Siddiqui 2017

Renewable and Sustainable Energy Reviews 74 (2017) 969–1002
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser
Efficient energy utilization through proper design of microchannel MARK

heat exchanger manifolds: A comprehensive review
⁎
Osman K. Siddiqui, Syed M. Zubair
Mechanical Engineering Department, King Fahd University of Petroleum & Minerals, KFUPM Box # 1474, Dhahran 31261, Saudi Arabia
A R T I C L E I N F O A BS T RAC T
Keywords: The presence of non-uniformities in the channels of a heat exchanger due to poor design of manifold and
Microchannel heat exchanger channel parameters has been the focus of the present paper. Unequal flow distribution results in an increased
Flow maldistribution flow resistance leading to increased pumping power. It also leads to an imbalance of thermal characteristics due
Manifold design, analytical models to the different mass flow rate in individual channels. The influence of manifold design, channel design, the
location of inlet and outlet connections to the heat exchanger along with the use of secondary header has been
discussed for both single and two-phase flow. The current achievements in the analytical modeling of such flow
distributing manifolds have also been discussed supporting the numerical and experimental literature. In
general, larger manifold area and a longer channel length are found to be resulting in better flow distribution.
The use of secondary header results in higher pressure loss but this loss is offset by the increased flow
uniformity. Two-phase flow behavior in distributing manifold is experimentally determined to be slug flow as a
dominating behavior with occasional reverse flow. These non-uniformities must be taken into consideration for
the proper design of heat exchangers, thus improving the performance of thermal systems.
1. Introduction cooling and the average surface temperature remains low. The
distribution of flow depends on the pressure losses associated with
A heat exchanger is used when the thermal potential of one each fluid path with the fluid taking the path of least resistance. So
medium is needed to be transferred to another medium, with and it is important to ensure that all paths have the same resistance so
without the direct interaction or mixing between them. The ability that flow will get uniformly distributed across all the channels. To
to transfer heat strongly depends on the contact area between the make flow uniform, the design of inlet and outlet header is of the
two fluids, larger area results in more heat transfer. The fluids highest significance, as its profile will determine how the flow gets
are typically transferred by a piping network of relatively smaller distributed.
size, upon entering the heat exchanger it has to be distributed The parameters that would govern the flow distribution and
within the heat exchanger so as to increase its contact surface area. heat transfer characteristics of heat exchangers include manifold
Particular care is needed to be focused on this fluid distribution, layout and dimensions, channel dimensions, channel-to-manifold
as the flow rate through each fluid path of the heat exchanger has area ratio, the aspect ratio of the manifold-to-channels, the phase
considerable effects on its performance with regards to pressure of the fluid, Reynolds number and secondary distributor systems.
drop and heat transfer [1,2]. The pressure drop in the heat These parameters have been studied by numerous authors to
exchanger manifold and channels can be measured experimentally improve our understanding of their impact on thermal-hydraulic
[3], thus providing the information about the flow distribution performance of micro-channel heat exchangers. This paper sum-
characteristics. marizes and clusters the literature with regards to the important
By making flow uniform throughout the device, it can be controlling parameters. In the end, a summary is provided for the
ensured that there is no potential area for development of a hot most important conclusions regarding heat exchanger design
spot. Furthermore, constant mass flow rate throughout the device and recommendations are given for areas that lack proper under-
ensures that all sections of the heat exchanger will provide uniform standing.
Abbreviations: CFD, computational fluid dynamics; HVAC, heating, ventilation and air conditioning; LAMEE, liquid-to-air membrane energy exchangers; LDV, laser doppler
velocimetry; NTU, number of transfer units; PIV, particle image velocimetry; PV, photovoltaic; PVT, photovoltaic thermal hybrid
⁎
Corresponding author.
E-mail address: [email protected] (S.M. Zubair).
http://dx.doi.org/10.1016/j.rser.2017.01.074
Received 2 July 2016; Received in revised form 11 December 2016; Accepted 12 January 2017
Available online 18 March 2017
1364-0321/ © 2017 Elsevier Ltd. All rights reserved.
O.K. Siddiqui, S.M. Zubair Renewable and Sustainable Energy Reviews 74 (2017) 969–1002
Nomenclature W width (m)

x manifold axial direction
A area (m2)
B constant, Eq. (27). Subscripts
Cd coefficient of discharge
Cp specific heat capacity (kJ kg−1 K−1) * represents parameters of outlet manifold, Eq. (21).
D hydraulic diameter (m) av average
E manifold length to width ratio, Lm/Wm c channel
f friction factor c-c channel to channel
h friction parameter, Eq. (19). eq sudden expansion loss coefficient
J constant, Eq. (27). i-o distance between inlet and outlet port
km momentum recovery factor, Eq. (19). i incremental channel number, Eq. (27).
KL minor pressure loss coefficient in inlet
L length (m) k kth segment of the manifold
m characteristic parameter, Eq. (22). m manifold
ṁ mass flow rate (kg s−1) out outlet
M ratio total channel area to manifold area(=Ac*N/Am) t turning loss
N number of channels
p pressure (Pa) Greek
P perimeter (m)
q volumetric flow rate in individual channel (m3 s−1) β ratio of axial velocity to total velocity in the manifold, Eq.
Q total volumetric flow rate (m3 s−1) (20).
Qw coefficient, Eq. (26). ε effectiveness
Rw coefficient, Eq. (26). ζ average total head loss coefficient for channel flow, Eq.
Re Reynolds number (22).
s constant with value +1 for U, −1 for Z-type manifold (Eq. η ratio of extreme velocities, Eq. (15).
(19)) θ angular position of the inner distributor (Fig. 20)
T temperature (K) µ fluid viscosity
u axial velocity (m s−1) ρ fluid density
U total velocity (m s−1) σD divided flow factor
V velocity in channel (m s−1) σC confluent flow factor
w normalized manifold axial velocity based on inlet velocity τW wall shear stress (N m-2), Eq. (23)
1.1. Applications channels, the resulting pressure drop is small along with reduced
thermal resistance. If the pressure drop increases in the flow domain,
The application of plate type flow distribution device includes then there exists an optimal value of thermal resistance which lies in
cooling of different components like electronic integrated circuits and the turbulent region of Reynolds number. An optimization was
photovoltaic panels where the one side of the channel is exposed to a obtained for the water cooled heat sinks by Goldberg [5], Tuckerman
heated surface. A uniform flow would result in uniform cooling and and Pease [6] and Philips [7]. The design optimization leads to a
equally distributed temperature throughout the surface. Typically, reduction in the thermal resistance within the range of 10–35% for
higher temperatures lead to performance degradation, hence cooling these heat sinks. Some similar applications to the heat sink are [8–15].
would enhance the performance. They are also used in catalytic In addition to heat sinks, flow distributing manifolds are also being
converters that need to increase the contact area for the reaction to utilized in cooling of solar photovoltaic panels or solar hybrid PVT
take place. systems, where cooling of PV panels increase the efficiency [16–25]
The application of flow distribution manifolds in heat sink has while the heated fluid is utilized in domestic heating applications [26–
received significant focus to increase the amount of heat that can be 33]. A review on the impact of cooling the PV panels to reduce the
dissipated from the heat sinks coupled with electronic components. negative impact of higher temperature on the performance was
Knight et al. [4] provided an optimization of the heat sink incorporat- presented by Bahaidarah et al. [34]. Chiou [35] studied the thermal
ing microchannels. The objective was to reduce the thermal resistance performance variation of a solar collector with non-uniform flow
of the heat sink to enable maximum heat transfer. The parameters of distribution, keeping in view that the uniform flow distribution may
Reynolds number, aspect ratio of the channels and fin-to-width ratio not be feasible due to economic and space constraints. A numerical
were studied. It was found that if the flow is low i.e. laminar within the method was developed to study the effect of this non-uniformity on
Fig. 1. A cross-section of a catalytic converter, showing flow distribution and convergence [40].
970
Table 1
A summary of the literature on bifurcation type manifold design characteristics.
Author Type of study Specifications Studies conducted Major outcomes

O.K. Siddiqui, S.M. Zubair
Alvarado et al. [49] Numerical: laminar flow, fluid used– Number of channels investigated–21. Eight Flow distribution uniformity, maximum and average • Distributor A and B designs preferred due to
water. configurations of serpentine, parallel and temperatures of the heating surfaces, temperature contours a) Low–pressure loss,
distributer flow patterns. Reynolds number–100– on the heating surfaces, thermal resistance, pressure loss, b) Lower maximum temperature difference,
600. pumping power. c) Low thermal resistance, and,
d) Good temperature distribution.
O–Charoen et al. Experimental and numerical: laminar Number of channels investigated–8. Headers Microreactor depth, length, and width, microchannel depth • Alternating columns are highly non–uniform.
[51] flow, fluid used–chemicals in DNA divided into half and then further half to get four and width, microreactor tapered inlet width and length. • More uniformity when
synthesis experiments. channels for inlet and four channels for outlet. a) Increase length of reactor,
b) Decrease width of reactor,
c) Decrease depth of reactor,
d) Increase channel depth,
e) Increase channel width.
Saber et al. [52] Multiscale network modeling: laminar Distributing/bifurcation manifolds utilizing Effect of different multi–scale flow distribution • Increase diameter ratio between manifold and
flow, fluid used–water. multiple passes. Number of channels investigated configurations on the overall flow resistance. channels.
up to 100. • Decrease length ratio.
• Optimum channel number calculation is given for
lower pressure loss.
Commenge et al. Multiscale network modeling: laminar Distributing/bifurcation manifolds utilizing Two–scale network used as base for building multi–scale • Design charts for flow characteristics.
[53] flow, fluid used–water. multiple passes. Number of channels investigated– network.
971
50 and 1000.
Amador et al. [54] Electric resistance modeling: laminar Manifold and bifurcation flow distribution. Pressure drop equalization, resistance to flow, circular and • Bifurcation better at higher Reynolds number.
and turbulent flow. Number of channels investigated–16. rectangular shaped channels.
Liu et al. [56] Numerical: laminar flow, fluid used– Bifurcation flow distribution. Number of channels 90°, rounded and circular bends in the bifurcation structure. • Longer length after a bend lead to uniformity.
air. investigated–16. Reynolds number–300–600. • Length to channel width ratio should be higher for
upstream channels as they have higher Reynolds
number.
• The rounded corner design is best, with length to
diameter ratio 5 for are sections.
Chen and Cheng Analytical: laminar flow, fluid used– Bifurcation flow distribution. Number of 90° bends, comparison with parallel channels. • Bifurcation has better heat transfer capability, larger
[57] general. bifurcation levels investigated–3,4 and 5. diameter and number of branching levels increase
heat transfer capability.
• Pumping power is less for bifurcation, decreases with
larger diameter and number of branching levels.
Chen and Cheng Experimental: laminar and turbulent Bifurcation flow distribution. Number of Comparison with parallel channels, impact of Reynolds • Thermal efficiency is more as compared to parallel
[58] flow, fluid used–deionized water. bifurcation levels investigated–4. Reynolds number. channels.
number 1,449–6,503. • The ratio of pumping power for bifurcation to parallel
channel decreases with increasing Reynolds number.
Senn and Numerical: laminar flow, fluid used– Bifurcation flow distribution. Number of Comparisons of bifurcation and serpentine for same surface • Bifurcation has half pressure loss as compared to
Poulikakos [59] water. bifurcation levels investigated–6. Constant wall area. serpentine.
heat flux, Reynolds number–20 and 200. local Nusselt numbers
• Laminar mixing leads to higher(continued on next page)
thermal performance with the introduction of a flow non-uniformity
in
Pressure loss for both bifurcation is 72.4% higher than
Inlet Pyramid and outlet bifurcation is found to be the

parameter to compare the flow distribution about the mean value. A
in heat transfer coefficient

relation was found that exists between the flow non-uniformity
parameter and the collector efficiency deterioration, which is expected
▪ Inlet bifurcation, outlet pyramid=15%.

▪ Inlet pyramid, outlet bifurcation=28%.
comparison to both pyramid manifolds to be helpful in reducing localized heat concentration. He used different
collector designs and operating conditions to determine the collector
efficiency based on ten flow channels, numbered 1–10 from left side.
The first flow model had maximum flow in tube 1, decreasing towards
tube 10. The second model had maximum flow in tubes 3 and 4 while
▪ Both bifurcation=30%.
decreasing in tubes on both sides. The third model had maximum flow
for bifurcation type.
best performing. in central tubes 5 and 6, leading to symmetrically decreasing distribu-

tion while the fourth model had some tubes plugged to have no flow in
both pyramid.
• Improvement
Major outcomes
them representing the most severe form of non-uniformity. The study

showed that the deterioration in collector efficiency was from 2% to
20% for the above flow distribution models, for flow non-uniformity
parameter in between 0.01 and 0.3. Some further research efforts have
•
•
also been published in the literature for photovoltaic and hybrid

thermal collectors [35–39].
The combination of inlet and outlet header to be either
Another application of flow distributing systems is in catalytic

converters which contain oxidation and reduction agents to reduce the
toxic emissions from engine to a lesser toxic form. They utilize flow
distributing systems, where the objective is to increase the contact
reaction area. A cross-sectional view of the catalytic converter is shown
in Fig. 1, where it can be seen that the inlet manifold incorporates a
diffuser to distribute the exhaust gasses followed by a ceramic monolith
pyramid or bifurcation shaped.
core that hosts the catalyst to react with the exhaust gasses. Towards
the end, a converging manifold is used to connect the exhaust pipe.
Agrawal et al. [40] studied the effect of flow maldistribution on the
Studies conducted
effectiveness of catalytic converter. They compared 85 channels with

the reduced scale of only 21 channels and found good agreement for
constant velocities. Flow expansion in manifold deteriorates flow
distribution while friction in channels improves the flow distribution.
Up to 20% deviation was discovered in the central channel as compared
to side channels for increasing Reynolds number. It was found that
Bifurcation flow distribution. Number of channels
conversion rates also reduce as the flow maldistribution increases;

investigated–128. Reynolds number– 50–260.
where the single channel models over-predict the actual conversion

rate. The consideration of energy effects improved the flow distribution
due to density changes as heat is lost to the surroundings. Some further
studies on performance enhancement by improving flow distribution in
catalytic converters are given in [41–47].
2. Flow distribution in heat exchanger
Various efforts have been put forward by some authors suggesting

Specifications
methods of analytical, numerical and experimental nature for ensuring

uniform flow distribution and optimization of heat exchangers so as to keep
the losses low with high effectiveness. These research efforts are summar-
ized in Tables 1–7, with detailed discussion in the following sections. The
manifold design, channel design, location of inlet and outlet connections to
the heat exchanger along with the use of secondary header, two-phase flow
Experimental: laminar flow, fluid
behavior, and the analytical modeling has been discussed in detail.
2.1. Manifold design
The basic manifold designs are conservative and bifurcation type, as

shown in Fig. 2. In a conservative design, there is a single manifold
Type of study
used–water.
region directly connected to the channels distributing the flow. The

dashed line in the manifold shows an alternate for improving flow
uniformity by making the manifold non-uniform or triangular. In
bifurcation type, the flow from the inlet is divided into two streams;
Table 1 (continued)
each is then further subdivided into two more till the number of
divisions matches the number of channels. This layout is inspired by
Luo et al. [62]
nature and a detailed description has been given by Bejan and Errera
[48], mentioning the presence of this design in trees, cracks in the dry
Author
ground, blood circulation system, and lungs, leading to flow uniformity

with low flow resistance.
972
Table 2
A summary of the literature on conservative type manifold design characteristics.
Gandhi et al. [63] Numerical: turbulent flow, fluid used– Inlet and outlet in–line with channels at middle of With and without a channel directly in line with the inlet • Non–uniformity increased with
air and water. manifold. Number of channels investigated–A (9, 27), and outlet port, effect of pitch diameter, header diameter, • Increase in tube diameter,
B (8, 24), C (10, 30, 50). Reynolds number–35,600– channel diameter, inlet outlet diameter, number of tubes • Increase in inlet flow rate,
100,500. inlet flow rate is discussed. •C1 Increase in total number of tubes.
• and C12 designs were the best configurations.
Mohammadi et al. Numerical: laminar flow, fluid used– Triangular manifold. Number of channels The size of manifold, angle of manifold. • Long narrow channels better.
[64] water. investigated–20. Reynolds number–5 to 25. • Optimum corner angle for each manifold width.
• Overall pressure loss minimum for biggest
manifold with smallest corner angle.
Minqiang et al. [66] Numerical: laminar flow, fluid used– Z–type. Number of channels investigated–20. Inlet Asymmetrical design layout, length, width, depth and • Symmetrical manifolds better than asymmetrical.
water. velocity 1 mm/s. spacing between channels, position and radius of inlet– • Corner angle significant for asymmetrical
outlet ports, width of manifold manifolds.
• More flow uniformity with
a) Larger length and depth,
b) Smaller width,
c) High aspect ratio of channels,
d) Larger manifold,
e) Larger inlet–outlet ports,
f) Thinner baffles.
Tonomura et al. Numerical: laminar flow, fluid used– Z–type. Number of channels investigated–5. Reynolds Length of channels, asymmetrical size of inlet outlet • More uniformity with
[67] water. number– 10–100. manifold, shape of manifold, moved baffles. a) Longer channels,
b) Outlet manifold width double to that of inlet
973
manifold,
c) Corners opposite to inlet/outlet port chamfered
to remove stagnant zone,
d) Moving baffles instead of chamfering corners.
Griffini and Numerical: laminar flow, fluid used– Z–type. Number of channels investigated–12. Comparison between 2D and 3D, effect of channel length, • Critical Reynolds number 0.37 for 2D and 12.83 for
Gavriilidis [68] liquid. Reynolds number– up to 50. plate thickness, design of manifold, different location of 3D.
inlet–outlet ports. • Increase channel length.
• Increase inlet/outlet manifold size.
• No impact of plate thickness due to low Reynolds
number.
• Move the inlet/outlet ports within the channel
domain, instead of offset.
Solovitz and Jeffrey Numerical and analytical design with Z type. Number of channels investigated–5 and 10. Hydraulic diameter from the analytical model, uniform Uniform and non–uniform manifold depth have no
[69] major pressure loss equalization: Reynolds number–5-500. and non–uniform manifold depth calculated from
• difference provided the hydraulic diameter remain
laminar flow, fluid used–water. hydraulic diameter, four manifold configurations–narrow, same.
wide, taper and power law manifold. • Behavior becomes significantly non–uniform above
Reynolds number of 20.
• Power law better for flow distribution at low
Reynolds number.
Pan et al. [70,71] Numerical and frictional resistance Z–type flow arrangement. Obtuse, right and acute triangular shapes considered, • Right angled triangular design better.
model: laminar flow, fluid used–liquid symmetrical and asymmetrical. • Symmetrical manifolds better.
and gaseous water. • For asymmetrical, rake angle significant.
Vásquez–Alvarez Numerical and fabrication: laminar Z–type and parallel. Number of channels Four configurations studied, obtuse, isosceles, circular • Circular manifold with baffles at the extreme sides
[72] flow, fluid used–nitrogen. investigated–19, 20 and 21. Reynolds number– with channels at extremes, circular with baffle at extremes. has the best flow distribution.
60,000. pressure loss.
• This design has slightly more(continued on next page)
Table 2 (continued)
• Higher flow rates increase non–uniformity.

