You are on page 1of 11

Chemical Engineering Science 62 (2007) 2386 – 2396

www.elsevier.com/locate/ces

Coiled flow inverter as a heat exchanger


Vimal Kumar, Monisha Mridha, A.K. Gupta, K.D.P. Nigam ∗
Department of Chemical Engineering, Indian Institute of Technology, Delhi, Hauz Khas, New Delhi 110016, India

Received 25 August 2006; received in revised form 9 January 2007; accepted 14 January 2007
Available online 1 February 2007

Abstract
In the present work attempts are made to investigate the hydrodynamics and heat-transfer characteristics of a coiled flow inverter (CFI) as
heat exchanger at the pilot plant scale. The experiments are carried out in counter-current mode operation with hot fluid in the tube side and
cold fluid in the shell side. Experimental study is made over a range of Reynolds numbers from 1000 to 16,000 using water in the tube side
of the heat exchanger. The shell side fluids used are either cooling water or ambient air. The coiled flow inverter is made up of coils and 90◦
bends and inserted in a closed shell. The shell side is fitted with three types of baffles to provide high turbulence and avoid channeling in
the shell side. The bulk mean temperatures at various downstream positions are reported for different flow rate on tube side, as well as the
heat transfer efficiency of the heat exchanger is also reported. Pressure drop and overall heat-transfer coefficient is calculated at various tube
and shell side process conditions. The outer and inner heat-transfer coefficients are determined using Wilson plot technique. The results show
that at low Reynolds numbers, heat-transfer is 25% higher as compared to coiled tubes. At high Reynolds numbers, the configuration has less
influence on heat transfer. New empirical correlations are developed for hydrodynamic and heat-transfer predictions in the coiled flow inverter.
䉷 2007 Elsevier Ltd. All rights reserved.

Keywords: Coiled flow inverter (CFI); Coiled tube; Bends; Heat transfer; Friction factor

1. Introduction fluid velocity. In the process industry the turbulence between


the fluid elements is increased by using either inserts or internal
Heat transfer is an essential component of nearly all industrial grooves for enhancing the mixing and heat-transfer (Nauman
processes, ranging from power production, chemical and food et al., 2002, 2003). The inserts and internal also leads into
industries, electronics, environment engineering, waste heat higher wall shear stresses and pressure drop since they intro-
recovery, manufacture industry, air-conditioning, refrigeration duce an additional no-slip surface or roughness, which results
and space applications. Consequences of improper heat-transfer into larger pumping costs. Therefore, it is advantageous to con-
include non-reproducible processing conditions and lowered sider devices, which do not increase the pressure drop and also
product quality, resulting in the need for more elaborate down- not affect the smoothness of the inner wall, but provide en-
stream process system and increased heat-transfer area. Despite hancement in the degree of fluid mixing.
its importance, however, heat-transfer performance is seldom Secondary flow in a plane normal to the principal flow
characterized rigorously for industrial systems. Detailed char- direction is very effective to enhance fluid mixing and heat-
acterization is important, particularly in laminar flows, which transfer. In the coiled tubes, the modification of the flow is due
have a serious potential to lead to inhomogeneity and poorly to the centrifugal forces caused by the curvature of the tube,
mixed regions within the flow systems. In case of laminar flow which produce a secondary flow field with a circulatory motion
condition, the heat-transfer is low as compared to turbulent flow pushing the fluid particles toward the core region of the tube
conditions. In a turbulent or unstable flow, on the other hand, (Fig. 1a). Because of the stabilizing effects of this secondary
the heat transfer is higher due to transverse fluctuations in the flow, laminar flow persists too higher Reynolds number value
in helical coils as compared to straight tubes. Consequently, the
differences in heat and mass transfer performance between coils
∗ Corresponding author. Tel./fax: +91 11 26591020. and straight tubes are particularly distinct in the laminar flow
E-mail address: nigamkdp@gmail.com (K.D.P. Nigam). region. Dean (1927, 1928) was the first to investigate flow in

0009-2509/$ - see front matter 䉷 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2007.01.032
V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396 2387

centrifugal
obtained under the condition of significant molecular diffusion
force and in the case of negligible molecular diffusion the value of
a b dimensionless time at which the first element of tracer appeared
at outlet was as high as 0.85.
Jones et al. (1989), Acharya et al. (1992, 1994a,b), Duchene
et al. (1995) and Peerhossaini and Le Guer (1992) presented an
centrifugal
alternative regime in laminar flow that has dispersive properties
force close to a turbulent regime using the phenomenon of chaotic
centrifugal centrifugal
force force
advection. In chaotic advection, the fluid–particle trajectories
are chaotic and enhance mixing, consequently increasing heat
Dean roll-cells
transfer. Such tools were addressed in Mokrani et al. (1997).
Regular flow (Dean flow) Alternate Dean flow The details of the chaotic and temporal flows are discussed by
Acharya et al. (1992) and Chagny et al. (2000).
Fig. 1. Generation of spatially chaotic particle paths in coiled tube by inserting In the present work, chaotic flow is generated by the phe-
a 90◦ bend. (Source: Kumar and Nigam, 2005; Castelain et al., 2000) nomenon of flow inversion by inserting 90◦ bends between reg-
ular coil tubes. The axis of each coil is rotated by 90◦ with
respect to the neighbouring coil; this makes the generation of
helically coiled circular tubes, and found that a pair of sym- roll-cells in a plane perpendicular to the previous one, due to
metric vortices was formed on the cross-sectional plane due to the reorientation of the centrifugal forces. This geometrical per-
centrifugal force. The strength of secondary flow is character- turbation is the main cause of flow inversion phenomenon.
ized by Dean number The aim of the present work is to characterize the perfor-
NRe mance of a coiled flow inverter (CFI) as heat exchanger for
NDe = √ , (1) a water–water or water–air (tube side fluid–shell side fluid)

