Applied Thermal Engineering 219 (2023) 119465

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Effect of single-phase flow maldistribution on the thermal performance of

brazed plate heat exchangers
Wenzhe Li a, b, Pega Hrnjak b, c, *
a
School of Energy Science and Engineering, Central South University, Changsha 410083, China
b
Department of Mechanical and Science Engineering, University of Illinois at Urbana-Champaign, 1206 West Green Street, Urbana, IL, USA
c
CTS (Creative Thermal Solutions, Inc.), Urbana, IL, USA

A R T I C L E I N F O A B S T R A C T

Keywords: This paper presents an experimental and numerical investigation of the effect of single-phase flow maldistri­
Brazed plate heat exchangers bution on the thermal performance of brazed plate heat exchangers (BPHEs). A thermal performance model of
Single-phase the BPHEs is developed with the consideration of flow maldistribution among the channels. The proposed model
Flow maldistribution
is validated against the experimental results. The model simulations reveal that the flow maldistribution has a
Capacity degradation
trivial impact on the overall thermal capacity of the BPHEs if two fluid streams are supplied from the same side of
the heat exchangers; if two fluid streams are supplied from the opposite side, the overall thermal performance is
significantly deteriorated due to mismatched flow distributions of two fluid streams. Moreover, other factors are
also proven to act in this issue, including the number of plates, plate length, header/port size, sudden expansion
flow at the heat exchanger entrance, and heat capacity rate ratio of two streams. They either directly change the
flow distribution or potentially affect the thermal effectiveness of the BPHEs. Based on the simulation results,
correlations are generated, which can evaluate the thermal capacity degradation due to flow maldistribution
without the necessity of knowing the detailed flow distribution and building heat exchanger models.

headers and that across the channels. For two-phase flow, in addition to
1. Introduction the pressure drop, the flow regime in the inlet header, or the phase
separation phenomenon in the inlet header also significantly impacts. In
Plate heat exchangers (PHEs) have the merits of compactness and this paper, the effect of single-phase flow maldistribution on the thermal
high heat transfer coefficients and thus have been used extensively in the performance of PHEs is first addressed.
heating, ventilating, air conditioning, and refrigeration industry. In a Many studies have investigated the single-phase flow maldistribu­
PHE, working fluid enters the heat exchanger through the inlet header, tion in PHEs. Most of them characterized the single-phase flow distri­
which is essentially a series of entering ports on the metal plates. The bution by the experimentally measured or numerically predicted
fluid is then distributed into multiple parallel channels for heat transfer. pressure profile in the heat exchangers. Bassiouny and Martin [2,3]
The parallel channels are formed by stacking the thin metal plates, obtained an analytical solution of channel velocity distribution in U- and
typically stamped with wavy chevron or herringbone patterns, and Z-type PHEs by applying the 1-D mass and momentum conservation
assembling them together in different ways (brazing, gasketing, or equation in the headers to correlate the flow rate and pressure profile
welding). Afterward, the working fluid leaves the heat exchanger inside. They also proposed a characteristic parameter, m2, to determine
through the outlet header. In such an arrangement, the flow rate of the flow distribution in PHEs. Later, Wang [15,16] employed the 1-D
working fluid through each channel is not necessarily equal and such a conservation equations in manifolds and developed an analytical
maldistribution can have an adverse influence on the heat exchanger model for the flow distribution in U- and Z-type manifold systems. The
performance [13,9]. developed models included both friction and inertial effects into
Depending on the state of the working fluid, the flow maldistribution consideration. The author claimed that the previous models developed
issue is divided into two categories: single-phase flow maldistribution in Bassiouny and Martin, [2,3,4,7], were special cases of the proposed
and two-phase flow maldistribution. In the case of single-phase flow, the model, after neglecting the friction or inertial effect. The 1-D assumption
flow distribution is mainly determined by the pressure drop along the oversimplified the complex flow condition in the headers. For example,

* Corresponding author.
E-mail addresses: [email protected] (W. Li), [email protected] (P. Hrnjak).

https://doi.org/10.1016/j.applthermaleng.2022.119465
Received 25 April 2022; Received in revised form 11 July 2022; Accepted 9 October 2022
Available online 17 October 2022
1359-4311/© 2022 Published by Elsevier Ltd.
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Nomenclature Eq. (A3)

∊ thermal effectiveness
A area, m2 ζ capacity degradation due to flow maldistribution
b pressing depth, mm ηep end plate efficiency
cp specific heat, J⋅kg− 1⋅K− 1 μ dynamic viscosity, Pa⋅s
C heat capacity rate, W⋅K− 1 or coefficients in Eq. (27) ρ density, kg⋅m− 3
Dp header diameter, mm σ non-dimensional standard deviation of flow distribution in
Df feeding tube diameter, mm Eq. (11)
f friction factor
h convective heat transfer coefficient, W⋅m− 2⋅K− 1 Subscripts
k thermal conductivity, W⋅m− 1⋅K− 1 avg average
Lw plate total width, mm c cold
Lp plate port length, mm ch channel
LMTD log-mean temperature difference, K counter counter-current flow
ṁ mass flow rate, kg⋅s− 1 exp expansion
n exponent in Eq. (20) f frictional
Nch total number of channels h hot
Np total number of thermal plates H header
NTU number of transfer units i inlet
Nu Nusselt number j layer index
P pressure, kPa k plate or channel index
Pr Prandtl number mal maldistributed
Q heat transfer rate, W max maximum
Re Reynolds number min minimum
T temperature, oC o outlet
U overall heat transfer coefficient, W⋅m− 2⋅K− 1 p plate
v velocity, m⋅s− 1 s stream
tot total
Greek Symbols uni uniform distribution
α ratio of heat transfer rate through two adjacent plates in w wall
Eq. (A2)
β ratio of heat transfer rate through two adjacent plates in

