International Journal of Thermal Sciences 42 (2003) 513–521

www.elsevier.com/locate/ijts

Analysis of entropy generation inside concentric cylindrical annuli

with relative rotation
Shohel Mahmud ∗ , Roydon Andrew Fraser
Department of Mechanical Engineering, University of Waterloo, 200 University Avenue West, Waterloo, ON, Canada N2L 3G1
Received 27 November 2001; accepted 20 June 2002

Abstract
The present article investigates analytically the First and Second Laws (of Thermodynamics) characteristics of fluid flow and heat transfer
inside a cylindrical annulus. A relative rotational motion presents between the inner and the outer cylinders which induces the flow. Two
different cases are considered: (a) both cylinders are isothermal and kept at different temperatures and (b) the outer cylinder is isoflux
and the inner is isothermal. Governing equations in cylindrical coordinates are simplified and solved to obtain analytical expressions for
dimensionless entropy generation number (NS ), irreversibility distribution ratio (Φ), and the Bejan number (Be) as a function of flow
governing and geometric parameters. Spatial distribution of velocity and temperature, volumetric and average entropy generation rate, and
heat transfer irreversibility are presented graphically. The effect of velocity ratio (λ), the group parameter (Br/Ω), and the Brinkman number
(Br) on the above parameters are tested.
 2002 Éditions scientifiques et médicales Elsevier SAS. All rights reserved.

Keywords: Bejan number; Entropy generation number; Irreversibility distribution ratio

1. Introduction coefficients in a rotating inner cylinder system with and

without axial flow. Bjorklund and Kaye [5] proposed a
Flow induced by a relative rotating motion or axial move- correlation for heat transfer measurement based on their ex-
ment between cylinders in a concentric arrangement has perimental data for zero axial flow. Later work of Tachibana
many significant engineering applications. In addition to et al. [6] showed excellent agreement with the correlation
heat transfer situations, the resulting flow is particularly of Bjorklund and Kaye [5]. Aoki et al. [7] predicted the-
applicable to rotating electrical machines, swirl nozzles, oretically the overall heat transfer coefficients limited to a
rotating disks, standard commercial rheometers, and other small gap widths for a fluid having Prandtl number equal
chemical and mechanical mixing equipment (see Maron and to 1. Effect of natural convection was studied by Leonardi
Cohen [1]). Since the pioneering work of Taylor [2], numer- et al. [8] for a finite annular gap with differentially heated
ous analytical and experimental works have been performed walls. El-Shaarawi and Sarhan [9], Gardiner and Sabersky
to predict flow, and thermal fields, stability, heat/mass trans- [10], and Gasley [11] have significant contributions related
fer characteristics, etc., inside the concentric annular space. to above-mentioned researches. The foregoing discussions
For one fixed and one rotating cylinder, Astill [3] and An- are some of the extensive research efforts concerning flow
dereck et al. [4] presented a developing flow with different and heat transfer between rotating surfaces. For a compre-
flow regimes inside a concentric cylindrical annulus. Heat hensive review, see the paper by Dorfman [12] or Childs and
transfer studies have been primarily stimulated by cooling Long [13].
problems in the design of electric motors of high power The foregoing discussions are a small part of wide re-
density (see Maron and Cohen [1]). Most experimental in- search efforts related to flow and thermal problems inside an-
vestigations focused on the measurements of heat transfer nular gap with rotating or axially moving cylinder. Although
these works have covered a wide variety problems involving
* Corresponding author. concentric cylinders, these problems have been restricted, in
E-mail addresses: [email protected] (S. Mahmud), thermodynamic point of view, to only the First Law (of Ther-
[email protected] (R.A. Fraser). modynamics) analyses. The contemporary trend in the field
1290-0729/02/$ – see front matter  2002 Éditions scientifiques et médicales Elsevier SAS. All rights reserved.
doi:10.1016/S1290-0729(02)00051-0
514 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521

