Journal of Physics: Conference Series

PAPER • OPEN ACCESS You may also like

- Dehydriding reaction of AlH3: in situ
Approximate analytical solution of the Graetz microscopic observations combined with
thermal and surface analyses
problem K Ikeda, S Muto, K Tatsumi et al.

- First-principles studies on the structural

stability of -AlH3 under pressure
To cite this article: A V Eremin 2021 J. Phys.: Conf. Ser. 1745 012116 Yi-Lin Lu and Hui Zhao

- Temperature distribution in a uniformly

moving medium
Joseph D Mitchell and Nikola P Petrov
View the article online for updates and enhancements.

This content was downloaded from IP address 180.180.43.208 on 14/06/2022 at 14:59

ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Approximate analytical solution of the Graetz

problem

A V Eremin1

1Samara State Technical University, Molodogvardeiskaya street 244, Samara, Russia,

443100

Abstract. The study on velocity and temperature distribution in fluid flows is

important both for the theory and practical applications. The design of efficient heat-
exchange equipment, the development of heat and thermomechanical modes of product
treatment, the determination of heat losses in the pipeline systems include the need to
determine the velocity and temperature fields in fluid flows. The key aspects of the
method have been considered with help of using the example of solving the Graetz
problem for parallel and cylindrical channels.
It is shown that finding the solution to a partial differential equation with respect to
the temperature function can be reduced to integrating an ordinary differential
equation with respect to the new unknown function q(η) which is the law of
temperature change in the center of the channel. The combined use of the heat balance
integral method and additional boundary characteristics made it possible to obtain a
simple in form analytical solution to the problem under consideration. It is noted that
the accuracy of the solutions obtained depends on the number of approximations used,
i.e. the number of terms of the approximating series.
When using only one term, i.e. already in the first approximation, the relative error
of the method is not more than 8 percent in the range of the longitudinal coordinate
change 0.1 ≤ η < ∞ and decreases to 4 percent in the second approximation. The
analytical form of the solutions obtained provides analyzing the isotherms fields,
calculating the average temperature, the Nusselt number, etc.

1. Introduction
Currently, heat exchangers using the incompressible fluid as a heat carrier have been widely
spread. Such heat exchangers can be used both for heating and for cooling of fluid flows.
It is widely used in heating, cooling, air conditioning, at petrochemical enterprises, in solar
energy, etc. The exchangers have various design features depending on the purpose. However,
in most cases, the channels through which the fluid moves have simple shapes that allows
applying the analytical methods to describe the heat transfer in them (parallel and cylindrical
channels, cylindrical annular, tubes). Such heat exchangers include, for example, double-pass
heat exchangers [1], plate heat exchangers, vacuum tube solar collector [2], microchannel heat
sinks [3] etc.
The mathematical description of heat and mass transfer processes in fluid flows includes
classical laws of fluid mechanics such as the Navier – Stokes equations, the continuity equation,
and the energy equation [4], [5]. Supplemented by the dependences of fluid physical properties
on temperature and pressure, these equations form a closed system of equations describing
convective heat exchange and fluid dynamics. The use of accurate analytical methods for solving

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

these problems is possible only in some simple cases. For example, when solving linear boundary
value problems for plane bodies and bodies with central axial symmetry, the variable separation
method (Fourier), Green’s functions method [6, 7] and integral transform methods with finite
and infinite limits of integration (Hankel, Laplace, Legendre transformations, etc.) [8, 9] are
used. Such solutions, as a rule, are expressed by complex analytical dependencies, infinite series
containing special functions that significantly limit their practical use.
Currently, numerical methods for studying the processes of heat and mass transfer in fluid
and gas flows are widely used [10, 11, 12, 13]. Modern software products make it possible to build
meshes, solve systems of linear equations and provide a wide range of tools for analyzing the
results obtained automatically. However, analytical solutions have some significant advantages
over numerical ones. For example, the solutions obtained in an analytical form provide
performing parametric analysis of the system under study, parametric identification, setup and
programming of measuring devices, planning of control actions in production processes, etc [14].
Thereby, approximate analytical methods for mathematical modeling of transfer processes in
fluid flows such as various modifications of the heat balance integral method [15, 16, 17, 18],
Ritz method [19, 20, 21], Kantorovich method [22], Galerkin method [23, 24, 25, 26] etc. have
rapid development.
This article presents the development results of an approximate analytical method for solving
one-dimensional heat and mass transfer problems which provide obtaining simple in form
solutions with accuracy sufficient for engineering applications. Using the example of solving
the Graetz problem [27, 28, 29, 30], it was shown that already in the second approximation the
solution error does not exceed 4 percent.

