striking

results since they can be added to equipment easily, and also they culminate in lower pressure

drop while their costs are relatively low too.

In the systems containing oil fluids, an increase in the coefficient is more considered

due to their low convective heat transfer coefficient. Chiou [10] investigated the effect of

coiled wire inserts in chilling horizontal heat-exchangers filled with an oil fluid. He observed

an increase in the convective heat transfer coefficient due to the cutting of the thin laminar

sub-layer in the fluid film and turbulency augmentation. Later, Garcia et al. [9] examined

heat transfer enhancement through coiled wire inserts in a heat-exchanger under laminar,

transition and turbulent regimes for the Reynolds numbers ranging from 80 to 90000 and the

Prandtl numbers ranging from 2.5 to 150. They used a water system as well as a water-

propylene mixture, and finally reported an increase in heat transfer by using the coiled wire

inserts. In 1989, Uttarwar et al. [11] investigated seven different types of coiled wire inserts

inside horizontal tubes. They used oil as a fluid with the Reynolds numbers ranging from 30

to 700 and the Prandtl numbers ranging from 300 to 675, and then they reported the effect of

the type and geometric characteristics of the wire inserts on heat transfer rates. Zimparov and

Penchev compared thermal and hydraulic performance of coiled wire inserts and twisted tape

inserts inside a heat-exchanger in turbulent flow regions [3]. The results showed that the

coiled wire inserts were more efficient and useful than the twisted tape inserts. Eiamsa-Ard et

al. [12] examined the effect of combining heterogeneous coiled wire inserts (with varying

pitch lengthes) and twisted tape inserts on the thermal performance, dimentionless Nusselt

number and friction coefficient, and then they compared the results with a plain heat-

exchanger. The results showed an improvement in terms of the thermal properties, Nusselt

number, and friction coefficient. A comprehensive review of the passive methods employed

3
to increase heat transfer rate within horizontal heat-exchangers has been presented by Dewan

et al. [4]. This study included the examination and comparison of the effect of twisted tape

inserts, coiled wire inserts, fin extended surface etc. inside the heat-exchangers under laminar

and turbulent flows. Following these works, Vahidfar and Kahrom [13] tried to scrutinize the

role of inserted wire coils and rings in a double pipe heat exchanger performance. They

placed the objects inside the inner tube through which air with Pr=0.7 was flowing and

becoming warm. So as to have findings with high accuracy, they repeated their tests with

three different d/D ratios and three various pitch ratios while the velocity of air was changed

from 1 to 6 m/s. In their experiments, it was the wire coil that rotated the flow and created a

centrifugal force whereby the flowing fluids with a high density (low temperature) gave way

to the fluids with a low density (high temperature), and then increased heat transfer rate of the

heat exchanger. Eventually, it was asserted that the overall enhancement efficiency of the

inserted wire coil with d/D = 0.11 and P/D = 1 at a high Reynolds number can be mounted

about 128%. To reveal the role of coiled wire inserts on fluid flow performance inside a

round pipe, Muñoz-Esparza and Sanmiguel-Rojas [14] employed OpenFOAM package which

is based on the finite volume method beside some of their experimental tests. Their

experiment was conducted on four wire coil inserts with four dimensionless pitch values of

1.5, 2.5, 3.37 and 4.50 inside a pipe. The authors' results show that there are four different

regions in the Re-f diagrams when using helical wire inserts. In region I, when Re < 500, the

flow regime is steady and laminar. In region II, when 500< Re < 600, the flow regime is not

steady, but it is still laminar. When the flow is in region III, 600< Re < 850, a high degree of

instability is defined for the flow. In the end, in region IV, 850< Re, there is a turbulency in

the flow, and it is impossible to simulate the flow by using the laminar model. A constant

friction coefficient value appeared within the Reynolds number range of 600 to 850, region

III, in their experimental results at each of the pitch values. They declared that the friction

4
factor intensity relates strongly on pitch size since for the flow with higher value of

dimensionless pitch, the growth of the relative friction coefficient is somewhat smaller than

that for the flow with lower pitch size. Solano et al. [15] presented a work in which fluid

dynamics and heat transfer in small oscillatory baffled reactors with helical coil inserts were

tested. Time-dependent flow on the basis of the periodic and laminar flow was used in the

simulation. They employed SIMPLE and first-order upwind algorithm to discretize the

governing equations. It was indicated that when the helical tube is used, the heat transfer

factor could increase by a factor of 4 at Re>300 compared to a plain tube. In a similar way,

McDonough et al. [16] presented a study in which they compared a helical baffled tube with

a simple tube in oscillatory reactors at laminar flow. Feng et al. [17] developed a numerical

investigation to study the heat transfer performance in a rectangular microchannel heat sink

through helical wire inserts. The used fluid was distilled water, and the range of defined

Reynolds number was between 100 and 1300. They employed some simplifying assumptions,

such as incompressible fluid, laminar and steady-state flow and constant thermophysical

properties, to validate their simulation. Finally, using the Nusselt and Friction results arisen

from the simulation they tried to determine constant coefficients to update the previous

Nusselt's and friction factor's correlations. In another case, the fluid flowing through a double

spiral spring insert was investigated by Zhang et al. [18] for the first time. The heat transfer

and friction loss inside the heat-exchangers in the Reynolds range of 6000-31000 were

analyzed with Fluent software. While assuming the steady state condition for the fluid flow,

they applied a turbulent model to the simulator to interpret the fluid dynamics. The enhanced

heat-exchangers had the length of 800 mm and the diameter of 21 mm. Accordingly, using

the tetrahedral geometry, they found 1,980,000 cells were the optimum number of elements

which accurately simulated the heat-exchangers. Their results show that the heat transfer

5
could enhance 1.78 times whilst the friction factor increased 8.79 times as compared to the

plain tube.

