Available online at www.sciencedirect.

com

International Communications in Heat and Mass Transfer 35 (2008) 535 – 544

www.elsevier.com/locate/ichmt

Fluid flow in microtubes with axial conduction including

rarefaction and viscous dissipation ☆
Barbaros Cetin a,⁎, Almila Guvenc Yazicioglu b , Sadık Kakac c
a
Department of Mechanical Engineering, Vanderbilt University, 37235-1592 Nashville, TN, USA
b
Department of Mechanical Engineering, Middle East Technical University, 06531 Ankara, Turkey
c
Department of Mechanical Engineering, TOBB Economics and Technology University, 06560 Ankara Turkey
Available online 20 February 2008

Abstract

Graetz problem inside the microtube is revisited considering rarefaction effect, viscous dissipation term and axial conduction in the fluid for
uniform wall temperature boundary condition in the slip flow regime. The flow is assumed to be hydrodynamically fully developed, thermally
developing, and the velocity profile is solved analytically. The temperature field is determined by the numerical solution of the energy equation.
The rarefaction effect is imposed to the problem via velocity-slip and temperature jump boundary conditions. The local and fully developed Nu
numbers are obtained in terms of dimensionless parameters; Pe, Kn, Br, κ. Fully developed Nu numbers and the thermal entrance length are
found to increase by the presence of the finite axial conduction.
© 2008 Elsevier Ltd. All rights reserved.

Keywords: Microtube; Slip flow; Knudsen number; Axial conduction; Viscous dissipation

1. Introduction to design, optimize, fabricate and utilize improved MEMS

devices.
The rapid development of microfabrication enabled the The special characteristics of fluid physics at microscale are
usage of devices having dimensions of micron, such as; micro the results of the microscale channel size, surface roughness
heat sinks, micro biochips, micro reactors (for modification effect, the interfacial electrokinetic effects near the solid–fluid
and separation of biological cells), micro-motors, micro interface (common for liquid flows), and rarefaction effect
valves and micro-fuel cells which are all called microfluidic (common for gaseous flows). As Knudsen number, which is
devices. These devices have found their applications in the ratio of the average distance traveled by the molecules
micro-electronics, micro scale sensing and measurement, without colliding with each other, which is known as the mean
spacecraft thermal control, micro-electro-mechanical-systems free path (λ), to the characteristic length of the flow (L)
(MEMS), medical, pharmaceutical, defense, and environ- increases, the continuum approach fails to be valid, and the
mental monitoring, as well as scientific investigations. Most fluid modeling moves from continuum model to molecular
of these applications include fluid flow in microchannels model.
in the orders of nanoliters. Therefore, the fundamental
k
understanding of the fluid physics at microscale is crucial Kn ¼ ð1Þ
L
The flow is considered as continuum flow for small values
of Kn (b0.01), and the well known Navier–Stokes equations
☆
Communicated by W.J. Minkowycz. together with the no-slip and no temperature jump boundary
⁎ Corresponding author. conditions are applicable for the flow field. Solutions of the
E-mail address: barbaros.cetin@Vanderbilt.Edu (B. Cetin). Navier–Stokes have been a wide area of research over the last
0735-1933/$ - see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.icheatmasstransfer.2008.01.003
536 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544

Nomenclature Boltzmann equations for dilute gases. Usually, the analytical

solutions are inaccessible due to the mathematical difficulties
x axial coordinate, m associated with the Boltzmann equation. Small-perturbation
Br Brinkman number, μum 2
/k(Ti − Tw) or μum
2
/q″D approach may lead to some approximate solutions near
cp constant pressure specific heat, J/kgK continuum or near collisionless regimes. Unfortunately, this
h convective heat transfer coefficient, W/m2K procedure has limited applications to problems encountered in
ū dimensionless velocity practice. For large disturbances, numerical methods are the
T fluid temperature, K only choice to deal with the problems. Molecular Dynamics
ℜ gas constant, J/kgK (MD) is the method to deal with dense gas or liquid flows, and
Kn Knudsen number, λ/L DSMC is the method to deal with the dilute gas flows [2]. The
Nu Nusselt number, hDh/k details of the fluid modeling at microscale can be found
FM momentum accommodation coefficient elsewhere [3].
Pe Peclet number, Re × Pr For values 0.01 b Kn b 0.1, which is the case for gas flows in
Pr Prandtl number, ν/α conduits with micron scale dimensions or gas flows in low
r radial coordinate, m pressures, the flow is in the slip flow regime. The Navier–
Re Reynolds number, umDh/ν Stokes equations are valid together with the modified boundary
FT thermal accommodation coefficient conditions at the wall due to the lack of thermodynamic
k thermal conductivity, W/mK equilibrium between the fluid particles and the wall. The fluid
R tube radius, m particles have a tangential velocity relative to the wall, and a
u velocity in axial direction, m/s finite temperature difference at the solid surface. First-order slip
flow and temperature jump boundary conditions are applicable
Greek symbols for slip flow regime [4].
ξ dimensionless axial coordinate The first-order slip velocity for a cylindrical geometry can be
η dimensionless radial coordinate expressed as [5],
θ dimensionless temperature


