Analysis of Winkler’s Model of Elastic Foundation Using Differential Transfrom

Method

Sema Bodur, H. Alpaslan Peker, Galip Oturanç

Selcuk University, Department of Mathematics, 42075, Campus, Konya, Turkey

Abstract

In this study, differential equation of beams on elastic foundation, which has a

great importance for engineering problems, has been analyzed [1,2]. The beams on
elastic foundation has widely been used in plenty of engineering areas. For instance,
railway engineering, pipes used in liquid and gas conduction lines, off-shore and port
foundations, some applications in airports, plane-space and petrochemical industries,
biomechanical and dentistry. Winkler, Pasternak, Vlasov, Kerr and some other models
have been developed for the foundations with different types of elements such as
beams, discs and shells [5,6]. In literature, the beams on Winkler foundation problem
has been solved for static, buckling and vibration analyses by finite differences, finite
elements, boundary elements, Ritz, Galerkin, differential quadrature, Monte Carlo
methods [7]. In this study, apart form the literature, the problems related to Winkler’s
model of elastic foundation with different conditions has been analyzed by the
differential transform method, which is based on Taylor series expansion and is easily
applied to linear or nonlinear problems and reduces the size of computational work.
With this method exact solutions may be obtained without any need of cumbersome
work and it is a useful tool for analytical and numerical solutions [3,4]. In addition, the
obtained solutions have been compared to analytical solutions. Besides, all
calculations have been made by Maple13 and codes were also given.

Acknowledgement. This study is supported by the Selcuk University Scientific

Research Project Coordinatorship (BAP).

Keywords: Differential Transform Method, Winkler’s Model, Beams on Elastic

Foundation
 1. DIFFERENTIAL TRANSFORMATION METHOD

Differential transformation method (DTM) is a semi-analytical method based

on Taylor expansion. This method tries to find coefficients of series expansion of
unknown function by using the initial data on the problem. This method is especially
good at finding the solutions of initial value problems. But, in general, it does not
find the solutions of boundary value problems. By DTM, algebraic equations of
initial value and boundary value problems are obtained and then transformation
coefficients are calculated. Thus, either closed form series solutions (i.e. analytical
solutions) or approximate solutions are obtained. This method was firstly introduced in
1954 by Pukhov who especially made some studies related on modelling of electrical
and electronical problems. After that Chen, Ho and Zou was applied and developed
this method for the partial differential equations [3,4].

Definition 1.1. The one-dimensional differential transform of a function w(x) at the

point x = x0 is defined as follows:

1  dk 
W (k ) =  k
w( x)  (1.1)
k !  dx  x= x
0

Where w(k) is the original function and W(k) is the transformed function.

Definition 1.2. The differential inverse transform of W (k ) is defined as follows

∞
w( x) = ∑ W (k ) x k (1.2)
k =0

From 1.1 and 1.2, we obtain

∞
1  dk 
=w( x) ∑ 
k = 0 k !  dx
k
w( x)  ( x − x0 ) k (1.3)
 x= x
0

The following theorems that can be deduced from the above definitions are given
below as a list:
 Original Function Transformed Function
1 f=
( x ) g ( x ) ± h( x ) F=
(k ) G (k ) ± H (k )

2 f ( x) = λ g ( x) F (k ) = λG (k )

3 f ( x ) = g ( x ) h( x ) k
=F (k ) ∑ G(l ) H (k − l )
l =0

4 d n g ( x) (k + n)!
f ( x) = =F (k ) G ( k + n)
dx n k!

5 f ( x) = x n 0, if k ≠ n
F ( k ) = δ ( k − n) = 
1, if k = n

Proof of 4 (for n=1):

1  dn  dg ( x)
We know that F (k ) =  n
f ( x)  . Since f ( x) = ,
k !  dx  x =0 dx

1  dn  d 
F (k ) =  n 
g ( x)  
k !  dx  dx   x =0

1  d n +1  1  d n +1 
=  n +1
g ( x)  = (k + 1)  n +1
g ( x) 
=
k !  dx  x 0= (k + 1)!  dx x 0

=+
(k 1)G (k + 1) .

This completes the proof of 4 for n=1.

Now we give some examples.

Example 1. Let’s find the solution of y′ = y , y (0) = 1 .