In general, most of the research conducted in this field has been experimental and

empirical in nature, but fewer works have employed numerical analysis and simulation [14].

The present study seeks to analyze heat transfer and fluid flow in a horizontal heat-exchanger

filled with coiled wire inserts. It should be mentioned that initially the results of simulation

have been compared against those of Akhavan Behabadi et al. [19] in 2010, to evaluate the

accuracy of the model. The data obtained from the plain heat-exchanger have been compared

against the heat-exchangers filled with the coiled wire inserts. Afterwards, the effect of the

type wire inserts as well as the Reynolds number on the Nusselt number and friction

coefficient through Computational Fluids Dynamics (CDF) have been examined. In the end,

it was tried to develop more practical correlations for the friction coefficient and the Nusselt

number when the used heat-exchanger is filled with helical wires.

2 Experimental Work

Akhavan Behabadi, et al. [19] reported experimental data on the effect of coiled wire

inserts on heat transfer and pressure drop when the engine oil was heated inside a thermal

exchanger under the laminar flow. Their utilized system consisted of a double-pipe

counterflow heat exchanger filled with the fluid having the properties shown in Table 1. The

engine oil was being heated by saturated steam flowing at the space between the tubes. The

inside diameter of the tube measured 26.04 mm and the outside diameter of the tube

measured 28.57 mm. The heat-exchanger length was 1700 mm, the inside tube was made of

Copper, and the heat exchanger shell was made of Steel Carbone. The inside diameter of the

6
shell measured 152.4 mm. The inlet and outlet temperatures were measured through installed

thermocouples on the heat-exchanger. The oil temperature across the heat-exchanger was

measured by some thermocouples installed on five holes in the heat-exchanger. The Nusselt

numbers were obtained through measuring temperature difference under varying conditions.

In the present study, several types of coiled wire inserts with different pitch sizes and

two varying diameters were used for the purpose of simulation. Table 2demonstrates the

properties of these simulated coiled wire inserts. Figure 1shows the geometry concerning the

wire inserts employed in the simulation process.

3 System Modeling through Computational Fluid Dynamics

Computational Fluid Dynamics (CFD) is a technique based on numerical methods

which aims at analyzing fluids movement, heat transfer, mass transfer, and chemical reaction.

The present study seeks to examine the fluid flow and heat transfer within a heat-exchanger

filled with the coiled wire inserts. Therefore, the governing equations including the mass

conservation (continuity), momentum conservation, and energy conservation equations are as

the following:

Mass conservation (continuity) equation:


 .     0 (1)
t

Momentum conservation equation:


    .     p   g  f (2)
t

Energy conservation equation:

7

t

 e   .  e  p    . k eff T   j h j J j   eff .   (3)

where t is the time, ρ is the mass density, stands for the flow velocity, p is the fluid

pressure, e is the inside energy per unit mass, shows the volumetric force, keff is the thermal

conductivity, hj is the enthalpy of species j, j represents the diffusion flux of species j and

finally eff exhibits the stress tensor.

One of the most important parts of simulation process is selecting a suitable meshing

process to calculate fundamental equations governing the heat-exchanger operation. The

softwares Gambit and Fluent were used to mesh the models and to simulate the system,

respectively. Selecting a proper meshing can contribute to a suitable convergence in solving

the equations while an improper meshing can lead to instability and divergence of

calculations. Figure 2 reveals some parts of the geometry used to simulate the system. The

grid design implemented by the authors in this work is an improvement of those structures

proposed by previous researchers [17, 20, 21]. Herein, a structured hexahedral mesh

algorithm was applied throughout the whole coiled inserts, for this type of mesh produces

more reliable results due to its logical aspect ratios and skewness. Furthermore, as this type of

mesh is less diffusive than the other types, doubtlessly it protects the results of simulation

from any inaccuracy, especially at the turbulent regime whose Reynolds number is higher

than 1000. In order for the structural mesh with a high quality to be created, each region of

the coiled wires with the length of one pitch was divided into 50 separate sub-regions (or sub-

volumes). The regions to be considered are shown in Figure 2. Each sub-region (as a small

volume) needs to be separately meshed with the command of Map and Cooper in Gambit

software. The number of nodes in each edge on the surface of the five sub-regions shown in

the figure is also an important parameter for creating a high-quality mesh and finally

8
converging to a precise solution. Therefore, a delicate effort was made to introduce a proper

mapped mesh using the optimal mapped mesh on a circle with a rectangular inner square

principle defined by Štigler [22]. Using this principle, all the surface of regions 1, 2, 3, 4 and

5 were meshed by rectangular cells in an optimal manner. As seen, the size of cells near the

vicinity of the tube walls and wires where there are large velocity and temperature gradients

were minimized as much as possible while the rest of the domain had coarser cells.