μ dynamic viscosity, kg/ms 2  FM du
ν kinematic viscosity, m2/s us ¼  k ð2Þ
FM dr r¼R
λ mean free path, m
κ parameter defined in Eq. (11)
γ specific heat ratio where λ is the mean free path, ℜ is the gas constant and FM is
α thermal diffusivity, m2/s the momentum accommodation factor which is close to unity
ε transformed axial coordinate for most of the gas solid couples used in engineering
applications [6], and taken so herein.
Subscripts Similarly, temperature jump for a cylindrical geometry can
s fluid properties at the surface be expressed as [5],
i inlet values


x local values 2  FT 2g k AT
m mean values Ts  Tw ¼  ð3Þ
FT g þ 1 Pr Ar r¼R
w wall values
where γ is the specific heat ratio, Pr is the Prandtl
number of the fluid and FT is the thermal accommodation
factor which varies between 0 and 1.0. One additional
century. As a result, analytical and closed form solutions are term to the temperature jump boundary condition, called
available for some model problems through simplifications by thermal creep, is neglected since it is proportional to Kn2 ,
some assumptions, depending on the nature of the problem; and very small for the slip flow regime, which is the case
and numerical solutions are available for practical problems in this study [6]. By imposing these two effects into the
via Computational Fluid Dynamics (CFD). The flow is boundary conditions as slip velocity and temperature jump,
considered as free-molecular flow for large values of Kn the Navier–Stokes equations become applicable for the slip
(N10), and the tool for dealing with this type flow is kinetic flow regime.
theory of gases. The governing equation is the collision- The analytical solution to the hydrodynamically fully
less Boltzmann equation. Analytical results and numerical developed, thermally developing flow of an incompressible
solutions via test-particle Monte Carlo (for steady problems), fluid inside a tube, which is known as Graetz problem, was
and Direct Simulation of Monte Carlo (DSMC) (for unsteady obtained by Graetz [7] and Nusselt [8] more than a century ago.
problems) methods are also available in this regime [1]. An excellent review on the solution of Graetz problem in
The flow is in the transition flow regime for values macroscale can be found elsewhere [9]. Some additional effects
0.1 b Kn b 10. The governing equation for this regime is the like rarefaction (due to small Kn number), axial conduction
 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544 537

