Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
FURTHER SOLUTIONS OF THE FALKNER-SKAN EQUATION
LAZHAR BOUGOFFAa, RUBAYYI T. ALQAHTANIb
Department of Mathematics and Statistics, College of Science, Al-Imam Mohammad Ibn SaudIslamic University (IMSIU), Riyadh, Saudi Arabia
E-mail a: [email protected] b: [email protected]
Received April 28, 2017
Abstract. In this paper, we propose a reliable treatment for the Falkner-Skanequation, which can be described as the non-dimensional velocity distribution in thelaminar boundary layer over a flat plate. We propose an algorithm of two steps that willintroduce an exact solution to the equation, followed by a correction to that solution.Three different special cases: Hiemenz flow (β=1), Homann flow (β= 1
2 ), and Blasiusproblem (β = 0) have been considered. When the pressure gradient parameter β takessufficiently large values, we use a transformation of variable that reduces the Falkner-Skan equation into an equivalent boundary value problem in a finite domain, and wesolve this problem by a different technique that allows us to find f ′′(0). Also, variousexact solutions can be obtained in a straightforward manner by using a direct methodwhen β = −1. The new technique, as presented in this paper, has been shown to bevery efficient for solving the Falkner-Skan equation.
Key words: Falkner-Skan equation; Hiemenz flow; Homann flow; Blasiusproblem; approximate solution; exact solution.
1. INTRODUCTION
The Falkner-Skan equation describes the class of the so-called similar laminarflows in boundary layer on a permeable wall, and at varying main - stream velocity [1,2]. This well-known nonlinear third-order differential equation is given by
f ′′′+ff ′′+β(1− (f ′)2
)= 0, 0< η <∞, (1)
subject to the boundary conditions
f(0) = γ, f ′(0) = 0 (2)
andf ′(∞) = 1, (3)
where f is a stream function, η is a distance from the wall, which is called similarityvariable. f ′ defines the velocity component in η-direction, f ′′ defines the shear stressin the boundary layer.The mass-transfer parameter γ in the boundary condition sets the measure for the
Romanian Journal of Physics 63, 102 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 2
mass flow rate through the wall boundary in either direction. Positive values de-termine flows with suction, negative with blowing through the wall boundary. Thezero value corresponds to flow along impermeable wall with zero mass transfer. Thenumerical parameter β (positive or negative) in the Falkner-Skan equation sets a de-gree of acceleration or deceleration of main stream [3]. The Falkner-Skan equation(1), together with (2)-(3), constitute the theoretical basis of convective heat and masstransfer. The existence and uniqueness question for the problem (1)-(3) was givenby [4] and [5]. Weyl [6] proved that for each value of the parameter β there exists aphysical solution with a positive monotonically decreasing, in [0,1), second deriva-tive that approaches zero as the independent variables goes to infinity. In the caseβ > 1, Craven and Pelietier [7] computed solutions for which df
dη < 0 for some valuesof η. Coppel [8] presented an elegant proof of the existence and uniqueness for β > 0and showed that the wall shear stress f ′′(0) is an increasing function of β. In theworks [9, 10], the extension for β < 0 was discussed and properties of solutions wereinvestigated further. New theoretical results for the solutions of the Falkner-Skanequation were given in Refs. [10] and [11]. Different methods have been proposedto solve this problem: finite difference [9, 12–14], shooting [15–20], finite element[21], group invariance theory [22, 23], Chebyshev spectral method [24, 25], diffe-rential transformation method [26], non-iterative transformation method (ITM) [27],homotopy analysis method [28], and Adomian decomposition method (ADM) [29].Other direct methods and numerical techniques to study nonlinear differential equa-tions of physical relevance have been recently reported [30–32].