There were a number of assumptions, some of which are presented in the following

lines, which simplified the simulation in the present study.

- The fluid utilized in this work was an oil engine (a Newtonian fluid), and its

properties were estimated depending upon the temperature of the media when it

flowed along the tube inside the heat-exchanger. The properties of the oil utilized in

both the experiments and simulation process are shown in Table 1. The fluid

thermophysical properties varied piecewise-linear with fluid temperature.

- At steady condition, the fluid flow was considered to be laminar at Reynolds

numbers of 100-1200 [19].

- No slip condition was considered for the system wall.

-The fully developed laminar profile in a pipe at a constant rate and temperature of

312 K, associated with a uniform static pressure, was applied to the tube’s inlet.

- In addition to considering no thickness for tube walls, a constant temperature 373 K

was simply set throughout the heat-exchangers.

- Except gravity force, other body forces and viscous dissipation were neglected.

- To calculate LMTD, only the surface average temperature of the fluid at both sides

of heat-exchangers was used.

9
 - The method of volume averaging was used for determining the properties of the
fluid in order for Nusselt and Friction Factors to be calculated.

Since the fluid entered the heat-exchanger has passed a long distance before its

entrance, it was necessary to impose the fully developed velocity profile condition at the inlet

of the heat-exchanger. It is needless to say that to adapt the inlet boundary condition

according to its experimental circumstances, the Fluent users have to introduce a user-

defined-function (udf) program and import it into the Fluent program.

To obtain the temperature and velocity field inside the heat exchangers, Semi-

Implicit Pressure Linked Equations (SIMPLE) scheme which is a branch of the pressure-

velocity coupling method was used for the simulations. The second order upwind

discretization scheme for momentum and energy was employed in order to determine the

variables inside each cell. As the Green-Gauss Node-Based is the most accurate method with

respect to others, it was chosen for the present work to find the variables' gradient.

Simulations were carried out until the value of residual for continuity and temperature

variables reached and , respectively. The computer used for this study

was a core i7 processor with 40 GHz CPU and 16 GB RAM-memory. Regarding the residual

criteria, the average time required for each run was about 12 hrs. The results and discussion

taken from several runs are presented in the proceedings.

4 Results and Discussion

This part shows the effect of the coiled wire inserts as well as the Reynolds number on the

Nusselt number and friction factor. Before that, to study the grid independence in the heat-

10
exchanger, several mesh structures with different mesh densities including 400000, 1200000,

2000000 and 3200000 elements for the 3D heat-exchanger were generated. The results of the

simulations for the pipe with the wire coil of 2 mm diameter and 65 mm pitch length are

shown in Figure 3. In the non-uniform mesh, the region near to the wire was meshed finer

than the rest of the section in order to compute the values of variables with high gradients

near the walls. Therefore, as the figure shows, the non-uniform mesh has to have more

accurate results than the uniform one. As the graphs show, the Nusselt numbers for the very

fine mesh size (with 3200000 cells), being near to the experimental data, are more accurate

with respect to the others. Moreover, the grid convergence index (GCI) was used in order to

examine the spatial convergence of the CFD simulations presented in this work. This

methodology which is based on Richardson extrapolation comes from Roache's theory.

Richardson extrapolation for the problem is defined as:

  (4)
g exact  g 1    g1z g 2  
 
r 1 


g refers to the grid solution (g is Nu here), r is the ratio of grid spacing on the two grids, and

z is the order of the convergence rate. While the result of g1 is followed by the coarse grid, g2

comes from the finer grid. In fact in this study, g1, g2, g3 and g4 indicate the grid solution for

very coarse, coarse, fine and very fine meshes, respectively.

 i 1  g i 1  g i (5)

R i
 i 1

Where:
0  R  1 means monotonic convergence,
R  0 means oscillatory convergence,
R  1 means divergence.

11
The GCI for the fine grid solution, then, can be written as:

 i 1 (6)
GCI i 1  G s
g i  r z  1

As listed in Table 4, there is a reduction GCI value for the successive grid refinements

(GCI2<GCI1). The GCI for the finer grid (GCI2) is relatively low as compared to the coarser

grid GCI1. It means the very fine mesh size in Figure 3 reaches the spatial convergence of the

CFD simulations. Therefore, the meshing processes for next coiled wire inserts were

performed under mesh elements more than 3200000 cells by means of the optimal mapped

mesh principle mentioned in the preceding section. Figure 4 was employed to show the

distribution of temperature and pressure over the inner-pipe fluid passing from the pipe.

Herein, to find the calculated Nusselt number with Fluent, the average value of temperatures

took into consideration. Indeed, there exists a radial temperature distribution at each cross

section of the heat-exchanger along axial locations and in order for the analysis of the

numbers and coefficients to be conducted, the average value of the inlet and outlet sections

was taken into consideration. In the proceedings, the reason why the bulk values (whether

pressure or temperature) for the Nusselt number and friction coefficient are important is

clarified.