counter-current flow system experimentally on the pilot plant
where  is the curvature ratio and is defined as the ratio of coil scale. The effect of the fluid flow rate on the heat transfer and
diameter to tube diameter i.e.,  = D/d. Extensive reviews on hydrodynamics were studied in the tube as well as in the shell
flow fields in curved ducts were reported by Berger et al. (1983), side of the heat exchanger. In the present work bent coils were
Shah and Joshi (1987), Nandakumar and Masliyah (1986) and considered as tube side which is inserted into a cylindrical shell.
Saxena and Nigam (1986). The shell of the heat exchanger is comprised of three types of
baffles. These features along with the pilot plant scale study are
1.1. Chaotic configuration (combination of coils and bends) being reported first time, which is not considered in the pre-
vious literature. The most important information in the design
There is global heat-transfer enhancement in the helical of the heat exchanger is the pressure drop and the heat-transfer
coiled tube; still the isotherms of temperature for different coefficient. Based on the fitting of experimental data, new cor-
fluids contain segregated cold and hot regions. The Dean roll relations of the heat-transfer coefficient and friction factor for
cells divide the cross-section into two zones in each of which the CFI are proposed for practical applications.
the isotherm forms the closed curves. Fluid particles inside the
Dean roll cells are prevented from approaching the hot walls;
thus mixing is poor, giving rise to a heterogeneous tempera- 2. Experimental study
ture field. Therefore in the coiled tubes, heat and mass transfer
can be further enhanced by inserting some perturbation in the The purpose of the study is to determine experimentally
geometry. the hydrodynamics and fully developed heat-transfer variations
Techniques commonly used to enhance mixing often involve with flow rates in the CFI as heat exchanger. Experiments on
the generation of turbulent flow. In some cases, however, flu- the heat exchanger were carried out in order to obtain a mea-
ids with long molecular chains can be damaged by high shear sure of their relative performance in terms of heat transfer and
stresses, and also energy is lost by turbulent agitation. In the pressure drop and the results obtained were compared with the
regular laminar regime, mixing is induced mainly by molec- data reported in the literature for coiled tube. Since the exper-
ular diffusion. The idea of generating a spatial (Lagrangian) iments were carried out over a range of Reynolds numbers,
chaotic behaviour from a deterministic flow by simple geomet- therefore the comparison was made based on the inner heat-
rical perturbations has attracted much attention in recent years. transfer coefficient as well as the pressure drop in this range.
Saxena and Nigam (1984, 1986) proposed a new technique, The pressure drops and heat-transfer experimental data analy-
“bending of helical coils,” to cause multiple flow inversion at sis were also carried out in shell side of the heat exchanger.
low flow rates. For the case of fully developed secondary flow,
a 90◦ bend induces a flow inversion, which narrows the RTD 2.1. Experimental apparatus
for equal arm lengths before and after the bend (Fig. 1b). They
assessed their device by the fact that even at a Dean number A schematic diagram of the heat exchanger test facility is
of 3 the value of the dispersion number as low as 0.0013 was shown in Fig. 2. The test facility was composed of a primary hot
2388 V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396

Fig. 2. Experimental setup for the pilot plant of coiled flow inverter heat exchanger.

