135
Heat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Faculty of Graduate S tudies The University of Western Ontario London. Ontario July 1996 @ Te- Moschandreou 1996

Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

  • Upload
    hacong

  • View
    213

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Heat Transfer with Pulsatiie Flow in a Tbbe

b~

Terry Moschandreou

Department of Appiied Mathematics

Submitted in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

Faculty of Graduate S tudies

The University of Western Ontario

London. Ontario

July 1996

@ Te- Moschandreou 1996

Page 2: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Acquisitions and Acquisitions et BiMiogmphic Senhœs services bibliographiques

The author has gmted a non- exclusive licence aiiowing the Naîionai L i of Canada to reproduce, loan, distriiuie or seU copies of his/her thesis by any means and in any fmm or fomat, m a h g this thesis adable to interested persons.

The author retains ownership of the copyright m his/her thesis. Neither the thesis nor substantial extracts fkom it may be p ~ t e d or othemise reproduced with the author's permission-

L'auteur a accordé une licence non excIusive permettant à la Bibliothèque nationale du Canada de nprodimc,~, distri'buerou vendre des copies de sa thèse de quelque manière et sous 9ue1que fome que ce soit pour mettre des exemplaires de cette thèse à la disposition des personnes intéressées.

L'auteur conserve la propriéte du droit d'auteur qui p d g e sa thèse. Ni la thèse ni des extraits substantieis de celle-ci ne doiveat être imprimés ou autrement rrproduits sans son autorisation,

Page 3: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

The partial differentid equations governing heat transfer with pulsatile flow in a

tube. which serves as a mode1 of a simple heat exchange device. are solved in this

t hesis.The governing equations axe the Navier Stokes equat ions and the convection-

diffusion equation. Two boundary value problems are solved.

tn the fint case a Xeumarin boundary condition is specified vvhich represents constant

heat flux at the waiI of the tube. with fluid entering a t h e m d region a t a constant

temperature. -4 regular perturbation e.xpansion is used to obtain higher order har-

monics downstream for the temperature field. The main assumption is that the

t ernperat ure field becomes fuli y developed downstream as the veloci t y field becomes

fully developed downstream. The perturbation parameter is the ratio of pressure

gradient amplitudes of unsteady 00w to that of steady flow. Using a Green's function

the first order solution is obtained. A s a measure of heat transfer enhancement. a

bulk temperature is formulated for the convective process involved and a change in

unsteady Yusselt number to that of steady Nusselt number is evduated anal-ytically

In the second part of the thesis the more difficult problem of heat transfer in pulsatile

flovv with constant wail temperature is considered. Although a complete solution is

not possible as in the first part, it is possible to use a combination of the Gener-

alized Integral Transforrn Techniques and Laplace transforms to solve this problem

upstream and downstream in the thermal region of the tube. The approximate solu-

tion indicates that a plane wave propogates down the tube and the phase of the wave

defines critical values in frequency and time dong the tube. Use of Dirac's distribu-

tion rnakes it possible to define a bulk temperature and a change in unsteady bulk

temperat ure from t hat of steady bulk temperat ure is presented. B y means of classical

analysis an inequality involving the two quantities is presented. -4s a result a measure .-. 111

Page 4: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

of heat transfer in pulsatile flow compared to that of steady flow is presented.

Page 5: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

To my father and mother

Page 6: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

I wouid Like to express my thanks and appreciation to my supervisor Dr.bI.Zamir

for his guidance and direction over the past few years. I also appreciate helpful

discussions with Dr. Stan Deaiiin. I: also would Iike to thank National Sciences and

Engineering Research Council of Canada for their Financial support.

Page 7: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

TABLE OF CONTENTS

CERTIFICATE OF EXAMINATION ii

ABSTRACT

DEDICATION v

ACKNOWLEDGEMENTS vi

TABLE OF CONTENTS vii

LIST OF TABLES x

LIST OF FIGURES 1

Chapter 1 Introduction 3

Li Motivation :3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L.2 Previous Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Present Work. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

C hapter 2 Governing Equations 7 C, 2.1 Basic Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 1 - - c. 2.3 Governing System of Equations . . . . . . . . . . . . . . . . . I

2.3 Preliminary Steady State problems . . . . . . . . . . . . . . . 11

2.3.1 Heat Thxirfer With SIug Flow in a Thbe . . . . . . . 11

2 - 3 2 Method of Solution . . . . . . . . . . . . . . . . . . . . . 12

Page 8: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

2.3.3 The Graetz Problem . . . . . . . . . . . . . . . . . . . . 16

. . . . . 3.3.4 Asymptotic Method to solve Graetz problem 16

C hapter 3 Heat Transfer with Pdsatile FIow and Constant H a t Flux 21

.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 39 3.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . .

13.3 Regular Perturbation Method . . . . . . . . . . . . . . . . . . 35

9.4 Steady Temperature . . . . . . . . . . . . . . . . . . . . . . . . 29

3.5 GreenYs Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 131

4.6 Oscillatory Temperature . . . . . . . . . . . . . . . . . . . . . . 33

3.7 Uniqueness of Solution . . . . . . . . . . . . . . . . . . . . . . . 39

. . . . . . . . . . . . . . . . . . . . . . . . . . i3.8 Riemann Surfaces 39

3.9 Verification of Boundary Conditions . . . . . . . . . . . . . . 43

3-10 Zero Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3.11 Higer Order Perturbation terms - Convergence Criteria . 46

3 . 1 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . 49

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13 Conclusions 53

Chapter 4 Heat Transfer with Pulsatile Flow and Constant Temperature 53

1 The Generdiaed Integral Transform Technique . . . . . . . 53

4 Heat Transfer in Pulsatile Flow with Constant Wdl Tem-

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . perature 60

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Distributions 65

. . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Method of Solution 66

. . . . . . . . . . . . . . . . . . . . . . . 4.5 Results and Discussion Cl

C hapter 5 Concluding Remarks 81

Appendix A Maple Code for the Graetz Problem 83

Page 9: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Appendix B Maple Code for Constant Heat Flux Problem

Appendi* C Axial Gradient of Temperature Downstream

Appendix D Picard-Lindelof Theorem

Appendix E Properties of the Dirac Delta Function

REFEmNCES

VITA

Page 10: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

LIST OF TABLES

2.1 Parameters associated rvith the Graetz problem . . . . . . . . . . . .

3.1 Zeros of the function !I! for zeroth order perturbation . . . . . . .

- 4.1 Zeros r.= of wave function and pulsatile velocity u = 1 + cost: t= n/4

- 4.2 Zeros r.= of wave function and pdsatile velocity u = 1 +cost:t =3i;/-L II

4.3 Zeros r. = of wave function and pulsatile velocity u = L + cost :t=3 i;/2

4.4 Zeros 7.5 of wave function and pulsatile velocity u = 1 + cost: t = 27r -

4.3 zeros ( r ). = of wave function and pulsatile velocity ti = 1 + cost + sint

A=,/;) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . -

4.6 zeros (7). = of wave function and pulsatile velocity u = 1 + cosf + sint:

- 4 zeros ( r ) . = of wave function and pulsatile velocity u = 1 + cost + sint

Page 11: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

LIST OF FIGPRES

2.1 lnterpoiated eigenfunction & of Graetz problern ...................... L9

2.2 interpolated eigenfunction RL of Graetz probkm ...................... 20

2.3 uiterpolated eigenfunction R2 of Graetz problem ...................... 21

*') 9 -1.4 [nterpolated eigenfunction R3 of Graetz problem ...................... -,

2.5 Temperature difference vs radial iength of Graetz problem ............. 23

2.6 Temperature difference vs axial Iength of Graetz problem .............. 24

2.7 Local Yusseit number vs axial length of Graetz problem ............... 25

3.1 Constant heat f l u model ................................... ,. .. .. .... 2 9

3 . 2 Bromwich contour associated with <p(x . r) ............................ 110

3.3 Riemann surface for log z ............................................. 15 .- 1-4 l.vu \ .sd : c = 0.1 ................................................... 3 1

........................... 3.3 A.\., VSJ* : ,C = 0.3 ..................... ,. 58

6 L V U vs C. : : = 0 . .j ................................................... 59

1.7 LI', V S ~ : : = L.0 ................................................... 60

.............................................................. 3.8 1 9 ~ v s u 61

....................................... ......... 3.9 Aivu vs LJ : P, = O . a .. 62

............................................... .3.10 AN, ~ S J . P,=.L.S 63

................................................ 1 &Vu v s u . P,=.j.O 64

4.0 Constant temperature mode1 ......................................... 72

4.1 Phase plane surfaces- Moderate Frequencies : g(r.t.r.sr)= const ......... 93

Page 12: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

4.2 Phase plane surfaces-Higher Frequencies : gp.t.::*. t = consr . . . . . . . . . . . . 94

xii

Page 13: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Chapter L

Introduction

1.1 Motivation

T here are several practical situations where heat is being transfered under conditions

of pulsatile Bow. Examples occur in industria applications and in blood flow. Also

the problern may be important in the design of control systems for heat exchanger

equipment. One of the objectives in industrial applications has been to determine

whether the performance of a heat-exchange device could be improved with pulsatile

flow.

When pulsations are irnposed on Bow in a tube. it is believed that heat transfer would

be changed because the pulsations would alter the thichess of the bound- layer

and therefore the thermal resistance. This vieiv was put fonvard by Richardson [II who &ad shown chat the velocity profile near the <val1 is steeper in pulsatile Bow than

in Poiseuille Bow. It then follows from a consideration of Reynold's anaiogy that the

heat traosfer should increase under such conditions.