2. Formulation of the problem

As a specific example of the use of this method, consider the heat transfer problem in a stabilized
incompressible fluid flow moving in a parallel channel of width 2h (see Fig. 1). To derive the
differential equation describing this process, assume the following: 1) the flow is steady-state
and stabilized; 2) the fluid is incompressible; 3) the thermophysical properties are constant; 4)
the fluid flow model is laminar; 5) internal heat sources, as well as energy dissipation, are not
taken into account. Under these assumptions, the process of non-stationary heat transfer is
described by the Navier - Stokes equations in combination with the energy equation and the
continuity equation
∂T
+ ω · ∇T = a∇2 T, (1)
∂τ
∂ω
ρ + ρω · ∇ω = −∇P + µ∇2 ω, (2)
∂t
divω = 0. (3)

It is known that the heat exchange process with constant physical properties of a fluid does
not effect on the fluid flow. In this case, the fluid moves as if the flow were isothermal.
The coordinate system was introduced as it shown in 1. In this case, equation (2) with regard
to (3) is as follows
d2 ωx (y) ∆p
2
=− ,
dy µl
where x, y – longitudinal and transverse spatial coordinates; ωx – velocity vector projection
on the Ox axis; ∆p/l = const – pressure drop in the channel section of length l; µ – dynamic
viscosity.

2
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Figure 1. Heat transfer scheme.

The well-known law of velocity distribution in a parallel channel was obtained with help of
using the solution of the motion equation and considering the fact that the flow velocity on the
channel surface is zero (ωx (h) = ωx (−h) = 0) [4, 5].
"  2 # "  2 #
∆ph2 y 3 y
ωx (y) = 1− = ωav 1 − ,
2µl h 2 h
2
where ωav = 13 ∆ph
µl - average flow rate.
The energy equation (1) with regard to the found velocity profile is as follows
"  2 #
3 y ∂T (x, y) ∂ 2 T (x, y)
ωaν 1 − = . (4)
2 h ∂x ∂y 2

Consider the heat transfer problem in a parallel channel where the inlet temperature is equal
to Tin and the wall temperature Tw varies linearly

Tw (x) = Tin + Bx,

where B = ∂T (x,±h)
∂x - coefficient characterizing the change intensity of the wall temperature
along the longitudinal coordinate x, K/m. Therefore, the inlet wall temperature is equal to the
initial temperature and changes in the direction of the fluid flow. If B > 0, then Tw increases
along the length of the channel, if B < 0, then it decreases.
Given the symmetry of the problem, the boundary conditions to equation (4) are as follows

∂T (x, 0)
T (0, y) = Tin ; T (x, h) = Tin + Bx; = 0. (5)
∂y

The problem (4), (5) can be presented in a non-dimensional form. To do this, introduce the
following non-dimensional parameters:

T − Tin y 2ax 3Bh2 ωaν

Θ= ;ξ = ;η = 2 ;A = ,
Tin h 3h ωaν 2T0 a

3
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

where Θ – non-dimensional temperature; ξ, η – non-dimensional transverse and longitudinal

coordinates. Problem (4), (5) with respect to the introduced notation is as follows