To our knowledge, a general analytic solution of Eq. (1) satisfying the bound-ary conditions (2)-(3) does not exist. The purpose of this work is to present an al-gorithm of two steps that will introduce approximate solutions to the Falkner-Skanequation followed by a correction to that solution. Furthermore, we will considerthree special cases: Hiemenz flow (β = 1), Homann flow (β = 1
2 ) and Blasius pro-blem (β = 0). We use a transformation of variables that reduces (1)-(3) into anequivalent boundary value problem (BVP) in a finite domain, Finally, we solve thisproblem by a different technique that allows us to find f ′′(0), when the pressure gra-dient parameter β takes sufficiently large values. Also, various exact solutions canbe obtained in a straightforward manner by using a direct method when β =−1. Theresults lead to very accurate approximate solutions. Listed below are some specialcases when Eqs. (1)-(3) are solvable.
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
3 Further solutions of the Falkner-Skan equation Article no. 102
2. CASE 1: LARGE VALUES OF β
The Falkner-Skan can be written in an equivalent form utilizing m as theFalkner-Skan pressure parameter instead of β :
f ′′′+ff ′′+2m
m+1
(1− (f ′)2
)= 0, (4)
where 0 ≤ m ≤ ∞. The authors have considered only the case of 0 ≤ β ≤ 2. Forlarge values of m→∞, Eq. (1) becomes
f ′′′+ff ′′+2(1− (f ′)2
)= 0. (5)
This case has been studied in Ref. [5]. One of the main problem appears whenβ →∞.
In this section, we will present a direct approach to solve this problem whenthe pressure gradient parameter β takes sufficiently large values, i.e. β →∞.Let us consider the following transformation [33]:
z(x) = f ′′(η) and x= f ′(η). (6)
Differentiating Eq. (1) with respect to η, we get
f ′′′′+ff ′′′+(1−2β)f ′f ′′ = 0 (7)
and from Eq. (6), we have
z(x) =df ′
dη=
dx
dη, (8)
z′(x) =dz
dx=
f ′′′
f ′′ =−f −β1−f ′2
f ′′ (9)
and
z′′(x) =d2z
dx2=
f ′′′′
f ′′2 −f ′′′2
f ′′3 . (10)
Substituting these equations into Eq. (7), we obtain
z′′(x)−β(1−x2)z′(x)
z2(x)+(1−2β)x
1
z(x)= 0. (11)
Since f ′(∞) = 1 and f(0) = 0, dealing, for example, with a simple case γ = 0, wehave z(1) = f ′′(∞) = 0 and z′(0) = −β
σ and z(0) = σ, where f ′′(0) = σ. Hence,the initial-boundary value problem for the Falkner-Skan equation can be transformedinto the new problem{
z′′(x)−β(1−x2) z′(x)
z2(x)+(1−2β)x 1
z(x) = 0, 0< x < 1,
z(0) = σ,z′(0) =−βσ ,z(1) = 0.
(12)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 4
Multiplying both sides of the first-equation of (12) by ξ(x) = e∫ (1−2β)x
β(1−x2)dx
and takinginto account that ξ′(x) = (1−2β)x
β(1−x2)ξ(x), we obtain
ξ(x)z′′(x)+β(1−x2)
(ξ(x)
1
z(x)
)′= 0. (13)
We have
ξ(x) = e(1−2β)
β
∫x
1−x2dx
= (1−x2)1− 1
2β . (14)Thus
z′′(x)+β(1−x2)12β
(ξ(x)
1
z(x)
)′= 0. (15)
The function (1−x2)12β can be approximated by 1 for sufficiently large values of β,
and then ξ(x)≈ 1−x2.More precisely, we obtain the approximated equation
z′′(x)+β
(ξ(x)
1
z(x)
)′= 0. (16)
Integrating Eq. (16) from 0 to x and taking into account that z′(0) =−βσ and z(0) =
σ, we get
z′(x)+βξ(x)1
z(x)= 0, (17)
which can be written as follows
(z2(x))′ =−2βξ(x). (18)
Integrating Eq. (18) from 0 to x and taking into account that z(0) = σ, we get
z2(x) =−2β
∫ x
0ξ(x)dx+σ2. (19)
Consequently,
z2(x) =−2β(x− x3
3)+σ2. (20)
Thus, it is required to find σ by the given condition z(1) = 0. Then we have
σ2 =4
3β. (21)
Thus
f ′′(0) = σ =±2
√β
3. (22)
It follows that
z(x) =±√
−2β(x− x3
3)+
4
3β. (23)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
5 Further solutions of the Falkner-Skan equation Article no. 102
3. CASE 2: HOMANN FLOW (β = 12 )
For β = 12 , Eq. (12) becomes{
z′′(x)− 12(1−x2) z
′(x)z2(x)
= 0,
z(0) = σ,z′(0) =−βσ ,z(1) = 0.
