lnt. J. Heat Mass Tmns&. Vol. 23, pp. 1399-1402 ~17-93i0/80~1~1-13~ $02.

00/O
Q Pergamon Pm Ltd. 1980. Printed in Great Britain

SHORTER COMMUNICATION

ON THE SOLUTION OF GRAETZ TYPE PROBLEMS

WITH AXIAL CONDUCTION

Y. BAYAZIT&LU* and M. N. (.)ZISIK?

(Received 10 July 1979 and in rev~sed~~~m 12 Feb~ary 1980)

NOMENCLATURE upstream and downstream regions of the origin [ 1% 171; a

dime~sionl~s heat generation ; combination of purely numerical scheme and an asymptotic
heat transfer coefficient, i = 1, 2; expansion technique was tried [IS].
= h&/k, Biot number, i = 1, 2; A scrutiny of all these analytical approaches discussed
a reference length ; above reveals that their application is rather involved and
thermal conductivity; their extension to more general situations is a complicated
= (/80&/a), PC&t number ; matter.
radial coordinate; We now present a straightforward, simple, yet a very
tube radius; general, finite integral transform technique for the solution of
temperature; Graetz type channel flow heat transfer problems including
infet temperature; the axial conduction, convective boundary conditions, and
mean velocity ; internal energy sources. The analysis is applicable to both
axial and normal coordinates, respectively; laminar and turbulent flow inside conduits. As our objective
molecular diffusivity of heat; in this work is to demonstrate the application of the method,
coefficientin u(t)) = fi8,A(n); we examine only the situation in which the axial heat
conduction is considered in the downstream direction. The
problems in which the axial conduction is considered in both
upstream and downstream directions can also be handled
with this approach.
turbulent diffusivity of heat;
ANALYSIS
= G, dimensionless temperature;
We consider a channel flow heat transfer problem with
axial conduction given in the dimensionless form as
= i, dimensionl~s normal coordinate ;

INTRODUCTION

NUMEROUS extensions of the original Graetz problem ap-

peared in the literature since its formulation by Graetz in
1885. However, the analysis becomes rather involved ma-
thematic~ly if the axial heat conduction in the fluid Cl-181 is
to be considered. Therefore, various elaborate mathemati~l
approaches have been tried to solve the problems with axial ae
ffi,e=o t>O,
heat conduction in the fluid. For example, the temperature
field was arbitrarily expanded in terms of the Fourier sine
att atrl=b

series [l] ; the nonorthogonal eigenfunctions were expanded e = F(q) atT = 0, y~iI tf I q2, (Id)
in terms of the eigenfunctions of an auxilliary orthogonal e-e,, as{-+ x:, (le)
system [2,3]. An expansion similar to that used in references
[2,3] was applied [4] to solve the problems with axial where the coefhcients H, and W, do not vanish simul-
conduction in the upstream and downstream regions of the taneously and the boundary condition (le) is as a consequence
origin. The Gramm-Schmidt orthonormahzation procedure of the requirement that for large 5 the temperature distri-
was used to construct orthogonal functions from the non- bution reduces to that of fully developed temperature profile,
orthogonal eigenfunctions [ 11,121; the same technique was B,,. Various other quantities are defined as
also applied to solve related problems involving internal
energy sources [13, 141. A two-sided Laplace transform yf_l)=l-bEHO),
j=xory; &. _,
technique was used to represent the temperature field in the

*Department of Mechanical Engineering, Rice University,

Houston, TX 77001, U.S.A.
tDepartment of Mechanical and Aerospace Engineering, Hi=y,i=lor2; lj=k; G=s;
North Carolina State University, Raleigh, NC 27650, U.S.A. 0

1399
1400 Shorter Communications

0 for slab,
S=
1 for cylinder

To solve the above problems we consider the eigenvalue where the summation index M of equation (4b) is replaced by
problem appropriate for the case in which the axial con- n in order to distinguish it from the eventual inversion-
duction is neglected formula summation-index m. The term J, is a known
quantity because G(n) is a specified function.
in n1 < n < VZ, When equations (6a), (6b) and (5d) are introducted into
equation (5a) we find

- % + H,IL(tl) = 0 atv = rll, (2b)

d&,,(i;)
+(I),.G)=~, forGO. (7a)
4

The boundary conditions for this equation are obtained by

taking the integral transform of the boundary conditions (Id)
The eigenfunctions satisfy the following orthogonality and (le); that is
condition
e,(C) = F, for < = 0, (7b)
0
“‘VA(n) M,,, V) +(L rl) dn = RI(r) + Q,D as[+ ~fI, (7c)
N(%,)
where the inner product of two functions is defined as
whete
rn>
((i/,.G)= ) s m
No_,) = j-T@A(+(&,,, n)I dn. (3b) * ‘I’ ’ * (l)G(n)d’L (8a)

