You are on page 1of 18

International Journal of Heat and Mass Transfer 158 (2020) 119973

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


journal homepage: www.elsevier.com/locate/hmt

Experimental and numerical investigation of the effect of number of


parallel microchannels on flow boiling heat transfer
Gaurav Hedau a, Prasenjit Dey b, Rishi Raj c, Sandip K. Saha a,∗
a
Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai - 400076, India
b
Department of Mechanical Engineering, National Institute of Technology Goa, Goa - 403401, India
c
Thermal and Fluid Transport Laboratory, Department of Mechanical Engineering, Indian Institute of Technology Patna, India

a r t i c l e i n f o a b s t r a c t

Article history: A combined numerical and experimental investigation to elucidate the two-phase flow behaviour and
Received 16 January 2020 heat transfer during subcooled boiling of water in 1 × 1 cm2 footprint area heat sinks with six, ten,
Revised 6 May 2020
and fourteen parallel microchannels is performed. A three-dimensional, laminar, multi-phase, transient
Accepted 17 May 2020
numerical model is developed to simulate flow boiling in microchannels. This study is one of the first
Available online 30 June 2020
studies which reports a three-dimensional numerical simulation of flow boiling in a large area multiple
Keywords: parallel microchannels heat sink including the effect of inlet and outlet plenums. We show an excellent
Two-phase flow agreement in heat transfer and pressure drop data with experimental results on a heat sink with fourteen
Microchannel parallel microchannels over a wide range of applied heat flux spanning various boiling regimes. The re-
Parallel sults show that the vapour blocking in the channel near the outlet is primarily responsible for instabilities
Heat transfer coefficient and oscillations in the pressure drop, the surface temperature, and the mass flux. Intensified confinement
Instability
due to the decrease in the number of the channels for a constant fin width results in increased surface
temperatures. Similarly, increase in the number of the channels from six to fourteen improved heat trans-
fer significantly wherein a significant drop of 45.5 °C in surface temperature with little increase of 37%
in pressure drop is observed for a mass flux of 500 kg/m2 s and a heat flux of 220 W/cm2 . Microchannel
heat sink with fourteen channels demonstrates on an average nearly 240% higher heat transfer coefficient
in comparison to the heat sink with six channels. The numerical modelling framework used in this study
can be used to provide design guidelines for microchannel heat sinks.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction Many experimental and numerical studies have been carried


out to observe global and local characteristics of flow boiling in
The application of compact microchannel is found to be effec- various combinations of microchannel arrangements. The flow re-
tive in dissipating a high heat flux from a comparatively smaller sistance at the microchannel walls was observed to be increased
area using two-phase flow boiling in the microchannel by ab- with increasing vapour bubble size [3]. Also, the developed vapour
sorbing heat in the form of latent heat. The two-phase flow heat bubbles have significant effect on the blockage of fluid flow in
transfer phenomenon has many advantages over single-phase heat the microchannel. A numerical study on a nucleate boiling was
transfer in the microchannel, viz. low pumping power and uniform conducted by Jia et al. [4] by simulating two-dimensional geom-
surface temperature [1]. Owing to these significant qualities of the etry with the volume of fluid (VOF) method, including micro-
microchannel heat sink, it has several applications in electronics layer model, and modified height function model. The shape of
cooling, nuclear reactor and cooling of transistors and high-power the bubble predicted by the numerical method agreed with the
lasers. However, flow boiling in microchannel gives rise to vari- experimental results; however, the departure time of the bubble
ous problems, like instabilities, dry-out, high-pressure drop, fluc- predicted by the numerical model was higher than the experi-
tuations in parameters like pressure drop, temperature and mass mental observation. A similar kind of numerical study showing
flux [2]. the impact on heat transfer due to the bubble growth and bub-
ble coalesce in a microchannel was conducted by Ling et al. [5].
The three-dimensional simulation of bubble growth in the rect-
angular microchannel revealed that the thin liquid film between

Corresponding author. the vapour bubble and the wall increases the heat transfer rate.
E-mail address: sandip.saha@iitb.ac.in (S.K. Saha).

https://doi.org/10.1016/j.ijheatmasstransfer.2020.119973
0017-9310/© 2020 Elsevier Ltd. All rights reserved.
2 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

idated with analytical results for the analysis on interfacial insta-


Nomenclature bility. Ferrari et al. [11] numerically investigated the bubble dy-
namics in a single square microchannel during slug flow regime
A Footprint area, m2 for R245fa. The bubble velocity in the channel was found to be
Ch Number of microchannels more as compared to that in the circular cross-section channel for
E Enthalpy, J the same inlet conditions. However, the liquid film thickness in the
G Mass flux, kg/m2 .s square channel varied along the perimeter and was found to be
g Acceleration due to gravity, m/s2 one order magnitude lesser than that in the circular cross-section
H Height of the microchannel, m microchannel due to the thickening of liquid film at the corner and
h Heat transfer coefficient, W/m2 .K the thinning of liquid film at the face of the square microchannel.
hL Latent heat of vapourization, J/kg The growth and transport of a single bubble in a single square
k Thermal conductivity, W/m.K microchannel were studied numerically by Mukherjee and Kand-
L Length of the microchannel, m likar [12]. It was observed that the bubble growth rate was directly
P Pressure, Pa proportional to the degree of superheat of incoming fluid, whereas,
q Heat flux, W/cm2 inversely proportional to Reynolds number (Re). The movement of
S Source terms in momentum and energy equations the interface of the bubble was found in both directions; forward
T Temperature, K and reverse during the bubble growth. Further, the heat trans-
t Time, s fer and bubble growth rate during nucleate boiling were anal-
u
 Velocity vector, m/s ysed numerically and found to be increased when the channel size
w Width, m was smaller than the bubble departure diameter [13]. A three-
x, y,z Axes dimensional numerical simulation on bubbly flow in a single mi-
crochannel with rectangular cross-section was carried out by Zu
Greek Symbols
et al. [14]. To simulate the nucleation site for the bubble growth,
ρ Density, kg/m3
vapour was injected from the wall through a small hole (pseudo
μ Viscosity, kg/m.s
nucleate boiling). Flow patterns appear in a circular microchannel
α Volume fraction
of diameter 0.5 mm for R-134a and R-22 fluids at different param-
σ Surface tension, N/m
eters were numerically studied by Zhuan and Wang [15]. Authors
λ Curvature of the interface
showed the importance of the mechanism of bubble growth and
Subscripts bubble coalesce on the flow pattern and flow transition. The tran-
c Copper sition in flow pattern at high heat flux was found due to the in-
ch Microchannel crease in bubble generation frequency, which increases the bubble
eff Effective coalesce.
l Liquid The effect of the microchannel cross-section on the hydrody-
v Vapour namic and thermal performance of microchannel was numerically
s Wall Surface studied by Jing and He [16]. Authors compared the performance of
sat Saturation a single microchannel with rectangular, elliptical and isosceles tri-
sub Sub-cooled angle cross-section considering the effect of conjugate heat trans-
loc Local fer. For the same hydraulic diameter of the microchannel, the mi-
avg Average crochannel with triangular cross-section has the smallest hydraulic
resistance and the lowest heat transfer coefficient value as com-
pared to other two cross-sections. An experimental work along
Jafari [6] conducted a numerical study on flow boiling with an ar- with a three-dimensional numerical study was conducted by Luo
tificial cavity, which acts as a nucleation site inside the microchan- et al. [17] to investigate the thermal and hydraulic performance
nel. With the constant surface temperature, as inlet mass flux in- of annular flow regime. The thickness of liquid film between the
creases upto 64 kg/m2 s, the bubble generation frequency also in- channel wall and the interface decreases as heat flux increases.
creases. Magnini and Thome [7] simulated the slug flow regime Further, the heat transfer coefficient increases with the decrease
and analysed the flow dynamics inside a horizontal circular mi- in thickness of the thin liquid film at the wall, when the interfa-
crochannel. The thin liquid film between the vapour and the wall cial temperature becomes equal to the saturation temperature. For
results into the high heat transfer; however, as mass flux increases, the same heat flux, the increase in mass flux tends to decrease
the liquid film thickness increases, thereby decreasing the heat the heat transfer coefficient in the microchannel. Chiapero et al.
transfer coefficient. High frequency of bubble generation increases [18] numerically analysed flow boiling in two parallel microchan-
the heat transfer rate as the liquid slug zone reduces. nels. Pressure drop oscillations were evaluated and compared with
The transition of the slug to the annular flow regime was in- the pressure drop oscillations in a single microchannel. Due to the
vestigated by Liu et al. [8]. The microchannel having a diameter different pressure drop oscillations in the two parallel channels,
of 0.4 mm was used under high heat flux with a mass flux of one of the two parallel channels contains superheated vapour. As
400 kg/m2 s. The disturbance in the thermal boundary layer during a result, the outlet of that channel always shows high surface tem-
transition enhances the heat transfer in the microchannel. In their perature. This causes a large difference between the outlet surface
other study [9], the authors numerically demonstrated the coalesce temperature of these two channels resulting in thermal stresses in
of bubble and its effect on heat transfer inside the microchannel. the microchannel. This condition further augmented by maldistri-
The rate of evaporation of liquid into vapour was found to be more bution of coolant in the parallel channels during flow boiling.
during bubble sliding and merging as these are near to the wall, Extensive experimental studies are available on the transition
where the presence of superheated fluid enhances the heat trans- of flow regime, flow patterns instabilities and heat transfer in
fer. parallel microchannels. Qu and Mudawar [2] conducted an ex-
A two-dimensional numerical simulation on a single mi- perimental study on a 21 parallel copper microchannel. Hydrody-
crochannel was conducted by Guo et al. [10]. In this study, a model namic instability in the form of severe pressure drop oscillation
of annular flow regime under hydrodynamic unstable flow was val- and mild parallel channel instability was observed, which results in
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 3

