Page Proof

-Please check throughout text for

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623
2050013 spelling errors, figures and tables.
-Please check title, author names,
affiliations and emails.
-Please check keywords and codes.
-Please check the left (on even pages)
and right (on odd pages) running title.

1 International Journal of Modeling, Simulation,

2 and Scientific Computing
3 (2020) 2050013 (22 pages)
4 c World Scientific Publishing Company
5 DOI: 10.1142/S1793962320500130

6 Analytical solution and numerical simulation

7 of the thermal entrance region problem for
8 laminar flow through a circular pipe

9 Ali Belhocine
10 Faculty of Mechanical Engineering
11 University of Sciences and the Technology of Oran
12 L.P 1505, El - Mnaouer, USTO 31000 Oran Algeria
13 belhocine.2018@gmail.com

14 Oday Ibraheem Abdullah

15 System Technologies and Mechanical
16 Design Methodology, Hamburg University
17 of Technology Hamburg Germany
18 odayia2006@yahoo.com

19 Received 7 November 2017

20 Revised 23 January 2020
21 Accepted 30 January 2020
22 Published

23 In this paper, the assumptions implicited in Leveque’s approximation are re-examined,

24 and the variation of the temperature and the thickness of the boundary layer were
25 illustrated using the developed solution. By defining a similarity variable, the govern-
26 ing equations are reduced to a dimensionless equation with an analytic solution in the
27 entrance region. This report gives justification for the similarity variable via scaling
28 analysis, details the process of converting to a similarity form, and presents a simi-
29 larity solution. The analytical solutions are then checked against numerical solution
30 programming by FORTRAN code obtained via using Runge–Kutta fourth order (RK4)
31 method. Finally, other important thermal results obtained from this analysis, such as;
32 approximate Nusselt number in the thermal entrance region was discussed in detail. A
33 comparison with the previous study available in literature has been done and found an
34 excellent agreement with the published data.

35 Keywords: Thermal entrance region; thermal boundary layer; dimensionless variables;

36 temperature; Nusselt number; Runge–Kutta method.

37 Nomenclature
[a, b] : Internal of integration of PDE’s
Cp : Heat capacity (J/kgK)
38
Cn : Coefficient of solution defined in Eq. (8)
F : Temperature field function which is the solution of Eq. (18)

2050013-1
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

f (x, y, z) : Function defined in two general first-order ODEs

Gn : Eigenfunctions of a proper Sturm–Liouville system
g(x, y, z) : Function defined in two general first-order ODEs
k : Thermal conductivity (W/mK)
k0 : Increment based on the slope at the beginning of the interval
k1 : Increment based on the slope at the midpoint of the interval
k2 : Increment based on the slope at the midpoint of the interval
k3 : Increment based on the slope at the end of the interval
l0 : Increment based on the slope at the beginning of the interval
l1 : Increment based on the slope at the midpoint of the interval
l2 : Increment based on the slope at the midpoint of the interval
l3 : Increment based on the slope at the end of the interval
h : Heat transfer coefficient (W/m2 K)
h : Step size
N : Number of steps
Nu : Nusselt number
Pe : Peclet number
qω : Heat flux from fluid wall (W/m2 )
1
R : Tube radius (m)
Re : Reynolds number
r, θ, z : Cylindrical coordinates (m)
T : Temperature of the fluid inside a circular tube (K)
Tb : Bulk temperature (K)
T0 : Temperature of the fluid entering the tube (K)
Tω : Temperature of the fluid on the wall of the tube (K)
t : Time (s)
x : Transversal coordinate, (m)
X : Dimensionless radial direction (m)
Y : Dimensionless radial direction (m)
yi , yi+1 : Values of y at xi and xi+1 respectively
ur , uθ , uz : Cartesian velocity components (m/s)
V : Fluid longitudinal velocity (m/s)
υ0 : Maximum axial velocity of the fluid (m/s)
Z : Dimensionless axial direction
z : Longitudinal coordinate (m)
zi , zi+1 : Values of z at xi and xi+1 , respectively

2 Greek Letters
α : Thermal diffusivity (m2 /s)
βn : Eigenvalues
3
Γ : Gamma function
δ : Thickness of the thermal boundary layer (m)

2050013-2
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

θ : Dimensionless temperature
θb : Dimensionless bulk average temperature
1 ρ : Density of the fluid (kg/m3 )
μ : Dynamic viscosity of the fluid (kg/ms)
η : Similarity variable

2 Acronyms
3 RK04 : 4th order Runge-Kutta method

4 1. Introduction
5 The experimental studies carried out by the researchers are generally in the field
6 of convective thermal transfers which several authors have addressed in their work,
7 heat transfer problems in a flow of fully developed laminar fluid through circular
8 conduits. Azimi et al.1 studied the effect of the nanoparticle size (diameter) on
9 the heat transfer coefficient of forced convective heat transfer of nanofluid in the
10 fully developed laminar region of a horizontal tube. Ibrahim and Shanker2 carried
11 out a numerical analysis based on thermal boundary condition to investigate the
12 problem of magneto hydrodynamic (MHD) boundary-layer flow and heat transfer
13 of a viscous incompressible fluid over a fixed plate. The governing boundary-layer
14 equations were reduced to couple higher-order nonlinear ordinary differential equa-
15 tions by using similarity transformation. Deepu et al.3 used three-stage Runge–
16 Kutta method for explicit time integration for the simulation of two-dimensional
17 axisymmetric/planar high speed compressible turbulent reacting shear layers. Zhu
18 et al.4 used fourth-order Runge–Kutta method to solve the dynamic model based
19 on lumped-mass method (LMM). Krivovichev5 performed stability analysis of lat-
20 tice Boltzmann equations (LBEs) on initial conditions for one-dimensional diffusion
21 and analytically analyzed by the method of differential approximation. Choudhary
22 and Gejji6 extended invariant subspace method for solving systems of multi-term
23 fractional partial differential equations (FPDEs) involving both time and space
24 fractional derivatives. Yaghobi Moghaddam7 employed Homotopy Perturbation
25 Method (HPM), Finite Volume Method and Analytical (Laplace) Method to pro-
26 vide proper solutions for some nonlinear differential equations which describe main
27 mechanisms governing heap (bio) leaching process. Chauhan and Srivastava8 used
28 Runge–Kutta (RK) three-stage geometric mean method to solve the initial value
29 problem that arises in autonomous systems. Bhaumik and Maity9 established a
30 two-dimensional axisymmetric thermal model using finite element method (FEM)
31 has been established for predicting the temperature distribution profile on the work
32 piece during electro discharge machining (EDM) from the temperature isotherm.
33 Padilla and Liceaga10 carried out a numerical study in order to show the improve-
34 ment that can be obtained with Chebyshev polynomials-based methods over the
35 classical finite difference schemes to obtain numerical solutions of cardiac models.

