Results in Engineering 19 (2023) 101324

Contents lists available at ScienceDirect

Results in Engineering
journal homepage: www.sciencedirect.com/journal/results-in-engineering

Thermal uniformity enhancement of the motorcycle exhaust thermoelectric

generator—Theory model for predicting heat exchanger fin profile
Thong Duc Hong a, b, *, Minh Quang Pham a, b, Quan Thien Phan Nghiem a, b
a
Faculty of Transportation Engineering, Ho Chi Minh City University of Technology (HCMUT), District 10, Ho Chi Minh City, 700000, Viet Nam
b
Vietnam National University Ho Chi Minh City (VNU-HCM), Thu Duc District, Ho Chi Minh City, 700000, Viet Nam

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

Keywords: This paper develops the theoretical model to predict the fin profile of the Hot-side heat exchanger (HHE) to
Thermoelectric generator enhance the uniformity of its longitudinal temperature distribution. The developed model can be applied to
Thermal uniformity general cases where heat exchangers have different numbers, heights, pitches, and materials of fins, as well as the
Fin profile
thermodynamic and heat transfer parameters of the working fluid. Numerical investigations of two fin profiles’
Heat exchanger
Muffler
thermal performance are conducted to validate the theoretical model. According to the results, the optimal
Prediction profile improves the temperature uniformity by 91.3% compared to the original profile. The improvement in the
temperature uniformity between thermoelectric modules (TEMs) can be achieved at the cost of reducing the
thermoelectric generator unit power. The reductions in heat absorption of the HHE in the current study are not
significant, varying from 1.3% to 2.9% at different engine speeds. However, the temperature uniformity can
restrict the output power loss due to the significant differences in voltages and amperages produced by TEMs and
also can eliminate the deterioration of TEMs’ lifespan.

results in the inconsistent temperature gradient of each thermoelectric

1. Introduction module (TEM); thus, the generated voltage, current, and power of TEMs
are unequal. In addition, because they are connected in series or parallel
The dramatic rises in fossil fuel prices and global warming issues are arrangements, the net system power will decrease [2]. The effect of
one of the most crucial concerns of countries in recent years. In that inhomogeneous temperature distribution on the hot side surface of TEG
circumstance, people must utilize energy more efficiently and also systems also lowers the TEMs’ longevity in areas having temperatures
recover waste energy sources. Currently, transportation is considerably exceeding the working temperature range.
affected by the issues of fuel price and global warming due to the Many studies have proposed approaches regarding the structures and
enormous quantity of vehicles powered by internal combustion engines. geometry of heat exchangers, which harvest exhaust heat and transfer it
Fortunately, thanks to the presence of energy recovery technologies, to TEMs, to enhance temperature uniformity. Musial et al. [3] numeri­
specifically the Rankine cycle and the thermoelectric generator (TEG), cally investigated the influence of a conical structure positioned at the
the utilization of waste heat in the exhaust of those vehicles is feasibly centre of the TEG system on its longitudinal temperature uniformity.
achieved. The findings showed that the dispersion structure effectively improved
Based on researching patents regarding waste heat recovery tech­ the consistency of temperature distribution in the heat exchanger,
nologies, Karvonen et al. [1] concluded that TEG is more developed than considerably enhancing the efficiency of the TEG system. Tang et al. [4]
the Rankine cycle; therefore, it will reach the commercial scale sooner. utilized the computational fluid dynamics (CFD) simulation method to
Meanwhile, the Rankine cycle’s application on engines has just begun, study heat exchangers’ exhaust flow field with three configurations of
and this technology still needs more development before being fully internal fins. The fins were proven to raise the temperature uniformity
commercialized. However, there are many issues that are hindering the and reduce the back pressure of the TEG system by enhancing the dis­
adaptation of TEG in practice. One of those is the non-uniform tem­ tribution of the exhaust flow field. Lu et al. [5] examined three heat
perature distribution on the hot side surface of TEG systems owing to the exchangers combined with the muffler in the form of one inlet and two
drop in exhaust temperature along the exhaust path. This phenomenon outlets, two inlets and two outlets, and an empty cavity. The model with

* Corresponding author. Faculty of Transportation Engineering, Ho Chi Minh City University of Technology (HCMUT), District 10, Ho Chi Minh City, 700000, Viet
Nam.
E-mail address: [email protected] (T.D. Hong).

https://doi.org/10.1016/j.rineng.2023.101324
Received 21 June 2023; Received in revised form 24 July 2023; Accepted 26 July 2023
Available online 28 July 2023
2590-1230/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
T.D. Hong et al. Results in Engineering 19 (2023) 101324

Nomenclature s fin spacing (m)

