You are on page 1of 11

Available online at www.sciencedirect.

com

Experimental Thermal and Fluid Science 32 (2007) 560–570


www.elsevier.com/locate/etfs

Heat transfer and pressure drop correlations of microchannel


heat exchangers with S-shaped and zigzag fins
for carbon dioxide cycles
Tri Lam Ngo, Yasuyoshi Kato *, Konstantin Nikitin, Takao Ishizuka
Research Laboratory for Nuclear Reactors, Tokyo Institute of Technology, 2-12-1-N1-2, O-okayama, Meguro-ku, Tokyo 152-8550, Japan

Received 14 March 2007; received in revised form 20 June 2007; accepted 22 June 2007

Abstract

A new microchannel heat exchanger (MCHE) with S-shaped fins was developed using the three-dimensional computational fluid
dynamics (3D CFD) FLUENT code. The MCHE provided 6–7 times lower pressure drop while maintaining heat-transfer performance
that was almost equivalent to that of a conventional MCHE with zigzag fins. This study was done to confirm the simulation results of
thermal-hydraulic performance using a supercritical carbon dioxide loop, and to propose empirical correlations of Nusselt numbers
and pressure-drop factors for a new MCHE with S-shaped fins and a conventional one with zigzag fins. This study is also intended to
confirm the independence of Pr obtained in the previous study by widely varying Pr from 0.75 to 2.2. Experimental results show that
the pressure-drop factor of the MCHE with S-shaped fins is 4–5 times less than that of MCHE with zigzag fins, although Nu is 24–
34% less, depending on the Re within its range. The Nusselt number correlations are expressed, respectively as NuS-shaped fins = 0.1740
Re0.593Pr0.430 and Nuzigzag fins = 0.1696 Re0.629Pr0.317 for the MCHEs with S-shaped and zigzag fins, and their pressure-drop factors
are given as fS-shaped fins = 0.4545 Re0.340 and fzigzag fins = 0.1924 Re0.091. The Nu correlation of the MCHE with S-shaped fins reproduces
the experimental data of overall heat transfer coefficients with a standard deviation (1 sigma) of ±2.3%, although it is ±3.0% for the
MCHE with zigzag fins. The calculated pressure drops obtained from pressure-drop factor correlations agree with the experimental data
within a standard deviation of ±16.6% and ±13.5% for the MCHEs with S-shaped and zigzag fins, respectively.
 2007 Elsevier Inc. All rights reserved.

Keywords: Microchannel heat exchanger; Heat transfer; Pressure drop; Supercritical CO2; Recuperator

1. Introduction (PCHE) which is a kind of micro channel heat exchanger


(MCHE). Typically, PCHEs are manufactured using two
Carbon dioxide is a non-toxic, non-flammable, and technologies of chemical etching and diffusion bonding
inexpensive natural fluid. Carbon dioxide cycles have been [4]. Flow channels are etched chemically into metal plates;
applied to automobile air conditioners, hot water supply, subsequently, etched plates are stacked and produced as
and gas turbine reactors [1,2]. Recuperator operating with one block by diffusion bonding. In contrast to typical plate
high-pressure, high-temperature-resistance, and high-heat- fin heat exchangers, a very small flow channel or micro-
transfer performance recuperators are the most crucial channel pattern is readily producible. Eqs. (1) and (2) show
components for a supercritical carbon dioxide (S-CO2) that reduction of the hydraulic diameter engenders a
gas turbine cycle [3]. A promising candidate for the high- decreased active length L or heat exchanger size at the same
efficiency recuperator is a printed circuit heat exchanger Colburn j factor, Pr, and NTU conditions [5].
Dh 2=3
j¼ Pr NTU; ð1Þ
* 4L
Corresponding author. Tel.: +81 3 5734 3065; fax: +81 3 5734 2959.
E-mail address: kato@nr.titech.ac.jp (Y. Kato). where NTU is the number of thermal units, and

0894-1777/$ - see front matter  2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermflusci.2007.06.006
T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570 561

Nomenclature

A surface heat transfer area, m2 Dt wall thickness, mm


Ac free flow area, m2 U overall HTC, W/m2K
Cp specific heat, J/kgK V wetted volume, m3
Dh hydraulic diameter, mm W flow rate, kg/h
f pressure-drop factor
FG geometric factor Greek symbols
h local HTC, W/m2K h fin angle
H enthalpy D difference
j Colburn factor q density, kg/m3
k thermal conductivity, W/mK l viscosity, kg/ms
L active flow length, mm
NTU number of thermal units Subscripts
Nu Nusselt number ave average value
P pressure, MPa calc calculated value
Pr Prandtl number cold, hot cold, hot side
Q heat load, W expt experimental value
Re Reynolds number in, out inlet, outlet
T temperature, C LMTD log-mean temperature difference