Nomenclature

Be Bejan number, = NR /NS Greek symbols

Br Brinkman number, = Ec × Pr
α thermal diffusivity . . . . . . . . . . . . . . . . . . . m2 ·s−1
Cn integration constants, n = 1, 2, . . .
Γn constants, n = 1, 2, . . .
Cp specific heat at constant pressure . kJ·kg−1 ·K−1
µ dynamic viscosity . . . . . . . . . . . . . . . . . . . . . . Pa·s
e exponential (2.718281)
ρ density of the fluid . . . . . . . . . . . . . . . . . . kg·m−3
Ec Eckert number, = (ω0 .r0 )2 /(Cp .T )
λ velocity ratio, = ω1 .r1 /(ω0 .r0 )
k thermal conductivity of fluid . . . . . W·m−1 ·K−1
ω angular velocity . . . . . . . . . . . . . . . . . . . . . rad·s−1
K argument of Lambert function
Π radius ratio, = r1 /r0
NF entropy generation number; fluid friction
Φ irreversibility distribution ratio, = NF /NR
NR entropy generation number; radial
Θ dimensionless temperature, = (T − T0 )/T
NS entropy generation number; total
Ω dimensionless temperature difference, = T /T0
P pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pa
∀ volume of the annular gap . . . . . . . . . . . . . . . . m3
Pr Prandtl number, = µ.Cp /k
q constant heat flux at wall . . . . . . . . . . . . . W·m−2 Subscript and superscript
r radial distance . . . . . . . . . . . . . . . . . . . . . . . . . . . m 0 value at the inner cylinder
R dimensionless radial distance, = r/r0 1 value at the outer cylinder
SG entropy generation rate . . . . . . . . . . W·m−3 K−1 C critical value
T temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ◦ C o average value
u tangential velocity . . . . . . . . . . . . . . . . . . . . m·s−1 T for isothermal boundary condition
U dimensionless tangential velocity, = u/(ω0 .r0 ) q for isoflux boundary condition

of heat transfer and thermal design is to perform a Second differentially heated isothermal boundary condition. Yilbas
Law (of Thermodynamics) analysis and its design-related [17] assumes a linear velocity profile and neglects the contri-
concept of entropy generation and its minimization (Bejan bution of fluid friction irreversibility to entropy generation.
[14]). This new trend is important and, at the same time, For other geometries, Second Law analyses as well as en-
necessary, if the heat transfer community is to contribute to tropy generation profiles are available in the references by
a viable engineering solution to the energy problems. Drost and Zaworski [18] and Bejan [15,16].
Entropy generation is associated with thermodynamic ir- In this paper, the governing equations in cylindrical
reversibilities, which is common in all types of heat transfer coordinates are simplified and solved using both isothermal
processes. Different sources are responsible for the genera- and isoflux boundary conditions assuming a relative angular
tion of entropy, for example, heat transfer down a temper- rotation is present between the cylinders in a concentric
ature gradient, viscous effects, etc. Bejan [15] has focused arrangement. Subsequently, expressions for dimensionless
on the different mechanisms behind entropy generation in entropy generation number, irreversibility distribution ratio,
applied thermal engineering. Generation of entropy destroys and Bejan number are derived.
the available work of a system. Therefore, it makes good
engineering sense to focus on irreversibilities (see Bejan
[14,15]) of heat transfer and fluid flow processes and try 2. Physical model and derivation
to understand the function of related entropy generation
mechanisms. Bejan [16] presented the Second Law aspect 2.1. First Law analysis
of heat transfer using different forced convection problem
examples. Bejan [16] introduced the concept of entropy Consider the steady flow maintained between two con-
generation number, irreversibility distribution ratio, and pre- centric cylinders, shown in Fig. 1, by a steady angular
sented spatial distribution profiles of entropy generation for velocity of one or both cylinders. Let the inner and outer
the example problems. Since then, numerous investigations
have been performed to determine the entropy generation
and irreversibility profiles for different geometric configu-
rations, flow situations, and thermal boundary conditions.
Most of the works are numerical calculations due to the non-
linear nature of flow the governing equations. Very few of
these works consider an analytical approach to a solution.
For a concentric cylindrical annulus, Yilbas [17] presented
an entropy analysis with a rotating outer cylinder and a Fig. 1. Schematic diagram of the problem under consideration.
 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521 515

cylinders have radii r0 and r1 , respectively. Steady angu- where Ec is the Eckert number and Pr is the Prandtl
lar velocities are denoted by ω0 and ω1 for the inner and number. Putting the expression for velocity U into Eq. (7)
the outer cylinders. Consider no relative axial movement and integrating, the solution to Eq. (7) for dimensionless
between the cylinders. Neglecting the radial velocity compo- temperature is
nent compared to the tangential component of the velocity,
the simplified momentum equation in cylindrical coordi- (Π − λ)2 Ec Pr
Θ = C3 + C4 ln(R) −
nates is (Π 2 − 1) R 2
∂ 2 uθ 1 ∂uθ uθ Ec Pr
+ − 2 =0 (1) = C3 + C4 ln(R) − Γ3 2 (8)
∂r 2 r ∂r r R
where uθ is the tangential velocity component. Neglecting In the above expression, C3 and C4 are constants of
the convection term, the energy equation becomes integration which depend on the boundary conditions. Γ3