loop and a secondary cold loop. The primary hot loop consists ibrated as per IS code 2056-1962. The thermocouples were
of a liquid storage vessel (V-101), made of SS-304 with a ca- also provided after every bank of the heat exchanger (a bank is
pacity of 500 l, a reciprocating plunger type pump (P-101) ca- comprised of 16 turns with three bends). For the pressure mea-
pable of flow rates from 30 to 320 1/h is provided to pump the surement of the tube side fluid, capacitance type pressure trans-
water (PL 2017, Positive Metering Pumps Pvt. Ltd., Nashik, mitter (PT 101 and 102) calibrated to a span of 0–5 bar were
India). To minimize the fluctuations in the tube side flow a used (Emerson Process Management Pvt. Ltd., Mumbai, India).
dampener of 10 l capacity made of SS316 is provided after the The accuracy of the pressure-measuring device was ±1.0%.
pump (P-101). An in-line electric resistance preheater (E-101) The cooling water pressure was measured by using pressure
of 22 KW rating is being designed for handling pressure up to gages (PT 104 and 104) (0–2 bar) having an accuracy class of
5 bar and temperature range from 40 to 180 ◦ C heats the liquid 1.6%. The flow rate is measured by a turbine rotameter type flow
to the desirable inlet temperature. An automatic PID regulator meter (FI-101) with a range of 30–300 l/h and signal is trans-
(TIC-101) was provided to control the outlet temperature of the mitted to an electrical transmitter, which provides an electrical
preheater. output in the range of 4–20 mA (Scientific Devices Pvt. Ltd.,
The secondary cold loop fluids consists either cooling wa- Mumbai, India). The turbine flow meter has only moving parts,
ter or ambient air. An induced draft type counter-flow-cooling a precise helical rotor. Rotation of the rotor is electronically
tower (CT-101), manufactured by Patel Cooling waters, Bar- detected and processed. The flow meters used were calibrated
oda, India was used for pumping cooling water directly into before, after and during the experiments. Afterwards the flow
the shell side of the heat exchanger. In case of ambient air meters were calibrated by weighing the water that was poured
in the shell side of the heat exchanger a blower (B-101) was into a bucket while the time was taken by a stop-watch. The ac-
used. curacy of the flow meters was found to be within ±2%. For the
The temperatures were measured with Teflon coated, gage shell side, the flow rate was measured by a turbine rotameter
30 and copper-constantan thermocouples (Yenkay Instruments type flow meter (FI-201) with the range of 500–3000 l/h.
and Controls Pvt. Ltd., Pune, India) at the inlet (TE 101 and The experiments were conducted at higher temperature
TE 109) and outlet (TE 111 and TE 112) of the tube as well as (60–120 ◦ C). At these higher temperatures the liquid may
the shell of the heat exchanger. These are T type thermocouples convert into vapour. Therefore to keep tube side fluid in
ranging from 0 to 300 ◦ C and an accuracy of 0.3 ◦ C or 0.4%. single phase only a pressure-regulating bottle is provided.
The thermocouple junction was kept small to achieve accuracy Through this vessel the process line was pressurized from
in local measurements. These thermocouples were soldered to 3 to 5 kg/cm2 g to maintain single phase in the tube side.
the bottom of 30 mm long pieces of 2.5 mm diameter brass tub- In the pressure-regulating bottle a level of liquid was main-
ing penetrating inside the tube. The thermocouples were cal- tained through a liquid level controller (VALFLO, Ahmedabad,
V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396 2389

Fig. 3. Chaotic configuration.


Fig. 4. Flow structure and the baffles in the shell side of the heat exchanger.

India) and the excess liquid was recycled back to the feed tank
(V-101).

2.2. Heat exchanger

Considerable efforts have been made into design and fabri-


cation of the CFI as heat exchanger. The heat exchanger was
constructed from SS-316 tubing and standard pipe connections.
The tube of heat exchanger had an outer diameter of 12.7 mm
with wall thickness of 1.2 mm. The radius of curvature (mea-
sured from the centre of the inner tube) was 102 mm. Fig. 3
shows the cross-sectional and lateral view of the exchanger.
The pitch of the coil was taken as 20 mm. There were total
eight banks having 31 bends in the heat exchanger. Each bank
comprises of four arms and one arm consists of four turns. The
total unfolded length of the heat exchanger was 44.134 m. All
the banks were connected with the Swagelok fittings. The shell
side of the heat exchanger had an outer diameter of 350 mm
and a wall thickness of 2 mm. The shell of the exchanger Fig. 5. Types of baffles and arrangement of tube and baffles in the shell side
was provided with three kind of baffles; (a) seven alternat- of the heat exchanger: (1) central rod for support, (2) cut plate, (3) cylindrical
ing vertical baffles that maintains the turbulent flow conditions rod rounded in square shape and (4) bent coils.
(Figs. 4 and 5), (b) one cylindrical rod across the heat ex-
changer to avoid channeling and (c) seven square shape pipes
within each bank to avoid channeling. The alternating vertical 2.3. Experimental procedure
baffles were in the form of circular plates fitted at a spacing
of 185 mm. The baffles not only maintained the turbulent flow Experiments were conducted under steady state conditions
condition but also avoid the channeling within the bank space. with DM water in the tube side and either tap water or am-
The axis of the heat exchanger was vertical and the primary bient air in the shell side as the working fluids. The flow rate
fluid was preheated before entering into the test section. in the tube side was varied over a range of 30–320 l/h for a
2390 V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396

constant shell side flow rate. The shell side flow rates were 20
varied from 800 to 2500 l/h. Temperature data were recorded Qs = 800 kg/hr
at every 10 s. Temperature measurements from the 300 s of the Qs = 1000 kg/hr
stable system were used, with temperature reading fluctuations Qs = 1500 kg/hr
within ±0.15 ◦ C. Though the type-T thermocouples had limits 10
Qs = 2000 kg/hr
of error of 0.5 ◦ C, when placed in a common water solution the

(qs - qt)/qt x 100


readings at steady state were all within ±0.1 ◦ C. All the ther-
mocouples were constructed from the same roll of thermocou-
0
ple wire, and hence the repeatability of temperature readings
is high. Similarly for the heat-transfer study in the shell side,
the tube side flow rate was kept constant at 220 l/h and varia-
tion in the shell side flow rate is made. At the inlet of the outer -10
tube of the heat exchanger the cooling water temperature was
26–28 ◦ C, and it rises by 4–5 ◦ C at the outlet of the outer tube.
During the experiments the ambient temperature was 27–28 ◦ C,
therefore, there was not much heat loss from the outer wall of -20
0 5000 10000 15000 20000
the heat exchanger. Reynoldsnumber (Tube side)