T temperature (K)
A area (m2) tf fin thickness (m)
Cp specific heat at constant pressure (J kg− 1 K− 1) v kinematic viscosity (m2 s− 1)
Dh hydraulic diameter (m) xd entry length (m)
e element xi, xj the Cartesian coordinates (m)
→g gravitational acceleration (m s− 2)
Greek symbols
Gb generation of turbulence kinetic energy due to buoyancy
α convection coefficient (W m− 2 K− 1)
Gk generation of turbulence kinetic energy due to the mean
δ boundary layer thickness (m)
velocity gradients
ε turbulence energy dissipation rate (m2 s− 3)
H overall high of the fins profile (m)
λ thermal conductivity (W m− 1 K− 1)
hi,j height of an element (m)
μt turbulent viscosity (Pa s)
Hj height of column j (m)
μ dynamic viscosity (Pa s)
i row number of an element
ρ density (kg m− 3)
j column number of an element
σ k, σε turbulent Prandtl numbers for k and ε
k turbulence kinetic energy (J kg− 1)
ω exhaust velocity in a fin spacing (m s− 1)
L overall length of the fins profile (m)
ωi velocity component (m s− 1)
l length of a column (m)
∇ gradient
ṁ mass flow rate (kg s− 1)
n number of columns Acronyms
p static pressure (Pa) CFD computational fluid dynamics
Nf fin number CHE cool-side heat exchanger
Nu Nusselt number HHE hot-side heat exchanger
Pr Prandtl number TEG thermoelectric generator
qgas_ij heat rate released by the exhaust flowing an element (W) TEMs thermoelectric modules
qi,j heat rate of an element (W) TGU thermoelectric generator unit
Qj Heat rate of a column (W)
Re Reynolds number

one inlet and two outlets produced higher surface temperatures and thermal uniformity than the twisted-fin. Wu et al. [12] modified the
thermal uniformity than the other; however, it also caused more exhaust heat-collection area of a cylindrical heat exchanger by replacing vertical
gas pressure drop. Zhao et al. [6] proposed a new thermoelectric inner fins with horizontal inner fins. The fin bottom width, height, and
generator system with the media fluid that transfers heat from the thickness were also optimized to raise the uniformity of temperature
exhaust to the TEMs. The results indicated that the temperature distri­ distribution. The numerical analyses showed that the temperature dis­
bution along the TEG system was more uniform than the conventional. tribution between fins was more uniform after the optimization. Quan
Luo et al. [7] presented a converging heat exchanger with inclined hot et al. [13] optimized temperature uniformity of a TEG system for mili­
sidewalls, which inhibited the decline of hot side temperature at tary vehicles by employing high conductive materials on the heat ex­
downstream areas of the heat exchanger. The findings revealed that this changer’s inlet side. Through experiments, it was concluded that the
design raised the TEG system’s output power and temperature distri­ method was effective in decreasing the temperature difference between
bution consistency while the backpressure was limited. Chen et al. [8] the inlet side and outlet side of the heat exchanger. Besides, the pressure
developed a segmented converging heat exchanger and optimized the drop of the system satisfied the limitation of backpressure for vehicles.
convergence angle of each segment with the target of enhancing the From the above studies, it is obvious that the heat exchanger’s structures
system performance. The convergence angles were determined so that are diverse and complicated; thus, there is a necessity to develop a
the TEMs’ temperature differences across the segments were equal by particular optimization method for each type of structure.
using mathematical analyses. The system output power was enhanced Considering the type of heat exchanger using straight fins for col­
thanks to the optimal convergence angle of the heat exchanger seg­ lecting heat, there are very few researches employing this structure in
ments. Rui et al. [9] studied the impacts of inserted fins’ parameters (i. TEG systems and optimizing it. Among the studies cited above, Luo et al.
e., length, width, spacing distance, and intersection angle of fins) on the [7] and Chen et al. [8] developed this type of heat exchanger, in which
performance of the TEG system with a polyhedral-shaped heat the outer walls of the component were made inclined to enhance ther­
exchanger. According to the simulation results, the rise in the width, mal uniformity. Nevertheless, that method is only suitable for TEG
length, and intersection angle of fins considerably improved tempera­ systems applied on automobiles and stationary engines, where the space
ture uniformity. Meanwhile, the temperature uniformity was insignifi­ limitation is less severe. In the case of motorcycles, the exhaust system
cantly raised by decreasing the fin spacing distance. Shu et al. [10] must be compact; therefore, it is required to develop an alternative
designed different heat-collecting fin configurations installed along the method of enhancing the thermal uniformity of longitudinal-fin heat
streamwise direction in a hexagonal-shaped heat exchanger. Based on exchangers while the compactness is unchanged.
numerical analyses, it was suggested that the configurations of three and Similar to the aforementioned studies, various structures were pro­
four fins achieved a more uniform temperature distribution compared to posed and investigated in the authors’ previous researches [14–17] to
other cases. Wang et al. [11] developed a twisted-fin and a spiral-fin minimize the temperature difference in the transverse cross-section of
heat exchanger and compared their thermal performance. According the TEG system. Firstly, different exhaust guide structures (e.g., guide
to simulation results, the spiral-fin heat exchanger obtained a higher fin, V-fin, double V-fin) were employed at the muffler inlet compartment