(due to finite Peclet number), viscous dissipation (due to high substituting the fully developed velocity profile, the energy
length to diameter ratio) which are important at microscale are equation is solved by using a finite difference scheme to
implemented to the analytical solutions successfully. Numerical determine the temperature distribution within the solution
solutions were also performed for the flows in microchannels. domain and to obtain the Nusselt number. The numerical
These microscale solutions include flow in slit channels [10– algorithm is verified by comparing the present numerical results
12], in microtubes [11–19], in rectangular microchannels [20] with the analytical results that are available for simplified cases
and in other geometries (e.g. trapezoidal) [21] with uniform in literature. The numerical methodology will be used to include
temperature or uniform heat flux at the boundary. the nonlinear effects in the future studies.
The effect of viscous dissipation on heat transfer is
significant for (1) flows with high velocities, (2) flows of 2. Analysis
highly viscous fluids at moderate velocities, or (3) flows
in microscale ducts with a moderate Pr number and The classical approach used for thermally developing flow is
moderate velocities but having small wall-to-fluid temper- not consistent with the presence of fluid axial heat conduction.
ature difference—which is the case for microscale duct flows Therefore, mathematically more consistent boundary conditions
[9]. Some studies include the effect of viscous dissipation are used in the analysis as stated by Shah and London [9], as
[11,17–19,21]. shown in Fig. 1.
For flows with small Peclet number, the axial conduction The fully developed velocity profile can be determined by
cannot be neglected since the characteristic time of convection solving momentum equations [39] together with the slip
and the diffusion become comparable, and the convection term velocity boundary condition at the wall for the hydrodynami-
no longer dominates the conduction term in the longitudinal cally fully developed, constant property fluid flow, as,
direction. The Graetz problem with the inclusion of the axial  
2
conduction term has been an interesting problem due to the u 2 1  ð r=R Þ þ 8Kn
presence of the non self-adjoint eigenvalue problem. Accord- ¼ : ð4Þ
um 1 þ 8Kn
ingly the linearly independent eigenfunctions become non-
orthogonal [22]. This interesting problem has been studied by
many researchers for macrochannels both analytically [23–32] Two-dimensional energy equation for hydrodynamically
and computationally [33,34] for more than three decades ago. fully developed, thermally developing flow with constant
More recently, Hadjiconstantinou and Simek [35] studied the properties can be written as,
effect of axial conduction for thermally fully developed flows in 
 

AT 1A AT A2 T m du 2
micro and nano channels; and Jeong and Jeong [36] studied the u ¼a r þ 2 þ ; ð5Þ
effect of axial conduction together with viscous dissipation in Ax r Ar Ar Ax cp dr
slit channels with micro spacing for thermally developing flow. at r ¼ R T ¼ Ts ; ð6  aÞ
Dutta et al. [37] and Horiuchi et al. [38] studied the thermal
characteristics of mixed electroosmotic and pressure-driven AT
at r ¼ 0 ¼ 0; ð6  bÞ
microflows with axial conduction. Ar
This present study revisits the Graetz problem (2D incom-
AT
pressible, constant property, hydrodynamically developed, at xYl ¼ 0; ð6  cÞ
thermally developing, single-phase laminar flow) considering Ax
rarefaction effect, viscous dissipation term and axial conduction at xY  l T ¼ Ti ; ð6  dÞ
in the fluid for uniform wall temperature boundary condition
inside a microtube. The fully developed velocity profile is where Ti and Ts are the temperature of the gas at the inlet and at
determined by solving momentum equations, analytically. By the surface, respectively.

Fig. 1. Geometry of the problem.

538 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544

The energy equation can be non-dimensionalized by the 3. Numerical solution

following dimensionless quantities,
The governing equation defined in Eq. (9) subjected to the
T  Tw u r x
h¼ ; ū ¼ ; g¼ ; n¼ ; associated boundary conditions, Eqs. (10a, b, c, and d), is solved
Ti  Tw um R R  Pe by using finite difference scheme [41]. 100 × 400 nodes are used
Au2m um D m for Br = 0, and Br b 0 cases, and 100 × 200 nodes are used for the
Br ¼ ; Pe ¼ Re  Pr; Re ¼ ; Pr ¼ Br N 0 case. The mean temperature is evaluated by using
k ð Ti  Tw Þ m a
trapezoidal rule, and local Nu, given by Eq. (13), is calculated
ð7Þ
by using finite difference formulation of the temperature gradient.
The coefficient and the force matrices are assigned as sparse
Since the boundary conditions at infinity are difficult to handle matrices [42], and matrix inversion is performed by MatLab®
in numerical analysis, a coordinate transformation is performed. built-in-function based on Bi-conjugate Gradient Method
By defining a new coordinate variable, ε, suggested by Verhoff (BGM) [43], which is an implicit method, by applying the LU
and Fisher [33], and Yang and Ebadian [40], decomposition as the pre-conditioner [44]. The initial guess,
p  which is dimensionless temperature being equal to 1 in the
n ¼ tan e ; ð8Þ region − ∞ b ξ b 0, and decreasing linearly from 1 to zero in the
2
region 0 b ξ b ∞, is assigned. Central differencing formulas are
and by introducing the dimensionless quantities, Eq. (7), the implemented for interior nodes, and backward differencing for
energy equation and the boundary conditions become, the gradients at the wall. For the nodes at the exit and the
P  centerline, the ghost node approach is implemented.
u BðeÞ Ah A2 h 1 Ah A2 ðeÞ A2 h 16Br The nodes at the exits tend to be equal to zero. Since BGM is
AðeÞ þ 2 ¼ þ þ þ g2 ;
2 Pe Ae Ag2 g Ag Pe2 Ae2 ð1 þ 8KnÞ2 an implicit method, it finds very small values at the exit and
stops the iteration procedure without reaching the desired
ð9Þ