2.4. Calculation of heat-transfer coefficient Fig. 6. Thermal equilibrium criterion vs. Reynolds number.

The heat-transfer study was carried out for both shell and tube
side, which was done by measuring the overall heat-transfer T1 , and the outlet temperature difference, T2 , using the fol-
coefficients. The total thermal resistance between the two flu- lowing equation of White (1984):
ids was due to the inner and outer heat-transfer coefficients, as (T2 − T1 )
well as the resistance of the tube wall, which is very small. The TLMTD =   . (5)
criterion for thermal equilibrium of the whole system was ver- ln  T2
T1
ified for each experiment before recording the measurements;
The inner and outer heat-transfer coefficients are usually ob-
measurements were taken when the heat balance was within
tained from the overall thermal resistance consisting of three
5%. This takes into account the heat-transfer losses from the
resistances in series: the convective resistance in the inner
shell to the surrounding air. To check the heat balance, the pri-
surface, the conductance resistance of the pipe wall and the
mary and secondary heat fluxes were compared. The primary
convective resistance on the outer surface by the following
and secondary heat flux is determined as
equation:
q = ṁt cp,t (tin − tout ) = ṁs cp,s (Tin − Tout ), (2) 1 A0 A0 ln(do /di ) 1
= + + , (6)
where the subscript ‘s’ refers to the shell fluid and ‘t’ refers Uo Ai hi 2kL ho
for tube side. Since the secondary fluid is the cold one, the where do is the outer diameter of the tube; di is the inner
secondary heat flux is negative. The fluid properties were de- diameter of the tube; k is the thermal conductivity of the wall;
termined at mean inlet–outlet temperature. The criterion for and L is the length of the tube.
thermal equilibrium is then written as Heat-transfer coefficients for the shell side, ho , and for the
qs − qt tube side, hi , were calculated using traditional Wilson plot tech-
× 100  ± 5%. (3)
qt nique as described in Rennie and Raghavan (2005) and Kumar
et al. (2006). For the calculation of inner heat-transfer coeffi-
The above-mentioned criterion for thermal equilibrium is
cient, the mass flow rate in the shell side was kept constant;
plotted against the Reynolds number in Fig. 6. The figure rep-
and assumed that the outer heat-transfer coefficient was con-
resents the thermal equilibrium criterion for various process
stant. The inner heat-transfer coefficient was assumed to be-
conditions in the tube side as well as in the shell side of the heat
have in the following manner with the fluid velocity in the tube
exchanger. The values scatter within ±5%. When the thermal
side, Vi :
equilibrium state is verified, the heat exchangers can be char-
acterized by temperature and pressure loss measurements. The hi = CV ni . (7)
overall heat-transfer coefficient was calculated from the tem-
perature data and the flow rates using White (1984) equation Eq. (7) was placed into Eq. (6) and the values for the constant
and is given below: C and the exponent n were determined through curve fitting.
The inner and outer heat-transfer coefficients could then be
q
Uo = , (4) calculated. This procedure was repeated for each shell side flow
A0 TLMTD rate. Similar procedure was adopted for the calculation of shell
where A0 is the surface area and TLMTD is the log-mean tem- side heat-transfer coefficient (by keeping tube side flow rate
perature difference, based on the inlet temperature difference, constant and shell side flow rate was varied).
V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396 2391

2.5. Fluid flowing in shell side: the equivalent diameter Area, m2


0 0.22 0.44 0.66 0.88 1.1 1.32 1.54 1.76 --
90
When a fluid flows in a conduit having other than a circular
cross-section, such as the present shell side of heat exchanger 80 Qt = 80 l/hr
consisting of baffles, it was convenient to express heat-transfer Qt = 100 l/hr
Qt =120 l/hr
coefficients and friction factors by the similar equations and 70

Temperature,°C
curves used for pipes and tubes. To permit this type of repre-
60
sentation for outer heat-transfer it has been found advantageous
to employ an equivalent diameter De concept. The equivalent 50
diameter is four times the hydraulic radius, and the hydraulic
radius is, in turn, the radius of a pipe equivalent to the annulus 40

cross-section. The hydraulic radius was obtained as the ratio 30


of the flow area to the wetted perimeter. It can be seen from
Figs. 3 and 4 that the flow path is very complex, therefore 20
Inlet 1 2 3 4 5 6 7 Outlet --
the equivalent diameter was calculated by using analogy from
Number of banks
packed bed as
4 ∗ void volume Fig. 7. Surface area vs. temperature at shell side flow rate of 1500 l/h.
De = 4rh = , (8)
wetted surface
where void volume is shell volume minus the tube and baf-
downstream the temperature decreases. The maximum change
fles volume, wetted surface is the surface used in heat transfer
in temperature was observed within the area 0.22 m2 (i.e., first
and pressure-drop studies. The wetted surface for heat transfer
bank) and there was no change in temperature after the 0.84 m2
and pressure drops are different. For heat transfer the wetted
for all the three-tube side flow rates. This is because of the phe-
perimeter is the outer circumference of the chaotic configu-
nomenon of flow inversion, which homogenizes the tempera-
ration with diameter do . Therefore, the heat transfer in shell
ture after the first bank (i.e., A0 =0.2 m2 ). In the helically coiled
side is
tube the Dean vortices transport cold fluid elements from the
4 ∗ void volume 4 ∗ void volume neighbourhood of the wall to the centre of the tube. However,
De = 4rh = = . (9)
wetted surface do L the fluid elements trapped in the centre of the Dean vortices
In pressure-drop calculations the friction not only results from were prevented from approaching the cold wall. Consequently
the resistance of the outer pipe but was also affected by the the centreline of the tube was overheated and the centre of the
outer surface of the inner pipe and the baffles. The total wetted Dean roll cells were not heated enough. Therefore the heat-
perimeter is (do Ls + Di Ls + db Lb ) and for the pressure drop transfer phenomenon in the helical coil was not very high. The
in annuli Dean vortices were also locally present after the bend, but after
each curvature plane the centrifugal force was reoriented, so
4 ∗ void volume
De = the Dean vortices of the previous arm vanished and reappeared
Frictional wetted perimeter in a plane perpendicular to the previous plane. Fluid elements
4 ∗ void volume that were trapped in the Dean vortices of the coiled tube, can
= , (10)
(do Ls + Di Ls + db Lb ) visit cold regions close to the walls after the reorientation of
the Dean vortices. The homogenization of heating or cooling in
where db and Lb are the baffles diameter and length,
chaotic configuration is the consequence of this phenomenon.
respectively.