∂Θ(η, ξ) ∂ 2 Θ(η, ξ)
(1 − ξ 2 ) = ; (6)
∂η ∂ξ 2

Θ(0, ξ) = 0; (7)
Θ(η, 1) = Aη; (8)
∂Θ(η, 0)
= 0. (9)
∂ξ
3. Method of solution

3.1. First approximation

According to the method developed a new unknown function - the law of temperature change
in the center of a parallel channel along its length was introduced

q(η) = Θ(η, 0). (10)

The problem solution should be found in the form of an algebraic or trigonometric series. To
solve the problem (6) - (9), use a trigonometric series
N
rπ
X  
Θ(η, ξ) = Aη + cos ξ bk (η), (11)
k=1
2

where r = 2k − 1 ; bk (η) – unknown coefficients.

Due to the chosen system of coordinate functions, the boundary conditions (8) and (9) are
satisfied at any values of the longitudinal coordinate η.
To obtain a solution to problem (6) - (9) in a first approximation, we require that relation
(11) satisfies the additional boundary condition (10). Note that an approximation number is
the number of terms under the sum sign in relation (11).
The condition (10) introduction doesn’t change the initial problem statement because the
temperature Θ(η, 0) on the axis of the channel is the new value. It is also worth noting that the
function q(x) is not equal to 0 at any value x > 0. This is due to the fact that the derivation
of the energy equation is based on the Fourier hypothesis, according to which a change in the
temperature gradient leads to the instantaneous heat flux. It was shown in [30, 31] that this
assumption results in the fact that the rate of heat transfer is infinite.
To obtain a solution in the first approximation (N = 1), substitute (11) into (10)

q(η) = Aη + b1 (η). (12)

Expressing b1 (η) using (12) and substituting into (11), we obtain

π
Θ(η, ξ) = Aη + cos ξ(q(η) − Aη). (13)
2
To determine the unknown function q(η), we require that relation (13) satisfies not the initial
differential equation but the integral equation – the heat balance integral [15, 16, 17, 18, 31]

Z1 Z1
2 ∂Θ(η, ξ) ∂ 2 Θ(η, ξ)
(1 − ξ ) dξ = dξ. (14)
∂η ∂ξ 2
0 0

4
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Calculating the integral (14), we obtain the following differential equation

∂q(η) π 4 π4A π3A

+ q(η) − η+ − A = 0,
∂η 32 32 24

so, the general solution is written as

π4 η 4
 
q(η) = C1 e− 32 +A η− π . (15)
3

Substituting (15) into (13), we obtain

πξ 4
4A
  
− π32η
Θ(η, ξ) = cos C1 e − + Aη. (16)
2 3π

Relation (16) exactly satisfies the boundary conditions (8), (9), additional condition (10), and
also the heat balance integral (averaged over the range of the transverse coordinate to equation
(6)). To fulfill the boundary condition (7), we find its residual and require the orthogonality of
the residual to the coordinate function, i.e.

Z1
πξ
 
Θ(0, ξ)cos dξ = 0. (17)
2
0

Using solution (17) we can find C1 = 4A

3π . Taking into account the found integration constant,
the solution of problem (6) – (9) in the first approximation can be written as

4A − π4 η π
   
Θ(η, ξ) = Aη + e 32 − 1 cos ξ . (18)
3π 2

The results of temperature calculations using formula (18) in comparison with a numerical
solution are shown in Fig. 2. Their analysis shows that already in the first approximation the
relative error of the method is not more than 8 percent. The true value of the temperature in
determining the resulting error was the value obtained by a numerical calculation. An algorithm
for the solution of the problem (6) - (9) is presented in the appendix.