2
T.D. Hong et al. Results in Engineering 19 (2023) 101324

to prevent the concentration of the exhaust flow in the heat exchanger and original optimization method provide a new approach to improve
middle area [14]. Those structures proved effective in increasing the the output power of the TEG systems.
temperature uniformity based on numerical and experimental analyses.
Moreover, the configuration using guide fins was the most efficient 2. Development of prediction model for HHE’s fin profile
model. Due to the concentration of exhaust flow in some areas of the
heat exchanger derived from the single-layer arrangement of prior de­ 2.1. Physical description
signs, numerous structures arranged in the double-layer configuration
were later developed [15]. The new configurations enhanced the tem­ In the authors’ prior papers [14–17], the HHE’s fins had a similar
perature uniformity of the heat exchanger; however, their effectiveness rectangular profile. Those aluminum fins, whose cross-section shape is
is not significantly higher than the single-layer configurations, and their rectangular, were distributed evenly on a base. As the exhaust flows
construction was more complicated to fabricate. As a result, the guide fin through the HHE’s fins, its heat energy declines because there is an
structure having the single-layer configuration was selected to further amount of heat absorbed into the fins. This phenomenon causes the
exploit its capability by optimizing its geometry [16]. Three geometrical temperature of HHE’s fins and base to distribute inconsistently. In
parameters of this structure (i.e., fin gap, fin position, and fin angle) particular, the regions near the exhaust inlet achieve higher tempera­
were numerically surveyed under different conditions to obtain optimal tures than the regions by the exhaust outlet. To produce a uniform
values. Based on the evaluation of exhaust flow uniformity, the effects of temperature distribution on the HHE’s base, the heat transfer area
the fin gap and fin angle were more significant than that of the fin po­ should be increased to compensate for the decline in the exhaust heat
sition. Besides the effectiveness of the optimal guide fins in improving energy along the muffler length. Consequently, the fin profile tends to
the temperature uniformity, experimental results also revealed that the increase its height along the HHE’s length.
uniformity was impacted by the muffler outlet structure, which led to a
study on optimizing this element. The effects of the outlet structures and
2.2. Proposed prediction model
geometrical parameters (i.e., outlet angle and position) on the exhaust
flow distributions were assessed to achieve the optimal outlet design
Based on the above physical description, the authors proposed a
[17]. The results revealed that the optimal outlet contributed to the heat
prediction model for HHE’s fin profile producing a uniform temperature
exchanger’s temperature uniformity. Ultimately, compared to the orig­
distribution on HHE’s base. The model’s configuration is illustrated in
inal model, the optimal structures of guide fins and muffler outlet raised
Fig. 1. The overall length and height of the fin profile are L and H,
the heat exchanger’s temperature uniformity on the transverse
respectively. The fin area is divided into elements arranged in a matrix
cross-section by 50%. Nevertheless, enhancing the temperature distri­
of n columns and n rows. The row order begins at the HHE’s base. The
bution uniformity on the longitudinal cross-section is essential because
column order starts at the first column to contact the exhaust. The po­
the difference in generated power could exist between TEMs located at
sition of each element is defined based on its row number, i, and its
different rows in the system.
column number, j. For example, element e1,2 locates in row 1 and col­
This study builds a theoretical model to determine a fin profile of the
umn 2. Elements located in the same row i have the same height hi. For
hot-side heat exchanger, which improves the longitudinal temperature
instance, row 1 has elements e1,1, e1,2, …, e1,n, which have the same
uniformity of a thermoelectric generator unit using the exhaust gas of a
height h1,1 = h1,2 = … = h1,n = h1. All columns have the same length of l
motorcycle. Based on the basic dimensions of the HHE used in the au­
= L/n.
thors’ previous studies [14–17], the theoretical model is employed to
Column 1 only has one element, e1,1; thus, its height is equal to the
develop a new optimal fin profile. The optimal and the original fin
element’s height, H1 = h1,1 = h1. As flowing through column 1, the
profile are then numerically investigated in terms of temperature dis­
exhaust loses an amount of heat, which is equal to the heat amount
tribution to verify the effectiveness of the theoretical model in
absorbed by column 1. As a result, column 2 must be higher than column
enhancing temperature uniformity. This novel profile prediction model
1 in order to gather the same heat amount as column 1. There are 2

Fig. 1. The proposed model for predicting the HHE’s fin profile.

3
T.D. Hong et al. Results in Engineering 19 (2023) 101324

elements e1,2 and e2,2 in column 2 with the respective height of h1,2 and
h2,2. Since h1,2 = h1,1, h2,2 is the additional height of column 2 compared
to that of column 1. The overall height of column 2 is H2 = h1,1 + h2,2 =
h1 + h2. Similarly, column j is higher than column j-1. Column j has j
elements, involving e1,j, e2,j, …, ej,j, and its height is Hj = h1,j+ h2,j + … +
hj,j = h1 + h2 + … + hj.
The energy balance equation is applied for each element to estimate
its heat rate, qi,j. The heat rate of a column, Qj, is the total heat rate of
elements in that column. For example, Q2 = q1,2 + q2,2. Following the
requirement of a constant heat rate for all columns, the height of each
column can be predicted based on the above model.
As the exhaust flows through the gap between two fins, the boundary
layers are developed due to the viscous effect [18]. The flow is divided
into two regions: the entry region and the fully developed region, which
are presented in Fig. 2. The entry region begins at the inlet and ends at
the point at which the boundary layers merge. The fully developed re­
Fig. 3. The assumption of model conversion for the entry region.
gion is where the boundary layers merge completely. Because of the
difference in the heat transfer coefficient in these two regions, the
convection coefficient of each element, αi,j, is calculated based on the - The exhaust flow regime in fin gaps is laminar, which is estimated
element position belonging to which region. based on the exhaust velocity in the muffler and fin spacing in pre­
vious studies [14–17].”
2.3. Assumptions
2.4. Development of equations
The prediction model is developed based on the following
assumptions: 2.4.1. Initial parameters of the model
The equations representing the geometrical specification of the
- Temperature distribution on fins is uniform and the exhaust velocity prediction model are presented below.
in fin gaps is the same. These two assumptions are based on the The heights of elements located in the same row i are equal:
enhancement of fin temperature uniformity, which was achieved hij = hi (1)
thanks to the optimization of muffler structures in previous studies
[16,17]. The height of any column:
- Heat transfer only occurs between the exhaust and the HHE, while j
∑
heat transfer between the exhaust and muffler surfaces is neglected. Hj = hi (2)
This assumption is derived from the requirement of supplying a i=1