high-pressure drop oscillations leading to premature CHF. These bles and different flow regimes, local wall superheat and local heat
instabilities can be reduced by using throttling the flow at the transfer coefficient. A numerical model is developed which is val-
upstream of the microchannel. Steinke and Kandlikar [19] stud- idated with the experimental results on a micro heat sink with
ied the thermal performance of six parallel copper microchannel fourteen channels. Further, the combined effect of the number of
with water as cooling fluid. Authors observed backflow of vapour parallel microchannels and heat flux on the average heat transfer
in the upstream direction of the liquid flow and dry-out of the mi- coefficient is investigated. The present numerical investigation of
crochannel. The heat transfer coefficient was observed to decrease flow boiling in parallel microchannels may likely to have potential
as quality increases. Experimental study on the flow boiling insta- implications toward designing of parallel multiple microchannels
bility was also conducted by Chen and Pan [20] in a fifteen parallel for efficient heat removal applications.
silicon microchannel with a hydraulic diameter of 86.3 μm. In sta-
ble flow boiling, the bubbly, slug and annular flow regime appear 2. Experimental setup
along the flow direction, whereas in unstable flow boiling, the slug
and annular flow regimes move in the upstream direction, which 2.1. Flow loop
causes fluctuations in pressure drop. For the stable flow, the bubble
grows exponentially along the flow direction, whereas, bubble os- A schematic of the closed-loop used for the experimental study
cillates in the unstable flow. Harirchein and Gerimella [21] studied of flow boiling inside parallel microchannels similar to earlier stud-
the confinement effect of flow regimes on the heat transfer coeffi- ies from the present authors [22–25] is shown in Fig. 1. The
cient in a microchannel using FC-77 as cooling fluid. The number loop consists of a degasifying unit, a peristaltic pump (Make: Cole
of parallel microchannels was varied from 2 to 61 by varying the Parmer), a dampener, a mass flow meter, a preheater, the test
width and hydraulic diameter of microchannel keeping the same section, a condenser, various power sources, pressure transducers,
footprint area of 12.7 × 12.7 mm. A new convective confinement thermocouples, and a data acquisition system. The degasifying unit
number, which depends on the channel dimensions, fluid proper- also acts as a reservoir for DI water used in the flow loop. The fluc-
ties and mass flux was proposed to show the confinement of the tuating output flow of the peristaltic pump is damped by a pulse
vapour bubble in the microchannel. A larger value of heat transfer dampener (Make: Masterflex, Model: HV-07596-20), which pro-
coefficient was observed for the convective confinement number vides a steady and constant mass flow rate. Coriolis mass flowme-
less than 160 due to thin-film evaporation inside the microchannel ter (Make: Micro Motion ELITE Coriolis Flow and density meter,
as compared to the value of heat transfer coefficient during uncon- Model: CMF010M) measures the mass flow rate in the flow loop.
fined flow and nucleate boiling. Heat generation in the preheater is controlled by a DC power
It can be noted from the aforementioned studies that most of source, which keeps the inlet temperature to the microchannel at
the numerical studies on flow boiling in the microchannel were the desired value. An air-cooled condenser connected between the
carried out on a single microchannel with an initially developed test section and the degasifying unit is used to cool the working
single bubble. In addition, the numerical studies were also mostly fluid before it is circulated back in the flow loop. The pressure
conducted on a single microchannel for bubbly flow regime and drop across the microchannel is measured by a differential pres-
slug flow regime. However, in practical application scenarios, mul- sure transducer (Make: Omega, Model: PXM459-350HDWUI). The
tiple parallel microchannels are found more efficient to carry heat surface and fluid temperatures are measured using T-type ther-
from the hot surface than the single microchannel [21]. A very few mocouples (Make: TC Ltd.). These measurements of pressure drop,
numerical studies are available on the three-dimensional flow boil- temperature and mass flow rate are acquired using a data acquisi-
ing in parallel microchannels [12–17], which are mainly focused on tion system (National Instruments cRio-9066). A high-speed cam-
the interaction of generated vapour bubble in parallel microchan- era (Phantom Miro M110) is used to capture videos of flow boiling
nels; however, the inlet and outlet plenums were not considered inside the microchannel.
in the previous numerical studies. It can be noted that the in-
let and outlet plenums significantly affect the flow distribution, 2.2. Microchannel assembly
thereby heat transfer in the multiple channels. Thus, the present
study aims to investigate the flow boiling phenomena in multi- A copper block with dimensions 30 × 30 × 50 mm
ple parallel microchannels with inlet and outlet plenums to ob- (width × length × height) is used. A top neck section of
tain detailed insight and characteristics of the development of bub- 10 × 10 × 10 mm, as shown in Fig. 2, is machined to fabricate

Fig. 1. Experimental flow loop.


4 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 2. Microchannel test section (All dimensions are in mm).

a microchannel with fourteen number of channels. The microchan- T1 , T2 and T3 are placed 2 mm below the top surface along the
nel is machined by a Computer Numeric Controlled (CNC) micro- channel length with T2 at the centre, as shown in Fig. 2(c). Ther-
milling machine on the top 10 × 10 mm2 footprint area of the mocouples T4 and T5 are placed below T2 in the vertical direction
neck portion of the copper block. The channel dimensions are kept 3 mm apart (Fig. 2(c)). These five thermocouples are used to cal-
as 315 ± 15 μm (width), 600 ± 10 μm (depth) and 10 ± 0.1 mm culate surface temperature and heat flux reaching the bottom of
(length). This heating unit is inserted inside a Teflon block having the microchannel. Two more thermocouples are placed to measure
dimensions of 70 × 70 × 50 mm (width × length × height) with a the inlet and the outlet temperatures of water/ steam used as the
cavity to accommodate the copper block (Fig. 2(a)). The dumbbell- working fluid.
shaped inlet and outlet plenums with depth same as that of the
microchannel (600 μm) are machined on the Teflon block (Fig.
2(b)). The dumbbell shaped plenum provides equal mass flow rate
2.3. Data reduction and uncertainty
to all the parallel microchannels for the single-phase heat trans-
fer [26]. The transparent polycarbonate cover is placed at the top
The temperature readings from the thermocouples located in
of the Teflon-copper block assembly for visualisation using the
the vertical direction (T2 , T4 and T5 ) are used to calculate the
high-speed camera. A very thin layer of silicone gel is applied be-
heat flux reaching the base of the microchannel assuming one-
tween the polycarbonate sheet and the assembly to make it leak
dimensional heat conduction, given by Fourier law of heat conduc-
proof. The polycarbonate sheet is secured on the copper-Teflon as-
tion as follows [22–25].
sembly using nut and bolt systems. Inlet and outlet ports to in-
let and outlet plenums, respectively, are drilled inside the polycar- dT
bonate sheet. A DC cartridge heater of 250 capacity is inserted q = −kc (1)
dx
in a hole having a diameter of 6.35 mm and a depth of 26 mm
drilled at the bottom surface of the copper block. The whole where kc is the thermal conductivity of copper, dT is the temper-
setup is covered with glass wool to reduce the heat loss to the ature difference between two thermocouple readings and dx is the
ambient. distance between thermocouples. In experiments, the heat input
Two pressure measurements port are machined 2 mm before to the cartridge heater is varied from 30 to 240 W in steps. Due to
and after the microchannels inside the Teflon block. Five holes are heat loss to the ambient, the corresponding heat flux value varies
drilled having 1 mm diameter and 5 mm depth for inserting T-type from 23 to 220 W/cm2 (footprint area (A) of 1 cm2 ). The uncer-
thermocouples, which are used to measure the temperature at five tainty related to measurements and derived parameters is provided
locations in the copper block as shown in Fig. 2(c). Thermocouples below in Table 1.
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 5

Fig. 3. Schematic diagram of the microchannel (a) with inlet and outlet plenums and (b) top view (All dimensions are in mm).

