EJMFLU

ARTICLE IN PRESS
JID:EJMFLU AID:2359 /FLA [m5G; v 1.65; Prn:14/10/2008; 12:29] P.1 (1-9)

European Journal of Mechanics B/Fluids ••• (••••) •••–•••
1 67
Contents lists available at ScienceDirect
2 68
3 69
4 European Journal of Mechanics B/Fluids 70
5 71
6 72
7 www.elsevier.com/locate/ejmflu 73
8 74
9 75
10 76
11 77
Transport of reactive solutes in unsteady annular flow subject to wall reactions
F
12 78
13 79
14 Suvadip Paul, B.S. Mazumder ∗ 80
O
15 81
Physics and Applied Mathematics Unit, Indian Statistical Institute, Kolkata 700 108, India
16 82
17 83
RO
18 a r t i c l e i n f o a b s t r a c t 84
19 85
20 Article history: The present paper concerns with the dispersion process in steady and oscillatory flows through an 86
21 Received 20 May 2008 annular pipe in presence of reversible and irreversible reactions at the wall. Method of homogenization, 87
22 Received in revised form 10 September a multiple-scale method of averaging, is adopted for deriving the effective transport equations. The main 88
2008
23
24
25
26
Accepted 29 September 2008
Keywords:
Pulsatile flow
DP
objective is to look into the effect of aspect ratio of the annular pipe on the dispersion coefficient
due to the combined effect of axial convection and radial diffusion in steady and oscillatory flows
along the annulus, subject to the kinetic reversible phase exchange and irreversible absorption at the
outer wall. Results demonstrate that upto a certain critical value of aspect ratio, dispersion coefficient
increases with increase of aspect ratio when the wall is retentive, though the wall inertness may lead
89
90
91
92
27 93
Aspect ratio to decrease of dispersion coefficient with increase of aspect ratio. The results would be useful to the
28 94
Phase exchange medical practitioners working in the domain of catheterized artery.
TE
29 95
Homogenization © 2008 Elsevier Masson SAS. All rights reserved.
30 Dispersion coefficient 96
31 Reaction 97
Catheter
32 98
33 99
EC
34 100
35 101
1. Introduction (Sarkar and Jayaraman [9], Mazumder and Mondal [10]) were also
36 102
considered. Dispersion coefficient affected by a reversible phase
37 103
The study of longitudinal dispersion of tracers through an annu- exchange has been studied by Davidson and Schroter [11]; Phillips
38 104
lar straight tube is of considerable interest due to its applications and Kaye [12]. But very few studies includes the effect of both re-
RR
39 105
in the field of chemical, environmental and biomedical engineer- versible and irreversible reactions. Though Purnama [13], Revelli
40 106
ing. The first fundamental study on dispersion was that of Tay- and Ridolfi [14,15] and Ng [16,17] carried out this effect for flow
41 107
42 lor [1] who showed that the dominant mechanism, whereby a through a tube or open-channel, but to the best of our knowl- 108
43 scalar contaminant cloud (or solute) is spread in the steady lam- edge no attempt has been made to examine dispersion phenom- 109
44 inar flow within straight tube, is dispersion – the interaction be- ena through an annular tube considering both reversible and irre- 110
CO
45 tween non-uniform velocity and diffusion across the flow. Aris [2] versible wall reactions. 111
46 extended Taylor’s theory to include longitudinal diffusion and de- Effect of aspect ratio on dispersion was studied by Mondal and 112
47 veloped an approach ‘method of moments’ to analyze the asymp- Mazumder [18], Mazumder and Mondal [10], Sarkar and Jayara- 113
48 totic behaviour of second order moment about the mean. man [9,19] and others. These studies reveal that aspect ratio of 114
49 An exact solution of the diffusion equation was obtained by annular pipe has important contribution in dispersion process. But 115
in all these cases, only the effect of irreversible reaction at the
UN
50 Chatwin [3] to study the dispersion in oscillatory flow. Watson [4] 116
51 used the concept proposed by Chatwin to study the passive con- boundary is considered, though the model demands to have ap- 117
52 taminant dispersion in an oscillatory pressure-driven flow. Jimenez plication in catheterized artery where the reversible reaction also 118
53 and Sullivan [5] used a probabilistic model to study the stream plays important role. 119
54 wise dispersion in unsteady laminar flow. Pedley and Kamm [6] The main objective of the present paper is to examine the in- 120
55 studied the axial mass transport in an annular region in presence fluence of annularity on the transport process under the combined 121
56 of an oscillatory flow field. effects of reversible and irreversible wall reactions, when the flow 122
57 There exist a large number of studies on Taylor dispersion un- is driven by a pressure gradient comprising of steady and peri- 123
58 der the sole influence of irreversible reaction. Some of them con- odic components. The inner wall of the outer tube is lined with 124
59 sidered channel flow (Mondal and Mazumder [7]) or tube flow a very thin layer made up of a retentive and reactive materi- 125
60 (Jiang and Grotberg [8]) while in few cases annular tube flows als. The substance that undergoes Taylor dispersion in the fluid 126
61 is subject to reversible phase exchange and irreversible absorption. 127
62 These boundary reactions are the most important processes con- 128
63 * Corresponding author. trolling the dispersion of solute in catheterized artery. A relatively 129
64 E-mail addresses: bsm46@yahoo.com, bijoy@isical.ac.in (B.S. Mazumder). fast rate of phase exchange and a relatively slow rate of absorption 130
65 131
0997-7546/$ – see front matter © 2008 Elsevier Masson SAS. All rights reserved.
66 doi:10.1016/j.euromechflu.2008.09.003 132
Please cite this article in press as: S. Paul, B.S. Mazumder, Transport of reactive solutes in unsteady annular flow subject to wall reactions, European Journal of Mechanics
B/Fluids (2008), doi:10.1016/j.euromechflu.2008.09.003
ARTICLE IN PRESS
2 S. Paul, B.S. Mazumder / European Journal of Mechanics B/Fluids ••• (••••) •••–•••
1 were considered. Results are shown how the spreading of tracers Here I 0 and K 0 are respectively the modified Bessel function of 67
2 is influenced by the aspect ratio, phase exchange parameter and first kind and second kind of order zero with imaginary arguments. 68
3 absorption. The function I 0 and K 0 can be expressed as I 0 (ri 1/2 ) = ber(r ) + 69
4 The insertion of catheter into an artery leads to a formation of i bei(r ) and K 0 (ri 1/2 ) = ker(r ) + i kei(r ) where ber, bei, ker and kei 70
5 an annular region between the catheter and the arterial wall (Mc- are Kelvins functions of order zero. 71
6 Donald [20]). The fact that the lung and blood vessels have con- For flow through a tube (i.e., when b = 0), the steady and un- 72
7 ductive walls (where phase exchange between the arterial wall and steady components of the velocity are given by 73
8 the flowing fluid (blood) and reaction with a reactant secreted by 2 74
r P a2
9 the wall tissue may take place) and the catheter, of course, acts as u s (r ) = 2u s 1 − with u s = 75
10 an impermeable boundary is reflected in our boundary conditions. a 8ν 76
11 The model will help us in understanding the indicator technique and 77