it neglected the non-uniform velocity profile in the headers and axial and adopted the flow distribution solutions in Bassiouny and Martin
momentum transfer to the lateral flow when branching. The afore­ [2,3], the simulation indicated that with the flow distribution parameter
mentioned models mostly used some empirical coefficients to correct the m2 increasing, representing a higher degree of flow maldistribution, the
error brought by such an oversimplification, which limits the univer­ thermal effectiveness of the single-pass PHEs significantly dropped (up
sality of the models. Regarding this limitation, Li and Hrnjak [5] built a to 50 % in some extreme cases). Later, the same research group evalu­
mechanistic model to predict the single-phase flow distribution in ated the effect of flow maldistribution on the thermal performance of
brazed plate heat exchangers (BPHEs), in which the effects of non- PHEs with different pass arrangements [9]. Based on their results, the
uniform velocity profile in the header, sudden expansion of the flow multipass arrangements could help overcome the adverse thermal effect
area at the heat exchanger entrance, and axial momentum transfer to the brought by flow maldistribution. More recently, Shokouhmand and
channel flow were mechanistically corrected by tracking the evolution Hasanpour [10] developed a computational method to investigate the
of axial velocity profile in the inlet header. In Li and Hrnjak [5], the effect of the number of plates on the thermal performance of single-pass
single-phase water flow distribution in the BPHEs was also experimen­ U-type PHEs. The authors also used the flow distribution solutions in
tally quantified based on the channel pressure drop measurements, and Bassiouny and Martin [2,3] in the simulation. Their calculation revealed
the results were used to validate the proposed model. the maldistribution-induced heat exchanger capacity degradation
There were limited studies in open literature discussing the effect of ranged from 8 % to 25 %, depending on different chevron angles and the
single-phase flow maldistribution on the performance of PHEs, and numbers of plates in the heat exchanger.
discrepancies were present in their results. Since the uniform distribu­ From the above literature overview, it is confirmed that limited
tion is hardly achieved in experiments, numerical approaches were studies addressed the effect of single-phase flow maldistribution on the
mostly adopted when studying the impact of flow maldistribution. thermal performance of PHEs, and disagreement existed in their results.
Thonon et al. [11] numerically investigated the single-phase flow dis­ Moreover, the existing studies mainly focused on the degradation of the
tribution in PHEs and its effect on thermal and hydraulic performance. overall thermal performance of PHEs, like thermal effectiveness or
Their results showed that the non-uniform flow distribution significantly overall heat capacity, lacking in-depth analysis of heat transfer inside
affects the uniformity of channel outlet temperature, but the influence the heat exchangers. Lastly, though several complicated heat exchanger
on the overall heat exchanger capacity was almost negligible (under the models were developed with the consideration of flow maldistribution,
test conditions, 3 % for the U-type, 6 % for the Z-type). Similar results it is not convenient to quickly evaluate the capacity degradation due to
were discussed in Wang et al. [17], in which the authors attributed the flow maldistribution for heat exchanger users.
trivial difference between the overall thermal performance with and This paper presents a numerical and experimental investigation of
without flow maldistribution to the averaging effect of different channel the effect of single-phase flow maldistribution on the thermal perfor­
outlet temperatures under flow maldistribution. While in the work of mance of BPHEs. The BPHEs are used in this study as a representative of
Rao et al. [8], who established a thermal performance model of PHEs PHEs, but the results can be applied to other PHE designs, like plate-and-

2
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

frame and plate-and-shell. A thermal performance model of BPHEs is In the BPHEs, each CV, other than those of the two channels at the
established, in which the flow maldistribution is accounted for by a ends, exchanges heat with two CVs of two adjacent channels. The heat
previously developed distribution model. The developed model is vali­ transfer rate between two adjacent CVs is calculated by:
dated by the experiments. Several key influential factors are analyzed
Qp,k,j = Up,k,j • Ap,j • LMTDp,k,j (1)
with the emphasis on heat transfer details inside the heat exchangers.
Based on the simulation results, correlations are generated to conve­ 1 1 1
niently evaluate the thermal capacity degradation of BPHEs due to = + (2)
Up,k,j hk,j hk+1,j
single-phase flow maldistribution.
( ) ( )
Tk+1,j,o − Tk,j,i − Tk+1,j,i − Tk,j,o
2. Model description LMTDp,k,j = (
T − Tk,j,i
) (3)
ln Tk+1,j,o
k+1,j,i − Tk,j,o

2.1. Flow distribution model

In Eq. (2), the convective heat transfer coefficient in the channel, hk,j
or hk+1,j , is estimated by an experimentally developed correlation for the
The previously developed single-phase flow distribution model in Li
BPHEs used in this study (Eq. (4)). More details regarding the devel­
and Hrnjak [5] is used in this study. The model predicts the single-phase
opment of the correlation can be found in Li and Hrnjak [6].
flow distribution in PHEs by imposing the condition of equal total
( )0.14
pressure drop for all flow paths starting at the heat exchanger entrance h • De μ
and ending at the exit. When calculating the pressure profile in the Nu = = 0.2913 • Re0.702
ch • Pr1/3 • (4)
k μw
headers, the 1-D mass and momentum conservation equations are used.
To compensate for the error brought by the 1-D assumption, this model with De = 2⋅b, Rech = (b•L
ch,k eṁ
.
•D

mechanistically tracks the evolution of the axial velocity profile in the

w )•μ

Eqs. (1)–(3) are valid for k = 2 ~ (Nch-2). For the heat transfer
inlet header to correct the effects of the non-uniform velocity profile in
through the two thermal plates at the ends, due to the end-plate effect,
the header, sudden expansion of flow area at the heat exchanger
the active heat transfer area is enlarged, thus Eq. (2) needs to be
entrance, and axial momentum transfer to the channel flow when
modified as:
branching.
1 1 1
= + (5)
Up,1,j h1,j • (1 + ηep,1 ) h2,j
2.2. Thermal performance model
1 1 1
The finite volume approach is used to calculate the thermal perfor­ = + (6)
Up,Nch− 1 ,j hNch− 1 ,j hNch,j • (1 + ηep,Nch )
mance of the BPHEs. As shown in Fig. 1, each channel is discretized into
multiple control volumes (CVs) or multiple layers in the plate length where ηep,1 and ηep,Nch are the end-plate efficiency of two end plates. The
direction. Several assumptions are made to simplify the calculation of method of estimating ηep,1 and ηep,Nch is also given in Li and Hrnjak [6].
each CV: (1) flow in the channels is 1-D along the plate length direction
It is worth noting that Eq. (1) adopts the log-mean temperature dif­
so that the temperature change only happens in the plate length direc­
ference (LMTD) method to calculate the heat transfer between two
tion and the convective heat transfer coefficient is constant throughout
adjacent CVs. The LMTD is commonly used to determine the mean
each channel; (2) heat conduction along the fluid flowing direction is
temperature difference between two fluids, of which the temperature
neglected; (3) thermal resistance of the metal plates is neglected due to
change is solely caused by the heat transfer between each other through
small thickness; (4) heat flux through the side walls of each CV is uni­
a wall. However, each CV in the BPHEs exchanges heat with two adja­
form due to small temperature change across a single CV.
cent CVs, except for the CVs of the two channels at the ends. Therefore,
It should be pointed out that the arrows in Fig. 1 do not represent the
the temperature change of the fluid in the CV is a consequence of the
actual direction of heat transfer, instead, they only indicate the “positive
double-side heat transfer, and in such a situation, the applicability of the
direction” of the heat transfer between two adjacent CVs. In another
LMTD method for calculation of the heat transfer rate between two
word, if the actual heat transfer direction is the same as the arrows (from
adjacent CVs, needs to be examined, which is given in Appendix.
the right side CV to the left side CV), Qp,k,j is positive, otherwise,
As mentioned before, for each CV in the BPHEs, other than those of
negative. The governing equations listed below also follow this
the two channels at the ends, the temperature change of the fluid in it is
regulation.
a consequence of heat transfer from two adjacent CVs:
( )
Qch,k,j = ṁch,k • cp • Tk,j,o − Tk,j,i = Qp,k,j − Qp,k− 1,j (7)