2050013-3
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

1 Jha et al.11 conducted numerical solution of three-space dimensions partial differen-

2 tial equations of elliptic type using high-order finite-difference compact scheme. He
3 et al.12 used the FEM to solve the problem for two-dimensional nonlinear modified
4 time-fractional fourth-order diffusion equation. Khan et al.13 presented an approx-
5 imate solution of nonlinear fractional differential equations (FDEs) by using HPM.
6 Li and Huang14 conducted new numerical technique for a class of 2D nonlinear
7 fractional diffusion-wave equations with the Caputo-type temporal derivative and
8 Riesz-type spatial derivative by using Galerkin finite element scheme. An analysis
9 of the heat transfer through a fluid flow and over the boundary layer was established
10 by Hamad and Ferdows.15 Another study was carried out by Wei and Al-Ashhab16
11 on boundary layers of a non-Newtonian fluid subject to new boundary conditions.
12 A study was conducted by Trı̂mbijas et al.17 to analyze a boundary layer in mixed
13 convection while employing a similarity technique to which partial differential equa-
14 tions are reduced to ordinary differential equations. Ahmed18 analyzed a boundary
15 layer in natural convection in the presence of transient wall temperatures using the
16 finite difference method. Shen and Lu19 have modeled the problem of free turbu-
17 lence using the Runge–Kutta method for the prediction of the three-dimensional
18 boundary layer. Mahanthesh et al.20 carried out a heat flow analysis on the basis
19 of a mathematical model managed by the boundary layer hypotheses while using
20 the similarity method to reduce the governing equations. Eldesoky et al.21 studied
21 the peristaltic pumping of a compressible fluid in a tube using a perturbation anal-
22 ysis. Baehr and Stephan22 and Stephan23 conducted research on heat transfer in
23 the input region with well-specified geometry. Additional work has been done by
24 Asako et al.24 Shah and London,25 Kakac et al.,26 Ebadian and Dong,27 and Kakac
25 and Yener28 on triangular, rectangular and circular geometry. In the literature,
26 we can find other thermal problems performed on other forms of tube geometry
27 such as; the circular channels, the circular and the parallel plate and a rectangular
28 channel. Thanks to these geometries, the thermal problems have been solved easily
29 using analytical methods, of which whose predictions of the thermal transfer of the
30 cylindrical walls were approached on several models. Hausen29 developed a model
31 to study the Graetz problem inside a circular tube. Churchill and Ozoe30,31 pro-
32 posed simple models to develop flux in a circular duct. With the fully developed
33 asymptote, and for the thermal input region the Leveque solution was combined by
34 Churchill and Ozoe.30,31 For the Graetz problem, and in order to predict the ther-
35 mal characteristics in an arbitrary form of the tube, models have been developed
36 by Yilmaz and Cihan.32,33 These two authors developed models for uniform wall
37 flow conditions (H) and a uniform wall temperature (T) in order to predict the fully
38 developed number of Nusselt. These models were fitted to these models with the
39 Leveque generalized solution so that the input offers an approved model along the
40 length of the tube. In the entrance area of the circular duct, two distinct problems
41 must be considered. One assumes the existence of a fully developed hydrodynamic
42 boundary layer while the other problem is more popular with developing thermal

2050013-4
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

1 boundary layers. In the case of Graetz’s classical problem, the velocity distribution
2 is fully developed and the temperature of the fluid tends to propagate fairly rapidly
3 inside the tube. In the input region, the use of the Levèque approach gives us better
4 convergent results in the approximate solution on which we can assume that the
5 velocity gradient is quite linear and the boundary layer is considered thin. Belhocine
6 and Wan Omar,34 Belhocine35 conducted an analysis to predict the distribution of
7 the dimensionless temperature in a fully developed laminar flow in a cylindrical
8 pipe. Recently, Belhocine and Wan Omar36 were able to develop the analytical
9 solution of the problem of convective heat transfer within a pipe whose solution
10 obtained is in the forms of the hypergeometric series.
11 The main objective of this work is to develop an exact solution of the thermal
12 boundary layer at the inlet of a circular pipe for a fully developed flow of laminar
13 fluid commonly called the Levèque approximation. The calculation methodology
14 that we have followed is based on the method of solution in similarity of the variables
15 in order to predict the dimensionless temperature as well as the thickness of the
16 thermal boundary layer near the entrance of the flow. Several steps have been
17 discussed here on the governing equation of the temperature field to reach the
18 solution such that the non-dimensionalization and the use similarity variables, the
19 transform the PDE to a set of PDE’s summarization of the boundary conditions and
20 the integration of the equation. We then compare the exact approximate solution
21 of the levèque problem, with the numerical results using a Runge-Kutta fourth
22 order (RK4) algorithm implemented by the FORTRAN code. The profiles of the
23 solutions are provided from which we infer that the numerical and exact solutions
24 agreed very well. Another result that we obtained from this study is the number of
25 Nusselt in the thermal entrance region to which a parametric study was carried out
26 and discussed well for the impact of the scientific contribution. The results obtained
27 are presented and compared with the previous results published in the literature
28 and found to be in good agreement.