• 3D modeling results in higher pressure loss as
compared to 2D modeling, due to additional walls
at top and bottom in 3D.
Hassan et al. [73] Experimental and numerical: turbulent Distributing manifold. Number of channels Pressure considerations in uniform and taper manifold. Linear taper for circular cross–section of manifold
flow, fluid used–water. investigated–5. Reynolds number– 100,000, 150,000
• and channels.
and 200,000.
Perlmutter [74] Analytical: turbulent flow, fluid used– Distributing manifold. Inlet and outlet headers with/without friction. • Good agreement with experiments.
dry air. • Large manifold length to height ratio.
• Lower overall pressure drop is possible with
manifold shapes manipulation.
Tereda et al. [75] Experimental: turbulent flow, fluid Z– and U–type flow arrangement. Number of channels Manifold diameter variation, flow rate variation. • Pressure variation due to flow branching dominant
used–water. investigated–25. Reynolds number–460, 644, 764 and over the friction pressure drop in manifold.
878. • Flow non–uniformity increase with flow rate.
• Larger manifold reduces pressure drop variation
pointing to better flow distribution.
• Effect of flow rate for larger manifold is less
significant.
Chen et al. [76] Numerical: turbulent flow, fluid used– U–type flow arrangement. Number of channels Effect of flow rate, permeability, inlet flow rate. • Low flow rate better distribution.
air. investigated–72. • Low permeability improve distribution but increase
overall pressure drop.
• More channel resistance leads to uniform flow.
974
Datta and Finite difference: turbulent flow, fluid U– and Z–type flow arrangement. Number of channels Comparison between U– and Z–type, effect of area ratio is better than Z.
Majumdar [77] used–air. investigated–10–20. and friction parameter.
• UDecrease
• distribution.channel to manifold area ratio, better
• Most flow in the last channel for Z, first channel for
U–type manifold.
• Effect of friction factor less significant.
Bajura and Jones Analytical: turbulent flow, fluid used– U– and Z–type flow arrangement. Number of channels Verification of the analytical model, area ratio and lateral • U–type better than Z–type.
[78] air. investigated–10–20. Reynolds number–60,000 to resistance manifold. • Decrease area ratio for better flow distribution, less
80,000. than one preferably.
• Larger lateral flow resistance is preferred, but it
increases overall pressure loss.
• For smaller manifolds, friction may be neglected,
for longer manifolds static pressure changes due to
channel branching may be neglected.
Choi et al. [79] Numerical: laminar flow, fluid used– Z–type flow arrangement. Number of channels Ratio between channel area and manifold area, area ratio • Friction always decrease the pressure while
water. investigated–8. Reynolds number–5, 50 and 250. 4, 8 and 16 studied. momentum decreases pressure in combining
header and increases pressure in dividing flow
header.
• Lowest area ratio of 4 is better for uniform flow
distribution, but still flow in the last channel was
2.75 time that in first.
Choi et al. [80] Numerical: laminar flow, fluid used– Z–type flow arrangement. Number of channels Width ratio between outlet and inlet manifold. • Larger width ratio 4 is the best, also less pumping
water. investigated–8. Reynolds number–5, 50 and 250. power.
Reynolds number effect is more significant than
• width ratio.
(continued on next page)
Table 2 (continued)
Kim et al. [81] Numerical: laminar flow, fluid used– Z–type flow arrangement. Number of channels Manifold shapes considered, trapezoid, triangular, and • Triangular better.
water. investigated–8. Reynolds number–50, 100, 200 and rectangular. • Higher Reynolds number decreases flow
300. distribution.
• Higher Reynolds number maldistribution more

significant for rectangular than triangular.
Jones and Lior [82] Analytical: turbulent flow, fluid used– Number of channels investigated–4, 8 and 16. Reynolds number and channel number with size of • Channel to manifold diameter ratio is significant,
water. Reynolds number–3210, 9640, 16100. channel. more uniform smaller ratio.
channels are better; non–uniformity
2
• Fewer
increases with n .
• Longer channels preferred.
• Effect of distance between channels is insignificant.
Tong et al. [83] Numerical: laminar flow. Z–type modified to parallel. Number of channels Manifold width, linear and non–linear taper for both inlet • Bigger manifold width.
investigated–10. Reynolds number–35 to 1,400. and outlet header, variable channels widths, shape of inlet, • Small taper angle for low Reynolds number, larger
three non–linear tapers considered, concave down, angle for higher Reynolds number.
concave up and combination of concave–down followed by • Concave down is best, but difficult fabrication
concave–up section. offsets the benefit.
• Outlet manifold taper led to increasing non–
uniformity.
• Making last channels of smaller widths than initial
channels is strong parameter.
• Smaller inlet than manifold leads to flow reversal.
The inlet and manifold size should be same.
of having inlet fluid entering at an angle
975
• inNothebenefit
manifold.
• Rounding of the channel inlet corners led to
increasing flow non–uniformity.
Miura et al. [84] Experimental and numerical: laminar 32 different flow arrangements. Number of channels Effect of flow arrangement on total pressure drop. • Linear dependence between number of passes and
flow, with some turbulent regions, fluid investigated up to 18. the pressure drop.
used–water.
Rao et al. [85] Theoretical and experimental: Z– and U–type flow arrangement, number of channels Effect of change in manifold size and number of plates for • Effect of maldistribution is significant on the heat
turbulent flow, fluid used–water. investigated–19 and 33. Reynolds number– 600– varying Reynolds number. exchanger performance.
5500. • U–type is more uniform in comparison to Z–type.
Z–type performance is least affected by flow
• distribution.
• Increasing plates reduces effectiveness.
• Increasing manifold size reduces flow
maldistribution.
Rao and Das [86] Experimental: turbulent flow, fluid Z– and U–type flow arrangement. Number of channels Effect of port to channel flow maldistribution. • Inlet–outlet Port size should be maximum possible.
used–water. investigated–10, 15 and 18. Reynolds number–700– • Total pressure drop –critical function of flow rate
7000. and port size.
• Flow more uniform in U–type arrangement.
Wasewar et al. [87] Numerical: laminar flow, fluid used– Flow distribution manifold. Number of channels Modified manifold shape. • Non–uniformity decreased by 70.71%.
air. investigated–11.
Gunnewiek et al. Numerical: turbulent flow, fluid used– Flow combining manifold. Vertical transpired air heater. • Flow reversal can occur for smaller volume flux.
[88] air. • Increasing manifold length to width ratio increases
the pressure loss.
• Higher suction velocity in lower channel is due to
Table 2 (continued)

buoyancy–driven flows, possible with wider

manifold.
• Heat transfer in non–uniform flow can lead to
better efficiency than uniform flow, for buoyancy
driven flows.
Mohan et al. [89] Numerical: laminar flow, fluid used– Z–type flow arrangement. Number of channels Effect of different gasses, flow rate, manifold size. • Hydrogen uniform.
hydrogen, oxygen, and air. investigated–80. • Air and oxygen non–uniform as Reynolds number
variation due to viscosity.
• Less uniform with higher flow rates.
• More uniform with larger manifold.
976
Table 3
A summary of the literature on channel parameters, affecting flow distribution.
Fu et al. [90] Experimental: turbulent flow, fluid used– Number of channels investigated–2. Reynolds Effect of flow rate. • Channel is more turbulent than manifold.
water. number–60,000. • Recirculation region at entry to channel.
• Flow rate more in the second channel.
• The second channel is more turbulent.
Commenge et al. Approximate resistive network model and Z–type flow arrangement. Number of channels Approximate pressure model compared with finite– • Longer length leads to uniform distribution, as
[92] finite difference: laminar flow, fluid used– investigated–13, 23 and 34. volume model, channel parameters of length and width pressure loss in channels significant as compared to
nitrogen. studied. loss in manifold.
• Wider channels result in increase in flow non–
uniformity.
Pan et al. [93] Equivalent electrical resistance network Z–type flow arrangement. Reynolds number–less Combined optimization of microchannel width, depth, • Microchannel width is most significant; larger is better
model: laminar flow, fluid used–general than 2,000. and number of channels. leading to fewer channels required.
incompressible and continuum. • Microchannel depth has an insignificant effect.
Pan et al. [94] Frictional resistance model: laminar flow, Z–type flow arrangement. Pressure drop and friction resistance studied. • Channel parameters more significant that manifold
fluid used–water. parameters.
• High aspect ratio channels preferred.
Weitbrecht et al. Analytical and experimental: laminar flow, Z–type flow arrangement. Number of channels Laser Doppler used to find flow in each channel, • Manifold may get turbulent but channels always
[95] fluid used–water. investigated–10. Reynolds number–807, 1491, analytical model used to study sensitivity of system to laminar.
2552 and 4604. various parameters. • The loss coefficient for combining manifold is larger
than that for dividing manifold.
Table 3 (continued)
• The total loss is mainly due to friction loss in channels.

• Higher Reynolds number increases maldistribution.
Alm et al. [96] Numerical: fluid used–nitrogen and water. Z–type flow arrangement. Number of channels Heat transfer efficiency for varying channel height and • Nitrogen, fairly constant efficiency for increase in
investigated–16. constant width, smaller channel height termed velocity.
geometry A, larger channel height termed geometry B. • Lower velocity, geometry B slightly better, higher
velocity geometry A better.
• Water, efficiency decreasing for increase in velocity,
Geometry B better at lower and higher velocity,
Geometry A better at intermediate velocity.
• Efficiency of liquid phase applications limited for
lower temperature ranges.
Abdel–Salam et al. Experimental: laminar and turbulent flow, Flat–plate LAMEE. Effect of membrane deflection on flow uniformity. Deflection of membrane reduces the air side flow area,
[98] fluid used–air.
• leading to flow non–uniformity and higher pressure
loss.
• Use of support grid suggested reinforcing the
membrane.
977
Abdel–Salam et al. Numerical: laminar flow, fluid used– Flat–plate LAMEE. Parallel channels for air and Effect of the pressure differential on membrane • Membrane deflection of up to 4 mm for 5 psig is
[102] desiccant solution and air. Z–Type for solution. Number of channels deflection, impact of manufacturing tolerances on observed.
investigated–21 for air and 20 for solution side. performance. • For 2 mm channel, increasing standard deviation in
thickness of 0.6 mm results in 12% drop in
effectiveness.
• For 6 mm channel, only 1.7% drop in effectiveness.
• For each standard deviation, an optimum width exists.
Abdel–Salam et al. Experimental: fluids used–water, air, and 3–fluid LAMEE. Water in Z arrangement and Effect of air channel variations on performance, impact • For three fluids, performance degraded by 7% for air
[103] desiccant solution. cross flow for air and desiccant. of pre–tensioned membrane. channel standard deviation of 0.2.
• Pre–tension in membrane significantly reduced the
membrane deflection.
Table 4
A summary of literature on number and location of inlet/outlet ports, affecting flow distribution.

Huang et al. [104] Numerical and experimental: laminar Z–type modified to parallel with additional inlet. Number of Charge–coupled device (CCD) camera to • Design IV with ten guiding vanes in manifold
flow, fluid used–water. channels investigated–12. capture flow distribution, comparison results in best flow distribution.
between four design layouts.
Balaji and Lakshminarayanan Numerical: laminar flow. Number of channels investigated–19. Reynolds number–5– Multiple inlets and outlets of circular shape, • Two inlets and four outlets better.
[105] 200. location and number of inlet and outlet • Results verified on low and highly density fluid.
ports.
Delsman et al. [106] Numerical: laminar flow, fluid used– Z–type flow arrangement. Number of channels investigated– Different angles of inlet and outlet port. • Best layout–inlet and outlet along the channels
gas. 19. Reynolds number–1600. axis.
• Doubling inlet–outlet port size reduced
standard deviation by factor of 4.
• Doubling channel length reduced standard
deviation by factor of 1.7.
• Optimal geometry depends on flow rate.
Lu and Wang [107] Numerical: laminar flow, fluid used– Number of channels investigated–20, 40 and 60. Reynolds Five flow arrangements, inlet velocity, • Arrangement (a) has highest pressure drop,
water. number–100–500. Heat flux 120–240W. number of channels. • Arrangement (c) has lowest pressure drop.
Arrangement (b) has flow recirculation, poor
• performance.
• Arrangements (a) and (e) have better heat
transfer performance.
978
• Arrangement (c) has best thermal performance
with lower pressure drop.
Bogojevic et al. [108] Numerical and resistance network: Number of channels investigated–40–100. Reynolds Comparison between approximate and • Approximate model only reliable at lower
laminar flow, fluid used–water. number–0.01–100. Pressure losses ignored in approximate numerical model, right angle and oblique Reynolds number.
model for higher Reynolds number and variable manifold. triangle. Better flow distribution for inlet/outlet port
• within the channel domain also results in
smaller manifold.
• Pressure loss increased for blocked channels
irrespective of location.
• Channels blocked opposite to inlet side causes
most maldistribution, Middle least.
• Bubbling disrupts flow distribution.
Zeng et al. [109] Numerical: laminar flow, fluid used–gas Z–type flow arrangement. Number of channels investigated– Right angle and oblique triangle. • Flow distribution independent of temperature.
mixture of methanol and water in 1:1.3 50. Temperature 400 °C and 450 °C. • Right angle manifold better flow distribution
molar ratio. with or without the chemical reaction
occurring.
Kim et al. [110] Numerical and experimental: laminar Parallel flow. Number of channels investigated–10. Reynolds CCD camera for flow visualization, 20 cases • New parameter defined with optimum value of
flow, fluid used–water. number–0.0332, 0.332, 3.32, 33.2. with different dimensions, inlet and outlet 0.8 for uniform flow.
offset. • Longer manifold decreases uniformity.
• Wider manifold increase uniformity.
• Inlet–outlet distance increment increases
uniformity up to a point then starts decreasing.
Table 5
A summary of literature on perforated plates for secondary flow distribution, affecting flow distribution.
Jiao et al. Experimental: fluid used–water. Two designs for secondary header by increasing the number of Effect of inlet pipe header, and secondary header diameter. • Flow maldistribution increases with Reynolds
[111] openings. Number of channels investigated–30. Reynolds number, a correlation is presented.
number–1100–3000. • Header C configuration is best, having seven
openings.
• Diameter ratio given for best results.
Zhang and Li Numerical: laminar and turbulent Number of channels investigated–13. Reynolds number–500– Secondary flow distributing mesh in the manifold. • More non–uniformity at higher Reynolds
[112] flow, fluid used–air. 3000. Inlet at middle of manifold. number.
• Secondary header better for flow distribution.
• Type C with seven holes, with width of secondary
header.
Habib et al. Numerical: turbulent flow. fluid Distributing manifold, flow perpendicular to channel plane. Inlet velocity, diameter, shape, number and location of nozzles. • Inlet velocity has no effect on flow distribution.
[113] used–oil. Number of channels investigated–32 (16 in axisymmetric • Inlet size reduction leads to higher velocity,
solution). causing more non–uniformity.
• Elliptic nozzle leads to better distribution.
• More channels result in better flow distribution
as less velocity in each.
979
• Increasing number of holes in secondary header
increases uniformity.
Wen and Li Numerical: highly turbulent flow, Number of channels investigated–43. Reynolds number– Effect on hole size of plate on flow distribution, plate with • Baffle with holes improve flow distribution.
[114] fluid used–nitrogen. 10,000–1,000,000. various sized hole distributed throughout to distribute the • Holes with staggered arrangement with punch
flow. ratio increasing.
Wen et al. Experimental and numerical: Number of channels investigated–21. Reynolds number– Comparison between PIV and numerical, effect of different • Staggered arrangement of holes better than
[115] turbulent flow, fluid used–water. 6,000. layout for holes in the secondary distributor. inline arrangement.
• Punched ratio increment from the central axis
reduced vortices.
Lalot et al. Analytical and numerical: turbulent Number of channels investigated–128. Reynolds number– Use of perforated plate as secondary, study of heat exchanger • Better distribution with perforated plate.
[117] flow, fluid used–air. 10,000–40,000. performance and variation in effectiveness calculated • More non–uniformity, less heat exchanger
depending on flow distribution. effectiveness.
• Decrease in effectiveness 7% for condensers and
counter flow.
• 25% for crossflow.
Ismail et.al. Numerical: turbulent flow, fluid Number of channels investigated–6. Reynolds number–2000 Effect of baffle plate on flow distribution and pressure drop. • Use of secondary baffle reduced pressure drop by
[119] used–air. to 15,000. 34% while improving the flow distribution.
Table 6
A summary of literature on flow distribution with two phase flow.