F
12 and other mechanisms in the branchial region. The insertion of a 1− i 78
iPψ J 0( δ r)
13 pipe with smaller diameter at the centerline of an artery brings u w (r ) = − 1− . 79
14 the asymmetry to the flow, and the increase of aspect ratio leads ω J 0 ( 1− i
δ a) 80
O
15 to the existence of symmetry of the annular flow. 81
16 3. Governing equation and boundary conditions 82
17 2. Velocity distribution 83
RO
18 Let us consider the transport of a chemical species through the 84
19 We consider a fully developed, axi-symmetric laminar flow of annular gap of a tube. If the species is completely miscible with 85
20 a homogeneous, incompressible viscous fluid through an annular the fluid and C (x, r , t ) is the concentration (mass of species dis- 86
21 pipe having inner radius b and outer radius a (i.e., a > b). We have solved per bulk volume of the fluid) of the mobile phase, then C 87
22 used a cylindrical coordinate system in which the radial and axial satisfies the mass transport equation as: 88

23
24
25
co-ordinates are r and x respectively. The flow is assumed to be
unidirectional and so the velocity has only axial component u (r , t )
which satisfies the Navier–Stokes equation as:

DP ∂C
∂t
+ u (r , t )
∂C
∂x
∂2C
=D 2 +
∂x
D ∂
r ∂r
r
∂C
∂r
along with the boundary conditions,
, b<r <a (4)
89
90
91
26 92
∂u 1 ∂p 1 ∂ ∂u
27 =− +ν r ∂C 93
28
∂t ρ ∂x r ∂r ∂r = 0, r = b, (5) 94
∂r
where ∂ p /∂ x is the axial pressure gradient, ρ is the fluid density
TE
29 ∂C ∂Cs 95
30 and ν is the kinematic viscosity. −D −ΓC = = k(α C − C s ), r = a, (6) 96
∂r ∂t
31 The horizontal pressure gradient, which drives the flow, consists 97
of steady and harmonically fluctuating components, where D is the molecular diffusion coefficient assumed to be con-
32 98
stant and C s (x, t ) is the concentration (mass of species retained
33 1 ∂p
99
− = P 1 + ψ Re e i ωt per unit surface area of the wall) of the immobile phase. Here Γ ,
EC
34 100
ρ ∂x k and α are the irreversible absorption rate, the reversible reaction
35 101
where P > 0 is the steady part of the pressure gradient, ψ is a rate and the partition coefficient respectively.
36 102
factor such that P ψ is the amplitude of the oscillatory part of the The boundary condition (5) states that there is no net transport
37 103
pressure gradient. of mass through the inner wall of the annular pipe. Last equality
38 104
The no-slip conditions on the surface of the inner and outer of the boundary condition (6) states that the rate of accumulation
RR
39 105
walls of the annular pipe i.e. u (b, t ) = 0 and u (a, t ) = 0, will pro- of the immobile phase in the outer wall is linearly proportional
40 106
duce the following velocity profile, to the departure from local equilibrium between concentrations of
41 107
i ωt
the two phases on the outer wall of the annulus, and the first
42 u (r , t ) = u s (r ) + Re u w (r )e (1) equality simply describes the irreversible reaction occurring at the
108
43 109
where the steady component (u s (r )) is given by, surface of the outer wall.
44 110
CO
2
45 r 1 − (b/a)2 r 4. Assumptions
111
46 u s (r ) = K u s 1 − − log (2) 112
a log(b/a) a
47 113
The following assumptions are made for carrying out the per-
48 where u s is the velocity averaged over the annular section given 114
turbation analysis,
49 by 115
UN
50 2 116
Pa 2 1. The length scale for the longitudinal spreading of the chemical
51 u s = and K= 117
4ν K 1 + (b/a)2 + (1 − (b/a)2 )/ log(b/a) cloud is much greater than the annular gap. It is meant that
52
x = O( L ) and r = O(a − b), where L is a characteristic longitu- 118
53 and the unsteady component (u w (r )) of the velocity is 119
dinal distance for the chemical transport. The ratio
54 1−i 1−i iPψ 120
55 u w (r ) = c 1 K 0 r + c2 I 0 r − (3)
= (a − b)/ L 1 (7) 121
δ δ ω
56 is small enough to use as ordering parameter. 122
57 where 2. The oscillation period of the flow is so short that within this 123
1− i 1− i
58
iPψ I 0 ( δ b ) − I 0 ( δ a) period there are no appreciable transport effects along the 124
59 c1 = , tube, though the effect of radial diffusion is not negligible. But 125
60
ω K 0 ( δ a ) I 0 ( 1−
1− i i 1− i 1− i
δ b ) − K 0 ( δ b ) I 0 ( δ a) 126
the annular gap of the tube is so fine that diffusion across the

61 iPψ K 0 ( 1− i 1− i
δ b ) − K 0 ( δ a) entire annular section may be accomplished within this short 127
62 c2 = 1− i 1− i 1− i 1− i time scale. 128
ω I 0 ( δ a) K 0 ( δ b ) − I 0 ( δ b ) K 0 ( δ a)
63 3. The two reactions are of different orders. The reversible phase 129
64 and exchange is much faster than the irreversible reaction. This 130

65
2ν ensures that local equilibrium can be largely achieved over a 131
66 δ= . finite number of oscillations. 132
ω
ARTICLE IN PRESS
S. Paul, B.S. Mazumder / European Journal of Mechanics B/Fluids ••• (••••) •••–••• 3

1 4. The rate of absorption is much slower than the rate of phase D ∂ ∂ C (0) 67
0= r , b < r < a, (19)
2 exchange and the former is comparable with the advection r ∂r ∂r 68
3 speed down the tube. ∂ C (0)
69
4 5. The Peclet number is equal to or greater than order of unity: = 0, r = b, (20) 70
5
∂r 71
6 Pe ≡ au s / D O(1). (8) ∂ C (0) (0) (0) 72
−D = k αC − Cs , r = a. (21)
7 ∂r 73
8 Under these assumptions, three distinct time scales may be de- Eqs. (19) and (20) obviously imply that the leading order con- 74
9 fined as (Ng [16]), centration is independent of r, i.e., 75
10 76
C (0) = C (0) (x, t 1 , t 2 ).
11 T 0 = 2π /ω = O (a − b)2 / D = O k−1 , (9) (22) 77

F
12 78
−1 The boundary condition (21) then gives
13 T 1 = L /u s = O (a − b)Γ = T 0 /
, (10) 79
(0)
14 2
T 2 = L / D = T 0 /
. 2 Cs = α C (0) . (23) 80
O
(11)
15 81
16 As expected, at the leading order the mobile phase of the chem- 82
Based on these time scales, we may introduce accordingly
17 ical is at local equilibrium with the immobile phase. 83
RO
18 t0 = t , t1 =
t , t2 =
2t (12) For first order (O(
)), Eqs. (13)–(15) give 84
19 85
which are, respectively, the fast, medium and slow time variables. ∂ C (0) ∂ C (1) ∂ C (0) D ∂ ∂ C (1)
20 + +u = r , b < r < a, (24) 86
21 ∂ t1 ∂ t0 ∂x r ∂r ∂r 87
22 5. Asymptotic analysis ∂ C (1) 88
= 0, r = b, (25)
23
24
25
26
The relative significance of the terms in the transport equation
with the boundary conditions (4)–(6) are indicated below with the
power of
:
DP −D
∂r
∂ C (1)
∂r
− Γ C (0) =
∂Cs
∂ t1
+
∂Cs
∂ t0

(0)
(1)
(1)
= k α C (1) − C s , r = a. (26)
89
90
91
92
27 Averaging Eqs. (24)–(26) w.r.t. the fast time variable t 0 , we get 93