Table 1 6, 10 and 14 keeping the footprint area of the combined channels


Uncertainty values of different parameters.
and fins constant at 1 × 1 cm2 as exhibited in Fig. 3(b). The ra-
Parameters Uncertainty tio of fin width (wf ) and microchannel width (wch ) is varied to
T type thermocouple ±0.2 °C obtain the different number of microchannels for a fixed height
Differential pressure transducer ±0.25% of full scale (±0.875 mbar) of 600 μm and a length of 10 mm. The geometrical parameters
Absolute pressure transducer ±0.25% of full scale (±0.0086 bar) of the microchannel configurations along with the mass and heat
Mass flux ±0.25% of flow rate (±25 kg/m2 .s) fluxes considered in the numerical study are tabulated in Table 2.
Heat flux (q") ±2%
Dumbbell shaped inlet and outlet plenums at both the ends of
the microchannel are incorporated for supplying and collecting the
working fluid to and from the microchannels (Fig. 3(b)). Both the
3. Numerical model plenums are maintained at the same level as of the bottom sur-
faces of the microchannel. The inlet and outlet ports are located at
3.1. Definition of physical problem a distance of 5 mm from the microchannel inlet and outlet, respec-
tively and the diameter of the ports is kept constant at 1.5 mm for
In the present numerical investigation, heat sinks with parallel all the cases.
microchannels and plenums at the inlet and the outlet are con-
sidered, as shown in Fig. 3(a). Most of the numerical studies con- 3.2. Mathematical formulation
sist of 1 or 2 microchannels [5–15,18]. For the footprint area of
1 × 1 cm2 , microchannel with fourteen number of channels (with Subcooled flow boiling in the parallel microchannels, as shown
320 μm width) shows less thermal resistance of the heat sink [27]. in Fig. 4, is numerically investigated. Subcooled water-liquid is
Therefore, the effect of the number of microchannels on flow boil- considered as the primary phase and water-vapour as the sec-
ing heat transfer is studied by varying the number of channels as ondary phase. The formation of vapour bubbles due to the
6 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 4. Schematic representation of the numerical domain with boundary conditions.

Table 2
Geometrical parameters of microchannels and mass and heat fluxes considered in the numerical study.

Case Number of channels Fin width (μm) Channel width (μm) Height of channel (μm) Mass flux (kg/m2 s) Heat flux (W/cm2 )

1 6 400 1250 600 500 23


2 6 400 1250 600 500 110
3 6 400 1250 600 500 220
4 10 400 580 600 500 23
5 10 400 580 600 500 110
6 10 400 580 600 500 220
7 14 400 320 600 500 220
8 14 400 320 600 500 110
9 14 400 320 600 500 23
10 14 400 320 600 400 23
11 14 400 320 600 250 23

evaporation of liquid coolant is modelled by a source term. The tions of all phases are written as,
governing equations based on mass, momentum and energy con-
∂ αl S
servation for two-phase unsteady flow are as follows. + ∇ . (u
 αl ) = − E (5)
∂t ρl
Continuity equation:
∂ αv S
+ ∇ . (u
 αv ) = − E (6)
∂ρ ∂t ρv
+ ∇ . (ρ u
) = 0 (2)
∂t
The sum of all phase volume fractions is always 1, i.e.,
Momentum equation:
αl + αv = 1 (7)
∂ρ u   
+ ∇ . (ρ u
u ) = −∇ P + ∇ . μ ∇ u
 + ∇u
 T + ρ g + Sm By solving either Eqs. (5) or (6), the volume fraction of other
∂t phase can be calculated from Eq. (7).
(3) The bulk properties (∅) of the fluids in the governing equations
are then calculated based on the phase volume fraction weighted
Energy equation:
values and is expressed as,
∂ρ E  
+ ∇ .[u
 (ρ E + P )] = ∇ . ke f f ∇ T + SE (4) ∅ = αl ∅l + αv ∅v (8)
∂t
where, ∅ can be ρ , μ and keff . Similarly, enthalpy, E in energy equa-
where ρ is the volume-averaged density, u  is velocity, P is pres- tion (Eq. (4)) is calculated based on the mass averaged values of
sure; μ represents the viscosity, and E is enthalpy. keff denotes the both phases as,
effective thermal conductivity.
A source term is introduced in the momentum equation, Sm and αl ρl El + αv ρv Ev
E= (9)
another one is introduced in the energy equation, SE , which repre- αl ρl + αv ρv
sent the surface tension force and change in phase, respectively.
The momentum source term, Sm in Eq. (3) represents sur-
The volume of fluid method (VOF) [28] is used to track the inter-
face tension, which is evaluated per unit volume by using the
face between the phases, which is accomplished by the solution
continuum-surface-force model as [29],
of the continuity equation for the volume fraction (α ) of one (or
more) of the phases. The continuity equations for the volume frac- Sm = σ λn (10)
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 7

where n is a unit vector normal to the interface that can be deter- Table 3
Grid independency test for microchannel with fourteen microchannels at
mined from,
q = 220 W/cm2 and G = 500 kg/m2 s.
∇∅ T (K) P (mbar)
n= (11) Total number of cells Deviation (%) Deviation (%)
|∇ ∅|
750,000 18.63 - 63.62 -
and λ is the curvature of the interface that can be calculated from, 950,000 19.81 5.98 59.9 6.21
1,150,000 19.98 0.84 59.29 1.01