sufficient heat amount to the HHE to achieve a high output power of The height of the last column:
the TEG system. In practice, the external surfaces of systems are
covered by thermal insulators to prevent heat loss. ∑
n
Hn = hi = H (3)
- On a fin surface, there is only one-dimension conduction from the fin i=1
tip to the base, whereas the conduction on the fin surface along the
exhaust flow direction is neglected. This assumption agrees with the The length of each column:
objective of a uniform fin temperature distribution along the exhaust L
l= (4)
flow direction. n
- The heat transfer model of parallel flow over a flat plate is applied to
the entry region. Within the fully developed region, the heat transfer 2.4.2. Defining the convection coefficient on the fin surface
model of an internal flow in a non-circular tube is employed. These
heat transfer models have similar flow regimes as respective regions a) The convection coefficient on the fin surface in the entry region
in the fin gap. The utilization of these models also ensures accurate
simulation results [18]. The calculation model of the parallel-laminar flow over a flat plate is
- The entry region of the fin has a step profile, which is illustrated in employed in the entry region. The boundary layer thickness at xd, δxd, is
Fig. 3 because the fin surface is divided into columns. calculated as follows [18]:
5xd
δxd = √̅̅̅̅̅̅̅̅̅ (5)
Rexd

where Rexd is the Reynolds number at xd in the entry region.

Reynolds number at xd is calculated below [18]:
ωxd
Rexd = (6)
ν

where ω is the exhaust velocity in a fin spacing, and v is the kinematic

viscosity of exhaust.
As the two boundary layers merge (See Fig. 2), the boundary layer
thickness is equal to half of the fin spacing:
s
δxd = (7)
2
Fig. 2. The boundary layer development of the laminar flow in a gap (δ is
boundary layer thickness, xd is entry length).

4
T.D. Hong et al. Results in Engineering 19 (2023) 101324

where s is the fin spacing.

By substituting Eqs. (6) and (7) into Eq. (5), the length of the entry
region, xd, is estimated as follows:

25 s2 ω
xd = (8)
4 ν
The average Nusselt number in the entry region, Nud , is calculated
below [18]:

Nud = 0.664Rexd 0.5 Pr1/3 (9)

where Pr is the exhaust Prandtl number.

The convection coefficient in the entry region is determined below
[18]:

Nud λ
αd = (10)
xd
Fig. 4. The original HHE.
where λ is the exhaust thermal conductivity.
The heat rate of each element, qi,j, is defined as [18]:
b) The convection coefficient on the fin surface in the fully developed ( )
Tin ij + Tout ij
region qij = αij Aij − Tc (15)
2
The heat transfer model of the fully developed laminar flow in a non- where qij is the heat rate of an element, and Tc is the temperature of the
circular tube is employed to calculate the convection coefficient in the fin surface.
fully developed region, as follows [19]: The heat rate released by the exhaust flowing through an element is
Nufd λ equal to the absorbed heat rate of that element.
αfd = (11)
Dh qgas = qij (16)
ij

where Nufd is the Nusselt number of the fully developed laminar flow in Substituting Eqs. (13) and (15) into Eq. (16), the temperature of the
a non-circular tube, namely, the parallel plates duct for the flow passes a exhaust exiting an element is expressed as:
fin’s element, referenced from Ref. [19]. Dh = 2s is the hydraulic ( )
diameter of the fully developed laminar flow in the parallel plates duct Tin ij ṁi Cp − 0.5αij Aij + αij Aij Tc
Tout ij = (17)
[19]. ṁi Cp + 0.5αij Aij
The convection coefficient of any element, αij, is estimated based on Substituting Tout_ij in Eq. (17) into Eq. (15), the heat rate of an
the element location belonging in the entry region or the fully developed element is obtained. Therefore, the heat rate of a column is defined
region: below:

αij = αd , l(j − i + 1) < xd or hij (j − i + 1) < xd

αij = αfd , l(j − i) > xd & hij (j − i) > xd
(12)
αd Ad ij + αfd Afd ij
αij = , l(j − i) < xd < l(j − i + 1) or hij (j − i) < xd < hi (j − i + 1)
Aij

where Ad_ij is the area of the entry region within an element, Afd_ij is the j
∑
area of the fully developed region within an element, and Aij is the total Qj = qij (18)
area of an element. i=1

Since the heat rate of a column is constant, it follows that:

2.4.3. Calculating the heat transfer between the exhaust and the HHE
Q1 = Q2 = Qj ... = Qn (19)
The heat rate released by the exhaust flowing through an element,
qgas_ij, is estimated as below [18]: By solving a system of equations involving from Eq. (1) to Eq. (19),
( ) the heights of all columns are achieved.
qgas ij = ṁij Cp Tin ij − Tout ij (13)

where qgas_ij is the heat rate released by the exhaust flowing through an
2.5. Calculation of a HHE fin profile
element, Tin_ij and Tout_ij are the temperatures of exhaust entering and
exiting an element, respectively, Cp is the specific heat of the exhaust,
In this work, the optimal fin profile, which acquires temperature
and ṁij is the exhaust mass flow rate passing through an element.
uniformity on the HHE’s base, is defined by utilizing the prediction
The mass flow rate of exhaust flowing through an element:
model in the prior section. Fig. 4 shows the original HHE construction,
ṁij = ρωhij s (14) which was proposed in the authors’ previous studies [14–17]. The
original HHE has the fin number, Nf, of 11, fin thickness, tf, of 1 mm, fin
where ρ is the mass density of exhaust. spacing, s, of 7 mm, overall fin length, L, of 160 mm and overall fin

5
T.D. Hong et al. Results in Engineering 19 (2023) 101324

height, H, of 68 mm. The fin surface is divided into 10 columns (n = 10).

The optimal fin profile is calculated with the exhaust properties of an
engine working at the rotational speed of 4925 rpm. The exhaust ve­
locity, ω, is 1.05 m/s. The exhaust kinematic viscosity, ν, is 7.71 × 10− 5
m2/s. The exhaust thermal conductivity, λ, is 0.0486 W/mK. The
exhaust mass density, ρ, is 0.474 kg/m3. The inlet temperature of the
exhaust, Tinlet, is 712 K. The Prandtl number, Pr, is 0.65.
The Reynolds number of exhaust flowing through the HHE, ReD, is
calculated as follows:
ωD h
ReD = (20)
v
The average hydraulic diameter, Dh, is expressed as:
4Hs
Dh = (21)
2(H + s)
The Reynolds number in this study is ReD = 153.44 ≤ 2300; thus, the
exhaust flow regime is laminar [18], which is compatible with the as­
sumptions of the prediction model. The length of the entry region, xd, is
0.0053 m. The convection coefficient of the entry region, αd, is 42.5
W/m2K. The convection coefficient of the fully developed region, αfd, is
27.9 W/m2K.
To simplify the calculation process, the authors selected the substi­
tution and iteration methods to solve the system of equations. The height
value of column 1, H1, varies from 0 to H. The convergence condition is
that the difference in heat rate between columns and the difference in
height value between the last column and the overall fin height must be Fig. 5. The dimensions of the optimal fin profile.
below 5%. The predicted heights of columns in the optimal fin profile
are listed in Table 1. Fig. 5 illustrates the optimal fin profile drawn from
height values presented in Table 1.

3. Simulation investigation of temperature distribution on hot-

side heat exchanger (HHE)

3.1. Configuration of thermoelectric generator unit (TGU)

Fig. 6 presents the configuration of the TGU used in this work. The
design consists of TEMs, a cool-side heat exchanger (CHE), and an HHE
that are all integrated into a motorcycle muffler. The muffler, which has
a 350 mm length and a rectangular cross-section of 90 mm × 68 mm, is
separated into three compartments. The first compartment is equipped
with guide fins, which were presented in the authors’ previous work
[16], to distribute the exhaust flow uniformly on the transverse
cross-section of the muffler. The center compartment consists of an HHE,
which contacts the exhaust and supplies heat energy to the TEMs and
CHE. The last compartment is empty. The compartments are separated
by the sound-absorbing units. This configuration of TGU is installed on
the exhaust line of a motorcycle to replace its original muffler.

Fig. 6. Configuration of TGU.

3.2. Fin profiles of HHE for simulation investigation
and 160 mm in height and length, respectively. Model 2 has the optimal
The two models of the HHE proposed to simulate their temperature parabolic fin profile achieved by applying the theoretical model. This
distribution are shown in Fig. 7. All HHE models have an overall optimal HHE model has a minimum and a maximum fin height of 16.4
dimension of 180 × 100 × 78 mm and have the same quantity, thick­ mm and 68 mm, respectively.
ness, and pitch of the fins. Model 1, which is the original HHE used in the
authors’ previous studies [14–17], has a rectangular fin profile with 68
3.3. CFD simulations
Table 1
The predicted heights of columns in the optimal fin profile. The average temperatures of eight positions corresponding to eight
Column height Value (mm) Column height Value (mm) TEMs on the base of two models are explored by using CFD simulations
H1 16.4 H6 46.4 on the ANSYS Fluent program. The simulation model used in this study
H2 23.9 H7 51.8 only includes the muffler and the HHE, while TEMs and CHE were
H3 29.9 H8 57.2 removed. The heat exchange in the simulation models occurs between
H4 35.5 H9 62.6 the exhaust, the HHE and the external environment. The remaining
H5 40.9 H10 68.0
surfaces of the muffler have no heat exchange with the surroundings.

6
T.D. Hong et al. Results in Engineering 19 (2023) 101324

Fig. 8. The meshing model of Model 1.