It is the purpose of this thesis to determine whether ~ulsatile flow can give rise to an

increase in heat transfer in a tube. which is a basic elernent in many heat exchange

devices.

Page 14: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

1.2 Previous Work

The mechanics of pulsatile Kow in a tube has received considerable attention starting

with the classical work of Uchida PI. Womersley [:3] and Atabek and Chung [;LI which

has led to an exact solution and much information about the flow problem. Studies of

the corresponding heat tramfer problem. hotvever. have been far less nurnerous and

existing results do not provide cornplete resolution of a l l the questions involved.

Siegel and Perlmutter (51 demonstrated the explicit dependence of overail heat trans-

fer on pulsatile frequenc- They found that for a constant d l temperature boundary

condition. t be resdting Nusselt number shotved penodk axial change which could en-

hance heat transfer.

Barnett and Vachon [6] analyzed fully developed pipe flow in the presence of lon-

gitudinal periodic pulsations by assuming that the radial part of the temperature

profile is additively independent from the axial part. Although they did not justify

this assumption they were able to conclude chat heat transfer effects with constant

flux at the mal1 are amplified by pulsations of fluids of Prandtl oumbers well below

unity for the limiting case of low frequency.

More recentiy CreE and Andce [ï] solved the Xavier Stokes and energy equations

using a finite difference met hod and an asymp totic development for the d ~ a m i c and

thermal quantities. Their mode1 shows the existence of an annular effect in the entry

region for the pulsatile part of velocity and temperature which is known as Richard-

son's effect. The work was based on the full Navier Stokes and energy equôtions which

made it difficult to obtain an explicit expression for the Nusselt number over one cycle.

Cho and Hyun [8] solved the time-dependent laminar boundaq layer equations nu-

merically over a large range of frequencies and amplitudes of pulsation. They found

that for certain frequencies the 'iusselt number increased over the steady-flow value.

Page 15: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

They also found that the Susselt number trend is amplified as the amplitude of pul-

sation increases and the Prandtl number is much Iess than unity.

Haneke. Laschefski. Gro be-Gorgemann and Mit ra [9] solved the Xavier-S tokes and

eoergy equations numericaily for laminar pulsating flow in a chaanel. At the en-

trance region of the channel a sinusoidal pulse was applied. Their results indicated

a positive and negative overshoot (Richardson's effect) in the avid velocity profile

and flow separation near the w d . They concluded that appreciable heat t r a d e r

enhancement occurs in the channel.

Kim. Kang and Hyun (101 conducted a study of the heat transfer characteristics of

lully-developed pdsatile flow in a channel and found numericdy that changes in the

Nusselt number were pronounced in the entrance region. and only minor for down-

stream. -ils0 the effects of pulsatile frequency on heat transfer \vas found to be

noticeable when the frequency is s m d or moderate but not when it is high. They

concluded that oscillations may produce both heat transfer enhancement as well as

reduction at different a~ial locations in the channel.

Base. Campbell S; Hobbs [II] studied the heat transfer characteristics of pulsatile

Aow in a pipe experimentally and found there to be an optimal frequency at which

there is an increase over the steady value in heat trânsfer for fluids OF Prandtl number

near unity.

Some other experimental work in heat transfer with pulsatile Bow has b e n con- - -

ducted by Fallen [Io] showing that superposition of a pulsating flow in a pipe with

zero mean velocity on steady flow increased significantly the overall Nusselt number.

Experimental work has also been done by Genin, Koval. Manchkha and Sviridov

1131. Hapke [l4]. Andre. Creff and Crabol [~j] and Peattie and Budwig [16] , dl of

whom have reported increases in heat transfer rates with pulsatile flow.

Page 16: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

1.3 Present Work

In this thesis the aim is to demonstrate the mechanics of heat transfer in pulsatile

ftow in a tube analytically. That is. ive must solve the energy equation coupled with

the Navier Stokes equat ions wit h a sinusoidal pressure gradient driving the flow.

In the first part of the thesis we use a regular perturbation expansion [1T] t o solve the

convection problem subject to a constant w d heat flux. CVe obtain. using Laplace

t ransforms and Green's functions. qualitative and quantitative insight into how the

above phenornenon depends on the nondimensional parameters involved such as pui-

sat ion frequenc- and amplitude. Prandtl and Reynolds numbers.

In the second part of the thesis we use an integral transform method known as the

geeoeeralized integral transform technique p3]. (-41. [25] to solve the problem of heat

transfer with pdsatile Bow in a tube as in the first part but with constant w d

temperature instead of constant heat flux. .An explicit fomulae is obtained for tem-

perature at any axial distance along the thermal region of the tube both upstream

and dosvnstream and the existence of a wave-like form in the solution is shotvn. The

bulk temperature for t his unsteady solution is iormulated and compared to the cor-

responding steady one.

Page 17: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Chapter 2

Governing Equations

2.1 Basic Assumptions

W e consider fully dewloped laminar pulsating flow of an incompressible. Sewtonia.

Buid. It is convenient to take cylindrical coordinates with x axis coinciding with the

center Line of the tube. The main assump tions are t hat ( 1) the velocity field is in the

r direction only and is assumed fu& developed: (2) the flow is axially syrnmetric: (3)

there is unifonn pulsating flow and hence the pressure gradient becomes a function

of time only: (4 ) viscous dissipation effects are negiigible compaxed to the convective

rate of heat transfer: ( 5 ) the fiuid is entering a thermal region with a constant uniforxn

t emp erat ure.

2.2 Governing System of Equations

The Xavier Stokes equations under the above assumptions and in cylindricai coordi-

nates are

a au àv I â p apv a 2 ~ 1 av - + u - + u - = --- at d~ br par

+Il(-+-+----) d x 2 dr* r d r r*

Page 18: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

with the equation of continuity

where x.r are axial and radial coordinates and u.u are velocity components in x and

r directions respectively. t is time and p is density of the fluid.

If the Bow direction is assumed CO be parallel to the r-axis. as stated in our as-

sumptions. velocity has x component only and c is zero. The equation of continuity

gives

mhich indicates the velocity to be constant in the direction p a r d e l to the center line

of the tube. The pressure gradient is assumed to be a harmonic of some Fourier series

in t ime.

The temperature T' within the tube is governed by the simplified convection dif-

fusion equation

dT' ~IT' k (B*T- MT=) - + uœ(rl t t*)- = - - dtœ + ,- a ~ - pC, a ~ - * r dr*

where C,, k are specific heat. and thermd conductivity of the fluid, and the starred

quantities indicate dimensional fom of these variables. The second derivative of T'

with respect to x that appears in the laplacian is neglected on the assumption that

it is small compared with the derivatives of Tm with respect to r in the downstrearn

Page 19: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

region of the fl ow being considered.

In the k t part of the thesis the solution is sought for a region of the tube down-

Stream. where the flow field is f d y developed and the heat BU dong the b o u n d q

of the tube is constant and fluid is entering the region at a uniform temperature Tt.

Introducing the nondimensional temperature dxerence

where q, is the constant heat B u at the wall. a is the radius of the tube. and k

is the thermal conductivity of the Buid. the problem can be put in the nondimen-

sionai form

with boundary conditions for B(x. r. t )

Page 20: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

In the second part of the rhesis a constant mdi temperature is prescribed on the

i v d of the tube rvith Buid entering a thermai region at a different constant temper-

ature. The velocity field is assumed to be that of fu- developed pulsatile Bow as in

the first part of the thesis.The goveming equation is the same as in 2.6 but with

and where TtU. To are the waH temperature of the tube and the entry temperature of

the fluid respect ively.

The boundary conditions in this case are

Page 21: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

3 Preliminacy Steady State pmblems

In what foiIows ive present resdts of steady state problems which are reguired for

solution of the heat transfer problem with pulsatile Bow.

2.3.1 Heat Transfer With Slug Flow in a Tube

in this section we consider the case of IaJninar flow in a circuiar tube with heat trans-

fer. Let the temperature of the fluid be To at the thermal entrance of the tube and the

tvall temperature be Tw. Also we assume in this section that a t the thermal entrance

to the tube the ve loc i - is uniform over the cross-section. t hat is

The energy equation for a fluid element is:

ivhere p is the Auid densit. C, is the specific heat at constant pressure. k is the

thermal conductivity of the fluid. cr = k/pC, is the thermal diffusivity and

in terms of nondimensional temperature 0 = (T' - Tw)/(To - Tw) and nondimen-

sional quantities x = x'/aP,- Re and r = r'/a the equation becomes:

86' a26' 1 de um- =a(-+--) ax dr* r d r

Page 22: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

wi th b o u n d q conditions:

2 - 3 2 Method of Solution

%y separation of mriables

one obtains

The second equat ion has a solution

JO(&) = O Vi.

Page 23: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

The constants -Ji are determineci by the condition that

Multiplying both sides of Eq. 2.16 by rJo(mr)

and integrating from O to L gives:

The following two results wiii be made use of in the solution.

If X and p are two different constants then.

Letting n = O and X = 3, in equation 2.19 above and given the identity rJn(r) =

Page 24: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

nJ,,(r) - rJ,+Lir) rvhich implies that.

JO'(.) = J:(P)

we get

where

The unifonn temperature assumption f ( r ) = 1 gives:

Page 25: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

and

The complete solut ion for slug flow becomes

Page 26: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3 The Graetz Problem

Heat transfer problems of forced convection in tubes have been studied for maoy y a x s

beginning with Graetz [lY] and later with Nusselt [191 and Leveque [-O].

In this section we assume poiseuille flow at the entrance of the thermal region and

to the left of this region we assume the Buid is isothermal. In the t hemal region a

prescribed temperature or heat flux is specified at the wall of the tube. This problern

is refened to as the Graetz problem.