3.2. The second approximation

To increase the accuracy of the solutions obtained, it is necessary to increase the number of terms
of the series (11), here the number of unknown coefficients bk (η) will increase. To determine
them (in addition to condition (10)) we use additional boundary conditions (characteristics).
To obtain them, the initial differential equation is repeatedly differentiated with respect to the
points ξ = 0 and ξ = 1 . It was shown in [32] that the satisfaction of the additional boundary
conditions obtained in this way is equivalent to the fulfillment of the initial differential equation
at the boundary points. In addition, it was noted in [32, 33] that the use of additional boundary
characteristics with an increase in the number of approximations leads to the fulfillment of the
equation inside the area as well.
To obtain the second additional boundary condition (the first one is relation (10)), write
equation (6) with regard to the point ξ = 0

∂Θ(η, 0) ∂ 2 Θ(η, 0)
= . (19)
∂η ∂ξ 2

5
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Figure 2. Calculation results of the temperature in a parallel channel (A = 1).

Relation (19) with respect to (10) can be written as

dq(η) ∂ 2 Θ(η, 0)
= . (20)
dη ∂ξ 2

We obtain one more additional condition by writing equation (6) at the point ξ = 1

∂ 2 Θ(η, 1)
= 0.
∂ξ 2

To obtain the following additional boundary characteristics, we differentiate the initial

differential equation (6) with respect to the variable ξ, as a result we have

∂Θ(η, ξ) ∂ 2 Θ(η, ξ) ∂ 3 Θ(η, ξ)

−2ξ + (1 − ξ 2 ) = . (21)
∂η ∂ξ∂η ∂ξ 3

We can now write relation (21) with regard to the point ξ = 0

∂ 2 Θ(η, 0) ∂ 3 Θ(η, 0)
= . (22)
∂ξ∂η ∂ξ 3

Relation (22) with respect to (9) is the third additional boundary condition

∂ 3 Θ(η, 0)
= 0.
∂ξ 3

Writing relation (21) at a point ξ = 1 , we can obtain the fourth additional boundary condition

∂ 3 Θ(η, 1)
−2A = .
∂ξ 3

6
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Additional boundary characteristics can also be obtained for subsequent approximations by

this method. To obtain a solution to problem (6) - (9) in the second approximation (N = 2),
we use two additional conditions. To do this, substitute (11) into (10) and (20)

 ∂q(η) π 2
b 1 (η) + b2 (η) − q(η) + Aη = 0, + [b1 (η) + 9b2 (η)] = 0. (23)
∂(η) 4

The solution of the system of algebraic equations is used to determine the unknown coefficients
b1 (η) and b2 (η)

9 1 dq(η) 1 1 dq(η)
b1 (η) = − [Aη − q(η)] + 2 , b2 (η) = [Aη − q(η)] + 2 . (24)
8 2π dη 8 2π dη

To determine the unknown function q(η), as in the first approximation, we require the
unknown solution (11) to satisfy the heat balance integral (14). Calculating (14), we obtain
an ordinary second-order differential equation with respect to the unknown function q(η)

d2 q(η) dq(η)
0.02711 2
+ 0.90123 + 2.3562(q(η) − Aη) + 0.08375 = 0. (25)
dη dη

Its solution is as follows

q(η) = C1 e−30.383η + C2 e−2.8606η + (η − 0.41804)A. (26)

The integration constants can be obtained by finding the residual of the boundary condition
(7) and the requirement of the orthogonality of the residual to all coordinate functions, i.e.

Z1
rπξ
 
Θ(0, ξ)cos dξ = 0(r = 2k − 1; k = 1, N ). (27)
2
0

For the second approximation (N = 2), relation (27) is a system of two algebraic equations
which are used to find
C1 = −0.0071153A, C2 = 0.42515A. (28)
Substituting (24), (26) and (28) into (11), we obtain the unknown temperature distribution
function in the parallel channel in the second approximation.
The calculation results of the non-dimensional temperature are presented in Fig. 2. A
significant increase in accuracy is observed in comparison with the first approximation. So, the
error in calculating the temperature in the given range of the variable η changes is not more
than 4 percent. To further increase the accuracy, it is necessary to use more terms of the series
(11), and additional boundary characteristics are used as well.