29 2. The Governing Heat Diffusion Equation

30 The total thermal energy balance, which is based on the use of equations of con-
31 tinuity and momentum, is simplified by the expression obtained by Bird, Stewart,
32 and Lightfoot37 is as follows;
     
∂T ∂T uθ ∂T ∂T 1 ∂ ∂T 1 ∂2T ∂2T
ρCp + ur + + uz =k r + 2 2 +
∂t ∂r r ∂θ ∂z r ∂r ∂r r ∂θ ∂z 2
 2   2  2   2
∂ur 1 ∂uθ ∂uz ∂uθ 1 ∂uz
+ 2μ + + ur + + 2μ +
∂r r ∂θ ∂z ∂z r ∂θ
 2   2
∂uz ∂ur 1 ∂ur ∂  uθ
+ + +μ +r . (1)
∂r ∂z r ∂θ ∂r r

2050013-5
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

1 3. The Graetz Poiseuille Flow Problem

2 The Graetz problem consists of determining the temperature in a steady state of
3 a fluid passing through a circular pipe whose fully developed laminar flow. Thus,
4 it is a transfer of heat by convection of a fluid approaching the inlet section of
5 a cylindrical tube with a constant temperature T0 whose wall is subjected to a
6 constant temperature Tω . The geometry of the problem is shown in Fig. 1.
7 The contour of the velocity of the flow becomes a stable contour after a certain
8 distance from the hydrodynamic inlet and it remains practically fully parabolic and
9 invariable along the circular tube. Our context for solving the thermal problem is
10 to find the behavior of the temperature field as it evolves to be uniform at the
11 temperature of the downstream wall. The distribution of the velocity of the flow is
12 not subordinated by the variation of the temperature as long as the nature of the
13 fluid is incompressible.

14 • The fluid flow is completely laminar in steady state and fully developed
15 • The flow is considered incompressible Newtonian whose properties ρ, μ, Cp, k.
16 are constant and do not depend on temperature.
17 • The temperature does not depend on the angular coordinate θ( ∂T∂θ = 0),
18 • Negligible viscous dissipation

19 The expression of the velocity of a fully developed flow is given by the following
20 form:
  r 2 
uz = v0 1 − , (2)
R
21 where v0 is the maximum speed (center of the tube), ur = 0, and uθ = 0.0.
22 The energy equation is subject to the assumptions mentioned above, Eq. (1)
23 can be written as follows:
     
r2 ∂T 1 ∂ ∂T ∂2T
v0 1 − 2 =α r + , (3)
R ∂z r ∂r ∂r ∂z 2
24 where α = k/ρCp is called the thermal diffusivity which has dimensions (m2 /s), our
25 problem is subjected to the following boundary conditions; at the inlet of the tube

z R
Fluid at
vr
T0
r

T(R, z )=Tω

Fig. 1. Illustration of Graetz problem.

2050013-6
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

1 T (r, 0) = T0 ; at the wall of the tube T (R, z) = Tω and at the centerline T (0, z) is
2 finite or ∂T
∂r (0, z) = 0 = 0.
3 Consider the following dimensionless terms:
⎧
⎪ T − Tω
⎪
⎪ θ=
⎪
⎪ T 0 − Tω
⎪
⎨
r
Y = ,
⎪
⎪ R
⎪
⎪
⎪
⎪
⎩Z = z
RPe
4 where the Péclet Number Pe = Rv α
0

5 By substituting the variables T, r, z for their expressions as a function of the

6 dimensionless variables θ, Y, Z in the heat equations, we obtain the following equa-
7 tions:
 
2 ∂θ 1 ∂ ∂θ 1 ∂2θ
(1 − Y ) = Y + 2 , (4)
∂Z Y ∂Y ∂Y Pe ∂Z 2
∂θ
θ(Y, 0) = 1, θ(1, Z) = 0, (0, Z) = 0. (5)
∂Y
8 The influence of the axial diffusion is totally neglected when we apply the assump-
9 tions of the boundary layer, which implies the resolution of the following dimen-
10 sionless equation
 
2 ∂θ 1 ∂ ∂θ
(1 − Y ) = Y . (6)
∂Z Y ∂Y ∂Y
11 This equation can be solved by the technique of the separations of the variables at
12 which the temperature that we seek will be found in terms of hypergeometic series
∞
 2
θ(Y, Z) = Cn e−βn Gn (Y ), (7)
n=1

13 where βn and Gn (Y ), are respectively the eigenvalues and the eigenfunctions asso-
14 ciated with the Sturm–Liouville problems. The coefficients Cn can be obtained by
15 using the orthogonality property of the eigenfunctions defined as follows:
1
Gn (Y )Y (1 − Y 2 )dY
Cn = 01 . (8)
2 2
0 Gn (Y )Y (1 − Y )dY

16 4. The Léveque Approximation

17 For all values of the axial position, the orthogonal function expansion solution
18 obtained in the resolution of the classical Graetz problem is quite convergent, but
19 the convergence is very slow as soon as one approaches the input tube. Indeed, for
2
20 very long values of Z, the factor e−λn Z has become converged. Lévêque38 examined
21 the thermal entrance zone in a cylindrical pipe while developing an approximate

2050013-7
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

z R
Fluid at
vr
T0
r
δt

T(R, z )=Tω

Fig. 2. Simplifying representation of the Lévêque approximation.

1 solution which is formally advantageous when the orthogonal function tends grad-
2 ually towards convergence (Fig. 2).
3 According to Lévêque’s assumption, we can take the thickness of the boundary
4 layer δt R, which leads to the following simplifications:

5 • In the radial conduction term, we can neglect here the effects of curvature. Thus,
1 ∂ ∂T 1 ∂ ∂T ∂2 T
6 derivative r ∂r (r ∂r ) is approximated by R ∂r (R ∂r )= ∂r 2
7 • We are interested in the thermal boundary layer of the velocity allocation of
8 which it can be developed in a Taylor series from the wall of the pipe according
9 to a measured position, if we keep the first non-zero term.

10 If we set x = R − r, the speed distribution will take the following form:

   
(R − r)2 x x2 x
vz (r) = v0 1 − = v0 2 − ≈ 2v0 . (9)
R2 R R2 R
11 • Let us know, the boundary conditions of the flow entering the pipe are those that
12 lie outside the boundary layer, we will exploit the boundary condition T (x →
13 ∞) → T0 instead of that market at the center of the tube to arrive at the Graetz
14 solution.