 is a constant and equal to (Π − λ)2 /(Π 2 − 1). For the
k ∂ ∂T ∂uθ uθ 2 isothermal boundary condition, Θ = 0 at R = 1 and Θ = 1
r = −µ − (2)
r ∂r ∂r ∂r r at R = Π . Using these values, non-dimensional temperature
subjected to the following boundary conditions distribution for the isothermal boundary condition becomes

 
At r = r0 : uθ = u0 = r0 ω0 and T = T0 1 ln(R) ln(R)
ΘT = Γ3 Ec Pr 1 − 2 1 − + (9)
∂T q (3) R ln(Π) ln(Π)
At r = r1 : uθ = u1 = r1 ω1 and T = T1 or =
∂r k For the isoflux boundary condition, a constant heat flux
Integrating Eq. (1) with respect to r, the solution to the q is applied to the outer cylinder, but the temperature at
momentum equation is the inner cylinder is kept constant as isothermal case. For
C2 this particular case, Θ = 0 at R = 1 and ∂Θ/∂R = 1 at
uθ = C1 r + (4) R = Π . Using these values, non-dimensional temperature
r
distribution for the isoflux boundary condition becomes
where C1 and C2 are two constants of integration. Applying  
boundary conditions described in Eq. (3), the tangential 1 2 ln(R)
Θq = Γ3 Ec Pr 1 − 2 − + Π ln(R) (10)
velocity uθ takes the following form: R Π2
r1 /r − r/r1 r/r0 − r0 /r
u = uθ = r0 ω0 + r1 ω1 (5) 2.2. Second Law analysis
r1 /r0 − r0 /r1 r1 /r0 − r0 /r1
In subsequent calculations, u is used in place of uθ as The foundations of our knowledge of entropy production
tangential velocity. Expressing the ratio r1 /r0 as Π and goes back to Clausius and Kelvin’s studies on the irreversible
u1 /u0 as λ, the dimensionless form of Eq. (5) is aspects of the Second Law of Thermodynamics. Since then

2 
2 
1 Π − R2 Πλ R −1 the theories based on these foundations have rapidly de-
U = 2 + 2 veloped. However, the entropy production resulting from
Π −1 R Π −1 R

2 
2  temperature differences has remained untreated by classi-
Π −R 2 R −1
= Γ1 + Γ2 (6) cal thermodynamics thus motivating many researchers to
R R conduct analysis of fundamental and applied engineering
where R is the dimensionless radial distance which is problems based on Second Law analyses. Review of such
equal to r/r0 . Tangential velocity u is made dimensionless analyses is beyond the scope of this paper; for a compre-
dividing by ω0 r0 . The constants Γ1 and Γ2 are equal to hensive review, see Bejan [15]. Based on the Second Law
1/(Π 2 − 1) and Πλ/(Π 2 − 1), respectively. In the above of thermodynamics and assumptions already made, the local
expression of velocity, constraints for different parameters volumetric rate of entropy generation, SG (W·m−3 ·K−1 ), in
are 1
R
∞, 1 < Π
∞, and −∞
λ
+∞. For cylindrical coordinates is shown in the following equation
positive λ, both cylinders rotate in the same direction and (for detail derivation see Bejan [15])
for negative λ, they rotate in the opposite directions. For
 

k ∂T 2 µ ∂ u 2
a stationary outer cylinder or very large rotation of the SG = 2 + r (11)
inner cylinder, the second part of Eq. (6) disappears. Before T0 ∂r T0 ∂r r
solving the energy equation (Eq. (2)) it is put into the The above form of entropy generation shows that the
dimensionless form. Velocity u is scaled with ω0 r0 , radial irreversibility is due to two effects, a conductive (k) effect
distance r is scaled with r0 , and dimensionless temperature and a viscous (µ) effect. Entropy generation rate (SG ) is
Θ can be expressed as (T − T0 )/T where T0 is the positive and finite as long as temperature and/or velocity
reference temperature and T is the reference temperature gradients are present in the medium. According to Bejan
difference. The dimensionless form of Eq. (2) is [14], the dimensionless form of SG is the entropy generation