NTU ¼ ðT out  T in Þ=DT LMTD : ð2Þ study has been applied for evaluation of the overall heat
transfer coefficient at a region of near-critical or near-
The diffusion bonding technology maintains the parent pseudo-critical temperature. Empirical correlations of heat
material strength because of the lack of flux, braze, or filler transfer and pressure drop performance for these MCHEs
in the heat exchanger core, which provides high corrosion are proposed and have been compared with empirical cor-
and temperature resistance. relations obtained in past studies.
A new microchannel heat exchanger (MCHE) with S-
shaped fins was developed for recuperators of S-CO2 gas
turbine nuclear reactors using the three-dimensional com- 2. Experimental facility
putational fluid dynamics (3D CFD) FLUENT code. This
MCHE provided 6–7 times lower pressure drop while Figs. 1 and 2 show an S-CO2 loop that was built to mea-
maintaining almost equivalent heat-transfer performance sure the thermal-hydraulic performance of MCHEs, which
compared to that of a conventional MCHE with zigzag fins includes the MCHEs test sections.
[6]. It was applied as recuperators of supercritical CO2 gas
turbine reactors [7].
The previous study was done to confirm the simulation 2.1. Supercritical carbon dioxide experimental loop
results of thermal-hydraulic performance experimentally
using an S-CO2 loop, and to investigate empirical correla- Fig. 1 shows that carbon dioxide from the low-pressure
tions of Nusselt numbers and pressure-drop factors for the tank is compressed by an oil compressor. The lubricant oil
new MCHEs with S-shaped fins and a conventional one type of the compressor is the model of TCS340/4-D, man-
with zigzag fins. The experiments were carried out at the ufactured by OFFICINE MARIO DORIN, Italy. High-
inlet temperature conditions of 280 C for the hot side pressure gas is heated using an electric heater 1 (1.9 kW)
and 108 C for the cold side; for that reason, the Prandtl after passage through the oil separators. The residual oil
number (Pr) was within the narrow range of content was confirmed as less than 0.5 wt. ppm through
0.75 < Pr < 1.04. From these experiments, Pr dependence the following two-stage oil separation. First-stage oil sepa-
was not found for the Nusselt number (Nu) and pressure- ration was provided with fine filter elements composed of
drop factor (f) correlations [8]. The results also showed that micro-fibers (dimensions: 94 cm ; · 308 cm H, and 0.01 l
the pressure drop of an MCHE with S-shaped fins is 4–5 filtration efficiency), which achieves a filtration efficiency
times less than that of MCHE with zigzag fluid channels, of 1.5 wt. ppm. The second stage of oil separation was done
although the Nu is 40% less. by absorption of activated charcoal (GG10/20, Kurare
This study was carried out to confirm, experimentally, coal; Kurare Co. Ltd.). The influence of carry-over oil to
the Pr independence from Nu and f, widely varying Pr heat transfer and pressure drop was judged to be negligibly
from 0.75 to 2.2 for the both MCHEs with S-shaped and small because no time dependence was found on the ther-
zigzag fins. The integral method developed in the previous mal-hydraulic performance and no oil was observed on
562 T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570

Chiller Cooler 2 ΔP

FR PT PT
MCHE
Warmer PT PT
BPCV ΔP

Cooler 1
Oil return Expansion
path valve
FR

Bypass
Compressor Oil Heater 1 valve Heater 2
separators
Fig. 1. Flow diagram of supercritical CO2 experimental loop.

used. The thermocouples were inserted just at the inlet


and outlet of the MCHE’s distributors. The flow meters’
accuracy was ±5% over the full range of 0–150 kg/h. All
parameters, including inlet/outlet temperatures and pres-
sures, flow rates, and pressure drops for both sides were
recorded automatically with a 0.5 Hz sampling frequency.
The data acquisition was started when the respective
fluctuations of temperature, pressure, and flow rate were
confirmed to be well within measurement accuracy for at
least 10 min. The temperatures, pressures, and pressure
drops were used directly in further calculations, whereas,
the flow rate was corrected according to the measured
CO2 temperature and pressure at the flow meter.
Fig. 2. Photograph of supercritical CO2 experimental loop.
2.2. Test sections

the heat transfer surface when the test MCHE was cut after Two types of MCHEs with S-shaped fins and zigzag fins
the experiments. were tested in the S-CO2 loop, which were designed to have
That heated CO2 is forwarded through a turbine-type the same free flow area, hydraulic diameter, and fin angle,
flow meter and cooler 1 to the low-temperature (cold; but were different mainly in flow channel configuration:
high-pressure) side of the MCHE. The heater 1 and cooler one has S-shaped fins and the other has a zigzag fin config-
1 are used to attain the required temperature test condi- uration, as shown, respectively in Fig. 3a and b. Both
tions. The CO2 gas pressure from the outlet cold side is MCHEs were produced from 316L stainless steel using
reduced using a flow rate control expansion valve and the two technologies of chemical etching and diffusion
transferred to an electric heater 2 (2.2 kW) to heat CO2 bonding, by which the fluid channel is processed by chem-
gas to the required temperature test conditions before flow- ical etching on stainless steel plates. Then the plates were
ing through the high-temperature (hot; low-pressure) side stacked together in one block, as shown in Fig. 4a, to form
of MCHE. After exiting the MCHE, the CO2 gas temper-
ature is adjusted using the gas cooler 2 and chiller/warmer
to achieve a suitable temperature when it is returned to the
low-pressure tank.
The CO2 inlet pressure of the MCHE was measured θ
wf
using a pressure gauge transducer with accuracy of
wf pfy
±0.25% over the full range of 13 MPa. The pressure drop θ
was measured using a differential pressure gauge with accu- gf
pfy gf
racy of ±0.15% over the full range of 200 kPa. For temper- pfx p x
f
ature measurement, T-type (Copper/Constantan)
thermocouples with accuracy of ±0.5 C at the range of
40 C–125 C were installed at the cold side inlet and S-shaped fins Zigzag fins
hot side outlet of the MCHE. For hot side inlet and cold Fig. 3. Fluid channel configurations of MCHEs. pxf ¼ finpitch in
side outlet, K-type (chromel/alumel) thermocouples with x direction; pyf ¼ fin pitch in y direction. wf = fin width; gf = fin gap; h =
accuracy of ±1.5 C at range of 40 C to 375 C were fin angle.
T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570 563