15 5. Governing Lévêque’s Equation

16 Starting from the reduced energy equation whose axial conduction has been
17 neglected yet, and considering the said hypotheses, for the temperature field, we
18 obtain the following governing equation
x ∂T ∂2T
2v0 =α 2. (10)
R ∂z ∂x
19 Using the string rule, in order to convert the second derivative of r into that of x,
20 we get
∂θ ∂2θ
X = . (11)
∂Z ∂X 2

2050013-8
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

1 5.1. Boundary conditions

2 The temperature T (x, z), is controlled by boundary conditions which are fixed like
3 this.
⎧
⎪
⎪ T (x, 0) = T0
⎨
T (0, z) = Tω .
⎪
⎪
⎩T (∞, z) = T0

4 5.2. Nondimensionalization
5 Now, we will use dimensionless variables for the simplification of the equation. For
6 this, we introduce the temperature and the axial coordinate of the following forms
T − Tω z
θ= , Z= . (12)
T0 − Tω RPe
7 The scaled governing equation placed on the wall by X = x/R and boundary
8 conditions are given as follows
∂θ ∂2θ
2X = , (13)
∂Z ∂X 2
⎧
⎪
⎪θ(X, 0) = 1
⎨
θ(0, Z) = 0 .
⎪
⎪
⎩θ(∞, Z) = 1

9 6. Analytical Methodology for Problem Solving: Temperature

10 Field and Thermal Boundary Layer
11 At the current problem, we are looking for a similarity solution for the temperature
12 field, we assume, θ(X, Z) = F (η) = F (η), where η = X/δ(Z) is the similarity
13 variable and δ(Z) is an ignored variable that provides us with the thickness of
14 the thermal boundary layer. Using the chain rule, we will perform the following
15 necessary transformations.
 
∂θ ∂η dF Xdδ dF ηdδ dF
= = − 2 =− , (14)
∂Z ∂Z dη δ dZ dη δdZ dη
∂θ ∂η dF 1 dF
= = , (15)
∂X ∂X dη δ dη
     
∂2θ ∂ 1 dF 1 ∂ dF 1 ∂η d dF 1 d2 F
= = = = . (16)
∂X 2 ∂X δ(Z) dη δ ∂X dη δ ∂X dη dη δ 2 dη 2
16 Using the important results, the previous equation for θ(X, Z), has been reduced
17 to the solution of the ordinary differential equation for F (η)F
 
d2 F 2 2 dδ dF
+ 2η δ = 0. (17)
dη 2 dZ dη

2050013-9
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

1 We put the term in parentheses (δ 2 dzdδ

) a constant is equal to 3/2 because it is
2 authentic that the supposition of similarity will lose unless this magnitude is indis-
3 pensable to be independent of Z. Finally, we get at the solution of a system of
4 equations composed of two unknowns to be found F (η) and δ(Z).
d2 F dF
+ 3η 2 = 0, (18)
dη 2 dη
dδ 3
δ2 = . (19)
dZ 2
5 Starting from the boundary condition on θ(X, Z), we can calculate the derivatives
6 the boundary conditions of these formulas. We notice that θ(0, Z) = 0 which implies
7 θ(∞, Z) = 1 and F (0) = 0 which tends to F (∞) = 1. The residual clause at the
8 entrance of the tube gives us:
 
X
θ(X, 0) = F . (20)
δ(0)
9 By favoring δ(0) = 0, this condition has been dismantled in the one obtained
10 recently F (∞) = 1, when the variable X tends towards the infinite, that is to say
11 X → ∞. By joining the two boundary conditions on F (η) and δ(Z), we get to:
F (0) = 0, F (∞) = 1 and δ(0) = 0.
12 From the equation, we can write
d2 F
dη 2
dF
= −3η 2 . (21)
dη
13 By integrating the two terms of the obtained equation
d2 F 
dη 2
dF
= − 3η 2 dη. (22)
dη
14 We arrive at the following expression
   
dF 3
ln = −η 3 + k = ln Ce−η , (23)
dη
15 where k and C are constants of the integral
16 By analogy, the following is drawn
dF 3
= Ce−η . (24)
dη
17 Finally, the solution of the equation will take the following expression:
 η
3
F (η) = C e−η dη. (25)
0
18 The function F (η) checks the initial condition for η = 0, F (0) = 0, and also consid-
19 ers the boundary condition for η → ∞, F (∞) = 1; which implies
 ∞
3
F (∞) = 1 = C e−η dη. (26)
0

2050013-10
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

1 From where
1
C = ∞ . (27)
0
e−η3 dη
2 So
η 3
e−η dη
F (η) =  0∞ . (28)
0
e−η3 dη
3 The scaled boundary layer thickness δ(Z) is calculated by the integration, which
4 gives the following solution
 1/3
9
δ(Z) = Z . (29)
2
5 Finally, the solution of our differential equation takes the following form
 η −γ 3  η
e dγ 1 3
0
F (η) =  ∞ −γ 3 = e−γ dγ. (30)
0
e dγ Γ(4/3) 0

6 where Γ(x) is the Gamma function (Abramowitz and Stegun,39 ), a MATLAB code
7 was used to approximate the values of the integral and the function F (η) for each
8 abscissa η.

9 7. Numerical Resolution of the Problem using RK04 Method

10 The original ODE of our problem is defined as follows:
d2 F dF
+ 3η 2 = 0. (31)
dη 2 dη
11 dη (0) = 0
With η = 0, F (0) = 0 and dF
12 We will use the fourth-order Runge–Kutta method, so we have the following
13 system:
⎧ dF
⎪
⎪
⎨ dη = P
. (32)
⎪
⎪
⎩ dP = −3η 2 P
dη
14 With F (0) = 0 and P (0) = 1
15 If we have two ordinary differential equations of the first order, we have:
⎧
⎪ dy
⎪
⎨ dx = f (x, y, z)
. (33)
⎪ dz
⎪
⎩ = g(x, y, z)
dx
16 By applying the RK04 method on this system, we give:
⎧
⎨yi+1 = yi + (k0 + 2k1 + 2k2 + k3 )/6
, (34)
⎩zi+1 = zi + (l0 + 2l1 + 2l2 + l3 )/6

2050013-11
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

Start

Set initial values

for x , y and, z
input N, a, b

h=(b−a)/N
steps size

Do 1 i=1, N

Have Yes
we calculated for Print * x , y ,z
the last internal ?