3.3. Determination of heat transfer coefficient

The average mass temperature can be found by formula

Rh
ωx (y)T (x, y)dy Zh
0 1
T (x) = = ωav (y)T (x, y)dy. (29)
Rh hωav
ωx (y)dy 0
0

7
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Figure 3. Distribution of average mass temperature and Nusselt number along the length of a
parallel channel (A = 1).

Relation (29) can be reduced to a non-dimensional formula

Z1
T (x) − Tin 3
Θ(η) = = Θ(η, ξ)(1 − ξ 2 )dξ. (30)
Tin 2
0

Nusselt criterion is determined according to the relation

2ah ∂Θ(η, ξ)
Nu = = −2(Θ(η) − Aη)−1 .
λ ∂ξ

The limit value of the Nusselt number x → ∞ in the second approximation was N u∞ = 4.104.
Graphs of changes in the Nusselt criterion and average mass temperature along the channel
length are shown in Fig. 3.

3.4. Analysis of isotherms

The temperature distribution functions obtained according to the above method have a simple
analytical form that makes it possible to study the heat transfer process in the fields of isotherms.
The law of distribution of isotherms in the first approximation can be obtained from relation
(18) by expressing the variable ξ as a function of temperature and longitudinal coordinate

ξ(Θ, η) = 2(1 − π −1 arccos(0.75A−1 π[Aη − Θ][exp(−3.044η) − 1]−1 ) (31)

The results of building isotherms using formula (29) are presented in Fig. 4. The analysis
shows that isotherms occur on the surface of the body (at a point (ξ = 0) and propagate along
the longitudinal coordinate η. Note, that for each isotherm it is possible to determine two
characteristic coordinates - the coordinates in which it appears and disappears.

8
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Figure 4. Distribution of isotherms in a parallel channel (A = 1).

4. Results and Discussion

The main advantage of this method is the possibility of solving the equations with non-separable
variables [34] with the temperature dependence of the thermophysical properties of substances,
under boundary conditions of the second and third kind, for channels of various shapes. For
example, for a cylindrical channel the heat transfer process is described by the following
differential equation
∂Θ(η, ξ) 1 ∂Θ(η, ξ) ∂ 2 Θ(η, ξ)
(1 − ξ 2 ) = + .
∂η ξ ∂ξ ∂ξ 2
This problem is called the Graetz problem. Its exact solution is expressed in an infinite
series in [35]. Using the method considered in the article, a simple in form approximate solution
of the Graetz problem can be found. Under the boundary conditions (7) - (9) in the first
approximation, it is as follows

Θ(η, ξ) = A[η − 0.125(1 − e−12η )(1 − ξ 2 )].

Thus, the proposed method is recommended for obtaining the analytical solutions that are
simple in form, in cases where an error (3 percent) can be considered as satisfactory. It is worth
noting that this method can be used to obtain high-precision solutions, however, in this case,
the formula of the resulting solution is rather complicated, and the amount of computational
work increases significantly.

5. Conclusion
Based on the applied additional boundary conditions and an additional unknown function, the
approximate solution to the heat transfer problem in a parallel channel has been obtained. It
is shown that the problem of finding a solution to a partial differential equation with respect to
the temperature function can be reduced to integrating an ordinary differential equation with
respect to the new unknown function. The error in determining the temperature, in this case, is
found by the number of performed approximations and is no more than 3 percent already in the
second approximation. The obtained results can be useful to designers and operators of heat
exchange equipment.

9
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at
https://data.mendeley.com/datasets/7mj2d3zw6n/1.

6. Acknowledgements
The reported study was funded by RFBR (project number 20-38-70021) and the Council on
grants of the President of the Russian Federation (project number MK2614.2019.8).