∂2C (1)
28 ∂C ∂C D ∂ ∂C 94
+
u =
2 D 2 + r , b < r < a, (13) ∂ C (0) ∂ C (0) D ∂ ∂ C̄
TE
29 95
∂t ∂x ∂x r ∂r ∂r + us = r , b < r < a, (27)
30 ∂ t1 ∂x r ∂r ∂r 96
∂C
31 = 0, r = b, (14) ∂ C̄ (1)
97
32 ∂r = 0, r =b (28) 98
33 ∂C ∂Cs ∂r 99
−D −
Γ C = = k(α C − C s ), r = a. (15)
EC
34 ∂r ∂t and 100
35 101
Following the asymptotic expansion introduced by Fife and Nic- ∂ C̄ (1) ∂Cs
(0)
(1)
36 −D − Γ C (0) = = k α C̄ (1) − C̄ s , r = a, (29) 102
holes [21], the concentrations C and C s are expressed as, ∂r ∂ t1
37 103
38 where the overbar denotes time average over one period of oscil- 104
C (x, r , t ) = C (0) (x, r , t 1 , t 2 ) +
C (1) (x, r , t 0 , t 1 , t 2 )
RR
39 lation and u s is the steady velocity component. We further take 105

40 2 (2)
3 cross-sectional average of (27) subject to the conditions (23), (28) 106
+
C (x, r , t 0 , t 1 , t 2 ) + O
, (16)
41 and (29): 107
(0) (1)
42 C s (x, t ) = C s (x, t 1 , t 2 ) +
C s (x, t 0 , t 1 , t 2 ) 108
3 ∂ C (0) u s ∂ C (0) 2aΓ
43
(2) + + C (0) = 0 (30) 109
44 +
2 C s (x, t 0 , t 1 , t 2 ) + O
(17) ∂ t1 R ∂x R (a2 − b2 ) 110
CO
45 111
and for the multiple-scale asymptotic analysis, the time derivative where
46 112
has been expanded as, 2aα
47 R =1+ (31) 113
48 ∂ ∂ ∂ ∂ a2 − b 2 114
49 = +
+
2 . (18) 115
∂t ∂ t0 ∂ t1 ∂ t2 is the retardation factor resulting from the reversible partitioning
UN
50 of the chemical into two phases. It is the ratio, when in equilib- 116
51 Here the terms involving a time scale T 1 or larger can be la- rium, of the total mass of the two chemical phases to the mass of 117
52 beled as ‘developed’, while those terms that vanish exponentially the mobile phase per unit length of the annulus. 118
53 as t → ∞ with relaxation time proportional to T 0 are called ‘tran- Using (23) in (26) and eliminating ∂ C (0) /∂ t 1 from (24)–(26) 119
54 sient’ terms. Therefore, transient terms can be neglected for a time and (30), we rewrite Eqs. (24)–(26) as: 120
55 that is a few time larger than T 0 . Since we are interested in a 121

56 time scale equal to or larger than T 1 , we may accordingly ignore ∂ C (1) u s ∂ C (0) 2aΓ 122
all the transient effects. The expansions (16) and (17) contain only + u− − C (0)
57 ∂ t0 R ∂x R (a2 − b2 ) 123
58 the so-called developed concentrations. The terms C (n) and C s
(n) 124
D ∂ ∂ C (1)
59 (n = 1, 2, 3, . . .) are purely oscillatory function of short time vari- = r , b < r < a, (32) 125
60 able t 0 but the oscillatory effect does not show up on the zeroth r ∂r ∂r 126
61 order (i.e., n = 0), and therefore the leading order concentrations ∂ C (1) 127
= 0, r = b, (33)
62 are taken to be independent of the time variable t 0 . ∂r 128
63 Using the expansions (16)–(18) in Eqs. (13)–(15) and equating (1) 129
∂ C (1) u s ∂ C (0) Γ (0) ∂ C s
64 the coefficients of like powers of
from both sides, a system of −D = −α + C + 130
65 differentials equations is obtained.
∂r R ∂x R ∂ t0 131
66 For zeroth order (O(1)) Eqs. (13)–(15) give = k(α C (1) − C s(1) ) + Γ C (0) , r = a. (34) 132
ARTICLE IN PRESS

1 Eqs. (32)–(34) suggest that C (1) and C s are linearly propor-
(1) 1 2 K 2 λ4 67
− r̂ − λ2 − λ −
2 tional to ∂ C (0) /∂ x and C (0) . Accordingly, one can express these first 4R 2 2 68
2 2
(1)
3 order concentrations C (1) and C s into a sum of two terms: the 1−λ λ r̂
69
4 first term is linearly proportional to ∂ C (0) /∂ x with a coefficient − λ2 log(λ) − log 70
5
log(λ) 2 λ 71
containing steady and oscillatory components, while the second
6 term is linearly proportional to C (0) with a steady coefficient: 1 2 r̂ 72
+ λ log , (46)
7 2R λ 73
8 (1)
i ωt 0
∂ C (0) (0) 74
C = N (r ) + Re B (r )e + M (r )C (35) r̂ 2 − λ2 λ2 r̂
9 ∂x M̂ (r̂ ) = M̂ (λ) − + log , (47) 75
10 2(1 − λ2 ) R (1 − λ2 ) R λ 76
and
11 α̂ 77
∂ C (0) N̂ s = α̂ N̂ (1) + (48)
F
(1)
12
Cs = N s + Re B s e i ωt0 + M s C (0) (36) Da R 78
13 ∂x 79
14 and 80
where the coefficients N (r ), N s , M (r ), M s , B (r ) and B s satisfy the
O
15 boundary value problems given below. 2 81
M̂ s = α̂ M̂ (1) + (49)
16 Substituting (35) and (36) into (32)–(34), and equating the Da (1 − λ2 ) R 82
17 steady term of the coefficient of ∂ C (0) /∂ x, we find the equation 83
RO
18 for N (r ) as: where N̂ (λ) and M̂ (λ) are undetermined constants unless a 84
19 uniqueness condition is specified. 85
D d dN u s Here D a , Γˆ and α̂ are respectively the Damkohler number (rep-
20 r = us − , b<r <a (37) 86
21 r dr dr R resenting the kinetics of the phase exchange at the outer wall), 87
22 absorption parameter (representing the rate of loss on the outer 88
with the boundary conditions
23
24
25
dN
dr
= 0, r = b, (38)
DP wall) and phase partition ratio (retention parameter).
The dimensionless form of the boundary value problem (43)–
(45) is
89
90
91
26 dN u s 92
27
−D = −α = k(α N − N s ), r = a. (39) 1 d d B̂ 93
2
dr R r̂ + η̂ B̂ = û w , λ < r̂ < 1, (50)
28 r̂ dr̂ dr̂ 94
Again equating the steady term associated with C (0) , we
TE
29 95
get M (r ) and M s : d B̂
30 = 0, r̂ = λ, (51) 96
31
dr̂ 97
D d dM 2aΓ
32 r =− , b<r <a (40) d B̂ 98
r dr dr R (a2 − b2 ) 2
= η̂ B̂ s = − D a (α̂ B̂ − B̂ s ), r̂ = 1, (52)
33 dr̂ 99
EC
34 with the boundary conditions 100

35 where B̂, B̂ s , û w and η̂ are the dimensionless parameters defined 101
dM as
36 = 0, r = b, (41) 102
dr
37 B Bs uw 103
dM Γ B̂ = , B̂ s = , û w = , η̂ = aη,
38 −D = = k(α M − M s ) + Γ, r = a. (42) u s a2 / D u s a3 / D u s 104
dr R
RR
39 105
40 Similarly, we obtain the following equations for the complex with 106
√
41 functions B (r ) and B s : 2i Sc ν δ 2 ω 107