'2.3.4 Asymptotic Method to solve Graetz problem

From Eq. 2.6 assuming a steady state form with u = 1 - r2. the Graetz problem

becomes

tvi t h boundary conditions :

rnhere 0 = (TR - T,) /(To - Tw) . Tu! To are rvall and entrance temperatures re-

spect iveI -

,\ssume O is in the form :

where A, are eigenvalues that satisfy the following differential equation:

Page 27: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

satisfying the boundary equations:

The Sturm-Liouville theory gives us onhogonality with respect to the weight function

r( 1 - r2)? that is

for rn # n

It can be shown eaçily that

and the folloming identities are known to hold [lS]

1 a m a , l1 r ( l - r2)%dr = (-)(-- lx, ar ax, .=,

where the boundaiy conditions above have been used.

Lïsing these ttvo results we get at once that

Page 28: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

Assume a solution to the above Sturm LiouviUe problem Eq. 2.29 in the fom:

R =

Substitution of R into the ordinary differential equation gives :

h" + hf2 + (L/+)hf + X2(1 - r") = O

An as-ymptotic solution is sought in the form:

h = Aho + hl + (A)-'h2 + ( ~ ) - ~ h ~ + . . .

where X is the eigenvalue of Eq. 2.35.

After straightfonvard substitutions a solution involving just ho and h is obtained in

the form

with A. B complex numbers.

Sellars. Tribus and Klein [-II have shown that the eigenvalues ase given as

and the eigenfunctions are as follows:

Year the center of the tube. r O

& ( r ) = Jo(kLr)

Page 29: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

for intermediate r =z 0.5

and near the wall, r = 1

These 3 approximations can be patched together to produce continuous results. W e

used a Maple routine [Appendix Al in which eigenfunctions and temperature promes

of the Graetz problem can be obtained.

Refer to Fig. 2.1- Fig. 2.7 for eigenfunctions and temperature profiles.

The constants C, are given by

( - L )Y + 6 2 / 3 r ( - p ) x - ~ 3 C, = n - n = 0 . 1 . L . .

I l

Table 2.1 lists several eigenvaiues and important constants for the Graetz problem.

The Xusselt number for constant wall temperature is

Page 30: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Table 2.1: Parameters associated with the Graetz problern

Refer to Fig. 2.7 For the Nusselt number variation dong the tube.

Page 31: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 32: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 33: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 34: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 35: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 36: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 37: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 38: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Chapter 3

Heat Transfer with Pulsatile Flow and Constant Heat Flux

3 1 Overview

In this chapter an analysis of fuliy developed Bow in a tube in the presence of longi-

tudinal periodic pulsations with heat transfer is presented.

CTsing a regular perturbation method [ l T ] an anal-ytical solution is sought for the tem-

perature field domnstream. This expression is written as a steady terrn with higher

order hannonic terms superimposed on it. These higher order t e m s are due to veloc-

ity fluctuations in pulsatile Bow. It can be shown that these velocity pulsations cause

harmonic osciilat ions in temperature to occur t here b- spli t t ing the temperat ure field

into a steady part plus an oscillatory part.

Our aim in this thesis is to deterrnïne if the Bow field for pulsatile flow eobances

the overall heat transfer rate as rneasured by the Nusselt number.

The bulk temperature over one cycle and the change in unsteady Nusselt number

to that of steady Nusselt number is shown to depend on the dimensionless frequency

4 and a number rvhich ive obtain through dimensional andysis which involves a ratio

of unsteady pulsation amplitude to that of steady amplitude. this number which we

derive below is denoted by c.

Page 39: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3 Governing Equations

The Temperature T' within the tube is governed by the foilowing coupled partial

differentiai equations

DT' k cl-=-

ot' pCP of T=

where the materiai derivative operator is defined as

D d --- - + ~ ' - y & Dt* bt

The function u' satisfies the Xavier Stokes equations for a sinusoidd pressure gradi-

ent driving the flow with a no slip boundaq condition at the wall of the tube and a

fully developed velocity profie along the tube. The temperature of the fluid at the

thermal entrance region is constant with constant heat flux at the wall of the tube.

See Fig 3.1 on page 23.

The following boundary value problem results.

Page 40: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

\ïscous dissipation effects as given by the dissipation function are assurned ne&

cible cornpared to the convective rare of heat transfer so that no nonlinear effects are z

present in the energy equation.

\Ve noow introduce the foiloiving nondimensional quantities

mhere q, is the constant heat flux at the boundary of the tube. a is the radius

of the tube and k is the thermal conductivity of the Buid.

The Prandtl number. Reynolds nurnber and the t hemal diffusivity are defined by

Page 41: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

Substitution of Eq.3.4. i3.5 and 3.6 into 3.3 the governing equation finail' becomes.

in nondimensionai Form

with b o u n d q conditions for B(x, r. t )

Fig. 3.1

Page 42: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

X3 Regular Perturbation Method

The flow field wit hin the thermal region of the tube is assumed to consist of a steady

part represented by f d y developed Poiseuille flow plus an oscillatory part represented

by the classical solution for pulsatile ff onr [3], PI. Thus with one h m o n i c of some

Fourier series elcpansion added to steady flow we mite

where A0 is the pressure gradient driving the steady part of the flow and AL is the

ampli tude of the oscillatory pressure gradient driving the oscillatory part of the flow.

The pressure gradient with respect to the x variable is written as a steady amplitude

plus one hannonic of some Fourier series in time.

Introducing a nondimensional frequency parameter.

the velocity downstream can be put in the nondimensional form

tr(r. t ) = uo(r) + fu t ( r . t )

Page 43: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

iv here

Variables have been nondimensionalized as before. wi t h the normalizing velocity tio

now being taken as the maximum velocity in Poiseuille flow. namely

Due to the nonlinearity of the product of the velocity and the gradient of temperature

difference in the convection equation and since the veiocity for pulsatile flow depends

on the ratio of LI/.^^. the following regular perturbation e-xpansion is used to solve

for temperat ure downst ream.

ivhere 6 is the perturbation parameter defined previously and will be assumed small

(i.e. less than unity).

Page 44: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

Substitut ing the velocity and temperature expressions in the governing systern yields

infinitel- ma- boundary kalue pmblem for Bo and 01 02. etc.

The first t hree are

a00 a2eo 1 ao, uo- = - + -- da: dr3 r dr

In view of the foim assumed for the temperature. these bound- d u e problems

govern the steady ( &(x. r) ) and oscillatory ( &(r, t ) , &(P. t) .etc.) parts of the tem-

perature field. respectively. and their solutions are considered separately in the next

t wo sections.

A solution to (:3.19) is sought downstream. where we assume that 2 = const . The

Page 45: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

bound- condition is physicaUy reasonable since far downstream the effects of idet

temperature are negiïgible where primary heating of Buid is due to the heat flac at

the maii done-

In the next section ive claim that the derivative of Bo with respect to x downstream

becornes a constant. Since the series is e'cpanded in { which is variable then the

partial derivatives of 62, etc with respect to x. approach O downstream.

Page 46: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3.4 Steady Temperature

The zerot h order problem. 60. is solved \vit h the use of Laplace transforms and residue

theo. W e have described the solution in appendix C . where we obtain the following

result for the temperature

where some talues of 32 and W(-32) are given in the following table

Table 3.1:

and

Zeros of the function Q for zeroth order perturbation

W e observe that for tb is zeroth order problem far downstream

Page 47: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

tVe make use of this result in the solution of the k t order problem.

Page 48: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

:Lj Greenk Eùnctions

The Green's function merhod is one of the most important approaches for solv-

ing b o u n d q d u e problems wi t h noohomogeneous ordinaq different ial equat ions.

Sturm Liouville pmblems w hich involve self adjoint different id operators are writ ten

as

The idea is to obtain a solution of the boundary value problem

Lh = g

in the form

where Ii turns out to be an integral operator. This kernel is cded the Green's

function for the boundary value problem.

Ordinarly q.g are real d u e d but the method can be extended to cornplex values

for q. g. The theory of global analytic functions allows one to study the complex

solution of ordinary different ial equations. [22]

Page 49: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

This extension is necessary in Our problem for the solution for oscillatory temper-

at ure. Construct iveiy. a standard technique for solving Sturm-Liouville problems is

t O use tariat ion of parameters. Suppose O: V are two linearly independent solutions

of the corresponding homogeneous equation L h = 0.

A solution is sought in the form h = oU + GV upon which substitution gives

If g is continuous and the CVronskian CV = p ( L V - VU') is not zero then h = OU+ c V

is a solut ion for the differential equation. more e-xp ki t -

is the Green's function associated wit h the problem.

In the next section the t-ype of Sturm-Liouville problem associated with oscillatory

temperature involves a differential equation with complex valued arguments as CO-

efficients with the nonhomogeneous part being complex d u e d as well. -4s already

mentioned a solution can be obtained in the complex phne-

Page 50: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3.6 Oscillatory Temperature

The result of the previous section ( Eq. :323 ) maks it possible to seek a solution far

downstream for BI in the form

Substituting this into boundaq value problem Eq 3.19 Leads to the following bound-

ary talue problern

CPh I d h - + -- - iwh = g ( r ) dr* r d r

h f ( l ) = 0 . h l (0 ) = O

In addition ive assume that at zero frequency. the solution for h is bounded and

finite.