7. References
[1] Lin G G, Ho C D, Huang J J and Chen Y C 2012 International Journal of Heat and
Mass Transfer 55 2261-2267.
[2] Badar A W, Buchholz R and Ziegler F 2012 Solar Energy 86 175-189.
[3] Azari M, Sadeghi A and Chakraborty S 2019 International Journal of Heat and Mass
Transfer 128 150-160.
[4] Landau L D and Lifshitz E M 1987 Fluid Mechanics (Pergamon, Oxford).
[5] Kays W M 1993 Convective heat and mass transfer (McGraw-Hill, New York).
[6] Sneddon I N 1995 Fourier transforms (Dover Publications, New York).
[7] Cherati D Y and Ghasemi-Fare O 2019 Geothermics 81 224-234.
[8] Tranter C J 1966 Integral transforms in matematikal physics (Methuen, London).
[9] Tsoi P V 2005 System methods for calculating the heat and mass transfer boundary value
problems (MEI Publishing House).
[10] Christie I, Griffiths D F, Mitchell A R and Zienkiewicz O C 1976 International Journal for
Numerical Methods in Engineering 10 1389-1396.
[11] Pigeonneau F, Jaffrennou B, Letailleur A and Limouzin K 2016 International Journal of
Heat and Mass Transfer 96 381-395.
[12] Campo A and Marucho M D 2017 Journal of Thermophysics and Heat Transfer 32
532-537.
[13] Ahmed M, Yaseen M and Yusoff M Z 2017 Case studies in thermal engineering 10 560-569.
[14] Haddout Y and Lahjomri J 2015 International Journal of Heat and Mass Transfer 80
673-687.
[15] Layeni O and Johnson J 2012 Applied Mathematics and Computation 218 7431-7444.
[16] Mitchell S L and Myers T 2010 International Journal of Heat and Mass
Transfer 53 3540-3551.
[17] Mitchell S L and Myers T 2010 SIAM review 52 57-86.
[18] Novozhilov V 2009 Thermal Science 13 73-80.
[19] Dutta S, Sil A N, Saha J K and Mukherjee T K 2018 International Journal of
Quantum Chemistry 118 e25577.
[20] Lotfi A and Yousefi S A 2017 Journal of Optimization Theory and Applications 174
238-255.
[21] Falk R S 1976 Revue française d’automatique, informatique, recherche opé rationnelle.
Analyse numérique 10 39-48.
[22] Kantorovich L V 1934 Reports of the USSR Academy of Sciences 2 532-536.
[23] Rao T and Chakraverty S 2017 Coupled systems mechanics 6 487-499.
[24] Nourgaliev R, Luo H, Weston B, Anderson A, Schofield S, Dunn T and Delplanque J P
2016 Journal of Computational Physics 305 964-996.
[25] Belytschko T, Lu Y Y and Gu L 1994 International journal for numerical methods in
engineering 37 229-256.
[26] Arnold D N, Brezzi F, Cockburn B and Marini L D 2002 SIAM journal on numerical
analysis 39 1749-1779.

10
ITNT 2020 IOP Publishing
Journal of Physics: Conference Series 1745 (2021) 012116 doi:10.1088/1742-6596/1745/1/012116

[27] Letelier M F, Hinojosa C B and Siginer D A 2017 International Journal of Thermal Sciences 111
369-378.
[28] Bennett T D 2019 International Journal of Heat and Mass Transfer 136 832-841.
[29] Eremin A 2019 International Science and Technology Conference EastConf 1-5.
[30] Eremin A, Kudinov V and Stefanyuk E 2018 Fluid Dynamics 53 S29-S39.
[31] Kudinov V, Eremin A and Kudinov I 2017 Thermophysics and Aeromechanics 24 901-907.
[32] Fedorov F and Egorov I 2000 Boundary method for solving applied problems of
mathematical physics (The science. Siberian Publishing Company RAS).
[33] Eremin A, Kudinov I, Dovgyallo A and Kudinov V 2017 Journal of engineering physics
and thermophysics 90 1234-1242.
[34] Kudinov I, Kudinov V, Kotova E and Eremin A 2017 Journal of Engineering Physics
and Thermophysics 90 1317-1327.
[35] Petukhov B S 1967 Energy 411.

11