1 ∂ ∂Θ ∂U U 2 number (NS ) and which is, by definition, equal to the ratio
R = −Ec Pr − (7) of actual entropy generation rate (SG ) to a characteristic
R ∂R ∂R ∂R R
516 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521

entropy transfer rate (SG,C ). According to Bejan [14,16] the problems both fluid friction and heat transfer have contribu-
characteristic entropy transfer rate is tions to the rate of entropy generation. Expression of entropy
 2    generation number (NS ) is good for generating spatial en-
q k(T )2
SG,C = or (12) tropy profile, but it fails to give any idea of whether fluid
kT02 Isothermal r02 T02 Isoflux friction or heat transfer is the dominating entropy generation
The first square bracketed term is used for isoflux mechanism. According to Bejan [14], the irreversibility dis-
boundary condition and the second square bracketed term tribution ratio (Φ) takes care the above problem and which
is used for isothermal boundary condition. Using the same is equal to the ratio of entropy generation due to fluid fric-
parameters, which are already used for scaling purpose, the tion (NF ) to heat transfer (NR ). Heat transfer dominates
dimensionless form of Eq. (11) is for 0
Φ < 1 and fluid friction dominates when Φ > 1.

 
 For Φ = 1, both heat transfer and fluid friction have the
∂Θ 2 Ec Pr ∂ U 2 same contribution for generating entropy. In many engineer-
NS = + R = NR + NF (13)
∂R Ω ∂R R ing designs and optimization problems (see Bejan [21]),
the contribution of heat transfer entropy on overall entropy
In the above equation, Ω is the dimensionless temperature
generation rate is needed. As an alternative irreversibility
difference, which is equal to T /T0 . On the right-hand
distribution parameter, Paoletti et al. [20] defines the Bejan
side of Eq. (13), the first term (= NR ) accounts for entropy
number (Be) which is the ratio of entropy generation due to
generation due to heat transfer in radial direction and the
heat transfer to the total entropy generation. Mathematically
second term (= NF ) is the fluid friction contribution to
Bejan number is
entropy generation. Combining Eqs. (6), (9), and (13),
the entropy generation number (NST ) for the isothermal NR 1
Be = = (16)
boundary condition is NR + NF 1+Φ
   The Bejan number ranges from 0 to 1. Accordingly, Be =
2BrΓ3 ln(R)
NST = 1− 1 is the limit at which the heat transfer irreversibility
R3 ln(Π)

 2 dominates, while Be = 0 is the opposite limit at which the
BrΓ3 1 1 irreversibility is dominated by fluid friction effect, and Be =
− 1− 2 +
R ln(Π) R R ln(Π) 1/2 is the case in which the heat transfer and fluid friction
 2 entropy generation rates are equal. Using Eqs. (14) and (16)
4Br Γ3 Π 2 − Γ2
+ (14) and after avoiding detail mathematical operation; the Bejan
Ω R2 number for isothermal boundary condition becomes
Combining Eqs. (6), (10), and (13), entropy generation  
number (NSq ) for the isoflux boundary condition is {2 ln(Π/R) − R 2 + R 2 /(BrΓ3 ) + 1} 2
BeT = BrΓ3

  {R 2 ln(Π)}
2BrΓ3 1 1 Π 2  
NSq = − + {2 ln(Π/R) − R 2 + R 2 /(BrΓ3 ) + 1} 2
R R2 Π 2 R × BrΓ3
  {R 2 ln(Π)}
4Br Γ1 Π 2 − Γ2 2 
+ (15) 4Br  2 −1
Ω R2 + Γ 1 Π 2
− Γ 2 (17)
ΩR 2
In Eqs. (14) and (15), Br is the Brinkman number, which Using Eqs. (15) and (16), the Bejan number for isoflux
is the product of the Eckert (Ec) and Prandtl number (Pr). boundary condition becomes
The Brinkman number determines the relative importance
 
between dissipation effects and fluid conduction effects (Π 2 − R 2 ) 2
(see White [19] for details). For each expression, the first Beq = Π + 2BrΓ3
R2 Π 2
square bracketed term at the right-hand side represents heat  
(Π 2 − R 2 ) 2
transfer contribution and the second square bracketed term × Π + 2BrΓ3
is the fluid friction contribution to entropy generation. Note R2 Π 2

that, for both isothermal and isoflux boundary conditions, 4Br  2 −1
fluid friction contribution to the entropy generation are + Γ 1 Π 2
− Γ 2 (18)
ΩR 2
same. Because flow and thermal fields are not coupled
like buoyancy flow and velocity field is independent of
temperature field. 3. Results and discussions