 accuracy for the whole domain. For large Pe values, the
Ah solution clearly reaches the fully developed Nu value, and
at g ¼ 1 hs ¼ 2jKn ; ð10  aÞ
Ag g¼1 experiences oscillations at the exit due to the pre-described fact.
Ah For low Pe values, instead of oscillation the solutions reaches
at g ¼ 0 ¼ 0; ð10  bÞ
Ag some bounded finite Nu value, where the fully developed Nu
value is desired. Therefore, for Br = 0 case 400 nodes are used to
Ah resolve the fully developed Nu region. Once an asymptotic
at e ¼ 1 ¼ 0; ð10  cÞ
Ae value (a plateau) is reached, it is accepted as the fully developed
Nu value. For Br ≠ 0 case, due to the viscous dissipation, even at
at e ¼ 1 h ¼ 1; ð10  dÞ
the exit nodes there have been finite dimensionless temperature
2  FT 2g 1 values, which make the numerical solution stable. 400 nodes are
where j ¼ : ð11Þ used for Br b 0, to resolve the entrance region properly, and 200
FT g þ 1 Pr
nodes become sufficient for Br N 0 case to resolve the whole
2 p  solution domain. 100 nodes in the radial direction are chosen to
and AðeÞ ¼ cos2 e ; get the accurate Nu values up to 2 decimal places compared
p 2
ð12Þ with the results in the literature for the classical Graetz problem
4 p  p  (Nu = 3.66). The present results are compared with the available
BðeÞ ¼ cos3 e sin e :
p 2 2 results in the literature and a good agreement has been achieved.
The comparison of the results is tabulated in Table 1.
κ is a dimensionless parameter that includes the effect of gas
and surface properties. 4. Results and discussions
Introducing dimensionless quantities, the Nusselt number is
determined as, The temperature field is determined by solving the energy
equation numerically. Kn is varied between 0 and 0.1 which are
  the applicability limits of the slip flow regime. Parameter κ is
Ah
2
he D Ag g¼1 taken as 0 and 1.667. κ = 0, is a fictitious case, to see the effect of
Nue ¼ ¼ ; ð13Þ
k hm the slip velocity in the absence of temperature jump. κ = 1.667 is a
typical value for air which is the working fluid in many
where θm is the dimensionless mean temperature, and it is engineering applications (κ = 10 results are only included in
defined as, Table 1). Pe number is varied between 1 and 50, and taken as
1000 to demonstrate Pe → ∞ case.
Z 1
 In Fig. 2, the effect of Pe variation on Nu is demonstrated for
u
hm ¼ 2 hðg; eÞgdg: ð14Þ macrochannels (i.e. Kn=0). For Pe=50, fully developed Nu number
0 um reaches to the 3.66 which is the result for the classical Graetz
 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544 539

Table 1
Comparison of fully developed Nu numbers for present study and the available results from the literature
Kappa Pe = 1.0 Pe = 2.0 Pe = 5.0 Pe = 10 Pe → ∞ Kn
(κ)
Nu∞ Nu⁎∞ Nu∞ Nu⁎∞ Nu∞ Nu∞⁎⁎ Nu∞ Nu⁎∞ Nu∞ Nu⁎⁎⁎
∞