λ = ∇ .n (12)
Lee [30] developed a mass transfer model concerning the pro- the heat flux is applied at the bottom wall at a distance of 100 μm
cess of evaporation and condensation. In their model, phase change from the bottom surface of the microchannel heat sink. This dis-
was assumed to occur at constant pressure and a quasi-thermal tance is chosen based on the locations of T1 , T2 and T3 thermo-
equilibrium state, and the mass transfer was mainly dependent on couples and the one-dimensional heat conduction assumption. The
the saturated temperature. The Lee model is applied herein to con- extra thickness of 100 μm is considered to reduce the computa-
sider the phase change process as described below, tional time and cost. Three different values of heat flux are adopted
T − T  as, q = 23, 110, and 220 W/cm2 . All other surfaces are kept adi-
m˙ s = re αl ρl
sat
, Evaporation when T ≥ Tsat (13) abatic. The thermal conditions at the solid-liquid interface are set
T
as,
T − T   
m˙ s = rc αv ρv
sat
, Condensation when T < Tsat (14) ∂ Tl  ∂ Ts 
T −kl = −ks (16)
∂ n inter f ace ∂ n inter f ace
where m˙ s is the rate of mass transfer per unit volume (kg/m3 .s).
In this investigation, the interfacial temperature is assumed at the Tl |inter f ace = Ts |inter f ace (17)
saturation temperature. For this purpose, the coefficients re and rc
are tuned and considered as at 10 0 0 s−1 in order to numerically Eqs. (16–17) ensure continuity of heat flux and temperature
maintain the interface temperature within Tsat ± 2 K. Excessively across the solid-liquid interface.
large values of re and rc cause a numerical convergence problem,
while very small values lead to a significant deviation between the 3.4. Grid independence study
interfacial temperature and the saturation temperature.
The source term, SE in Eq. (4) takes care of the changes in en- The grid sensitivity test is conducted for numerical accuracy
ergy during the phase transformation process. This source term is and computational time. A trade-off between the low and high
expressed as the product of latent heat of vapourisation (hL ) and number of grids is required to achieve the numerical accuracy with
the rate of mass transfer (m˙ s ) as, minimum computational costs. In order to study the grid indepen-
dency, three grid sizes are considered, and the relative error of
SE = m˙ s hL (15) wall superheat (T), and the pressure drop (P) is summarised
in Table 3 for 14 parallel microchannels at q = 220 W/cm2 and
3.3. Numerical methodology and Boundary conditions G = 500 kg/m2 s. It can be observed from the table that the per-
centage deviation of T and P between the grids, 950,0 0 0 and
The numerical analysis is carried out in a commercial Finite 1,150,0 0 0 is 0.84% and 1.01%, respectively, which is insignificant
Volume Method (FVM) based solver ANSYS Fluent V16. The vol- compared to the percentage deviation of T and P between the
ume of fluid (VOF) method is used to model the two-phase flow grids, 750,0 0 0 and 950,0 0 0, that is 5.98% and 6.21%, respectively.
boiling in the parallel channels. The governing equations, which Thus, a grid of 950,0 0 0 is considered for the present numerical
are in the partial differential form, are solved by applying the con- analysis for better accuracy with the least computational cost.
trol volume-based collocated grid approach. The laminar viscous
model is chosen to represent the low Reynolds number flow con- 4. Results and discussion
dition. PISO (pressure implicit with the splitting of operators) al-
gorithm is implemented for coupling pressure and velocity field. A The experimental and numerical analyses are carried out for
pressure-based coupled approach is found to yield improved sta- multiple parallel microchannels at the mass flux of G = 250 and
bility as against the segregated approach for the pressure–velocity 500 kg/m2 s and the heat flux is varied as, 23 ≤ q ≤ 220. Multi-
coupling, and hence this approach is used along with the PRESTO ple parallel microchannels in the heat sink are formulated in two
(pressure staggering option) scheme for pressure term discretisa- ways. In numerical analysis, fin width is kept constant at 0.4 mm,
tion [31]. The unsteady term is resolved using the first-order im- and the microchannel width is varied to design three different par-
plicit scheme, where a very small-time step (10−8 s) is used. This allel microchannel sets, viz. 6, 10 and 14 as given in Table 2. Three
time step is chosen to avoid numerical instability and to obtain different conditions of heat flux, q = 23, 110 and 220 W/cm2 are
convergence of the numerical code. The geometric reconstruction considered in the present numerical study.
scheme represents the interface between fluids using a piecewise-
linear approach is considered for volume fraction calculation. The 4.1. Validation of the present numerical model
QUICK (quadratic upstream interpolation for convection kinematic)
scheme is used to discretise the momentum and energy equa- Before proceeding to the numerical analyses, it is important to
tions. The solution is considered to be converged once the resid- validate the present numerical model with the experimental study.
uals reach 10−6 for the continuity and momentum equations, and The numerical domain and initial and boundary conditions con-
10−9 for the energy equation. sidered for the validation study are the same as the microchan-
The numerical study in this work is focused on the boiling flow nel and prevailed conditions in the experiments. The comparison
of sub-cooled liquid in parallel microchannels with inlet and out- of the experimental and numerical pressure drops (P) and wall
let plenums. At the entry section in the inlet plenum, sub-cooled superheat (T = Ts - Tsat , where, Ts = average surface temperature
liquid water with Tsub = 20 C is allowed to enter with different and Tsat = saturation temperature) is depicted in Fig. 5 for the heat
mass flux ranges, 250 – 500 kg/(m2 s). In the numerical simulation, sink with fourteen microchannels at the mass fluxes of G = 250
8 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 5. Validations of the numerical and experimental results of (a) pressure drop (P) and (b) temperature difference between surface temperature and inlet temperature
(T) for heat sink with fourteen microchannels.

and 500 kg/m2 s. Three different values of heat flux are considered nels is shown in Fig. 6. It is found that the flow remains in the
for validation as, q = 23, 110 and 220 W/cm2 . It can be noted single-phase condition at low values of heat flux, 23 and 42 W/cm2
that the numerical simulations are performed for a certain time for the mass fluxes of 250 and 500 kg/m2 s, respectively. During
after which no significant deviation is observed on P and T, i.e. the bubbly flow, a large number of vapour bubbles appears near
the time of dynamically steady-state condition. However, this time the outlet of the microchannel. It can be noticed from the Fig. that
of dynamically steady-state condition is found different for differ- flow regime cycle exists within the range of heat flux, 109 ≤ q ≤
ent working conditions considered in the numerical analysis as the 137 W/cm2 for the mass flux of 500 kg/m2 s, whereas this range
mass flux, heat flux and number of microchannels have notewor- of heat flux for 250 kg/m2 s is 62 ≤ q ≤ 137 W/cm2 . Flow regime
thy influence on the flow boiling characteristics. The surface aver- cycle occurs due to unstable annular flow regime at moderate heat
age value is then calculated for all the quantitive data presented in flux as the surface temperature is not sufficient to maintain the
the numerical analysis as the local properties are less dependent flow regime. When the heat flux is further increased above 137
on time after attaining the dynamically steady-state condition. The W/cm2 for both the mass flux conditions, the flow regime becomes
discrepancy between the experimental and numerical P is shown annular, with thin liquid film at the wall and vapour core at the
in Fig. 5(a), and it is observed that the present numerical data is in centre of the microchannel, as shown in the inset view of flow vi-
good agreement with the experimental results. A maximum devia- sualisation with the schematic of annular flow regime in Fig. 6.
tion of 13% is found between the experiment and numerical pres- The reduction in fluctuations in temperature and mass flux also
sure drops. In addition, heat transfer characteristic is also validated confirms the presence of annular flow regime. In this flow regime,
with the experimental results of the heat sink with fourteen mi- heat transfer is mainly due to convection through thin liquid film
crochannels. The comparison between the numerical and experi- maintained at the walls.
mental T at different heat flux and mass flux conditions is de- In addition, different two-phase flow regimes obtained exper-
picted in Fig. 5(b). The surface temperature at the bottom surface imentally (top view) and numerically (side view- vapour fraction
of the microchannel is recorded in both numerical and experimen- contour and top view – vapour isosurface) in the heat sink with
tal cases. The surface temperature of microchannel decreases with fourteen microchannels for the mass flux of 250 kg/m2 s and heat
increasing mass flux, and more heat is convected by the working flux of 220 W/cm2 are shown in Fig. 7. Volume fractions of vapour
fluid. As a result, the temperature difference, T decreases. It is and liquid phases are indicated by the red and the blue colours,
observed that the numerical temperature data agree well with the respectively in the figure. In general, a small amount of bubble
experimental values with a maximum deviation of 16%. nucleates and grows on the wall and adheres to the wall initially
(Fig. 7(a)). Frequent nucleating bubbles, which appear concurrently
from several nucleating sites, grow. Nucleation of the bubbles oc-
4.2. Flow characteristics
curs at the wall when its temperature is higher than the satura-
tion temperature of the fluid. In Fig. 7(a), at the time of t = 350
4.2.1. Two-phase flow pattern
ms, several nucleating bubbles appear on the wall, except at the
Flow boiling heat transfer is known to depend on the type of
entrance. Large bubbles nucleate on the bottom surface and side-
flow regime present in the channel along the length, which is a
wall due to the application of heat flux on that surface. The rate of
function of input heat flux and mass flux for a particular num-
nucleating bubble increases along the flow direction due to phase
ber of microchannels. As heat flux increases for a specific mass
change. Initially, few bubbles are nucleated at very low wall su-
flux, different flow regimes appear as single-phase liquid, bub-
perheat (T). It can be observed in Fig. 7(b) that at t = 500 ms,
bly/slug flow regime, annular flow regime with partial dry-out. For
the overall size of the bubbles is increased; however, these bub-
a better understanding of the effect of flow regimes on the boil-
bles adhere to the microchannel wall, which is due to rise in the
ing heat transfer, experimental visualisation of flow regimes along
bulk fluid temperature, which increases evaporation rate compared
with the boiling curve of the heat sink with fourteen microchan-
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 9

Fig. 6. Flow regimes at different heat flux for the heat sink with fourteen microchannels [25].