The differential equation is of regular cornplex valued Sturm-Liouville type

Page 51: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

where G is the Green's fuoction of the above Sturm-Liouville problem and K is a

specid integral operator '

The most general solut ion is

The Wronskian is defined as follows

K is a compact cornplex Hermitian operator acting on L2(0. l ) , 1; : L2(0. 1) + L'[O. 1)

Page 52: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Xow the boundaq conditions. of s y m e t r y and flux are

The first condition gives

and since the derivative of the Xeumann function is unbounded at the centreline

of the tube.

Differentiation of the function h ( r ) ivith respect to r gives

Page 53: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

The second boundary condition gives us at once that

c1U'(l) + V ' ( I ) G V - ~ LL U ( r ) g ( r ) d r = O

from which cl is determined as

Hence the solution for the function h becomes

The functions U. V are Iinearly independent solutions of the corresponding homo-

geneous different id equat ion

Page 54: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

Lh = O

The function H is Heaviside's function. defined on [O.Il.

Substitution of the linearly independent sohtions L-. C;' the function g, the Wron-

skian W and the Green's function G leads to the following solution:

where.

Jo(r\l-iwl Pr) 1' 1W

Page 55: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3.7 Uniqueness of Solution

The solution obtained in the previous section involves Bessel functions 1; of a com-

plex argument =. and since these in tum depend on log2 each part of the solution is

multi\dued and thus not unique. In this section we discuss the notion of a Riemann

surface upon which each part of the solution h l , hz, h3 becomes single-dued and thus

rvell-defined.

W e introduce the idea of a branch in order to study rnuiti vaiued functions properly.

Two function elements (fi. R I ) and ( f2, R2) determine the same branch at a point in

the complex plane z0 E ( R I n RÎ) whenever fl = f2 in a neighbourhood of the point

3, which holds if the functions are analytic continuations of each other. The branch

at the point 20 given by the function element (f. R) is denoted as (f: 20) . The set of

branches ( f. t ) is defined as the Riemann Surface.

-4 Riemann surface is simply a generalization of the complex plane where a multiple

valued function is defined on a surface of many sheets where the function becomes

single d u e d on this surface. To describe a surface of many sheets we mil1 consider

the Riemann surface for the function logr which is comected to our Green's function

solut ion obtained previously.

A Surface for Log 2

For z in polar form

- = re - id

log= = Logr + i9

Page 56: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

where Log r is the principal value of log =.

The function log z is obviously many d u e d at each point in the complex plane

since the angle 19 is increased by multiples of 2 1 and -Zn respectively.

To turn this Function into a single valued function. the rplane is replaced by its

Riemann surface on wbich new points are represented denever the argument of the

point z is either increased or decreased by 2n radians.

By considering the =-plane as a sheet &, cut along the positive x axis. let the angle

O range frorn O to 2;;. -4 second sheet Ri is cut ideotically and placed in front of the

sheet 5. The lower edge of slit in & is joined to the upper edge of the dit in RI.

On RI the angle 0 ranges from 21 to 4;1. The same is done for sheets R2, R3, etc.

A sheet R-I placed behind Ra. on which B varies from O to -21. is cut with the lower

edge of it's dit connected to the upper edge of the slit in Ra. and this is coatinued

for R 4 , R+ etc-

'iow the origin is cornmon to ail sheets. In the above construction as the point com-

pletes a cycle around the origin on &, the angle ranges frorn O to 27i. -1s it rnoves

across 6 = 2;; the point passes to the sheet Ri ol the surface. As the point completes

a cycle in RI. the angle 6 ranges from 27 to -Lx and so on. Refer to Fig 3.3 for the

Riemann surface for log:.

Page 57: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 58: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

A U parts of the solution h. chat is hlth2.h3 are ma- valued functions in the

complex plaae and therefore require a Riemann surface consisting of infinirely many

sheets for the Iogarithmic function involved. As mentioned previously the functions

becorne single d u e d on the Riemann surface and hence are well defined functions.

The solution obtained however for h via the Green's Function is unique. Despite

the multivaluedness of each function h 1, h2, hJ, their sum h adds up to be single

valued. .-i qui& calculation shows that the logarithmic parts cancel out as they are

summed up to obtain h.

To show this note that since logr = Logr + i0. the multi-valued part of hl is

L L - ~ i ~ j ~ > & i 3 1 2 . j o i ~ i 3 ~ 2 r ) ~ ~ 0 ( & i ~ / ~ 7 ) ~ [l - Joc~-, -iu/ P r ) d r (3.12)

the multi-valued part of h2 is

and the multi-valued part of h3 is

Adding the three parts together gives us zero identically and thus the function h

becomes single-kalued in the complex plane. Hence we m i v e at a unique solution.

Page 59: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

1.9 Verification of Boundary Conditions

One of the properties of the solution for h obtained previously is that in addition to

satis@ing the complex differential equation. it satisfies both flux and symmetry Line

boundary conditions.

Different iation and use of the first version of the fuidamental t heorern of calculus

gives us the followiog

then it follows immediately t hat

Page 60: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

&(rd-iw/ P r ) -- ah, - &< fii312r)(a/z)~i3/2 dr 1

Adding up the three tems above gives

Next the flux boundary condition is checked. Differentiating and using the first

version of the fundamental theorem of cdcuius we get

Page 61: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Adding the above parts gives.

Hence bot h boundary condit ions are satisfied by the Green's funct ion solution.

Page 62: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3.10 Zero Frequency

At zero frequency the term. - V ( l ) / U ' ( I ) l in hl ( a.39 ) is unbounded because its

imagin- pa.rt is unbounded. The real part provides a d i d solution however and

therefore. O& the real part of -C"(1)/Uf('(1) is taken for calculation purposes.

3.11 Higer Order Perturbation terms - Convergence Criteria

The perturbation e-vansion in section 3.3 led to infinitely many coupled partial dif-

ferentiai equations of which we have solved only the first two downstream.

A naturd question to ask is whether or not the perturbation series converges down-

streaxn for s m d perturbation parameters c.

In order to answer this we must first solve a,il of the remaining higher order terms

in the perturbation series starting with the second order and proceeding to the kth

order one and so on.

This of course means that infinitely many boundary value problems must be solved

from the second order system and so on.

Recall the second order probIem

ae, ae, adz a2e2 i de2 - + u ~ ( r . t ) - + u&)- = p at + -- dx t3x r

Page 63: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Downstrearn this simplifies to

In fact the generai k'th order boundq value problern downstrearn becomes

The solution of this problem is readily obtained as

where

Page 64: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

and prime denotes differentiation with respect to r.

The eigenfunctions R(&, r ) are the solutions to the FoHowing eigenvalue problem:

where the 3, are the zeroes of J L .

Yow one sees that the k'th order solution is bounded over one cycle and across the

tube radially.

Since the k'th order problem in the perturbation e.xpa,nsion is bounded with 5 < 1

t hen the series converges downstream.

Hence the perturbation e-xpansion converges downstream.

Page 65: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3-12 Results and Discussion

To determine the effect of the oscillations on the rate of heat t ramier ive define the

local Xusselt number

and the corresponding Yusselt number in steady Boiv

where R denotes the real part of a cornples valued function. the subscript 6 refers to

bulk properties of the fluid. and the overbar indicates time average over one cycle.

that is

In terms of the nondimensional temperature difference 0 the expressions for the Nus-

selt numbers become

Page 66: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

in these. we have that

and

w here

where only the Çst two terms are considered in the perturbation expansion since the

expression for veiocity is exact. (Eq. 3.12)

As a measure of the effect of oscillations on the heat transfer rate we now consider

the relative difference quant ity

Substituting for the Nusselt numbers from equation 3.52 and 3.53 and for the unsteady

bulk temperature from equat ion 3.57 t his becomes finally

Page 67: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

In Fig 3.4-Fig :3.ll ive gaphed our resuits for &ous frequencies and Prandtl num-

ben as iveil as for different values of C.

In particular. frequencies in the range of 0-50 are considered wit h Prandtl numbers

P, at 0.5. 2.5 and 5.0. The perturbation parameter is taken as 0.1. 0.3. 0.5 and 1.0.

Page 68: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

3.13 Conclusions

Csing a regdar perturbation method. the coavection equation coupled to the clas-

sical pdsatile solution was solved using Green's huictions in the cornplex plane and

Laplace transfonns upon which a bulk temperature and relative Nusselt nmber were

obtained. Since heat transfer fiom the tube is constant because of the boundaxy

condition. the effect of osciiiations can only occur in tems of change of buik ternper-

ature tvithin the tube- From the definition of Nusselt number we see that the Xusselt

number and the relative 'iusselt number depend on the change in bulk temperature

due to the pulsations. It is believed that Iower viscosity and hence srnalier Prandtl

numbers are associated with s m d momentum and thermal b o u n d q Iayers hence

by Reynold's andogy an increase in heat t ransfer shodd occur. The results obtained

indicate that pdsatile floiv contributes to an enhancement in the overalI heat transfer

rate. An important observation is that there exists a critical value in frequency where

a maximum occurs in the relative Nusselt n u b e r . The Prandtl number dependency

is such that k a t transfer increases by pulsations of fluids with Prandtl number less

t han unity and decreases for Prandtl numbers greater than unit.

Since the Green's function requires us to evaiuate a number of tedious integrais in-

rolving Bessel functioos it was appropriate to wi te a Maple program to compute the

bulk temperature and relative 'iusselt number. [AppendixB]

Page 69: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 70: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 71: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 72: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 73: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 74: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 75: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 76: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

t i r

Page 77: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Chapter 4

Heat Transfer with Pulsatile Flow and Constant Temperature

4.1 The Generalized Integral 'Itansform Technique

In the previous chapter a constant flw b o u n d q condition [vas specified and we

were able to obtain an anaiflical solution. The corresponding problem wit h constant

temperature specified at the 1vaU is more difficult to sohe and hence a compIete as-

alyticai solution is not possible as in the previous section. However because of the

discontinuity due to the boundary conditions at the entrance of the thermal region

of the tube the use of the Laplace transform seerns appmpriate for this ~robIem as

well as the following integral transform rnethod.

