Varational mehods (Rayeigh-Ritz Method)

fentahun ayalneh
 Outlines

 Varational mehods
 Rayeigh-Ritz Method
 Rayeigh-Ritz Method for Finite element method
 Varational mehod
 The governing equations in FE formulation is in the forms of
stationarity of a functional
 A functional is used to represent a function of a function.
𝒃
𝒅𝒖
𝑰 =𝑰 (𝒖)=∫ 𝑭 (𝒖, , 𝒙) 𝒅𝒙
𝒂 𝒅𝒙 I is a function of u, which

{( ) ( ) }
𝟏 𝒅𝒖 𝟐 itself is a function of x
𝑰 =𝑰 (𝒖)=∫ 𝑨𝑬 −(𝒒)(𝒖) 𝒅𝒙
𝟐 𝒅𝒙

 “Finding ū(x) that makes I stationary with respect to small

admissible varations in u(x) is equivalent to finding u(x) that
satisfies the governing diffrential equation”
 It is also konwn that the extrem value of a function f(x) is given
by df(x)/d(x)=0 (i.e. a function of several variables f(x 1, x2,...xn)
gives an extrimum value at x1, x2....xn) if
==....==0
 Varational mehod
 Now let us asumue the functional and find ū(x) that makes I(u)
stationary subject to end conditions u(a)=ua and u(b)=ub.

 The continues line shows the

exact function u(x) and dotted
line shows an approximate
function ū(x), both satisfys the
end condition .
 The diffrence between
ū(x) and u(x) is called the
varation of u(x) denoted
by
𝛿𝑢(𝑥)=𝑢̄ (𝑥)− 𝑢(𝑥)=𝜀 𝑣(𝑥)
  The varation is the diffrence
between ū(x) and u(x) at a given
ponit of x while du refers to the
diffrential change in u(x) as x
changes to x+dx

 Now for a given x as we move from u(x) to ū(x) we have

Expanding with taylor’s series

]+....
 Thus
]+....
The 1st varation of F is
The 2nd varation of F is ]
=
Hence the same is true with the functional I(u)
-
dx=

 For small values of the 2nd varation is niglected and the condition for
I to be stationary =0
 Rayleigh- rize method

 Rayleigh- rize method is an approch used to develop an approximate

solution to exterimaizing functional
 Instade of diffrential equation we arive at a system of algebric
equations
 For an elastic solid subject to a set of loads for which the
displacements and stresses are to be determined and u, v and w be
the displacements in x, y and z coordinate directions. Then for each
of displacement component an approximate solution is taken as

u= The function φi are usually

v=
taken as polynomials
satisfying the boundary
conditions. ‘a’ are the
amplitudes of the functions.
 Rayleigh- rize method

Rayleigh-Ritz Method consists of three basic steps

 Asume a displacment field given by ) for i=1,2...n, Ni is shape
function and ci undetermined cofficent
) should satisfy essential boundary conditions and internal
compatibility
 Evaluation of total potential consistent with the asumed
dispacment field (express it interms of ci)
 Set up and solve the system of equations (total potential will be
stationary with respect to small variations in the displacment field.
=i=1,2,...n
 This method of assembling stiffness equations is widely used in
elasticity problem.
 In calculus we know a function has extreme value when its first
derivative with respect to variables is zero.
 The function is maximum, if the second derivative is negative and
is minimum, if its second derivative is positive.
 The function of a function is termed a functional and its first
derivative is called first variance, we can say the first variance
attains a stationary value if its value is zero.
 In many engineering problems there are such functional, the first
variance of which attain stationary values, the potential energy of
the body of the structure is an example in elasticity problems.
 A deformable body is in equilibrium when the total potential
energy is having stationary value, and the condition that value of
total potential energy is stationary correspond to minimum value.
 Hence the principle of minimum potential energy stated as “of all
the possible displacement configurations a body can assume which
satisfy compatibility and boundary conditions, the configuration
satisfying equilibrium makes the potential energy a minimum
value”.
The potential energy of a structure is
Π = Internal potential energy – External potential energy
=U–W
 Rayeigh-Ritz Method for Finite element method

 In variational principle, a body is in equilibrium when potential

energy is minimum.
 The first derivative of a functional is called first variance, we can
say the first variance attains a stationary value if its value is zero.
 Using Rayleigh-Ritz method we have found approximating
functions satisfying the boundary conditions can be used to get the
solutions.
 In finite element analysis we use approximating functions for the
elements but not for entire body and use the principle of
minimizing potential energy to get the solutions for complex
structures.
Now let us see the variational method in finite element analysis
 The potential energy of a structure is

Π = Internal potential energy – External potential energy

= Strain energy – Work done by external forces
=U–W
 The above expression refers to entire structure, we can apply the
principle to each element separately, as the integration of the
summation is the same as the sum of the individual integrals.