Analytical Solution: y ( x) = e x
 DTM Solution: Let Y (k ) be the transformed equation of y ( x) . Then the

corresponding equation is as follows:

(k + 1)Y (k + 1) − Y (k ) =
0

Y (k ) =+
(k 1)Y (k + 1)

Y (k )
Y (k + 1) =
k +1

and corresponding initial condition is Y (0) = 1 .

For k=0, Y(1)=1.

For k=1, Y(2)=1/2.

For k=2, Y(1)=1/6.

For k=3, Y(1)=1/24.

∞
y ( x) =∑ Y (k ) =Y (0) + Y (1) x + Y (2) x 2 + Y (3) x 3 + ....
k =0

Thus, it is easily found that Y(k)=1/k! and then solution is as follows [4]:

∞ ∞
1 k ∞ xk x
y ( x) = ∑ Y (k ) ( x − x0 ) =
∑ x = ∑ = e.
k

=k 0 = k!
k 0= k 0 k!

Example 2. (Riccati differential equation) Let’s find the solution of the below
equation with a special solution y1 = 1 and initial condition y (0) = 2 ,

y '= xy 2 + (1 − 2 x) y + x − 1 .

1
Analytical Solution: y = 1+ −x
ce + 1 − x

The special solution suitable for the initial conditions is as follows:

 1
y (0) =
1+ =
2 then c=0.
c +1

Therefore the special solution of the given equation is as follows [4]:

∞
y = 1 + ∑ xn .
n =0

DTM Solution: Let Y (k ) be the transformed equation of y ( x) . Then the

corresponding equation is as follows:

(k + 1)Y (k + 1)= δ (k − 1) ⊗ U (k ) + F (k ) ⊗ Y (k ) + G (k )

δ (k − 1) ⊗ U (k ) + F (k ) ⊗ Y (k ) + G (k )
Y (k + 1) =
k +1

where

u ( x) =
y 2 , f ( x) =
1 − 2 x, g ( x ) =
x −1

U (k )= Y (k ) ⊗ Y (k ), F (k )= δ (k ) − 2δ (k − 1), G (k )= δ (k − 1) − δ (k ) .

For k=0

Y (1) = δ (−1) ⊗ U (0) + F (0) ⊗ Y (0) + G (0)

Y (1) = 0 + 2 − 1, Y (1) = 1

For k=1

2Y (2) = δ (0) ⊗ U (1) + F (1) ⊗ Y (1) + G (1)

2Y (2) = 4 − 3 + 1 = 2, Y (2) = 1

For k=2
 3Y (3) =δ (1) ⊗ U (2) + F (2) ⊗ Y (2) + G (2)

3Y (3) = 4 − 1 = 3, Y (3) = 1

If it is continued by this, then it is obtained that Y(0)=2 and Y=

(k ) 1 (k > 1) .

Then, by combining this results, y ( x) can be find as follows:

∞
y ( x) =∑ Y (k ) =Y (0) + Y (1) x + Y (2) x 2 + Y (3) x3 + ....
k =0

∞
y ( x) = 2 + x + x 2 + x3 + ... =1 + ∑ x n .
n =0

This solution is exactly same as the analytical solution.

Example 3. (Blasius equation) Consider the below boundary value problem,

1
f ′′′( x) + f ( x) f ′′( x) =
0
2

= 0 , f ′(0)
f (0) = 1 , f ′(−∞=
) 0 , −∞ < x < 0 .

This is a boundary value problem where boundary condition is at infinity. This is a

difficulty for the DTM. To overcome this difficulty, Pade approximants can be applied
to manipulate the obtained series for DTM.

Padé approximant is the best approximation of a polynomial approximation of

a function into rational functions of polynomials of given order. Let f be a function and
m,n ≥ 0 be integers. Then the Pade approximant of order [m / n] , denoted by
[m,n](f x) is the following rational function:
 m

∑a x i
i
a0 + a1 x + a2 x 2 +  + am x m
f ( x) ≅ R=
( x) =
i =0
n
1 + b1 x + b2 x 2 +  + bn x n
1 + ∑ bj x j
j =1

Note that
f (0) = R (0) ,

f ′(0) = R′(0) ,

f ′′(0) = R′′(0) ,

........
f ( m + n ) (0) = R ( m + n ) (0)

In order to obtain better numerical results, the combination of the differential

transform method and the diagonal approximants [N/N] will be used [12].