29

745.2

76

Stacking of hot and cold plates (not in scale)

76

hot
cold
hot
hot
cold
29 hot
hot
cold
hot
hot
cold
hot

Cross-sectional view (not in scale)


Fig. 4. Configuration of the MCHE; unit: mm ‘‘hot’’ = hot side plates; ‘‘cold’’ = cold side plates.

a heat exchanger core with required heat capacity. Each wool in total. Fig. 5 shows that ribbon heaters were
MCHE includes eight hot plates and four cold ones inserted into the ceramic wool thermal insulator. During
arranged in a sandwiched configuration, as illustrated in experimental runs, these heaters were controlled automati-
Fig. 4b: a so-called double banking model. The detailed cally, as described below.
specification parameters are listed in Table 1. The test section was divided into six regions along the
The MCHEs were covered thermal insulator materials flow direction. One set of four thermocouples were
with heaters in order to minimize heat loss from their sur- mounted on the surfaces of the test sections, and another
faces to the ambient environment. They consist of 80 mm set of four thermocouples were installed at the outer sur-
thick ceramic wool, 50 mm rock wool, and 30 mm glass face of the glass wool in each region, as shown in Fig. 5.

Table 1
Specification parameters of MCHEs
Items Side S-shaped fin Zigzag fin
MCHE dimensions, mm – H745.2 · W76 · D29
Number of plates Hot/cold 8/4
Plate thickness, mm Hot/cold 1.5/1.5
Number of channels Hot/cold 96/44
Fin pitch pf in x–y axis, mm Hot/cold 7.565–3.426/7.565–3.426
Fin angle, h,  Hot/cold 52/52
Fin width, wf, mm Hot/cold 0.8/0.8
Fin depth, df, mm Hot/cold 0.94/0.94
Fin gap, gf, mm Hot/cold 1.31/1.31
Hydraulic diameter, Dh, mm Hot/cold 1.09/1.09
Free flow area, Ac, m2 Hot/cold 11.82 · 105/5.42 · 105
Heat transfer area, A, m2 Hot/cold 0.5099/0.2559 0.4653/0.2353
564 T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570

Thermo-couples
Ribbon heater
(Thickness: 4 mm)
Ceramic wool
(Total thickness: 80 mm)
MCHE
Rock wool
Glass wool (Total thickness: 50 mm)
(Thickness: 30 mm)

Fig. 5. Cross-sectional view of MCHE with insulator materials.

Maximum capacity is 200 W each for the heaters located at cific heat in the cold side is a nonlinear change with temper-
the three upstream hot regions, and 100 W each for heaters ature near pseudo-critical according to each pressure of
located at the three downstream regions. The ribbon heater experimental conditions shown in Table 3; a large differ-
is switched on to raise Tins to Ttest when the temperature of ence was apparent compared to that of the specific heat
the test section surface Ttest is higher than temperature of on the hot side, as illustrated in Fig. 6. Therefore, the mass
the insulator Tins. These ribbon heaters are controlled by flow rate through the cold side should be well adjusted to
the PID control system. The heat loss from the surfaces obtain different temperatures between the hot outlet–cold
of the test section becomes zero in principle; for that rea- inlet side and the cold outlet–hot inlet side of at least about
son, the heat loss is neglected in evaluation of the heat 2 C, using a bypass controllable valve, as shown in Fig. 1.
transfer performance.
3. Experimental data analysis
2.3. Experimental conditions
3.1. Integral method
The experiments were performed in a previous study [8]
under the experimental conditions described in Table 2. A logarithmic mean heat transfer equation, like that
These conditions provided a wide range of Re, given in Eq. (3), is commonly used in heat exchanger design
3.0 · 103 < Re < 2.0 · 104, although the range of Pr was [9].
limited: 0.75 < Pr < 1.04. The experiments showed no Pr Q ¼ UAF G DT LMTD ; ð3Þ
dependence on Nu. To confirm the Pr independence, addi-
tional experiments were carried out under the conditions therein, Q is the overall heat transfer rate of the exchanger,
shown in Table 3, which widened the Pr range to U is the overall heat transfer coefficient, and DTLMTD is the
0.75 < Pr < 2.2. logarithmic mean temperature difference (LMTD) for the
The MCHEs were tested for a wide range of conditions two fluids defined as for the case of a counterflow:
using these two sets of experimental data. Note that great ðT hot;in  T cold;out Þ  ðT hot;out  T cold;in Þ
care is necessary when making tests near critical or DT LMTD ¼ : ð4Þ
ln ½ðT hot;in  T cold;out Þ=ðT hot;out  T cold;in Þ
pseudo-critical regions, because thermo-properties of CO2
change drastically with temperature. For example, the spe- The dimensionless geometric factor FG is taken as unity for
the both MCHEs because of the fluid velocity increase at
distributors, which results in local heat transfer
Table 2
Previous experimental conditions augmentation.
However, because of a large temperature dependence on
Inlet pressure (MPa) Cold side 2.2 2.5 2.7 3.0 3.5
Hot side 6.5 7.5 8.5 9.5 10.5 constant-pressure specific heat of CO2 around its pseudo-
Inlet temperature (C) Cold side 108 critical temperature, as shown in Fig. 6, the temperature
Hot side 280 difference between the hot and cold side shows nonlinearity
Flow rate (kg/h) Hot side 30–85 kg/h with 5 kg/h increment against its heat load, as illustrated in Fig. 7, for a typical