In the foilowing the General Integral Transform technique ['Z3]. ['XI. ['25] [30]-[X]

is discussed which \ d l be used to solve the convection equation with pulsatile Bow.

This technique is a hybrid rnethod [34]. [43].[44] applied to the convection diffu-

sion equation.

Consider the following general transient Iinear convection diffusion problem in the

Page 78: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

where cc. I<. d. P have continuous partial derivat ives in some closed bounded volume

C'. rvith the initiai condition

and b o u n d q conditions

The technique of the generalized integral transform method is to start by choos-

ing the foilowing au'ciliary problem.

tvi t h boundary conditions

a d it is assumed that the solution of the above auxiliary problem is known.

Page 79: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

tif

-

The appropriate integral transfonn pair obtained from this e igendue problem is

Inverse:

Operating on the general transient convection-diffusion equation with the operator.

we obtain

Page 80: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

- where the function g i ( t ) can be shown to be

The integrai appearing in the transformed convection-diffusion equation is cdculated

by making use of the inversion formula to give the foiiowing system of coupled ordi-

nùry differentid equat ions

where the matrix -4; as a function of t is

and the normaiization integral. Xi, is given by

Page 81: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

The initiai condit ion equation is also transformed using the operator

which gives us the following

These equations. are an infinite system of coupled ordinary differential equations

for the transformed funct ions. ie the 'S.

Since the system is infinite it is not always possible to obtain an exact solution.

If this system can be solved however. the inversion formula above can be used to give

LIS the solution T(.E t ) of the convection-diffusion equation. In practice the system to

be solved can be tmncated at the N'th row and column. with N siifiiciently large for

the required accurac. and then the fuute system can be solved by standard numerical

techniques. [3S]

In matrix form. the tmncated version of the system becornes

Page 82: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-4 special case occurs when the velocity u in Eq. 4.11 is independent of t . In this

case the matrix of coefficients. -4. becomes constant. and the solution of the system

of differential equations above can be expIicitly written as the following

where the eaponential marrix can be computed once the eigenvalues and eigenvectors

of -4 have been obtained through solving the dgebraic problem

Scientific subroutines are available to accurately accornpiïsh this ta&.

In applications obtaining an approximate explicit solution c m be very important

and of special interest in getting qualitative insight into a problem. -4 Iowest order

Page 83: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

solut ion can b e O btained provided the non-diagonal elements of the coefiicients matrix

-4 are neglible as compared to those of the main diagonal tvhich wouid then approx-

imately correspond to a decoupled system. Therefore by keeping o d y the diagonal

elements of the matrix an appro-cimate solution is obtained from the fouowing initial

d u e problem

This type of problem arises for the case of heat transfer ivith pdsatile flow and

constant wall temperature which is considered in the next section.

Page 84: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

4.2 Heat %ansfer in Pulsatile Flow with Constant Wall Temperature

In t his section the convection equation with pulsatile Bow and constant temperature

prescribed dong the wall of the tube is solved using the integral transform technique

discussed in the previous section. The governing equation derived from Eq. 4.1 be-

cornes

where 0 = ( T - TW)/(To - Tw). T, is the constant wall temperature of the tube

and To is the constant temperature of the fluid at the thermal entrance region. Refer

to Fig. 4.0 for a sketch of the model.

Page 85: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Fig. 4.0

-4s in Chapter 2 the velocity u satisfies the Navier Stokes equations for full44 developed

axiai flow with the pressure gradient set equd to a harmonic of some Fourier series

in time, It foilows as before that

Taking the Laplace transform of Eq. 4-18 with respect to x. that is operating on

this partial differential equation with Jo e-"'( -)dx mhere s is cornplex and Rs >

for some 7 positive . we get

where we have integrated by parts and used the boundary condition at x = O .

Yow using the Generalized Integral Transform technique [4l].[Q]. Ive introduce the

Page 86: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

following alLui1iary problem ivhich is independent of the parameter s and whose solu-

tion is the zero order Bessel hnctioa as a function of 3 where the 3 s are the zeros

of JO.

1 d dd($, - - [r rdr dr

"1 + d211r(13. r ) = O

The following integral transform pair is dehed as:

Eq. 4.19 is now operated on wieh the operator

Page 87: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

to obtain the foIloiving:

Yow. using the auxiliary problem the Laplacian on the right hand side can be written

as

From this we obtain the following initial value problern with s cornplex.

-L L 9; (O) = -

S

Page 88: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

where the matris A , is caiculated from the following

and the function .\;, is simply the n o m so that system of eigenfuoctions becomes

an ort honormal system.

Recall in the discussion of the generaiized integral transform technique that if the

rnatrix -4 is diagoody dominant then a lower order solution is possible. The main

difficulty is that the matrix -4 is infinite and hence an infinite system of coupled dif-

ferential equations in s arises for which the solution is impossible to obtain. In what

follows. to make the problem tractable. we assume that the off diagonal terms of the

matris =Li j can be neglected. This is clearIy an approximation to the solution of Eq.

4. LS. Based on t his approximation. we arrive at the following system of O D E k

-L 1 6; ( O ) = -

S

where u is left general for now. .ln existence and uniqueness theorem for this type of

problem is found in Appendix D.

Page 89: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

4-3 Distributions

The solution to Eq. 4.26. a s wiU be shown Iater involves a generalized solution b o w n

as Dirac's distribution-

P hysicai quant it ies are usuaiIy t hought of as funct ions. But in an experimental set ting

it is quite difficult if not impossible to observe instantaneous values of the function at

every instant of space or time. A measuring instrument would record the effect chat

the function produces on it over some interval of space or time with nonzero length.

An alternate description for some physical quantity is to specify it as a functional. a

rule which assigns a number to each function in a set of testing functions. Continuous

linear functionals d e h e d on the space of testing functions are cdled distributions.

[-61 The distribution theory is a very powerfd tool for studying linear partial differen-

tial equations mith smooth kariable coefficients. It t m s out that using distributions

allows for anal!-zing types of physical phenornena more n a t u r d y than using the func-

tion concept. The Dirac delta function is an example of a distribution rvhich aises

naturaily in physical settings in particular in the following section.

The delta function ( S ) is defined hy

where ~ ( t ) is a test function. an infinitely smooth function that vanishes outside a

finite interval.

The delta function 6(t - r). selects the value of a test function at the point t = T.

Refer to the Appendix E for some important identities involving the delta function.

Page 90: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

4.4 Method of Solution

From Eq.4.26. the infinite system of cornplex o r d i n q differential equations in s with

initiai condition in s is

The solution of this system is

Let

Page 91: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Refer to Appendix E for the inverse Laplace transform of the exponential function in

s, Kence we obtain

where CS is Dirac's generalized function.

As in the case of a Green's function the intenal is extended onto the real axis by

introducing the Heaviside function as a kernel.

Page 92: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

From t his we obtain

ivhere a formula for the delta funct ion as a funct ion of a general argument has been

used from Appendix E and the phase g is

If Ive Let the phase equal a constant then a set of critical points is defined in the

Page 93: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

- - phase plane (r. T ) . Since the velocity also depends on frequency a set of critical fre-

quencies eBst in the phase plane.

The sequence of terms ~ i ( t . t ) are the zeros of the function g(r), assuming they eGst

for a given function - A i i ( r ) .

Tables of zeros. Z and velocity are included for different axial d u e s as well as for

different values of time in a cycle. Various velocities in puisatire flow are considered.

Phase plane surface plots for frequency variation are included.

In the foilowing section the results for pulsatile flow are discussed. Recail from the

integral transfonn pair

The series in the solution for 8 is unifonnly convergent since r j < t and thus the

inverse Laplace transform of gL ( r. t. s) exists.

From the generalized integral transform method. the solution for forced convection

heat tramfer ivith a general velocity term u(t ) becomes

After t e m s rlii(r) concel out we are left with the following expression for temperature

difference which depends explicitly on the zeros of g.

Page 94: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

rvhere rve are surnming up over a l the zeros of g.

Page 95: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

4.5 Results and Discussion

Ln this section a bulk temperature is defined as

-4 solution for the corresponding steady problem was given in chapter 2: the bulk

temperature in that case is given by

where

As in the constant flux solution. a measure of the effect of pulsations on the overall

heat transfer rate is required. The following ratio of steady buik temperature to un-

steady buik temperature will be considered.

o b s i=- @h

From the steady state solution of Chapter 2 and the generalized integrai transform

Page 96: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

technique of t his chapter rve have

It can be shown chat there exists a zero ri = r of g ( ~ ) for each z. t for the pul-

satile flow u ( t ) = I + c r c o ~ ( ~ f ) ) where g is specified in Eq. 4.35

In fact ive can generaiize this to the case of an arbitrary Fourier series in t . It

can be shown that there always exists a zero of g(r) at any axial distance dong the

tube and at any t ime t in a complete cycle.

If there is only one zero then

For some r positive and

Since the 7,'s are cornputable and positive then the following inequaiity results

Page 97: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

This proves that i is strict- less than unity and thus we have heat transfer en-

hancement due to the pulsations in the flow field.

Yumericd integration techniques can be used to calculate i for different foms of

w-

It is also interesting to note that if we define the bulk temperature as in Eq. 3.56,

3.5; of Chapter 3 then we have heat transfer enhancernent as well.

