• Tidak ada hasil yang ditemukan

An approximate analytic solution of the Blasius problem

N/A
N/A
Protected

Academic year: 2025

Membagikan "An approximate analytic solution of the Blasius problem"

Copied!
4
0
0

Teks penuh

(1)

Short communication

An approximate analytic solution of the Blasius problem

Faiz Ahmad

a,*

, Wafaa H. Al-Barakati

b

aCentre for Advanced Mathematics and Physics, National University of Science and Technology, EME Campus, Peshawar Road, Rawalpindi, Pakistan

bDepartment of Mathematics, Faculty of Science, King Abdulaziz University, P.O. Box 15905 Jeddah 21454, Saudi Arabia Received 19 September 2007; received in revised form 26 December 2007; accepted 31 December 2007

Abstract

The [4/3] Pade approximant for the derivative is modified so that the resulting expression has the required asymptotic behavior. This gives an analytical result which represents the solution of the classical Blasius problem on the whole domain.

Ó2008 Elsevier B.V. All rights reserved.

PACS: 47.15.Cb; 02.30.Mv

Keywords: Viscous flow; Blasius problem; Analytical solution; Pade approximation

1. Introduction

The two dimensional steady state laminar viscous flow over a semi-infinite flat plate is modeled by the non- linear two-point boundary valueBlasius problem

f000ðgÞ þ1

2fðgÞf000ðgÞ ¼0; gP0 ð1:1aÞ

fð0Þ ¼0; f0ð0Þ ¼0; f0ð1Þ ¼1 ð1:1bÞ

wheregandfðgÞare, respectively, the dimensionless coordinate and the dimensionless stream function. Bla- sius[1]found the following analytic solution for the problem

fðgÞ ¼X1

k¼0

1 2 k

Akrkþ1

ð3kþ2Þ!g3kþ2; ð1:2Þ

whereA0¼A1¼1 and

1007-5704/$ - see front matterÓ2008 Elsevier B.V. All rights reserved.

doi:10.1016/j.cnsns.2007.12.010

* Corresponding author. Tel.: +92 3455334211.

E-mail address:[email protected](F. Ahmad).

Available online at www.sciencedirect.com

Communications in Nonlinear Science and Numerical Simulation xxx (2008) xxx–xxx

www.elsevier.com/locate/cnsns

ARTICLE IN PRESS

Please cite this article in press as: Ahmad F, Al-Barakati WH, An approximate analytic solution of the Blasius problem, Commun Nonlinear Sci Numer Simul (2008), doi:10.1016/j.cnsns.2007.12.010

(2)

Ak ¼Xk1

r¼0

3k1 3r

ArAkr1; kP2: ð1:3Þ

In(1.2)rdenotes the unknown f00ð0Þ:In spite of the presence ofð3kþ2Þ!in the denominator, the above series converges only within a finite interval½0;g0whereg01:8894r , thus making it impossible to use(1.2)to findrby applying the last condition in(1.1b)for a largeg. Howarth[2]solved the Blasius problem numerically and foundr0:33206:Asaithambi [3]found this number correct to nine decimal positions as 0.332057336.

Although the Blasius problem is almost a century old, it is still a topic of active current research[4–11].

The series(1.2)fails to converge outside a finite interval. This difficulty is inherent in several physical prob- lems governed by a nonlinear differential equation in that although a solution exists over an unbounded domain, but a power series representation of the solution converges only within a finite interval. For such problems finding an analyticalsolution which is uniformly valid over the whole domain is of fundamental interest. For the Blasius problem such a solution did not exist until 1999, when Liao, in a land mark paper, published a solution by using the homotopy analysis method[5]. His 35th order solution differs from How- ath’s numerical solution[2], forgP5, only in the fourth decimal position. However, Liao’s solution contains a large number of terms so that an explicit expression for the 35th order solution will require several pages to write upon.

In this short note, we derive ashortanalytical expression for the derivative of the solution. The idea is to form a hybrid expression which takes care off0ðgÞnot only for smallgbut also whenggrows very large. The

½4=3Pade approximant forf0ðgÞis slightly modified to the effect that the resulting expression represents the function on the entire domain½0;1Þwith remarkable accuracy. The actual solution is found by a simple quad- rature. A comparison with the numerical results shows that our solution gives accurate results over the entire domain. For largegour expression yields results in agreement with the numerical up to four decimal positions.