42 η2 = − , Sc = , δ̂ = = , Λ=a . 108
D d dB a2 δ̂ 2 D a Λ ν
43 r = iω B + u w , b<r <a (43) 109
44 r dr dr Because of the complex structure of û w (r̂ ) in Eq. (50), central 110
CO
45 with the boundary conditions, difference scheme has been adopted to solve the equation under 111
46 the prescribed boundary conditions. 112
dB Here Sc is the Schmidt number which is the ratio of viscous
47
= 0, r = b, (44) 113
48 dr diffusion (ν ) to the molecular diffusion (D) and δ̂ is the ratio of 114
√
49 dB the Stokes boundary thickness ( 2ν /ω ) to the outer radius of 115
−D = i ω B s = k(α B − B s ), r = a. (45)
UN
50 dr the annular tube (a). It should be noted that δ̂ is inversely pro- 116
51
Introducing the dimensionless quantities portional to the dimensionless frequency parameter or Womersley 117
52 number Λ which is the product of two dimensionless param- 118
53 r b N Ns eters, the Reynolds number (u s a/ν ) and the Strouhal number 119
r̂ = , λ= , N̂ = , N̂ s = ,
54 a a u s a2 / D u s a3 / D (ωa/u s ). The parameter Λ2 is a measure of the ratio of the time 120
55 M Ms (a2 /ν ) required for viscosity to smooth out the transverse varia- 121
M̂ = , M̂ s = , tion in vorticity to the period of oscillation (1/ω). The number δ̂
Γˆ Γˆ a
56 122
57
2 increases with the oscillation period. Therefore, the higher the fre- 123
ka α Γa
58 Da = , α̂ = , Γˆ = quency, the smaller the value of δ̂ . The Schmidt number Sc is of 124
59 D a D order unity when the fluid is a gas, and is much greater than unity 125
60 the solutions (in dimensionless form) of Eqs. (37) and (40) subject when the fluid is a liquid. Keeping in mind the possible application 126
61 to the prescribed boundary conditions are given by, of this model in catheterized artery, some investigations are made 127
62 taking Sc = 103 . For dispersion to be significant, one may expect 128
1 4 K
63
N̂ (r̂ ) = N̂ (λ) + r̂ − λ − r̂ − λ4 2 2 that, δ̂ O(1) or the period of oscillation must be long enough for 129
64 4 4 the Stokes boundary-layer thickness to be at least comparable with 130

65 1 − λ2 2 the annular gap of the tube. 131
66 − r̂ log(λ) − λ2 log(λ) − r̂ 2 − λ2 For second order (O(
2 )), (13)–(15) give 132
log(λ)
ARTICLE IN PRESS
1 ∂ C (0) ∂ C (1) ∂ C (2) ∂ C (1) + 708λ6 R log(λ) D a − 36 log(λ) D a + 144R λ6 log(λ)2 67

+ + +u
2 ∂ t2 ∂ t1 ∂ t0 ∂x 2 2 2 4 8 2
68
3 − 144R log(λ) λ + 144R log(λ) λ − 99D a λ R 69
∂ 2 C (0) D ∂ ∂ C (2)
4 =D + r , b < r < a, (53) + 162D a λ8 R + 648λ4 R D a + 180λ6 D a − 216λ4 D a + 108λ2 D a 70
∂ x2 r ∂r ∂r
5 71
6 ∂ C (2) − 372R λ2 log(λ) D a + 72D a log(λ)3 λ8 + 72D a log(λ)3 λ4 72
= 0, r =b (54)
7 ∂r 3 8
− 18D a − 144D a log(λ) λ R + 36D a log(λ) λ 2 2 73
8 74
and
9 − 144D a log(λ)3 λ6 R − 219D a R 2 log(λ)2 λ8 75
(0) (1) (2)
10 ∂ C (2) ∂Cs ∂Cs ∂Cs 8 2 4 2 2 76
11
−D − Γ C (1) = + + + 420D a λ R log(λ) + 108D a λ R log(λ)
77
∂r ∂ t2 ∂ t1 ∂ t0
(2) (2) + 108D a λ4 R 2 log(λ) − 144R log(λ)2 + 48D a R log(λ)2
F
12 78
= k α C − C s , r = a. (55)
13 79
2 2 8 2 2
14 Averaging Eq. (53) w.r.t time followed by space subject to
− 33D a R log(λ) − 198D a λ log(λ) − 76D a R log(λ) 80
O
15 the boundary conditions (54) and (55) and using Eqs. (30), (35) + 144D a log(λ)3 λ6 + 180D a λ8 log(λ) + 72D a log(λ)3 λ8 R 2 81
16 and (36) in appropriate places one gets, 82
17 − 60D a R log(λ)2 λ2 − 324D a R log(λ)2 λ4 − 438D a λ8 R log(λ) 83
RO
18 ∂ C (0) u s ∂ C (0) 2aΓ 84
+ζ +χ C (0) − 464D a λ6 R 2 log(λ) + 184D a λ2 R 2 log(λ)
19 ∂ t2 R ∂x R (a2 − b2 ) 85
20 2 (0) + 248D a λ8 R 2 log(λ) + 54R D a + 144D a log(λ)2 λ6 R 2 86
D ∂ C
21 = + DTs + DT w (56) 87
22
R ∂ x2 − 324R λ2 D a − 144λ4 log(λ)2 + 144λ2 log(λ)2 88
23
24
25
26
with

ζ =− M
1
R
−
us
u s

−
2aM s
R (a2 − b2 )

DP − 84 log(λ)2 λ6 R D a − 144λ6 log(λ)2 + 144 log(λ)2
+ 288 log(λ) + 288R log(λ)λ4 − 288λ6 R log(λ)
+ 288R λ2 log(λ) − 45R 2 D a + 288λ6 log(λ) + 234R 2 λ2 D a
89
90
91
92
27 2aΓ N 2aN s 93
− + − N (a ) , (57)
− 432R 2 λ4 D a + 342R 2 λ6 D a − 144λ6 − 288 log(λ) R
28 u s (a2 − b2 ) R R (a2 − b2 ) 94
TE
29 M 2aM s 2 4
+ 216 log(λ) λ D a − 36 log(λ) λ D a − 288 log(λ)λ 2 6 4 95
30 χ =− − + M (a), (58) 96
R R (a2 − b2 )
31 − 432λ2 + 432λ4 − 288λ2 log(λ) + 432R λ2 − 432λ4 R 97
1 u s 2au s N s
32 DTs = N − us + 2 2 , (59) + 144λ6 R 144R 3 − log(λ)2 − λ2 log(λ)2 − 2 log(λ)
98
33 R R R (a − b2 ) 99
EC
34 and + λ4 log(λ)2 + 2 log(λ)λ4 − 1 + 3λ2 − 3λ4 + 2λ2 log(λ) 100