Page 98: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

- Table 4.1: Zeros 7.- of wave function and pulsatile velocity u = 1 + cost: t= ; ~ / 4

x

0.1

0.2

0.3

0.3

0.5

Zeros r

0.727.5

0.6709

0.6153

0.360'7

0.5069

- - = - 1

. -1 - -3

-4

. 5

- - - - -

Velocity u ( r )

1.0953

1, LS22

1.26198

1,3334

1.4033

Page 99: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

a 1 Zeros r 1 S ( Velocity u ( r ) / ,

Table 4.2: Zeros r.S of wave function and pulsatile velocity u = 1 + cost:t = 3a/4

Page 100: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Table 4.3: Zeros r. Z of wave function and pulsatile velocity u = 1 + cost :t=3 r / 2

Page 101: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

x Zeros r E 1 Velocity u ( r )

Table 1.4: Zeros r . of wave function and pdsatile velocity u = 1 + cost: t = Zn

Page 102: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

- Table 4.5: zeros ( r ) . = of wave function and pulsatile veiocity u =

Page 103: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

- Table 4.6: zeros ( T ) , = of wave function and pulsatile velocity u =

Page 104: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

x Zeros T S Velocity u

0.1 2.2127 . I A013 - 0.2 12.1332 .- -4668

--

0.3 2.0598 -3 -5302 i I

- Table 4.7: zeros (r) ' ; of wâve function and pulsatile velocity u =

L + cost + sint : t = 3x12

Page 105: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Phase p h e sur faces irequencies

Page 106: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 107: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Chapter 5

Concluding Remarks

In the first part of the thesis. a method of anaiysis is presented for the problem of heat

transfer in pulsatile flow with constant heat transfer in a tube. -4 regdar perturbation

expansion is used to solve for the temperature field downstream. A cornplex-valued

Green's function is utilized to brmuiate a b u k temperature and a relative Nusselt

number. Since heat transfer from the tube is constant the effect of pulsation only

occurs by means of a change in the b u k temperature within the tube. The resuits

indicate that a pulsating Bow field enhances the overd heat transfer rate. In par-

ticular there exists a critical value in frequency where the relative Nusselt number

attains a maximum value. The trend in relative Nusselt number with Prandtl number

is that there is an increase with decreasing values of Prmdtl number less than unity

and a decrease mith increasing values greater than unit- The perturbation param-

eter plays an important role in the mechanism of heat transfer with pulsatile flow.

The results indicate that increases of pulsatile pressure gradient amplitude over that

of steady pressure gradient amplitude give rise to increases in relative Nusselt number.

In the second part of the thesis. ac approxïrnate solution is presented for the problem

of heat transfer in pulsatile Borv with constant wall temperature in a tube. Using

the generalized integral transform method and the Laplace transform a system of

ordinaxy cornplex differential equations arises tvhose solution is presented with the

aid of Dirac's distribution t h e o . A bulk temperature of this solution is formulated . - and an increase of unsteady buk temperature over that of steady bulk ternperature

is presented.

Courant and Hilbert [45] have defmed a prog-ressing plane wave as a solution of the -

fonn u = f(S) where J = x - d is the phase of the solution and f is any function.

The form of Eq. 1.35 indicates that rve have a plane wave propagating down the tube.

Page 108: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

For future rvork any possible connections that can b e made between the convection

equation and the wave equation fiom this solution can be studied. The existence of

mave-like phenornena could prove to be usefd in describing physicdy the increase in

Yusselt number of pulsatile flow over that of steady Bow. .Us0 the frequency chacac-

teristics for the convection equation coupled to the full Navier-S tokes equations with

a general sinusoidal pressure gradient can be investigated. In addition solutions us-

ing the generalized integrai transform technique can be studied for the Xavier-Stokes

equations without neglecting off diagonal t e m s in the method.

Page 109: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Appendix A

Maple Code for the Graetz Problem

# Program to calculate Eigenfunctions for Graetz Problem

# and to calculate Temperature Distribution t heta

# R:=array(U-.3.l..fOl): # Subroutine to calculate point of intersection of graphs

# of funetions B(delta) and G(de1ta)

My'iewton := proc(rn.delto)

deltold := delto:

for i to rn do

deltnerv:= deltoid - evalf(H(de1told) )/evalf(K(deltold) ):

deltold:=deltnew:

od:

end:

# Subroutine to calculate the matching point eps of the # solution

# of the asymptotic formula and the sirnilarity transformation

# solution of the Graetz problem.

'iIyNewton2:= proc(s.eps10)

epslold := epslo:

for i to s do

epslnew := epslold - ed(f(epslold))/evaIf(fp(epslold));

epslold:=eva.lf(epslnew ) ;

od:

end:

# Dimensionalize array for successive approximations via Newton's # Method

deltt:=array(l..3):

delto:=0.11:

# Loop calculates the Eigenfunct ions for the Graetz problem

Page 110: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

# for n=0..3 using Lagrangian Interpolation.

for n from O to 3 do

delta:='delta':

Digits:=I6:

lamda :=4'n+d/3:

G1:= delta - > sqrt(2/(PiXlamda"delta)); G2:= delta - > cos((lamda/Z)*(delta*sqrt(1- deltaz) + atcsin(delta))-(Pi/4))/(1 - delta2)(1/1);

G:= GI"G2:

E:= x - > exp(-lamda' * c); FI:= delta - > diff(G(delta)?delta);

B : = delta - > Bessel J(O.larnda'de1t a):

F-:= delta - > diff(B(delta).delta):

H:= unapply( (G-B) (delta).delta);

K:= unappl-( ( F 1-F'I)(delta).delta);

F3:= delta - > sqrt(2'(I-de1ta)/3)*(-l)"'Besse1J(1/3.1amda"sqrt(S)' ( 1-delta)(3/'2)/3)

F4:= delta - > diff(F'3(~ielta)~delta);

L:= unapply( ( G-F3 )(delta) .delta):

M:= unapply((F1-F.L)(delta).delta): SI:=plot(B(delta).C;(delta).F3(delta),delta=O..l.titIe=-R(O)'):

m:=10:

Boolean BIock assigns array values deItt[n] in Xewton's

Method according to the value of n.

if n = O then

intl:= 0.10:

int2:= 0.45:

int3:= 0.50:

int4:= 0.53:

int5:= 0.65:

int6:= 0,933:

elif n = I then

deltt [n]:=MyYewton(m.delto):

Page 111: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

delto:=deltt [n]:

int4:= 0,.53:

int5:= 0.65:

int6:= 0.935;

else

hdel:=evalf(delto/'l) :

delt t [n]:=MyNewton(rn+hdel):

delto:=deltt[nj:

int4:= 0.53:

int.5:= 0.65:

int6:= 0.935;

fi:

Lagrangian Interpolating p o l ~ o m i d s for interpolation of

functions B(de1ta) G(de1ta). F3(delta) on [O. 11.

Quadratic fit

L 1 := delta - > ( (delt a-int2)'(deIta-intJ) )/( (int 1-int.l)'(int 1-int3) ) ;

L2:= delta - > ((delta-int l )'(delta-int3))/((in%int l)*(int2-int3)):

L3:= delta - >((delta-intl) '(delta-int2))/((int3-intlint2));

L4:= delta - >((delta-int5)'(delta-int6))/((int.L-int5)'(intPintG)); L5:= delta - >((del ta- int4) ' (del ta- int6)) / (( int5-int~int6)) ;

L6:= delta - >((delta-int4)x(delta-int5))/((int6-int4)x(int6-int5)): writeto('eig4.dath):

if n = O then

k:=.Ol:

for i From 1 to 101 do

delta:=(i-L)"k:

if delta >= O and delta <= intl then

R[n.i]:= B(delta);

ri nt (delta,evalf(R[n.i]) ) :

elif delta > int 1 and delta <= int3 then

R[n.i]:= B(int l)'Ll(delta)+G(int2)*L3(delta)+

C(int3)*L3(delta);

print (delta.evalf(R[n.i]) );

elif delta > int3 and delta <= int4 then

Page 112: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted
Page 113: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

od:

Loop to print out 2 dim a a y R for n Tom O to -5 do

for i from 1 to 100 do print(ed(R[n.i]));

od;

od ;

Loop to print out temperature theta(x,r) vaild for n>=eps.

We can prove that eps is in the neighbourhood of 0.01

Define on array for theta(x?r)

p:=may(l..lOl);

q:=array(l..lOl):

theta:= arra+y(1..100.1..101);

writeto('3 D.datb):

k:=.01;

1:= 2;

j:=l:

for j from 1 to 50 do &:=l"j;

for i from 1 to 101 do i:=101:

delta:=(i-l)'k:

init:=O.O;

for n from O to 3 do

theta[i,i]:= init + ((2/Pi) * (-l)n * 6(2/3) * GAMMA(2/3) (4 * n + 8/3)( - 213) ) * R[n, il * exp(-(4 * n + 813)' * axi);

init:=thetab,i];

od; print (axi,delta,evaL€(thetab,i]));

od;

od: p[i]:=delta;

Page 114: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

q[i]:=ed(thetaDj]):

od:

od:

writeto(terminal);

writeto( 'out lg-dat') :

for j fkom 1 to 101 do

print( plil 7 q U ' I 1; od:

writeto(terminal);

Routine to calculate local Nusselt modulus for t emperat ure

distribution given above

C := n- > (-l)n * 2 * 6(2/3) * GAMM.4(2/3) * (4 * o + 8/3)( - 2/3)/Pi:

r:=array(1..3000);

1:=0*001:

init:=O.O;

for j from L by 10 to 3000 do wi:=j*l;

for n fiom O to 100 do

Nus := init + (-2 * C(n) * (- l )(n + 1) * 2(2/3) * (4 * n + 5 / 3 ) ( 1 / 3 ) / ( G ~ ~ ~ r \ ( 4 / 3 )

9 ( 5 / 6 ) ) ) * e q ( - ( 4 * n + 8/3)* * a.); init:= Nus:

od:

init:=0.0;

rb] :=(axi,eMir(Nus));

od;

writeto('Nus2.dat');

for i from 1 by 10 to 3000 do

print(r[i]);

od; writeto(termina1) ;

Routine to coldate local Nusselt modulus for the temperature

distribution theta for constant heat flux q.