2.3. Fluid friction versus heat transfer irreversibility In many practical situations, one of the two cylinders in
the concentric orientation is kept fixed (λ = 0). This special
Entropy generates in a process or system due to the pres- case is presented graphically in Fig. 2 where dimensionless
ence of irreversibility (see Bejan [14,15]). In convection velocity profiles are plotted as a function of radial distance
 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521 517

Fig. 2. Velocity profiles at different λ. Fig. 3. Temperature distribution for the isothermal boundary condition.

for radius ratio Π = 5.0 and different velocity ratios (λ). cylinder. The peak location of temperature is also important,
As long as the gap between the cylinders remains small because at this position entropy generation due to radial
(Π ≈ 1) relative to their radii, the velocity profile between temperature gradient (∂T /∂R) is zero. For any value of
the cylinders approaches the idealized linear shear flow λ, Π , and Br, the radial location for the peak value of
profile (see Leal [22]). At higher Π , velocity profiles temperature can be obtained from the following expression
become non-linear due to the simultaneous contributions of    
1 (Γ3 Br − 1)Π 2 e 1
the R and 1/R terms in Eq. (6). Velocity decreases along RΘmax = Π exp − LambertW +
2 Γ3 Br 2
the radial direction and shows a minimum value at the outer
(21)
cylinder up to λ ≈ 0.3846 which can be determined from the
following relation In the above expression, the special function ‘LambertW’
2Π with a general argument ‘K’ can be evaluated using the
λ = λC = (19) simplified expression given in Eq. (22).
1 + Π2
The above relation actually determines the value of λ (= λC ) LambertW{K}
  
at which velocity gradient (∂U/∂R) is zero at the outer 
 0.665 1 + 0.0195 ln(K + 1) ln(K + 1)
cylinder for a particular value of radius ratio (Π). For ≈ + 0.04, 0
K
500    (22)
λC < λ
1, minimum velocity occurs inside the annular 
 ln(K − 4) − 1 − 1/ ln(K) ln ln(K) ,
gap and the radial location of the minimum velocity can be K > 500
determined from the following relation Entropy generation number (NST ) is plotted as a function
of radial distance in Fig. 4 for Π = 2.0, λ = 0, Br = 1.0,
Π(Π − λ)
RUmin = (20) and for group parameters (Br/Ω) ranging 0 to 1. The group
Πλ − 1 parameter determines the relative importance of viscous
The above relation is important, because at this radial posi- effects and has significant effect on entropy generation. For
tion, fluid friction contribution to the rate of entropy genera- all group parameters, the inner cylinder acts as a strong
tion is zero due to the zero velocity gradient (∂U/∂R). concentrator of irreversibility. Entropy generation number
At the first part of this section, discussion is restricted to is high in magnitude near the inner cylinder due to the
the isothermal case and the isoflux case will be discussed high gradient of temperature and velocity. NST then falls
later. Fig. 3 shows the dimensionless temperature profiles exponential like along the radial direction and approaches
as a function of radial distance for λ = 0 and Π = 5.0 an asymptote near the outer cylinder. NST profiles are
at different Br (= Ec × Pr). Temperature is equal to 0 similar in shape and almost parallel to one another whatever
and 1 at the inner and the outer cylinder, respectively. the value of group parameter, but vary in magnitudes. For
For Br > 1, temperature rise inside the fluid is significant constant Br, Π , and λ, the magnitude of NST depends
due to the dissipation effect. A marked peak value of only on the group parameter (Br/Ω) which is a part of the
temperature occurs inside the annular space, which is higher fluid friction irreversibility (NF ) term in entropy generation
in magnitude than the hot wall temperature. The wall heat number (see Eq. (14)). So, an increasing or decreasing Br/Ω
flux actually is into the outer cylinder even though the just increases or decreases the magnitude of the NST profile
imposed temperature difference would initially have been without changing its shape. At a particular radial position,
considered to cause wall heat flux to be out of the outer magnitude of NST is higher for higher group parameter.
518 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521

Fig. 4. Isothermal entropy generation profiles at different group parameters. Fig. 6. Isothermal Bejan number distribution at different group parameters.

Fig. 5. Isothermal entropy generation profiles at different velocity ratios. Fig. 7. Isothermal Bejan number distribution at different velocity ratios.