(24)
We start first by noting that the assumption
z(x) = k, (25)
where k is a parameter to be determined, satisfies the first equation of (24). Usingthe boundary condition z(0) = σ, we obtain k = σ. However, this solution does notsatisfy both boundary conditions z′(0) = −β
σ and z(1) = 0. In view of this, we willemploy the given conditions together with (25) to achieve the second goal of ourapproach by making a correction to (25). We do this by writing the first equation of(24) in an equivalent form as
z′′(x) =1
2(1−x2)
z′(x)
z2(x). (26)
Inserting the obtained solution (25) in the right hand side (RHS) of Eq. (26), weobtain
z(x) = ax+ b, (27)where a and b are two constants of integration. Taking into account the boundaryconditions z(0) = σ, z′(0) =−β
σ =− 12σ , z(1) = 0, we immediately find that
a=− 1
2σ, b= σ (28)
with σ = 1√2= 0.7071. The value of the skin friction for Homann stagnation was
determined to be σ = 0.927682 [15].Returning to the original transformation, we get
f ′′(η) = af ′(η)+ b, (29)
so that
f ′(η) = ceaη− b
a(30)
and
f(η) =c
aeaη− b
aη+d, (31)
where c and d are two constants of integration. Using the boundary conditions (2)-(3), we obtain c=−2σ2 and d=−4σ3.Consequently, the approximate solution of the Homann flow problem is given as
fp(η) = 4σ3e−12σ
η−2σ2η−4σ3, σ = 0.7071. (32)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 6
3.1. THE ERROR REMAINDER
An error analysis ER(η) for fp(η) is considered in Table 1, where the residualER(η) or error remainder is calculated for different values of η by substituting fp(η)for f(η) in Eq. (1) when β = 1
2 :
ER1(η) =−1
2e−
12σ
η+σe−12σ
η(4σ3e−
12σ
η+2σ2η−4σ3)
+1
2
[1−4σ4
(1−e−
12σ
η)2
]. (33)
This demonstrates that the approximate solution is a good approximation of the actualsolution. The complete error analysis is given in Tables 1-4.
4. CASE 3: HIEMENZ STAGNATION FLOW (β = 1)
We note that the solution of the form
f(η) = βη+λ, (34)
where λ is a parameter to be determined, satisfies Eq. (1). Then we have
f ′(η) = β,f ′′(η) = f ′′′(η) = 0. (35)
Inserting this solution into Eq. (1) to get
β(1−β2) = 0. (36)
Thus β =−1, 1, 0. For the Hiemenz flow problem (β = 1), we have
f(η) = η+λ. (37)
Inserting now the boundary condition f(0) = γ we get λ= γ.We write Eq. (1) in an equivalent form as
f ′′′+ff ′′ =−(1− (f ′)2
). (38)
Inserting the obtained solution Eq. (37) in the RHS of Eq. (38), we obtain
f ′′′
f ′′ =−f (39)
Integrating Eq. (39) twice, we obtain
f ′(η) = c
√π
2e
γ2
2 erf
(η+γ√
2
)+d, (40)
where erf is the error function given by
erf(x) =2√π
∫ x
0e−t2dt. (41)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
7 Further solutions of the Falkner-Skan equation Article no. 102
Consequently,
fp(η) = c
√π
2e
γ2
2
[(η+γ)erf
(η+γ√
2
)+
√2
πe−
(η+γ)2
2
]+dη+e, (42)
where c,d, and e are constants of integration.Using the boundary conditions f ′(0) = 0 and f ′(∞) = 1, we obtain
c
√π
2e
γ2
2 erf
(γ√2
)+d= 0 (43)
and
c
√π
2e
γ2
2 +d= 1. (44)
Solving this system for c and d, we obtain
c=1√
π2 e
γ2
2 −√
π2 e
γ2
2 erf(
γ√2
) (45)
and
d=
√π2 e
γ2
2 erf(
γ√2
)√
π2 e
γ2
2 erf(
γ√2
)−√
π2 e
γ2
2
. (46)
The constant e can be determined from the condition f(0) = γ. Thus
e= γ− c
√π
2e
γ2
2
[γerf
(γ√2
)+
√2
πe−
γ2
2
]. (47)
Hence the approximate solution of the Hiemenz flow problem (β = 1) is given byEq. (42). Figures 1 and 2 show the approximate solution.