to the condensation rate of the vapour [13]. At t = 650 ms, the only three different values of heat flux are considered as q = 23,
bubble growth rate is quite high, and the size of the bubble is 110 and 220 W/cm2 . As discussed earlier for G = 500 kg/m2 s (Fig.
larger. It is also observed that few nucleating bubbles are formed 5(a)), the numerically obtained P is also found in good agree-
at the upper surface of the microchannel. This is due to the fact ment with the experimental results for G = 250 kg/m2 s for all
that as the interface area of the bubble grows, the rate of vapour the considered heat flux conditions. It is observed that the P in-
formation increases and a thin liquid film, which is trapped be- creases with increasing heat flux. When vapour generation starts
tween the bubble interface and the channel walls, starts to evap- within the microchannels at a certain heat flux condition, two fric-
orate. At this stage, most of the bubbles at the bottom surface tional components are presented at the microchannels surfaces,
grow to the size of the microchannel. This behaviour of bubbles viz. one component of the liquid phase and another one is for the
affects the flow and heat transfer in the microchannel as these developed vapour phase. The formation of vapour increases with
bubbles try to block the flow passage [6]. In Fig. 7(d), at t = 850 increasing heat flux, and both two-phase frictional and accelerating
ms, the bubble grows more, and the bubble moves in the flow components of pressure drop increases [32]. As a result, the over-
direction, and it occupies the entire height of the microchannel all pressure drop increases with the increase in heat flux. Also, it
and blocks the flow passage. Moreover, the bubble coalesces with is found that P increases with increasing mass flux. At the higher
each other with a large number of nucleate bubbles at the top sur- mass flux, the flow velocity is higher, resulting in more frictional
face of the microchannel. At t = 1 s, it is observed that the bub- resistance within the microchannels. However, a discrepancy is ob-
ble elongates due to bubble coalescence along the flow direction served in the range of heat flux, 62 ≤ q ≤ 81 W/cm2 , where P
and it occupies the larger space within the microchannel. Further, is less for the mass flux of 500 kg/m2 s than the 250 kg/m2 s. As
the vapour is observed at the outlet section of the microchannels. discussed earlier, within this heat flux range, the vapour genera-
Also, the nucleation of the bubble is found more near the inlet at tion rate is more at the lower mass flux due to the presence of the
t = 1 s compared to t = 350 ms as the difference between the bulk flow regime cycle. Whereas, for the same heat flux range, the bub-
fluid and the saturation temperature of water increases with time. bly flow regime exists at the mass flux of 500 kg/m2 s, resulting in
Moreover, the occupancy of microchannel flow passage by bubbles a lesser two-phase frictional resistance.
restricts the fluid movement in the downstream direction, which
results in the instability and flow reversal of coolant during flow 4.2.3. Effect of mass flux and number of parallel microchannels
boiling in the microchannel. The influences of mass flux and the number of parallel mi-
crochannels at a constant heat flux condition are analyzed numeri-
4.2.2. Pressure drop cally and are depicted in Fig. 8. Three different values of mass flux
Experimental and numerical pressure drops (P) for the heat are considered as G = 250, 40 0 and 50 0 kg/m2 s for 14 parallel
sink with fourteen microchannels at the mass fluxes of G = 250 microchannels at q = 23 W/cm2 . As discussed earlier, it is found
and 500 kg/m2 s are shown in Fig. 5(a). The heat flux is varied in Fig. 8(a) that the P increases with increasing mass flux. Also,
as 23 ≤ q ≤ 220 W/cm2 . However, for numerical investigation, the variation of P with the number of parallel microchannels is
10 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 7. Flow patterns in the heat sink with fourteen microchannels with side view (numerical- red represent volume fraction contour) and top view experimental and
numerical (blue represent vapour isosurface) for 250 kg/m2 s and 220 W/cm2 at different time instants of (a) 350 ms, (b) 500 ms, (c) 650 ms, (d) 850 ms and (e) 1 s.

shown in Fig. 8(b) at G = 500 kg/m2 s and q = 220 W/cm2 . The [2], and a similar characteristic is also observed in the present nu-
number of micro-channels is varied as 6, 10 and 14 by changing merical study. When the number of parallel microchannels is in-
the microchannel width as 320, 580 and 1250 μm, respectively, creased, the pressure drop increases as the acceleration effect of
while the fin width is kept constant at 400 μm. It can be seen the generated vapour becomes pronounced [33]. Also, the effect
that the P increases with the increasing channel number. The P of bubble confinement and occupancy of flow passage by elon-
is inversely proportional to the microchannel hydraulic diameter gated bubbles is more in the smaller width microchannels [34].
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 11

Fig. 7. Continued

Fig. 8. Pressure drop variation with heat flux for varying (a) mass flux and (b) number of microchannels.

However, the two-phase frictional pressure drop is less significant 4.3. Heat transfer characteristics
over the accelerated pressure drop as at the same mass flux and
heat flux, the smaller channel results in the higher vapour qual- 4.3.1. Boiling curve
ity, and at the higher vapour quality, the contribution of frictional The variations of experimental and numerical wall superheat
pressure drop decreases because of the influence of viscosity de- (T) with the applied heat flux for two mass fluxes of 250 and
creases. For mass flux of 500 kg/m2 s and heat flux of 220 W/cm2 , 500 kg/m2 s for the heat sink with fourteen microchannels are dis-
the pressure varies nonlinearly with the increase in the number played in Fig. 5(b). The corresponding flow regimes can be ob-
of parallel channels from 6 to 14 keeping width of fin constant. served in Fig. 6. It is perceived that the wall superheat, T in-
The heat sink with fourteen microchannels shows 37% increment creases with increasing applied heat flux, q . The same behaviour
in pressure drop with a difference of 16.3 mbar as compared to is also noticed in the present numerical analysis and is shown in
the heat sink with six parallel microchannels due to more cross- Fig. 5(b) for three different heat flux values, q = 23, 110 and
sectional area of 6 channels which gives less pressure drop. 220 W/cm2 . The average surface temperature of the microchannel
12 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 9. Variation of surface superheat with heat flux with varying number of parallel microchannels (keeping mass flux as constant).

mainly depends on the flow regimes, applied heat flux, mass flux W +W
plane at z = ch 2 f (right view) at t = 500 ms. The behaviour of
and number of microchannels on a given footprint area. As seen surface temperature is significantly affected by the mass flux, heat
in the figure, there is no temperature undershoot for all the mass flux and number of parallel microchannels. Thus, different mass
fluxes, and boiling initiates at very low wall superheat (< 8 K). The flux conditions, G = 250, 400 and 500 kg/m2 s for q = 23 W/cm2
onset of nucleate boiling (ONB), where the slope of curve changes and Ch = 14, different heat flux conditions, q = 23, 110 and
due to the change in flow regime from the single-phase liquid to 220 W/cm2 for G = 500 kg/m2 s and Ch = 14, and different number
the bubbly regime, is observed experimentally at the heat flux of of parallel microchannels, Ch = 6, 10 and 14 for q = 220 W/cm2
42 and 62 W/cm2 for G = 250 and 500 kg/m2 s, respectively. It can and G = 500 kg/m2 s are considered and are shown in Fig. 10(a –
be noticed both experimentally and numerically that the surface c), Fig. 10(c – e) and Fig. 10(e – g), respectively. It is observed that
temperature decreases with increasing mass flux, as at the higher for all the considered cases, the temperature near the inlet of the
mass flux, more fluid interacts with the hot surface and takes away microchannel is less and increases along the flow direction within
more heat with it. Also, the vapour extraction capability improves the microchannels. This is because of the sub-cooled liquid enters
due to higher inertial effect of the incoming fluid at the higher the microchannels through the inlet plenum where more heat is
mass flux [35]. removed from the wall through convection by the sub-cooled liq-
The variation of wall superheat, and the number of parallel uid. As discussed earlier, in Fig. 7(a, b) that near the inlet, the
microchannels with heat flux at the mass flux of 500 kg/m2 s is size of the bubble is small and also very few. These few bubbles
shown in Fig. 9. The effect of channel width is found to be sig- are the result of condensation of vapourised fluid in the subcooled
nificant. For the same operating conditions, q = 220 W/cm2 and bulk fluid that flows at a distance from the hot wall. The corre-
G = 500 kg/m2 s, the wall superheat increases non-linearly with sponding temperature distribution can be observed in Fig. 10. The
a decreasing number of parallel microchannels. Due to the con- flow region is superheated near the outlet, where bubbles grow
finement effect of the microchannels, flow conditions can be dif- rapidly and coalesce to form larger slug bubbles. In order to bet-
ferent even at the similar heat flux and mass flux condition [36]. ter visualise the transverse distribution of temperature within the
As discussed earlier, the flow regime becomes annular with a thin microchannel, temperature contour on the x-y plane located at the
liquid film at the microchannel surfaces and vapour core at the mid of the centre channel is depicted in Fig. 10 (right view). The
centre of the microchannel at the higher heat flux (Fig. 6). This bulk temperature in the top region of the microchannel is mini-
liquid film thickness increases with increasing microchannel width mum. Significant fluctuation in the temperature can be observed
resulting in higher thermal resistance. As a result, the liquid film in Fig. 10(e-g). When the fluid temperature is higher than the sat-
evaporation rate is less in the larger width channels, and the sur- uration temperature, vapour bubble nucleates and gradually grows
face temperature becomes more. For all the considered mass flux and then lifts off from the hot surface [37]. The newly created va-
conditions in the present numerical analysis, the single-phase flow cant spaces are then occupied by the liquid where the temperature
is observed at the heat flux of q = 23 W/cm2 . At the mass flux lower than the temperature at which bubbles lifts off due to the
of 500 g/m2 s and the heat flux of 220 W/cm2 , the difference of mass conservation. The minimum bulk fluid temperature is found
wall superheat flow between the 6 and 14 parallel microchannels for three cases of mass flux (250, 400 and 500 kg/m2 s) for the heat
is found to be 45.5 °C. sink with fourteen microchannels at q = 23 W/cm2 for each con-
sidered mass flux condition in the present numerical analysis. The
4.3.2. Temperature distribution intermediate value of the bulk temperature is noticed for Ch = 14
Fig. 10(a – g) shows the temperature distribution at two differ- at G = 250 kg/m2 s and q = 110 W/cm2 . The maximum bulk
ent planes, viz. (i) x-z plane at y = 100 μm (left view) and (ii) x-y
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 13