The negative sign is added before Qp,k− 1,j since Qp,k− 1,j represents the
heat transfer rate from channel k to channel k-1.
For the CVs of the two end channels, heat is only transferred through
one side, then:
( )
Qch,1,j = ṁch,1 • cp • T1,j,o − T1,j,i = Qp,1,j (8)
( )
Qch,Nch ,j = ṁch,Nch • cp • TNch ,j,o − TNch ,j,i = − Qp,Nch − 1,j (9)

Eqs. (1)–(9) give the governing equations for all CVs in the BPHEs. As
for the boundary conditions, the inlet temperature of each CV is the
outlet temperature of the upstream CV:
⎧
⎪
⎪ Tc,i , k = odd, j = 1;
⎨
Tk,j− 1,o , k = odd, j ∕
= 1;
Tk,j,i = (10)
⎪
⎪ Th,i , k = even, j = NL ;
⎩
Fig. 1. Discretization of channel flow in the BPHEs for heat transfer Tk,j+1,o , k = even, j ∕= NL
calculation.

3
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

In the heat transfer calculation, the thermal properties are estimated experiments, the same average channel mass flow rate (ṁch,avg ) is
based on the average temperature of the inlet (Tk,j,i ) and outlet tem­ maintained in the hot and cold streams of the BPHEs. The schematic of
perature (Tk,j,o ) for each CV, and the pressure is fixed at the atmospheric the corrugated plate and its geometric parameters are given in Fig. 4 and
pressure. Table 2. More detailed facility design and data reduction processes can
An algorithm is developed to solve the above equation system iter­ be found in Li and Hrnjak [5].
atively. As shown in Fig. 2, the temperature profile of the cold water side
is first assumed and fixed. Then, along the flow direction of the hot water 3.2. Validation of the models
(downward), the heat transferred to each hot water side CV, as well as
the temperature change in the CV, is calculated based on Eqs. (1)–(10). The single-phase flow distribution model has been experimentally
The equation system is solved numerically by the nonlinear equation validated in Li and Hrnjak [5].
solver fsolve in MATLAB(R2017a). The cold water side temperature To validate the proposed thermal performance model, the predicted
profile is then updated based on the heat transfer calculation. The up­ and measured overall heat exchanger capacity are compared in Fig. 5.
date is along the flow direction of the cold water (upward). Then, the The model predictions and experimental data agree well with each
next round of the hot water side calculation starts, and the iteration other, with the differences between them being within 2 %. In addition,
continues until the temperature profile of both sides stabilizes (the the absolute temperature change across the channels, measured by the
change in the outlet temperature of each CV between two iterations is probes placed into the headers, are also compared with the modeling
smaller than 0.01 ◦ C). results. As demonstrated in Fig. 6, it seems that the deviations between
the measurements and simulations are not negligible: the absolute dif­
3. Experimental validations ference could reach up to 2 ◦ C with the channel temperature change
ranging from 4 to 18 ◦ C. Moreover, it can be seen in Fig. 6, that the
3.1. Facility model tends to overestimate the temperature change across the chan­
nels. This is related to the flow maldistribution within the channels. As
The same experimental facility as in Li and Hrnjak [5] is used in this shown in the schematic in Fig. 6 and the picture in Fig. 3, when
study, of which the schematic is shown in Fig. 3. Two independent water measuring the channel outlet temperature, the thermocouple tip was
loops (hot stream /cold stream) are in the system. To avoid air pockets aligned with the direction connecting the inlet and outlet ports of the
trapped inside, each loop is charged with distilled water from the plates (vertical direction in Figs. 3 and 6). Along that direction, the flow
highest location with the system being held vacuum. For each loop, a rate is higher and the flow path is shorter, resulting in a smaller tem­
centrifugal pump is used, along with a variable frequency drive, to perature change, compared with the predictions in which uniform 1-D
control the water flow rate. Two immersion electrical heaters are used to flow is assumed in the channel.
adjust the water inlet temperature of the hot stream; for the cold stream,
the water temperature is regulated by tap water and a valve. Coriolis 4. Results and discussions
type flow meters, absolute and differential pressure transducers, and
type-T thermocouples are installed at locations as indicated in Fig. 3. 4.1. General flow distribution profile
The sensors have been calibrated. Their range and uncertainty are listed
in Table 1. Static mixers are installed at the entrance and exit of the Fig. 7 plots the predicted non-dimensional flow distribution profiles
tested BPHEs so that the measurement accuracy of energy balance in the in the BPHEs with different numbers of plates. To compare the degree of
heat exchangers is guaranteed. flow maldistribution under different working conditions, the dimen­
To measure the temperature change and pressure drop across the sionless standard deviation of flow rate through the channels is used as a
channels, four probes are placed into the headers of the BPHEs. The metric. Taking the cold water side as an example:
probe is also shown in Fig. 3. In the probe, a lengthened thermocouple is √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
inserted into a stainless-steel tube, of which one end is blocked and a ( )2̅
1 1 ∑k=Nc,ch
through-hole is drilled on the side surface close to that end for measuring σc = ṁ c,ch,k − ṁ c,ch (11)
ṁc,ch Nc,ch k=1
the pressure in the header. The tip of the thermocouple is bent into a
channel through the open through-hole, to record the inlet/outlet tem­ The results in Fig. 7 are calculated with a cold-side total mass flow
perature of the channel. By moving the probes forward and backward, rate of 0.5 kg⋅s− 1. According to Li and Hrnjak [5], the non-dimensional
the measurement is done for different channels. flow distribution profiles under different total mass flow rates are quite
Three BPHEs with identical plate geometry but different numbers of similar, due to synchronous changes of the pressure drop along the
plates (10, 50, and 100) are tested in the experiments. They are all in a headers and that across the channels. As shown in Fig. 7, along the
1pass-1pass, U-type arrangement. The tested BPHEs are well insulated flowing direction in the inlet header, the channel flow rate first increases
so that the energy balance discrepancy (measured heat transfer rate for the first several channels due to a sudden expansion of the flow area
difference between the hot and cold stream) is within ± 4 %. In the at the heat exchanger entrance. For the rest channels, the flow rate
generally decreases with the distance away from the BPHE entrance/

Fig. 2. Heat transfer calculation algorithm.

4
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Fig. 3. Experimental facility and the pressure and temperature probe.