Table 3
Additional experimental conditions
Inlet pressure (MPa) Cold side 7.7 8.0 8.5 9.0 10.0 11.0 12.0
Hot side 6.0
Inlet temperature (C) Cold side 35 38 41 44 47 51 55
Hot side 120
Flow rate (kg/h) Hot side 40–150 kg/h with 10 kg/h increment
T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570 565

100 Then, integration of the left-hand side by Q and right-hand


P=6.0 MPa side by A of Eq. (5) gives the following equation in an inte-
P=7.7 MPa gral form:
Specific heat, Cp [kJ/(kgK)]

P=8.0 MPa Z Q0 Z A0
P=8.5 MPa dQ
P=9.0 MPa ¼ U ðAÞdA: ð6Þ
0 DT ðQÞ 0
P=10.0 MPa
10 P=11.0 MPa When an average overall heat transfer coefficient U is de-
P=12.0 MPa fined as
Z A0
1
U¼ U ðAÞdA; ð7Þ
A0 0
the following equation is obtained after manipulation of
1
0 20 40 60 80 100 120
Eqs. (6) and (7).
Temperature, T [°C]
Z Q0
1 dQ
U¼ : ð8Þ
Fig. 6. Change of specific heat with temperature. A0 0 ðT hot ðQÞ  T cold ðQÞÞ
Eq. (8) can be rewritten introducing the non-dimensional
geometric factor FG as
16
Q0
Temperature difference, T [°C]

14 U¼ ; ð9Þ
A0 F G DT ave
12
where the average temperature difference DTave is
10
Q0
8
DT ave ¼ R Q0 dQ
: ð10Þ
0 ðT hot ðQÞT cold ðQÞÞ
6
The heat load, Q0, is the average heat loads of hot and cold
4
side at each test condition. The heat load is calculated from
2 the mass flow rate, W, and different enthalpy of outlet,
Hout, and inlet Hin in each side, as given in Eq. (11).
0
0 500 1000 1500 2000 2500 3000 3500 4000 Qhot=cold ¼ W hot=cold ðH hot=cold;out  H hot=cold;in Þ: ð11Þ
Hot side Inlet Hot side Outlet
Cold side Outlet Heat Load, Q [W]
Cold side Inlet
Thermo-physical properties of CO2 were taken from the
Fig. 7. Temperature difference as a function of heat load. PROPATH database package [10]. On computation of
the temperature, Re, Pr, etc. distributed along MCHEs,
condition, which is given as case 1 shown in Table 4. For the pressure is assumed to change linearly with the heat
that reason, the logarithmic mean temperature difference load.
method is no longer adequate to evaluate the overall heat
transfer coefficients around its pseudo-critical temperature.
3.2. Derivation of Nusselt number correlation
Therefore, the logarithmic mean rate equation is modi-
fied to an integral form using what we call an integral
The local heat transfer coefficient h is given as
method. The methodology is described below.
An amount of heat exchange dQ in an infinitesimal kNu
h¼ : ð12Þ
region dA of a heat exchanger can be written from Eq. Dh
(3) as
The Nusselt number is usually described as a function of
dQ Re and Pr, which are given for both hot and cold sides as
dQ ¼ U DT dA or ¼ U ðAÞdA;
DT ðQÞ Nuhot ¼ C hot  Remhot1 Prnhot
1
; and ð13Þ
where DT ¼ T hot  T cold : ð5Þ Nucold ¼ C cold  Remcold
2
Prncold
2
; ð14Þ

Table 4
Typical experimental conditions of the MCHE with S-shaped fins
Case Cold side Hot side
Tin (C) Tout (C) Pin (MPa) DP (kPa) W (kg/h) Tin (C) Tout (C) Pin (MPa) DP (kPa) W (kg/h)
1 48.27 111.59 10.04 22.04 88.57 120.42 51.10 6.04 35.45 151.54
2 34.98 118.04 7.75 7.06 34.60 120.38 37.88 5.998 2.59 51.66
566 T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570