Page 115: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

!-:=am-( 1 .-1000):

0~l:=25.639;

agn2:=Y4.624:

gm3:=1'76.40;

H1:=0.008854:

H2:=O1002O62;

H3:=OO0009435:

1:=0.001:

for j from 1 by 10 to 1000 do &:=j'l:

?Iu:=1/((11/4S) + (1/2)*(exp(-gml"axi)/(-gm12 ' Hl)+ exp(-gm2 ' axi)/(-gd2 ' H2)+

exp(- gm3' axi)/(-gma2 ' H3)):

yij]:=(axi.evalf(Nu));

od:

print('Local Nusselt Valueso):

wri teto(cNuss.dat6);

lor j from L by 10 to 1000 do

print (Y Li] ) ; od:

writeto( terminai):

Page 116: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Appendix B

Maple Code for Constant Heat Flux Problem

writeto('output'):

hsum := proc(w.Pr,N):

Iambda := s ~ ~ ( w ) V ~ / ~ :

alpha := sqrt ( w / ~ r ) * i ~ / ~ :

ulc := convert(series(BesseIJ(0Jambda*r) .r=0~m):

u2c := convert (series(Besse1Y (0,lambda'r) , r = O , N o m ) :

ulprime := diff(u1c.r):

u2prime := diff(u2c.r):

b := evalc(subs(r=i.u'lprime)):

u l tc := convert (series( BesselJ(Otaiphaxt ),t =O.N),poiynom):

u3tc := convert (series(BesselJ(OJambdart),t=O~N)~pol~~om):

u4tc:= convert (series(Besse1 J(O.alphatr) ? r = O . N ) m ) :

u2tc := convert(series(BesselY(O~IambdaLt) . t=O~m):

c l := evd(sum (((-1). * (~qrt(w/Pr)/2)'~~)/(factorial(2 * n ) 2 ) , n = 0.25)) :

c2 := evaif(sum (((-1)" * (~qrt(w/Pr)/2)'~+~)/(factoria.l(2 * n + I ) ~ ) , n = 0.25)) :

c := c l +I*cS:

hl l :=Re(edc(-b/a)):

hl2 := evalc( Pi*~~rt(~)*0.5*1~/~)*~1~):

h l 3 := edc(int((u3tc/(I*w))'(l-ultc/c),t=0..l)):

h l := evalc(hll*hl2*hl3):

ha := ulc*Pi*sqrt(w)*0.5* I3l2: hb := int((u2tc/(I*w))*(l-ultc/c),t=r..l):

h2 := evalc(ha*hb):

hc := ~2c*Pi*s~rt(w)*0.5'1~/~:

hd := int((u3tc/(I'~v))*(l-ul tc/c),t=O..r):

Page 117: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

h3 := evalc(hc'hd):

h l + h3 +h3 :

h := evdc("):

ha1 := Re(h): ha2 := Im(h):

g := ( l/(I'm))'( 1-u4tc/c):

u := edc(g):

u l := Re(u):

u2 := Im(u): part0 := ulWhal:

part':, := uPha2:

part I := evalf(ht((part0) * (r), r = 0.J. 5. NCrule) ) :

part3 := e d ( h t ( ( p a r t 2 ) * (r). r = O--1.5. NCruie)) :

#print ('the b d k temperature isœ):

Yx(part 1) + Yx(part3):

end:

for i from 1 to 10 do

P := i*0.5:

for j from 1 by 10 to 50 do

print ('The frequency wb:The Prandt 1 No-?*The b u k temperature is ' ) :

print(j IP.hsum(j.P.25)):

od:

od:

quit:

Page 118: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Appendk C

Axial Gradient of Temperature Downstream

In this appendix we show that the pa.rtial derivative of the zerogth order solution

Bo tvith respect to the x Mliable approaches a constant downstream. The result is

used in section 3.6 and makes it possible to obtain a solution of the first order problem.

To show this let the Laplace transform of t - to be :

If the wdl temperature is given by t,(z), using the principle of superposition the temperature difference t - to is expressed as the foliowing Stieltjes Integral

where 0 is a solution of the Graetz problern:

Define the foIlowing transforms

Page 119: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Applying the convolution theorem to Eq. C 2 ' and using the boundary conditions

above we get the foilowing

Let

A pplying the convolution t heorem and using the foilowing identities

we arrive at the foilowing

Letting

lIff(s) = C { F ( r ) } and g(s) = Ç{G(t)} then f(s)g(a) = C { H ( x ) } where H ( x ) = I: F(<)G(x-

W C

Page 120: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

t hen:

t hen

Yext ive let,

and

(C. 15)

Page 121: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Using the convolution theorem it can be shown that

k L - ' L { / = ~ ( = ) ~ ( z - t, r)dc} = C 1 { r ( s 7 r)-1

O ro

which irnplies directly t hat:

efore. if we are given the heat flux q, then the temperatur

( C . 18)

s given by

(C.20)

To End @(x, r ) we apply the following inversion formula and use residue theo- in

order to perform the complex integration * - - - - - - - -

?o(x, r) = Res(esfO(s. r)} at the poles of @(S. r)

Page 122: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Cornplex htegration is taken dong the Bromwich contour:

Fig. 3.2

R e c d t hat .

where A, = 472 + Y/3

Refer to section 2.3.1 and Table 2.1 for the eigenfunctions &, eigendues A, and constants C,.

This leads to:

t hen t his imp lies t hat :

Page 123: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Xow Q(s) and i y ( s ) have poIes at s = O and s = -A:. Therefore 8/9 hos p o l e

o d y at s = O and the zeros of %(s). Let 9: be nich that 4(-dd:) = O

Then the following holds:

The zeros of q ( s ) can ody be found nurnericdy.

'iow we compute the residues.

First we set.

Computing the residue Res[G. O] we get:

L Res[G. O] = lim -sesz[s C,&(l)-]

s 4 0 n=O s +A:

Thus the residue of G at zero is:

Page 124: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

-

W e wiU m d e use of the result to cornpute the residue of the entire expression

associated with o.

Since ik has a simple zero at -,Jm then the residue of G at 2 = -3; is

Consider the other part of the term

in Eq. C.27 that is

The residue of this part is

Hence rve obtain the following solution for c$(x. r )

where we can solve For 0: numericdy for ad values of m from the following

31f f. g are andytic at so and if f (so) # O and g has a simple zero at so, then Res [ f/g,so ] = f (so ) l9'h

Page 125: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

where An = 4n + 813

It can be shown that @(O) = 114 [46]

Now returning to the formula denved above for temperature t - to in Eq. C.20.

for q = const.

Therefore.

t hen

Page 126: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Appendix D

Picard-Lindeiof Theorem

In this section an existence theorem for first order differential equations of the form

given in Eq. 1.36 is given.

Consider the Cauchy Problem

where (7. €) E D is a nonempty open subset of R x C

.-\ssurne that f is continous on D and satisfies the Lipschitz condition

I / ( W - f ( t , ~ ) ISCR l x - y 1

on each rectrangle R c D and CR is a constant depending on the size of the rectangle.

Then there exists a unique solution to the Cauchy problem above on I = (r-a. r+a),

w here a, 6 are chosen appropriately s m d so t hat for the rectangle centered at ( r, c ) ,

and

Page 127: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

where

iM = max 1 f ( t , x ) 1

and the maximum is taken over ( t - x ) E ka.

Proof

LVe show that it is possible to define

- ( I r = 0.1.2 ....) for t E I = [r-a.r +a] and derive the estimates

for some c > O. This would prove that (&) is a Cauchy sequence in the normed

Linear space of continous complex valued functions C(7), with the uniform metric

nom. Nor since this space is known to be a complete metric space (ie. Banach

Space) t hen

dt -t d .for some . q5 E C(T),

Page 128: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Taking Limits of the functions & it foUows that

It is a simple matter to see that r$ provides the required solution.

W e prove the above rigorously

Define c i o on 7 using the initial condition already given.

The sequence of functions t$k defhed above continue to successively defhe O,. &, 4, . . . on T Trivial l~ ive have that

for ail t E 7. Hence. because of the definition of a . the c u v e ( t. ~ ( t ) ) . t E remains

in Ra,b and. in particular. the integral to be used in defining 01 exists. Induction is

used for all k. Thus assume that

Now 4k+l must be shown to satisfy the same inequality. But from the first step

of induction method? &+, is defined on Ï, and using the definition of the maximum

LM, leads to

Page 129: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

The next step is to estabiish the estimate

( k = 1,2, . . .) , t E 7. where c is the Lipschitz constant associated with R = ki.

Now using the Lipschitz inequality, it follows t hat

Hence the above inequality is proven by induction. Reapplication of this inequd-

ity gives:

Page 130: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Hence the sequence of hinctionri oc is a Cauchy sequence and the sequence must

converge to a sohtion d .

The uniqueness of the solution foIIows readily using the Lipschitz condition for f

Page 131: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

Appendix E

Properties of the Dirac Delta hnction

Page 132: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

REFERENCES

[II Richardson. The Transverse Velocity gradient near the mouths of pipes in which an alternating or continuous jiow of air is establisheà, Proc. Roy-Soc. 42.pt.l. (231) 1-15 1929.