Table 1 Table 2
Measurement uncertainty. Plate geometry of the BPHE.
Measured Parameter Uncertainty Parameters Value

Temperature (T type) 0.1 ( C)

◦
Chevron angle, φ, ◦
60
Absolute pressure (0–172 kPa) 0.11 % (full scale) Corrugation depth, b, mm 2.0
Differential pressure (0–7.5 kPa) 0.25 % (full scale) Corrugation pitch, Pc, mm 7.0
Differential pressure (0–6.9 kPa) 0.25 % (full scale) Plate thickness, d, mm 0.30
Mass flow rate 0.15 % (reading) Port length, Lp, mm 172
Total length, Lv, mm 206
Calculated Parameter Uncertainty
Port width, Lh, mm 42
Heat exchanger capacity <1.5 % Total width, Lw, mm 76
Heat transfer area per plate, Ap, m2 0.014
Port diameter, Dp, mm 20.0
Feeding tube diameter, Df, mm 16.0

exit. That is associated with the axial momentum transfer to the chan­
nels in the inlet headers. In addition, with more plates in the heat
exchanger, σc increases, meaning the flow maldistribution is more se­
vere. This is because the pressure drop in the headers becomes
increasingly dominant in the total pressure drop with a larger number of
plates in the BPHEs.
With the predicted flow distributions above, the effect of flow mal­
distribution on the thermal performance of BPHEs will be discussed in
the following sections. Several key factors to this issue will be analyzed,
which either directly change the flow distribution or indirectly act by
affecting the thermal effectiveness of the heat exchanger.

4.2. Results for the BPHEs with the “same-side supply” arrangement

ζ, defined in Eq. (12), is used as a measure of the thermal capacity

degradation due to flow maldistribution:
QBPHE,mal
ζ = 1− (12)
QBPHE,uni
Fig. 4. Schematic of the plate in BPHE.
where QBPHE,mal and QBPHE,uni denote the heat exchanger capacity of the

5
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

maldistributed case and the ideally uniform distribution case. With a

higher ζ, the degradation of heat exchanger capacity induced by flow
maldistribution is higher.
Fig. 8 shows the variation of ζ as a function of the number of plates at
three different total flow rates. The simulations are based on the BPHEs
with the “same-side supply” arrangement, meaning the hot and cold
water are supplied from the same side of the heat exchanger, as shown
by the schematic in Fig. 8. The following discussion will reveal the
importance of such an arrangement to this issue. It can be seen that,
under the tested conditions, the capacity degradation due to flow mal­
distribution is very low (within 2.5 %), even with a very larger number
of plates. This result agrees with Thonon et al. [11] and Wang et al. [17].
Though the absolute value of ζ is small, the trend that ζ increases with an
increasing number of plates is clear. This matches with the observation
in Fig. 7 that severer flow maldistribution is present in the BPHEs with
more plates. Meanwhile, when the total mass flow rate increases, the
capacity degradation slightly increases. Because the total mass flow rate
does not change the non-dimensional flow distribution profile too much,
this result might be attributed to the smaller thermal effectiveness of the
BPHE with a higher total flow rate. The BPHE with 180 plates is taken as
Fig. 5. Heat exchanger capacity: experiments vs model predictions.
an example: with the uniform flow distribution, the thermal effective­
ness of the heat exchanger (∊uni ) decreases from 0.72 to 0.67 with the
total mass flow rate (cold side) increasing from 0.25 kg⋅s− 1 to 0.5 kg⋅s− 1.
The smaller the thermal effectiveness under the uniform distribution,
the higher capacity degradation flow maldistribution can cause.
To understand the effect of flow maldistribution on heat transfer
details inside the BPHEs, for each thermal plate in the 180-plate BPHE
(ṁc,tot = 0.5 kg⋅s− 1, ṁh,tot = 0.49 kg⋅s− 1), averaged overall heat transfer
⃒ ⃒
coefficient (Up ), absolute log-mean temperature difference (⃒LMTDp ⃒), as
⃒ ⃒
well as absolute heat transfer rate (⃒Qp ⃒), are calculated and plotted in
Figs. 9–11. For a certain plate (plate k), Up is calculated by averaging
⃒ ⃒
Up,k,j in Eq. (2) over the plate; ⃒LMTDp ⃒ is evaluated based on the inlet
and outlet temperature of two channels adjacent to the plate (channel k
⃒ ⃒ ⃒ ⃒
and k + 1); ⃒Qp ⃒ is the summation of ⃒Qp,k,j ⃒ in Eq. (1) over the plate.
⃒ ⃒
It can be seen from Figs. 9 and 10, that Up and ⃒LMTDp ⃒ per plate
basically follow the profile of mass flow rate distribution, except for
some abrupt changes at two ends due to the end-channel/end-plate ef­
fect. This feature is largely related to the arrangement that the cold and
hot water are supplied from the same side of the BPHEs. In other words,
the flow rate distributions of two streams of the BPHEs match up with
each other, which guarantees the flow rates in two adjacent channels do
Fig. 6. Channel temperature change: experiments vs model predictions. not differ too much. In such an arrangement, when smaller flow rates

Fig. 7. Flow distribution profiles in the PHEs with different numbers of plates, Fig. 8. Capacity degradation due to flow maldistribution: with the “same-side
withṁc,tot = 0.5 kg⋅s− 1. supply” arrangement.

6
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Fig. 9. Averaged overall heat transfer coefficient per plate in the 180-plate Fig. 11. Absolute heat transfer rate per plate in the 180-plate BPHE with the
BPHE with the “same-side supply” arrangement. “same-side supply” arrangement.

theoretical analysis is developed. Here, for the convenience of analysis,

each channel in the BPHEs is assumed to contain only one CV, instead of
multiple CVs as in the thermal performance model in section 2.2. As a
result, each thermal plate in the heat exchanger can be approximately
treated as a counter-flow heat exchanger, of which the absolute heat
transfer rate is estimated by:
⃒ ⃒ ⃒ ⃒
⃒Qp,k ⃒ ≈ ∊p,k • Cmin,k • ⃒Tk+1,i − Tk,i ⃒ (13)

Considering the flow distributions of two streams match up with

each other, and the mass flow rates of two adjacent channels of each
plate do not differ too much, the thermal effectiveness is approximated
as that in a counter-current flow heat exchanger with the heat capacity
rate ratio equal to 1:
NTU p,k
∊k ≈ (14)
1 + NTU p,k