where n1, m1, Chot, n2, m2, and Ccold are constants. Using for the further analyses of heat transfer and pressure drop
the local heat transfer coefficient and hot side heat transfer performance.
area, Ahot, an overall heat transfer coefficient, U calc
hot , of a Eq. (16) includes one dependent variable ðU calc
hot Þ and six
heat exchanger is written as independent variables (khot, kcold, Rehot, Prhot, Recold,
1 Prcold) with six searched parameters (Chot, Ccold, m1, n1,
U calc
hot ¼ 1 Ahot Dtwall Ahot
; ð15Þ m2, n2), which are solvable using the following least-squares
hhot
þ Awall k wall
þ Acold hcold method:
where Dtwall is fluid-channel-wall thickness and kwall is the X
N expt
1  calc 2
thermal conductivity of the wall. 2
U hot i  U expt
hot i ! minimum: ð21Þ
i¼1
xU i
From Eqs. (12)–(15) the following equation is obtained:
k wall Awall Acold k hot k cold C hot C cold Remhot1 Prnhot
1
Remcold
2
Prncold
2 Therein, Nexpt is the number of experimental data, and xU i
U calc
hot ¼ ; denotes the uncertainty of the overall heat transfer coeffi-
X þY þZ
cient for the ith experimental datum, which was calculated
ð16Þ
numerically as
where, vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 9  
uX oU i 2
X ¼ k wall Awall k cold Acold Dhot m2 n2
xU i ¼ t x2Rg ; ð22Þ
h C cold Recold Prcold ;
g¼1
oR g
Y ¼ Dtwall Ahot Acold k hot k cold C hot C cold Remhot1 Prnhot
1
Remcold
2
Prncold
2
; and
m1 n1
Z¼ k wall Awall k hot Ahot Dcold
h C hot Rehot Prhot : where Rg is the gth measured parameter (namely tempera-
tures, pressures at the inlet and outlet, pressure drops and
Average values were used for the Re, Pr, thermal conduc- flow rate) and xRg represents the corresponding measure-
tivity, and density of the fluids in Eqs. (13)–(16), (24), ment uncertainty.
(25). Averaging was taken from the inlet to the outlet, as
defined by a set of Eq. (17). 3.3. Derivation of pressure-drop factor correlation
Z Q0
WDh 1 1
Re ¼ dQ; Eq. (23) shows that the pressure drop of fluids through
Ac Q0 0 lðQÞ surfaces are given in terms of the pressure-drop factor by
Z Q0 Z Q0
1 1 lðQÞC p ðQÞ neglecting the small contributions of pressure drop around
Pr ¼ PrðQÞdQ ¼ dQ; ð17Þ
Q0 0 Q0 0 kðQÞ the inlet and outlet structures:
Z Q0 Z Q0
1 1 2LW 2
k¼ kðQÞdQ; q ¼ qðQÞdQ: DP ¼ f: ð23Þ
Q0 o Q0 o Dh A2c q
The average thermo-physical values defined by Eq. (17) Eq. (24) shows that the pressure-drop factor f is a function
were calculated in the following manner. It is assumed that of Re.
the pressure, P, in MCHE changes linearly with the heat
load, Q, from the inlet to the outlet. P changes from the in- f ¼ BReb : ð24Þ
let (Q = 0) to the outlet (Q = Q0) for the hot side as Combining Eqs. (23) and (24), Eq. (25) is obtained.
Q 2LW 2
P hot ðQÞ ¼ P in
hot  DP hot : ð18Þ DP calc ¼  BReb : ð25Þ
Q0
Dh A2c q
Then, the fluid enthalpy, H, is calculated from the inlet to
The coefficients B and b were found using the following
the outlet as a function of Q.
least-squares method:
Q
H hot ðQÞ ¼ H in X 1  i 2
N
hot  DH hot : ð19Þ expt
Q0 DP  DP i
! minimum; ð26Þ
calc expt
i¼1
x2DP i
Subsequently, the fluid temperature, T, is calculated from
the inlet to the outlet as a function of Q using the FTPH where xDP i is uncertainty of the pressure drop measurement
function of the PROPATH database. The FTPH function for the ith experimental datum.
reconstructs T from P and H.
4. Results and discussion
T hot ðQÞ ¼ F TPH ½P hot ðQÞ; hhot ðQÞ: ð20Þ
Finally, knowing Phot(Q) and Thot(Q), the thermo-physical 4.1. Empirical correlations of the Nusselt number
properties of CO2 (viscosity, l; specific heat, Cp; heat con-
ductivity, k; and density, q) are calculated as a function of The fluid channels have the same geometrical configura-
Q using PROPATH database. The average thermo-physi- tions for both hot and cold sides. Therefore, the Nu corre-
cal values are calculated from Eq. (17), which are used lations are assumed to be identical in hot and cold sides:
T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570 567

C hot ¼ C cold ¼ C; 1000 Zigzag fin MCHE


0.629±0.009 0.317±0.014
m1 ¼ m2 ¼ m; ð27Þ 900
Nu = (0.1696±0.0144)Re Pr
n1 ¼ n2 ¼ n:
800
+6.0% (2σ)
The three parameters (C, m, n) were determined simulta-

Ucalc, [W/(m K)]


700
neously using the least-squares method as given in Eq. -6.0% (2σ)