No

Iteration steps :
first Stop
approximation
second
slop approximation
Fin
third
slop approximation

fourth
slop approximation

Calculate for next iteration

Take next interval

Fig. 3. Flowchart of the RK-4 method for resolving the second ODE’s systems.

2050013-12
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

Fig. 4. FORTRAN code of Runge–Kutta for set of first order differential equations.

1 where
⎧
⎪
⎪ k0 = hf (xi , yi , zi )
⎪
⎪  
⎪
⎪ h k0 l0
⎪
⎪
⎪
⎨ k1 = hf x i +
2
, y i +
2
, z i +
2
 
⎪
⎪ h k1 l1
⎪
⎪ k2 = hf xi + , yi + , zi +
⎪
⎪ 2 2 2
⎪
⎪
⎪
⎩
k3 = hf (xi + h, yi + k2 , zi + l2 )

2050013-13
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

1 and
⎧
⎪
⎪l = hg(xi , yi , zi )
⎪0
⎪  
⎪
⎪ h k0 l0
⎪
⎪
⎪
⎨l1 = hg xi + 2 , yi + 2 , zi + 2
 .
⎪
⎪ h k1 l1
⎪
⎪l 2 = hg x + , y + , z +
⎪
⎪
i
2
i
2
i
2
⎪
⎪
⎪
⎩
l3 = hg(xi + h, yi + k2 , zi + l2 )
2 We have adopted the Runge–Kutta algorithm for finding the solution of our system
3 of equations;

4 • The interval for the integration of the equations is chosen to perform our calcu-
5 lations: [a, b] if we take a = 0, and b = 3
6 • The number of iterations N = 30,
7 • The size of the iterations will be estimated as follows: h = (b−a)/N = 3/30 = 0.1

8 The flowchart for the above process is shown in Fig. 3.

9 The main program was drafted by FORTRAN, which will solve the problem of
10 Levèque whose procedure initiated to solve simultaneously two differential equa-
11 tions of the order by the method of Runge–Kutta RK04. This program relies on a
12 definition of two functions whose subroutine RK04 is called at each repetition of
13 the loop that intervenes in the calculations. The code edited in the machine that
14 was executed is illustrated in detail in Fig. 4.

15 8. Resultants and Discussions

16 8.1. Validation of the numerical results via the analytical solution
17 of the problem
18 The analytical solution that we have developed above is compared here with the
19 numerical results derived from the FORTRAN V.05 calculation code. The results
20 of the two methods are condensed in detail in Table 1.
21 Figure 5 shows a comparison between the resolution results of the equation
22 predicted by the analytical method and the numerical data derived from the FOR-
23 TRAN code, the two sets of results of which are plotted in the same figure. On
24 the basis of Fig. 5, it can be seen that the two curves are fairly identical, while
25 observing that the dimensionless temperature θ gradually and gradually increases
26 to the abscissa Z = 0.7, then loops and arches a little, by varying its path until
27 it reaches the position Z = 1.7 where it stabilizes at a constant value 0.79 along
28 the tube until the outlet of the fluid stream. In the same figure, the derivative
29 function dF
dη which is physically interpreted as the variation of the thermal trans-
30 fer coefficient (h) indicated in blue color which is a solution of our set equations
31 decreases exceptionally as it moves away from the inlet region and then reaches the
32 value zero on the abscissa Z = 1.6 until the exit of the flow. Figure 5 shows clearly

2050013-14
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

Table 1. Exact results and the numerical solution.

Numerical solution via fourth-order

Runge–Kutta (RK4) method using a
Exact analytical solution FORTRAN Code
η F (η) = θ(X, Z) X Y estimated Z estimated
0 0 0.0000 0.0000000 0.8929367
0.1 0.08927136 0.1000 0.0892714 0.8920442
0.2 0.17823109 0.2000 0.1782309 0.8858217
0.3 0.26608715 0.3000 0.2660866 0.8691499
0.4 0.35156264 0.4000 0.3515626 0.8375790
0.5 0.43300027 0.5000 0.4329998 0.7880137
0.6 0.50853023 0.6000 0.5085291 0.7194705
0.7 0.57631574 0.7000 0.5763146 0.6336619
0.8 0.63483615 0.8000 0.6348343 0.5351335
0.9 0.68314582 0.9000 0.6831438 0.4307465
1 0.72105634 1.0000 0.7210538 0.3284985
1.1 0.74916957 1.1000 0.7491656 0.2359384
1.2 0.76875346 1.2000 0.7687482 0.1586460
1.3 0.78149478 1.3000 0.7814872 0.0992794
1.4 0.78918921 1.4000 0.7891808 0.0574867
1.5 0.79347888 1.5000 0.7934697 0.0306282
1.6 0.79567283 1.6000 0.7956641 0.0149366
1.7 0.79669613 1.7000 0.7966892 0.0066372
1.8 0.79712921 1.8000 0.7971242 0.0026785
1.9 0.7972944 1.9000 0.7972914 0.0009805
2 0.79735155 2.0000 0.7973494 0.0003264
2.1 0.79736852 2.1000 0.7973676 0.0000995
2.2 0.79737298 2.2000 0.7973729 0.0000283
2.3 0.79737387 2.3000 0.7973742 0.0000077
2.4 0.79737387 2.4000 0.7973746 0.0000021
2.5 0.79737477 2.5000 0.7973747 0.0000006
2.6 0.79737477 2.6000 0.7973747 0.0000002
2.7 0.79737477 2.7000 0.7973747 0.0000001
2.8 0.79737477 2.8000 0.7973747 0.0000000
2.9 0.79737477 2.9000 0.7973747 0.0000000
3 0.79737477 3.0000 0.7973747 0.0000000