The expression (7) is used to calculate friction coefficient as follows:

pD
f  (7)
2 u m2 L

According to the above correlations, friction coefficient is a function of Reynolds

number as follows:

16
f  , 100  Re  2000 (8)
Re

12
 Figure 5 (a) illustrates the results of the simulation, the experiment and the related

theory. Table 3 shows the velocities used as inputs for each case. Depending upon the figures,

it is observable that as the Reynolds number increases, the friction coefficient in the whole

data range decreases. It should be pointed out that the calculated numbers with Fluent are

somehow lower than those for the experimental process. However, there is no large

discrepancy between results to impact negatively the current analyses and conclusions. As

mentioned, the friction coefficient with various Reynolds numbers taken from Eq. (8) is

shown in Figure 5 (a). Almost, the outcomes overlap one another at each Reynolds number

in the defined ranges.

What distinguishes the Reynolds-friction coefficient figure for a plain heat-exchanger

from that for the heat-exchanger filled with coiled wire inserts is that the diagram of Figure 5

(a) is always descending for the plain heat-exchanger. However, in heat-exchangers

containing coiled wire inserts, this downswing turns into a straight line after a special

Reynolds number. In other words, after the mentioned Reynolds number, the friction

coefficient reaches a constant value. While the theoretical relations cannot predict these

variations, CFD can predict them well. To illustrate why the presence of coiled wires leads to

an increase in the friction coefficients, it was tried to sketch the vortex flow of the fluid

passing through the tube with the wire diameter of 2 mm on a cross section along axial

locations. Figure 6 illustrates the fluid stream lines at the vicinity of the tube's walls in

different Reynolds numbers during the heat-transfer operation. Initially, as seen in Figure

6(a), there is a small vortex in stream lines. As the Reynolds number increases the produced

vortex in the fluid widens along the walls' vicinity. Therefore, it can be expected that this

turbulency makes a tangible change in the final value of friction coefficients especially at

very high Reynolds numbers. The appeared friction increment when using the wire inserts

13
with pitch lengths of 65 mm and 69 mm and wire diameter of 2 mm and 3.5 mm are

presented in Figure 5 (b) and (c), respectively.

To examine whether the application of the wire coils fitted inside the pipe could be

enhanced more or not, other four object inserts with different properties were used to analyze

more. The geometries of wire coils inside the pipe used in the present study are displayed in

Table 2. In the table, tubes 4, 5 and 6 show the coiled wire inserts consisting of two distinct

pitch lengths, one has the diameter of 2 and the other has the diameter of 3.5 mm.

A comparison among the various coiled wire inserts inside the inner pipe beside the

plain pipe is presented in Figure 7. It can be perceived that the presence of the wires

culminates in an increase in pressure drop and friction coefficients in all of the heat transfer

processes. Even, the difference between the plain's friction coefficients and the coiled wire

insert tubes' coefficients increases when the value of the Reynolds number increases.

Depending upon the general rule, the bigger wire volume (as a turbulence promoter) the tube

has the more pressure drop the process faces up to. This issue is somewhat valid in this study.

Of coiled wire inserts, sample 4 has highest friction coefficients while tube 1 with P = 65 mm

and d =2 mm has the lowest amount over a wide range of Reynolds numbers. One reason

behind this fact is that tube 4 has totally small pitch size with respect to others. Indeed, the

coiled wire with pitch of 26-65 mm has longest wire interwoven inside the tube and prevents

the fluid to move as easy as the flowing fluid in the other tubes move. As a result of that, we

expect more pressure drop for this case.

Now that the impact of wires on the friction coefficient was clarified, search for

finding a better correlation seems more convenient. Eq. (8) is the suggested correlation for a

plain tube at the Reynolds numbers lower than 1500. Herein, to estimate friction coefficients

for coiled wire inserts accurately Eq. (8) was manipulated. In this way, an extension in terms

14
of the Reynolds number and wire surface area was added in Eq. (8). Based on the numerical

simulation done on the engine oil under the defined conditions in Table 1, Table 2 and Table

3, the following correlation for the friction coefficient ranging from 100 to 1200 was deduced

using genetic algorithm method in Matlab toolbox. In developing the desirable relation, two

facts were considered. First, wires with short pitch sizes produce more pressure drops.

Secondly, as it is implied from friction coefficient graphs, the value of coefficients is raised

with increasing the Reynolds number. Taking the aforementioned facts into account, it can be

expected that the modified equation should be accompanied by these two variables. Thus,

regarding Eq. (8), the extended term made up of the two factors was generated as follows.

16
f   0.0188 Re 0.075NS  , 100  Re  1200
Re
(9)
Wire Surface Area Tube Surface Area
NS 
Tube Surface Area

where NS is a ratio coming from the surface area of coiled wires occupying the tubes' space.