Datta and Majumdar [120] Finite difference: fluid used– U– and Z–type flow arrangement. Number of channels With and without heat transfer, steam formation • Mixed flow arrangement found to be best.
water and air. investigated–6. Heat flux studied–50, 100, 150 and depending on heat flux. • Non–uniformity increase with heating load.
200 kW. Exit steam pressure 68.9 bar. Inlet water • Dryness fraction increases with decreasing flow rate.
temperature 204 °C. • Higher flow rate results in steam formation in some
channels while non in other.
Balasubramanian and Experimental: fluid used–boiling Number of channels investigated–6. Two–phase flow patterns and pressure drop, effect • Nucleate boiling under annular flow conditions.
Kandlikar [121] water. of temperature on bubbling frequency and flow • High interface velocity, growth reduces upon contact
rate on overall pressure drop. with wall.
• Pressure drop increases with increasing flow rate.
• Frequency of pressure drop fluctuations increases
with increase in surface temperature up to 109 C,
due to higher bubble nucleation frequency.
• Higher than 109 C, slug formation reduces the
frequency.
• Reverse slug flow in some channels.
Bobbili et al. [122] Analytical (Bassiouny and Martin U–type plate condenser. Number of channels investigated Effect of non–uniformity on thermal performance • Effectiveness and condensation rate decrease for
[133]): fluid used–process steam. –98. and exit vapor conditions, effect of mass flow rate, increasing non–uniformity.
flow configuration, number of channels. • Output steam quality remained virtually same.
• Performance and condensation deteriorated as
maldistribution increases.
980
Martin et al. [123] Numerical and fractional–step Number of channels investigated–3. Reynolds number– Flow maldistribution due to vapor bubbles in • Numerically simulated results match with
scheme: fluid used–water and air, 196 and 1895. channels, valve activation based on flow rate, experimental.
laminar flow. utilized to actively remove bubbles in channels. • Use of valves actively reduced flow maldistribution.
Marchitto et al. [124] Experimental: fluid used–water Distributing manifold for upwards vertical flow in Effect of area ratio, nozzle at inlet to manifold to • Increase area ratio, gas distribution improved but
and air. channels. Number of channels investigated–16. create jet. liquid distribution decreased.
• Nozzle improved liquid flow distribution at higher
gas superficial velocities, but worsening effect for
intermediate velocities.
Marchitto et al. [125] Experimental: fluid used–water Distributing manifold for upwards vertical flow in Effect of liquid and gas superficial velocities. Gas flow rate more important for flow distribution
and air. channels. Number of channels investigated–16.
• than liquid flow rate.
• Increase in gas flow rate improves its distribution
while reducing liquid distribution.
• Secondary distributor found to be better for uniform
flow distribution.
Marchitto et al. [126] Experimental: fluid used–water Distributing manifold for upwards vertical flow in Impact on cylindrical distributor with 8 or 16 in– • Improvement with the secondary distributor, best
and air. channels. Number of channels investigated–8. line holes as secondary distributor. when hole opposite to channel flow (holes facing
down for upwards flow in channels β=180°).
• Eight holes instead of 16 yields better distribution.
• Smaller hole size does lead to higher pressure in
manifold, but no effect on flow distribution.
• Header pressure drop decreases with increasing gas
velocity; flow distribution is not dependent on it.
2.1.1. Bifurcation type manifold
zero protrusion, more liquid flow into initial

flow rate increased the flow in later channels.
For protruded channels, less liquid flow in first
shape has better flow distribution than

of mass quality diminishes as protrusion
air velocity forces more water in the last

protrusion reduces the size of recirculation.
and increasing towards the last channels.
channels and then decreasing to the last channels.
Bifurcation type manifold is a nature-inspired design for flow
distribution and is found to be an efficient way of distributing flow in
multiples channels. A summary on bifurcation manifolds is given in
circular; volume fraction remains same.

Table 1, with a detailed discussion in the following paragraphs.
increases, due to increased mixing.

3 mm protrusion found.
Alvarado et al. [49] studied eight different layouts including
serpentine, parallel and bifurcation flow patterns for a microchannel
based heat sink (Fig. 3). The numerical modeling was done using
ANSYS Fluent [50], a commercial software for numerical solution of
the basic mass, momentum and energy equations, resulting in infor-
mation about the flow distribution in channels, pumping power needed
Major outcomes
as well as the uniformity of surface temperature. These parameters are

• Increasing
• Optimum
channels.
useful in quantifying the performance of such heat exchangers. The
• channels
• Higher
• Square
• Larger
findings revealed that the distributor/bifurcation type of manifold

• Effect
• For
results in the lowest pressure loss and surface temperature difference,

thus recommended for heat sink applications where liquid based
cooling is needed.
Another type of layout for bifurcation manifold with inlet and outlet
Effect of shape of manifold, circular or square.
Effect of channels protruding into manifold
channels parallel to each other is presented by O-Charoen et al. [51],

where four inlet and four outlet channels are studied leading to a total
of eight channels. Computational Fluid Dynamics (CFD) software was
used for computing the results and had two experimental arrangements
for validation. It was found that the alternating channels have flow
maldistribution. Fig. 4(a) shows the layout along the channel axis (x-
coordinate) velocity, where it can be seen that near the entry and exit,
flow is relatively symmetric while at the middle section stagnation or
Studies conducted
reverse flow can occur. This is due to the fact that the manifold and
tube are of the same size, which causes the majority of flow directed in
studied.
the first channels while the remaining would just flow towards the ends
as it gains momentum in the manifold. Fig. 4(b) shows a modification
where the channels have smaller inlet and outlet, which increases the
Distributing manifold. Number of channels investigated–
Distributing manifold. Number of channels investigated–
pressure loss and improves the flow distribution by reducing the

impact of pressure loss in the manifold as compared to that in the
channel. It was further revealed that more flow uniformity was
obtained when length of the reactor is increased with a decrease in
width and depth while increasing the channel width and depth.
A design similar to distributor layout in Fig. 3 was examined by
Saber et al. [52] by terming it as a multi-scale flow distribution. The
development of large-scale network of microchannels for application to
macro-scaled applications was analyzed by a basic layout and then
creating networks with laminar flow and constant temperature opera-
tion. The two scale configuration resulted in lower flow maldistribution
and pressure loss, with a limit on the number of channels for each
Specifications
segment for the least pressure loss. The connection between different
sections of the geometries was given special attention, as the proper
connection can result in uniform flow for any number of sections.
10.
Furthermore, increasing the number of flow scales leads to better flow

4.
distribution. An additional contribution was published by Commenge

Numerical: fluid used–water and
Experimental: fluid used–water
et al. [53] including design charts developed to design a multi-scale

network based on a two-scale network. The model was based on flow
resistances at every scale and step-by-step procedure that allows
calculation of the global parameters, thus assisting in design and
evaluation of multiple systems to find the optimum configuration.
Type of study
A comparison between the two manifold arrangements i.e. con-

servative and bifurcation was made by Amador et al. [54] by modeling
and air.
of the geometries as electrical resistances forming a network and then

air.
validated with finite element simulated results. The model has been
found to apply to both rectangular and circular cross-sections for the
channels, while using two approaches to attain uniform distribution.
Premkumar and Mahesh
Firstly, by the usage of larger uniform cross-section for the manifold to

Table 6 (continued)
Lee and Lee [127]
minimize the pressure drop while the second method was to make the
pressure drop uniform for all the channels with the use of variable sized
channels. The results show that for lower Reynolds number applica-
[128]
tion, the first method is suitable, while for non-uniform depth cases,
Author
the second method is recommended. Also if the Reynolds number is

high, the bifurcation type of manifold is suitable. For the impact of
981
Table 7
A summary of the literature on analytical modeling for flow distributing and combing manifolds.
Acrivos et al. [129] Experimental and analytical, Distributing and combining manifolds. Reynolds number– Flow distribution, effect of number of channels. • Pressure recovery and pressure drop due to friction
fluid used– air. 22,830. are factors determining non–uniformity.
• Reducing channels enhances flow distribution until
friction becomes dominant for increased channel

gap.
Kubo and Ueda [130] Analytical and experimental: Distributing and combining manifolds. Central inlet parallel to channels. Dimensionless parameters of divided flow factor
fluid used–water.
• and confluent flow factor presented.
• Formulae are given to calculate flow rate ratio for
each branch pipe.
• Pressure and flow rate ratio increases with the
number of channels.
Majumdar [131] Analytical–finite difference: Number of channels investigated–10. Design charts generated, two non–dimensional Chart is given for flow imbalance and overall
fluid used–general. parameters presented.
• pressure drop for combining and dividing manifold.
• Can be extended to two–phase flow.
1 1
• Low values of F and P for flow uniformity.
Pigford et al. [132] Analytical: turbulent flow, U– and Z–type flow arrangement. Number of channels Effect of pressure difference on flow uniformity. • Application of analytical model.
fluid used–air. investigated–5 and 50 for U–type. 10, 20 and 100 for Z–type. • Header should be greater than all tubes.
Reynolds number 53,730. • Larger tubes allow more tolerance in header
pressure changes.
Bassiouny and Martin Analytical: fluid used– U–type flow arrangement. Dimensionless distance to study Velocity distribution in channels, area ratio based on • More uniformity by varying the area ratio,
[133] general. presence of channels. new parameter ‘m’. regardless of number of channels and shape.
2 2
982
m =0, pure uniform flow, m should be less than
• 0.1 for practical cases.
Bassiouny and Martin Analytical: fluid used– Z–type flow arrangement. Dimensionless distance to study Velocity distribution in channels. • Pressure drop in channels increase along the
[134] general. presence of channels. manifold direction.
Shen [135] Analytical: turbulent flow, Combining and distributing manifold. Dimensionless distance Two parameters α (area ratio) and β (L/D ratio), effect • Friction most effect on the moderate area ratios.
fluid used–air. to study presence of channels. of friction on flow distribution. • Friction increases maldistribution in combining.
greater
• βmaldistribution. than π/2, friction reduces
• β less than π/2, friction increases maldistribution.

Kikas [136] Analytical: laminar flow, fluid Z– and U–type flow arrangement. Number of channels Experimental pressure loss coefficients. • Z– is better than U–type design.
used–water. investigated–10 and 20. Isothermal non–isothermal. • Independent on channel numbers, rather position
of channels.
Sparrow et al. [137] Quasi–analytical with Number of channels investigated–20. Overall pressure drop and volumetric flow rate with flow • Overall pressure drop increases with volume flow.
numerical. pattern.
Fang et al. [138] Analytical and experimental: Number of channels investigated–15 for dividing and 8 for Pressure distribution. • The model can accommodate local disturbances in
turbulent flow. combining. segments like T–junction.
• Pressure distribution in manifold in good
agreement with the experiments.
Maharudrayya et al. Analytical and numerical: Z– and U–type configurations. Number of channels Manifold to total channel area ratio, smaller channels • Smaller channels width increases pressure drop
[139] laminar flow. investigated–38 and 75. Reynolds number tested–58, 233, width for same header areas increases channel number and enhancing flow distribution.
584 and 1871. from 38 to 75. U–type has monotonic flow variation while Z–type
• has minimum flow in the center.
• U–type flow distribution is independent of flow rate.
Table 7 (continued)
• U–type better than Z–type for higher flow rates.

Wang [140] Analytical model. Flow distributing manifold. Three parameters of friction losses, length to diameter • Uniformity can be achieved when
ratio and channel to manifold area ratio. a) Reduce friction losses of bend and friction ζ,
b) Decrease E=L/D ratio for manifold for better

uniformity,
c) Decrease M=(Aport*n/AManifold) ratio, it reduces
the port size to force flow in initial channels.
• Effect of E not very significant effect as compared to
M.
Wang [141] Analytical model. U–type. Application of model. • Good agreement with previous work, more
generalized model.
• Friction increases the pressure drop, and
momentum effect increases the pressure.
Wang [142] Analytical model. Z–type. Application of model. • Good agreement with previous work, more
generalized model.
• Smaller values of M results in better flow
distribution for the Z–type flow.
Wang [143] Analytical model. Single serpentine, multiple serpentine and Z–type Application of the model to design of fuel cells, single • Pressure drop and flow uniformity increases, for a
arrangement. serpentine longest single channel, Z–type most number decrease in channel number.
of smallest channels. • Balance between the number of channels and
channel length for optimum design.
983
Wang [144] Review paper and analytical Review on different flow arrangements with application to Two additional parameters studied after previous work a) Ratio of the areas of the inlet header to the outlet
model. fuel cells. (Wang [140]). header, (A/Ae).
b) Pitch between channels, S.
Wang and Wang [145] Analytical model. Z– and U–type flow arrangements. Comparison of Z– and U–type flow arrangement. Z–type flow more sensitive to structural
• parameters.
• U–type was found to better for small resistance ζ
and larger M.
• Z–type was found better for larger ζ and smaller M.
• Decrease in E reduces the pressure dro with an
increase in flow uniformity.
Fig. 2. Schematic for (a) conservative and (b) bifurcation type manifold structures (Amador et al. [54]).
Fig. 3. Eight flow distributing layouts considered by Alvarado et al. [49].
manufacturing tolerance, an assumed design tolerance of 5% was conductivities, the effect of maldistribution is reduced as conduction
considered by the authors. It was found that for the uniform manifold contributes significantly to heat transfer.
design, the impact of manufacturing tolerance is not very significant, To study the effect of rounded corners and circular bifurcation
an increase in non-uniformity by five times results in just 0.5% change channels, Liu et al. [56] studied three configurations with a base 90°
in standard deviation. For smaller hydraulic diameter non-linear bend in bifurcation design. The modification to this base case was to
manifold, the change in standard deviation is from 19.19% to round off the bends while the second modification was to make the
25.12%, while for large diameter manifold the change is from 1.90% channels circular. The results showed that a longer length of port after
to 5.43%. For 128 channel optimized linear manifold shape, the flow a bend assists in flow uniformity. Also, the channel length-to-width
standard deviation was 38.88% without the manufacturing tolerance, it ratio should be higher for upstream channels as they have higher
increased to 65.44% when the design tolerance is accounted for. For Reynolds number, with Reynolds number decreasing downstream due
the bifurcation type, the impact of design tolerance is small. Their to the split. Finally, it was found that rounded corner design was the
analysis of design tolerance suggested that constant manifold should be best design for flow distribution, with length-to-diameter ratio of five
used for large or unknown manufacturing tolerances especially in cases for all segments.
when channel blockages are also expected. For smaller manufacturing An analytical comparison was presented by Chen and Cheng [57]
tolerances, a variable manifold design was suggested. between parallel channel and bifurcation type design. The results were
In addition, Nielson et al. [55] showed the significance of uniform/ for three to five branching levels and fractal dimension (D) from 1.1 to
non-uniform channel size on the performance when subjected to flow 2.0, where the fractal dimension quantifies the relative length of each
maldistribution. It was found that the impact of maldistribution is channel with increasing level of branching. It was found that for fractal
more amplified when there are non-homogeneous channels. For low dimension of two and five branching levels, heat transfer ability
thermal conductivities, there is not conduction between channels; increases by two folds, while the pumping power reduces by about
hence the impact of non-uniformity is significant. At higher thermal 1.8 times. Hence larger size and more branching levels increase the
984
heat transfer ability while minimizing the pumping power required. In

continuation of their work, Chen and Cheng [58] provided experi-
mental verification of the analytical model for a heat sink with
dimensions of 20×20×1.4 mm fabricated on a silicon wafer with
micromachining. It was found that the ratio of pumping power of
bifurcation to parallel flow channel reduces by 2.8 times while thermal
efficiency increases by 2.8 times by increasing the Reynolds number.
Another numerical comparison between serpentine and bifurcation
type design was presented by Senn and Poulikakos [59], for water flow
at Reynolds number of 20 and 200. It was noticed that a lower pressure
loss is for bifurcation in comparison to serpentine for the same surface
area. It was further reported that due to multiple turns in the
bifurcation, there exists secondary flow motion leading to laminar
mixing and better local heat transfer coefficient. Furthermore, heat
transfer coefficient increases as the number of branches increase,
thereby the total heat transfer was enhanced by increasing the number
of branches.
Some design optimization studies were presented by [60,61], and
the results were used by Luo et al. [62] to present an experimental
comparison of having either pyramid type or bifurcation type manifold
at the inlet and outlet of the channels. They used water in laminar flow
region Re=50–260), with pyramid manifold at both inlet and outlet as
the base case for comparison purpose. Improvements were found with
the use of bifurcation manifold; for example, 30% when both inlet and
outlet are bifurcated. For inlet bifurcation and outlet pyramid 15%
improvement was observed and 28% with opposite arrangement. With
regards to pressure loss, it increased to 72.4% from both pyramids to
both bifurcation types. It was concluded that the best design with
regard to both heat transfer and pressure loss is with a pyramid type at
the inlet and bifurcation type manifold at the outlet.
2.1.2. Conservative manifold design