3. Result and discussion 3.2. Heat-transfer efficiency of the CFI as heat exchanger

The heat exchanger has been analysed in terms of tempera- Experiments were carried to out examine the effect of flow
ture variation along the downstream length and heat exchanger inversion on global heat transfer by assessing the efficiency of
efficiency before discussing the friction factor and fully de- heat exchange in the CFI as heat exchanger. This requires the
veloped heat-transfer coefficient in the tube and shell side, measuring of bulk fluid temperature at the entrance and exit of
respectively. both tube and shell side. The physical properties of the tube and
shell side fluid were calculated at average temperatures. The
3.1. Variation of bulk mean temperature at various banks heat-transfer rate for the tube side was calculated using Eq. (2).
The efficiency was calculated by the following expression:
Fig. 7 shows the cross-sectional average temperature at var- qt
ious locations in the heat exchanger. In the figure the shell side = . (11)
(ṁcp )mm (Tin − tin )
flow rate was kept constant for various tube side flow rates.
For all the three experiments the tube inlet temperatures were In the present work the term (ṁcp )mm corresponds to the tube
maintained at 80 ◦ C. From Fig. 7 it can be seen that as flow go side fluid. Therefore the above equation reduces to the ratio of
2392 V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396
ε = (tout - tin)/(Tin - tin)

1.1 1.40

Ttube,in 80 degee C; Qs = 1500


1.20
1 Ttube,in 100 degee C; Qs = 1500

Pressure drop, kg/cm2.g


1.00 Ttube,in 120 degee C; Qs = 1500
Tangent 2
0.9 Ttube,in 80 degee C; Qs = 1000
0.80
Heat exchnager efficiency,

0.8 0.60

0.40
0.7
Tangent 1
0.20

0.6
0.00
0 50 100 150 200 250 300 350
Flow rate, lt/hr
0.5
2000 7000 12000 17000
Reynolds number Fig. 9. Pressure drop vs. tube side flow rate on various process conditions in
the heat exchanger.
Fig. 8. Efficiency of coiled flow inverter as heat exchanger vs. Reynolds
number.
Gupta (1979) for coiled tube and is given below:

fc /fs = 1 + 0.033[log10 NHe ]4 for 1 < NHe < 3000, (14)


two temperature differences:
0.0075
(tout − tin ) fc − fs = √ for 4500 < NRe < 105 , 6.7 <  < 346
= . (12) 
(Tin − tin ) and 0 < (H /2Rc ) < 25.4, (15)
Fig. 8 shows the plot of heat exchanger efficiency vs. Reynolds where NHe is the Helical number and is represented by
number for the heat exchanger. Vertical bar indicates the un-  1/2
certainties of the values. The uncertainty interval for  is due to dv d/Dc
NHe = (16a)
the errors in measuring the temperatures at the inlet and outlet  1 + (H /Dc )2
of tube and shell sides, respectively. The uncertainties for the
and fs is the friction factor in straight tube
temperature have been estimated to be equal to 0.1 ◦ C. Using
the method described by Kline and MCClintock (1950), the fs = 0.079/NRe
0.25
. (16b)
relative uncertainty intervals for  have been calculated. The 
values vary from 0.5% for high values to the 1.5% for lower The fully developed heat-transfer predictions in chaotic
values. Fig. 8 also shows that at lower values of Reynolds num- configuration in the present work were compared with the
bers the efficiency is very high and as the Reynolds number in- experimental measurements of Shchukin (1969) for coiled
creases the efficiency decreases. It can be observed from Fig. 8 tube. According to their experimental data, Shchukin (1969)
that chaotic advection is more advantageous at lower Reynolds proposed the following correlation:
number. In the chaotic flow the pathlines is independent of the
dynamics of the flow. In chaotic flow, the enhancement in mix- NNu = 0.0575NRe0.33 0.42 0.43
NDe NP r
ing does not require high Reynolds number if the secondary for 6.2 <  < 62.5 and 26 < NDe < 7000, (17)
flow reached in stable state.
NNu = 0.0266[NRe0.85 0.15
/ + 0.2251.15 ]NP0.4r
for 6.2 <  < 104 and NRe.Crit < NRe < 67, 000. (18)
3.3. Friction factor in CFI
The experimental data of present study were also compared
The inner and outer Fanning friction factor were deter- with Sieder and Tate equation of heat transfer in straight tube
mined as
NNu = 0.023NRe
0.8 0.3
NP r . (19)
P D e
f= , (13) 3.3.1. Friction factor in heat exchanger
2V 2 L
3.3.1.1. Tube side pressure drop and friction factor Fig. 9
where L is the length of the heat exchanger and De is the shows the pressure drop for different values of flow rates in
equivalent diameter for the tube side and shell side, respectively. the chaotic tube. It can be seen from the figure that the pres-
The friction factor measurements in the chaotic configuration sure drop increases with increase in the tube side flow rate. It
were compared with the empirical correlations of Mishra and can be seen from Fig. 9 that the pressure drop curves has two
V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396 2393