Π=e=e-e ==

Where {δ} is vector of nodal displacement in the structure.

{δ} e is the vector of nodal displacements in the element.
Now, Strain energy of the element for elastic body
𝟏
𝑈 𝑒=∭ {𝜺 }𝑻 [ 𝛔] 𝒅𝑽 where
𝟐

Ue =
=
If {X }T={Xb Yb Zb} T and {Xs}T={Xs Ys Zs}T , then the work done by
these forces is given by

Wp =
where
but

=
 Πe=Ue+Wp

 From principle of minimum potential energy (the 1st variance of

Πe=0)== 0=

 Thus element equilibrium equation is

[k]e=(F)e

where [k]e stiffness matrix of the element [k]e

= displacement vector =[u1 u2...un ] and

(F)e load vector called consistent load vector, (F)e=

 Since [D] and [B] are constant matrixes
[k]e
where V= At
[D] is the elasticity matrix, In case of isotropic materials, case,
For plane stress For plane strain

[B] is strain displacement vector [B]=

 The above equation of equilibrium is to be assembled for entire
structure and boundary conditions are to be introduced.
 Then the solution of equilibrium equations result into nodal
displacements of all the nodal points.
 Once these basic unknowns are known, then displacement at any
point may be obtained by the relation
u=[N]{δ}e .

 The strains are assembled using the relation {ε}= [B]{δ}e and then

stresses also can be found {σ} = [D]{ε}e

 Varational method, is a versatile method.
 Using this method stiffness matrices and consistent load vectors
can be assembled easily.
 This method has made finite element analysis a versatile method.
Consider a linear bar element shown above
 Displacment field =u
 Nodal d.o.f vector
 Shape function [N]=[N1 N2]=[(1-) ()]
 Strain {
 Strain displacment vector [B]=
 [k]e=
 Consider a three noded quadratic bar element shown above, the
displacment at any point within the element is interpolated using
shape functions

 Displacment field =u(x)= N1(x)u1+ N2(x)u2+N3(x)u3

 The shape function Ni varies quadratically within the element
u(x)=c0+c1x+c2x2 where BC, u(0)=u1, u(l/2)=u3, u(l)=u2
 Hence solving for BC we have c0=u1,
c1=(4u3-u2-3u1)/l,
c2=(2u1+2u2-4u3)/l2
 u(x)=u1+()x+(2 rearranging we get
u(x)=(1-u1+(-)u2+()u3
  N1=1-,
 N2=-,
 Hence  N3=

 Nodal d.o.f vector

 Strain { ,
 Hence strain displacment matrix
[B]=
 Hence we have element stiffness matrix
 [k]e

=
 Hence we have element stiffness matrix
 [k]e

[k]e =
• Stiffness matrix for spring element
• For any deformation x of the spring corresponding to force F,
and Ω

Total potential energy is Π

• Applying the principle of minimum potential energy

=
=0 as is arbiterary and may not been zero
===0=

where {f} is nodal force (f1,f2,...fn)

d is nodal displacment (u1, u2,...un)
The strain energy stored in spring is given by
=
strain energy of spring 2
strain energy of spring 3
The potential energy due to external nodal forces given by
Ω =F1u1+F1u1+F1u1+F1u1=
Total potential energy is
Π= -(F1u1+F1u1+F1u1+F1u1)

Appying the priciple of potetional

= - k1u3 + k1u1 - F1=0

= k3u2 - k3u4 - F2=0

= k1u3 - k1u1 - k2u4 + k2u3+F3=0

= k2u4 - k2u3 - k3u4 + k3u4-F4=0

In matrix
Formulation of stiffness equation for bar elemnt

where [k]= [D]=E [B]=

Steps in stiffness matrix formulation in bar elelemt

 Formulate an expression for the total potential energy.
 Assume the displacement pattern to vary with a finite set of
undetermined parameters (here these are the nodal displacements
ui), which are substituted into the expression for total potential
energy.
 Obtain a set of simultaneous equations minimizing the total
potential energy with respect to these nodal parameters. These
resulting equations represent the element equations.
Thank you!!!