where NTUp,k is the number of transfer units of the plate, calculated by:
Up,k • Ap
NTU p,k = (15)
Cmin,k
Fig. 10. Absolute log-mean temperature difference per plate in the 180-plate
Referring to Eqs. (2) and (4), and the flow rates of the two sides are
BPHE with the “same-side supply” arrangement.
approximately equal (ṁch,k ≈ ṁch,k+1 ), it can be inferred:
( )− 1
flow through two adjacent channels, Up of the plate in the middle is 1 1
Up,k = + ∝hk ∝ṁach,k (16)
reduced due to a smaller convective heat transfer coefficient on both hk hk+1
⃒ ⃒
sides. As for the change of ⃒LMTDp ⃒, it can be explained by treating each
plate and its two adjacent channels as a counter-flow heat exchanger. here, the temperature-induced thermal property change is ignored. a in
When the flow rates through two adjacent channels of a plate decrease, Eq. (16) denotes the exponent on the Reynolds number (Rech) in the heat
the reduction in Up is smaller than that in the flow rates of the two transfer correlation. For the tested BPHEs, a = 0.702, as given in Eq. (4).
channels, therefore, the number of transfer units (NTU) of such a “two- For the Cmin , we can also have:
channel, one-plate, counter-flow heat exchanger” increases. As a result, Cmin,k ∝ṁch,k (17)
the effectiveness of the heat exchanger (∊uni ) is larger, thus, the tem­
⃒ ⃒ It is worth noting that, for most channels, the temperature change is
perature changes of the two channels are larger and ⃒LMTDp ⃒ across the
⃒ ⃒ ⃒ ⃒ caused by the heat transfer from both sides, thus, the effective Cmin
plate is smaller. The synchronous Up and ⃒LMTDp ⃒ profiles lead to ⃒Qp ⃒
should be smaller than either ṁch,k ⋅cp or ṁch,k+1 ⋅cp when treating each
profile also follows the flow rate distribution. It can be seen in Fig. 11,
thermal plate as an independent counter-current heat exchanger.
that the plates near the heat exchanger entrance/exit transfer more heat
However, in a BPHE with the “same-side supply” arrangement, the heat
in the maldistributed case than in the uniform distribution case, while
transfer rates through two adjacent plates are approximately equal for
for the plates near the rear, less heat is transferred through them in the
most channels, as shown in Fig. 11. Therefore, when treating the thermal
maldistributed case than in the uniform distribution case.
plate as an independent counter-current heat exchanger, the effective
However, the local difference in the heat transfer rate per plate be­ ṁ ṁ
Cmin could be estimated by ch,k2 p or ch,k+1 p
, and considering
⋅c ⋅c
tween the uniform and maldistributed cases does not explain the trivial 2

difference in the overall heat capacity. For this purpose, the following ṁch,k ≈ ṁch,k+1 , Eq. (17) is thus justified. Combing Eqs. (15), (16), and
(17):

7
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

ṁach,k ( )
NTU p,k ∝ = ṁa−ch,k1 (18) ∑k=Np ⃒⃒ ⃒ ∑k=Np n
ṁch,k ⃒ k=1 ṁch,k
QBPHE,mal k=1 Qp,k mal
ζ = 1− = 1 − ∑ Np ⃒ ⃒ ≈ 1− ( )mal
Looking back at Eq. (14) and considering two extreme cases: QBPHE,uni ⃒ ⃒
1 Qp,k uni
∑k=Np n
k=1 ṁch,k
NTUp,k ≫1 and NTUp,k ≪1: uni
{ ( )
∑j=Np /2
NTU p,k 1, withNTU p,k ≫1 ṁnch,2j
∊k ≈ ≈ (19) j=1
1 + NTU p,k NTU p,k , withNTU p,k ≪1 ≈ 1− ( )mal (23)
∑j=Np/2
Combing Eqs. (13) and (19) gives: j=1 ṁnch,2j
uni
⃒ ⃒
⃒Qp,k ⃒ ≈ Using the fitted value of n (n = 0.893) in the case of 180-plate, ζ
{ ⃒ ⃒ estimated by Eq. (23) is 1.62 %, which is close to the result in Fig. 8
Cmin,k • ⃒Tk+1,i ⃒
⃒ − Tk,i , withNTU p,k ≫1 (1.93 %). However, the purpose of Eqs. (13)–(23) is not to derive a new
⃒ (20)
NTU p,k ⋅Cmin,k • ⃒Tk+1,i − Tk,i ⃒, withNTU p,k ≪1 equation for estimating ζ, but to theoretically verify that the matched-up
⃒ ⃒ flow distributions of two streams may be the reason behind the almost
Under a certain working condition, ⃒Tk+1,i − Tk,i ⃒ is a constant. Then,
negligible influence of flow maldistribution on the heat exchanger
with Eqs. (17), (18), and (20), we have: overall capacity.
⎧
⎪ More generally, n is bounded in the range a ~ 1. If n approaches its
⃒ ⎨ ṁch,k , withNTU p,k ≫1 ⃒ ⃒
⃒
⃒Qp,k ⃒∝ (21) upper limit (n = 1), then ⃒Qp,k ⃒ is a linear function of ṁch,k and there is a
⎪ ṁach,k , withNTU p,k ≪1
⎩ trivial influence of flow distribution on the overall thermal capacity. The
lower limit of n is a, which is heavily related to the design of the
Eq. (21) demonstrates the relationship between the heat transfer rate chevron/herringbone pattern of plates, as well as the flow conditions.
of a plate and the mass flow rate through its adjacent channel. However, Based on the literature survey by Ayub [1], a could change from 0.33 to
Eq. (21) only represents two extreme cases. With the fact ∊k is a 0.87. With n changing in the above possible range, the variation of ζ for
monotonic function of NTUp,k (Eq. (14)), a more general relationship can the 180-plate BPHE is estimated by Eq. (23). As Fig. 13 shows, though ζ
be written as: first slightly increases and then decreases as n increases, its absolute
⃒ ⃒ value is still relatively small (within 5 %), which generalizes our finding
⃒Qp,k ⃒∝ṁn , a ≤ n ≤ 1 (22)
ch,k that flow maldistribution has a trivial impact on the overall thermal
To validate the above analysis, the modeling results of the 180-plate capacity of heat exchangers if two fluid streams are supplied to the BPHE
BPHE, ṁc,tot = 0.5 kg⋅s− 1, ṁh,tot = 0.49 kg⋅s− 1 are used to find a suitable n from the same side.
in Eq. (22). It turned out that n = 0.893 is best fitted in this case, which is
⃒ ⃒
indeed in the range of a (0.702) ~ 1. The relationship between ⃒Qp ⃒ and
4.3. Results for the BPHEs with the “opposite-side supply” arrangement
ṁnch,k is plotted in Fig. 12. It can be seen in Fig. 12 that most of the data
points (for both maldistributed and uniform distribution cases) lay on a The BPHEs with two fluid streams supplied from the opposite sides,
straight line through the origin, which verifies the derived relationship or the “opposite-side supply” arrangement, are also investigated. The
(Eq. (22)). There are a few outliers in Fig. 11 coming from the first/last results are given in Fig. 14. Compared with the “same-side supply”
several plates due to the end-channel/end-plate effect, and the abrupt arrangement in Fig. 8, the flow maldistribution induced capacity
flow rate change in the first several channels. degradation is significantly higher with the “opposite-side supply”
With most of the data points of both maldistributed and uniform arrangement. Similar to the results in Fig. 8, ζ with the “opposite-side
distribution cases laying on the same straight line through the origin, the supply” arrangement also increases with an increasing number of plates,
capacity degradation due to flow maldistribution can be approximated as flow maldistribution becomes severer. However, three curves under
as follow: different total flow rates seem to overlay each other. This might be
attributed to a relatively trivial influence of the change in overall ther­
mal effectiveness compared with the effect of the “opposite-side supply”
arrangement.
⃒ ⃒ ⃒ ⃒
Figs. 15–17 plots Up , ⃒LMTDp ⃒, and ⃒Qp ⃒ in the 180-plate BPHE with
the “opposite-side supply” arrangement. As Fig. 15 demonstrates, for
half of the plates near the heat exchanger entrance/exit, Up with the
“same-side supply” arrangement is significantly higher than that with
the “opposite-side supply” arrangement; while for the rest plates near
the rear, though a higher Up is present with the “opposite-side supply”
arrangement, the difference between these two arrangements is much
smaller. This feature could be attributed to that Up of a plate is signifi­
cantly determined by the smaller h of the streams in two adjacent
channels. That means, for a plate, when the flow rates in two adjacent
channels significantly differ, a small h of the low-flow-rate stream is
dominant and overwhelms the benefit of a high h brought by the high-
flow-rate stream. The shortcoming of the “opposite-side supply”
⃒ ⃒
arrangement is more obvious in the comparison of ⃒LMTDp ⃒ in Fig. 16.
⃒ ⃒
For most of the plates in the BPHE, ⃒LMTDp ⃒ with the “opposite-side
supply” is smaller than that with the “same-side supply” arrangement.
Especially for the plates near two ends with a very small heat capacity
rate ratio, in which the temperature of the low-flow-rate stream quickly
approaches the inlet temperature of the high-flow-rate stream and
⃒ ⃒ ⃒ ⃒
Fig. 12. Relationship between ⃒Qp ⃒ and ṁ0.893 ⃒LMTDp ⃒ is remarkably reduced. As a consequence of the disadvantages
ch,k in the 180-plate BPHE.