1 that the results of the analysis solution are very excellent convergence with those
2 of the numerical results performed by the Visual FORTRAN v5.0 calculation code
3 during which the use of the RK04 method obviously gives us a severely accurate
4 assessment.
5 Figure 6 shows the variation in the thickness of the thermal boundary layer as
6 a function of the longitudinal coordinate where the latter increases slowly from the
7 zero position towards the direction of flow of the fluid as it penetrates the pipe
8 through its center and arrogates its total space. At the inlet of the tube and its
9 wall, the shear stress is greater during which the thickness of the boundary layer is
10 very short and slowly decreases to the fully developed value. In fact, the collapse of
11 the pressure is increased in the inlet zone of the tube under the effect which may

2050013-15
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

1,0

0,9

0,8

0,7
Dimensionless temperature

0,6

0,5

0,4

0,3

0,2

0,1

0,0
0,0 0,5 1,0 1,5 2,0 2,5 3,0
Similarity variable

Fig. 5. Comparison of exact and fourth-order Runge–Kutta (RK4) numerical solutions.

2,5
Thermal boundary layer thickness (Z)

2,0

1,5

1,0

0,5

0,0
0,0 0,5 1,0 1,5 2,0 2,5 3,0
Dimensionless axial coordinate Z

Fig. 6. Thermal boundary layer thickness distribution by analytical method.

1 cause the phenomenon of friction over the whole of the tube. This elevation can
2 be negligible for long and important tubes in short lengths. A thin layer can be
3 observed on the wall at which the velocity of flow is less than the wall. By going
4 from front to back, the thickness of the thermal boundary layer lengthens along the
5 channel.

2050013-16
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

1 8.2. The heat transfer coefficient

2 Depending on the axial position, we try to understand the heat fluxes of the wall
3 when a fluid flow is involved, we can calculate it directly using the following formula:
∂T
qω (z) = k (R, z). (35)
∂r
4 By usual notation, the convective transfer coefficient h(z) is known from the fol-
5 lowing expression

qω (z) = h(z)(Tω − Tb ), (36)

6 where Tb is the bulk or cup-mixing average temperature.

7 The average bulk temperature is mathematically defined as:
R
2πrV (r)T (r, z)dr
Tb = 0  R , (37)
0 2πrV (r)dr

8 where V (r) = v0 (1 − r2 /R2 ) is the velocity field. The temperature gradient at the
9 wall is commonly subordinate to the heat exchange coefficient, we can estimate it
10 as follows:
k ∂T
∂r (R, z)
h(z) = . (38)
(Tω − Tb )
11 The Nusselt number is defined as a dimensionless heat exchange coefficient.
∂θ
2hR (1, Z)
Nu(Z) = = −2 ∂Y , (39)
k θb (Z)
12 where θb θb is the without dimensional bulk temperature along pipe
13 We approximate the mean temperature Tb by the temperature of the liquid
14 entering the pipe T0 and this, in the region of entry where the boundary layer is
15 thin.
16 As a result, and through the thermal input region, the heat exchange coefficient
17 (h) is expressed as
∂T
qω = k (R, z) = h(Tω − T0 ). (40)
∂r
18 We know the Nusselt number Nu = 2hR/k, and by introducing the dimensionless
19 variables, we obtain the following:
∂θ 2 ∂F
Nu(Z) = 2 (0, Z) = (0). (41)
∂X δ(Z) ∂η

20 By substituting δ(Z) and ∂F ∂η (0), the final formula of the Nusselt number as a
21 function of the variables Pe, Z, and R which we evaluated in the input region can

2050013-17
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

14

R=1
12 R=5
R=6
10 R=7
Nusselt Number Nu

R=8
R=9
8 R=10
R=20

0 1 2 3 4 5
Dimensionless axial coordinate Z

Fig. 7. Nusselt number as a function of axial position for different tube radius.

1 be expressed as follows:
 1/3
R
Nu(Z) ≈ 1.357Pe1/3 . (42)
z
2 By comparing with the exact solution, we can now appreciate that this calculation
3 is a better evaluation in the range
Pe  z Pe
≤ ≤ . (43)
2500 R 50
4 Figure 7 shows the variation obtained in the input region of the Nusselt number as
5 a function of the axial distance Z obtained in the thermal input region for various
6 radius of the pipe. We can observe that the number of Nusselt, Nu (Z), rises as
7 a function of the increase of the radius of the tube and that this influence is very
8 noticeable enlarged at the entrance. When Z is greater than a certain distance,
9 all the bundles of curves have become intensified and they stabilize horizontally
10 flat, this explains why the fully developed boundary layer is reached. Indeed, the
11 boundary layer triggers to increase when the fluid enters the tube in the walls of the
12 walls having a temperature distinct from that of the fluid. The developed thermal
13 condition is achieved after the flow passes a certain position.
14 Figure 8 shows the Nusselt number as a function of the longitudinal coordinates
15 for different values of the Peclet number. It is observed that the increase in the
16 number of Peclet leads to an increase in the number of Nusselt. As can be seen,
17 the Péclet number has a much more pronounced effect on the Nusselt values for
18 positions near the tube entrance. However, the curve exhibits the same overall
19 behavior — larger Nu at small Z and more or less constant value of large Z. In the
20 tube entry region, where the boundary layer has expanded, we can see the reduction
21 of the Nusselt number where it stabilizes in the fully developed thermal zone to a

2050013-18
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

18

16

14
Nusselt number Nu

12

10

0 1 2 3 4 5
Dimensionless axial coordinate Z

Fig. 8. Nusselt number as a function of axial position with various Peclet numbers.

1 constant value does not depend on the Reynolds number and the heat flux. Hence,
2 the thermal coefficient (h) appeared unlimited at the birth of the thermal boundary
3 layer, and then gradually decreases to a stable value when the flow is fully developed
4 at the origin. The numerical results clearly illustrate that the value of the Nusselt
5 number increases and then decreases sharply over the entire longitudinal position
6 of the tube.