2
(21); the following empirical correlations were obtained 600
for the MCHEs with S-shaped and zigzag fins.
500
S-shaped fin:
400
Nu ¼ ð0:1740  0:0118ÞRe0:5930:007 Pr0:4300:014 ;
300
3:5  103 < Re < 2:3  104 ; 0:75 < Pr < 2:2: ð28Þ
300 400 500 600 700 800 900 1000
Zigzag fin: 2
Uexpt, [W/(m K)]
Nu ¼ ð0:1696  0:0144ÞRe0:6290:009 Pr0:3170:014 ;
Fig. 9. Calculated OHTCs vs. the experimental values for zigzag fin type.
3:5  103 < Re < 2:2  104 ; 0:75 < Pr < 2:2: ð29Þ
Eqs. (28) and (29) show that the Nu depends on the power
Table 5
indices of Pr of 0.430 and 0.317, respectively, whereas, the Uncertainty analysis of overall heat transfer coefficients at the typical
Nu correlation obtained by Nikitin [8] or Ishiyama [11] is experimental conditions
independent of Pr. These respective independencies might Case 1 Case 2
be ascribed to the narrow range of Pr in their experiments, oU
oW xW ; ð%Þ 5.0 5.0
e.g., 0.75 < Pr < 1.04. The Nu correlation (Eq. (28)) of the oU
MCHE with S-shaped fins reproduces the experimental oT hot;in xT hot;in ; ð%Þ 3.3 16.8
oU
data of the overall heat transfer coefficients with a standard oT hot;out xT hot;out ; ð%Þ 6.2 5.4
oU
deviation of ±2.3%, although it is ±3.0% (Eq. (29)) for the oT cold;in xT cold;in ; ð%Þ 7.1 9.9
oU
MCHE with zigzag fins. The calculated overall heat trans- oT cold;out xT cold;out ; ð%Þ 4.3 18.4
oU
fer coefficients vs. the experimental ones are plotted, oP hot;in xP hot;in ; ð%Þ 0.3 0.4
oU
respectively in Figs. 8 and 9 for MCHEs with S-shaped oP cold;in xP cold;in ; ð%Þ 0.2 1.6
and zigzag fins. oU
oDP hot xDP hot ; ð%Þ 0.003 0.0006
Uncertainty analysis was done for the overall heat trans- oU
oDP cold xDP cold ; ð%Þ 0.002 0.0005
fer coefficients of the MCHE with S-shaped fins in the two xU (%) 11.97 27.85
typical cases summarized in Table 4. Case 1 is represented
for the case of low uncertainty of the overall heat transfer
coefficient, and case 2 is for the high one. The nine compo- heat transfer coefficient uncertainties of case 1 and 2 are
nents in the uncertainties of these two cases were calculated calculated as 11.97% and 27.85%, respectively. The experi-
using Eq. (22); the results are shown in Table 5. The overall mental data uncertainties are similar for both MCHEs. The
uncertainties of overall heat transfer coefficients were esti-
mated as approximately from 12% to 28%.
S-shaped fin MCHE
800 0.593±0.007 0.430±0.014
The heat exchangers with S-shaped fins and zigzag fins
Nu = (0.1740±0.0118)Re Pr
both provide very high thermal efficiency. For that reason,
700 the temperature difference between the hot side and the
cold side at the inlet and the outlet is rather small, espe-
600
+4.6% (2σ) cially at the low flow rate, as in case 2. Therefore, errors
Ucalc, [W/(m K)]

in the temperature measurements give rise to such large


-4.6% (2σ)
2

500 uncertainties of 28% in the overall heat transfer coeffi-


cients. However, actual accuracy in the temperature mea-
400 surement may be expected to be a few times higher than
the guaranteed one. Consequently, the uncertainties may
300 be expected to be a few times lower than the uncertainties
estimated above for the overall heat-transfer coefficients.
200
200 300 400 500 600 700 800
2 4.2. Empirical correlations of pressure-drop factor
Uexpt, [W/(m K)]

Fig. 8. Calculated OHTCs vs. the experimental values for S-shaped fin The dependence of the pressure-drop factor on Re is
type. inferred to be same for hot and cold sides of each MCHE
568 T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570

because of their identical fin patterns. The coefficients B S-shaped fins. This behavior is explainable that the S-
and b in Eq. (25) were found by solving the least-squares shaped fin MCHE provides a very small pressure drop;
method described as Eq. (26) for the MCHE with S-shaped the error of pressure drop measurement increases in the
and zigzag fins, respectively. The results are given below. pressure differential gauges at this region.
S-shaped fin: The following correlations were obtained for the hot
and cold sides of the MCHEs with S-shaped zigzag fins
f ¼ ð0:4545  0:0405ÞRe0:3400:009 ;
when the pressure-drop factors were correlated separately
3:5  103 < Re < 2:3  104 : ð30Þ for the hot and cold sides of the MCHEs. S-shaped fin
Zigzag fin: MCHE:

f ¼ ð0:1924  0:0299ÞRe0:0910:016 ; Hot side : fhot ¼ ð0:6070  0:0779ÞRe0:3720:014 ;