Compared with former relations [14, 17], this correlation only needs the surface area of

helical wires inside tubes. Therefore, enabling us to estimate the friction only with the surface

term, in contrast to preceding correlations, Eq. (9) obviates the difficulty to calculate friction

coefficients when using non-uniform wire coils for the Reynolds numbers measuring from

100 to 1200. Figure 8 compares the friction values yielded from Eq. (9) in different Reynolds

numbers with those taken from the numerical study (CFD). One could see that except tube 4,

most of the estimated values lie within ±20% error bands, and the mean absolute percentage

error for the correlation is 9.7% which means it can predict friction coefficients reasonably

regarding wide ranges of geometry conditions Eq. (9) covers.

15
 The values of convective heat transfer coefficient for the different coils were

calculated through the temperature distribution yielded from simulation as well as expression

(10) as follows:

mC p T in T out 
h (10)
 DL T im

where ΔT is the logarithmic average of temperature difference between the hot and cold feeds

at each end of the double pipe heat-exchangers. Having heat transfer coefficient, the value of

Nusselt number can be found by:

hk
Nu  (11)
Dh

where k is the thermal conductivity of fluid and D is the diameter of the heat-exchanger's

tube.

According to the theory of convection heat transfer, for the laminar regime, the

Nusselt number in plain tubes can also be computed by the following expression as an

empirical relation:

1 0.14
1
 D 3   
Nu  1.5  Re.Pr  3
   
(12)
 L   w 

Using the existing data, the Nusselt numbers were calculated and also the Reynolds-

Nusselt figures concerning the plain heat-exchanger and coiled wire insert heat-exchangers

were drawn as Figure 9. It should be noted that to calculate Nusselt number using Eq. (12),

all fluid properties were evaluated at the mean bulk temperature of the flowing fluid defined

in Table 1. The Nusselt numbers yielded from the simulation process for the heat-exchangers

are lower than those yielded from the experimental data. It should be admitted that there exist

larger errors in lower values of the Reynolds numbers, which are mostly due to the

16
discretization and iterative convergence errors besides the considered assumptions for the

simulation process. However, the Nusselt numbers taken from Fluent Package are able to

follow the track of those numbers yielded from the experimental work. It is quite clear that if

the problem were solved by the direct numerical simulation (DNS) method, much more

accurate results would produce with Fluent. Of course, it is a really time-consuming process

to find the converged results by means of DNS when the geometry is complex.

According to the diagrams of Figure 9, as the Reynolds number increases, the

convective heat transfer coefficient and the Nusselt number increase. It should be reminded

that the wire coils generate a swirl flow by rotating the flow, and the fluid’s centrifugal force

automatically put pressure on the cold section of the fluid with higher density to move

outward, near to the hot solid boundary layers. Rotating around the center line, the fluid can

be easily mixed and the energy can be transferred faster. Consequently, one of reasons why

higher Reynolds numbers can increase the Nusselt number is the change occurred in the fluid

motion in helical wires, which leads to a rotating flow and eventually a remarkable increase

in the Nusselt number. The swirl motion to be considered is shown in Figure 10. Three

Reynolds numbers of 100, 400 and 1000 were chosen to compare their pathlines. Plus, three

cross sections with the axial locations of z = 0.5, z = 1 and z = 1 m in a coiled wire insert with

Re = 1000 were sketched in Figure 11. It does not need an intense focus to observe the

rotation of the pathlines along the tubes' lengths (Z direction). It is also noted that the

increment of Reynolds number raises the swirl motion. For instance, from Figure 10 it can be

perceived that one pathline of fluid for Re=1000 revolves three times around the center line

along its tube while that for Re=100 only has enough time to rotate twice around the center

line (Z direction). The swirl motion is more obvious when the Reynolds number is on the

threshold of turbulent flow regime (which is higher than Re = 2100). Aside from that, at the

farther distance along the tubes, the pathlines are all mixed with one another. Totally, as the

17
Reynolds number increases, the streamlines in a streamtube swirl around and spread through

the heat-exchangers faster, and the fluid mixing increases; consequently, the heat transfer rate

is augmented.

Figure 12 demonstrates distribution of Reynolds-Nusselt numbers for the plain heat-

exchanger and the other coiled wire heat-exchangers according to CFD results. It is observed

that the heat transfer coefficient and the Nusselt number increased in the heat-exchangers

filled with coiled wire inserts as compared to the plain one. Concerning the coiled wire

inserts measuring 69 mm in pitch length with 3.5 mm diameter, when Reynolds numbers

increase, the Nusselt numbers are enhanced by 1.77 times with respect to plain heat-

exchangers. As seen from the figure X, the Nusselt number of tube 3 grows much more than

the others. Probably, there exist two reasons behind this fact, one of which is the rotation of

the flow while the other is the vortices generated by the separation of the flowing fluid

around the wires. These mechanisms lead to more fluid mixing and then more heat

convection. Figure 13 represents the average value of tangential velocity at the end of the

heat-exchangers with different Reynolds number. In addition, Figure 14 shows the growth of

vorticities for tube 3 when Reynolds number increases from 100 to 1200. There is an

increment for the tangential velocity in tube 3 too. It does not need a big effort to perceive

that tube 3, tube 5 and tube 6 in Figure 13 have more slopes than the three other tubes.