0 4.028 4.030 3.922 3.925 3.767 3.767 3.695 3.697 3.656 3.656 0.00
1.667 4.028 4.030 3.922 3.925 3.767 3.767 3.695 3.697 3.656 3.656
10 4.028 4.030 3.922 3.925 3.767 3.767 3.695 3.697 3.656 3.656
0 4.358 – 4.270 – 4.131 – 4.061 – 4.020 4.020 0.04
1.667 3.604 – 3.517 – 3.387 – 3.325 – 3.292 3.292
10 1.706 – 1.678 – 1.643 – 1.630 – 1.624 1.624
0 4.585 – 4.509 – 4.386 – 4.319 – 4.279 4.279 0.08
1.667 3.093 – 3.036 – 2.949 – 2.909 – 2.887 2.887
10 1.029 – 1.021 – 1.012 – 1.009 – 1.008 1.008
Nu∞: Results for present study.
Nu⁎∞: Results from Shah and London [9].
Nu⁎⁎
∞ : Results from Lahjomri and Qubarra [32].
Nu⁎⁎⁎
∞ : Results from Cetin et al. [11].

problem [39]. For Pe=1, the solution reaches the 4.03 which is finite axial conduction. All local Nu tends to reach infinity at x=0
again the result for the macrochannels [9]. Convective heat transfer for all Pe numbers because the temperature gradient at the wall is
is in the direction of flow, but the diffusion is in both the +x and −x, infinite.
as well as in the r-direction. Therefore, the wall temperature also Fig. 3 illustrates the variation of local Nu number along the
affects the temperature profile at the inlet which is not the case for channel for positive Br numbers and for different Pe numbers.
Pe→∞. For the heating situation, by the presence of axial Positive Br number means fluid is being cooled—inlet
conduction, the heat flow from the wall causes the local temperature temperature is higher than the wall temperature. As expected
rise as well as increasing the temperature at the upstream. From this [9], Nu number experiences a jump because of the heat
aspect our mathematical model allows the heating of the fluid at the generated due to the viscous dissipation. The direction of the
inlet. If temperature profile were uniform at x=0, it would not let the diffusion of the heat in the downstream, therefore the
temperature rise at the inlet. The propagation of the heat to the conductive heat transfer is assisting to the convective one.
upstream leads to a decrease in the local mean temperature. Since the assistance is more dominant by decreasing Pe
Moreover, the gradient at the wall decreases as Pe decreases. With number, the jump-point moves in the downstream direction
the combined effect of these two aspects, local Nu tends to increase like the thermal entrance length. Before the jump, Nu values in
with decreasing Pe as well as the fully developed Nu. This means the entrance region would overlap with the Br = 0 case for
the thermal entrance length also increases with decreasing Pe due to infinite Pe [11]. However by the introduction of the finite Pe

Fig. 2. Variation of local Nu as a function of ξ for different Pe numbers (Kn = 0).

540 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544

Fig. 3. Variation of local Nu as a function of ξ for different Br and Pe numbers (Kn = 0).

number, the curves deviate from the Br = 0 case before the the bulk mean temperature of the fluid becomes equal to the
jump-point, and the deviation increases with decreasing Pe. wall temperature. Beyond that location, fluid starts to heat the
This shifting of the local Nu curve upwards also increases the wall, which would be the undesired case for most of the
mean Nu value for a finite length. Therefore, the mean Nu value practical applications. Fluid temperature continues to develop
in the entrance region increases with decreasing Pe number. thermally and finally reaches the same fully developed Nu as
Fig. 4 shows the variation of local Nu number along the the Br N 0 case. For Br b 0, the diffusive flux is in the opposite
channel for negative Br numbers and for different Pe numbers. direction to the convective flux. Therefore the total wall heat
Br b 0 means fluid is being heated—inlet temperature is lower flux tends to heat the upstream. As Pe number increases, this
than the wall temperature. There exists a singular point for all tendency increases. As a result, the point of singularity moves in
curves where the Nu goes to the infinity. This is the point where the downstream direction, likewise the dimensionless thermal

Fig. 4. Variation of local Nu as a function of ξ for different Br and Pe numbers (Kn = 0).
 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544 541

Fig. 5. Variation of dimensionless mean temperature as a function of ξ for different Br and Pe numbers (Kn = 0.04, κ = 1.667).