3:5  103 < Re < 2:2  104 : ð31Þ 3:5  103 < Re < 2:0  104 : ð32Þ
0:2420:0201
Cold side : fcold ¼ ð0:1769  0:0346ÞRe ;
Figs. 10 and 11 show that the pressure-drop factor correla-
3 4
tions of Eqs. (30) and (31) predict the experimental pres- 6:5  10 < Re < 2:3  10 : ð33Þ
sure drops with standard deviations of ±16.6% and Zigzag fin MCHE:
13.5% for the MCHEs with S-shaped and zigzag fins,
respectively. Fig. 10 shows that a ‘‘bumpy’’ behavior Hot side : fhot ¼ ð0:3390  0:0285ÞRe0:1580:009 ;
appears at the low pressure drops of the MCHE with 3:5  103 < Re < 2:0  104 : ð34Þ
0:1540:036
Cold side : fcold ¼ ð0:3749  0:1293ÞRe ;
3 4
S-shaped fin MCHE 6:0  10 < Re < 2:2  10 : ð35Þ
40 -0.340±0.009
f = (0.4545±0.0105)Re
The pressure drops calculated by the correlation equations
were compared to experimental ones for the hot and cold
30
+33.2% (2σ) sides of MCHEs with S-shaped and zigzag fins in Figs.
Δ Pcalc, [kPa]

12–15. The standard deviation of calculated pressure drops


20 from experimental ones is reduced to ±9–11.2%.
-33.2% (2σ)
4.3. Comparison with past correlations
10
In this section, the respective thermal-hydraulic perfor-
mances of MCHEs with S-shaped and zigzag fins are com-
0
0 10 20 30 40 pared together and to those of other heat exchangers in
terms of Nu and pressure-drop factor.
Δ Pexpt, [kPa]
The Nusselt number and pressure-drop factor correla-
Fig. 10. Calculated pressure drop vs. experimental ones for S-shaped fin tions of a high-performance plate fin heat exchanger
MCHE. (PFHE) type for recuperator model (core width · heigh-
t · length = 100 · 100 · 100 mm) with ultra fine off-set fins

40
200 Zigzag fin MCHE
-0.372±0.014
f = (0.1924±0.0299)Re
-0.091±0.016 fhot = (0.6070±0.0779)Re

160 30
+27% (2σ) +19% (2σ)
Δ Pcalc, [kPa]

Δ Pcalc [kPa]

120
20
-19% (2σ)
-27% (2σ)
80

10
40

0 0
0 40 80 120 160 200 0 10 20 30 40
Δ Pexpt, [kPa] Δ Pexpt [kPa]

Fig. 11. Calculated pressure drop vs. experimental ones for zigzag fin Fig. 12. Calculated pressure drop vs. experimental ones of the hot side for
MCHE. S-shaped fin MCHE.
T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570 569

(fin height · thickness · off-set-length · pitch = 1.2 · 0.2 ·


-0.241±0.020
40 fcold = (0.1769±0.0346)Re 5.0 · 1.6 mm) were obtained by Ishiyama [11] as follows.
Ishiyamað2001Þ : Nu ¼ 0:0645Re0:693 ; ð36Þ
+18% (2σ) 0:402
30 f ¼ 0:512Re
Δ Pcalc [kPa]

5:0  102 < Re < 3:0  103 : ð37Þ


20 -18% (2σ)
Dittus and Boelter derived the Nu correlation for a circular
tube type heat exchanger for a wide range of Re of
10 2 · 103 < Re < 1.5 · 105 [12].
Nu ¼ 0:026Re0:8 Pr0:3 ; for cooling; ð38Þ
0 0:8 0:4
0 10 20 30 40
Nu ¼ 0:024Re Pr ; for heating: ð39Þ
Δ Pexpt [kPa] Moreover, the pressure drop was measured on a conven-
tional PCHE with zigzag or corrugated flow channels at
Fig. 13. Calculated pressure drop vs. experimental ones of the cold side
for S-shaped fin MCHE. the Re range of 4 · 103 < Re < 105. The wavelength-to-
width ratio of channel is about 7 [13]. The pressure-drop
factor correlation suggested by Hesselgreaves [5] is given as
Oyakawað1989Þ : f ¼ 11:0Re0:53 ; 4  103 < Re < 105 :
-0.158±0.009
160 fhot = (0.3390±0.0285)Re ð40Þ

These correlations are shown, respectively, as plotted


+18.4% (2σ)
120 against the Re number in Figs. 16 and 17 for the Nu num-
ber and pressure-drop factor. Actually, Fig. 16 shows that
Δ Pcalc, [kPa]

-18.4% (2σ) the Nu number of the MCHE with S-shaped fins most clo-
80
sely resembles that provided by the Dittus–Boelter’s corre-
lations for the ranges of 3.5 · 103 < Re < 2.3 · 104. The
40 heat transfer performance of PFHE investigated by Ishiy-
ama is about 24.5% less than that of MCHE with S-shaped
fins, as shown in Fig. 16, although its pressure-drop factor
0 shows only a 59% difference, as illustrated in Fig. 17.
0 40 80 120 160
Fig. 13 also shows that the pressure drop factors of MCHE
Δ Pexpt, [kPa]
with zigzag fins are 12% higher than that reported by
Fig. 14. Calculated pressure drop vs. experimental ones of the hot side for Oyakawa in 1989 over the Re range.
the zigzag fin MCHE. In addition, Figs. 16 and 17 show that the Nu number of
MCHE with zigzag fins is 24–34% higher than that of the
S-shaped fins type, but the pressure-drop factor is, remark-
ably, four to five times larger depending on the Re
within its range. This difference is mainly attributable to
-0.154±0.036
200 fcold = (0.3749±0.1293)Re
110
100
160
+22.4% (2σ) 90
Nusselt number, Nu