Therefore, it expects that they have higher Nusselt numbers at the defined Reynolds range. It

is worth saying that this issue is more tangible for tube 3. According to Figure 13 and Figure

14, the amount of tangential velocity and vorticity for tube 3 are not so different with the

others when Re < 400; thus, similar to tube1 and tube 2, they produce a low Nusselt number

at low Reynolds numbers. However, Nusselt numbers in tube 3 grew at high Reynolds

numbers since tube 3 developed higher vorticity and swirl flow in the flowing fluid when

Reynolds number is higher than 400.

18
 From collected results it would be easy to understand that there is a direct relation

between the Reynolds number and Nusselt number, but it is somewhat difficult to detect a

significant relation between the kinds of coiled wires and their corresponding Nusselt

numbers at a constant Reynolds number. For instance, tube 6 with a longer and bigger wire

has lower Nusselt numbers than tube 3 with shorter and smaller wire for the Reynolds

numbers ranging between 600 and 1200. To estimate a rough relation in the range of the six

tube samples mentioned in Table 2, Eq. (12) which was introduced for a plain tube was

modified in this study. It took for granted that with increasing of the Reynolds number and

wire’s surface, the value of Nusselt number enhances with the help of an addition term to the

right hand side of Eq. (12),

1 0.14
1
 D 3   
Nu  1.5  Re.Pr   0.68 Re
0.48NSE 
Pr 
0.0002NS 
3
    100  Re  1200 (13)
 L   w 

This equation consists of the summation of two terms, the first of which is the simple

equation resulting from plain heat-exchangers, and the second of which coming from the

effect of coiled objects fitted inside the heat-exchangers. This equation states that the Nusselt

number could increase somewhat with inserting a wire insert with high surface area. Eq. (13)

was derived for the flowing fluid property limited in 200 < Pr < 350 and is applicable for

wire inserts in the range of 26 mm < P < 69 mm and 2 mm < d < 3.5 mm. Moreover, as seen

in Figure 15, the developed equation predicts the CFD Nusselt number in error bands of

±15%. The mean absolute percentage error for the predicted numbers against CFD ones is

5%. Thus, Eq. (13) seems to be a good way of predicting Nusselt numbers with regard to the

wide range of conditions it covers.

The overall enhancement efficiency is defined as the ratio of the heat transfer within

enhanced pipes against plain pipes to the friction coefficient in enhanced pipes against plain

19
pipes [23]. This ratio is an effective factor in the heat transfer field and reveals the

importance of heat transfer to pressure drop in heat exchangers. The following equation

represents the general correlation to calculate this ratio:

Nu e / Nu p
 (14)
f /f p
1/3
e

Figure 16 illustrates variations of this parameter relative to the Reynolds numbers. As

it is seen, this parameter is a value larger than one in most part of Reynolds ranges. This

implies that the ratio of heat transfer to pressure drop in the heat-exchangers containing the

helical wires is larger than the plain pipe. As shown in the figure, as Reynolds numbers

increase, this ratio for the heat-exchangers containing coiled wire inserts starts decreasing in

value. In other words, by increasing the Reynolds number the values of friction coefficients

for the mentioned heat-exchangers overcome the growth of the Nusselt numbers. Thus, it can

be expected that the heat-exchangers' overall efficiencies go down gradually along the

Reynolds number's axis. The figure also points out the fact that the best choice for enhancing

the heat transfer is the heat-exchanger with coiled wire of P = 69 mm whose overall

efficiencies are higher than others especially at the high Reynolds range. Since this heat-

exchanger not only could boost the Nusselt number (according to Figure 12), but also had

low pressure drops during the process.

5 Conclusion

A numerical study was introduced based on CFD application. The influence of coiled

wire inserts with various arrangements on friction coefficients and heat transfer rates was

investigated. The reasons why the object inserts make an impact on those values were

20
explained by taking the advantage of the several diagrams which clarify the augmentation

mechanism. It was shown that helical wires impede fluids to move easily on the surface of

tube walls. Higher friction coefficients are followed by this hindrance. In addition, a swirl

motion was observed during heat transfer processes by means of coiled wire inserts. The

higher Reynolds numbers the process has, the higher intensive swirl motion it produces.

Totally, the disturbance promoted by helical wires leads to higher heat transfer rate in the 6

heat exchangers studied in this work. Two correlations, the first of which was for the friction

coefficient and the second of which was for the augmented Nusselt number, were developed

based on the plain tube’s correlation. The MAPD of the correlation of friction coefficient was

9%, and that of Nusselt number was 5%. Regarding the heat-exchanger containing coiled

wire inserts P = 69 mm, the Nusselt number was enhanced by 1.77 times as compared with

the plain heat-exchanger. To find the optimum heat-exchanger with wire coiled inserts, two

criteria should be taken into account: 1) high Nusselt number 2) low pressure drop.

Regarding both of the factors, the introduced heat-exchangers were tested by the means of

overall enhancement efficiency relation. The results indicate that the coiled wire with pitch of

69 mm attributed a promising yield with respect to others.