7 8.3. Validation against published data

8 We found deeply in the specialized literature, the work of Shah and London,40 where
9 they presented the numerical solutions of the well-known Nusselt–Graetz problem
10 of heat transfer to an incompressible fluid with constant properties flowing through
11 a circular duct having a uniform wall temperature and a fully developed laminar
12 velocity profile. The asymptotes for the local Nusselt number correlation in the
13 thermal entrance region of a circular tube.
14 Figure 9 plots the present analytical results for Nusselt number versus Z at
15 the thermal entrance region of circular conduit and comparison to existing liter-
16 ature correlations established by Shah and London.40 These results are presented
17 in this figure, where the Nusselt number is plotted against the axial coordinate Z.
18 Examination of this one reveals the expected trend whereby higher Nusselt numbers
19 correspond to very small values of the axial position Z at the entrance of the tube.
20 It has been found that for the unblocked tube, the local Nusselt number decreases
21 gradually as Z increases. From Fig. 9, it can be seen that the results from both the
22 analytical and the previous work are very similar which implies that the present
23 work with the analytical method gives us a good correlation with the investiga-
24 tions found in the literature. Given the sources of variability of the Nusselt number
25 calculated in the input thermal region, this result is considered satisfactory.

2050013-19
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

3,0

2,5

2,0
Nusselt number Nu

Present work
1,5 Previous work Shah and London (1978)

1,0

0,5

0,0
0 1 2 3 4 5
Dimensionless axial coordinate Z

Fig. 9. Comparison between the present work and the previous study.40

1 9. Conclusion
2 This paper presented an analytical and numerical solution to the Levèque approx-
3 imation problem in order to predict the evolution of the thickness of the boundary
4 layer as well as the temperature of the fluid at thermal entrance fully developed
5 region through a circular tube with boundary condition at the axial coordinate
6 origin. The exact solution methodology was based on the similarity variable and
7 the generalized integral transform technique while the numerical approach is based
8 on the integration technique of two differential equations with the Runge–Kutta
9 method of order 4 (RK4) programmed in Visual FORTRAN v5.0. The solution
10 method was verified to lead to converging values which are in accordance with
11 physically expected results. The numerical results obtained are in excellent agree-
12 ment with the published results found in literature in limiting sense of the present
13 work. After demonstrating the convergence of the solution, the Nusselt number
14 distribution of different Péclet values was analyzed, and the results are also in
15 accordance with expected literature values. As final comments one should mention
16 that the same solution procedure can be used for any dynamically developed veloc-
17 ity profile, as it occurs in many other occasions. Also, the methodology can be easily
18 extended to other configurations such as another channel geometries, different wall
19 heating conditions, and vicious and other flow heating effects.

20 References
21 1. Azimi S. S., Kalbasi M., Namazi M. H., Effect of nanoparticle diameter on the forced
22 convective heat transfer of nanofluid (water + Al2 O3 ) in the fully developed laminar
23 region, Int. J. Model. Simul. Sci. Comput. 5(3):1450008, 2014.
24 2. Ibrahim W., Shanker B., MHD boundary-layer flow and heat transfer over permeable
25 plate with convective surface boundary condition, Int. J. Model. Simul. Sci. Comput.
26 5(1):1350021, 2014.

2050013-20
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

Analytical solution and numerical simulation of the thermal entrance region problem

1 3. Deepu, M., Dhrishit, M. P., Shyji S., Numerical simulation of high speed reacting
2 shear layers using AUSM+ + − up scheme-based unstructured finite volume method
3 solver, Int. J. Model. Simul. Sci. Comput. 8(3):1750020, 2017.
4 4. Zhu, Z. X., Yin Y., Movania M. M., A novel parallel algorithm for computing the
5 mooring line based on lumped-mass method, Int. J. Model. Simul. Sci. Comput.
6 8(1):1750004, 2017.
7 5. Krivovichev G. V., On the stability of lattice boltzmann equations for one-dimensional
8 diffusion equation, Int. J. Model. Simul. Sci. Comput. 8(1):1750013, 2017.
9 6. Li, M., Huang C., ADI Galerkin FEMs for the 2D nonlinear time-space fractional
10 diffusion-wave equation, Int. J. Model. Simul. Sci. Comput. 8(3):1750025, 2017.
11 7. Yaghobi Moghaddam M., Shafaei Tonkaboni S. Z., Noaparast M., Doulati Ardejani F.,
12 A mathematical model to simulate Heap (bio)-leaching process: An exact conceptual
13 model, Homotopy theory and comparative insights with conventional methods, Int.
14 J. Model. Simul. Sci. Comput. 8(1):1750018, 2017.
15 8. Chauhan V., Srivastava P. K., Trio-Geometric mean-based three-stage Runge–Kutta
16 algorithm to solve initial value problem arising in autonomous systems, Int. J. Model.
17 Simul. Sci. Comput. 9(4):1850026, 2018.
18 9. Bhaumik M., Maity K., Experimental investigation and finite element simulation of
19 AISI 304 during electro discharge machining, Int. J. Model. Simul. Sci. Comput.
20 9(4):1850022, 2018.
21 10. Padilla J. R., Liceaga D. O., Chebyshev multidomain pseudospectral method to solve
22 cardiac wave equations with rotational anisotropy, Int. J. Model. Simul. Sci. Comput.
23 9(4):1850025, 2018.
24 11. Jha N., Venu Gopal V., Singh B., Geometric grid network and third-order compact
25 scheme for solving nonlinear variable coefficients 3D elliptic PDEs, Int. J. Model.
26 Simul. Sci. Comput. 9(6):1850053, 2018.
27 12. He H., Liang K., Yin B., A numerical method for two-dimensional nonlinear modified
28 time-fractional fourth-order diffusion equation, Int. J. Model. Simul. Sci. Comput.
29 10(1):1941005, 2019.
30 13. Khan N. A., Hameed T., Ahmed S., Homotopy perturbation aided optimization pro-
31 cedure with applications to oscillatory fractional order nonlinear dynamical systems,
32 Int. J. Model. Simul. Sci. Comput. 10(4):1950026, 2019.
33 14. Choudhary S., Gejji V. D., Solving systems of multi-term fractional PDEs: Invariant
34 subspace approach, Int. J. Model. Simul. Sci. Comput. 10(1):1941010, 2019.
35 15. Hamad M. A. A., Ferdows M., Similarity solutions to viscous flow and heat trans-
36 fer of nanofluid over nonlinearly stretching sheet, Appl. Math. Mech. 33(7):923–930,
37 2012.
38 16. Wei, D. M., Al-ashhab S., Similarity solutions for non-Newtonian power-law fluid flow,
39 Appl. Math. Mech. 35(9):1155–1166, 2014.
40 17. Trı̂mbijas R., Grosan T., Pop I., Mixed convection boundary layer flow past vertical
41 flat plate in nanofluid: Case of prescribed wall heat flux, Appl. Math. Mech. 36(8):
42 1091–1104, 2015.
43 18. Ahmed S. E., Modeling natural convection boundary layer flow of micropolar
44 nanofluid over vertical permeable cone with variable wall temperature, Appl. Math.
45 Mech. 38(8):1171–1180, 2017.
46 19. Shen L., Lu C., Mechanism of three-dimensional boundary-layer receptivity, Appl.
47 Math. Mech. 38(9):1213–1224, 2017.
48 20. Mahanthesh B., Gireesha B. J., Shehzad S. A., Abbasi F. M., Gorla R. S. R., Nonlinear
49 three-dimensional stretched flow of an Oldroyd-B fluid with convective condition,
50 thermal radiation, and mixed convection, Appl. Math. Mech. 38(7):969–980, 2017.