8
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Fig. 13. ζ estimated by Eq. (23) vs n (for the 180-plate BPHE).

Fig. 15. “opposite-side supply” arrangement vs “same-side supply” arrange­

Fig. 14. Capacity degradation due to the flow distribution: with the “opposite- ment: averaged overall heat transfer coefficient.
side supply” arrangement.

⃒ ⃒ ⃒ ⃒ dominant. The capacity degradations with different header diameters

in Up and ⃒LMTDp ⃒, ⃒Qp ⃒ with the “opposite-side supply” arrangement are given in Fig. 19. The calculations are based onṁc,tot = 0.5 kg⋅s− 1,
shows an inferior result compared with the “same-side supply” ṁh,tot = 0.49 kg⋅s− 1. Fig. 19 shows that with a smaller header, the ca­
arrangement (Fig. 17). pacity degradation is higher as severer flow maldistribution is present.

4.4.2. Effect of the plate length

4.4. Other factors affecting the flow maldistribution induced thermal
The plate length (Lp) is another crucial parameter to the single-phase
capacity degradation
flow maldistribution. A longer plate can improve the flow distribution
since the higher pressure drop across the channels evens out the pressure
4.4.1. Effect of the header diameter
drop difference between the headers. This is verified by the predictions
As mentioned before, the single-phase flow distribution depends on
in Fig. 20. A less maldistributed flow rate profile, brought by a longer
the relative magnitude of the pressure drop along the headers and that
plate, is expected to reduce the capacity degradation. Moreover, with a
across the channels. Thus, the size of the header in BPHEs plays an
longer plate, the overall thermal effectiveness increases, which further
important role in determining the flow distribution, and consequently
reduces the adverse influence of flow maldistribution, as Fig. 21 in­
the capacity degradation due to flow maldistribution. Fig. 18 plots the
dicates. The results in Fig. 21 are also forṁc,tot = 0.5 kg⋅s− 1, ṁh,tot = 0.49
flow distribution profiles in the 180-plate BPHE with different header
kg⋅s− 1.
diameters (Dp). In the calculations, the size of the feeding tube (Df)
before the heat exchanger is also changed accordingly, to keep a fixed
4.4.3. Effect of the sudden expansion flow at the heat exchanger entrance
radius difference from the header (Dp - Df = 2 mm). It can be seen clearly
As discussed in Li and Hrnjak [5], a smaller feeding tube before
in Fig. 18, that when the header shrinks, the flow maldistribution be­
BPHEs could introduce a sudden expansion flow near the heat exchanger
comes severer because the pressure drop along the headers is more

9
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Fig. 16. “opposite-side supply” arrangement vs “same-side supply” arrange­

Fig. 18. Flow distributions in the 180-plate BPHE with different head­
ment: absolute log-mean temperature difference.
er diameters.

Fig. 17. “opposite-side supply” arrangement vs “same-side supply” arrange­

ment: absolute heat transfer rate. Fig. 19. Capacity degradation due to flow maldistribution: effect of the head­
er diameters.
entrance and thus reduce the flow rate in the first several channels. This
phenomenon is not uncommon in practice and has been reported in Cmin/Cmax would lead to higher thermal effectiveness of the BPHEs,
other studies [12], Wang et al. [14]. The influence length of this sudden which overcomes the adverse effect brought by the mismatched distri­
expansion flow is positively related to the radius difference between the bution profiles of two fluid streams. Taking the 180-plate BPHE as an
feeding tube and the header. As depicted in Fig. 22, under the tested example again: with the uniform flow distribution, the thermal effec­
conditions, such a sudden expansion of the flow area at the heat tiveness of the BPHE (∊) increases from 0.67 to 0.88 with Cmin/Cmax
exchanger entrance seems to improve the flow distribution slightly. decreasing from 0.98 to 0.49. The higher the thermal effectiveness under
Especially, when the “opposite-side supply” arrangement is adopted, the the uniform distribution, the smaller penalty that flow maldistribution
reduction of the flow rate through the first several channels could reduce would lead to.
the mass flow rate differences between two adjacent channels in this
region, which benefits the thermal performance of the BPHEs. The
4.5. Correlations to evaluate the capacity degradation due to flow
simulation results in Fig. 23 support such an analysis.
maldistribution
4.4.4. Effect of the thermal capacity rate ratio
Several key factors related to the effect of single-phase flow mal­
The discussion in section 4.3 indicates the importance of the thermal
distribution are discussed above. However, it is still inconvenient for
capacity rate ratio to this issue. The simulations are carried out to
heat exchanger users to quickly evaluate the capacity degradation due to
compute ζ for the BPHEs with different Cmin/Cmax. In the calculations,
flow maldistribution. Correlations are developed for this purpose.
ṁc,tot is fixed at 0.5 kg⋅s− 1, while ṁh,tot is changing from 0.245 to 0.49
As mentioned previously, the discussed influential factors either
kg⋅s− 1. The data in Fig. 24 reveals that the influence is more obvious in
directly change the flow distribution or indirectly act by potentially
BPHEs with the “opposite-side supply” arrangement since a smaller
affecting the thermal effectiveness of the heat exchangers. Therefore, at

10
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Fig. 20. Flow distributions in the 180-plate BPHE with different plate lengths.
Fig. 22. Flow distributions in the 180-plate BPHE with different sudden
expansion lengths at the entrance.