The effect of velocity ratio (λ) on entropy generation is Bejan number (BeT ) is plotted in Fig. 6 as a function
presented in Fig. 5 for Π = 2.0, Br = 1.0 and Br/Ω = 1.0. of radial distance for Π = 2.0, λ = 0, Br = 2.0, and group
Value of λ is kept between 0 and 1 because of practical parameters 0 to 1. For Br/Ω = 0, fluid friction contribution
importance. The inner wall still acts as a strong concentrator to entropy generation is zero. The Bejan number is equal
of irreversibility, but now the magnitude of NST drops to its maximum value (= 1) and independent of radial
significantly at the inner wall for higher λ due to the lower distance at this group parameter. For Br/Ω > 0, BeT is still
temperature and velocity gradients. With the increase of maximum at the inner wall, but its magnitude drops with the
radial distance, NST profiles approaches one another and increase of Br/Ω. With the increase of radial distance, BeT
merge at or near R = 1.70. NST is inversely proportional to falls and approaches zero at R ≈ 1.55 for all values of group
the R, R 2 , R 3 terms (see Eq. (14)). For constant Br, Br/Ω, parameters. The zero value of BeT is due to the maximum
and λ, NST falls with increasing R. It is interesting to note radial temperature gradient at this point and this location can
that a big contribution to entropy generation comes from be determined by Eqs. (21) and (22).
the first of the three parts inside the first square bracketed The effect of velocity ratio λ on BeT is shown in Fig. 7
term (NR ) of Eq. (14). When R approaches Π , the first one where BeT is plotted against R for Π = 2.0, Br = 4.0, and
of these three parts approaches zero. Initially (at low R) Br/Ω = 1. Heat transfer irreversibility is still maximum at
NST shows different magnitude for different λ, but with the inner wall for all values of λ due to high temperature
the increasing R contribution to the entropy generation and velocity gradient. The magnitude of BeT is higher for
from the dominating term decreases resulting a trends of lower value of λ at the inner wall. BeT then falls along
merging profiles with each other. The magnitude of entropy the radial direction. For all values of λ, profiles of BeT
generation number is same for rest of the radial distance. intersect with one another at R ≈ 1.2. Contribution of heat
 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521 519

Fig. 8. Isothermal Bejan number distribution at different Brinkman num- Fig. 9. Temperature distribution for the isoflux boundary condition.
bers.

transfer irreversibility to entropy generation is the same at

this point for all velocity ratios (λ) provided that λ
1. This
behavior depends on the Brinkman number as long as the
radius ratio (Π) is kept constant. The lowest location of
BeT profile, which represents the zero contribution of heat
transfer irreversibility, shifts towards the outer cylinder for
higher λ. In Fig. 8, the effect of the Brinkman number on
heat transfer irreversibility is presented for Π = 2.0, λ = 0,
Br/Ω = 1.0. For Br = 1, the Bejan number falls along the
radial direction and no distinct minimum point is observed
inside the annular gap due to the absence of any peak in
temperature distribution. For Br > 1, the Bejan number falls
rapidly along the radial direction, touches the minimum
value (= 0) and then increases towards the outer cylinder.
The minimum location for BeT profile shifts towards the
Fig. 10. Isoflux entropy generation profiles at different group parameters.
outer cylinder for lower Br. BeT is higher at the higher value
of Br at both the inner and outer cylinders.
For the isoflux boundary condition, dimensionless tem- similar to NST profiles except the slower falling rate that is
perature profiles are plotted as a function of radial distance in observed for the isoflux boundary condition. The effect of
Fig. 9 for different Brinkman numbers. Temperature profiles velocity ratio on entropy generation is presented in Fig. 11
in this case are not similar in shape to the isothermal case. and the profiles are very much similar to the isothermal case.
The temperature gradient is higher near the inner cylinder The dimensionless volumetric averaged entropy gener-
ation rate (NSTo ) can be evaluated using Eq. (23) for the
and this gradient increases with Br. Temperature increases
along the radial direction with the decreasing radial tempera- isothermal case
 
ture gradient. Temperature gradient is inversely proportional 1 1
NST =
o
NST d∀ = NST r dθ dr dz
to R (see Eq. (10)). So, with increasing R, slope of the ∀ ∀
Θq – R profile decreases. For a particular radial position, ∀ ∀