35 101
36 DT w = −
1
Reu w B ∗ (60)
+ λ6 log(λ)2 − 2λ6 log(λ) + λ6 D a . 102
37 2R 103
When λ → 0, the dimensionless dispersion coefficient due to
38 where the asterisk denotes the complex conjugate. steady flow reduces to 104
RR
39 Here ζ and χ are the higher order correction factors to the 105
40 advection velocity and decay rate respectively. D T s and D T w are 1 11 1 1 2α̂ 106
lim D̂ T s = − + + with R = 1 + 2α̂
41 the respective dispersion coefficients due to steady and oscillatory λ→0 R 48 3R 8R 2 R 3 Da 107
42 part of the fluid motion and these are positive definite. 108
43 Introducing dimensionless quantities, which agrees with Ng [16]. 109
44 The dispersion coefficient D̂ T w due to oscillatory flow has been 110
CO
45
( D̂ T s , D̂ T w ) = ( D T s , D T w )/ u s 2 a2 / D (60a) evaluated numerically from the expression (60a) through (60) with 111
46
the known values of B̂ (r̂ ) and û w (r̂ ). Like steady component, os- 112
and using N̂ (r̂ ), N̂ s , M̂ (r̂ ) and M̂ s from Eqs. (46)–(49), expres- cillatory component of the dispersion coefficient D̂ T w also agrees
47 113
sions (57)–(59) can be written in dimensionless form as, well with Ng [16] when λ → 0.
48 114
It is mentioned here that two different normalization scales
49 ζ = −Γˆ −48 + 48R − 8 log(λ) R D a + 42λ6 log(λ) D a 115
may be used for the dispersion coefficient due to oscillatory flow –
UN
50 116
51 − 18λ2 log(λ) D a − 30λ4 log(λ) D a + 27λ6 R D a − 46λ6 R log(λ) D a (i) when the amplitude of the oscillatory pressure gradient is kept 117
constant and the frequency is varied, in which a direct comparison
52
+ 6 log(λ) D a − 63λ4 R D a − 18λ6 D a + 42λ4 D a − 30λ2 D a between D T s and D T w can be possible, (ii) the flow can be under
118
53 119
54 + 36R log(λ)λ4 D a + 18R λ2 log(λ) D a + 6D a − 9R D a + 45R λ2 D a a volume-cycled oscillation such that the tidal or Stroke volume 120
55 2 6 4
is maintained constant when the frequency is varied in oscillatory 121
+ 24 log(λ) λ R D a − 48 log(λ) − 48R log(λ)λ + 48 log(λ) R pumping.
56 122
57 − 24 log(λ)2 λ4 D a − 24 log(λ)2 λ6 D a + 48 log(λ)λ4 + 96λ2 For the second case, it is desirable to normalize D T w as: 123

58
− 48λ4 − 96R λ2 + 48λ4 R 12R 2 D a log(λ) − λ2 log(λ)λ4 Ď T w = D T w / D V T2 /a6 (60b) 124
59 125
60 − log(λ) + λ6 log(λ) + 1 − 3λ2 + 3λ4 − λ6 , where V T (= (2π a2 /ω)|u w |) is the tidal volume that can be 126
61 defined equal to twice the amplitude of the volume fluctuation 127
χ = −3D a λ4 − 8R + 4D a λ2 − 8λ2 − D a + 8 + 8R λ2
62
2 (Watson [4]). 128
63 + 4λ4 D a log(λ) /4D a R 2 −1 + λ2 , The undetermined constants N̂ (λ) and M̂ (λ) in the expressions 129
64 N̂ (r̂ ) and M̂ (r̂ ) respectively have no effect on the evaluation of ζ , 130
D̂ T s = 144 − 144R + 102 log(λ) R D a − 288λ6 log(λ) D a
65 χ and D̂ T s . One can easily show that all terms involving the un- 131
66 + 144λ2 log(λ) D a − 540λ6 R D a − 18D a log(λ)2 − 54D a λ8 determined constants cancel out. 132
ARTICLE IN PRESS

1 Eliminating ∂ C (0) /∂ t 2 from (53)–(56) and assuming the second 1 u s 2au s U s 2aM s DTs 67
D ΓT s = U + 2 2
− us − M +
2 order concentration of the form R R R (a − b2 ) a2 − b 2 R 68

3
∂ C (0) ∂ 2 C (0) 2aN s u s 69
4 C (2) = U (r ) + Re X (r )e i ωt0 + V (r ) + Y (r ) (61) + ζ N + 2 70
∂x ∂ x2 a − b2 R 2
5 71
and 2aV s 2aΓ
6
+ V − R V (a) + 2 (75) 72
7 a − b2 R 2 (a2 − b2 ) 73
(2) ∂ C (0) ∂ 2
C (0)
8 Cs = U s + Re X s e i ωt0 + [V s + Y s ] (62) and 74
∂x ∂ x2
9 75
1 2aM s DT w
10 where U and V are forced by the steady flow and X and Y by the D ΓT w = − Reu w X ∗ − M + 76
11 oscillatory flow, we find the boundary value problems for U as: 2R a2 − b 2 R 77

F
12 2aY s 2aΓ 78
D d dU u s u s 2aΓ + Y − RY (a) + . (76)
13 r = M us − −ζ − N, b<r <a a2 − b 2 (a2 − b2 ) R 79
14 r dr dr R R R (a2 − b2 ) 80
O
15 (63) It may be mentioned here that D ΓT s and D ΓT s are the higher or- 81
16 der corrections to D T s and D T s respectively due to wall absorption. 82
with the boundary conditions Finally, combining the first (Eq. (30)), second (Eq. (56)) and
17 83
RO
18 dU third order transport equations (not given in the text), we get the 84
19 = 0, r = b, (64) overall effective transport equation (using the time variable t): 85
dr
20
dU u s 2aΓ ∂ C (0) u s ∂ C (0) 2aΓ
86
21 −D − Γ N (r ) = −(α ζ + M s ) − Ns + (1 + ζ ) + (1 + χ ) C (0) 87
dr R R (a2 − b2 ) ∂t R ∂x R (a2 − b2 )
22 88
= k[α U − U s ], r =a (65)
23
24
25
and that for V is
D d

dV

u s

DP
= D/R + D T s + D T w + D T s + D T w Γ Γ
∂ 2 C (0)
∂ x2
.
DΓ
(77)
89
90
91
26 r = N us − + D T s, b<r <a (66) The dimensionless dispersion coefficient D̂ ΓT s (= u 2Tas2 / D ) can 92
s
27 r dr dr R 93
be evaluated from (75) on substituting the values of Û , V̂ , Û s
28 94
with the boundary conditions and V̂ s . Again one can show that the undetermined constants have
TE
29 95
no effects on D̂ ΓT s . The expression for D̂ ΓT s is too lengthy to provide
30 dV 96
= 0, r = b, (67) it in the text. We have employed the symbolic computer alge-
31 dr bra package MAPLE to perform the tedious algebraic manipulations 97
32 dV u s and computations. 98
−D = αDTs − N s = k[α V − V s ], r = a. (68)
33 dr R In the limit when λ → 0, the expression for D̂ ΓT s , after making 99
EC
34
Similarly the function X satisfies the boundary value problem use of R = 1 + 2α̂ , reduces to 100
35 101