21
 Nomenclature

Cp Heat capacity (kJ/kg K)

d Diameter of wire (mm)

D Inside diameter of tube (mm)

Dh Hydraulic diameter (mm)

f Friction coefficient

volumetric force

GCI Grid convergence index

Gs A constant number (1.25)

gi Grid converged results (For example Nu)

h Heat transfer coefficient (W/m2 K)

k Thermal conductivity (W/m K)

L Heat exchanger length (mm)

Mass flow rate (kg/m3)

Nu Nusselt number

NS Wire Normalize surface area

P Pitch length (mm)

p Pressure (Pa)

Pr Panhandle number

Re Reynolds number

T Temperature (K)

udf User defined function

z Order of convergence rate

Velocity (m/s)

t Time (s)

ρ Mass density (kg/m3)

Bulk viscosity (Pa.s)

22
 Water viscosity (Pa.s)
w

Overall enhancement efficiency

23
References

[1] He Y, Zheng H, Zhang Q, Tang Y, Zhang J. Analysis of Enhanced Heat Transfer

Performance Through Duct with Constant Wall Temperature. Journal of Thermophysics and
Heat Transfer 2012;26(3):472-5.

[2] Liu S, Sakr M. A comprehensive review on passive heat transfer enhancements in pipe
exchangers. Renewable and Sustainable Energy Reviews 2013;19:64-81.

[3] Zimparov VD, Penchev PJ. Performance evaluation of some tube inserts as heat transfer
enhancement techniques. Heat transfer engineering 2006;27(8):39-46.

[4] Dewan A, Mahanta P, Raju KS, Kumar PS. Review of passive heat transfer augmentation
techniques. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power
and Energy 2004;218(7):509-27.

[5] Webb RL, Kim N-H. Principl of enhanced heat transfer. Taylor Francis: New York, NY, USA
1994.

[6] Liebenberg L, Meyer JP. In-tube passive heat transfer enhancement in the process industry.
Applied Thermal Engineering 2007;27(16):2713-26.

[7] Saqr KM, Shehata AI, Taha AA, ElAzm MMA. CFD modelling of entropy generation in
turbulent pipe flow: Effects of temperature difference and swirl intensity. Applied Thermal
Engineering 2016;100:999-1006.

[8] Naphon P, Sriromruln P. Single-phase heat transfer and pressure drop in the micro-fin tubes
with coiled wire insert. International communications in heat and mass transfer
2006;33(2):176-83.

[9] Garcia A, Vicente PG, Viedma A. Experimental study of heat transfer enhancement with wire
coil inserts in laminar-transition-turbulent regimes at different Prandtl numbers. International
Journal of Heat and Mass Transfer 2005;48(21):4640-51.

[10] Chiou J. Experimental investigation of the augmentation of forced convection heat transfer in
a circular tube using spiral spring inserts. Journal of heat transfer 1987;109(2):300-7.

[11] Uttarwar S, Rao MR. Augmentation of laminar flow heat transfer in tubes by means of wire
coil inserts. Journal of heat transfer 1985;107(4):930-5.

[12] Eiamsa-Ard S, Nivesrangsan P, Chokphoemphun S, Promvonge P. Influence of combined

non-uniform wire coil and twisted tape inserts on thermal performance characteristics.
International Communications in Heat and Mass Transfer 2010;37(7):850-6.

[13] Vahidifar S, Kahrom M. Experimental Study of Heat Transfer Enhancement in a Heated Tube
Caused by Wire-Coil and Rings. Journal of Applied Fluid Mechanics 2015;8(4):885-92.

[14] Muñoz-Esparza D, Sanmiguel-Rojas E. Numerical simulations of the laminar flow in pipes

with wire coil inserts. Computers & Fluids 2011;44(1):169-77.

[15] Solano J, Herrero R, Espín S, Phan A, Harvey A. Numerical study of the flow pattern and
heat transfer enhancement in oscillatory baffled reactors with helical coil inserts. Chemical
Engineering Research and Design 2012;90(6):732-42.

24
[16] McDonough J, Ahmed S, Phan A, Harvey A. A Study of the Flow Structures Generated by
Oscillating Flows in a Helical Baffled Tube. Chemical Engineering Science 2017.

[17] Feng Z, Luo X, Guo F, Li H, Zhang J. Numerical investigation on laminar flow and heat
transfer in rectangular microchannel heat sink with wire coil inserts. Applied Thermal
Engineering 2017;116:597-609.

[18] Zhang C, Wang D, Zhu Y, Han Y, Wu J, Peng X. Numerical study on heat transfer and flow
characteristics of a tube fitted with double spiral spring. International Journal of Thermal
Sciences 2015;94:18-27.

[19] Akhavan-Behabadi M, Kumar R, Salimpour M, Azimi R. Pressure drop and heat transfer
augmentation due to coiled wire inserts during laminar flow of oil inside a horizontal tube.
International Journal of Thermal Sciences 2010;49(2):373-9.

[20] Guo J, Fan A, Zhang X, Liu W. A numerical study on heat transfer and friction factor
characteristics of laminar flow in a circular tube fitted with center-cleared twisted tape.
International Journal of Thermal Sciences 2011;50(7):1263-70.

[21] Shabanian S, Rahimi M, Shahhosseini M, Alsairafi A. CFD and experimental studies on heat
transfer enhancement in an air cooler equipped with different tube inserts. International
Communications in Heat and Mass Transfer 2011;38(3):383-90.