Contrary to bifurcation type manifold, a conservative type design is
simple in fabrication leading to significant efforts in a detailed analysis
of every aspect of the manifold design. These efforts are summarized in
Table 2, with a detailed discussion in this section.
A comprehensive parametric study was conducted by Gandhi et al.
[63], with steam acting as the working fluid. The study was focused on
variations of pipe diameters for inlet and outlet, header and tube size
along with the number of tubes and corresponding pitch. The two
major design variations studied included the presence or absence of a
channel immediately next to the inlet. Third variation was to add two
more tubes to make the total number of tubes ten. Their findings
revealed that the flow maldistribution increased when the tube
diameter and the flow rate was increased. The first layout was showing
most non-uniformity of about 76%, while 3% and 2% non-uniformity
for the second and third type, respectively. It was also reported that by
increasing the number of tubes, the flow non-uniformity increases
while the reduction in tube diameter improves the flow distribution.
The design configuration for ten tubes having header length of 100 mm Fig. 4. The x-velocity contours for a) normal channel b) channels having smaller size
with inlet and outlet diameter of 50 mm and distance between each inlet and outlet (O-Charoen et al. [51]).
tube of 170 mm was deemed the best for uniform flow distribution
while also providing more heat transfer area. The experimental pressure loss was found to be minimum when the manifold size was
investigation of the authors was focused on both water and air as the maximum (1000 µm) with 15° taper. Furthermore, if the size of inlet
operating fluid, and the results were within 5–10% deviation with the manifold is increased, the flow in opposite channels increases and vice
numerical modeling. It was also concluded that the effect of having versa for the outlet manifold.
different fluid did not have a significant effect on the flow distribution. A similar study was provided by Minqiang et al. [66] for a three-
With regards to Z-type flow arrangement, Mohammadi et al. [64] dimensional model having the triangular manifold in Z-type flow
studied triangular manifolds having the inlet and exit port perpendi- arrangement. It was concluded that more uniform flow distribution
cular to the plane of channels (Fig. 5). Reynolds number ranged was obtained by increasing length and depth while decreasing the
between 5 and 25, and numerical simulations were performed with a width of channels, thus leading to higher aspect ratio. Larger manifold
commercial software [65]. It was found that for flow distribution, the along with larger radius of inlet and outlet ports also resulted in better
channel parameters were less significant as compared to the geome- flow uniformity. Furthermore, a symmetrical design leads to lesser
trical parameters of the manifold, which includes the smallest section maldistribution as compared to asymmetrical.
of the manifold and angle of taper. As the Reynolds number becomes A detailed optimization for Z-type flow arrangement in plate-fin
low, the presence of optimum corner angle becomes evident. The micro heat exchangers was provided by Tonomura et al. [67]. The first
985
important aspects to take into consideration for enhanced performance

of heat exchangers. The simulations pointed that two-dimensional
numerical simulations are not representing the true flow behavior, thus
a complete three-dimensional modeling must be used to understand
the overall flow characteristics. The presence of a critical value of
Reynolds number was found in both two- and three-dimensional
simulations, but it was noticed that there was two order of magnitude
difference between the values of critical Reynolds number. This showed
that two-dimensional results were giving a higher estimate of the flow
maldistribution. Stagnation regions were identified to be forming at the
corners of the flow region which can be reduced by utilizing baffle
directing the flow towards the stagnation regions. The thickness of the
baffles had no effect on the flow distribution mainly due to the lower
Reynolds number while increasing the size of both inlet and outlet
manifold showed improved flow distribution up to a particular limit
after which the flow distribution remained constant. Previous models
had the flow inlet and outlet outside the flow channel regions, so the
Fig. 5. The schematic of the structural parameters studied by Mohammadi et al. [64]. design was modified to bring the inlet and outlet in line with the
channel and noticed improved distribution of flow as compared to the
study showed that the longer length results in more uniform distribu- original layout.
tion. The reason is that the pressure drop, in the channels, becomes Solovitz and Jeffrey [69] provided an analytical model to optimize
more significant as compared to the pressure loss in the manifold the flow distribution by equating the pressure loss in all the channels.
region. Next, the impact of manifold width was studied with four cases, They considered the major pressure loss (frictional, Eq. (1)) while
the baseline scenario, double inlet manifold, double outlet, and both ignoring the minor pressure losses (geometrical, Eq. (2)) for the
inlet and outlet with double manifold width as illustrated in Fig. 6. The analytical model.
study further showed that the inlet header increment has an adverse
fLρV 2
influence on the flow distribution while the larger outlet manifold has a Δpmaj =
2D (1)
positive impact on the flow uniformity by reducing the velocity peaks in
the last channels (i.e. opposite to the inlet side). Hence the outlet KLV 2
should be twice the inlet. Subsequently, a corner optimization by Δpmin =
2 (2)
cutting segments of the corner was performed to find the optimum 2
shape. After finding the optimum shape, they proceeded to make the For fully developed condition, f = 64/Re and V = (4 × Q)/(πD ) led
original size same and vary the baffle location according to the to simplified general expression for the major pressure loss,
optimized channel length obtained earlier. This design (termed Type 128μLQ
Δpmaj =
C) was found to give the best flow distribution for the three velocities πD4 (3)
considered.
A similar study was conducted by Griffini and Gavriilidis et al. [68] For uniform distribution, the flow rate in each channel should be
with focus on lower Reynolds number to explore the flow distribution same. Therefore, for five identical parallel channels the inlet flow rate is
in a manifold for several design layouts. To ensure uniform cooling, the five times the flow in each channel. The major pressure loss compo-
residence times as well as the fluid flow rate in each of the channels, are nents in each section along the paths are added, leading to the
following four equations, where the largest segment is not included
as it is common for all paths and has no impact on flow distribution.
128μLm(4)Qc 128μLm(3)Qc 128μLm(2)Qc 128μLm(1)Qc
Δppath,1 = + + +
πDm4,4 πDm4,3 πDm4,2 πDm4,1
128μLcQc
+
πDc4 (4)
128μLm(4)Qc 128μLm(3)Qc 128μLm(2)Qc 128μLcQc

Δppath,2 = + + +
πDm4,4 πDm4,3 πDm4,2 πDc4
128μLm(4)Qc
+
πDm4,4 (5)
128μLm(4)Qc 128μLm(3)Qc 128μLcQc 128μLm(3)Qc

Δppath,3 = + + +
πDm4,4 πDm4,3 πDc4 πDm4,3
128μLm(4)Qc
+
πDm4,4 (6)
128μLm(4)Qc 128μLcQc 128μLm(2)Qc 128μLm(3)Qc

Δppath,4 = + + +
πDm4,4 πDc4 πDm4,2 πDm4,3
128μLm(4)Qc
+
πDm4,4 (7)
Now for each set of two consecutive paths, only one term is different
while all other terms are same. Hence for uniform pressure drop, those
Fig. 6. The different cases studied by Tonomura et al. [67]. two terms should be equal. Equating the terms leads to the following
986
simplification flow distribution for a triangular manifold. The optimization led to the
128μLm(1)Qc 128μLm(N − 1)Qc conclusion that the right angle manifold is the best performing shape
= for identical structural parameters. A further study [71] included the
πDm4,1 πDm4, N −1 (8) effect of having asymmetrical inlet and outlet manifolds using the same
Further simplification of the common terms leads to the expression model and found that better flow distribution occurs for the symme-
for diameter ratio. trical manifold. For asymmetrical manifolds, the taper angle has a very
significant influence on the flow distribution; larger angles enhance the
Dm, N −1 = (N − 1)1/4 Dm,1 (9) distribution.
Similarly, for path two and three, we get the following relation. Another study for triangular manifolds was presented by Vásquez-
Alvarez et al. [72], with four manifold designs i.e., obtuse, isosceles,
(2) Dm, N −2 = (N − 2)1/4 Dm,2
1/4
(10) circular with channels at extremes and circular with baffles at extremes
as shown in Fig. 8. The study revealed that making the manifold
Hence, a general relation can be written as
circular with inlet at the center had better flow distribution character-
⎛ N − k ⎞1/4 istics, but the presence of channels at extremes cause lower flow in
Dm, N − k = ⎜ ⎟ Dm, k them. Further analysis showed that by having baffles at the extremes,
⎝ k ⎠ (11)
flow becomes even more uniform with same velocity at all channels
where m denotes the manifolds, N is the total number of channels, including the extreme channels.
while k is the local channel for which the diameter is to be calculated. In another study regarding the manifold shape, Hassan et al. [73]
To further simplify the ratio, the minimum diameter is used as a compared numerical and experimental results for five channels of
reference and all other diameters are calculated from it using the power circular cross-section with a diameter of 5 cm, spacing of 22 cm and
law manifold diameter of 10 cm. The base case had the manifold with
Dm, k = (k )1/4 Dm,1 constant diameter and results show that the flow has non-uniform
(12)
distribution with maximum pressure occurring towards the last
This rule can be applied to any shape of the channel, as long as the channel in the manifold. The modification of tapered manifold was
hydraulic diameters are used in these expressions. Thus the equaliza- applied, and it was found that the pressure is now more uniformly
tion of major pressure losses in all flow paths resulted in the relation distributed in the manifold, and results in a relatively uniform flow
termed as power law, which is given by Eq. (13) within the channels. The Reynolds number studied were 100000,
150000, and 200000 which are in the turbulent region, and the results
Dk = (k )1/4 D1 (13)
from numerical were found to be matching with the experimental
where k is the manifold segment number corresponding to the channel values. For such higher Reynolds numbers, the effect of inlet flow
number. With the use of this power law, the results show that step-wise variation becomes less significant for the flow uniformity into the
increments in the manifold size result in excellent flow distribution. channels.
Another modification was done by linearizing the stepped changes into Another study on the equalization of pressure within the manifold
a straight line joining the smallest and largest segment. This reduced was presented by Perlmutter [74] with an analytical model for the
the minor losses due to the sudden shape changes, leading to just 0.8% manifold flow. The resulting optimization study was validated by
standard deviation of flow in channels. experiments. The design was focused on making the static pressure
After understanding the utility of triangular manifolds, Pan et al. constant throughout the manifold to ensure flow uniformity. The
[70] studied three triangular configurations, i.e., obtuse, right and momentum balance was used with Blasius law for wall shear stress
acute angle for the manifold (Fig. 7). The resistance networks modeling to come up with an equation to represent the outlet header, and a
was used, dividing the manifold into rectangular channels and calcu- similar momentum balance for the inlet header. The variations studied
lated the corresponding resistance for each segment to determine the were, with and without friction, with parallel- and counter-flow
arrangements. The optimum results were found when the manifold
length-to-height ratio is larger. The overall pressure drops have also
been found to be in accordance with the experimental values.
The experimental determination of pressure variation within the
manifold was presented by Tereda et al. [75] in addition to velocity
variation in the manifold and channels for a plate heat exchanger. A
mandrel was inserted into the manifold with taps available for pressure
measurements. It was found that as the flow rate increases, the
maldistribution of flow increases. This effect was not very significant
when a larger manifold is utilized which also improved the flow
distribution and reduced the pressure loss. It was found that pressure
variation in the manifolds was dominated by the flow branching effects
in comparison with the friction pressure drop.
The effect of manifold width has also received attention for flow
uniformity. In this regard, Chen et al. [76] presented a CFD simulated
results to understand flow distribution in the stack of Proton exchange
membrane. The model considered of 72 channels that were filled with
porous medium, while studying the effect of permeability, inlet velocity
of air along with the manifold dimensions. The findings revealed that
higher inlet velocity increases the flow maldistribution. Decreasing
permeability enhances the flow distribution but it also increases the
overall pressure drop. Also for channels with more flow resistance,
more pressure drop will occur and result in flow uniformity. The
manifold width is a significant parameter, increasing it leads to more
Fig. 7. Three types of triangular manifolds studied by Pan et al. [70]. flow uniformity while reducing the overall pressure drop.
987
Fig. 8. The layout and velocity distribution in the four cases presented by Vásquez-Alvarez et al. [72].
The effect of area ratio between the manifold and the channel has attributed to the static pressure difference along the flow paths which
been studied by Datta and Majumdar [77]. A numerical finite varies due to friction and momentum. For combining manifold both
difference procedure of iterative nature was presented to solve the friction and momentum always decreases the pressure whereas for the
differential equations of continuity and momentum in the longitudinal dividing manifold, friction reduces the pressure while momentum
direction while the integral equation of momentum in the lateral increases it. The study also revealed that area ratio of four produced
direction. The results have been verified by experimental data available better flow distribution, the difference in flow rate was 2.75 times
by Bajura and Jones [78]. The model was applied to U- and Z-type flow between the first and last channels. In continuation of their work, Choi
pattern termed as parallel and reverse flow. The flow in manifold was et al. [80] presented the effect of Reynolds number and the manifold
previously assumed to be parabolic i.e. the end conditions have no width for the flow distribution in cooling modules to be used in
effect on the initial flow. An elliptic model was presented in which the electronic packaging. The largest width ratio of 4.0 resulted in the best
downstream conditions do have an effect on the flow in the upstream. flow distribution, where the flow uniformity ratio was 1.2. This larger
They studied the flow distribution for various set of area ratio and manifold area also results in less overall pressure drop further
friction parameters, with two-phase flow modeling possible by using increasing the benefits. The dependence on Reynolds number was
effective fluid properties. The experimental data used for the validation studied for Re=5, 50 and 250, with high non-uniformity at a higher
and comparison of results was published by Bajura and Jones [78], Reynolds number.
while also including the basic momentum balance and a mathematical In another application on cooling of electronic components, Kim
foundation for the analytical solution of the flow distribution equa- et al. [81] compared rectangular, triangular and trapezoidal manifolds.
tions. The experimental setup used air as working fluid and consisted of Eight channels were used with the Reynolds number variation from 50
eight different flow configurations using 10 or 20 channels with U- and to 300 studied using a numerical solution software. The rectangular
Z-type arrangements. The parameters studied were, varying area ratio manifold has the problem of concentrating the flow in the last channel
and flow resistance, the later was varied by inserting orifice in the flow which leads to a large flow maldistribution. It was found that for the
area. range of Reynolds number considered, triangular manifold resulted in
Another study on the influence of area ratio on the distribution of the best flow distribution. Reynolds number effect on flow distribution
flow was presented by Choi et al. [79] for liquid cooling modules. The was more significant for rectangular manifold as compared to trian-
objective of uniform flow arrangement is to reduce the development of gular, which resulted in virtually uniform distribution for the range of
hot spots in the modules, using a finite element code to numerically Reynolds number considered.
solve the mass and momentum equations for different area ratios. It Another study on area ratio between channel-to-manifold was
was shown that as the area ratio increases, the flow in first and last studied by Jones and Lior [82] with the use of numerical solution of
channel increases while decreasing in the middle channels. This is non-linear equations by neglecting the effect of buoyancy. The study
988
revealed three major flow distribution controlling parameters: (i) the