Table 4
Comparison of the average HHE’s temperatures and errors with different grid
numbers.
Mesh size (mm) Grid number Average temperature of HHE (K) Error (%)

2 5,018,976 690.9 0.582

1.7 6,250,547 689.5 0.379
1.6 7,201,674 688.3 0.204
1.4 8,017,682 687.5 0.087
1.2 9,025,763 687.0 0.015
1.1 10,134,592 686.9 0.000

Each model is surveyed at three engine speeds of 990, 2955, and

4920 rpm. Exhaust properties are determined in the authors’ previous
study [15], as presented in Table 2. The temperature is measured in
experiments on the Honda Wave Alpha 110 cc motorcycle at the sta­
tionary mode. The inlet flow rate, density, thermal conductivity and
viscosity are obtained by solving the ideal combustion process of the
engine.
The outlet pressure of the exhaust is set at 101,325 Pa. The HHE is
made of aluminum, and its physical properties are revealed in Table 3.
Fig. 8 shows the meshing model of Model 1, which has a rectangular
fin profile. The unstructured Tetrahedral grid is employed to define the
numerical model. The mesh sensitivity analysis is performed for Model 1
Fig. 7. Investigated fin profiles of HHE: a) Model 1 – original rectangular fin to select a suitable grid number, which balances the accuracy of simu­
profile, b) Model 2 - optimal parabolic fin profile. lation results and the computing resources. In addition, the mesh size is
set equal for both the exhaust and the HHE at each grid number in this
test. The average temperatures of the HHE base according to the grid
Table 2 numbers are presented in Table 4. The error of the grid level of nine
Physical properties of exhaust at different engine speeds. million elements is 0.015% compared with the case of ten million ele­
Property Speed (rpm) ments. Therefore, the nine-million-element grid model is utilized in all
990 2955 4925 investigated cases to ensure accuracy and save computer resources. In
addition, the Skewness mesh metric is applied to assess the meshing
Inlet flow rate (l/s) 0.81 2.41 4.02
Inlet temperature (K) 387 584 712 quality, which is essential for the accuracy and stability of the numerical
Density (kg/m3) 0.981 0.583 0.474 simulations. The average Skewness values of Models 1 and 2 are
respectively 0.236 and 0.245, which are in the excellent quality range.
5 5 5
Dynamic viscosity (Pas) 2.018 × 10− 2.818 × 10− 3.655 × 10−
Thermal conductivity (W/mK) 0.03 0.0435 0.0518

3.4. Governing equations

Table 3
Physical properties of HHE. The conservation equations are the continuity, the momentum, and
Property Value the energy equation, which are presented as follows:
Density (kg/m3) 2700 ∇.ω = 0 (22)
Thermal conductivity (W/mK) 202.4
Specific heat (J/kgK) 870 1
∇.(ωω) = − ∇p + ∇.(μ∇ω) (23)
ρ
The base surface of the HHE is set to convective heat transfer with the
∇.(λ∇T) = ρcp ω.∇T (24)
outside environment at an ambient temperature of 303 K and a con­
vection coefficient of 5 W/m2K.
Where T and p denote the exhaust temperature and exhaust static

7
T.D. Hong et al. Results in Engineering 19 (2023) 101324

pressure, respectively.
The k-ε turbulent model is used because it accommodates flow with
high curvature and swirl. The transportation equations of this model
[20] are shown below:
[ ( )]
∂ ∂ ∂ ∂k μ
(ρε) + (ρkωi ) = μ + t + Gk + Gb − ρε (25)
∂t ∂xi ∂xj ∂xj σk
[ ( )]
∂ ∂ ∂ ∂ε μ ε ε2
(ρε) + (ρεωi ) = μ+ t + C1ε Gk − C2 ε ρ (26)
∂t ∂xi ∂xj ∂xj σε k k

The turbulent (or eddy) viscosity μt is:

k2
μt = C μ ρ (27)
ε

Where xi and xj are respectively rectangular coordinates; μ, k, ε, and ωi

are the dynamic viscosity, turbulence kinetic energy, turbulence energy
dissipation rate, and velocity components, respectively. Detailed de­
scriptions of σ k, Gk, Gb, σε, C1ε, C2ε, and Cμ can be found in Ref. [20].
In addition, the criterion for convergence of residual is set at 10− 6 for
the energy and 10− 4 for the remaining variables of the calculation
equations. The error of mass flow rate between the inlet and outlet of the
TGU is kept at less than 10− 8.

4. Results and discussions

4.1. Temperature distribution on the base of HHE Fig. 10. Average temperature distribution at TEM locations on the HHE base at
the engine speed of 2955 rpm for a) Model 1, and b) Model 2.

Figs. 9–11 illustrate the average temperature distribution at TEMs’

locations on the HHE base of different investigated models at the engine
speed of 990, 2955, and 4925 rpm, respectively. It is easy to recognize
from these three figures that the temperature variation of Model 1 along
the TGU centerline is wider than that of Model 2 at all engine speeds.
In Model 1, the HHE temperature is highest in the areas near the
muffler inlet, and it decreases towards the outlet. Model 1, with the
rectangular fin profile, has a higher front-end temperature due to
receiving greater thermal energy; later on, the heat loss to the envi­
ronment increases, leading to a decrease in exhaust energy. As a result,

Fig. 11. Average temperature distribution at TEM locations on the HHE base at
the engine speed of 4925 rpm for a) Model 1, and b) Model 2.

the temperature at the muffler’s rear area also declines. To analyze the
temperature uniformity on the HHE quantitatively, the maximum tem­
perature differences between TEMs’ locations of the two investigated
models at different engine speeds are examined and shown in Fig. 12.
Model 1, which is the original model, has maximum temperature
differences on the HHE base of 0.8, 2.3, and 3.1K at engine speeds of
990, 2955, and 4925 rpm, respectively. Meanwhile, those investigated
Fig. 9. Average temperature distribution at TEM locations on the HHE base at results of Model 2, which is the optimal model, have very low values of
the engine speed of 990 rpm for a) Model 1 and b) Model 2.