0.1 figure that the friction factor in the tube side of the heat ex-
Tt,in = 80 °C;Qs = 1500 l/hr changer is higher as compared to the coiled tube and straight
T
t,in
= 100 °C;Q = 1500 l/hr
s
tube, respectively.
Tt,in = 120 °C;Qs = 1500 l/hr
T
t,in
= 80 °C;Q = 1000 l/hr
s 3.3.1.2. Shell side friction factor Fig. 11a shows pressure drop
Proposed empirical correlation
vs. flow rate in the shell side. It can be seen from the Fig. 11a
that as the flow rate in the shell side increases the pressure drop
f

increases linearly. Based on the definition of the equivalent di-


Co
iled ameter the friction factor in the shell side of the heat exchanger
tub
St
e is analysed. The experimental predictions of friction factor in
ra the shell side of heat exchanger are shown in Fig. 11. The
ig
0.01 ht
tu
be Coiled tu
be
friction factor in the shell side decreases with decrease in the
Straigh
t tube Reynolds number. The friction factor in the CFI is also com-
pared with the straight tube friction factor. The friction factor
1000 10000
Reynolds number
in the CFI is higher as compared to the straight tube, which
is due to the presence of baffles in the shell side of the heat
Fig. 10. Friction factor vs. Reynolds number on various process conditions exchanger. To the best of our knowledge, this is the first study
in the heat exchanger. made at pilot scale in the CFI as heat exchanger with consid-
ering baffles in the shell side.
slopes, which shows the transition in the flow regime from
3.3.2. Fully developed heat transfer
laminar regime to turbulent regime. From the experimental
Fig. 12 shows the overall heat-transfer coefficient (Uo ) for
measurements of pressure drop in tube side a new empirical
different values of flow rates in the chaotic tube and in the shell
correlation is developed for friction factor predictions in the
side. It can be seen from the figure that the overall heat-transfer
heat exchanger. The correlation for tube side was developed for
coefficient increases with increase in the tube side flow rate for
two different ranges of Reynolds number as
a constant flow rate in the shell side. About 30% increase in
fCH E,1 = fs (1 + 0.0456NDe
0.603
) for NRe < 10, 000, (20a) the overall heat-transfer coefficient was observed over two-fold
change in the tube side flow rate. It can also be concluded from
fCH E,2 = fs (1 + 0.3377NDe
0.042
) for NRe > 10, 000, (20b) Fig. 12 that the overall heat-transfer coefficient increases with
where fCH E,1.2 are friction factor in the tube side of heat increase in the flow rate in the shell side for a constant tube
exchanger. side flow rate.
Fig. 10 shows the experimental predictions in the tube side at
various process conditions of tube and shell side of the heat ex- 3.3.2.1. Inner Nusselt number Based on the experimental mea-
changer. The percentage error of the proposed correlation with surements new empirical correlations was developed for two
experimental data was within ±10%. Fig. 10 also illustrates different range of Reynolds number as
the comparison of present predictions of friction factor in heat 0.7 0.4 −0.1
NNu = 0.08825NRe NP r  for NRe < 10, 000, (21a)
exchanger with the experimental data for coiled tube (Mishra
0.85 0.4 −0.1
and Gupta (1979)) and straight tube. It can be seen from the NNu = 0.0271NRe NP r  for NRe > 10, 000. (21b)

0.4 0.0120
Bent Coil heat Exchanger
0.35
Straigh tube
0.0110
0.3
Pressure drop, kg/cm2

Friction factor

0.25
0.0100

0.2

0.0090
0.15

0.1
0.0080
0.05

0 0.0070
0 500 1000 1500 2000 3000 4000 5000 6000 7000 8000 9000
Flow rate, lt/hr Reynolds number

Fig. 11. (a) Pressure drop and (b) friction factor in the shell side of the heat exchanger.
2394 V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396

800 tube in the literature (Shchukin, 1969). The mechanism respon-


sible for the enhancement in heat transfer in the tube side is the
700 periodic switching of coil axis in the downstream direction. In
order to observe a noticeable effect of chaotic mixing on heat
600 transfer, the device should have a large number of 90◦ bends.
But, on the other hand, there is a limitation on the frequency at
500 which this can be done. After a 90◦ bend, the flow field needs a
Uo, W/m2.sec