Fig. 10. Temperature (in K) distribution at the base of the parallel microchannel (left view, x-z plane at y = 100 μm) and at the middle microchannel (right view, x-y plane
W +W
at z = ch 2 f ) at 500 ms for different heat flux, mass flux and microchannel width conditions.

temperature is observed in the heat sink with six parallel mi- in a few cases. As discussed earlier, the bubbles are nucleated and
crochannels for G = 500 kg/m2 s, q = 220 W/cm2 . move along the flow direction. During the movement, the bubbles
It is observed that the thickness of superheated layer increases and slugs enhance the wall shear stress, thereby increasing the
with increasing heat flux for Ch = 14 at a constant mass flux of heat transfer from the wall to the bulk flow region. The bubbles
G = 500 kg/m2 s (Fig. 10(c) - (e)). As a result, the degree of wall disturb the thermal boundary layer and drag the low momentum
superheat also increases, which escalates the rapid development fluid into the bulk flow region. In this case, convective heat trans-
of bubbles on the hot wall, thereby resulting in the formation of fer is enhanced. In microchannel flow boiling, bubble evolution in-
larger bubbles. It is also found that bubble behaviour is signifi- duces flow pattern transitions along the channel from the bubbly
cantly influenced by non-uniform temperatures in the fluid. When flow regime to the slug flow, and from the slug flow to the annular
nucleate bubbles move in such region of coolant, where the tem- flow regimes. The transition region in the flow pattern map is in-
perature is less than the saturation temperature; the vapour tem- fluenced by the bubble growth and coalescence, which are, in turn,
perature decreases and consequently condensation ensues at the affected by mass flux, heat flux, vapour velocity, the frequency of
interfaces. Thus, the reduction in bubble size is observed (Fig. 7) bubble generation and lift-off diameters [15].
14 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 11. Variation of mean local bottom wall temperature along the microchannel length with (a) mass flux, (b) heat flux and (c) number of parallel microchannels.

4.3.3. Variation of local heat transfer parameters mass flux. Fig. 11(b) depicts the effect of different heat fluxes (q )
The variation of local wall superheat (Tloc = Tx - Tsat ) is on Tloc in the heat sink with fourteen parallel microchannels for
shown in Fig. 11 for different mass flux, heat flux and number the constant mass flux of G = 500 kg/m2 s. The value of Tloc in-
of microchannels conditions. The local wall superheat is calcu- creases with increasing heat flux for a given mass flux condition.
lated as the difference between the local wall temperature of the Further, the influence of the number of parallel microchannels on
microchannel bottom surface (Tx ) and the saturation temperature the local wall superheat is shown in Fig. 11(c). Three values of mi-
(Tsat ) of the working fluid. It is worth to mention that the local crochannel width are considered as Wch = 320, 580 and 1250 μm
temperature of the microchannel surface presented is estimated as resulting in 14, 10 and 6 parallel microchannels, respectively, on
the average of local temperature of all channels. It is observed that the same footprint area of 1 × 1 cm2 . The Tloc is inversely pro-
for every considered case, the Tloc increases along the flow direc- portional to the channel width, i.e. Tloc decreases with the in-
tion and attempts to reach a constant value at a certain time of op- crease in the channel width or the decrease in the number of the
eration. The effect of different mass fluxes (G) on Tloc for the heat parallel microchannels. With increasing channel width, the area of
sink with fourteen parallel microchannels is shown in Fig. 11(a) for the bottom surface of the microchannel increases and thus the sur-
a constant heat flux of q = 23 W/cm2 . It is found that the value of face temperature increases on the larger width channels for a con-
Tloc decreases with increasing the mass flux. A larger amount of stant heat flux condition. If the surface temperature of larger width
heat is removed from the hot surface by the coolant due to more microchannels is required to maintain at the same surface temper-
quantity of coolant interacts with the hot surface at the higher ature of the smaller width microchannel, a higher mass flux is then
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 15

Fig. 12. Variation of mean local heat transfer coefficient along the microchannel length with (a) mass flux, (b) heat flux and (c) number of parallel microchannels.

required. However, a constant mass flux is considered in the vary- coefficient decreases along the flow direction and reaches a nearly
ing microchannel width cases. Thus, the same quantity of coolant constant value at the outlet of the microchannel. It can be seen
is not sufficient to cool the larger width channel surfaces compared from Fig. 12(a) that the hloc increases with increasing mass flux
to the smaller width channels. As a result, higher Tloc is observed in the heat sink with fourteen parallel microchannels for the con-
when the number of parallel microchannel decreases. stant heat flux conditions, which is a result of convective heat
The variation of spatial average local heat transfer coefficient transfer effects. In subcooled flow boiling, both nucleate boiling
(hloc ) along the channel length is shown in Fig. 12. Three differ- and single-phase convective heat transfer effects are present [32].
ent number of parallel microchannels along with different mass For a constant heat flux, more heat is taken away by the coolant
flux and heat flux conditions are considered for evaluating the be- when mass flux increases. As a result, a thinner thermal bound-
haviour of the local heat transfer coefficient. It is observed that the ary layer is formed at the higher mass fluxes compared to the
hloc decreases along the microchannel length, whereas the maxi- lower mass flux conditions. Thus, to achieve a high heat removal
mum value of hloc is at the entrance of the microchannel for all rate, a higher mass flux of coolant is required. The effect of heat
the considered cases. At the entrance of the channel, the temper- flux for a constant mass flux on the local heat transfer coefficient
ature boundary layer is thinner and the heat transfer coefficient is in the 14 parallel microchannels is shown in Fig. 12(b). It is ob-
higher. The temperature of the thin film approaches the wall tem- served that the hloc increases when heat flux is increased. This
perature as the fluid flows through the microchannel. As a result, is due to more nucleation sites at the higher heat fluxes along
the thermal boundary layer becomes thicker and attains the thick- the microchannel, which enhances the nucleate boiling. The ef-
ness of the liquid film. It is observed that the local heat transfer fect of the number of channels on the local mean heat transfer
16 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Fig. 13. Average heat transfer coefficient variation with heat flux for different mass fluxes, heat fluxes and number of parallel microchannels