temperature is higher for higher value of Br. Fig. 10 shows = f (Br, Br/Ω, Π, λ) (23)
the variation of entropy generation number (NSq ) with ra- In the above equation, ∀ is the volume of the annular
dial distance for Π = 2.0, λ = 0, and Br = 1.0 at different space. Average entropy generation rate is a function of
group parameters. The magnitudes of entropy generation geometric parameter Π , flow parameter λ, dimensionless
numbers for both isothermal and isoflux cases are very much numbers Br, and group parameter Br/Ω. For a significant
comparable though the temperature is higher for isoflux practical interest, we focus our attention on parameters Π
boundary condition. Because the basic equation (Eq. (13)) and λ. Fig. 12 shows the distribution of average entropy
which is used for generating expression for entropy gener- generation as function of velocity ratio for Π = 2.0 and
ation number deals with gradient of temperature instead of Br = 2.0 at different group parameter. Velocity ratio (λ) is
the magnitude of temperature. NSq profiles are very much kept between −1 to 5 for convenience. Both cylinders rotate
520 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521

Fig. 11. Isoflux entropy generation profiles at different velocity ratios. Fig. 13. Average entropy generation profiles at different Brinkman numbers.

Fig. 12. Average entropy generation profiles at different group parameters. Fig. 14. Minimum average entropy generation profiles.

in same direction for positive λ and opposite directions for where K0 –K3 are coefficients at different power of λ and
negative λ. The dashed line represents the Couette flow function of Br, Br/Ω, and Π . Eq. (24) has three solutions
situation where λ = 0 (fixed outer cylinder). For this range and two of them give imaginary results. The only real
of λ, average entropy generation rate (NSTo ) shows a concave solution is λ = Π . Fortunately, prior graphical presentations
shaped distribution pattern. For all group parameters, the (Figs. 12 and 13) satisfy this solution (λ = Π) for minimum
magnitude of minimum NST o
is same (≈ 0.6937) and which o
NST . For λ = Π , the angular velocity of the outer cylinder
occurs at λ = 2.0. Except λ = 2.0, NST o is higher for (ω1 ) is equal to the inner cylinder (ω0 ), i.e., no relative
o angular motion exists between the outer and inner cylinders.
higher group parameter. NST is plotted against λ at different
Brinkman numbers in Fig. 13. A similar concave shaped After substituting λ = Π in Eq. (23), the expression for
o
pattern is observed and each Br, profile is symmetrical about minimum NST becomes
the λ = 2.0 line. For all values of Br, minimum NST o
is the
  2
same and equal to 0.6937. Except λ = 2.0, magnitude of o
NST = (25)
NSTo is higher for higher Br. To locate the minimum N o , min Π(Π + 1)[ln(Π)]2
ST
Eq. (23) is differentiated with respect to λ resulting a third o ]
[NST min is only a function of geometric parameter Π and
order polynomial as follows independent of Br, Br/Ω. This can be easily verified from
o Figs. 12 and 13 where for all values of the group parameter
∂NST
= 0 = fλ (Br, Br/Ω, Π, λ) and the Brinkman number, the magnitude of minimum
∂λ entropy generation is same. A similar scenario is observed
= K0 + K1 λ + K2 λ2 + K3 λ3 (24) for the isoflux boundary condition except for the magnitude
 S. Mahmud, R.A. Fraser / International Journal of Thermal Sciences 42 (2003) 513–521 521

of minimum average entropy generation rate ([NSq o

]min ). Canadian Department of Foreign Affairs and International
[NSq ]min can be calculated using the following equation
o Trade’ for the scholarship and the financial support to
  conduct the present research.
2Π
o
NSq = (26)
Π +1
min

Fig. 14 shows the combined distributions of [NST o ]