ratio of channel diameter to the manifold diameter, (ii) the number of
channels, and (iii) the length of these channels. The flow non-
uniformity increased with an increase in the first and second parameter
while it was decreasing with the last parameter. The findings revealed
that the flow in initial channels is lowest and then increases along the
channels in flow direction for a Z-type configuration, while it reduces
for the U-type configuration. The ratio of channel diameter to the
manifold diameter is a significant factor, the value of which should be
kept as small as possible. After fixing this ratio, the maldistribution
increases with the square of the number of channels. The effect of
distance between the channels has insignificant effect for the case
under study. Increasing the length distributed the flow more uniformly
as it decreases the friction pressure changes within the channels. In
conclusion, the pressure drop in the channels should be significant as
compared to the losses of frictional and inertial losses to make the flow
distribute itself uniformly amongst the channels.
A numerical study into non-linear manifolds as a further improve-
ment to triangular manifolds was presented by Tong et al. [83], with
the use of eight different parameters that can affect the flow distribu-
tion. It was found that channel width increments make the flow Fig. 10. Layouts showing a) smaller inlet than the manifold leading to reverse flow in
distribution better. A linear taper was then studied to demonstrate first channel, b) angled inlet, showing recirculation by Tong et al. [83].
that as the Reynolds number increases, the optimum taper angle also
increases to ensure better distribution. A further study included non- experiments were used to modify the friction factor model with the
linear patterns, having concave down, concave up and the combination assumption of inlet velocity to be directly affecting the pressure loss.
of concave-down followed by concave-up section, as shown in Fig. 9. It The flow types 2×1 and 1×2 were numerically modeled, and pressure
was found that the concave down was the best design but the difference drop was found in a good agreement with the predicted values. Rao
between this design and linear design was not significant and linear et al. [85] presented their study on U- and Z-type arrangement. It was
may be preferred keeping in view the fabrication cost of a non-linear found the effect of flow maldistribution is significant on the heat
design. It was found that if a similar taper is introduced in the outlet exchanger performance, with Z-type having more maldistribution in
manifold, this leads to flow non-uniformity thus it is not preferred. comparison to U-type. Increasing number of plates reduces effective-
Then the width of channels was varied, making last channel smaller ness while increasing manifold size reduces flow maldistribution.
than the first channel and found this to be a significant parameter. The Another study on overall pressure drop was carried out by Rao and
flow inlet was made smaller than the inlet manifold (Fig. 10) and found Das [86], with experimental pressure drop as a consequence of non-
a negative impact of the smaller inlet suggesting that the inlet and uniform flow distribution. A comparison between U- and Z-type plate
manifold should be of same dimensions. The inlet at an angle to the heat exchangers showed that fluid flow rate, the number of channels
manifold (Fig. 10) led to reverse flow in first few channels while and their size all contribute to flow maldistribution. The U-type
rounding off the corners of the entrance of each channel had a negative arrangement has better flow distribution, while higher operating
influence discouraging further investigations. temperature leads to more flow maldistribution. Finally, an industrial
The overall pressure drop in the heat exchanger was studied by heat exchanger is studied showing that up to 300 plates, flow
Miura et al. [84], with manipulation of the flow patterns within the heat distribution is normal, but further additions would result in no flow
exchanger. Experiments were performed on 32 different possible fluid in those following channels.
pass patterns including the Z-type. They found the pressure loss to be Wasewar et al. [87] studied manifold shape modification where a
linearly dependent on the number of passes while the effect of flow conventional longitudinal manifold was modified with a curved shape
channel per pass was not clear. The results obtained from the design as shown in Fig. 11. It was found that the flow uniformity index
reduced by 70.71% which is a significant improvement. The velocity
contours in the manifold showed how the diverging manifold changes
the velocity vector direction, thus enabling better flow distribution.
Gunnewiek et al. [88] presented a numerical study of turbulent air flow
in combining manifold. It was found that increasing manifold length to
width ratio increases the pressure loss. A higher suction velocity in the
lower channel was due to buoyancy-driven flows which increase with
the wider manifold, leading to better efficiency in non-uniform flow
than uniform flow.
The effect of different fluids on the flow distribution in a Z-type flow
arrangement was studied numerically by Mohan et al. [89]. It was
found that for same flow rate, the Reynolds number is different for
different gasses due to the variation in the kinematic viscosity. Hence,
the Reynolds number for Hydrogen is much smaller as compared to air
and oxygen. These results are in almost perfect uniformity seen for
hydrogen whereas other gasses showed non-uniformity. Between air
and oxygen, oxygen was more uniformly distributed as compared to air.
It was also found that higher flow rates and smaller manifold size lead
to a non-uniform distribution.
Fig. 9. The three manifold designs considered by Tong et al. [83].
989
as the working fluid with inlet and outlet placed at the opposite
diagonal corners of the heat exchanger. The whole heat exchanger was
divided into smaller segments and then the approximate model was
used to solve for the fluid velocity and pressure drop, and the results
were compared with the numerical solution [50]. The comparison
showed that the results from the approximate model are in good
agreement with the detailed modeling, and can be used to optimize the
flow distribution. The channel length is found to be inversely affecting
the maximum velocity difference, where longer channels lead to better
distribution. The increase in channel width leads to increase in flow
non-uniformity.
The channel width was also found to be dominating factor for flow
distribution by Pan et al. [93] with a mathematical model for
determination of fluid flow characteristics in complex shaped mani-
folds. The analytical model was complimented by an electrical resis-
tance equivalent modeling. The solution validates the analytical
foundation of dividing the manifold into several smaller rectangular
segments. The optimization algorithm results in an optimum channel
width to be between 300 and 500 µm, while it becomes clear that the
dominating factor for flow uniformity optimization is the channel width
for the case under study, while channel depth has an insignificant
effect.
In continuation to the above work, Pan et al. [94] developed a
method to calculate the resistance due to friction in microchannels and
obtained the flow distribution characteristics. The reason of flow
maldistribution is due to the presence of different frictional pressure
Fig. 11. a) Conventional b) modified manifold design by Wasewar et al. [87]. losses in various flow paths. One of their findings was that uniform flow
might be attained for larger channel length and shorter manifold length
with shorter channel width. It was found that microchannel parameters
2.2. Channel parameters are more significant than manifold parameters, and high aspect ratio is
resulting in more uniform distribution. Utilizing this model, it was
As with the manifold design, channel design is also an important suggested a manifold with a centro-symmetric shaped triangle.
consideration in ensuring uniform flow distribution within the heat The effect of channel friction was studied by Weitbrecht et al. [95],
exchanger. A summary of channel design impact on flow distribution is using the technique of Laser-Doppler-Velocimetry (LDV) to determine
given in Table 3, while the detailed discussion is given in the following the channel velocities for a Z-type flow arrangement utilizing water as
paragraphs. operating fluid. It was found that Reynolds number increases the flow
A comprehensive study of flow distributing in channels from the maldistribution. They used an analytical model to study the effect of
manifold was provided by Fu et al. [90]. The study focused on utilizing various controlling parameters and found that the losses in the
two channels and characterized the flow behavior with attention to the combining manifold are larger than the losses in the dividing manifold.
difference between the first and second channel. They studied square It was further found that the friction loss in the channel is the
manifolds with turbulent flow to explore the flow behavior in the controlling parameter for the total pressure loss in the heat exchanger.
channels. In this regard, a laser Doppler was used for experimental The impact of channel height was studied by Alm et al. [96], by the
investigation and CFD for its numerical comparison. The effect of flow development of ceramic micro heat exchanger with regards to its
split ratio and inlet Reynolds number was investigated. It was found design and fabrication in Z-type configuration. They studied the
that these parameters significantly influencing the flow rate in the variation of channel height with increasing flow velocity. It was found
channels. The presence of secondary flow motion downstream of the that for nitrogen gas, the heat transfer efficiency was almost constant
channel was found to be a common phenomenon, especially if the flow with larger height giving better results at a lower velocity and worst at
split ratio is high contributing to significant pressure loss. As the flow higher. For liquid (water), the efficiency was decreasing for higher
splits into the channel, some recirculation occurs for the flow im- velocity, and smaller height better at intermediate velocity, while larger
mediately after the inlet region. The fluid after turning into the channel height better at higher and lower velocities.
becomes more turbulent as compared to the turbulence level in the This impact of channel height was also observed in a new class of
manifold, mainly due to the recirculation region at the channel heat exchangers used in heating ventilating and air-conditioning
entrance, but it changes to laminar as it progresses in the channel. In (HVAC) systems, where the membrane collapses into the channel,
the study by Koh et al. [91], It was found that gas flow in manifold was reducing the available flow area or completely blocking the flow. In
turbulent while it was laminar in channels, this is primarily due to the HVAC systems, the use of liquid desiccant results in a reduction of
smaller diameter of channels as compared to the manifold. The region power consumption by 26–80% in comparison to conventional systems
in manifold just before the channel has the maximum pressure, while [97]. Liquid-to-air membrane energy exchangers (LAMEE) are used in
the region of recirculation in the channel has the minimum pressure. It HVAC systems to simultaneously transfer heat and moisture through
was found that for the second channel, the vortex region is smaller the semi-permeable membrane that prevents the desiccant carry over.
while the flow rate is higher as compared to the first channel. Also, the In these heat exchangers, the membrane is keeping the two fluid
fluid is more turbulent in the second channel as compared to the first separate, which tends to deflect towards the air side owing to the
channel. The overall flow pattern was found to remain same irrespec- pressure differential between the high-pressure liquid and low-pres-
tive of the order of channel. sure air. Abdel-Salam et al. [98] showed the deformed membrane due
The channel width and length are also an important parameter for to the pressure differential and a support grid was introduced to
flow uniformity, studied with an approximate pressure drop model increase the membrane strength and prevent deflection [99]. The
developed by Commenge et al. [92]. Incompressible nitrogen was used membrane deflection reduced flow and contact area, increasing the
990
velocity and pressure drop while increasing the convective heat channels were used, then optimized to 19 by removing the solid no-
transfer. flow regions. Two flow patterns were identified depending on flow
In addition, maldistribution was also found in hollow-fiber LAMEE velocity. Below 30 m/s, the flow rate does not affect the flow distribu-
where the circular fibers vibrate at higher air velocities leading to tion while being governed by the wall friction. At higher than 30 m/s,
contact between the fibers and a reduction in contact surface area. It the momentum effects become significant to affect the flow distribu-
was found that as the air velocity increases, the impact of flow tion. By making the inlet and outlet of double size, the flow distribution
maldistribution becomes severe leading up to 58% reduction in latent deviation reduced four times while doubling the length of the flow
effectiveness at flow rate of 250 m3/h for packing fraction of 0.1 [100] channels reduced the standard deviation by a factor of 1.7. The best
while at lower flow rates of around 0.5 m3/h, about 38% reduction in design was having flow inlet and outlet ports parallel to the channels
performance is observed [101]. These outcomes show the importance producing a standard deviation of 3% for the flow rate of 0.12 m3/h.
of having uniform flow distribution in LAMEE heat exchangers as the The effect of variation of the inlet position for flow distribution in a
impact on performance is significant. parallel channel was studied by Lu and Wang [107], focusing on five
Abdel-Salam et al. [102] also reported a decrease in performance configurations as shown in Fig. 14. The variations included the channel
for increasing the standard deviation of hydraulic diameter (represent- hydraulic diameter of 1, 1.5 and 3 mm with a heat flux of 120 and
ing the effect of manufacturing tolerance) and decreasing the width. It 240 W. When both inlet and outlet were perpendicular to the flow
was indicated that for air channel thickness of 2 mm and standard channels (a), higher inlet flow rate increased the flow maldistribution
deviation variation from 1.2 to 1.8 mm results in a reduction of while having more channels made the flow uniform. The (b) and (d)
effectiveness from 49% to 37%. While for 6 mm channel, only 1.7% arrangements resulted in reverse flow in some of the channel making
drop in effectiveness was observed. An optimum channel width was these arrangements ineffective. The (c) arrangement did not have flow
also observed for each standard deviation in the hydraulic diameter. recirculation and resulted in better flow distribution. Regarding
In another study Abdel-Salam et al. [103] found that for a 3-fluid pressure drop; (a) type resulted in highest pressure drop while (c)
LAMEE, flow maldistribution can lead to performance degradation up type led to the lowest pressure drop. The type (b) showed flow
to 7%. To reduce the flow maldistribution in LAMEE, an experimental recirculation leading to poor performance. The types (a) and (e) had
study on pre-tensioned membrane was conducted, to prevent its flow impinging the surface resulting in better heat transfer perfor-
deformation due to the pressure differential. The experimental mem- mance. The best thermal performance with lower pressure drop was
brane was operated for eight months and no significant deformation of obtained with the (c) arrangement.
the membrane was found, avoiding the flow non-uniformity. Bogojevic et al. [108] studied two configurations whether the inlet/
outlet manifolds are in-line with the channel domain or outside by
2.3. Number and location of inlet/outlet ports comparing between the approximate resistance model with a numerical
solution. It was found that deviations occur for higher Reynolds
The number of inlet and outlet port and their location along with number and variable manifold, as additional pressure losses are
the orientation play a significant role in flow distribution. A summary ignored for the approximate solution. It was found that ports in-line
of literature is given in Table 4, while the discussion is continued with the channels result in a better distribution with an additional
below. benefit of reduced manifold area. The effect of blocking some flow
Huang et al. [104] studied use of additional flow distributing channels and studying the resulting change in flow distribution was
manifold with application to planar solid oxide fuel cell (SOFC). It also studied. It was found that pressure loss increased irrespective of
was suggested modifying the previously used Z-type design (Design I) the location of blocked channels with the significance of number of
with two inlets symmetrically placed from the central axis and one exit blocked channels, not their location. The location of blocked channels
at the center of the manifold, as can be seen in Fig. 12. Design II had affects the standard deviation of flow distribution, where the blockage
two connected inlet manifolds, while Design III has isolated the two of middle channels results in the least deviation. The most deviation
manifolds such that fluid only combines in the exit manifold. Design IV was found for blockages at the opposite side of the inlet port, with
is a modified version of the Design II with ten baffles or guides in the deviation for the adjacent to inlet channels to be in between the other
manifolds to help distribute the flow in the extreme channels. two cases. Experimental evaluation of the heat sink revealed that the
Experiments were performed by mixing water and dye to allow a flow distribution without heat flux was uniform as fluid was a single
charge-coupled device (CCD) camera to record the flow distribution. It phase, but with high heat flux, bubble formation resulted in two phase
was found that the design IV was the best performing design with flow, leading to severe maldistribution.
regards to flow uniformity, which removed the local hot spots and Zeng et al. [109] presented a similar study with two cases, one
power was increased up to 11%. termed right angle manifold where the inlet and outlet are in line with
Influence of multiple inlet and outlets were studied by Balaji and the channels while the oblique angle manifold has the manifold
Lakshminarayanan [105], with a CFD based design configuration for extending beyond the channels. They used a gas mixture of methanol
microchannel applications. A commercial software [65] was used for and water in 1:1.3 M ratio as the working fluid as it has application in
the analysis of locating the outlet at corners while inlet at the center on steam reforming for hydrogen production. It was found that the
opposite side of the channels. However, this design created a dead zone velocity distribution was independent of the operating temperatures
at the corners where the flow was minuscule and susceptible to hot taken to be 400 and 450 °C and the flow distribution remained same
spots. The Reynolds number range of 5–200 was studied while also whether the reaction was occurring or not. The results showed that
varying the fluid density from 1 to 100 kg/m3. After studying various more flow uniformity could be achieved if the right angled manifold is
inlet and outlet combinations, they recommended to have two inlets used. Lastly, the flow uniformity was decreasing for an increase in
and four outlets to ensure uniform flow distribution for the majority of entrance velocity when methanol steam reforming was not taking place
flow conditions. while if the reaction takes place an increase in entrance velocity leads to
The impact of inlet and outlet size was presented by Delsman et al. better uniformity.
[106], with a three-dimensional CFD analysis on flow distribution for A numerical investigation on the impact of asymmetrically placed
microchannels using a commercial software [50]. They studied nine inlet and outlet ports was studied with a channel parameter (G)
different arrangements for the flow distributing microchannel as shown introduced by Kim et al. [110] along with an experimental verification
in Fig. 13. The inlet velocity was varied from 0.1 to 100 m/s Re=1600) of two typical setups. They selected smaller channel length so that the
for a gas having a density of 0.68 kg/m3. Validation of some of the pressure loss is occurring mainly due to the configurations of the
results using turbulent model was found to be within 8%. Originally 58 manifold. They defined a new parameter G = (Li − o / Lm )2 + Wm / Lm
991
Fig. 12. The three manifold designs considered by Huang et al. [104] for flow uniformity.
Fig. 13. The design layouts for inlet and exit ports studied by Delsman et al. [106] and their corresponding relative standard flow deviation.
992
Fig. 14. Schematic of the cold-plate, with different inlet configurations. (a) I-arrangement. (b) Z-arrangement. (c) ]-arrangement. (d) L-arrangement. (e) Γ-arrangement. (f) Channel
configuration of the present simulations. (g) X-Y-Z direction and Dh. Dimensions are shown in mm (Lu and Wang [107]).
(where L is the longer dimension and W is the shorter dimension of the inlet diameter of the second header as given by Eq. (14)
manifold) and identified that the best result for flow uniformity is
obtained when this parameter has a value of 0.8, while keeping the D1(out ) D2(out )
=
Reynolds number up to 33.2. D1(in) D2(in) (14)
3. Perforated plates for secondary flow distribution Zhang and Li [112] presented improvements in flow maldistribu-
tion for an existing heat exchanger with the use of CFD software [50].
In some application, a simple manifold may not be satisfactory The study was focused on an existing arrangement being used in the
enough to lead to a uniform flow distribution. In such cases, secondary industry, which is followed by experimental verification of the simula-
manifolds incorporating baffle plate with openings/holes are utilized. tions. They recommended two stage baffle arrangement to distribute
The flow splits first in the primary manifold, then further distributes in the flow more uniformly (Fig. 16) and provided experimental valida-
the secondary manifold. This leads to an increased pressure drop, with tion. It was demonstrated that at Reynolds number of 1,000, the
improved performance due to uniform flow. A summary of literature is standard deviation of flow in channels reduced from 3.47 to 0.71. Also,
given in Table 5, with the discussion in the following paragraphs. the equivalent diameters should be equal to get even better flow
Jiao et al. [111] provided experimental results for flow distribution distribution.
utilizing a secondary header with five and seven openings as shown in A numerical study including the impact of secondary manifold was
Fig. 15. It was found that as Reynolds number increase, flow presented by Habib et al. [113] for an air cooled heat exchange
maldistribution increases. The use of secondary flow distribution systems. The various parameters studied were including the flow
header with seven opening was found to be the best configuration. velocity, a secondary distributor at the inlet, the diameter, location
Furthermore, it was found that the flow distribution is uniform when and geometry for the nozzle. The study showed that by reducing size of
the ratio of outlet diameter of the first header and inlet diameter of the the inlet, flow maldistribution increased while doubling inlet nozzles
first header is equal to the outlet diameter of the second header and from 2 to 4 has significantly improved the distribution by 62.5%. The
993
Fig. 15. Design layout for the perforations in the secondary header. a) header B is having five while header C has seven openings, b) shows the three-dimensional layout for the
secondary header arrangements. (Jiao et al. [111]).
presence of the second header reduces the flow maldistribution by 50%. placed mid-way in the manifold to distribute the flow evenly. Layout B
For this case, an insignificant effect of inlet flow velocity while slight has holes in line while layout C places the holes in a staggered
effect of the shape of the nozzle was found on flow distribution. arrangement. Good agreement was found between PIV and numerical
The utility of variable sized perforations in the baffle for secondary results, with the layout C found to be the most efficient one in
flow distribution was studied by Wen and Li [114]. The use of different distributing the flow evenly. The punch ratio increment from the
hole diameters in a baffle was proposed, to restrict flow in some central axis resulted in lower vortices. Hence any pressure loss due
channels while allowing more flow in other channels consequently to the presence of secondary baffle was compensated by the reduction
distributing the pressure equally. The baffle consists of a single plate in vortices. Wen et al. [116] presented a similar paper utilizing PIV to
with holes in it to distribute the fluid evenly. This arrangement results study the flow distribution in manifolds. It was concluded that the
in an increase in pressure drop leading to a reduction in mean baffle is cost effective and convenient, resulting in improved flow
velocities. They also studied the effect of hole distribution, baffle distribution. With the use of baffle, the ratio of maximum to minimum
dimension, total area ratio, and effect of punched ratio. Later the effect channel velocity reduced from 23.2 to just 1.8, which is a substantial
of this improved design on the thermal performance was studied. It improvement.
showed that the effectiveness increases with the improved layout. It Lalot et al. [117] focused on an electrical heater to study the effect of
was concluded that reduction of mean velocities is compensated by flow distribution on its performance with a secondary distributor in the
better flow distribution leading to enhanced overall performance. middle of manifold. It was found that for some cases if the inlet is not
A similar arrangement was experimentally investigated by Wen designed properly, it leads to reverse flow in some of the regions rendering
et al. [115] using the technique of Particle Image Velocimetry (PIV) to them ineffective. Eq. (15) was used to find the ratio (η) of highest velocity
find the flow distribution behavior in a plate fin heat exchanger and the to the lowest velocity within the channels, utilizing the average pressure
results were compared with a numerical model. They presented a base drop across the heat exchanger and the inlet dynamic pressure.
case, and two modified layouts for the holes (Fig. 17) in the baffle
1
2
ρV02 + Δpav
η=
Δpav (15)
It is evident from the equation that if the average pressure drop is