Fig. 21. Capacity degradation due to flow maldistribution: effect of the

plate length.
Fig. 23. Capacity degradation due to flow maldistribution: effect of the sudden
expansion flow at the entrance.
least two independent variables should be included in the correlations.
The first one represents the degree of flow maldistribution. The ratio
between the average channel frictional pressure drop and the dynamic [ ]
Up,avg • Np • Ap
pressure at the heat exchanger entrance, of the stream with a higher NTU counter = (26)
( ) Cmin
total mass flow rate, , is used here. ΔPf,avg is calculated by:
ΔPf,avg
2 ( ( ))
ρ•vH,i /2
1 − exp − NTU counter • 1 − CCmax
min

( )2 ∊counter = ( ( )) (27)
2 • f • ṁch,avg /Acs • Lp 1 − CCmax
min
• exp − NTU counter • 1 − CCmax
min

ΔPf ,avg = (24) [ ]

ρ • De In Eq. (26), Np • Ap represents the total heat transfer area in the
BPHE, and Up,avg is calculated by Eqs. (2) and (4) based on ṁch,avg of each
where De = 2 • b is the channel effective diameter, Acs = b • Lw is the
stream.
flow area of the channel, and f is the Fanning friction factor for a single
In addition, a third independent variable is added in the correlations
channel. And vH,i is given by:
to consider the influence of the sudden expansion flow near the
( )
ṁtot N
vH,i = (25) entrance, which is: 1 − Nch,exp
ch,s
. Nch,exp denotes the number of channels
ρ • π • D2p /4
the sudden expansion length occupies. Based on Li and Hrnjak [5], it can
The second independent variable is selected to be the thermal be estimated by:
effectiveness of the heat exchanger, calculated by treating it as a pure
counter-current flow heat exchanger, ∊counter :

11
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

Fig. 24. Capacity degradation due to flow maldistribution: effect of the thermal
capacity rate ratio. Fig. 25. Correlation predictions vs the model simulations, for the “same-side
supply” arrangement.
⎧⌈ ( ) ⌉
⎨ 7 • Dp − Df /2 , withD > D
⎪
(28)
p f
Nch,exp = 2•b
⎪
⎩
0, withDp ≤ Df

and Nch,s is the total number of channels of the stream with a higher total
( )
N
mass flow rate. 1 − Nch,exp
ch,s
is selected because it will go to 1 if there is no
sudden expansion flow in the inlet header (Dp ≤ Df ). The final form of
the correlations is:
( )C2 ( )C
ΔPf ,avg Nch,exp 4
ζ = C1 • • ∊ C3
counter • 1 − (29)
ρ • v2H,i /2 Nch,s

The simulated data by the models described in Sections 2.1 and 2.2
are used to obtain C1–C4 in the correlations. Two correlations are
generated for the different arrangements. For the “same-side supply”
arrangement, 267 data points are used to curve-fit the correlation,
( )
among which, ρ•v2f,avg/2 changes from 0.13 to 1.99, ∊counter changes from
ΔP
H,i
( )
N
0.54 to 0.89, and 1 − Nch,exp
ch,s
changes from 0.9 to 1. The developed
correlation is given in Eq. (30). By using the correlation, the capacity
degradation ζ is predicted and compared with the model simulation, as
shown in Fig. 25. The mean absolute deviation between the model
simulations and the correlation-fitted results is 0.13 %. Fig. 26. Correlation predictions vs the model simulations, for the “opposite-
side supply” arrangement.
( )− 1.19 ( )6.99
ΔPf ,avg Nch,exp
ζ[%] = 0.336 • • ∊ − 0.80
counter • 1 − (30) Eqs. (30) and (31), the total heat exchanger capacity can also be
ρ • v2H,i /2 Nch,s
conveniently calculated with the consideration of single-phase flow as
For the “opposite-side supply” arrangement, 270 data points are used follows:
( )
( )
for the curve-fitting, with ρ•v2f ,avg/2 changing from 0.13 to 1.99, ∊counter
ΔP
H,i
QBPHE = Cmin • Th,i − Tc,i • ∊counter • (1 − ζ) (32)
( )
N
changing from 0.54 to 0.89, and 1 − Nch,exp changes from 0.9 to 1. The Fig. 27 compares the QBPHE predicted by the models in Sections 2.1,
ch,s

developed correlation is given in Eq. (31) and its predictions are shown 2.2, and that calculated by Eqs. (30)–(32). It can be concluded that they
in Fig. 26. The mean absolute deviation is 2.8 %. agree reasonably well with each other. For the “same-side supply”
arrangement, the deviations are within 1 %, with the mean absolute
( )− 1.03
ΔPf ,avg
(
Nch,exp
)4.36 deviation of 0.28 %. Slightly larger deviations are obtained for the
ζ[%] = 8.296 • • ∊ − 0.09
counter • 1 − (31) “opposite-side supply” arrangement, in which the majority of the data
ρ • v2H,i /2 Nch,s
points is within 5 % difference, and the mean absolute deviation is 3.67
By using the developed correlations in Eqs. (30) and (31), the ther­ %.
mal capacity degradation due to flow maldistribution is conveniently
evaluated, without the necessity of knowing the detailed flow distribu­
tion and building heat exchanger models. Further, with ζ evaluated by

12
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

impact on the overall thermal capacity of brazed plate heat exchangers if

two streams are supplied from the same side; if two streams are supplied
from the opposite side of heat exchangers, the overall thermal perfor­
mance is significantly deteriorated due to mismatched flow distribution
profiles of two streams. Moreover, the thermal capacity degradation due
to flow maldistribution can be moderated with fewer but longer plates, a
larger header, a larger size difference between the header and the
feeding tube, and a smaller heat capacity rate ratio between two fluid
streams. Those factors act by either directly changing the flow distri­
bution or/and potentially affecting the overall thermal effectiveness of
brazed plate heat exchangers. Correlations are generated based on the
simulation results, which can predict the thermal capacity degradation
due to flow maldistribution without the necessity of knowing the
detailed flow distribution and building heat exchanger models.

Declaration of Competing Interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
the work reported in this paper.
Fig. 27. Comparison of the heat exchanger capacities obtained by the models
in Sections 2.1, 2.2, and by Eqs. (30)–(32). Data availability

5. Conclusions The authors do not have permission to share data.

In this study, the effect of single-phase flow maldistribution on the Acknowledgments

thermal performance of brazed plate heat exchangers is investigated by
an experimentally validated thermal performance model. The results The authors thankfully acknowledge the support provided by the
reveal that the way two fluid streams are supplied to the heat exchangers Creative Thermal Solutions, Inc. and the Air Conditioning and Refrig­
significantly matters in this issue: the flow maldistribution has a trivial eration Center at the University of Illinois at Urbana-Champaign.

Appendix

Applicability of the log-mean temperature difference (LMTD) method to the CVs of BPHEs is proven as follows: considering a finite length of a
thermal plate in a BPHE, along with its two adjacent channels, as shown in Fig. A1, the heat transfer equation of the plate, as well as the energy balance
equations of each fluid stream, over the finite length, are given as:
|Q| = U • A • ΔTm (A1)
( )
|Q| = ṁh • cp,h • Th,A − Th,B − α • |Q| (A2)
( )
|Q| = ṁc • cp,c • Tc,A − Tc,B − β • |Q| (A3)

Fig. A1. A finite length of the plate and its adjacent channels for the LMTD derivation.