36 D d dX 2aΓ Γˆ 61 199 5 3 102
r + iω X = − B + Mu w , b < r < a, (69) lim D̂ T s =
Γ
− + −
37
r dr dr (a2 − b2 ) R λ→0 R2 240 288R 8R 2 16R 3 103
38 104
dX 11 2 3 3 2 4α̂
+ − + − − 1−
RR
39 = 0, r = b, (70) 105
40 dr 24 R 2R 2 R Da R R Da 106
dX 2aΓ which coincides with the expression obtained by Ng [16].
41
−D −Γ B =− B s − i ω X s = k(α X − X s ), r = a (71) 107
42 dr (a2 − b2 ) R The dispersion coefficient D̂ ΓT w (= D ΓT w /(u s 2 a2 / D )) given 108
43
while Y satisfies by (76) can be evaluated upon solving (69)–(74) for X , Y and Y s . 109
44 Again it may be shown that the undetermined constants have no 110
CO
45 D d dY 1 ∗ effects on D ΓT w . Since no closed-form analytical solutions are avail- 111

r = Re(u w B ) + D T w , b < r < a, (72)
46 r dr dr 2 able for X and Y in general, it may be solved numerically by a 112
47 dY second order central finite-difference scheme. But for the present 113
48 = 0, r = b, (73) study we have omitted the discussions on D ΓT w . It is only a higher 114
dr
49 order correction to D̂ T w due to irreversible wall absorption. In fact, 115
dY
−D = α D T w = k(α Y − Y s ), r = a.
UN
50 (74) our main emphasis is on the effect of aspect ratio. 116

dr
51 It should be mentioned here that when the frequency of flow 117
52 Introducing dimensionless quantities oscillation is sufficiently low, its period becomes comparable to the 118
53 advection time-scale. This leads to a considerable simplified model. 119
54 (Û , X̂ ) = (U , X )/ Γˆ u s a2 / D , The steady and time varying component of the dispersion coeffi- 120

55
(Û s , X̂ s ) = (U s , X s )/ Γˆ u s a3 / D , cient resulting from the oscillatory flow can be evaluated. Because 121
56 of the complex form of u w (r ) due to the aspect ratio, the analyt- 122
57 ( V̂ , Ŷ ) = ( V , Y )/ u s 2 a4 / D 2 , ical expressions of transport coefficients are very lengthy and are 123
58 124
( V̂ s , Ŷ s ) = ( V , Y )/ u s 2 a5 / D 2 not presented here. The fast oscillation limit of the transport coef-
59 ficients are also omitted for the interest of space. 125
60 Eqs. (63), (66), (69) and (72) are solved subject to the boundary 126
61 conditions. 6. Results and applications to catheterized artery 127
62 Averaging the transport equation of third order w.r.t. time t 0 128
63 and space and using (35), (36) and (61), (62) in appropriate places The study of mass transport has potential applications to pul- 129
64 and collecting the steady and oscillatory terms associated with monary and cardiovascular flow of living systems (Grotberg [22]). 130
65 ∂ 2 C (0) /∂ x2 , one gets the corresponding dispersion coefficients D ΓT s This paper aims to develop a conception on how catheterization 131
66 and D ΓT w due to steady and oscillatory part of the fluid motion as: may influence the dispersion in physiological systems. With this in 132
ARTICLE IN PRESS
1 view the reversible and irreversible reactions are considered to oc- 67

2 cur simultaneously at the outer boundary (on the walls of blood 68
3 vessel). The catheters are made of inert materials, where no re- 69
4 actions of the blood particles take place. A zero flux boundary 70
5 condition is considered in this study for the inner boundary. The 71
6 results of the present analysis are important in understanding the 72
7 dispersion process through a catheterized artery with a reactive 73
8 and retentive arterial wall. The tube of radius a can be considered 74
9 as a blood vessel and the insertion of another tube of radius b 75
10 (i.e., the catheter) into the blood vessel causes the formation of an 76
11 annular region between the catheter and the arterial walls. Apart 77
F
12 from the pulsatile nature of the blood flow, since lung and blood 78
13 vessels involve conductive walls, to understand the indicator di- 79
14 lution technique and other mechanisms in the bronchial region, 80
O
15 it is important to analyze the dispersion procedure considering 81
16 the wall characteristics. The objective is to provide a correction 82
17 for the catheter induced errors in the measured values based on 83
RO
18 the concept of longitudinal diffusion of tracer materials due to the 84
19 combined action of the convection and diffusion. The parameter λ, 85
20 the ratio of the catheter radius b to the arterial radius a, is varied 86
21 from 0.1 to 0.8 to analyze the effect of the catheter size on dis- 87
22 persion phenomena. The catheterization increases the frictional re- 88
23
24
25
26
sistance to flow through the artery and hence alters the flow field
owing to changes of hemodynamic conditions in the artery. The
result of this analysis may provide a correction for catheter inser-
tion in the light of the behaviour of the dispersion coefficient. The
DP 89
90
91
92
27 fast and slow oscillation limits of the dispersion coefficients due 93
28 to oscillatory flow may be of interest in the area of cardiovascu- 94
TE
29 lar system. In the artificial ventilation of the lung (Grotberg [22]), 95
30 the fast oscillation limits of the dispersion coefficients have signif- 96
31 icant importance. Slow and fast oscillation of the heart is directly Fig. 1. Steady component of dispersion coefficient: (a) against the aspect ratio λ, 97
32 related to the systolic and diastolic pressure of living-body. when the wall is inert (b) against the aspect ratio λ, when the wall is retentive and 98
33 In this paper an asymptotic analysis has been presented for (c) against the retention parameter α̂ for different values of the aspect ratio λ. 99
EC
34 the advection–diffusion transport of a chemical species in a flow 100

35 through an annular tube, where the flow is generated using pul- ert wall compared with the retentive wall. It is interesting to note 101
36 satile pressure gradient with non-zero mean. The species is under- that the presence of catheter with increase in size inhibits the dis- 102
37 went linear reversible and irreversible reactions at the inner wall persion process when the wall is inert, whereas for retentive wall, 103
38 of the outer tube. Analytical expressions are derived for the various increase of catheter size enhances the dispersion coefficient if the 104
RR
39 dispersion coefficients due to steady flow with retention and the retention is weak, though strong retention in the wall may dimin- 105
40 combined effect of retention and absorption at the wall; whereas ish the dispersion coefficient. Fig. 1(c) depicts the variation of D̂ T s 106
41 a numerical scheme has been adopted to compute the dispersion with α̂ for different values of the aspect ratio λ. It is observed from 107
42 coefficient due to oscillatory flow. The behaviour of these compo- the figure that for a fixed aspect ratio λ, D̂ T s increases for small 108
43 nents of the dispersion coefficient under the variation of various values of α̂ within the range 0.2–0.3, reaches a maximum value 109
44 controlling parameters like aspect ratio λ, retention parameter α̂ , and then it decreases with α̂ , because they are much weighted 110
CO
45 oscillation parameter δ̂ (inverse of Womersley number Λ), Schmidt down by the retardation factor R. With the increase of aspect ratio, 111
46 number Sc, Damkohler number D a is described in this section. To the maximum value of D̂ T s decreases. The deviation of D̂ T s from 112
47 relate the present study with arterial blood flow, the Schmidt num- its maximum value increases with the increase of λ i.e., more the 113
48 ber Sc is restricted to the order of 103 . The main emphasis is given value of λ, more sharp the decrease of D̂ T s . 114
49 on how the components of dispersion coefficient change with the The variation of oscillatory component of the normalized dis- 115
persion coefficient D̂ T w with λ for Sc = 103 , δ̂ = 1, D a = 1 is plot-
UN
50 aspect ratio of the annular tube. 116