larger, it will lead to better flow distribution. If the flow distribution is not
uniform among the two heating elements, then the power has to be
increased to achieve the same temperature. They used air as the operating
fluid with Reynolds number variation from 10,000 to about 40,000 i.e.
within the turbulent region. Their numerical study was performed with
the use of CFD software [118], utilizing the k-epsilon turbulence model.
The use of bigger manifolds has been a way to improve the flow
distribution up to a certain limit. For better flow distribution, a perforated
mesh is used so that the flow get distributed by the mesh/grid before
entering the channels. The use of this type of grid at the middle of
manifold allows the flow to be diverted along the channel axis and then
flows towards the channels with axial velocity component. They applied
Bernoulli's equation to two regions, first for the region between the inlet
tube and channel outlets and second from the outlet channels to the outlet
of the manifold where the flow exits the heater. They applied the velocity
ratio parameter η to a condenser, counter flow heat exchanger, and
crossflow heat exchanger and showed that as the velocity ratio decreases,
the effectiveness reaches to a value of unity, thus showing the impact of
proper flow distribution on the overall performance of the heat exchanger.
The maximum impact was for the cross flow heat exchanger, where flow
maldistribution can lead to a decrease in effectiveness by as much as 25%.
>A similar study on impact of performance with baffled secondary
header was provided by Ismail et al. [119]. They used alternate large and
Fig. 16. a) Type B having five holes and b) type C having seven holes in secondary small holes in the perforated plate similar to the design studied by Wen
distributor (Zhang and Li [112]). et al. [115]. The resulting velocity distribution is shown in Fig. 18, with
994
Fig. 17. Layout B and layout C for the perforated baffles used for flow distribution by Wen et al. [115].
baffles case showing better flow distribution. It was found that the overall Datta and Majumdar [120] developed a finite difference procedure
pressure drop reduced by 34% when the baffle was used, indicating the to model the flow behavior of two-phase medium with and without the
usefulness of baffle. heat transfer effects. The mass and momentum equations were solved
using the numerical iterative techniques, with application to three
arrangements including U-type, Z-type and mixed type (Fig. 19) by
4. Flow distribution with two phase flow varying the boundary conditions. They validated the model by an
experimental layout for water and air mixture representing two-phase
Flow distribution in manifolds with the two-phase flow is a complex flow and found an excellent agreement. The results for mixed were
behavior. Very few analytical models have been developed owing to the found to follow the basic principles and logic while giving the best flow
complexity, with majority of the work focused on experimental studies distribution among the three arrangements. Increasing in heating load
for evaluation of flow patterns and distribution behavior. A summary is led to steam formation in some channels resulting in higher maldis-
given in Table 6, while the discussion is summarized hereafter.
Fig. 18. The velocity contours in a) base case and b) secondary baffle (Ismail et al. [119]).
995
rate as compared to the channel nucleation. In addition to film boiling,

there was some bubble generation from the bulk of the liquid itself. They
observed the maximum velocity of the bubble to be 5.3 m/s and reported
the frequency of pressure drop increased as the temperature of the
surface was increased. The presence of slug flow was observed for
majority of the cases where the surface temperature was high. For
particular conditions, it even led to backflow in some of the channels
which resulted in increased flow maldistribution.
In contrast to boiling flows, Bobbili et al. [122] studied flow
distribution keeping in view the thermal-hydraulic aspects of the
design for condensing flows. A mathematical model is presented,
having flow rate based heat transfer coefficient relation. They validated
their model with ε-NTU assuming pure uniform two phase flow. The
parameters of flow distribution index, pressure and mass flux ratio
have been studied. The cooling water temperature was found to be
increasing if the NTU was increased, along with increasing pressure
leading to higher heat transfer due to condensation. The mass flux ratio
was also found to increase the heat transfer. The flow distribution also
plays a significant role; uniform flow increases effectiveness and
condensation.
Controlling of flow distribution with the use of microvalves in each
channel for two-phase flow at low Reynolds number was studied by
Martin et al. [123], seeded with solid or bubbly impurities. They also
used a direct numerical solution technique to model the bubble as a
solid particle, which is a reasonable assumption for lower Reynolds
number flow and validated the results with experimental data.
Microvalves were incorporated such that if any flow maldistribution
is detected from exit velocities, they actuate accordingly to make the
flow uniform. One application of this model is to the flow maldistribu-
tion due to fouling or clogging of sections of the channels.
A circular baffle in a circular manifold was discussed by Marchitto
et al. [124], with the two-phase flow in 16 vertical channels being fed
by horizontal manifold. It was found that an increase in area ratio
between the manifold and total channel area resulted in an increase in
liquid flow maldistribution while a decrease in gas maldistribution was
found. The change in area was achieved by the placement of nozzles at
the junction between the manifold and channels. They reduced the size
of the inlet to the manifold by making a nozzle inlet resulting in jet
formation in the manifold, this increased the gas superficial velocity
and was found to be beneficial at higher gas velocities, while at
intermediate velocities, a deteriorating case was observed. Hence the
combined use of nozzle at manifold inlet and orifice at channels inlet
would lead to a better flow distribution within the heat exchanger.
Marchitto et al. [125] further mentioned that gas flow rate is the
controlling factor for flow distribution, increasing it improves gas flow
distribution while reducing liquid flow distribution.
A secondary cylindrical distributor with holes was also introduced
by Marchitto et al. [126], who presented upward flow in vertical
channels for air and water mixture with the principal focus being the
use of secondary header in the form of a cylindrical tube with inline
holes on one side. It was found that the distribution was enhanced for
all configuration of the tubes in comparison to its absence. Further
Fig. 19. a) Parallel b) reverse and c) mixed flow manifold used by Datta and Majumdar optimization revealed that best distribution was possible by making the
[120]. holes facing downwards (θ=180° as shown in Fig. 20) and by having
eight openings (8hA or 8hB) instead of 16. The size of the hole was also
tribution between the flow channels. Higher flow rates also result in varied, it was found that smaller hole does increase the pressure inside
steam formation in some channel while not in others. The dryness the manifold, but this has no effect on the flow distribution. Likewise,
fraction was also found to be decreasing with increasing flow rate. the flow distribution is not dependent on the header pressure drop
Balasubramanian and Kandlikar [121] studied flow distribution and which decreases with increase in gas velocity.
pressure drop for a rectangular channel heat exchanger with hydraulic Lee and Lee [127] presented work on protruding channels in the
diameter of 333 µm undergoing flow boiling. Images were used to follow manifolds on the two-phase flow distribution. They compared the
the movement of bubbles while measuring the temperature and results with the base case of no protrusion and found that an optimum
pressure. A relation was established between the pressure drop varia- protrusion depth exists for the best flow distribution. For no protru-
tions and the associated flow patterns. The experiments revealed a liquid sion, more liquid was flowing in first channels, but with protruded
layer on the surface while the bubble moves through the channel as a channels, more liquid started flowing in latter channels. The protru-
slug flow. The nucleation from the liquid film was occurring at a faster sions increase the flow mixing, resulting in reduced effect of variation
996
method. The continuity equation was simplified to

d ṁ
(ρVAm ) = −
dx Lc − c (16)
where Lc-c is the distance between the two consecutive channels and the
mass flow rate is given by
⎡⎛ ⎞ ⎤
ṁ = ρ⎢⎜⎜ ⎟ 2Δp / ρ ⎥A
1
⎢ 1 + K + (fL / D ) ⎟ ⎥ c
⎣ ⎝ eq ⎠ ⎦ (17)
The momentum equation leads to
d ⎛ ρV 2 ⎞ dp fρV 2P
⎜KL ⎟= −
dx ⎝ 2 ⎠ dx 8A m (18)
where KL is the bending loss from the manifold to channel and P is the
perimeter. The experimental verification was carried out for turbulent
flow regime for viscous fluids. Two flow governing parameters were
defined that control the flow distribution in the manifolds
(P1 = NDc2 KL Cd / Dm2 ) and (F1 = 4fLDm / NDc2 ). It was demonstrated that
flow uniformity can be achieved with low values of both parameters.
Pigford et al. [132] presented a model for an analytical solution for
turbulent flow distributing manifolds. The experimental setup was
Fig. 20. The secondary cylindrical flue shown with openings at angle θ for upwards needed to evaluate some flow coefficients, later used in the analytical
vertical flow in channels. The arrangement of with all 16 openings and two configurations model which is based on momentum balance and Bernoulli's equation.
for eight opening are also shown. (Marchitto et al. [126]).
The governing equation that was derived, can be expressed as
in mass quality and larger protrusion led to small localized recircula- ⎛ km − kmin ⎞ kmouts(s − 1) h ⎡ ⎛1 − s ⎞2 ⎤
tion with a distribution pattern. Finally the impact of channel shape for u′u″ − 2⎜ out ⎟uu′ − u′ − m ⎢ u 2 + s ⎜ + u⎟ ⎥
⎝ hc ⎠ hc 2hc ⎢⎣ ⎝ 2 ⎠ ⎥⎦
two phase flow distribution was studied by Premkumar and Mahesh
[128], with circular and square shapes for air and water mixture =0 (19)
flowing in distributing manifolds. It was found that better flow where km is momentum recovery factor that quantifies the portion of
distribution is obtained with square shaped manifold, with the same initial momentum that is conserved as fluid crosses the control volume
volume fraction. Furthermore, an increase in the air velocity results in and s is +1 for U-type and −1 for Z-type of flow field. hc and hm are
an increase in water flow and reduction in air flow in the last channels. friction parameters for channels and manifold, respectively. The
resulting flow distribution was validated by experimental results with
5. Analytical modeling a good agreement, eventually applied to the two design configurations
of U and Z-type. It was concluded that larger manifolds are better while
Analytical models are based on fundamental mathematical formu- for larger channels, the impact of the smaller header is more tolerable.
lations representing the physical behavior. The momentum balance in Bassiouny and Martin [133] presented the first analytical approach
the manifold is used to account for flow branching in addition to the in understanding the flow distribution in a plate heat exchanger. Their
governing flow equations. A summary of the analytical models devel- model was useful in calculating the velocity and pressure distribution
oped for flow distribution manifolds is given in Table 7 while the in the manifolds as well as the flow distribution among different
forthcoming text provides a detailed discussion. channels along with the total pressure drop in the heat exchanger. The
Acrivos et al. [129] presented work on distributing and combing procedure started with the mass balance and momentum balance and
manifolds where they assumed that downstream conditions have no then utilized Bernoulli's equation to formulate the basic fundamental
effect on the upstream flow. They came up with an iterative method to flow distribution equation. For the distributing type manifold, the
solve the governing set of equations and found that the larger area of governing equation was simplified to
manifold results in better flow distribution. They also pointed that as
1 dP f 2 dU
the flow leaves the manifold, there is some increase in pressure termed + U + (2 − β )U =0
ρ dx 2D dx (20)
as pressure recovery due to decrease in velocity along the manifold. It
was found that non-uniformity depends on this pressure recovery and Similarly, for the combining type manifold, the governing equation
the frictional pressure drop. can be expressed as
Kubo and Ueda [130] presented formulae to calculate the flow rate
1 dP* f * *2 dU *
ratio for a dividing and combining manifold, with the use of two − U + (2 − β*)U * =0
dimensionless parameters of divided flow ‘σD’ and confluent flow ‘σc’.
ρ dx 2D* dx (21)
These parameters included components like pressure recovery factor, where β is the ratio of axial velocity component to total velocity in the
the resistance coefficient of branch pipes and the area ratio of header to manifold and ‘*’ is used for the parameters of the outlet manifold. A
the pipe. They further presented equations and procedure to solve the parameter “m” was defined whose value is based on the type of flow if
flow distribution when the inlet is in line with the channels rather than the flow rate is decreasing in distributing manifold m2 will be positive
perpendicular. The results were validated against experiments per- while it would be negative for flow gathering manifold. It is expressed
formed using water as the working fluid, and a good agreement was as,
found.
⎡⎛ ⎤
2 − β * ⎞⎛ A m ⎞
2
2 − β ⎛ NAc ⎞
2
Majumdar [131] presented an analytical model for combining and
m 2 = ⎢⎜ ⎟⎜ ⎟ − 1⎥ ⎜ ⎟
dividing the flow header for inviscid flow. The continuity, momentum, ⎢⎝ 2 − β ⎠⎝ Am* ⎠ ⎥ ζc ⎝ Am ⎠
⎣ ⎦ (22)
and orifice equations were applied to a control volume for both dividing
and combing manifold and channels and solved using finite difference where ζ is the average total head loss coefficient for channel flow. A zero
997
value for m2 means purely uniform distribution. Therefore, based on

this model the value of m2 should be kept below 0.1 to ensure uniform
distribution. The results for U-type heat exchanger suggested that there
exists an area ratio of inlet to outlet manifold, which would lead to
uniform flow. Hence by varying the area ratio, uniform flow can be
obtained for almost all heat exchangers. In continuation to the
analytical modeling of flow distribution in U-type heat exchanger,
Bassiouny and Martin [134] presented analytically modeling for the Z-
type of flow arrangement. It was concluded that the difference occurs
only when the m2 value is negative, otherwise the results for U and Z-
type are identical.
Shen [135] focused on combining and dividing manifolds to study
the effect of friction on flow distribution. They used the analytical
model of governing equations to determine the flow distribution in the Fig. 21. Momentum balance on the manifold with branching channel (Wang [140]).
channels and noted that as the fluid velocity varies, friction factor also
varies accordingly. Thus a constant friction factor assumption may not in Fig. 21, results in the following equation:
be applicable for lower Reynolds number. However, in the turbulent
⎛ dp ⎞ ⎛ du ⎞2
region, friction factor variation follows the 1/4th power of velocity and pAm − ⎜p + Δx ⎟Am − τW PhΔx = ρAm ⎜um + m Δx ⎟ − Am um2 + ρAc Vcuc
⎝ dx ⎠ ⎝ dx ⎠
its effect is less significant. Therefore, a constant friction factor may be
a good approximation for turbulent flow regions. They introduced a (23)
new parameter λ which is a ratio of the friction loss to the turning loss. Using mass balance and shear stress formulation, this equation is
This factor is then used to observe the effect of friction on the flow further simplified to the following governing differential equation:
maldistribution. It was found that for dividing manifolds, maldistribu-
1 dP f 2 du
tion may decrease or increase based on the area ratio of the channel to + um + (2 − β )um m = 0
ρ dx 2Dm dx (24)
the manifold while for combining manifold the friction always results
in additional flow maldistribution. To find the pressure loss, Bernoulli's equation is applied to
Kikas [136] studied solar water heating systems using variable and incorporate the entrance, channel and bend loss terms resulting in:
constant temperature fluid at lower Reynolds number having reverse
1 ⎛ L ⎞
and direct circuits. The junction pressure losses were obtained from the P − Pc = ρζuc2, ζ = ⎜1 + KL + fc c ⎟
experimental data available while modeling the flow as algebraic 2 ⎝ Dc ⎠ (25)
equations. The model revealed that the flow uniformity does not
Introducing the above pressure loss equation in Eq. (24) and
depend on the number of channels or the total flow rate in the
introducing two coefficients to get:
manifold, rather the flow distribution is only a function of the location
of each channel with regard to the other. The application of such model dw d 2w dw
+ 3Qww + 2R w w 2 = 0
in laminar flow range can be used as an initial guidance to estimate the dx dx 2 dx (26)
performance of the flow in the turbulent region. The pressure loss
(2 − β ) ⎛ Ac N ⎞ fL ⎛ Ac N ⎞
2 2
occurring in the branching flow needs some further investigation to be where Qw = ⎜ ⎟ and R w = − ⎜ ⎟
3ζ ⎝ Am ⎠ 4Dmζ ⎝ Am ⎠
understood properly, while some parametric studies of flow rate and
The solution of (26) yields the following equation for the flow rate
non-uniform temperature have not been included in the study.
in the channels [140]:
Sparrow et al. [137] studied a combining manifold while presenting
a virtual analytical model on spreadsheet to predict the flow distribu- ⎛ ⎛ x ⎞⎞
tion. The model was validated with numerical results showing the vc, i +1 = ⎜0.5exp⎜ −Bi i +1 ⎟⎟
⎝ ⎝ 2 ⎠⎠
closeness of 1% for local and 4% for global parameters. The upstream
⎛ (B (sin( 3 J 1 − xi +1 ))) + ( (3) J (Cos( (3) J 1 − xi +1 ))) ⎞
channel flow results in flow circulation while the downstream jets get ⎜ i i 2 i i 2 ⎟
deflected along the manifold channel because of the upstream manifold ⎜ sin ( (3)
Ji
) ⎟
⎝ 2 ⎠ (27)
fluid flow. Fang et al. [138] used a similar discrete model to find the
analytical evaluation of the pressure distribution within the manifold. This analytical model was earlier validated in two separate articles.
Their work was for combining and dividing manifold sections and The application on U-type manifold was done by Wang [141]. This
comparison with experimental results for the selected cases. model includes both frictional and inertial effects on flow distribution.
Maharudrayya et al. [139] provided an analytical solution, which is The effect of friction and momentum was found to be working opposite,
validated by numerical solution, for U- and Z-type of flow patterns, where friction increases the pressure drop and momentum effect
while studying the effect of area ratio and Reynolds number on the flow increases the pressure. Hence a balance of these two effects can result
distribution. The individual channel size was reduced with increasing in uniform flow distribution. They showed that Bassiouny and Martin
number of channels for the same header dimensions and found that an [133] model was neglecting friction effect while Maharudrayya et al.
increase of pressure drop with better flow distribution. The U-type has [139] neglected the momentum effect. The model application on Z-type
monotonic flow i.e. more flow in first channels and less flow in the last was studied by Wang [142] and compared with the existing models
channels, in contrast to Z-type which shows less flow in the middle revealing the generality of this model. He found that smaller values of
channel and more flow in both side of the channels. It was further channel to manifold area ratio (M=(Ac N )/ Am ) results in better flow
noticed that Z-type has flow distribution dependent on flow rate, while distribution for the Z-type flow.
U-type is independent of flow rate. That is, U-type has better flow Wang [140] discussed three main parameters in the current model
distribution as compared to Z-type for higher flow rate. that control the flow distribution and pressure drop, showing that
Wang [140] unified the models available for flow distribution better flow distribution can be obtained if the combined losses of
within the manifolds. He used the momentum balance and friction and bend (ζ) are reduced by smooth surfaces and rounded
Bernoulli's equation to present an analytical model to calculate
different flow parameters within the manifold and flow channels. The
momentum balance in the manifold with a branched channel, as shown
998
bends. Furthermore, a decrease in parameter E (the length to width finite difference method, and then an iterative method of solution is
ratio of manifold) i.e. larger manifold width leads to better uniformity, used to reduce the residuals. The residuals are typically monitored and
but the impact is not as significant in comparison to the area ratio kept to a value less than 10−6 for all equations. Both laminar and
parameter M. A decrease in M value reduces the port size to force flow turbulent regimes can be modeled in these softwares. In general, the
in initial channels. pressure at outlet can be taken to be zero gauge pressure while the inlet
The application of this model to design and analysis of fuel cell has was based on mass flow rate or velocity inlet based.
been discussed by Wang and Wang [143], by providing the steps that
must be followed to use the model for the solution of a particular
design problem. They studied single serpentine, multiple serpentines 6. Concluding remarks
and straight parallel channel configurations (Fig. 22), where a serpen-
tine acts as single channel twisted to cover the whole area, while A micro-channel heat exchanger has seen numerous applications on
multiple channels reduce the individual channel lengths for the same different scales, as studied by several researchers that are discussed in
base area. If the channel length was increased, it would result in this paper. Most of the work is focused on making the flow uniform
reduction of channel number. It was found that for a decrease in across all the channels, ensuring flow uniformity is the first step in
channel number, pressure drop and flow uniformity increases. A ensuring thermal uniformity. The flow uniformity reduces the pressure
balance between the number of channels and channel length is needed losses along with the enhancement in thermal performance and
to result in the lowest pressure drop with acceptable flow distribution effectiveness of these heat exchangers.
for application to fuel cell. It has been found that in the design of manifolds, bifurcation design
Wang [144] further reviewed the flow field distributions for fuel is found to be an excellent choice for flow distribution in the channels
cells and discussed numerous design configurations. He concluded that but the design is rather complicated as compared to a single manifold.
the flow distribution and pressure drop could be adjusted by the five For single flow distributing manifold, the right-angled triangular
characteristic parameters, namely, design has been found to be better for uniform flow distribution in
comparison with a rectangular section. A parabolic shaped manifold
⎛ L ⎞ was also found to be useful, but the benefit was negligible as compared
1) The ratio of total loss coefficient of channels, ζ ⎜=1 + KL + fc Dc ⎟.
⎝ c⎠ with fabrication cost. Circular manifolds with inlet/outlet in the center
2) The ratio of header length to header diameter, E (=Lm/Wm). are also studied but are difficult to fabricate limiting their practical
3) The ratio of areas of the inlet header to the outlet header, (Am, in/ application. The use of secondary header/baffles increase the total
Am, out). pressure drop, but the impact on flow distribution normally overcomes
4) The ratio of the total flow area of channels to the cross-sectional the pressure loss impact. U-type of flow arrangement has been found to
area of the header, M (=Ac N / Am ). generally having more uniform distribution as compared to Z-type of
5) The pitch between channels, Lc-c. flow arrangement, but for some particular cases, Z-type can lead to
better distribution. The effect of Reynolds number is less significant on
Wang and Wang [145] further utilized this model to compare the U-type as compared to Z-type.
effect of various design parameters on U- and Z-type flow distribution. The channel geometric characteristics are more significant in their
It was found that Z-type flow is in general more sensitive to the effect on the flow distribution as compared to the manifold. Longer
variation of structural parameters. U-type was found to be better for channels make the majority of pressure drop occurring in the channels,
small resistance ζ and larger M, however Z-type was found better for which reduce the impact of pressure variations in the manifolds. For
larger ζ and smaller M. Also a decrease in E reduced the pressure drop smaller channel length, the manifold design becomes critical.
with an increase in flow uniformity. Regarding the channel dimensions, wider channels with smaller depth
For numerical modeling using simulation software [50,65,118], the is found to be leading to a uniform distribution. The channel was also
flow within the channels is modeled by using the mass and momentum found to be more turbulent than the manifold region. A higher ratio of
conservation equations: manifold cross-section area to total channel cross-sectional area also
Continuity equation: results in better flow distribution.
→ For two-phase flow, slug flow behavior was found, with flow
∇. V = 0 (28) nucleation occurring in the heating application. The presence of bubble
Momentum equation: increase non-uniformity, to the extent of reverse flow in some cases.
The analytical modeling has mainly been focused on flow resistance
→→ →
∇ ⋅ (ρV V ) = −∇p + μ∇2 V (29) network modeling and the use of basic Navier-Stokes equations. The
analytical momentum balance approach is found to be useful, particu-
Energy equation
larly to increase the range of applications. It's coupling with the energy
→ equation is expected to provide better understanding of the heat
ρCp( V .∇)T = k ∇2 T (30)
exchanger performance with regard to both fluid and energy distribu-
The discretization is carried out by using finite volume method or tion characteristics. Numerical simulations have recently gained po-
Fig. 22. The three flow field configurations, a) single serpentine with the longest channel length, b) multiple serpentines, with four channels and comparatively smaller channel length,
and c) a conventional Z-type arrangement with smaller channel length and a maximum number of channels (Wang and Wang [143]).
999
pularity, with results matching those of the analytical studies along Heat exchange at this level is the way forward for applications like
with the ability to extract almost any type of flow or thermal electronic chip cooling, where it becomes necessary to dissipate
characteristics over the whole heat exchanger domain. It is found that localized heat generation.
experimental evaluation of heat distribution based on flow maldistri-
bution has gained popularity with the use of particle image velocimetry Acknowledgements
(PIV) and charge-coupled device (CCD) camera. Further studies are
recommended for the advanced and complex geometric layouts. The authors acknowledge support from the King Fahd University of
Petroleum and Minerals (KFUPM) through the project IN151001.
7. Future work directions
References
Following recommendations are made keeping in view the gaps
found in the literature by the authors. [1] Srihari N, Prabhakara Rao B, Sunden B, Das SK. Transient response of plate heat
exchangers considering effect of flow maldistribution. Int J Heat Mass Transf
• Use of temperature dependent properties for the working fluids is