[22] Štigler J. Optimal Mapped Mesh on the Circle. ARCADEA; 2009.

[23] Webb R. Performance evaluation criteria for use of enhanced heat transfer surfaces in heat
exchanger design. International Journal of Heat and Mass Transfer 1981;24(4):715-26.

25
 Table 1: Properties of the flowing fluid used in the simulated model [19].

properties Temperature (°C)

30 50 70 90

Viscosity (Pa.s) 0.0379 0.0192 0.011 0.007

Specific heat (J/kg.K) 1925.65 1999.8 2074 2148.2

Thermal conductivity (W/m.K) 0.1621 0.1600 0.1585 0.1568

Density (kg/m3) 855.6 841.1 826.6 812.1

26
Table 2. Properties of different types of coiled wire inserts used in the experiment [19] beside
other non-uniform coiled wire inserts used in this CFD study.

Tube Diameter of the coiled Pitch Tilted angle Wires surface area
number wire length (grade) (mm2)

d (mm) P (mm) α

1 2 65 49.2 17169

2 2 47 58.1 21441

3 3.5 69 45.7 28991

4 3.5-2 26-65 69.8-49.2 29620

5 3.5-2 69-47 45.7-58.1 25660

6 3.5-2 69-28 45.7-69.7 30100

27
Table 3: Different input velocities used in this work in order to study the variation of Nusselt
number/Friction factor with Reynolds number.

u1 u2 u3 u4 u5 u6 u7

0.077581 0.150157 0.400418 0.616883 0.800835 1.053773 1.401462

28
Table 4: Grid convergence variables and order of accuracy for very coarse, coarse, fine and
very fine mesh cells defined in Figure 3.

1 2 3 R1 R2 P GCI1 GCI2

46 36 24 0.666 0.782 0.879 0.041 0.040

29
Figure 1: The configuration of the utilized heat-exchanger in the simulation process [19]. Tin
is 312 K, and Tw is 373 K. The value of the inlet velocity is changed depending upon the
required Reynolds number. All of the walls are assigned to the no-slip boundary condition.

30
a)

b)

Figure 2: A sample of the geometry utilized in computational fluids dynamics with the
related meshing in this study. a) a sketch of one pitch of the heat-exchanger along z axis
position. b) a cross sectional view of the heat-exchanger.

31
Figure 3: The Nusselt diagrams produced from five different mesh elements for the wire coil
heat-exchanger the with pitch length of 65 mm (tube 1). Acceptable results are developed by
the very fine mesh elements with non-uniform mesh sizes in the vicinity and center of the
tube.

32
a)

b)

c)

Figure 4: a) Temperature distribution in the heat-exchanger with wire coils of 2 mm

diameter and 65 mm pitch length at Reynolds 90. b). Pressure distribution in the heat-
exchanger. c) Velocity distribution in a small portion of the heat-exchanger.

33
(a)

(b)

34
 (c)

Figure 5: Friction coefficient according to the Reynolds number for the plain tube and coiled
wire tubes. a) the plain pipes b) the wires with pitch length 65 mm and diameter of 2 mm c)
the wires with pitch length 69 mm and diameter of 3.5 mm.

35
(a)

(b)

(c)

(d)

Figure 6: A repeating pattern of vortices caused by the separation of the flowing fluid around
the wires for a) Re = 90 b) Re = 400 c) Re = 800 d) Re = 1000.

36
Figure 7: Reynolds numbers vs. friction coefficients for the plain heat-exchanger and the
exchangers containing coiled wire inserts.

37
Figure 8: The comparison of predicted friction values using Eq. (9) with CFD outcomes.

38
(a)

(b)

39
 (c)

Figure 9: Reynolds-Nusselt variations for the plain tube and coiled wire tubes. a) the plain
pipes b) the wires with pitch length 65 mm and diameter of 2 mm c) the wires with pitch
length 69 mm and diameter of 3.5 mm.

40
Figure 10: The swirl motion of pathlines, and the effect of Reynolds number on the location
of pathlines along the heat-exchanger.

41
Figure 11: Plots of the tangential velocity vectors beside contour diagrams of the axial
velocity on the cross sections of Z = 0.5, 1 and 1.5 m.

42
Figure 12: A comparison of Nusselt numbers for different coiled wire inserts along varied
Reynolds numbers.

43
Figure 13: A comparison of tangential velocity for different coiled wire inserts along varied
Reynolds numbers.

44
Figure 14: A comparison of vorticity for different coiled wire inserts along varied Reynolds
numbers.

45
Figure 15: The comparison of predicted Nusselt values using Eq. (12) with CFD outcomes

46
Figure 16: CFD data on overall enhanced efficiency for the heat-exchangers filled with
coiled wire inserts.

47
Research Highlights

 The influence of coiled wire inserts on various parameters of a double pipe heat-
exchanger is studied.
 The latter parameters are the Nusselt number, friction coefficient and overall
efficiency.
 The CFD simulations have been performed using two commercial softwares.
 Taking the advantage of proper wire coils could improve the Nusselt values to 1.77
times.
 Proper friction coefficient and Nusselt number correlations are proposed.

48