51 The behaviour of the steady component of normalized disper- ted in Fig. 2(a) for different values of α̂ . It is observed that D̂ T w 117
52 sion coefficient D̂ T s with variation of aspect ratio λ is described increases with the increase of aspect ratio over a wide range (0.0– 118
53 in Fig. 1(a), (b) for different values of α̂ when D a = 1.0. It is ob- 0.85), reaches a maximum value and then within the remaining 119
54 served from the figures that, D̂ T s shows its decrement [Fig. 1(a)] short range of λ it shows a sudden drop. So unless the aspect ratio 120
55 with the increase of aspect ratio when the wall is perfectly inert of the tube exceeds a critical value, D̂ T w increases as the annu- 121
56 (i.e., α̂ = 0), but for retentive wall (α̂ > 0) the dependence of D̂ T s lar gap of the tube decreases (increase of aspect ratio). The rate of 122
57 on λ shows a little irregular behaviour [Fig. 1(b)]. Fig. 1(b) shows increment of D̂ T w with λ depends on the retention parameter α̂ . 123
58 that for weak retention (α̂ = 0.05, 0.10) D̂ T s is growing with λ The rate is high when the retention is weak. Also increase of α̂ 124
59 over the range 0.0–0.8 and then it suddenly drops in the narrow leads to a decrease of D̂ T w (Fig. 2(b)). It is observed from the fig- 125
60 region of the annulus, but when the retention is strong (α̂ > 0.1), ure that for all λ, D̂ T w decreases monotonically with the increase 126
61 D̂ T s decreases with λ over the whole annular gap. So increase or of α̂ . The effects of phase exchange on D̂ T s and D̂ T w are similar. 127
62 decrease of steady component of the dispersion coefficient with λ Stronger kinetics of the phase exchange (i.e. smaller D a ) will give 128
63 depends upon the retention parameter α̂ . For inert wall, the de- rise to a larger value of either dispersion coefficients (Ng [16]). 129
64 crease of D̂ T s with λ may be due to decrease of mean velocity The dispersion coefficient D̂ T w has been plotted in Fig. 3(a) 130
65 with aspect ratio λ. It may be mentioned here that the magnitude against the aspect ratio λ for different values of δ̂ when Sc = 103 , 131
66 of the steady component of dispersion is very less for perfectly in- D a = 1, α̂ = 0. It is observed from the figure that for inert wall 132
ARTICLE IN PRESS
1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
F
12 78
13 79
14 80
O
15 81
16 82
17 83
RO
18 84
19 85
20 86
21 87
22 88
23
24
25
Fig. 2. Oscillatory component of the dispersion coefficient D̂ T w : (a) against the as-
pect ratio λ for different values of the partition coefficient α̂ and (b) against the
retention parameter α̂ for different values of the aspect ratio λ.
DP 89
90
91
26 92
27 93
28 94
TE
29 95
30 96
31 97
32 98
Fig. 4. Variation of oscillatory component of the dispersion coefficient for different
33 values of the aspect ratio λ; against the Schmidt number Sc for δ̂ = 1, (b) dispersion 99
EC
34 coefficient for pressure gradient-cycled oscillation ( D̂ T w ) as functions of δ̂ when 100

35 Sc = 103 and α̂ = 1 and (c) the same as (b), but for dispersion coefficient due to 101
volume-cycled oscillation ( Ď T w ).
36 102
37 103
38 influence of oscillation parameter δ̂ , when it is small, is not that 104
effective on dispersion coefficient for 0 < λ < 0.4 (Fig. 3(b)). How-
RR
39 105
40 ever large oscillation parameter δ̂ has appreciable contribution on 106
41 the oscillatory component of the dispersion coefficient for all λ. 107
42 The behaviour of D̂ T w with the variation of λ is found to be con- 108
43 sistent with the results obtained from Fig. 3(b) even when the wall 109
44 is inert with small δ̂ or when the wall is retentive with large δ̂ . It 110
CO
45 is observed that like steady component, oscillatory component of 111

46 the dispersion coefficient is very less in case of non-retentive wall 112
Fig. 3. Oscillatory component of the dispersion coefficient against the aspect ratio compared to that when the wall is retentive.
47 113
of the annular tube for different values of the oscillation parameter δ̂ : (a) when the
48 wall is inert and (b) when the wall is retentive.
The variation of D̂ T w with Schmidt number Sc is shown in 114
49 Fig. 4(a) for different values of α̂ and λ. It is seen that higher 115
value of Sc leads to smaller D̂ T w for retentive (α̂ = 1) as well
UN
50
(α̂ = 0) and low oscillation parameter (δ̂ ≈ 1), D̂ T w is practically 116
51 as inert (α̂ = 0) wall. As we discussed earlier, based on differ- 117
ineffective with the variation of λ except for a narrow annular
52 ent normalization scales, there exists two dispersion coefficients: 118
gap. Figure shows that for perfectly inert wall, D̂ T w may increase
53 D̂ T w (given by (60a)) and Ď T w (given by (60b)). The former is 119
with λ over a considerable range when the oscillation parame-
54 for a pressure gradient-cycled oscillation, while the latter is for 120
ter δ̂ = 4. When δ̂ is sufficiently large, it is observed that D̂ T w
55 a volume-cycled oscillation. As shown in Fig. 4(b), (c), the oscil- 121
56
decreases with λ throughout its domain and it agrees well with lation frequency have qualitatively different effects on these two 122
57
the results of Sarkar and Jayaraman [9] and Mazumder and Mon- normalized dispersion coefficients. While D̂ T w increases monoton- 123
58 dal [10]. For fixed but small λ, D̂ T w increases with increase of δ̂ . ically with δ̂ (i.e., the increase of Womersley number Λ inhabits 124
59 As the annular gap becomes narrower, dispersion coefficient de- the dispersion process (Mazumder and Mondal [10])), the oppo- 125
60 creases with λ and apparently ineffective with the variation of δ̂ . site is true for Ď T w . Therefore under a volume-cycled oscillation, 126
61 Fig. 3(b) shows the variation of D̂ T w with λ for different values a larger dispersion coefficient will result from a higher Womers- 127
62 of the oscillation parameter δ̂ when the wall is retentive (α̂ > 0). ley number, because a stronger pressure force needs to be applied 128
63 It is clear from the figure that the increase of δ̂ leads to an in- to maintain a constant tidal volume. From Fig. 4(b), it is seen that 129
64 crease of D̂ T w . This is because the decrease of δ̂ i.e., increase of D̂ T w asymptotically reaches a steady state after a certain δ̂ and the 130
65 frequency of oscillation results a reduction in the flow within the value of δ̂ depends on α̂ and λ. For a fixed retention parameter α̂ , 131
66 annular region, hence the reduction of dispersion coefficient. The D̂ T w approaches to stationary state earlier if λ is large. 132
ARTICLE IN PRESS
1 3. Unless the aspect ratio of the annular tube exceeds a certain 67