2005;48:3231–43.
[2] Karwa R, Karwa N, Misra R, Agarwal PC. Effect of flow maldistribution on thermal
necessary to have reliable flow models, as the viscosity and density performance of a solar air heater array with subcollectors in parallel. Energy
variations is expected to change the flow resistance affecting the flow 2007;32:1260–70.
[3] Yin JM, Bullard CW, Hrnjak PS. Single-phase pressure drop measurements in a
distribution in micro-channel heat exchangers.
•
microchannel heat exchanger. Heat Transf Eng 2002;23:3–12.
Manufacturing process for bifurcation type of manifolds needs [4] Knight RW, Hall DJ, Goodling JS, Jaeger RC. Heat sink optimization with
enhancement as they have better performance, but are complicated application to microchannels. IEEE Trans Compon Hybrids Manuf Technol
1992;15:832–42.
in design thus limiting their utility. This design is inspired from
[5] Goldberg N. Narrow channel forced air heat sink. IEEE Trans Compon Hybrids
nature, indicating that it is an optimized method to allow uniform Manuf Technol 1984;7:154–9.
flow distribution. [6] Tuckerman DB, Pease RFW. High-performance heat sinking for VLSI. IEEE
• Most of the work assumes fully developed conditions, but the impact Electron Device Lett 1981;2:126–9.
[7] Phillips RJ. Micro-channel heat sinks. In: Bar-Cohen A, Kraus AD, editors.
of developing flow characteristic especially in short channels needs Advances in thermal modeling of electronic components and systems, 2. New
to be studied keeping in view the recirculation regions after any York, NY: ASME Press; 1990.
structural change. [8] Phillips R. Microchannel heat sinks. Linc Lab J 1988;1:31–48.
• Coupling of analytical flow models with heat transfer model to get

[9] Kim SJ, Kim D. Forced convection in microstructures for electronic equipment
cooling. J Heat Transf 1999;121:639–45.
complete understanding of the flow behavior. The attainment of [10] Fedorov AG, Viskanta R. Three-dimensional conjugate heat transfer in the
uniform flow distribution is a step towards uniform temperature microchannel heat sink for electronic packaging. Int J Heat Mass Transf
2000;43:399–415.
distribution. Any non-uniformity in temperature must be strictly [11] Qu WL, Mudawar I. Experimental and numerical study of pressure drop and heat
prevented as it can lead to further complications and performance transfer in a single-phase micro-channel heat sink. Int J Heat Mass Transf
degradation. 2002;45:2549–65.
• Experimental verification should be provided for the representative

[12] Ryu JH, Choi DH, Kim SJ. Numerical optimization of the thermal performance of
a microchannel heat sink. Int J Heat Mass Transf 2002;45:2823–7.
cases from analytical models. Computational models have also been [13] Ryu JH, Choi DH, Kim SJ. Three-dimensional numerical optimization of a
used extensively, and should be complimented by experimental manifold microchannel heat sink. Int J Heat Mass Transf 2003;46:1553–62.
[14] Colgan EG, Furman B, Gaynes M, Graham WS, LaBianca NC, Magerlein JH, et al.
validation.
•
A practical implementation of silicon microchannel coolers for high power chips.
Use of wavy and perforated channels has been explored, with wavy IEEE Trans Compon Packag Technol 2007;30:218–25.
showing better heat transfer at the expense of increased pressure [15] Sarangi S, Bodla KK, Garimella SV, Murthy JY. Manifold microchannel heat sink
design using optimization under uncertainty. Int J Heat Mass Transf
loss. More research efforts are needed for these types of channels
2014;69:92–105.
and hybrid (straight and wavy) channels may be explored in [16] Mathur RK, Mehrotra DR, Mittal S, Dhariwal SR. Thermal non-uniformities in
locations susceptible to hot spots. concentrator solar cells. Sol Cells 1984;11:175–88.
• As demonstrated by various authors regarding two phase flow, the [17] Radziemska E, Klugmann E. Thermally affected parameters of the current–voltage
characteristics of silicon photocell. Energy Convers Manag 2002;43:1889–900.
distribution is very un-predictable and severe non-uniformities are [18] Radziemska E. The effect of temperature on the power drop in crystalline silicon
present. A systematic procedure should be established to explore the solar cells. Renew Energy 2003;28:1–12.
behavior. [19] Coventry JS. Performance of a concentrating photovoltaic/thermal solar collector.
• The field of cryogenics is complex. The temperature range is large

Sol Energy 2005;78:211–22.
[20] Royne A, Dey CJ, Mills DR. Cooling of photovoltaic cells under concentrated
and fluid used is mixture of various components. Two phase flow can illumination: a critical review. Sol Energy Mater Sol Cells 2005;86:451–83.
occur and the behavior of the mixture can be dramatic as compared [21] Amin N, Lung CW, Sopian K. A practical field study of various solar cells on their
performance in Malaysia. Renew Energy 2009;34:1939–46.
to single-phase flow conditions. All these parameters make the [22] Tiwari GN, Mishra RK, Solanki SC. Photovoltaic modules and their applications: a
cryogenic heat exchanger applications, challenging and needs review on thermal modelling. Appl Energy 2011;88:2287–304.
further investigation. [23] Moharram KA, Abd-Elhady MS, Kandil HA, El-Sherif H. Enhancing the perfor-
• For micro-channel heat exchangers, it is generally assumed that the mance of photovoltaic panels by water cooling. Ain Shams Eng J 2013;4:869–77.
[24] Bahaidarah H, Subhan A, Gandhidasan P, Rehman S. Performance evaluation of a
surfaces are smooth and have no impact due to surface character- PV (photovoltaic) module by back surface water cooling for hot climatic condi-
istics. Contributions can be made in the literature to study the tions. Energy 2013;59:445–53.
[25] Du D, Darkwa J, Kokogiannakis G. Thermal management systems for photo-
impact of surface roughness and manufacturing tolerances on flow
voltaics (PV) installations: a critical review. Sol Energy 2013;97:238–54.
distribution in the heat exchangers. [26] Dubey S, Tiwari GN. Thermal modeling of a combined system of photovoltaic
• The advances in LAMEE show the secondary source of non- thermal (PV/T) solar water heater. Sol Energy 2008;82:602–12.
[27] Wu S, Xiong C. Passive cooling technology for photovoltaic panels for domestic
uniformity in the form of deflection of the channel walls/mem-
houses. Int J Low-Carbon Technol 2014;9:118–26.
branes. This shows how multi fluid systems can lead to pressure [28] Cremers J, Mitina I, Palla N, Klotz F, Jobard X, Eicker U. Experimental analyses of
differential on either side of a heat exchanger and need further different PVT collector designs for heating and cooling applications in buildings.
investigation. Energy Procedia 2015;78:1889–94.
• Nanofluids containing phase change particulates can be used in heat