4.1. THE ERROR REMAINDER
An error analysis ER(η) for
fp(η) = ηerf
(η√2
)+
√2
πe−
η2
2 −√
2
π(48)
when γ = 0 is considered in Table 3, where the residual ER2(η) or error remainderis calculated for different values of 1≤ η <∞.
ER2(η)=
√2
πe−
η2
2
[−√
2
πη+ηerf(
η√2)+
√2
πe−
η2
2 −√
2
π
]+1−
(erf(
η√2)
)2
.
(49)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 8
0
0.5
1
1.5
2
2.5
0 0.5 1 1.5 2 2.5
fp(
η)
η
Fig. 1 – The approximate solution of the Hiemenz flow problem fp(η) when γ = 0.
Table 1
The error remainder ER4(η) for the value of γ = 0.
η ER4(η)
0 -0.1951342281 -0.07102780912 0.04203181485 0.123399024010 0.0390240252420 0.00111218682550 0.5342136800×10−8
100 0.2810156498×10−17
200 0.3725913644×10−36
500 0.2625174347×10−93
1000 0.6291376971×10−189
2000 0.1799367855×10−380
5000 0.1311231850×10−955
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
9 Further solutions of the Falkner-Skan equation Article no. 102
(a) β = 0
0
5
10
15
20
25
0 5 10 15 20 25
fp(
η)
η
Approximation solution Numerical solution
(b) β = 0.5
0
5
10
15
20
25
0 5 10 15 20 25
fp(
η)
η
Approximation solution Numerical solution
(c) β = 1
0
5
10
15
20
25
0 5 10 15 20 25
fp(
η)
η
Approximation solution Numerical solution
Fig. 2 – The approximate solution compared to numerical solution, γ = 0.
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 10
(a) 0≤ η ≤ 20
0
5
10
15
20
0 5 10 15 20
fp(
η)
η
α=0.0α=0.5α=1.0α=1.5α=2.0
(b) 0≤ η ≤ 5
0
1
2
3
4
5
0 1 2 3 4 5
fp(
η)
η
α=0.0α=0.5α=1.0α=1.5α=2.0
Fig. 3 – The approximate solution of the Blasius equation fp(η) when γ = 0 and α= 0, 0.5, 1, 1.5, 2.
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
11 Further solutions of the Falkner-Skan equation Article no. 102
(a) 0≤ η ≤ 30
0
5
10
15
20
25
30
0 5 10 15 20 25 30
fp(
η)
η
α=0.0α=-0.5α=-1.0α=-1.5α=-2.0
(b) 0≤ η ≤ 5
-4
-3
-2
-1
0
1
2
3
4
0 1 2 3 4 5
fp(
η)
η
α=0.0α=-0.5α=-1.0α=-1.5α=-2.0
Fig. 4 – The approximate solution of the Blasius equation fp(η) when γ = 0 and α =0, −0.5, −1, −1.5, −2.