coefficient is also investigated and is depicted in Fig. 12(c) for the region. However, in the present case, the flow is developing for the
mass flux of G = 500 kg/m2 s and the heat flux of q = 220 W/cm2 . majority of the length of the microchannel for the mass fluxes of
It is found that the hloc increases with the increasing number of 250 and 500 kg/m2 s. In addition, the collective effect of heat flux
parallel microchannels. As discussed earlier, the thermal bound- and the number of parallel microchannels is shown in Fig. 14(c)
ary layer is thinner in a narrower microchannel for a given mass for the constant mass flux of G = 500 kg/m2 s. At high heat flux
flow rate. As a result, more heat can be transferred by employing of 220 W/cm2 and mass flux of 500 kg/m2 s, the value of aver-
smaller width microchannels. age heat transfer coefficient of the heat sink with six and fourteen
parallel microchannels is found to be 34.02 and 115.8 kW/m2 K, re-
4.3.4. Variation in the average heat transfer coefficient spectively. Therefore, by increasing the number of channels from 6
The variation in the average heat transfer coefficient (havg ) with to 14 by keeping the same width of the fin, the heat transfer coef-
heat flux for two mass flux values of G = 250 and 500 kg/m2 s and ficient increases by 240%. It is observed that to achieve high havg ,
the constant number of parallel microchannels, Ch = 14 is shown the microchannel width is to be minimum (or the number of par-
in Fig. 13. The havg of a particular case is evaluated as the mean of allel microchannels is to be maximum), resulting in more surface
spatial average of all individual microchannels. It can be seen that area to volume ratio of microchannels.
the havg increases with increasing heat flux. The minimum havg is Further, the combined effect of heat flux and mass flux on pres-
found at q = 23 W/cm2 for both the mass flux conditions. As dis- sure drop, P is evaluated. The pressure drop is a very significant
cussed earlier, the single-phase flow is observed at q = 23 W/cm2 parameter to quantify the pumping power required to drive the
resulting in the lesser value of havg . However, as the heat flux in- working fluid through the microchannel and is exhibited in Fig.
creases, the transition from the single-phase to the two-phase is 14(b) for Ch = 14. It can be noted that the P is minimum when
found (Fig. 6). The vapour generation rate increases with increas- both the mass flux and heat flux are minimum for the considered
ing heat flux due to enhanced nucleate boiling, which results in cases in the present study. The frictional resistance is less for the
significantly higher havg for both the mass flux conditions. As heat minimum mass flux condition as the velocity of the fluid is less in
flux is increased further to 220 W/cm2 , the vapour generation rate the microchannel. It is observed that when the mass flux is var-
further increases and the area near outlet region of the microchan- ied for constant heat flux, q = 23 W/cm2 , the percentage incre-
nel is filled with a large volume vapour slug (Fig. 7(e)), resulting in ment of P is 81% and 120% is observed, respectively when G is in-
a small drop in havg ; however, the maximum value of havg is found creased to 400 kg/m2 s and 500 kg/m2 s from 250 kg/m2 s. However,
at q = 220 W/cm2 . A similar trend of the variation of havg with for the constant mass flux condition, P increases with increasing
heat flux can be observed in the present numerical analysis also heat flux. As discussed earlier, the generation of vapour and the ac-
and matches well with the experimental results for both the mass celeration of generated vapour increase with increasing heat flux.
flux cases. A maximum deviation of 15% is found between the nu- Furthermore, the combined effect of heat flux and the number of
merical and experimental results for havg . channels for constant mass flux, G = 500 kg/m2 s is depicted in
The combined effect of heat flux and mass flux on havg is Fig. 14(d). It is found that the P increases with increasing both
shown as a contour for a constant number of parallel microchan- the heat flux and the number of microchannels or by decreasing
nels, Ch = 14 in Fig. 14(a). It shows that the maximum havg can be the microchannel width. The P is found to be increased by 37%
achieved when both the mass flux and heat flux are maximum for when the number of channels is increased from 6 to 14. However,
the cases considered in the present study. The havg increases as the as discussed earlier, by increasing the number of channels from 6
mass flux increases, because of the reduction in thermal boundary to 14 by keeping the same width of the fin, the heat transfer co-
layer thickness, which increases the temperature gradient at the efficient increases by 240%. Thus, for obtaining an optimised set of
wall. When the heat flux is minimum (q = 23 W/cm2 ), the havg is parallel microchannels, a trade-off between pressure drop and heat
minimum and is independent of mass flux for the developed flow transfer coefficient is required to be investigated.
G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973 17

Fig. 14. Combined effect of mass flux and heat flux on (a) havg and (b) P; and microchannel width and heat flux on (c) havg and (d) P.

5. Conclusions microchannel. At higher heat flux, the heat sink with fourteen mi-
crochannels shows higher heat transfer coefficient as compared to
A three-dimensional simulation of two-phase boiling inside the heat sink with six and ten number of microchannels. More-
multiple parallel microchannels is carried out for the different over, the surface temperature of the heat sink with fourteen mi-
number of channels, mass and heat fluxes considering the effect of crochannels is less by 45.5 °C with pressure drop penalty of 37%
inlet and outlet plenums on the flow boiling. Experiments are per- (16.3 mbar) at the mass flux of 500 kg/m2 .s and the heat flux of
formed on a heat sink with fourteen microchannels. The present 220 W/cm2 compared to six microchannels keeping the fin width
numerical model is validated with experimental results with a constant and the footprint area same. The heat transfer coefficient
maximum deviation of 13% and 16%, respectively between the pre- of the heat sink with fourteen parallel microchannels is enhanced
dicted pressure drop and the temperature difference between sur- by 240% as compared to the heat sink with six parallel microchan-
face temperature and inlet temperature with the experimental re- nels at the same operating conditions due to the thin film evap-
sults. The bubble generation and blocking of the channel by bub- oration during annular flow regime at the higher heat flux with
ble elongation cause resistance to the fluid flow, resulting in an more surface area to volume ratio as compared to the heat sink
increase in instability and pressure drop during flow boiling. The with six microchannels. The complete blocking of the channel with
numerical simulation shows that as the number of channels de- the vapour bubble causes the incoming fluid ceases to flow in the
creases, the surface temperature increases due to the increase in channel, which results in instability, as also observed in the exper-
liquid film thickness at high heat flux for larger width of mi- iments.
crochannel because of the confinement effect of bubble inside the
18 G. Hedau, P. Dey and R. Raj et al. / International Journal of Heat and Mass Transfer 158 (2020) 119973