2 critical value, the oscillatory component of the dispersion co- 68
3 efficient D̂ T w increases with decrease of annular gap (increase 69
4 of λ). The rate of increment of D̂ T w with λ depends on the re- 70
5 tention parameter and the rate is high when the retention is 71
6 weak. 72
7 4. The frequency of flow oscillation shows qualitatively differ- 73
8 ent effects on two types of normalized dispersion coefficients 74
9 resulting from the pressure gradient-cycled oscillation and 75
10 volume-cycled oscillation. 76
11 5. With the possible application in catheterized artery, the re- 77
F
12 sults show that the presence of catheter with increase in size 78
13 inhibits the dispersion process when the wall is non-retentive, 79
14 whereas for retentive wall, the insertion of catheter enhances 80
O
15 the dispersion if the retention is weak, though strong retention 81
16 in the wall reduces the dispersion coefficient. 82
17 83
RO
18 Acknowledgement 84
19 85
20 One of us (Suvadip paul) is thankful to CSIR, India for financial 86
21 support for pursuing this work. Suggestions of two anonymous ref- 87
22 erees are gratefully acknowledged. 88
23
24
25
26
Fig. 5. (a) Variation of dispersion coefficient D̂ ΓT s against
when D a = 1, Γˆ = 1 and (b) variation of D̂ ΓT s against
kinetics of the phase exchange D a .
α̂ for different values of λ
λ for different values of the
DP
References
[1] G.I. Taylor, Dispersion of soluble matter in solvent flowing slowly through a
tube, Proc. Roy. Soc. London A 219 (1953) 186–203.
89
90
91
92
27 93
[2] R. Aris, On the dispersion of a solute in a fluid flowing through a tube, Proc.
28 94
The steady component of normalized dispersion coefficient D̂ ΓT s Roy. Soc. London A 235 (1956) 67–77.
TE
29 [3] P.C. Chatwin, On the longitudinal dispersion of passive contaminant in oscilla- 95
due to the combined effect of wall retention and absorption has
30 tory flows in tubes, J. Fluid Mech. 71 (1975) 513–527. 96
been plotted in Fig. 5(a) against the retention parameter α̂ for var- [4] E.J. Watson, Diffusion in oscillatory pipe flow, J. Fluid Mech. 133 (1983) 233–
31 97
ious values of aspect ratio λ when D a = 1 and in Fig. 5(b) against 244.
32 98
λ for various values of reversible reaction rate parameter D a . Fig- [5] C. Jimenez, P.J. Sullivan, Contaminant dispersion in some time-dependent lam-
33 inar flows, J. Fluid Mech. 142 (1984) 57–77. 99
ure shows that mild retention in the wall results a sharp increase
EC
34 [6] T.J. Pedley, R.D. Kamm, The effect of secondary motion on axial transport in 100
in D̂ ΓT s . It is clear that for small values of α̂ and D a = 1, the peak oscillatory tube flow, J. Fluid Mech. 193 (1988) 347–367.
35 101
36 of D̂ ΓT s increases with the increase in λ. D̂ ΓT s first shows an in- [7] K.K. Mondal, B.S. Mazumder, On solute dispersion in pulsatile flow through a 102
crement with α̂ , then it decreases upto a certain value and then channel with absorbing walls, Int. J. Non-Linear Mech. 40 (2005) 69–81.
37 103
[8] Y. Jiang, J.B. Grotberg, Bolus contaminant dispersion in oscillatory flow with
38 it remains almost independent of α̂ . This phenomenon becomes conductive walls, J. Biomech. Eng. 115 (1993) 424–431. 104
more significant with the increase in λ. From Fig. 5(b), D̂ ΓT s shows
RR
39 [9] A. Sarkar, G. Jayaraman, The effect of wall absorption on dispersion in oscilla- 105
40 a slight increase with λ over the range 0.0–0.8 and within the re- tory flow in an annulus: application to a catheterized artery, Acta Mech. 172 106
maining short range it decreases sharply. With the decrease of D a , (2004) 151–167.
41 107
[10] B.S. Mazumder, K.K. Mondal, On solute transport in oscillatory flow through an
42 the effect of λ on D̂ ΓT s appears to be more significant. It is observed annular pipe with reactive wall and application to a catheterized artery, Quart. 108
43 that D̂ ΓT s decreases with increase of D a over the whole range of λ J. Mech. Appl. Math. 58 (3) (2005) 349–365. 109
44 [11] M.R. Davidson, R.C. Schroter, A theoretical model of absorption of gases by the 110
except at the narrow gap of the annular tube, where D̂ ΓT s increases.
CO
45 bronchial wall, J. Fluid Mech. 129 (1983) 313–335. 111

[12] C.G. Phillips, S.R. Kaye, Approximate solutions for developing shear dispersion
46 112
7. Conclusions with exchange between phases, J. Fluid Mech. 374 (1998) 195–219.
47 [13] A. Purnama, The dispersion of chemically active solutes in parallel flow, J. Fluid 113
48 Mech. 290 (1995) 263–277. 114
49 The stream-wise dispersion of tracer materials released in an [14] R. Revelli, L. Ridolfi, Influence of suspended sediment on the transport pro- 115
oscillatory flow through an annular pipe (termed as catheterized cesses of nonlinear reactive substances in turbulent streams, J. Fluid Mech. 472
UN
50 116
(2002) 307–331.
51 artery) with reversible and irreversible reactions at the outer wall 117
[15] R. Revelli, L. Ridolfi, Transport of reactive chemicals in sediment-laden stream,
52 has been studied using the method of homogenization. The present Adv. Water Resour. 26 (2003) 815–831. 118
53 results are compared with the existing results available in the lit- [16] C.O. Ng, Dispersion in steady and oscillatory flows through a tube with re- 119
54 erature when the aspect ratio approaches to zero and are found in versible and irreversible wall reactions, Proc. Roy. Soc. London A 462 (2006) 120
good agreement. The following important conclusions that can be 481–515.
55 121
[17] C.O. Ng, Dispersion in open-channel flow subject to the process of sorptive
56 drawn from the present study: exchange on the bottom and air–water exchange on the free surface, Fluid Dy- 122
57 namics Res. 38 (2006) 359–385. 123
58 1. The steady component of D̂ T s decreases with increase of λ [18] K.K. Mondal, B.S. Mazumder, On the solute dispersion in a pipe of annular cross 124
for perfectly inert wall, whereas for retentive wall (α̂ > 0), de- section with absorbing boundary, ZAMM 85 (6) (2005) 422–430.
59 125
[19] A. Sarkar, G. Jayaraman, The effect of wall absorption on dispersion in annular
60 pending upon the degree of retention, it may increase with λ. 126
flows, Acta Mech. 158 (2001) 105–119.
61 2. For non-retentive wall (α̂ = 0) and low oscillation parameter [20] D.A. McDonald, Blood Flow in Arteries, Edward Arnold, 1974. 127
62 δ̂ , D̂ T w is practically ineffective with the insertion of catheter [21] P.C. Fife, K.R.K. Nicholes, Dispersion in flow through small tubes, Proc. Roy. Soc. 128
63 except in a narrow gap, whereas for sufficiently large δ̂ (low London A 344 (1975) 131–145. 129
[22] J.B. Grotberg, Pulmonary flow and transport phenomena, Annu. Rev. Fluid
64 Womersley number), D̂ T w decreases with λ throughout its do- 130
Mech. 26 (1994) 529–571.
65 main. 131
66 132

EJMFLU

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

EJMFLU

Загружено:

Авторское право:

Доступные форматы

ARTICLE IN PRESS

JID:EJMFLU AID:2359 /FLA [m5G; v 1.65; Prn:14/10/2008; 12:29] P.1 (1-9)

39 lation and u s is the steady velocity component. We further take 105

34 with the boundary conditions 100

41 functions B (r ) and B s : 2i Sc ν δ 2 ω 107

1 ∂ C (0) ∂ C (1) ∂ C (2) ∂ C (1) + 708λ6 R log(λ) D a − 36 log(λ) D a + 144R λ6 log(λ)2 67

34 and + λ4 log(λ)2 + 2 log(λ)λ4 − 1 + 3λ2 − 3λ4 + 2λ2 log(λ) 100

45 D d dY 1 ∗ effects on D ΓT w . Since no closed-form analytical solutions are avail- 111

50 (74) our main emphasis is on the effect of aspect ratio. 116

1 view the reversible and irreversible reactions are considered to oc- 67

34 the advection–diffusion transport of a chemical species in a ﬂow 100

50 aspect ratio of the annular tube. 116

34 coeﬃcient for pressure gradient-cycled oscillation ( D̂ T w ) as functions of δ̂ when 100

45 is observed that like steady component, oscillatory component of 111

1 3. Unless the aspect ratio of the annular tube exceeds a certain 67

45 bronchial wall, J. Fluid Mech. 129 (1983) 313–335. 111

Вам также может понравиться