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 12
(a) 0≤ η ≤ 30
0
5
10
15
20
25
30
0 5 10 15 20 25 30
fp(
η)
η
γ=0.0γ=0.5γ=1.0γ=1.5γ=2.0
(b) 0≤ η ≤ 5
0
1
2
3
4
5
0 1 2 3 4 5
fp(
η)
η
γ=0.0γ=0.5γ=1.0γ=1.5γ=2.0
Fig. 5 – The approximate solution of the Blasius equation fp(η) when α= 0 and γ = 0, 0.5, 1, 1.5, 2.
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
13 Further solutions of the Falkner-Skan equation Article no. 102
Table 2
The error remainder ER3(η) for different value of γ.
η ER3(η), γ=0.5 ER3(η), γ=1 ER3(η), γ=1.5 ER3(η), γ=20 0 0 0 01 0.3678794412 0.1353352832 0.04978706837 0.018315638892 0.5032147244 0.1536509221 0.05226582055 0.018651101525 0.3350779415 0.02699718793 0.002212643382 0.1816017802×10−3
10 0.06068692292 0.0004086014290 0.2753120978×10−5 0.1855038260×10−7
20 0.0008626007266 0.3916191882×10−7 0.1777948364×10−11 0.8071873084×10−16
50 0.6805092491×10−9 0.9450874255×10−20 0.1312532111×10−30 0.1822837228×10−41
100 0.1909462350×10−19 0.3682875216×10−41 0.7103345013×10−63 0.1370057562×10−84
200 0.1909462350×10−19 0.2753954089×10−84 0.1024491844×10−127 0.3811187498×10−171
500 0.1331925918×10−105 0.3555163627×10−214 0.9489407965×10−323 0.2532903490×10−431
1000 0.7117451831×10−214 0.5070882939×10−431 0.3612789294×10−648 0.2573959337×10−865
2000 0.1014684184×10−430 0.5150495210×10−865 0.2614370199×10−1299 0.1327043567×10−1733
5000 0.9176527672×10−1082 0.1684510105×10−2167 0.3092209160×10−3253 0.5676283841×10−4337
Table 3
The error remainder ER2(η) for the value of γ = 0.
η ER2(η)
0 11 0.3262573048
2 0.0482564534
5 0.1779034859×10−5
10 0.2187308194×10−21
20 0.3692604684×10−86
50 0.1010532795×10−541
100 0.5224985463×10−2170
200 0.4077397794×10−8684
500 0.1238358721×10−54284
1000 0.9223046463×10−217145
2000 05000 0
5. CASE 4: BLASIUS FLOW (β = 0)
5.1. γ = 0
For β = 0, we havef(η) = γ. (50)
Proceeding as before, inserting the obtained solution Eq. (50) in the RHS of Eq. (39),we obtain
f(η) =c
γ2e−γη+dη+e, γ = 0. (51)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 14
Table 4
The error remainder ER1(η) for the value of γ = 0.
η ER1(η)
0 01 0.2236783820
2 0.2518178278
5 0.889029848×10−1
10 0.55999659×10−2
20 0.29018999482×10−4
50 0.19179816079×10−4
100 0.19179816064×10−4
200 0.19179816064×10−4
500 0.19179816064×10−4
1000 0.19179816064×10−4
2000 0.19179816064×10−4
5000 0.19179816064×10−4
Using the boundary conditions (2)-(3), we obtain
c= γ, d= 1, e= γ− 1
γ. (52)
Consequently, the approximate solution of Eqs. (1)-(3) when β = 0 is given as
fp(η) =1
γe−γη+η+γ− 1
γ, γ = 0. (53)
If f ′(0)=α, where α is a constant, then c= γ(1−α). The initial condition f ′(0)=αindicates the slip condition at the wall. The case where α= 0 represents no-slip.