entrance length. The effect of both positive and negative Br same dimensional temperature values. For Br b 0 case, the
number on the flow in macrochannels [9] and microchannels curves start from a positive value and reach a negative value
[11] can be found elsewhere in details. which causes the singular behavior of the local Nu curve. The
Fig. 5 demonstrates the change of dimensionless mean location of the singular point on the local Nu curve is the point
temperature along the microtube for different Pe numbers. where the dimensionless mean temperature becomes equal to
Since the Pe numbers are finite, the mean temperature at the zero in the figure. As Pe number decreases, the curves reach the
inlet is less than unity due to conducted heat transfer to the asymptotic values further downstream because of the presence
upstream. For Br = 0 case, the mean temperature reaches the of the heat conduction.
wall temperature value at the downstream. For Br ≠ 0, mean In Fig. 6, the effect of Pe number on Nu number is shown for
temperatures reach their fully developed values. Actually these rarified cases where Kn = 0, 0.04 and 0.08 with κ = 1.667. As
two dimensionless mean temperature values correspond to the Kn decreases the fully developed Nu values decreases. The

Fig. 6. Variation of local Nu as a function of ξ for different Kn and Pe numbers (Br = 0.1, κ = 1.667).
542 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544

Fig. 7. Variation of local Nu as a function of ξ for different Kn and Pe numbers (Br = − 0.1, κ = 1.667).

decreasing Pe number has the same effect for different Kn not mention the clearness of the graph. The behavior in that region
number cases. It shifts both the jump-point and the thermal would be similar to that of the curves in Fig. 4. The curves reach
entrance length to the right due to the axial diffusion. The the fully developed Nu number associated with the positive Br
thermal lengths for different Nu numbers are approximately case, which are decreasing with increasing Nu number. Pe
equal for the same Pe numbers. In other words, the effect of Pe number only affects the thermal entrance length.
number on the thermal entrance length is independent of the Kn Fig. 8 illustrates the effect of Pe number for Kn = 0.04 and
number. κ = 1.667 for both positive and negative Br numbers. The
Fig. 7 demonstrates the effect of Kn and Pe numbers on Nu region prior to the singularity does not shown for negative Br
number for Br b 0. The region before the point of singularity does numbers for the clarity of the figure. The effect of the Pe

Fig. 8. Variation of local Nu as a function of ξ for different Br and Pe numbers (Kn = 0.04, κ = 1.667).
 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544 543

Fig. 9. Variation of local Nu as a function of ξ for different κ parameters and Pe numbers (Br = 0, Kn = 0.04).