13
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

In Eqs. (A1)–(A3), |Q| is the absolute heat transfer rate through the finite length of the plate in between the two channels in Fig. A1. Considering the
fluid streams in these two channels transfer heat through both the plates on both sides, it is assumed in Eq. (A2), (A3) that, in the two channels, the
absolute heat transfer rates through the plate other than the middle plate are α • |Q| and β • |Q| respectively, as Fig. A1 demonstrated. If the finite
length is sufficiently small, it is rational to assume that the heat flux over the length is constant, αand β are thereby constant. It also should be noticed
that a negative sign is added to δ|Q| • α and δ|Q| • β (the incremental amount of α • |Q| and β • |Q|) in Fig. A1 to indicate that the heat is actually
transferred from the left side channel to the right side channel.
Then, focusing on a small incremental length of the plate and channels, the governing equations are:
δ|Q| = U • (Th − Tc ) • dA (A4)

δ|Q| = ṁh • cp,h • dTh − α • δ|Q| (A5)

δ|Q| = ṁc • cp,c • dTc − β • δ|Q| (A6)

Re-organizing Eqs. (A5) and (A6):

(1 + α) • δ|Q|
dTh = (A7)
ṁh • cp,h

(1 + β) • δ|Q|
dTc = (A8)
ṁc • cp,c

Combing Eqs. (A7) and (A8):

⎛ ⎞
1 + α 1 + β
d(Th − Tc ) = δ|Q| • ⎝ − ⎠ (A9)
ṁh • cp,h ṁc • cp,c

Using Eq. (A4) to substitute δ|Q| in Eq. (A9):

⎛ ⎞
d(Th − Tc ) ⎝ 1+α 1+β ⎠
=U• − • dA
Th − Tc ṁh • cp,h ṁc • cp,c

Integrating Eq. (A10) from B to A:

⎛ ⎞
1+α 1+β ⎠
ln(Th − Tc ) |AB = U • A⋅⎝ −
ṁh • cp,h ṁc • cp,c

Revisiting Eqs. (A2) and (A3):

( )
1+α Th,A − Th,B
=
ṁh • cp,h |Q|
( )
1+β Tc,A − Tc,B
=
ṁc • cp,c |Q|

Combing Eqs. (A11)–(13):

( )
Th,A − Th,B − Tc,A + Tc,B
ln(Th − Tc ) |AB = U • A⋅
|Q|
Defining:
ΔTA = Th,A − Tc,A

ΔTB = Th,B − Tc,B

Then, Eq. (14) becomes:

(ΔTA − ΔTB )
ln(ΔTA − ΔTB ) = U • A⋅
|Q|
Re-organizing Eq. (17):
(ΔTA − ΔTB )
|Q| = U • A⋅
ln(ΔTA − ΔTB )
Comparing Eq. (A18) and (A1), we have:
(ΔTA − ΔTB )
ΔTm =
ln(ΔTA − ΔTB )
Eq. (A19) is the same as the conventional expression of LMTD. The above derivation justifies the use of the LMTD method to calculate heat transfer
between two adjacent CVs in BPHEs.

14
W. Li and P. Hrnjak Applied Thermal Engineering 219 (2023) 119465

References [10] H. Shokouhmand, M. Hasanpour, Effect of number of plates on the thermal

performance of a plate heat exchanger with considering flow maldistribution,
J. Energy Storage 32 (2020) 101907.
[1] Z.H. Ayub, Plate heat exchanger literature survey and new heat transfer and
[11] B. Thonon, P. Mercier, M. Feidt, Flow distribution in plate heat exchangers and
pressure drop correlations for refrigerant evaporators, Heat Transfer Eng. 24 (5)
consequences on thermal and hydraulic performances, Des. Oper. Heat Exchang.
(2003) 3–16.
18 (1992) 245–254.
[2] M.K. Bassiouny, H. Martin, Flow distribution and pressure drop in plate heat
[12] J.C.K. Tong, E.M. Sparrow, J.P. Abraham, Geometric strategies for attainment of
exchangers-I: U-type arrangement, Chem. Eng. Sci. 39 (4) (1984) 693–700.
identical outflows through all of the exit ports of a distribution manifold in a
[3] M.K. Bassiouny, H. Martin, Flow distribution and pressure drop in plate heat
manifold system, Appl. Therm. Eng. 29 (17–18) (2009) 3552–3560.
exchangers-II: Z-type arrangement, Chem. Eng. Sci. 39 (4) (1984) 701–704.
[13] H. Tuo, P. Hrnjak, Effect of the header pressure drop induced flow maldistribution
[4] R.J. Kee, P. Korada, K. Walters, M. Pavol, A generalized model of the flow
on the microchannel evaporator performance, Int. J. Refrig. 36 (8) (2013)
distribution in channel networks of planar fuel cells, J. Power Sources 109 (1)
2176–2186.
(2002) 148–159.
[14] C.-C. Wang, K.-S. Yang, J.-S. Tsai, I.Y. Chen, Characteristics of flow distribution in
[5] W. Li, P. Hrnjak, Single-phase flow distribution in brazed plate heat exchangers:
compact parallel flow heat exchangers, Part I: Typical inlet header, Appl. Therm.
experiments and models, Int. J. Refrig. 126 (2021) 45–56.
Eng. 31 (16) (2011) 3226–3234.
[6] W. Li, P. Hrnjak, Compensating for the end plate effect on heat transfer in brazed
[15] J.Y. Wang, Pressure drop and flow distribution in parallel-channel of
plate heat exchangers, Int. J. Refrig. 126 (2021) 99–108.
configurations of fuel cell stacks: U-type arrangement, Int. J. Hydrogen Energy 33
[7] S. Maharudrayya, S. Jayanti, A.P. Deshpande, Flow distribution and pressure drop
(21) (2008) 6339–6350.
in parallel-channel configurations of planar fuel cells, J. Power Sources 144 (1)
[16] J.Y. Wang, Pressure drop and flow distribution in parallel-channel of
(2005) 94–106.
configurations of fuel cell stacks: Z-type arrangement, Int. J. Hydrogen Energy 35
[8] B. Prabhakara Rao, P. Krishna Kumar, S.K. Das, Effect of flow distribution to the
(11) (2010) 5498–5550.
channels on the thermal performance of a plate heat exchanger, Chem. Eng.
[17] L. Wang, B. Sunden, R.M. Manglik, Plate Heat exchanger: Design, Application and
Process. 41 (1) (2002) 49–58.
Performance, WIT Press, Great Britain, 2007.
[9] B.P. Rao, B. Sunden, S.K. Das, An experimental and theoretical investigation of the
effect of flow maldistribution on the thermal performance of plate heat exchangers,
J. Heat Transfer 127 (2005) 332–343.

15