We remark that whereas Liao’s solution is completely analytical, ours is partly numerical in that we make no attempt toevaluater, but use its value as found numerically in[2,3]or analytically by Liao[5]or Ahmad [8]. Also we use the numerical resultfð7Þ ¼5:27924. Our emphasis is on finding a simple expression capable of producing accurate results over the entire domain½0;1Þ.

2. Uniformly valid analytic solution

We start with the analytical solution(1.2)with four terms fðgÞ ¼rg2

2 r2g5

240þ 11r3g8

161;280 5r4g11

4;257;792þ ð2:1Þ

On differentiation, we get f0ðgÞ ¼rgr2g4

48 þ11r3g7

20;160 5r4g10

387;072þ ð2:2Þ

We shall use a Pade approximant to represent the above expression[12]. A Pade [4/3] approximant of(2.2) gives

f0ðgÞ ’rgþ5603 r2g4

42011rg3 ð2:3Þ

In the above expression letr¼0:332057 and modify it by addingag5expðg421Þto the numerator as well as the denominator. Denote the new expression bygðgÞ.

We get

gðgÞ ¼0:332057gþ0:00059069g4þag5expðg42

1þ0:00869674g3þag5expðg421Þ ð2:4Þ

The above step is motivated by the idea that the functiongwill representf0with fair amount of accuracy for smallg and g will quickly approach unity as g becomes large. The parameter a is to be chosen in such a

2 F. Ahmad, W.H. Al-Barakati / Communications in Nonlinear Science and Numerical Simulation xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Ahmad F, Al-Barakati WH, An approximate analytic solution of the Blasius problem, Commun Nonlinear Sci Numer Simul (2008), doi:10.1016/j.cnsns.2007.12.010

(3)

manner that we get close agreement in sofar as possible in the transition from small g to large g. Also the choice of the exponential function is dictated by the following heuristic argument. Eq. (1.1a)can be written in the form

f000ðgÞ f00ðgÞ ¼ 1

2fðgÞ

An integration from 0 tog gives f00ðgÞ ¼rexp 1

2 Z g

0

fðuÞdu

ð2:5Þ For large u, f0ðuÞ !1, therefore, as an approximation we choose fðuÞ ¼u in (2.5) which gives f00ðgÞ rexpð12g2Þ:An integration fromg to1will give

1f0ðgÞ r Z 1

g

e12u2du ð2:6Þ

If we express the right side in the form of an asymptotic series and keep only the first term, we obtain f0ðgÞ 1þrexpð12g2Þ

g ð2:7Þ

An inspection of(2.4)shows that, for largeg,gðgÞ ¼f0ðgÞbehaves as required by(2.7).

If we chooseaso thatRb

0gðgÞdgis close to the value of the exact solution atbsuch thatgðbÞ 1, the inte- gral will continue to be close to the exact solution in ½b;1Þ. We arbitrarily selectb¼7 and find that with a¼2:88106

Table 1

Comparison of analytical and numerical results

g Approximate solution Numerical solution

0 0 0

0.4 0.0266 0.0266

0.8 0.1061 0.1061

1.2 0.2379 0.2379

1.6 0.4203 0.4203

2.0 0.6500 0.6500

2.4 0.9223 0.9223

2.8 1.2311 1.2310

3.2 1.5693 1.5691

3.6 1.9297 1.9295

4.0 2.3058 2.3057

4.4 2.6922 2.6924

4.6 2.8879 2.8882

4.8 3.0848 3.0853

5.0 3.2827 3.2833

5.2 3.4813 3.4819

5.4 3.6805 3.6809

5.6 3.8799 3.8803

5.8 4.0796 4.0799

6.0 4.2794 4.2796

6.4 4.6793 4.6794

6.8 5.0792 5.0793

7.0 5.2792 5.2792

7.4 5.6792 5.6792

8.0 6.2792 6.2792

10 8.2792 8.2792

20 18.2792 18.2792

100 98.2792 98.2792

F. Ahmad, W.H. Al-Barakati / Communications in Nonlinear Science and Numerical Simulation xxx (2008) xxx–xxx 3

ARTICLE IN PRESS

Please cite this article in press as: Ahmad F, Al-Barakati WH, An approximate analytic solution of the Blasius problem, Commun Nonlinear Sci Numer Simul (2008), doi:10.1016/j.cnsns.2007.12.010

(4)

Z 7 0

gðgÞdg¼5:27923

while the exact value of the solution, found numerically, is 5.27924. We substitute the above value ofain(2.4) and expect that the resulting expression will provide an approximate analytical expression for the derivative of the solution on theentiredomain½0;1Þ. A comparison of the analytical results with the numerical inTable 1 justifies this expectation.