(in,>$.) = i”’ v” G,(vMn(v)de. WI

The integral transform pair with respect to the n variable is . ‘II
defined as and the bar is used to denote the integral transform, i.e.
Transform :
J
w>4d = ‘.rl’“AW)W,,, 7’) Ott,4) W,
J
I,/
(W
‘II
Other abbreviations used include g”,(t) = @(I.,, t), l/l,,,(r) =
Inversion : t@,,n), N, = N(i.,) and m = I, 2, 3,
Thus, the original problem (1) is transformed to the
(4b) solution of an infinite set of coupled, second degree ordinary
differential equations for the transforms 6,,,(t), m = 1,2,3,
We take integral transform ofsystem (1) by the application of Once 8,,,(t) are determined from the solution of this system,
the transform (4a); namely, we operate on both sides of the temperature distribution 0(& 9) is immediately obtainable
equation (la) by the operator from the inversion formula (4b). However, it is not practical
to solve an infinite set of coupled differential equations; but,
simple approximations yielding highly accurate results are
obtainable by considering only a finite number equations in
the system as now described
and obtain
Lowest order analysis
d&, A,,)
J, +&J, +J3=- (54 The lowest order analysis is obtainable from this system by
d5 ’ taking only one term in the series in equation (7a) by setting m
where = n. We find

d6,SO
+(IL,,G)=- for c > 0, @a)
dl ’
J 91
The term J, is evaluated by integrating it by parts twice and
utilizing the eigenvalue problem (2) and the boundary 8,=P, fort=O, Pb)
conditions (lb) and (1~). 8, -+ 8,, fort + X, (9c)

J, = - 1; 8(<,i.,). (6a) for m = 1, 2, 3,

To evaluate the term J,, the function e(n, 5) is replaced by its The system (9) provides a set of uncoupled, second-degree
equivalent inversion formula (4b) ; We obtain ordinary differential equations for the transforms 8,,,, m = 1,
2, 3, Once the transforms B are determined from the
1
J,= f- ~“Wm~~)W,~~)d~ d”, solution of this system, the temperature distribution S(& n) is
n= LWA I immediately obtained by the inversion formula (4b).
 Shorter Communications

The lowest order equations are also obtainable by other

choices of the terms in the series ; however, the choice given by
equations (9) yields better results.
The solution of equations (12) is
Higher order analysis
Higher order analysis is obtainable by considering a 8,(t) = Lembmc, m = 1,2,3,. . (13)
simultaneous solution of the first two or more of the
where
differential equations; for the remainder of the functions %,,
the system is reduced to a set of uncoupled second-degree 1 + J(I + 4A, 1.3
differential equations as discussed above for the case of lowest b,= -
2A, .
order analysis.
Then the tem~~ture distribution %(n,0 is determined as
LAMINAR FLOW INSIDE A CIRCULAR TUBE

To illustrate the application of the foregoing method of

analysis and the accuracy of the lowest order solution, we
consider laminar flow inside a circular tube with uniform wall Once @,5) is known, the average temperature e.,(t) is
temperature and a uniform inlet temperature at the origin of calculated from
the axial coordinate. We allow for axial conduction in the
downstream region of the origin, The mathematical for- %.,(<) = 4 ’ ~(1 - a’)@s,t)ds, (15)
mulation of the problem, in the dimensionless form, is given as s0
Id d% 1 s0 and the Nusselt number from
II& ( aa; ! +Pets
Nu(‘$)= - &.!_-% . (16)
@A3 dq *=1
= (I - a’), ae(“r) inO<?< , r>O > (lOa)
To illustrate the accuracy of the lowest order solution, we
present in Table 1 a comparison of%,&) and Nu(t) obtained
-=i from the present analysis with thosecalculated by Singh [3]*.
ae 0 ata=O, c>O, (lob) Clearly, the agreement is sufficiently close.
a?
%(s,e) = 0 at q = 1, 5 > 0, (lOc) Table 1.A comparison oflowest order solution for %,, and Nu
%(rl,{)= I at5=0, Ostl11, (lOd) with Singh’s [3] results for Pe = 50t

e-+0 at{-+ X, (IOe) Lowest order solution Singh’s [3]