5.2. γ = 0
Proceeding as before, we obtain
f(η) =c
k2e−kη+dη+e. (54)
Using the boundary conditions (2)-(3), we obtain
c= k, d= 1, e=−1
k. (55)
Consequently, the approximate solution of Eqs. (1)-(3) when β = 0 is given as
fp(η) =1
ke−kη+η− 1
k. (56)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
15 Further solutions of the Falkner-Skan equation Article no. 102
To find the value of k, we use the transformation [33]:{z(η)z′′(η)+η = 0,z′(0) = 0,z(1) = 0
(57)
with z(0) = σ.Proceeding as in [34], we define the solution z(η) by an infinite series in the form
z(η) =
∞∑n=0
Cnηn. (58)
Thus
z′(η) =
∞∑n=0
(n+1)Cn+1ηn, (59)
z2(η) =
∞∑n=0
ηnn∑
k=0
CkCn−k (60)
and
z′2(η) =
∞∑n=0
ηnn∑
k=0
(k+1)(n+1−k)Ck+1Cn+1−k. (61)
The substitution into Eq. (57) yields
∞∑n=0
ηnn∑
k=0
CkCn−k = σ2− 1
3η3
−2
∞∑n=2
ηn
n(n−1)
n−2∑k=0
(k+1)(n−1−k)Ck+1Cn−1−k. (62)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 16
We equate the coefficients of the like powers of η on the left side and on the rightside to arrive at recurrence relations for the coefficients. Thus
C20 = σ2,∑1
k=0CkC1−k = 0,∑2k=0CkC2−k = −C2
1 ,∑3k=0CkC3−k = −1
3 +13
∑1k=0(k+1)(2−k)Ck+1C2−k,∑4
k=0CkC4−k = +13
∑2k=0(k+1)(3−k)Ck+1C3−k,∑5
k=0CkC5−k = +16
∑3k=0(k+1)(4−k)Ck+1C4−k,∑6
k=0CkC6−k = +16
∑4k=0(k+1)(5−k)Ck+1C5−k,
...
(63)
The coefficients C0,C1, ... are
C0 = σ,
C1 = C2 = 0,
C3 = − 16σ ,
C4 = C5 = 0,
C6 = − 1180σ3 ,
...
(64)
We can now compute the truncated decomposition series z(η) = z0(η) + z1(η) +z2(η)+ z3(η)+z4(η)+z5(η)+ z6(η). Consequently, the solution is given by
z(η) = σ− η3
6σ− η6
180σ3. (65)
Thus, it is required to find σ by the given condition z(1) = 0; then we get
σ− 1
6σ− 1
180σ3= 0. (66)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
17 Further solutions of the Falkner-Skan equation Article no. 102
Then, we get σ = 0.4417, that is f ′′p (0) = 0.44174. The numerical value of f ′′(0)
by the Runge-Kutta method is given as f ′′(0) = 0.46960. Finally, substituting thisvalue into Eq. (56), we obtain k = 0.44174. Consequently, the approximate solutionof the initial-boundary value problem for the Blasius equation when γ = 0, which ischaracterized by k = 0.44174, is given by Eq. (56).
5.3. THE ERROR REMAINDER
The error remainder ER(η) is calculated for different values of η :
5.3.1. Case 1: γ = 0
ER3(η) = e−2γη+e−γη(η−1). (67)
5.3.2. Case 2: γ = 0
ER4(η) = e−2kη+e−kη(kη−1−k2), k = 0.44174. (68)where the residuals ERi(x), i = 3,4 or errors remainders are calculated for smalland large values of η, which show that the present solutions are highly accurate.The figures 3-5 have been drawn to show the approximate solutions of the Blasiusequation.