Witha¼2:88106;gðgÞbecomes

gðgÞ ¼0:332057gþ0:00059069g4þ0:00000288g5expðg42

1þ0:00869674g3þ0:00000288g5expðg421Þ : ð2:8Þ

InTable 1, we compare the approximate resultsRg

0 gðuÞduwith the ones obtained by a numerical solution of the Blasius problem. We observe that the two results match to four decimal positions for 06g62:8:In the transition stage2:86g67:0 the two results agree to three decimal positions and even in this interval the max- imum error is less than two parts in ten thousand. ForgP7:0 the two results again match to four decimal positions. In view of this, we can claim that the functionRg

0gðuÞdurepresents the solution of the Blasius prob- lem on thewhole domain½0;1Þwith maximum error less than one part in five thousand.

Acknowledgement

Part of this work was done while the first author was at the King Abdulaziz University. He wishes to thank the University for its hospitality.

References

[1] Blasius H. Grenzschichten in Flussigkeiten mit kleiner Reibung. Z Math Phys 1908;56:1–37.

[2] Howarth L. On the solution of the laminar boundary layer equations. Proc London Math Soc A 1938;164:547–79.

[3] Asaithambi Asai. Solution of the Falkner–Skan equation by recursive evaluation of Taylor coefficients. J Comput Appl Math 2005;176:203–14.

[4] Liao SJ. An approximate solution technique not depending on small parameters part 2: an application in fluid mechanics. Int J Non- Linear Mech 1997;32:815–22.

[5] Liao SJ. An explicit, totally analytic approximate solution for Blasius’ viscous flow problems. Int J Non-Linear Mech 1999;34:759–78.

[6] He JH. A simple perturbation approach to Blasius equation. Appl Math Comput 2003;140:217–22.

[7] Fang T, Guo F, Lee CF. A note on the extended Blasius problem. Appl Math Lett 2006;19:613–7.

[8] Ahmad F. Application of Crocco–Wang equation to the Blasius problem. Electron J ‘‘Tech Acoust”2007:2. Available from:<http://

www.ejta.org>.

[9] Cortell R. Numerical solution of the classical Blasius flat-plate problem. Appl Math Comput 2005;170:706–10.

[10] Wang L. A new algorithm for solving classical Blasius equation. Appl Math Comput 2004;157:1–9.

[11] Ahmad F. Degeneracy in the Blasius problem. Electron J Differ Equations 2007;2007(92):1–8.

[12] Baker Jr GA, Graves Morris P. Pade approximants. Cambridge University Press; 1996.

4 F. Ahmad, W.H. Al-Barakati / Communications in Nonlinear Science and Numerical Simulation xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Ahmad F, Al-Barakati WH, An approximate analytic solution of the Blasius problem, Commun Nonlinear Sci Numer Simul (2008), doi:10.1016/j.cnsns.2007.12.010

Referensi

Dokumen terkait

This paper is organised as follows: Section 2.0 provides selected examples on numerical simulation using unstructured meshes, while Section 3.0 describes in detail the two

Therefore, we implement an adaptive finite difference moving mesh method as an alternative numerical method to solve the equation. The advantages of implementing the

Malaysian Journal of Science 42 1: 62-67 February 2023 https://mjs.um.edu.my AN INTEGRAL TRANSFORM TOGETHER WITH TAYLOR SERIES AND DECOMPOSITION METHOD FOR THE SOLUTION OF NONLINEAR

In this paper, we will use the inverse multiquadric radial basis function approximation for numerical solution of nonlinear mixed Volterra-Fredholm integral equations on a non-

Analysis of the Onset Process of Spontaneous Oscillations in a Standing Wave Thermoacoustic Engine, Using the Experimental Method and an Improved Numerical Solution Method..

2009 http://www.eurojournals.com/ejsr.htm Numerical Solution of the Boundary Layer Flow Over an Exponentially Stretching Sheet with Thermal Radiation Biliana Bidin Institute of

Application 1 - Least square method Feasible direction theorem Nonlinear Programs SincekAx−bk22 = Ax−bTAx−b =xTATAx−2bTAx+bTb, 11.18 can be formulated as the unconstrained problem