certain transition length before it attains the “developed” form


400
of Dean vortices and adjusts itself to the new curvature. If the
axis is switched much more frequently than this, the effect of
300
chaotic mixing diminishes and the pressure drop also increases
simultaneously. This points to an optimum number of bends
200
maximum heat-transfer enhancement.
100
3.3.2.2. Shell side heat-transfer coefficient Fig. 14a shows the
0 variation of overall heat-transfer coefficients with the shell side
0.00 0.20 0.40 0.60 0.80 1.00 1.20 flow rate for different tube side flow rates. It can be seen from
Velocity, m/s Fig. 14a that as the shell side flow rate increases the over-
Qs = 800 kg/hr; Water Qs = 2000 kg/hr; Water all heat-transfer coefficient increases. Similar phenomenon was
observed for tube side flow rate, the overall heat-transfer coef-
Qs = 1000 kg/hr; Water Qs = 730 kg/hr; Air
ficient increases with increase in tube side flow rate. The outer
Qs = 1500kg/hr; Water Qs = 900 kg/hr; Air Nusselt numbers were evaluated at the modified Reynolds num-
ber that uses the concept of equivalent diameter. Fig. 14b shows
Fig. 12. Overall heat-transfer coefficient vs. tube side velocity various process the shell side Nusselt number vs. Reynolds number. It can be
conditions in the heat exchanger. seen from the figure that the Nusselt number values increase
with increase in the shell side flow rates. This is due to increase
in turbulence between the fluid elements. The shell side heat-
200 transfer study was also carried out ambient air. A heat-transfer
Straight tube (Seader and Tate's equation)
180 Coiled tube (Shchukin (1969)) correlation was proposed based on the heat-transfer data for the
QS = 800 l/hr; Water
QS = 1000 l/hr; Water
shell side as
160
QS = 1500 l/hr; Water
140 QS = 1500 l/hr; Water NNu = 15.44Re0.345 Pr0.33 . (22)
Nusselt number

QS = 2000 l/hr; Water

120 QS = 2000 l/hr; Water The maximum deviation between the model results and the
QS = 750 Nm3/hr; Air
QS =620 Nm3/hr; Air experimental data was approximately ±1%.
100 Proposed empirical correlation

80 4. Conclusion
60
Experimental study of a new-coiled configuration was per-
40
formed on the pilot plant scale. The new design of CFI as heat
20 exchanger is presented which comprised of coils and 90◦ bends.
The insertion of 90◦ bends between the coiled tube generates
2000 4000 6000 8000 10000 12000 14000 16000 18000
Reynolds number
the phenomenon of flow inversion that increases the mixing be-
tween the fluid elements and enhance the heat transfer. Experi-
Fig. 13. Inner heat-transfer coefficient vs. Reynolds number various process mental study was made over a large range of Reynolds numbers
conditions in the heat exchanger. (from 1000 to 16,000) using water in the tube side of the heat
exchanger. The shell side fluids used were either cooling wa-
ter (cooling tower) or ambient air (blower). The shell side was
Fig. 13 illustrates the comparison of experimental values of fitted with baffles to provide high turbulence and avoid chan-
fully developed Nusselt number in the tube side of the heat neling in the shell side. The effect of flow inversion on temper-
exchanger with the empirical correlation predictions. It can be ature distribution at the outlet of the various banks, as well as
observed that the model results agree fairly well with the mea- on the overall efficiency of the heat exchanger was reported. It
surements. The maximum difference between the model results was found that the maximum decrease in the temperature was
and the experimental data was approximately ±4%. within the area of 0.22 m2 (first bank). Further the efficiency
The present experimental and empirical correlation predic- of the heat exchanger at low Reynolds number was near one
tions were also compared with the data reported in the literature and as the Reynolds number increases the efficiency decreases.
for coiled tube (Shchukin, 1969). In the present study, higher Pressure drop and overall heat-transfer coefficient was calcu-
values of heat transfer are obtained as compared to the coiled lated at various tube and shell side process conditions. It was
V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396 2395

900 450

Outer heat transfer coefficient (W/m2.°C)


Qt = 150 l/hr
400
800 Qt = 220 l/hr
350
Qt = 300 l/hr

Nusselt number
700 300

250
600
200

500 150

100 Experimental data


400
Proposed correlaton
50

300 0
0 500 1000 1500 2000 2500 0 1000 2000 3000
Flow rate (l/hr) Reynolds number

Fig. 14. (a) Overall heat transfer heat coefficient and (b) Nusselt number in the shell side of the heat exchanger.

observed that the overall heat-transfer coefficient increases with U overall heat-transfer coefficient, W/m2 K
increase in the tube side Reynolds number for a constant flow v velocity, m/s
rate in the shell side. Similar trends in the variation of overall Greek letters
heat-transfer coefficient were observed for different flow rates
in the shell side for a constant flow rate in the chaotic tube. A  curvature ratio, D/d
new empirical correlation was developed for friction factor and  density of fluid, kg/m3
heat transfer for shell and tube of the heat exchanger. The in-  dynamicviscosity, kg/ms
ner heat transfer in the heat exchanger was compared with the Subscripts
coiled tube heat-transfer data reported in the literature. It was
observed that at low Reynolds number the heat transfer was CFI coiled flow inverter
25% higher and at higher Reynolds number it was 12% higher t tube side
as compared to the coiled tube data reported in the literature. s shell side