[2] S-Uchida. The Pulsating Viscous Flow Supen?nposed on the Steady Laminar Mo- tion of hcompressible Fluid in a Circular Pipe. Tokyo Japaa. ZAMP Vol.VII. 1956

[3] J .R. Womersley Osnllatory Motion of a Viscous L iquid in a Thin- GValled E h t i c Tube - I: The Linear Approz. Phil.Mag (16) 199-221 1955.

[4] H.B. Atabek and C.C. Chang, Oscillatory jlow near the entry of a circular tube. ZAMP. (12).155201 1961.

[5] R-Siegel and iM-Perlmutter, Heat Transfer for pulsating Laminar duct flou. Trans. ASàIE J. Heat Trans 84 (2), 111-123 1962.

(61 D.O. Barnett and R.I. Vachon, An Analysis of convective heat transfer for pul- sating jlow in a tube. ProcAth Int. Heat Transfer Conference. Paris.1-11. 1970.

[7] R.Creff and P-Ancire. Dynamic and Conuectiue Results for a Developing Laminar unsteady Flow. Internationd Journal for Numerical Methods in Fluids. (5) ' 745- 760 1985.

[SI H.W.Cho and J.M. Hyun. Numerical solutions of Pulsating Flow and Heat Trans- fer Characteristics in a Pipe, ht.J.Heat and Fluid Flow. Vol 11. (4), 1990.

[9] H-Haneke. H.Laschefski, A-Grobe-Gorgemm a ~ d N.K. Mitra, Aduanced Com- putationai Methods in Heat Transfer II, VoL.2: iVatural/Forced Convection and Combustion Simulation, Editors: L.C. Wrobel, C.A. Brebbia, Wessex Institute of Tecnology, University of Portsmouth. UK . A.J. Nowak, Silesian Technical University, Poland. Computational Mechanics Publications, 287-298, 1992.

[IO] S.Y.Kim. B. K. Kang and J. M. Ryun. Heat Transfer in the thennally deueloping region of a pulsating channel jlaw, int. J. Heat Mass T r d e r . Vo1.36. No.17. pp.4257-4266, 1993.

[Il] T.E. Base. M.R. Campbell and J.Hobbs. The effects of a pulsating flou, on the forced convection heat transfer inside a circular heated tube. Proceedings 15th Canadian Congress of Applied Mechanics.Cancam 95,Vo1.2,pp.612-613,1995.

Page 133: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

[12] M.Fallen. Heat Transfer in a Pipe udh Superimposed Pulsating Flow. CVame- und Stofibertragung 16. 59-99 1982.

[13] L.G.Genin. A.P. Loval. S-P-Manchkha. V-G-Sviridov, Hydrodynamics and heat transfer tmth pulsating fluid flow ira tubes, Thermal Engineering 39 5 1992.

[Ml J.Hapke. CVameu6ergang bei pulsierender larninarer Stromung, Bremsti Warme-Kraft 26 1974.

[15] P.hdre. R.Creff and J-Crabol, Etude des Conditions Particulieres De Frepuence Fauorisant Les Transferts Thermiques En Ecoulements Pulses En Canalisation Cylindrique. Int.J.Heat Mass Transfer. Vol 24, No7 12114219,1981.

[16j R. Peattie and R-Budwig, Heat Transfer in laminar. osdatory Jow in cylindn- cal and conical tubes, Int. J. Heat Mass Transfer Vol32 No5 923-934. 1989.

[lï] M.Vm D yke Perturbation Methods in Fluid Mechanics. The Parabolic Press, S tandford. California. 197-5.

[IS] L. Graetz. Uber die Wameleitungsfahigkeit von Flwsigkeiten.. Annalen der Physik Chem.. ~701.25.pp.335-357,1885.

(191 W. Nusselt. The dependence of the heat-twnsfer coement on tube length. Zeit. Ver.deut. hg.? 54. 1154-1158 1910.

r20] A. Leveque. Les lois de la transmission de chaleur par convection. . h d s Mines. Paris- Mem. Ser. 12. 13. 253-290 1925.

pl] J.R. Sellars. M. Tribus, and J.S. Klein, Heat Transfer to Laminar Flow in a Round tube or Flat Conduit- The Graetz Problem Eztended, Trms.Am.Soc. Mech.Engrs.. iS.UlJ956.

[-21 Lars V. Ahlfors.Complex -4nalysis.an introduction to the theory of analytic func- tions of one complez variable. McGraw-Hill , second edition.

[23] R.M. Cotta. Integral Transfonns in Computational Heat and Fluid Flow. CRC Press, Boca Raton 1993.

[-41 R.M. Cotta, Benchmark results in computational heat and Puid Jowrthe integral transfom method. 1nt.J.Heat Mass Transfer. Vol.37. Supp1.l. pp.381-393?1994.

[25] R.M.Cotta and M.D. Mikhailov,The integral transfom method, Appl. Math. - - Modeliïng 17, 156461,1993.

[26] A.B.Zernanian. Distribution Theory and Transfom Analysis: An Introduction to Generalixd Functions, toith Applications. Dover Publications. Inc..1987.

[07] A.V. Luikov. Heat and Mass Transfer.. Mir Publishers. Moscow.!973.

[28] M.D. Mikhailov and M.N.Ozisik. Wnijied analysis and Solutions of Heat and Mass Diffusion. Wiley. New York.1984.

Page 134: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

[-191 R-34-Cotta. Co.mputationa1 integml transfom approach in nonlinear diflusion and convection-diffusion problems. Laboratono d i Ingegneria Xucleare di Mon- tecuccoiino, Sene Scientifica LIN-1202. Universita degli S t udi di Bologna. Itdy, J d y 1993.

[:3O] M-D-Mikhailov and R.hl.Cotta, Cinified integral transfonn method. J . Braz. soc. Mech. Sci., RBCM, 12(3).301-310 1990.

[Ji] RSerfaty and R-M-Cotta. Integral t ransfom solutions of d i fw ion probiems urith nonlinear equation coeficients, Int.Comm.Heat Mass Transfer 17. 552-564 1990.

[32] R.M. Cotta and RSerfaty, Intepal transfonn algorithm for pambolic di/- m o n probiems with noniinear boundary and equation source terms, Proc. of the 7th lnt.Conf.on Num-Meth-for Thermal Problems. part9,pp.9& 926,Stanford.CA,July 1991.

[33] R.M.Cotta. R.Serfaty and R.O.C.Guedes. Integral t ransfom solution of a cZass of tmnsient convection-diffwion problems. Proc. of the Int. C o d on Advanced Computational Methods in Heat Transfer. Heat Transfer 90. Vol.l.pp.234 250.Southampton.UKJuly 1990.

[Ml RSerfaty and R.M.Cotta. Hybrid analysis of transient nonlinear convection- diffusion problems, int.J.Num.Meth.Keat Fluid Flow 2,5562 1992.

[35] -4.J.K. Leiroz and R.M.Cotta. On the solution of nonlinear elliptic convection- diffusion problems through the integral transfom nethod. Num-Heat Transfer. Part B-Fundamentals 23. 401311, 1993.

[36] J.S. Perez Guerrero and R-.CI-Cotta. Integral transfom method for Navier-Stokes equations in strearn-function only formulation. Int.J.Num.Meth.Fluids 15.399- 409 1992.

[ 3 7 ] R.M.Cotta.J.S.Perez Guerrero and F-Scofano Neto. Hybrid solution O/ the in- compressible Navier-Stokes epat ions via integral transformation. Proc-of the 2nd Int. C o d Advanced Computational Methods in Heat Tramfer-Heat Transfer 92. Vol. 1 .pp.735-i50,Mi1angItaly, July 1991.

[39] R.M.Cotta and MN. Ozisik. Diffusion Problems &th General Time-Dependent Coeficients, J.Braz.Assoc.Mech.Sciences, v.9.4,pp.264292,1987.

[4O] R.M.Cotta. -4. J.K. Leiroz and J.B.Aparecido, Steady-State Diffusion P d - lems with Variable Equation C o e f i e n t s . Int.J.Heat and Technology,~. lO.no.3- l,pp.lO4,l99Z

[31] R.M. Cotta and M.N.Ozisik, Transient Forced Conuection in Laminar Channel Flow with Stepwise Variations of Wall Temperature, Can. J . Chem. Eng., v.64. pp.734T42.1986.

Page 135: Te- · PDF fileHeat Transfer with Pulsatiie Flow in a Tbbe b~ Terry Moschandreou Department of Appiied Mathematics Submitted

[QI W.S.I<im and M.N.Ozisik. Transient Laminar Forced Convection in Ducts uiith Suddenly Applied L ~ n i f o m Wall Heat F[w, Lnt . J. Heat and Mass Trander. v.30. pp.1753-l756,1987.

[43] R.Serfaty and R.M.Cotta. Hybrid analysis of trunsïent non-linear conuection- diffusion problems, Int.J.Num.Meteth.Heat Fluid Flow 2,5562,1992.

[ R.M.Cotta and T.M.B. C d o . Kijbrid anal@ of boundary layer equations for interna1 flow problems, Proc-of the 7th 1nt.Conf.on Num. Meth. in Laminas and Turbulent Flow. Part l,pp.l06415,Stanford,CA,Jdy 1991.

[45] R.Courant and D.Hilbert, Methodf of mathematical Physics. Volume I I , Wi- IeyJ962.

[46] M.Jakob. Heat transfer John Wiley, Xew York 1949.