6. EXACT SOLUTIONS OF THE FALKNER-SKAN EQUATION WHEN β =−1
The solutions for the value of β < 0 represent decelerating flows. We integrateEq. (1) from 0 to η, using the fact that
ff ′′ = (ff ′)′−f ′2. (69)
The substitution of Eq. (69) into Eq. (1) leads to
f ′′′+(ff ′)′ = 1. (70)
Integrating Eq. (70) from 0 to η and taking into account that f(0) = γ and f ′(0) = 0,we obtain
f ′′+ff ′ = η+f ′′(0). (71)Consequently Eq. (71) can also be integrated analytically, and we obtain the nonli-near Riccati equation
f ′+1
2f2 =
1
2η2+f ′′(0)η+
γ2
2. (72)
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 18
The first result in the analysis of Riccati equation (72) is: if one particular solution toEq. (72) can be chosen as follows f1 = η+f ′′(0), then f ′′(0) must satisfy
f ′′2(0)+2 = γ2, (73)
that is
f ′′2(0) = γ2−2, γ2−2≥ 0 (74)
and the general solution is obtained as
f(η) = f1(η)+1
w(η), (75)
where w satisfies the corresponding first - order linear equation
w′−f1w =1
2, (76)
and we can obtain its closed form solution
w(η) =
√π2 e
f ′′2(0)2 erf
(η+f ′′(0)√
2
)+2C
2e−η2
2−f ′′(0)η
, (77)
where erf is the error function and C is a constant of integration.A set of solutions to the Riccati equation is then given by
f(η) = η+f ′′(0)+2e−
η2
2−f ′′(0)η√
π2 e
f ′′2(0)2 erf
(η+f ′′(0)√
2
)+2C
, (78)
and therefore f satisfies the boundary condition (3), that is f ′(∞) = 1.Inserting now the boundary condition f(0) = γ into Eq. (78) to get
f ′′(0)+2√
π2 e
f ′′2(0)2 erf
(f ′′(0)√
2
)+2C
= γ. (79)
Consequently, this leads to
C =1
γ−f ′′(0)− 1
2
√π
2e
f ′′2(0)2 erf
(f ′′(0)√
2
). (80)
Thus we have proved.Lemma 1
The solution of the Falkner-Skan equation when β = −1 subject to the boundaryconditions (2)-(3) can be exactly obtained by Eq. (80), where C is determined by Eq.(79) and γ2−2≥ 0.
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
19 Further solutions of the Falkner-Skan equation Article no. 102
7. CONCLUSION
In this work, we have considered the Falkner-Skan equation, which can bedescribed as the non-dimensional velocity distribution in the laminar boundary layerover a flat plate. Very good approximate analytic solutions are obtained by a directmethod when β takes large values. Also, three special cases: Hiemenz flow (β = 1),Homann flow (β = 1
2 ), and Blasius problem (β = 0) have been considered. We havealso demonstrated that the exact solution can be obtained in a straightforward mannerby using a direct method when the pressure gradient parameter takes the value β =−1.
REFERENCES
1. V. M. Falkner, S. W. Skan, Some approximations of the boundary layers equations, Philos. Mag.Soc. 12, 865-896 (1931).
2. H. Blasius, Grenrschichten in Flussigkeiten mit kleiner Reibung, Z. Math. u. Phys. 56, 1-37 (1908).3. A. Solodov, V. Ochkov, Differential Models An Introduction with Mathcad, Springer (2005).4. D. R. Hartree, On the equation occurring in Falkner-Skan approximate treatment of the equations
of the boundary layer, Proc. Camb. Phil. Soc. 33, 223-239 (1937).5. K. Stewartson, Further solutions of the Falkner-Skan equation, Proc. Camb. Phil. Soc. 50, 454-465
(1954).6. H. Weyl, On the differential equation of the simplest boundary-layer problems, Ann. Math. 43,
381-407 (1942).7. A. H. Craven, L. A. Pelietier, Reverse flow solutions of the Falkner-Skan equation for λ > 1,
Mathematika. 19, 135-138 (1972).8. W. A. Coppel, On a differential equation of boundary layer theory, Philos. Trans. Roy. Soc. London
Ser. A. 253, 101-136 (1960).9. A. Asaithambi, A finite-difference method for the solution of the Falkner-Skan equation, Appl.
Math. Comput. 92, 135-141 (1998).10. G. C. Yang, New results of Falkner-Skan equation arising in boundary layer theory, Appl. Math.