2050013-21
 Page Proof

March 17, 2020 21:10 WSPC/262-IJMSSC/S1793-9623 2050013

A. Belhocine & O. I. Abdullah

1 21. Eldesoky I. M., Abdelsalam S. I., Abumandour R. M., Kamel M. H., Vafai K., Inter-
2 action between compressibility and particulate suspension on peristaltically driven
3 flow in planar channel, Appl. Math. Mech. 38(1):137–154, 2017.
4 22. Baehr H., Stephan K., Heat Transfer, Springer-Verlag, 1998.
5 23. Stephan K., Warmeubergang und Druckabfall bei Nicht Ausgebildeter Laminar Stro-
6 mung in Rohren und in Ebenen Spalten, Chem. Ingenieur Tech. 31(12):773–778, 1959.
7 24. Asako Y., Nakamura H., Faghri, M., Developing laminar flow and heat transfer in the
8 entrance region of regular polygonal ducts, Int. J. Heat Mass Transf. 31(12):2590–
9 2593, 1988.
10 25. Shah R. K., London A. L., Laminar Flow Forced Convection in Ducts, Academic
11 Press, New York, NY, 1978.
12 26. Kakac S., Shah R. K., Aung W., Handbook of Single Phase Convective Heat Transfer,
13 Wiley, New York, 1987.
14 27. Ebadian M. A., Dong Z. F., Forced convection internal flows in ducts, in Handbook of
15 Heat Transfer, 3rd edin., McGraw-Hill, New York, pp. 5.1–5.137, 1998.
16 28. Kakac S., Yener, Y., Laminar forced convection in the combined entrance region of
17 ducts, in Kakac S., Shah R. K., Bergles A. E. (eds.), Low Reynolds Number Heat
18 Exchangers, Hemisphere Publishing, Washington, pp. 165–204, 1983.
19 29. Hausen H., Darstellung des Wärmeübergangs in Rohren durch verallgemeinerte
20 Potenzbezie-hungen, VDI-Zeitung, Suppl. “Verfahrenstechnik” 4:91–98, 1943.
21 30. Churchill S. W., Ozoe H., Correlations for laminar forced convection with uniform
22 heating in flow over a plate and in developing and fully developed flow in a tube,
23 ASME J. Heat Transfer 95:78–84, 1973.
24 31. Churchill S. W., Ozoe H., Correlations for laminar forced convection in flow over an
25 isothermal flat plate and in developing and fully developed flow in an isothermal tube,
26 ASME J. Heat Transfer 95:416–419, 1973.
27 32. Yilmaz T., Cihan E., General equation for heat transfer for laminar flow in ducts of
28 arbitrary cross-sections, Int. J. Heat Mass Transfer 36(13):3265–3270, 1993.
29 33. Yilmaz T., Cihan E., An equation for laminar flow heat transfer for constant heat
30 flux boundary condition in ducts of arbitrary cross-sectional area, J. Heat Transfer
31 117(3):765–766, 1995.
32 34. Belhocine A., Wan Omar W. Z., Numerical study of heat convective mass transfer in
33 a fully developed laminar flow with constant wall temperature, Case Stud. Thermal
34 Eng. 6:116–127, 2016.
35 35. Belhocine A., Numerical study of heat transfer in fully developed laminar flow inside
AQ: Please 36 a circular tube, Int. J. Adv. Manuf. Technol. 85(9):2681–2692, 2016.
37 36. Belhocine A., Wan Omar W. Z., An analytical method for solving exact solutions of
provide Vol. No.
38 the convective heat transfer in fully developed laminar flow through a circular tube,
39 Heat Transfer—Asian Res. 1–12, 2017.
40 37. Bird R. B., Stewart W. E., Lightfoot E. N., Transport Phenomena, John Wiley and
41 Sons, New York, 1960.
42 38. Lévêque M. A.. Les lois de la transmission de chaleur par convection, Annales des
43 Mines, Memoires, Series 12, 13, 201–299, 305–362, 381–415 (1928) {as cited by J.
44 Newman, Trans. ASME J. Heat Transfer 91: 177, 1969.
45 39. Abramowitz M., Stegun I., Handbook of Mathematical Functions, Dover, New York,
46 1965.
47 40. Shah R. K., London A. L., Laminar Flow Forced Convection in Ducts, Academic
48 Press, New York, NY, 1978.

2050013-22