where
5 e I?” NU e 0” NU

0.005 0.896 1.905 0.91 8.079

0.025 0.718 4.697 0.73 4.759
The eigenvalue problem (2) takes the form 0.05 0.580 4.027 0.59 4.034
0.10 0.396 3.716 0.40 3.715
0.25 0.132 3.657 0.134 3.66

t Pe = 50 in the present definition is equivalent to Pe

dw) - 0 atq=O, UW = 100 in Singh’s definition.
drl
Acknowledgement--This work was supported in part through
a) =0 at? = 1, (llc) the National Science Foundation Grant Eng. 77-12949.
which is the eigenvalue problem associated with the original
Graetz problem. REFERENCES
The lowest order analysis for the transform 8,,,(t) of
temperature is obtained from equations (9) as
1. S. C. R. Dennis and G. Poots, A solution of the heat
_ A, d%,(5)+ d&At) 14B&) = 0,
dr + 4 > 0, (12a)
transfer equation for laminar flow between parallel
dtz plates, Q. AppJ. Math. 14, 231-236 (1956).
2. S. N. Singh, The determination of eig~-functions of a
&ce)= J* for f = 0, (l2b) certain St~~Liou~IIe equation and its application to
asT+ T, (l2c) problems of heat transfer, Appt. Sci. Res. A7, 237-250
%,(i;) + 0
(1957-58).
for m = 1, 2, 3 3. S. N. Singh, Heat transfer by laminar flow in a cylindrical
tube, AppJ. Sci. Res. A7, 325-340 (1957-58).
where
4. H. C. Agrawal, Heat transfer in Iaminar flow between
parallel plates at small P6cIbt numbers, Appl. Sci. Res. A9,
A _ (ti*, tim),
m
N, Pe’ ’
T,= ~(1- aZM,,Ws l 177-189 (1960).
c0 5. B. S. Petukhov and F. F. Tsvetkov, Inz. - Fiz. Zh. 4,
IO-17 (1961). Transl. No. FTD TT61-321, Jan. 29
(1962).
6. D. K. Hennecke, Heat transfer by Hagen-Poiseuille Row
in the thermal development region with axial conduction,
* it was pointed out in [19] that an error was involved in ~~rrne-Stu~e~ragu~g Bd. 1, 177-184 (1968).
the original analysis by Singh [Z]; but that error does not 7. F. H. Verhoff and D. P. Fisher, A numerical solution of
invalidate his later results since he never actualiy used it in tbe the Graetz problem with axial conduction, J. Heat
subsequent developments. Transfer 95, 132-134 (1973).
1402 Shorter Communications

8. M. L. Michelsen and J. Villadsen, The Graetz problem transfer, Can. J. Chem. Engng 51, 562.-572 (1973).
with axial conduction, Int. J. Heat Mass Transfer 17, 14. C. E. Smith, M. Faghri and J. R. Welty, On the
1391-1402 (1974). determination of temperature distribution in laminar
9. J. P. Sorensen and W. E. Stewart, Computation of forced pipe flow with a step change in wall heat flux, J. Heut
convection in slow flow through ducts and packed beds - Transfer 97C, 137-139 (1975).
extensions of the Graetz problem, Chem. Engng Sci. 29, 15. A. S. Jones, Extensions to the solutions of the Graetz
811-817 (1974). problem, Int. J. Heat Mass Transfer 14,619.-623 (1971~.
10. A. Campo and J. C. Auguste, axial conduction in laminar 16. A. S. Jones, Laminar forced convection at low P&let
pipe flows with nonlinear wall heat fluxes, Int. J. Heat number, Bull. Austral. Math. Sot. 6, 83- 105 (1972).
-Galss Transfer 21, 1597-1607 (1978). 17. A. S. Jones, Two-dimensional adiabatic forced con-
11. Chia-Jung I
Hsu. Theoretical solutions for low-P&l&- vection at low-P&l& number, Appl. Scient. Res. 25,
number thermalentry-region heat transfer in laminar 337-348 (1972).
flow through concentric annuli, Inc. J. Heat Mass 18. C. A. Deavours, An exact solution for temperature
Transfer 13, 1907-1924 (1970). distribution in parallel plate Poiseuille Row, -J. Heur
12. Chia-Jung Hsu, An exact analysis of low-P&l&-number Transfer %C. 489-495 (1974).
thermal-entry-region heat transfer in transversely non- 19. J. C. iirkle and V. G. Sig‘illito:Calculation of coefficients
uniform velocity fields, A.1.Ch.E. Jl17,3,732-740 (1971). of certain eigenfunction expansions, Appl. Sci. Res. 26,
13. E. J. Davies, Exact solutions for a class of heat and mass 105-107 (1972).