Comput. 202, 406-412 (2008).11. O. Pade, On the solution of Falkner-Skan equation, J. Math. Anal. Appl. 285, 264-274 (2003).12. A. Asaithambi, A second-order finite-difference method for the Falkner-Skan equation, Appl.
Math. Comput. 156, 779-786 (2004).13. P. M. Beckett, Finite difference solution of boundary-layer type equations, Int. J. Comput. Math.
14, 183-190 (1983).14. B. Oskam, A. E. P. Veldman, Branching of the Falkner-Skan solutions for λ < 1, J. Engg. Math.
16, 295-308 (1982).15. A. Asaithambi, A numerical method for the solution of the Falkner-Skan equation, Appl. Math.
Comput. 81, 259-264 (1997).16. A. Asaithambi, Solution of the Falkner-Skan equation by recursive evaluation of Taylor coeffi-
cients, J. of Comput. Appl. Math. 176, 203-214 (2005).17. T. Cebeci, H. B. Keller, Shooting and parallel shooting methods for solving the Falkner-Skan
boundary-layer equation, J. Comp. Phys. 7, 289-300 (1971).
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f
Article no. 102 Lazhar Bougoffa, Rubayyi T. Alqahtani 20
18. C. W. Chang et al., The Lie-group shooting method for boundary layer equations in fluid mechan-ics, J. Hydrodynamics Ser. B, 18, 103-108 (2006).
19. C. S. Liu, J. R. Chang, The Lie-group shooting method for multiple-solutions of Falkner-Skanequation under suctioninjection conditions, Int. J. Non-Linear Mech. 43, 844-851 (2008).
20. I. Sher, A. Yakhot, New approach to the solution of the Falkner-Skan equation, AIAA J. 39, 965-967 (2001).
21. A. Asaithambi, Numerical solution of the Falkner-Skan equation using piecewise linear functions,Appl. Math. Comput. 159, 267-273 (2004).
22. R. Fazio, The Falkner-Skan equation: Numerical solutions within group invariance theory, Calcolo31, 115-124 (1994).
23. R. Fazio, A novel approach to the numerical solution of boundary value problems on infinite inter-vals, SIAM J. Numer. Anal. 33, 1473-1483 (1996).
24. E. M. E. Elbarbary, Chebyshev finite difference method for the solution of boundary-layer equa-tions, Appl. Math. Comp. 160, 487-497 (2005).
25. H. Nasr et al., H.M. Chebyshev solution of laminar boundary layer flow, Int. J. Comput. Math. 33,127-132 (1990).
26. B. L. Kuo, Application of the differential transformation method to the solutions of the Falkner-Skan wedge flow, Acta Mech. 164, 161-174 (2003).
27. R. Fazio, Blasius problem and FalknerSkan model: Topfers algorithm and its extension, Comput.Fluids. 73, 202-209 (2013).
28. S. J. Liao, A non-iterative numerical approach for two-dimensional viscous flow problems gov-erned by the Falkner-Skan equation, Int. J. Numer. Meth. Fluids. 35, 495-518 (2001).
29. N. S. Elgazery, Numerical solution for the Falkner-Skan equation, Chaos Solit. Fract. 35, 738-746(2008).
30. A. M. Wazwaz et al., Reliable treatment for solving boundary value problems of pantograph delaydifferential equations, Rom. Rep. Phys. 69, 102 (2017).
31. L. Bougoffa, Exact solutions of a generalized Bratu equation, Rom. J. Phys. 62, 110 (2017).32. A. M. Wazwaz, The successive differentiation method for solving Bratu equation and Bratu-type
equations, Rom. J. Phys. 61, 774–783 (2016).33. L. L. Wang, A new algorithm for solving classical Blasius equation, Appl. Math. Comput. 157,
1-9 (2004).34. L. Bougoffa, A. M. Wazwaz, New approximate solutions of the Blasius equation, Internat. J. Nu-
mer. Methods Heat Fluid Flow. 25, 1590-1599 (2015).
(c) RJP 63(Nos. 1-2), id:102-1 (2018) v.2.1*2018.2.16#c2fdcf3f