min and References
[NSq ]min with radius ratio Π . At the two extremes, Π → 1
o

and Π → ∞, [NST o ]
min shows an asymptotic nature. In [1] D.M. Maron, S. Cohen, Hydrodynamics and heat/mass transfer near
contrast, [NSq ]min shows an asymptotic nature only when
o rotating surfaces, Adv. Heat Transfer 21 (1991) 141–183.
Π → ∞. At Π = 1.763, minimum entropy generation for [2] G.I. Taylor, Distribution of velocity and temperature between con-
centric rotating cylinders, Proc. Roy Soc. London Ser. A 151 (1935)
both boundary conditions are equal at 1.2761. At the upper
494–512.
extreme of Π (Π → ∞), minimum entropy generation rate [3] K.N. Astill, Studies of the developing flow between concentric
approaches 0 for the isothermal case and 2 for the isoflux cylinders with the inner cylinder rotating, J. Heat Transfer 86 (1964)
case. 383–392.
[4] C.D. Andereck, S.S. Lui, H.L. Swinney, Flow regimes in a circu-
lar Couette systems with independently rotating cylinder, J. Fluid
4. Conclusions Mech. 164 (1983) 155–183.
[5] I.S. Bjorklund, W.H. Kaye, Heat transfer between concentric rotating
cylinders, J. Heat Transfer 81 (1959) 175–186.
The First and Second Laws of Thermodynamics are
[6] F. Tachibana, S. Fukui, H. Mitsumura, Heat transfer in an annulus with
applied to forced convection inside a cylindrical annular an inner rotating cylinder, Bull. JSME 8 (1960) 119–123.
space with both isothermal and isoflux boundary conditions [7] H. Aoki, H. Nohira, H. Arai, Convective heat transfer in an annulus
to obtain general expressions for velocity and temperature with an inner rotating cylinder, Bull. JSME 10 (1967) 523–532.
distribution, entropy generation number, and Bejan number. [8] E. Leonardi, J.A. Reizes, G. de Vahl Davis, Heat transfer in a vertical
Flow inside the annular gap is induced by the relative rotating annulus—a numerical study, in: Proc. 7th Internat. Heat
transfer Conf., Munich, 1982.
rotating motion between the inner and the outer cylinders.
[9] M.A. El-Shaarawi, A. Sarhan, Developing laminar free convection in
Near the inner cylinder, the entropy generation rate is higher an open ended vertical annulus with a rotating inner cylinder, J. Fluid
due to the higher temperature and velocity gradients. The Mech. 102 (1983) 552–558.
entropy generation rate shows asymptotic behavior near the [10] S.R.M. Gardiner, R.H. Sabersky, Heat transfer in an annular gap,
outer cylinder provided that the rotation of the inner cylinder Internat. J. Heat Mass Transfer 21 (1978) 1459–1466.
is higher than the outer. For different group parameters, [11] C.J.R. Gasley, Heat transfer characteristics of the rotating and an axial
NS profiles are nearly parallel to one another. Difference flow between concentric cylinders, J. Heat Transfer 80 (1958) 79–90.
between the magnitudes of two consecutive NS decreases [12] L.A. Dorfman, Heat and mass transfer near rotating surfaces, J. Engrg.
Phys. 22 (1972) 247–257.
very slowly along the radial distance. Velocity ratio affects
[13] P.R.N. Childs, C.A. Long, Review of forced convective heat transfer
entropy generation number significantly. Difference between in stationary and rotating annuli, J. Mech. Engrg. Sci. 210 (1996) 123–
the magnitude of two consecutive NS decreases rapidly 134.
along the radial direction and approaches zero near the outer [14] A. Bejan, Second-law analysis in heat transfer and thermal design,
cylinder. For the isothermal boundary condition, the Bejan Adv. Heat Transfer 15 (1982) 1–58.
number is maximum at the inner cylinder. The Bejan number [15] A. Bejan, Entropy Generation Minimization, CRC, Boca Raton, NY,
1996.
decreases along the radial direction showing a minimum
[16] A. Bejan, A study of entropy generation in fundamental convective
and then increase towards the outer cylinder as long as heat transfer, J. Heat Transfer 101 (1979) 718–725.
Br/Ω > 0. The magnitude of the Bejan number is zero at [17] B.S. Yilbas, Entropy analysis of concentric annuli with rotating outer
its minimum location which can be determined by using the cylinder, Exergy Internat. J. 1 (2001) 60–66.
expression given in Eq. (21). The minimum value of average [18] M.K. Drost, J.R. Zaworski, A review of second law analysis tech-
entropy generation rate depends only on radius ratio and it niques applicable to basic thermal science research, ASME AES 4
occurs when numerical value of λ is equal to the numerical (1988) 7–12.
value of Π . [19] F.M. White, Viscous Fluid Flow, McGraw-Hill, New York, 1974.
[20] S. Paoletti, F. Rispoli, E. Sciubba, Calculation of exergetic losses in
compact heat exchanger passages, ASME AES 10 (1989) 21–29.
[21] A. Bejan, Shape and Structure, from Engineering to Nature, Cam-
Acknowledgements bridge University Press, Cambridge, 2000.
[22] L.G. Leal, Laminar Flow and Convective Transport Process: Scaling
One of the authors (Shohel Mahmud) wishes to thank Principles and Asymptotic Analysis, Butterworth–Heinemann, Stone-
‘The Canadian Commonwealth Scholarship Program of the ham, 1992.