8
T.D. Hong et al. Results in Engineering 19 (2023) 101324

achieve the temperature uniformity of the HHE base may have to be

traded off with output power. However, the absorbed heat energy of
Model 2 is 2.9%, 2.2%, and 1.3% lower than that of Model 1 at engine
speeds of 990, 2955, and 4925 rpm, respectively.
Meanwhile, it is obvious that the significant temperature difference
between the TEMs leads to considerable differences in their voltages and
amperage. As a result, the TEMs producing lower voltages and amperage
can consume the electrical power generated by TEMs creating higher
voltages and amperages under certain operating conditions, especially
when the electrical load on the circuit is low. It leads to a reduction in
TEMs’ lifespan, high output power loss and poor efficiency. Further­
more, the large difference in the temperature distribution of the HHE
base can potentially cause thermal stress on the TEMs and reduce their
lifespan.

5. Conclusions

This research proposed a prediction model to construct the optimal

fin profile of the hot-side heat exchanger (HHE) to enhance the thermal
uniformity on the longitudinal cross-section of a motorcycle exhaust
thermoelectric generator. The effects of the optimal and original fin
Fig. 12. The maximum temperature differences between the TEMs’ locations
profiles on the temperature distributions and heat absorption of the heat
on the bases of the two investigated HHE models at different engine speeds.
exchanger were assessed using CFD models. The following conclusions
are summarized from the results:

➢ The optimal fin profile is effective in enhancing the thermal unifor­

mity of the heat exchanger. According to the results, the optimal fin
profile produces the highest temperature difference across the ther­
moelectric modules (TEMs) arranged on the HHE base of only 0.3K.
It can improve temperature uniformity by up to 91.3% compared to
the original fin profile.
➢ The temperature uniformity between TEMs can be traded off by
reducing the output power of the thermoelectric generator unit.
Compared with the original model, the reduction of heat absorption
of the optimal model is not significant, varying from 1.3% to 2.9%
depending on engine speeds.
➢ The theoretical model developed in this study can be applied to
general cases where heat exchangers have different numbers,
heights, pitches, and materials of fins, as well as the thermodynamic
and heat transfer parameters of the working fluid used in this study.
In addition, the method of optimizing the heat exchanger’s fin profile
contributes to the literature as a solution to improve thermal uni­
formity while the system compactness remains unchanged.
➢ The accuracy of predicting the optimal fin profile can be improved by
employing the computational programming method with more
computational elements. This model development will be conducted
in future works. The optimal fin profile will also be fabricated for
Fig. 13. The heat absorption of two investigated HHE models at
different speeds.
experiments validating the profile prediction model.

Credit author statement

only 0.1–0.3K at all considered engine speeds. It shows that the para­
bolic fin profile model, which is developed by the method mentioned in
Thong Duc Hong: Conceptualization; Methodology; Building the fin
this study, has been effective.
profile prediction model of HHE; Data curation; Project administration;
The maximum temperature difference of Model 2 is 0.7K, 2.1K, and
Supervision; Validation; Writing – review & editing. Minh Quang Pham:
2.8K lower than that of Model 1 at engine speeds of 990, 2955, and 4925
Investigation; Data curation; Resources; Software; Visualization;
rpm, respectively. It means that the optimal model can improve tem­
Writing – original draft. Quan Thien Phan Nghiem: Data curation;
perature uniformity by 87.5%, 91.3%, and 90.3% at the corresponding
Formal analysis; Validation; Visualization; Writing – original draft;
engine speeds, compared to the original model.
Writing – review & editing.

4.2. Heat absorption of HHE Declaration of competing interest

Fig. 13 reveals the heat absorption of the HHEs, which represents the The authors declare that they have no known competing financial
amount of heat transferred to the TEMs as well as the TGU output power. interests or personal relationships that could have appeared to influence
The results indicate that the larger the area of the fin exposed to the the work reported in this paper.
exhaust flow, the higher the heat absorption of the HHE. Furthermore,
the heat absorption increases as the engine speed rises. It means that to