In order to find numerical values of f ′′(0) , the series obtained by the

differential transform method and the diagonal Padé approximants will be combined.
Several diagonal Padé approximants of different degrees will be used to form an
opinion about the behavior of the solution of the Blasius equation.
Note that f ′′(0) = A , where A is a positive constant.
If DTM is applied the given equation, then

1 k
(k + 1)(k + 2)(k + 3) F (k + 3) + ∑ (k − r + 1)(k − r + 2) F (r ) F (k − r + 2) =0
2 r =0

1 k
− ∑ (k − r + 1)(k − r + 2) F (r ) F (k − r + 2)
2
F (k + 3) = r =0
(k + 1)(k + 2)(k + 3)

is found. For the initial and boundary values, differential transform equivalences are as
follows:
A
=
F (0) 0=
, F (1) 1=
, F (2) .
2
Then, the solution is obtained as follows:

A 2 A 4 A2 5 A 6 11A2 7  1 11 
f ( x) = x + x − x − x + x + x +− A+ A3  x8 − 
2 48 240 960 20160  21504 161280 

Now, our aim is to determine a numerical value for A using the boundary
condition lim f ′( x) = 0 . After the formation of derivative of above polynomial solution
x →−∞

by Padé approximation, this condition is applied to the obtained rational function.

  1 
3  4 + 3 Ax +  − A2 +  x 2 
 3 
For [2 / 2] , lim  =−3 A2 + 1 =0 , then A = 0.5773502693 .
x →∞ 12 − 3 Ax + x 2

For [3/3], A = 0.5163977795 .

For [4/4], A = 0.5227030798 .

Adomian decomposition method and variational iteration method will be used for
the comparison [11].

In order to see the consistency of solutions, the series solution up to the 12th
term of the problem are calculated by the differential transform method, Adomian
decomposition method and variational iteration method respectively with respect to the
numerical value of A for the [4/4] diagonal approximant for which 0.5227030796 is
the obtained value by variational iteration method [8]. Furthermore, they are calculated
with step size 0.1 on the interval [0,1] and the results are shown graphically in the
below figure [9].
 Results obtained by ADM, DTM and VIM

2. ELASTIC CURVE

The studies related on structures on elastic foundations (rail lines, road paving,
building foundations, etc.) are quite high due to the width of the application area.
Especially, a wide range of beams problems resting on elastic foundations were
studied in the literature. This is because of such problems in interaction with the floor
beams, some of the assumptions made primarily to examine the behavior of the
system, ie, a floor model should be considered from the beginning. The most
commonly used hypothesis for beams on elastic foundation studies is the hypothesis of
Winkler which assumes that the ground reaction is proportional to the beam collapses.
In this study, the beam problem on Winkler’s model of an elastic foundation is solved
by DTM and the results are compared to analytical ones [1,2,5,6].

On the other hand, as a M.Sc. Thesis, we will make a visual computer

program that will calculate all kind of beams on elastic foundation.
2.1. Winkler’s Model of an Elastic Foundation [7]

Winkler’s model of an elastic foundation assumes that the deflection y at any point on
the surface of the foundation is proportional to the stress σ at that point, i.e., σ = k0 y ,
where k0 is called the modulus of the foundation with dimension [force/length3]. In
the study of beams on elastic foundation, let p be the intensity per unit length of the
distributed load on the foundation along the length of the beam, i.e., p=σb, where b is
the width of the beam. Hence, as shown in Figure 2.1 and from Winkler’s assumption,
p=ky, where k=k 0 b with dimension [force/length2].

Figure 2.1 Winkler’s model of elastic foundation

For a beam on an elastic foundation under the action of a distributed load w(x) as
shown in Figure 2.1(b), the flexural deflection y(x) is governed by the equation

d4y
EI = w( x) − p ( x) ,
dx 4
where EI is the flexural rigidity of the beam. Substituting p(x)=ky(x) into the equation
leads to a fourth-order linear ordinary differential equation

d4y w( x) k
4
+ 4 β 4 y =, 4β 4 =
dx EI EI
The constants in the solution of the differential equation are determined from the
boundary conditions of the beam, which are given by the end supports of the beam.
Some typical boundary conditions are listed in Table 2.1.