Notation Acknowledgement

a radius of the helical pipe, m The authors gratefully acknowledge the Ministry of Chemi-
A area, m2 cal and Fertilizers, GOI, India for funding the project.
C constant in Eq. (6)
Cp specific heat, kJ/(kg K) References
d diameter of tube, m
h heat-transfer coefficient, W/m2 K Acharya, N., Sen, M., Cheng, H.-C., 1992. Experimental and numerical
H pitch, m investigation of heat transfer enhancement in coiled tubes by chaotic
k thermal conductivity, W/m K mixing. Ph.D. Dissertation, University of Notre Dam, Notre Dam, IN.
L length of heat exchanger, m Acharya, N., Sen, M., Cheng, H.-C., 1994a. Thermal entrance length and
Nusselt numbers in coiled tubes. International Journal of Heat and Mass
LMTD log-mean temperature difference, K or ◦ C Transfer 37, 336–340.
n exponent in Eq. (6) √ Acharya, N., Sen, M., Cheng, H.-C., 1994b. Analysis of heat transfer
NDe Dean number (=NRe / ) enhancement in coiled-tube heat exchangers. International Journal of Heat
NDe∗ modified Dean number, dimensionless and Mass Transfer 44, 3189–3199.
NN u Nusselt number Berger, S.A., Talbot, L., Yao, L.S., 1983. Flow in curved pipes. Annual
Review of Fluid Mechanics 15, 461–512.
NP r Prandtl number
Castelain, C., Berger, D., Legentilhomme, P., Mokrani, A., Peerhossaini,
NRe Reynolds number (=U0 dh /) H., 2000. Experimental and numerical characterisation of mixing in a
p pressure, N/ m2 steady spatially chaotic flow by means of residence time distribution
q heat-transfer rate, J/s measurements. International Journal of Heat and Mass Transfer 43,
Rc radius of the coil, m 3687–3700.
Chagny, C., Castelain, C., Peerhossaini, H., 2000. Chaotic heat transfer for
T temperature, K
heat exchanger design and comparison with a regular regime for a large
T1 temperature difference at inlet K range of Reynolds numbers. Applied Thermal Engineering 20, 1615–1648.
T2 temperature difference at outlet K Dean, W.R., 1927. Notes on the motion of fluid in a curved pipe. Philosophical
u0 inlet velocity, m/s Magazine 4, 208–233.
2396 V. Kumar et al. / Chemical Engineering Science 62 (2007) 2386 – 2396

Dean, W.R., 1928. The stream line motion of fluid in a curved pipe. Nauman, E.B., Thakur, R.K., Vial, Ch., Nigam, K.D.P., Djelveh, G., 2003.
Philosophical Magazine 5, 673–695. Static mixers in the process industries. Chemical Engineering Research &
Duchene, Ch., Peerhossaini, H., Michard, P.J., 1995. On the velocity field Design 81, 787–826.
and tracer patterns in a twisted duct flow. Physics of Fluids 7, 1307–1317. Peerhossaini, H., Le Guer, Y., 1992. Effect of curvature plane orientation on
Jones, S.W., Thomas, O.M., Aref, H., 1989. Chaotic advection by laminar vortex distortion in curved channel flow. In: Andereck, C.D., Hayot, F.
flow in twisted pipe. Journal of Fluid Mechanics 209, 335–357. (Eds.), Ordered and Turbulent Patterns in Taylor-Couette Flow. Plenum
Kumar, V., Nigam, K.D.P., 2005. Numerical simulation of steady flow fields Press, New York, pp. 263–272.
in coiled flow inverter. International Journal of Heat and Mass Transfer Rennie, T.J., Raghavan, G.S.V., 2005. Experimental studies of a double-
48, 4811–4828. pipe helical heat exchanger. Experimental Thermal and Fluid Science 29,
Kumar, V., Saini, S., Sharma, M., Nigam, K.D.P., 2006. Pressure drop and heat 919–924.
transfer study in tube-in-tube helical heat exchanger. Chemical Engineering Saxena, A.K., Nigam, K.D.P., 1984. Coiled configuration for flow inversion
Science 61, 4403–4416. and its effect on residence time distribution. A.I.Ch.E. Journal 30,
Mishra, P., Gupta, S.N., 1979. Momentum transfer in curved pipes. I. 363–368.
Newtonian fluids. Industrial & Engineering Chemistry Process Design and Saxena, A.K., Nigam, K.D.P., 1986. Residence time distribution in straight
Development 18, 130–137. and curve tubes. In: Cheremisinoff, N.P. (Ed.), Encyclopedia of Fluid
Mokrani, A., Castelain, C., Peerhossaini, H., 1997. The effects of chaotic Mechanics, vol. 1. Gulf Publishing, USA, p. 675.(Chapter 22).
advection on heat transfer. International Journal of Heat and Mass Transfer Shah, R.K., Joshi, S.D., 1987. Convective heat transfer in curved ducts.
40, 3089–3104. In: Kakac, S., Shah, R.K., Aung, W. (Eds.), Handbook of Singlephase
Nandakumar, K., Masliyah, J.H., 1986. Swirling flow and heat transfer in Convective Heat Transfer. Wiley Interscience, New York, pp. 5.1–5.46.
coiled and twisted pipes. In: Mujumdar, A.S., Masliyah, R.A. (Eds.), Shchukin, V.K., 1969. Correlation for experimental data on heat transfer in
Advances in Transport Process, vol. 4. Wiley Eastern. curved pipes. Thermal Engineering 16, 72–76.
Nauman, E.B., Kothari, D., Nigam, K.D.P., 2002. Static mixers to promote White, F.M., 1984. Heat Transfer. Addison-Wesley Publishing Company Inc.,
axial mixing. Chemical Engineering Research & Design 80, 1–5. New York, NY.

You might also like