9
T.D. Hong et al. Results in Engineering 19 (2023) 101324

Data availability [8] J. Chen, R. Wang, D. Luo, W. Zhou, Performance optimization of a segmented
converging thermoelectric generator for waste heat recovery, Appl. Therm. Eng.
202 (2022), 117843, https://doi.org/10.1016/j.applthermaleng.2021.117843.
Data will be made available on request. [9] R. Quan, Y. Li, T. Li, Y. Chang, H. Yan, Numerical and experimental study on
performance of a low-backpressure polyhedral thermoelectric generator for waste
Acknowledgment heat recovery, J. Therm. Sci. 32 (2023) 109–124, https://doi.org/10.1007/s11630-
022-1698-2.
[10] G. Shu, X. Ma, H. Tian, H. Yang, T. Chen, X. Li, Configuration optimization of the
We acknowledge Ho Chi Minh City University of Technology segmented modules in an exhaust-based thermoelectric generator for engine waste
(HCMUT), Vietnam National University Ho Chi Minh City (VNU-HCM) heat recovery, Energy 160 (2018) 612–624, https://doi.org/10.1016/j.
energy.2018.06.175.
for supporting this study. [11] T. Wang, S. Ma, Thermoelectric generator heat performance study about improved
fin structures, Therm. Sci. 22 (2018) 101–112, https://doi.org/10.2298/
References tsci150801064w.
[12] J. Wu, Y. Chen, Y. Dou, C. Ma, Q. Du, Q. Liu, Optimal design and performance
analysis of thermoelectric power generation device based on multi-objective
[1] M. Karvonen, R. Kapoor, A. Uusitalo, V. Ojanen, Technology competition in the
genetic algorithm, Adv. Theory Simulations. 4 (2021) 1–11, https://doi.org/
internal combustion engine waste heat recovery: a patent landscape analysis,
10.1002/adts.202000271.
J. Clean. Prod. 112 (2016) 3735–3743, https://doi.org/10.1016/j.
[13] R. Quan, G. Liu, C. Wang, W. Zhou, L. Huang, Y. Deng, Performance investigation
jclepro.2015.06.031.
of an exhaust thermoelectric generator for military SUV application, Coatings 8
[2] Z.G. Shen, L.L. Tian, X. Liu, Automotive exhaust thermoelectric generators: current
(2018), https://doi.org/10.3390/coatings8010045.
status, challenges and future prospects, Energy Convers. Manag. 195 (2019)
[14] T.D. Hong, Q.T.P. Nghiem, T.V. Mai, L.T. Le, A numerical simulation and
1138–1173, https://doi.org/10.1016/j.enconman.2019.05.087.
experimental study on thermal uniformity of heat exchanger in motorcycle
[3] M. MusiaŁ, M. Borcuch, K. Wojciechowski, The influence of a dispersion cone on
thermoelectric generator unit, JP J. Heat Mass Transf. 22 (2021) 89–105, https://
the temperature distribution in the heat exchanger of a thermoelectric generator,
doi.org/10.17654/HM022010089.
J. Electron. Mater. 45 (2016) 1517–1522, https://doi.org/10.1007/s11664-015-
[15] T.D. Hong, Q.T.P. Nghiem, B.T. Nguyen, T.V. Mai, Study on the effects of guide fins
4090-0.
structures on the flow and thermal uniformity of the motorcycle exhaust
[4] Z.B. Tang, Y.D. Deng, C.Q. Su, X.H. Yuan, Fluid analysis and improved structure of
thermoelectric generator, Multidiscip. Model. Mater. Struct. 18 (2022) 430–444,
an ATEG heat exchanger based on computational fluid dynamics, J. Electron.
https://doi.org/10.1108/MMMS-02-2022-0027.
Mater. 44 (2015) 1554–1561, https://doi.org/10.1007/s11664-014-3472-z.
[16] T.D. Hong, Q.T.P. Nghiem, T.T. Le, T.M.H. Tran, L.T. Le, Thermal uniformity
[5] X. Liu, C.G. Yu, S. Chen, Y.P. Wang, C.Q. Su, Experiments and simulations on a heat
enhancement of the motorcycle exhaust thermoelectric generator—Part 1: guide
exchanger of an automotive exhaust thermoelectric generation system under
fins optimization, Heat Transf 51 (2022) 6177–6196, https://doi.org/10.1002/
coupling conditions, J. Electron. Mater. 43 (2014) 2218–2223, https://doi.org/
htj.22587.
10.1007/s11664-014-3015-7.
[17] T.D. Hong, Q.T.P. Nghiem, P.B. Hua, T.V. Vo, Thermal uniformity enhancement of
[6] Y. Wang, C. Wu, Z. Tang, X. Yang, Y. Deng, C. Su, Optimization of fin distribution
the motorcycle exhaust thermoelectric generator—Part 2: muffler outlet
to improve the temperature uniformity of a heat exchanger in a thermoelectric
optimization, Heat Transf 51 (2022) 6470–6488, https://doi.org/10.1002/
generator, J. Electron. Mater. 44 (2015) 1724–1732, https://doi.org/10.1007/
htj.22600.
s11664-014-3527-1.
[18] T.L. Bergman, A.S. Lavine, F.P. Incropera, D.P. DeWitt, Fundamentals of Heat and
[7] D. Luo, R. Wang, W. Yu, W. Zhou, A numerical study on the performance of a
Mass Transfer, seventh ed., John Wiley & Sons, Inc, Hoboken, 2011.
converging thermoelectric generator system used for waste heat recovery, Appl.
[19] A.F. Mills, C.F.M. Coimbra, Basic Heat and Mass Transfer, third ed., Temporal
Energy 270 (2020), 115181, https://doi.org/10.1016/j.apenergy.2020.115181.
Publishing, San Diego, 2015.
[20] ANSYS Inc, ANSYS Fluent Theory Guide, ANSYS Inc, Canonsburg, 2013.

10