Declaration of Competing Interest [15] R. Zhuan, W. Wang, Flow pattern of boiling in micro-channel by numerical
simulation, Int. J. Heat Mass Transfer 55 (5–6) (2012) 1741–1753. https://doi.
org/10.1016/j.ijheatmasstransfer.2011.11.029.
We wish to confirm that there are no known conflicts of inter- [16] D. Jing, L.He Dalei, Numerical studies on the hydraulic and thermal perfor-
est associated with this publication and there has been no signifi- mances of microchannels with different cross-sectional shapes, Int. J. Heat
cant financial support for this work that could have influenced its Mass Transfer 143 (2019) 118604. https://doi.org/10.1016/j.ijheatmasstransfer.
2019.118604.
outcome. [17] Y. Luo, W. Li, K. Zhou, K. Sheng, S. Shao, Z. Zhang, J. Du, W.J. Minkowycz, Three-
dimensional numerical simulation of saturated annular flow boiling in a nar-
CRediT authorship contribution statement row rectangular microchannel, Int. J. Heat Mass Transfer 149 (2020) 119246.
https://doi.org/10.1016/j.ijheatmasstransfer.2019.119246.
[18] E.M. Chiapero, M. Fernandino, C.A. Dorao, Numerical analysis of pressure drop
Gaurav Hedau: Visualization, Investigation, Methodology, For- oscillations in parallel channels, Int. J. Multiphase Flow 56 (2013) 15–24. https:
mal analysis, Data curation, Validation, Writing - original draft. //doi.org/10.1016/j.ijmultiphaseflow.2013.05.010.
Prasenjit Dey: Software, Data curation, Validation, Formal anal- [19] M.E. Steinke, S.G. Kandlikar, An experimental investigation of flow boiling char-
acteristics of water in parallel microchannels, J. Heat Transfer 126 (4) (2004)
ysis, Writing - original draft. Rishi Raj: Conceptualization, Writ- 518–526. https://doi.org/10.1115/1.1778187.
ing - review & editing, Funding acquisition, Project administration. [20] K.H. Chang, C. Pan, Two-phase flow instability for boiling in a microchannel
Sandip K. Saha: Conceptualization, Writing - review & editing, Su- heat sink, Int. J. Heat Mass Transfer 50 (11–12) (2007) 2078–2088. https://doi.
org/10.1016/j.ijheatmasstransfer.2006.11.014.
pervision, Funding acquisition, Project administration. [21] T. Harirchian, S.V. Garimella, A comprehensive flow regime map for microchan-
nel flow boiling with quantitative transition criteria, Int. J. Heat Mass Transfer
Acknowledgement 13–14 (2010) 2694–2702 53. https://doi.org/10.1016/j.ijheatmasstransfer.2010.
02.039.
[22] D. Sharma, D.P. Ghosh, S.K. Saha, R. Raj, Thermohydraulic characterization of
Authors would like to acknowledge the financial support pro- flow boiling in a nanostructured microchannel heat sink with vapour venting
vided by SERB, Department of Science and Technology, Govern- manifold, Int. J. Heat Mass Transfer 130 (2019) 1249–1259. https://doi.org/10.
1016/j.ijheatmasstransfer.2018.11.005.
ment of India, through grant no. SB/S3/MMER/0023/2014. We
[23] D.P. Ghosh, D. Sharma, D. Mohanty, S.K. Saha, R. Raj, Facile fabrication
thank Prof. Amit Agrawal for providing experimental support for of nanostructured microchannels for flow boiling heat transfer enhance-
this research work. ment, Heat Transfer Eng. 40 (7) (2019) 537–548 2019. https://doi.org/10.1080/
01457632.2018.1436399.
[24] D.P. Ghosh, D. Sharma, A. Kumar, S.K. Saha, R. Raj, An ingenious fluidic ca-
Supplementary materials pacitor for complete suppression of thermal fluctuations in two-phase mi-
crochannel heat sinks, Int. Commun. Heat Mass Transfer 110 (2020) 104347–
Supplementary material associated with this article can be 1–8. https://doi.org/10.1016/j.icheatmasstransfer.2019.104347.
[25] G. Hedau, P. Dey, R. Raj, S.K. Saha, Combined effect of inlet restrictor
found, in the online version, at doi:10.1016/j.ijheatmasstransfer. and nanostructure on two-phase flow performance of parallel microchannel
2020.119973. heat sinks, Int. J. Therm. Sci. 153 (2020) 106339. https://doi.org/10.1016/j.
ijthermalsci.2020.106339.
References [26] A. Siddique, J.M. Bhaskar, A. Agrawal, A. Singh, S.K. Saha, Design of a collector
shape for uniform flow distribution in microchannels, J. Micromech. Microeng.
[1] T.G. Karayiannis, M.M. Mahmoud, Flow boiling in microchannels: fundamentals 27 (7) (2017) 075026. https://doi.org/10.1088/1361-6439/aa73e.
and applications, Appl. Therm. Eng. 115 (2017) 1372–1397. https://doi.org/10. [27] W. Zhimin, C.K. Fah, The optimum thermal design of microchannel heat
1016/j.applthermaleng.2016.08.063. sinks, in: Proceedings of the 1997 1st Electronic Packaging Technology Con-
[2] W. Qu, I. Mudawar, Measurement and prediction of pressure drop in two- ference (Cat. No.97TH8307), Singapore, 1997, pp. 123–129. https://doi.org/10.
phase micro-channel heat sinks, Int. J. Heat Mass Transfer 46 15 (2003) 2737– 1109/EPTC.1997.723898.
2753. https://doi.org/10.1016/S0017- 9310(03)00044- 9. [28] C.W. Hirt, B.D. Nichols, Volume of fluid (VOF) method for the dynamics of free
[3] Z. Dong, J. Xu, F. Jiang, P. Liu, Numerical study of vapour bubble effect on flow boundaries, J. Comput. Phys. 39 (1) (1981) 201–225. https://doi.org/10.1016/
and heat transfer in microchannel, Int. J. Therm. Sci. 54 (2012) 22–32. https: 0021- 9991(81)90145- 5.
//doi.org/10.1016/j.ijthermalsci.2011.11.019. [29] J.U. Brackbill, D.B. Kothe, C. Zemach, A continuum method for modeling sur-
[4] H.W. Jia, P. Zhang, X. Fu, S.C. Jiang, A numerical investigation of nucleate boil- face tension, J. Comput. Phys. 100 (2) (1992) 335–354. https://doi.org/10.1016/
ing at a constant surface temperature, Appl. Therm. Eng. 88 (2015) 248–257. 0021- 9991(92)90240- Y.
https://doi.org/10.1016/j.applthermaleng.2014.09.022. [30] W.H. Lee, A pressure iteration scheme for two-phase flow modeling, Multi-
[5] K. Ling, G. Son, D.L. Sun, W.Q. Tao, Three dimensional numerical simulation on phase Transport Fundamentals, Reactor Safety, Applications, Hemisphere Pub-
bubble growth and merger in microchannel boiling flow, Int. J. Therm. Sci. 98 lishing, Washington, DC, 1980.
(2015) 135–147. https://doi.org/10.1016/j.ijthermalsci.2015.06.019. [31] H. Ganapathy, A. Shooshtari, K. Choo, S. Dessiatoun, M. Alshehhi, M. Ohadi,
[6] R. Jafari, T. Okutucu-Özyurt, Numerical simulation of flow boiling from an ar- Volume of fluid-based numerical modeling of condensation heat transfer and
tificial cavity in a microchannel, Int. J. Heat Mass Transfer 97 (2016) 270–278. fluid flow characteristics in microchannels, Int. J. Heat Mass Transfer 65 (2013)
https://doi.org/10.1016/j.ijheatmasstransfer.2016.02.028. 62–72. https://doi.org/10.1016/j.ijheatmasstransfer.2013.05.044.
[7] M. Magnini, J.R. Thome, A CFD study of the parameters influencing heat trans- [32] A.K. Sadaghiani, A. Koşar, Numerical and experimental investigation on the
fer in microchannel slug flow boiling, Int. J. Therm. Sci. 110 (2016) 119–136. effects of diameter and length on high mass flux subcooled flow boiling in
https://doi.org/10.1016/j.ijthermalsci.2016.06.032. horizontal microtubes, Int. J. Heat Mass Transfer 92 (2016) 824–837. https:
[8] Q. Liu, W. Wang, B. Palm, A numerical study of the transition from slug to //doi.org/10.1016/j.ijheatmasstransfer.2015.09.004.
annular flow in micro-channel convective boiling, Appl. Therm. Eng. 112 (2017) [33] P.S. Lee, S.V. Garimella, Saturated flow boiling heat transfer and pressure drop
73–81. https://doi.org/10.1016/j.applthermaleng.2016.10.020. in silicon microchannel arrays, Int. J. Heat Mass Transfer 3–4 (2008) 789–806
[9] Q. Liu, W. Wang, B. Palm, Numerical study of the interactions and 51. https://doi.org/10.1016/j.ijheatmasstransfer.2007.04.019.
merge of multiple bubbles during convective boiling in micro channels, [34] G.P. Celata, M. Cumo, D. Dossevi, R.T.M. Jilisen, S.K. Saha, G. Zummo, Flow pat-
Int. Commun. Heat Mass Transfer 80 (2017) 10–17. https://doi.org/10.1016/j. tern analysis of flow boiling inside a 0.48 mm microtube, Int. J. Therm. Sci. 58
icheatmasstransfer.2016.11.009. (2012) 1–8. https://doi.org/10.1016/j.ijthermalsci.2012.03.014.
[10] D.F.Fletcher Z.Guo, B.S. Haynes, Numerical simulation of annular flow hydrody- [35] W. Li, Z. Chen, J. Li, K. Sheng, J. Zhu, Subcooled flow boiling on hy-
namics in microchannels, Comput. Fluids 133 (2016) 90–102. https://doi.org/10. drophilic and super-hydrophilic surfaces in microchannel under different ori-
1016/j.compfluid.2016.04.017. entations, Int. J. Heat Mass Transfer 129 (2019) 635–649. https://doi.org/10.
[11] A. Ferrari, M. Magnini, J.R. Thome, Numerical analysis of slug flow boiling in 1016/j.ijheatmasstransfer.2018.10.003.
square microchannels, Int. J. Heat Mass Transfer 123 (2018) 928–944. https: [36] Y. Wang, K. Sefiane, Effects of heat flux, vapour quality, channel hydraulic di-
//doi.org/10.1016/j.ijheatmasstransfer.2018.03.012. ameter on flow boiling heat transfer in variable aspect ratio micro-channels
[12] A. Mukherjee, S.G. Kandlikar, Numerical simulation of growth of a vapour bub- using transparent heating, Int. J. Heat Mass Transfer 55 (2012) 2235–2243.
ble during flow boiling of water in a microchannel, Microfluid. Nanofluid. 1 (2) https://doi.org/10.1016/j.ijheatmasstransfer.2012.01.044. 9-10.
(2005) 137–145. https://doi.org/10.1007/s10404-004-002. [37] I. Mudawar, M.B. Bowers, Ultra-high critical heat flux (CHF) for subcooled
[13] W. Lee, G. Son, Bubble dynamics and heat transfer during nucleate boiling in water flow boiling—I: CHF data and parametric effects for small diameter
a microchannel, Numer. Heat Transfer Part A 53 (10) (2008) 1074–1090. https: tubes, Int. J. Heat Mass Transfer 8 (1999) 1405–1428 42. https://doi.org/10.
//doi.org/10.1080/10407780701789898. 1016/S0017- 9310(98)00241- 5.
[14] Y.Q. Zu, Y.Y. Yan, S. Gedupudi, T.G. Karayiannis, D.B.R. Kenning, Confined bub-
ble growth during flow boiling in a mini-/micro-channel of rectangular cross-
section part II: approximate 3-D numerical simulation, Int. J. Therm. Sci. 50 (3)
(2011) 267–273. https://doi.org/10.1016/j.ijthermalsci.2010.09.004.

You might also like