[29] Reddy SR, Ebadian MA, Lin C-X. A review of PV–T systems: thermal management
and efficiency with single phase cooling. Int J Heat Mass Transf 2015;91:861–71.
exchangers to reduce the temperature with added benefit of energy [30] Aste N, Del Pero C, Leonforte F, Manfren M. Performance monitoring and
storage, the particulate have to be small enough so as to avoid modeling of an uncovered photovoltaic-thermal (PVT) water collector. Sol Energy
2016;135:551–68.
settling while having higher specific heat capacity.
•
[31] Ratnakar RR, Dindoruk B, Wilson L. Experimental investigation of DME–water–
With the advancement in technology, nano-scale is the new frontier.
1000
crude oil phase behavior and PVT modeling for the application of DME-enhanced com/comsol-multiphysics〉.
waterflooding. Fuel 2016;182:188–97. [66] Minqiang P, Dehuai Z, Yong T, Dongqing C. CFD-based study of velocity
[32] Evola G, Marletta L. Exergy and thermoeconomic optimization of a water-cooled distribution among multiple parallel microchannels. J Comput 2009;4:1133–8.
glazed hybrid photovoltaic/thermal (PVT) collector. Sol Energy 2014;107:12–25. [67] Tonomura O, Tanaka S, Noda M, Kano M, Hasebe S, Hashimoto I. CFD-based
[33] Shyam GN, Tiwari O, Fischer RK, Mishra IM. Al-Helal, performance evaluation of optimal design of manifold in plate-fin microdevices. Chem Eng J
N-photovoltaic thermal (PVT) water collectors partially covered by photovoltaic 2004;101:397–402.
module connected in series: an experimental study. Sol Energy 2016;134:302–13. [68] Griffini G, Gavriilidis A. Effect of microchannel plate design on fluid flow
[34] Bahaidarah HMS, Baloch AAB, Gandhidasan P. Uniform cooling of photovoltaic uniformity at low flow rates. Chem Eng Technol 2007;30:395–406.
panels: a review. Renew Sustain Energy Rev 2016;57:1520–44. [69] Solovitz SA, Mainka J. Manifold design for micro-channel cooling with uniform
[35] Chiou JP. The effect of nonuniform fluid flow distribution on the thermal flow distribution. J Fluids Eng 2011;133:51103.
performance of solar collector. Sol Energy 1982;29:487–502. [70] Pan M, Tang Y, Yu H, Chen H. Modeling of velocity distribution among
[36] Jones GF. Consideration of the heat-removal factor for liquid-cooled flat-plate microchannels with triangle manifolds. AIChE J 2009;55:1969–82.
solar collectors. Sol Energy 1987;38:455–8. [71] Minqiang P, Yong T, Wei Z and Longsheng L. Flow distribution among micro-
[37] Ghani F, Duke M, Carson JK. Effect of flow distribution on the photovoltaic channels with asymmetrical manifolds, 2007 IEEE International Conference on
performance of a building integrated photovoltaic/thermal (BIPV/T) collector. Sol Control and Automation, Guangzhou, 2007, pp. 193-7.
Energy 2012;86:1518–30. [72] Vásquez-Alvarez E, Degasperi FT, Morita LG, Gongora-Rubio MR, Giudici R.
[38] Laskar J, Nuttinck S, Pinel S. Therm Manag High Power Dev 2004;30332:0–3. Development of a micro-heat exchanger with stacked plates using LTCC technol-
[39] Naewngerndee R, Hattha E, Chumpolrat K, Sangkapes T, Phongsitong J, Jaikla S. ogy. Braz J Chem Eng 2010;27:483–97.
Finite element method for computational fluid dynamics to design photovoltaic [73] Hassan JM, Mohamed TA, Mohammed WS, Alawee WH. Modeling the uniformity
thermal (PV/T) system configuration. Sol Energy Mater Sol Cells 2011;95:390–3. of manifold with various configurations. J Fluids 2014;2014:1–8.
[40] Agrawal G, Kaisare NS, Pushpavanam S, Ramanathan K. Modeling the effect of [74] Perlmutter M. Inlet and exit-header shapes for uniform flow through a resistance
flow mal-distribution on the performance of a catalytic converter. Chem Eng Sci parallel to the main stream. J Basic Eng 1961;83:361.
2012;71:310–20. [75] Tereda FA, Srihari N, Sunden B, Das SK. Experimental investigation on port-to-
[41] Jeong S and Kim T. CFD investigation of the 3-dimensional unsteady flow in the channel flow maldistribution in plate heat exchangers. Heat Transf Eng
catalytic converter, SAE technical paper 971025, 1997. 2007;28:435–43.
[42] Guojiang W, Song T. CFD simulation of the effect of upstream flow distribution on [76] Chen CH, Jung SP, Yen SC. Flow distribution in the manifold of PEM fuel cell
the light-off performance of a catalytic converter. Energy Convers Manag stack. J Power Sources 2007;173:249–63.
2005;46:2010–31. [77] Datta AB, Majumdar AK. flow distribution in parallel and reverse flow manifolds.
[43] Badami M, Millo F, Zuarini A, and Gambarotto M. CFD analysis and experimental Int J Heat Fluid Flow 1980;2:253–62.
validation of the inlet flow distribution in close coupled catalytic converters, SAE [78] Bajura RA, Jones EH, JR . Flow distribution manifolds. J Fluids Eng
technical paper 2003-01-3072, 2003. 1976;98:654–65.
[44] Zygourakis K. Transient operation of monolith catalytic converters: a two- [79] Choi SH, Shin S, Cho YI. The effect of area ratio on the flow distribution in liquid
dimensional reactor model and the effects of radially nonuniform flow distribu- cooling module manifolds for electronic packaging. Int Commun Heat Mass
tions. Chem Eng Sci 1989;44:2075–86. Transf 1993;20:221–34.
[45] Wollin J and Benjamin SF. A study of the flow performance of ceramic contoured [80] Choi SH, Shin S, Cho YI. The effects of the Reynolds number and width ratio on
substrates for automotive exhaust catalyst systems, SAE Technical Paper 1999-01- the flow distribution in manifolds of liquid cooling modules for electronic
3626, 1999. packaging. Int Commun Heat Mass Transf 1993;20:607–17.
[46] Lai MC, Kim JY, Cheng CY, Li P, Chui G, Pakko JD. Three-dimensional [81] Kim S, Choi E, Cho YI. Effect of header shapes on the flow distribution in a
simulations of automotive catalytic converter internal flow, SAE Technical Paper manifold for electronic packaging applications. Int Commun Heat Mass Transf
910200, 1991. 1995;22:329–41.
[47] Lemme CD and Givens WR. Flow through catalytic converters - an analytical and [82] Jones GF, Lior N. Flow distribution in manifolded solar collectors with negligible
experimental treatment, SAE Technical Paper 740243, 1974. buoyancy effects. Sol Energy 1994;52:289–300.
[48] Bejan A, Errera MR. Deterministic tree networks for fluid flow: geometry for [83] Tong JCK, Sparrow EM, Abraham JP. Geometric strategies for attainment of
minimal flow resistance between a volume and one point. Fractals 1997;5:685–95. identical outflows through all of the exit ports of a distribution manifold in a
[49] Ramos-Alvarado B, Li P, Liu H, Hernandez-Guerrero A. CFD study of liquid- manifold system. Appl Therm Eng 2009;29:3552–60.
cooled heat sinks with microchannel flow field configurations for electronics, fuel [84] Miura RY, Galeazzo FCC, Tadini CC, Gut JAW. The effect of flow arrangement on
cells, and concentrated solar cells. Appl Therm Eng 2011;31:2494–507. the pressure drop of plate heat exchangers. Chem Eng Sci 2008;63:5386–93.
[50] Product documentation, Fluent 15. ANSYS Inc.. 2014. 〈http://www.ansys.com/ [85] Prabhakara Rao B, Sunden B, Das SK. An experimental and theoretical investi-
Products/Fluids/ANSYS-Fluent〉. gation of the effect of flow maldistribution on the thermal performance of plate
[51] O-Charoen S, Srivannavit O, Gulari E. Simulation and visualization of flow pattern heat exchangers. J Heat Transf 2005;127:332.
in microarrays for liquid phase oligonucleotide and peptide synthesis. Biotechnol [86] Rao BP, Das SK. An experimental study on the influence of flow maldistribution
Prog 2008;23:755–61. on the pressure drop across a plate heat exchanger. J Fluids Eng 2004;126:680.
[52] Saber M, Commenge JM, Falk L. Rapid design of channel multi-scale networks [87] Wasewar KL, Hargunani S, Atluri P, Naveen K. CFD simulation of flow distribu-
with minimum flow maldistribution. Chem Eng Process Process Intensif tion in the header of plate-fin heat exchangers. Chem Eng Technol
2009;48:723–33. 2007;30:1340–6.
[53] Commenge J-M, Saber M, Falk L. Methodology for multi-scale design of [88] Gunnewiek LH, Brundrett E, Hollands KGT. Flow distribution in unglazed
isothermal laminar flow networks. Chem Eng J 2011;173:541–51. transpired plate solar air heaters of large area. Sol Energy 1996;58:227–37.
[54] Amador C, Gavriilidis A, Angeli P. Flow distribution in different microreactor [89] Mohan G, Rao BP, Das SK, Pandiyan S, Rajalakshmi N, Dhathathreyan KS.
scale-out geometries and the effect of manufacturing tolerances and channel Analysis of flow maldistribution of fuel and oxidant in a PEMFC. J Energy Resour
blockage. Chem Eng J 2004;101:379–90. Technol 2004;126:262.
[55] Nielsen KK, Engelbrecht K, Bahl CRH. The influence of flow maldistribution on [90] Fu H, Watkins AP, Yianneskis M. The effects of flow split ratio and flow rate in
the performance of inhomogeneous parallel plate heat exchangers. Int J Heat manifolds. Int J Numer Methods Fluids 1994;18:871–86.
Mass Transf 2013;60:432–9. [91] Koh J-H, Seo H-K, Lee CG, Yoo Y-S, Lim HC. Pressure and flow distribution in
[56] Liu H, Li P, Van Lew J. CFD study on flow distribution uniformity in fuel internal gas manifolds of a fuel-cell stack. J Power Sources 2003;115:54–65.
distributors having multiple structural bifurcations of flow channels. Int J Hydrog [92] Commenge JM, Falk L, Corriou JP, Matlosz M. Optimal design for flow uniformity
Energy 2010;35:9186–98. in microchannel reactors. AIChE J 2002;48:345–58.
[57] Chen Y, Cheng P. Heat transfer and pressure drop in fractal tree-like microchannel [93] Pan M, Tang Y, Pan L, Lu L. Optimal design of complex manifold geometries for
nets. Int J Heat Mass Transf 2002;45:2643–8. uniform flow distribution between microchannels. Chem Eng J 2008;137:339–46.
[58] Chen Y, Cheng P. An experimental investigation on the thermal efficiency of fractal [94] Pan M, Wei X, Zeng D, Tang Y. Trend prediction in velocity distribution among
tree-like microchannel nets. Int Commun Heat Mass Transf 2005;32:931–8. microchannels based on the analysis of frictional resistances. Chem Eng J
[59] Senn S, Poulikakos D. Laminar mixing, heat transfer and pressure drop in tree-like 2010;164:238–45.
microchannel nets and their application for thermal management in polymer [95] Weitbrecht V, Lehmann D, Richter A. Flow distribution in solar collectors with
electrolyte fuel cells. J Power Sources 2004;130:178–91. laminar flow conditions. Sol Energy 2002;73:433–41.
[60] Tondeur D, Luo L. Design and scaling laws of ramified fluid distributors by the [96] Alm B, Knitter R, Haußelt J. Development of a ceramic micro heat exchanger –
constructal approach. Chem Eng Sci 2004;59:1799–813. design, construction, and testing. Chem Eng Technol 2005;28:1554–60.
[61] Luo L, Tondeur D. Optimal distribution of viscous dissipation in a multi-scale [97] Mohammad AT, Bin Mat S, Sulaiman MY, Sopian K, Al-abidi AA. Survey of hybrid
branched fluid distributor. Int J Therm Sci 2005;44:1131–41. liquid desiccant air conditioning systems. Renew Sustain Energy Rev
[62] Luo L, Fan Z, Le Gall H, Zhou X, Yuan W. Experimental study of constructal 2013;20:186–200.
distributor for flow equidistribution in a mini crossflow heat exchanger (MCHE). [98] Abdel-Salam MRH, Ge G, Fauchoux M, Besant RW, Simonson CJ. State-of-the-art
Chem Eng Process Process Intensif 2008;47:229–36. in liquid-to-air membrane energy exchangers (LAMEEs): a comprehensive review.
[63] Gandhi MS, Ganguli AA, Joshi JB, Vijayan PK. CFD simulation for steam Renew Sustain Energy Rev 2014;39:700–28.
distribution in header and tube assemblies. Chem Eng Res Des 2012;90:487–506. [99] Hemingson H. Preliminary testing for run around heat and moisture exchanger.
[64] Mohammadi M, Jovanovic GN, Sharp KV. Numerical study of flow uniformity and Canada: Department of Mechanical Engineering, College of Engineering,
pressure characteristics within a microchannel array with triangular manifolds. University of Saskatchewan; 2005, summer work report.
Comput Chem Eng 2013;52:134–44. [100] Zhang L-Z, Li Z-X, Zhong T-S, Pei L-X. Flow maldistribution and performance
[65] COMSOL multiphysics®. Version 4.2. COMSOL, Inc.. 2011. 〈https://www.comsol. deteriorations in a cross flow hollow fiber membrane module for air humidifica-
1001
tion. J Memb Sci 2013;427:1–9. [122] Rao Bobbili P, Sunden B, Das SK. Thermal analysis of plate condensers in
[101] Li Z-X, Zhang L-Z. Flow maldistribution and performance deteriorations in a presence of flow maldistribution. Int J Heat Mass Transf 2006;49:4966–77.
counter flow hollow fiber membrane module for air humidification/dehumidifi- [123] Martin M, Patton C, Schmitt J, Apte SV. Direct simulation based model-predictive
cation. Int J Heat Mass Transf 2014;74:421–30. control of flow maldistribution in parallel microchannels. J Fluids Eng
[102] Abdel-Salam MRH, Besant RW, Simonson CJ. Sensitivity of the performance of a 2009;131:111201.
flat-plate liquid-to-air membrane energy exchanger (LAMEE) to the air and [124] Marchitto A, Devia F, Fossa M, Guglielmini G, Schenone C. Experiments on two-
solution channel widths and flow maldistribution. Int J Heat Mass Transf phase flow distribution inside parallel channels of compact heat exchangers. Int J
2015;84:1082–100. Multiph Flow 2008;34:128–44.
[103] Abdel-Salam MRH, Besant RW, Simonson CJ. Performance testing of 2-fluid and [125] Marchitto A, Fossa M, Guglielmini G. Distribution of air–water mixtures in
3-fluid liquid-to-air membrane energy exchangers for HVAC applications in cold– parallel vertical channels as an effect of the header geometry. Exp Therm Fluid Sci
dry climates. Int J Heat Mass Transf 2016. 2009;33:895–902.
[104] Huang CM, Shy SS, Lee CH. On flow uniformity in various interconnects and its [126] Marchitto A, Fossa M, Guglielmini G. The effect of the flow direction inside the
influence to cell performance of planar SOFC. J Power Sources 2008;183:205–13. header on two-phase flow distribution in parallel vertical channels. Appl Therm
[105] Balaji S, Lakshminarayanan S. Improved design of microchannel plate geometry Eng 2012;36:245–51.
for uniform flow distribution. Can J Chem Eng 2006;84:715–21. [127] Lee JK, Lee SY. Distribution of two-phase annular flow at header–channel
[106] Delsman ER, Pierik A, De Croon MHJM, Kramer GJ, Schouten JC. Microchannel junctions. Exp Therm Fluid Sci 2004;28:217–22.
plate geometry optimization for even flow distribution at high flow rates. Chem [128] Premkumar M, Mahesh G. Distribution analysis of fluid in circular and square
Eng Res Des 2004;82:267–73. header. Int J Adv Mar Eng Renew 2015;1:51–7.
[107] Lu M, Wang C. Effect of the inlet location on the performance parallel-channel [129] Acrivos A, Babcock BD, Pigford RL. Flow distributions in manifolds. Chem Eng
cold-plate. IEEE Trans Compon Packag Technol 2006;29:30–8. Sci 1959;10:112–24.
[108] Bogojevic D, Sefiane K, Walton AJ, Christy JRE, Cummins G, Lin H. Investigation [130] Kubo T, Ueda T. On the characteristics of divided flow and confluent flow in
of flow distribution in microchannels heat sinks. Heat Transf Eng headers. Bull Jsme 1969;12:802–9.
2009;30:1049–57. [131] Majumdar AK. Mathematical modelling of flows in dividing and combining flow
[109] Zeng D, Pan M, Tang Y. Qualitative investigation on effects of manifold shape on manifold. Appl Math Model 1980;4:424–32.
methanol steam reforming for hydrogen production. Renew Energy [132] Pigford R, Ashraf M, Miron Y. Flow distribution in piping manifolds. Ind Eng
2012;39:313–22. Chem Fundam 1983;22:463–71.
[110] Kim D, Yu CH, Yoon SH, Choi JS. Effects of manifold geometries on flow [133] Bassiouny MK, Martin H. Flow distribution and pressure drop in plate heat
distribution to parallel microchannels. J Mech Sci Technol 2011;25:3069–74. exchangers-I U-type arrangement. Chem Eng Sci 1984;39:693–700.
[111] Jiao A, Zhang R, Jeong S. Experimental investigation of header configuration on [134] Bassiouny MK, Martin H. Flow distribution and pressure drop in plate heat
flow maldistribution in plate-fin heat exchanger. Appl Therm Eng exchangers-II Z-type arrangement. Chem Eng Sci 1984;39:701–4.
2003;23:1235–46. [135] Shen PI. The effect of friction on flow distribution in dividing and combining flow
[112] Zhang Z, Li Y. CFD simulation on inlet configuration of plate-fin heat exchangers. manifolds. J Fluids Eng 1992;114:121–3.
Cryography 2003;43:673–8. [136] Kikas NP. Laminar flow distribution in solar systems. Sol Energy
[113] Habib MA, Ben-Mansour R, Said SAM, Al-Qahtani MS, Al-Bagawi JJ, Al-Mansour 1995;54:209–17.
KM. Evaluation of flow maldistribution in air-cooled heat exchangers. Comput [137] Sparrow EM, Tong JCK, Abraham JP. A quasi-analytical method for fluid flow ina
Fluids 2009;38:677–90. multi-inlet collection manifold. J Fluids Eng 2007;129:579.
[114] Wen J, Li Y. Study of flow distribution and its improvement on the header of [138] Fang L, Young-hao L, Shi-ming Y. Analytical and experimental investigation of
plate-fin heat exchanger. Cryography 2004;44:823–31. flow distribution in manifolds for heat exchangers*. J Hydrodyn Ser B
[115] Wen J, Li Y, Zhou A, Zhang K. An experimental and numerical investigation of 2008;20:179–85.
flow patterns in the entrance of plate-fin heat exchanger. Int J Heat Mass Transf [139] Maharudrayya S, Jayanti S, Deshpande AP. Flow distribution and pressure drop
2006;49:1667–78. in parallel-channel configurations of planar fuel cells. J Power Sources
[116] Wen J, Li Y, Zhou A, Ma Y. PIV investigations of flow patterns in the entrance 2005;144:94–106.
configuration of plate-fin heat exchanger. Chin J Chem Eng 2006;14:15–23. [140] Wang J. Theory of flow distribution in manifolds. Chem Eng J
[117] Lalot S, Florent P, Lang SK, Bergles AE. Flow maldistribution in heat exchangers. 2011;168:1331–45.
Appl Therm Eng 1999;19:847–63. [141] Wang J. Pressure drop and flow distribution in parallel-channel configurations of
[118] STAR-CD, CD-adapco. 2016. 〈http://mdx.plm.automation.siemens.com/star-cd〉. fuel cells: u-type arrangement. Int J Hydrog Energy 2008;33:6339–50.
[119] Sheik Ismail L, Ranganayakulu C, Shah RK. Numerical study of flow patterns of [142] Wang J. Pressure drop and flow distribution in parallel-channel configurations of
compact plate-fin heat exchangers and generation of design data for offset and fuel cells: z-type arrangement. Int J Hydrog Energy 2010;35:5498–509.
wavy fins. Int J Heat Mass Transf 2009;52:3972–83. [143] Wang J, Wang H. Discrete approach for flow field designs of parallel channel
[120] Datta AB, Majumdar AK. A calculation procedure for two phase flow distribution configurations in fuel cells. Int J Hydrog Energy 2012;37:10881–97.
in manifolds with and without heat transfer. Int J Heat Mass Transf [144] Wang J. Theory and practice of flow field designs for fuel cell scaling-up: a critical
1983;26:1321–8. review. Appl Energy 2015;157:640–63.
[121] Balasubramanian P, Kandlikar SG. Experimental study of flow patterns, pressure [145] Wang J, Wang H. Discrete method for design of flow distribution in manifolds.
drop, and flow instabilities in parallel rectangular minichannels. Heat Transf Eng Appl Therm Eng 2015;89:927–45.
2005;26:20–7.
1002

Siddiqui 2017

Uploaded by

Copyright:

Available Formats

Siddiqui 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Siddiqui 2017

Uploaded by

Copyright:

Available Formats

Renewable and Sustainable Energy Reviews 74 (2017) 969–1002

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

Eﬃcient energy utilization through proper design of microchannel MARK

Nomenclature W width (m)

Author Type of study Specifications Studies conducted Major outcomes

thermal performance with the introduction of a ﬂow non-uniformity

Pressure loss for both bifurcation is 72.4% higher than

Inlet Pyramid and outlet bifurcation is found to be the

in heat transfer coeﬃcient

▪ Inlet bifurcation, outlet pyramid=15%.

best performing. in central tubes 5 and 6, leading to symmetrically decreasing distribu-

them representing the most severe form of non-uniformity. The study

also been published in the literature for photovoltaic and hybrid

Another application of ﬂow distributing systems is in catalytic

eﬀectiveness of catalytic converter. They compared 85 channels with

conversion rates also reduce as the ﬂow maldistribution increases;

where the single channel models over-predict the actual conversion

2. Flow distribution in heat exchanger

Various eﬀorts have been put forward by some authors suggesting

methods of analytical, numerical and experimental nature for ensuring

behavior, and the analytical modeling has been discussed in detail.

2.1. Manifold design

The basic manifold designs are conservative and bifurcation type, as

region directly connected to the channels distributing the ﬂow. The

ground, blood circulation system, and lungs, leading to ﬂow uniformity

Author Type of study Specifications Studies conducted Major outcomes

Author Type of study Specifications Studies conducted Major outcomes

• Higher ﬂow rates increase non–uniformity.

Author Type of study Specifications Studies conducted Major outcomes

• Higher Reynolds number maldistribution more

Author Type of study Specifications Studies conducted Major outcomes

buoyancy–driven ﬂows, possible with wider

Author Type of study Specifications Studies conducted Major outcomes

Author Type of study Specifications Studies conducted Major outcomes

• The total loss is mainly due to friction loss in channels.

Author Type of study Specifications Studies conducted Major outcomes

Author Type of study Specifications Studies conducted Major outcomes

Author Type of study Specifications Studies conducted Major outcomes

2.1.1. Bifurcation type manifold

zero protrusion, more liquid ﬂow into initial

For protruded channels, less liquid ﬂow in ﬁrst

shape has better ﬂow distribution than

air velocity forces more water in the last

circular; volume fraction remains same.

increases, due to increased mixing.

as well as the uniformity of surface temperature. These parameters are

ﬁndings revealed that the distributor/bifurcation type of manifold

results in the lowest pressure loss and surface temperature diﬀerence,

channels parallel to each other is presented by O-Charoen et al. [51],

Distributing manifold. Number of channels investigated–

pressure loss and improves the ﬂow distribution by reducing the

Furthermore, increasing the number of ﬂow scales leads to better ﬂow

distribution. An additional contribution was published by Commenge

et al. [53] including design charts developed to design a multi-scale

A comparison between the two manifold arrangements i.e. con-

of the geometries as electrical resistances forming a network and then

Firstly, by the usage of larger uniform cross-section for the manifold to

Lee and Lee [127]