number on local Nu is qualitatively same as the effect of the Pe length are found to be increasing with the presence of finite
for Kn = 0 case. Decreasing Pe number shifts the location of the axial conduction (i.e. decreasing Peclet number).
singular point, jump-point and the thermal entrance length to the
downstream. From the graph it seems to be Nu reaching to a Acknowledgment
finite value at x = 0 because of the convexity of the curves
around the inlet, which is a misleading conclusion. All curves Financial support from the Turkish Scientific and Technical
tend to reach infinity at the inlet regardless of the degree of Research Council, Grant No. 106M076, is greatly appreciated.
rarefaction and Pe number. To resolve the region closer to the
entrance better, more nodes should be use in the computation, References
and special attention should be considered to solve this region.
Since microchannels have moderate or high length-over- [1] G.A. Bird, Molecular Gas Dynamics and Direct Simulation of Gas Flows,
diameter ratio, the resolution of this region is not that crucial. Claredon Press, Oxford, 1994.
[2] M. Gad-El-Hak, The fluid mechanics of microdevices, J. Fluid Eng. 121
Fig. 9 shows the effect of Pe number for different κ (1999) 5–33.
parameters on local Nu value (Br = 0, Kn = 0.04). By increasing [3] M. Gad-El-Hak, Gas and liquid transport at the microscale, Heat Transfer
κ, fully developed Nu number decreases due to the higher Eng. 27 (4) (2006) 13–29.
temperature jump at the wall, which decreases the heat transfer [4] Y. Yener, S. Kakac, M. Avelino, T. Okutucu, Single phase forced
convection in microchannels—state-of art-review, in: S. Kakac, L.
at the wall. By increasing Pe number, the fully developed Nu
Vasiliev, Y. Bayazitoglu, Y. Yener (Eds.), Microscale Heat Transfer—
number increases, so does the mean Nu number. The singularity Fundamentals and Applications in Biological Systems and MEMS,
point moves in the downstream direction due to the diffusion of Kluwer Academic Publisher, The Netherlands, 2005, pp. 1–25.
the heat in the upstream direction. Beyond the singular point, [5] F.E. Larrode, C. Housiadas, Y. Drossinos, Slip-flow heat transfer in
the curves have the exactly same nature as the positive Br cases. circular tubes, Int. J. Heat Mass Transfer 43 (2000) 2669–2680.
[6] E.G.R. Eckert, R.M. Drake Jr., Analysis of Heat and Mass Transfer,
McGraw-Hill, New York, 1972, pp. 467–486.
5. Conclusions [7] L. Graetz, Uben die Wärmeleitungsfähigkeit von Flüssigkeiten (On the
thermal conductivity of liquids), Part 1, Ann. Phys. Chem. 18 (1883)
In this study, the classical Graetz problem inside a microtube 79–94;
is solved by considering rarefaction, viscous dissipation and Part 2, Ann. Phys. Chem. 25 (1885) 337–357.
axial conduction effects. The effect of the Peclet number on the [8] W. Nusselt, Die Abhängigkeit der Wärmeübergangszahl von der
Rohrlänge (The dependence of the heat transfer coefficient on the tube
local Nusselt number is discussed. The energy equation is length), VDI Z. 54 (1910) 1154–1158.
solved numerically to be able to extend the present study for [9] R.K. Shah, A.L. London, in: T.F. Irvine Jr., J.P. Hartnett (Eds.), Laminar
further more complicated cases in microscale heat transfer (e.g. flow forced convection in ducts, advances in heat transfer, Academic Press,
variable thermo-physical properties). The solution domain is New York, 1978, pp. 78–152.
extended to − ∞ and +∞ to formulate the problem and boundary [10] H.P. Kavehpour, M. Faghri, Y. Asako, Effects of compressibility and
rarefaction in gaseous flows in microchannels, Numer. Heat Transf. 32 (A)
conditions mathematically consistent. Some numerical instabil- (1997) 677–696.
ities are addressed in the numerical methods for Br = 0 case. [11] B. Cetin, H. Yuncu, S. Kakac, Gaseous flow in microchannels with viscous
Fully developed Nusselt number and the thermal entrance dissipation, Int. J. Trans. Phenom. 8 (4) (2006) 297–315.
544 B. Cetin et al. / International Communications in Heat and Mass Transfer 35 (2008) 535–544