80
Δ Pcalc, [kPa]

120
70
-22.4% (2σ) 60
80 50
40 S-shaped fin MCHE
40 30 Zigzag fin MCHE
Ishiyama's PFHE (2001)
20
Dittus-Boelter (1930) for heating
0 10
0 40 80 120 160 200 Dittus-Boelter (1930) for cooling
0
Δ Pexpt, [kPa] 0 5000 10000 15000 20000 25000
Reynolds number, Re
Fig. 15. Calculated pressure drop vs. experimental ones of the cold side
for the zigzag fin MCHE. Fig. 16. Variation of Nusselt number with Reynolds number.
570 T.L. Ngo et al. / Experimental Thermal and Fluid Science 32 (2007) 560–570

0.16 S-shaped fin MCHE data with the standard deviation of ±16.6% and 13.5% for
Zigzag fin MCHE the MCHEs with S-shaped and zigzag fins, respectively.
0.14 Ishiyama's PFHE (2001) Both MCHEs have higher heat-transfer performance
than either PFHE or circular tube heat exchangers. The
Pressure drop factor, f

0.12 Oyakawa's PCHE (1989)


Nusselt number of the MCHE with zigzag fins is 24–34%
0.10 higher than that of S-shaped fins, but the pressure-drop
factor (or pressure drop itself) is 4–5 times larger, depend-
0.08
ing on Re.
0.06
References
0.04

0.02 [1] Y. Kato et al., A carbon dioxide direct cycle with partial condensa-
tion for nuclear reactors, in: Proc. ICAPP-02, 2002.
0.00 [2] V. Dostal et al., A super-critical CO2 gas turbine power cycle for next-
0 5000 10000 15000 20000 25000 generation nuclear reactors, in: Proc. ICONE- 10, Arlington, TX 1-8,
Reynolds number, Re 2002.
[3] Y. Muto et al., Design of turbomachinery for supercritical CO2 gas
Fig. 17. Variation of pressure-drop factor with Reynolds number. turbine fast reactor, Paper 6094, in: Proc. ICAPP 2006, Reno, NV
USA, 2006.
[4] K. Nikitin et al., Printed circuit heat exchanger thermal-hydraulic
the difference in flow channel configuration. The continu-
performance in supercritical CO2 experimental loop, International
ous zigzag flow channel configuration generates swirl, Journal of Refrigeration 29 (2006) 807–814.
eddy, and recirculation flows around their bend corners, [5] J.E. Hesselgreaves, Compact Heat Exchanger, Selection, Design and
thereby resulting in large pressure drop. In contrast, the Operation, first ed., PERGAMON an imprint of Elsevier Science,
S-shaped-fin flow channel configuration causes neither 2001.
[6] N. Tsuzuki et al., High performance printed circuit heat exchanger,
swirl, eddies, nor recirculation flows [6]. This engenders less
in: Proc. HEAT-SET 2005, Grenoble, France, 2005.
pressure drop by a factor of four to five and a moderate [7] Y. Kato et al., Design of recuperator for the supercritical CO2 gas
degradation of local heat transfer. turbine fast reactor, Paper 5196, in: Proc. ICAPP 2005, Seoul, Korea,
2005.
5. Conclusions [8] K. Nikitin et al., Experimental thermal-hydraulics comparison of
microchannel heat exchangers with zigzag channels and S-shaped fins
for gas turbine reactors, in: Proc. of Fifteenth International Confer-
Thermal-hydraulic characteristics of microchannel heat ence on Nuclear Engineering, at Nagoya, Japan, on April 22–26,
exchangers (MCHEs) with S-shaped and zigzag fins were (2007) ICONE15-10826.
investigated experimentally by varying Pr widely from [9] W.M. Kays, A.L. London, Compact Heat Exchangers, 3rd ed.,
0.75 to 2.2 to confirm Pr dependence on heat transfer per- McGraw-Hill, New York, 1984.
[10] T. Ito et al., PROPATH: A Program Package for Thermo-physical
formance and pressure drop for carbon dioxide cycles.
Properties of Fluids Version 10.2, Corona Publishing Co. Ltd.,
The Pr dependence was found for the empirical correla- Tokyo, Japan, 1990.
tions of Nu and not for the pressure-drop factors of both [11] S. Ishiyama et al., Development of the compact heat exchanger for
MCHEs. The Nu correlations reproduce the experimental the HTGR, (II) heat transfer and fluid characteristics test, Journal of
data of the overall heat transfer coefficients with the stan- the Atomic Energy Society of Japan 43 (7) (2001).
[12] P.W. Dittus, L.M.K. Boelter, Heat transfer in automobile radiators
dard deviation of ±2.3% and ±3.0%, respectively, for the
of the tubular type, Univ. Calif. Pub. Eng. 12 (1985) 3–22.
MCHEs with S-shaped and zigzag fins. The calculated [13] K. Oyakawa et al., The effect of the channel width on heat-transfer
pressure drop obtained from the pressure-drop factor cor- augmentation in a sinusoidal wave channel, JSME International
relations agree with the experimental data of pressure-drop Journal Series II 32 (3) (1989) 403–410.

You might also like