Table 2.1 Boundary Conditions

2.2. Solution of The Problem

The deflection of a beam free at both ends under a trapezoidally distributed load as
shown in the following figure will be determined.
Using the Heaviside step function, the distributed load can be expressed as

w2 − w1
w( x) = [ w1 + ( x − a )][ H ( x − a ) − H ( x − b)],
b−a

The differential equation becomes as follows [7]

d4y w( x) ^ −
+ 4 β 4
y = = [ w1 + w ( x − a )][ H ( x − a ) − H ( x − b)],
dx 4 EI

where
^ ^
^ w ^ w −w −w
w1 = 1 , w2 = 2 , w= 2 1.
EI EI b−a

In this equation, k is the coefficient of the spring, w is slump, E is Young’s

modulus (Pa) and I is the moment of the inertia (m4), x is the horizontal distance
measured along beam.
In this study, the assumptions are as follows:
w(x)= e − x
a=0
b=1
E=I=1
Heavyside step function=1.

According to these, the DTM correspondence of the equation and the conditions are
as follows:
y(x) 
→ Y(k)

y ' (0) 
→ Y(1)

y '' (0) 
→ 2Y(2)

y ''' (0) 
→ 6Y(3).
From the conditions, we have

Y(0)=A, Y(1)=B, Y(2)=0, Y(3)=0.

The equation is as follows [10]:

(k+1)*(k+2)*(k+3)*(k+4)*Y(k+4)+Y(k)=W(k)

and so

W (k ) − Y (k ) * k !
Y (k + 4) = .
(k + 4)!

For the calculations, Maple 13 is used. The analytical solution is as follows:

The DTM solution is as follows:
By graphically, both results are compared as follows:
plot({g,f},x=-2..8);

It is easily seen from the graphic that DTM solution is very close to the
analytical solution. [10]

3. CONCLUSION
1. DTM is very efficient semi-analytical method especially for the initial value
problems. This method is also used for the boundary value problems.
2. The result obtained in this study by DTM is very close to the analytic solution
which can be easily seen from the last graphic.
3. However, convergenceny analysis has not been solved exactly, so this remains
as a future project for the young researchers.
REFERENCES

1.M. İnan, Cisimlerin Mukavameti, 1967 (In Turkish).

2. İ. Kayan, Cisimlerin Mukavemeti, 1992 (In Turkish).

3. Zhou, J. K. 1986, Differential transformation and its applications for electical

circuits.

4.G.Oturanç,Y. Keskin, Diferensiyel Dönüşüm Yöntemiyle Diferensiyel Denklemlerin

Çözülmesi, 2010 (In Turkish).

5. Vlasov, V. Z. and Leont’ev N. N., Beams, Plates and Shells on Elastic foundations,
Translated from Russion to Enghlish by Barouch, A., Israel program for scientific
translations, Jerusalem, 1966.

6. Daloğlu, A. , Doğangün, A. , and Ayvaz, Y., Dynamic analysis of foundation plates

using a consistent Vlasov Model , Journal of Sound and Vibration, 224(5), 941-951,
(1999).

7. Wei-Chau Xie , Differential Equations for Engineers , Cambridge University Press ,

2010.

8. Abdul-Majid Wazwaz, The variational iteration method for solving two forms of
Blasius equation on a half-infinite domain, Applied Mathematics and Computation,
188, 485–491, 2007.

9. H. A. Peker, O. Karaoğlu and G. Oturanç , The Differential Transformation

Method and Pade Approximant For a Form of Blasius Equation , Mathematical and
Computational Applications, Vol. 16, No. 2, pp. 507-513, 2011
10. S. Bodur , M.Sc. Thesis , Computer-Aided Analysis of Beam Models Resting on
Elastic Foundations by Differential Transform Method , Selcuk University Department
of Mathematics (To Appear).

11. A. Wazwaz , A comparison between the variational iteration method and Adomian
decomposition method , Journal of Computational and Applied Mathematics , Volume
207, Issue 1, 1 October 2007, Pages 129–136

12. http://en.wikipedia.org/wiki/Padé_approximant