[12] M.D. Mikhailov, R.M. Cotta, S. Kakac, Steady state and periodic heat [27] E. Papoutsakis, D. Ramkrishna, H.C. Lim, The extended Graetz problem
transfer in micro conduits, in: S. Kakac, L. Vasiliev, Y. Bayazitoglu, Y. with Dirichlet wall boundary condition, Appl. Sci. Res. 36 (1980) 13–34.
Yener (Eds.), Microscale Heat Transfer—Fundamentals and Applications [28] E. Papoutsakis, D. Ramkrishna, H.C. Lim, The extended Graetz problem
in Biological Systems and MEMS, Kluwer Academic Publisher, The with prescribed wall flux, AIChE J. 26 (1980) 779–787.
Netherlands, 2005, pp. 75–92. [29] A. Acrivos, The extended Graetz problem at low Peclet numbers, Appl.
[13] S.B. Choi, R.F. Barron, R.O. Warrington, Fluid flow and heat transfer in Sci. Res. 36 (1980) 35–40.
microtubes, micromechanical sensors, actuators, and systems, ASME DSC [30] B. Vick, M.N. Ozisik, Y. Bayazitoglu, A method of low Peclet number
32 (1991) 123–134. thermal entry region problems with axial conduction, Lett. Heat Mass
[14] A. Beskok, G.E. Karniadakis, A model for flows in channels, pipes, and Transf. 7 (1980) 235–248.
ducts at micro and nano scales, Microscale Thermophys. Eng. 3 (1999) [31] B. Vick, M.N. Ozisik, An exact analysis of low Peclet number heat transfer
43–77. in Laminar flow with axial conduction, Lett. Heat Mass Transf. 8 (1981)
[15] R.F. Barron, X.M. Wang, R.O. Warrington, T.A. Ameel, The Graetz 1–10.
problem extended to slip flow, Int. J. Heat Mass Transfer 40 (8) (1997) [32] J. Lahjomri, A. Qubarra, Analytical solution of the Graetz problem with
1817–1823. axial conduction, J. Heat Transfer 121 (1999) 1078–1083.
[16] T.A. Ameel, R.F. Barron, X.M. Wang, R.O. Warrington, Laminar forced [33] F.H. Verhoff, D.P. Fisher, A numerical solution of the Graetz problem with
convection in a circular tube with constant heat flux and slip flow, axial conduction, J. Heat Transfer 95 (1973) 12–134.
Microscale Thermophys. Eng. 1 (1997) 303–320. [34] M.L. Michelsen, J. Villadsen, The Graetz problem with axial heat
[17] G. Tunc, Y. Bayazitoglu, Heat transfer in microtubes with viscous conduction, Int. J. Heat Mass Transfer 17 (1974) 1391–1402.
dissipation, Int. J. Heat Mass Transfer 44 (13) (2001) 2395–2403. [35] N.G. Hadjiconstantinou, O. Simek, Constant-wall-temperature Nusselt
[18] G. Tunc, Y. Bayazitoglu, Convection at the entrance of micropipes with number in micro and nano-channels, J. Heat Transfer 124 (2002) 356–364.
sudden wall temperature change, in: Proceedings of IMECE 2002, [36] H.E. Jeong, J.T. Jeong, Extended Graetz problem including streamwise
November 17–22, 2002 New Orleans, Lousiana. conduction and viscous dissipation in microchannels, Int. J. Heat Mass
[19] B. Xu, K.T. Ooi, C. Mavriplis, M.E. Zaghloul, Evaluation of viscous Transfer 49 (2006) 2151–2157.
dissipation in liquid flow in microchannels, J. Micromech. Microeng. 13 [37] P. Dutta, K. Horiuchi, H.M. Yin, Thermal characteristics of mixed
(2003) 53–57. electroosmotic and pressure-driven microflows, Comput. Math. Appl.
[20] G. Tunc, Y. Bayazitoglu, Heat transfer in rectangular microchannels, Int. J. 52 (2006) 651–670.
Heat Mass Transfer 45 (2002) 765–773. [38] K. Horiuchi, P. Dutta, A. Hossain, Joule-heating effects in mixed
[21] G.L. Morini, Viscous heating in liquid flows in microchannels, Int. J. Heat electroosmotic and pressure-driven microflows under constant wall heat
Mass Transfer 48 (2005) 3637–3647. flux, J. Eng. Math. 54 (2006) 159–180.
[22] W.M. Deen, Analysis of Transport Phenomena, Oxford University Press, [39] S. Kakac, Y. Yener, Convective heat transfer, Pub. No: 65, METU, Ankara,
Oxford, 1998, pp. 391–392. 1980.
[23] R.L. Ash, J.H. Heinbockel, Note on heat transfer in laminar fully [40] G. Yang, M. Ebadian, Thermal radiation and laminar forced convection in
developed pipe flow with axial conduction, Math. Phys. 21 (1970) the entrance region of a pipe with axial conduction and radiation, Int. J.
266–269. Heat Fluid Flow 12 (3) (1991) 202–209.
[24] C.J. Hsu, An exact analysis of low Peclet number thermal entry region heat [41] S.C. Chapra, R.P. Canale, Numerical Methods for Engineers, 3rd Edition,
transfer in transversely non-uniform velocity field, AIChE J. 17 (1971) McGraw-Hill, 1998, pp. 632–635.
732–740. [42] I.S. Duff, A.E. Erisman, J.K. Reid, Direct Methods for Sparse Matrices,
[25] V. Taitel, A. Tamir, Application of the integral method to flows with axial Clarendon Press, Oxford, 1986.
conduction, Int. J. Heat Mass Transfer 15 (1972) 733–740. [43] G.H. Golub, J.F. VanLoan, Matrix Computations, 3rd Edition, The John
[26] V. Taitel, M. Bentwich, A. Tamir, Effects of upstream and downstream Hopkins University Press, 1996, pp. 508–544.
boundary conditions on heat(mass) transfer with axial diffusion, Int. J. [44] K. Chen, Matrix Preconditioning Techniques and Applications, Cambridge
Heat Mass Transfer 16 (1973) 359–369. University Press, Cambridge, 2005, pp. 68–75.