Rayleigh-Ritz Method:

The Rayleigh–Ritz method of expressing field variables by approximate method

with minimization of potential energy.

Assume a body in the x-y plane is subjected to all kinds of forces. There is no exact
solution for such problems, but an approximate solution can be used.
Let u and v be the displacements in x and y direction. Then for each of displacement
component an approximate solution is taken as:
𝑛

𝑢(𝑥, 𝑦) = ∑ 𝑎𝑖 . 𝑔𝑖 (𝑥, 𝑦) = 𝑎1 . 𝑔1 (𝑥, 𝑦) + 𝑎2 . 𝑔2 (𝑥, 𝑦) + ⋯.

𝑖=1

𝑣(𝑥, 𝑦) = ∑ 𝑏𝑗 . ℎ𝑗 (𝑥, 𝑦) = 𝑏1 . ℎ1 (𝑥, 𝑦) + 𝑏2 . ℎ2 (𝑥, 𝑦) + ⋯.

𝑗=1
Where:
𝑎𝑖 , 𝑏𝑖 are the unknown coefficient (generalized coefficient)
𝑔𝑖 (𝑥, 𝑦) , ℎ𝑖 (𝑥, 𝑦) are functions, can be selected in any form such as polynomial, trigonometric
….etc. Substituting these expressions for displacements in total potential energy expression can
be assembled. Then the total potential energy:

𝛱𝑃 = 𝛱𝑃 (𝑎1 , 𝑎2 , 𝑎3 … … … 𝑎𝑛 , 𝑏1 , 𝑏2 … … . 𝑏𝑚 )
From the principle of minimum potential energy,

𝜕𝛱𝑃
= 0 …… i=1→n
𝜕𝑎𝑖
𝜕𝛱𝑃 (Rayleigh-Ritz Equations)
= 0 …… j=1→m
𝜕𝑏𝑗
Note: the Rayleigh-Ritz method is different from the minimum potential energy, in
which the last one is dependent on the derivative of 𝛱𝑃 with respect to the
displacement itself while the Rayleigh-Ritz method depends on the derivative of 𝛱𝑃
with respect to the coefficients, but in concept both are the same.

9
Exampe1: For the shown structure, find maximum displacement and moment? Using
Rayleigh-Ritz method.

Solution:
For an approximate solution, we can assume a single displacement field, such as:
𝑣(𝑥) = 𝑎0 + 𝑎1 𝑋 + 𝑎2 𝑋 2 + 𝑎3 𝑋 3 + ⋯ (Polynomial)

The boundary conditions should be satisfied:

At x=0 → 𝑣(0)=0 & 𝑣 ′ (0)=0 (rotation=0)
Therefore 𝑎0 =0 &𝑎1 =0
𝑣(𝑥) = 𝑎2 𝑋 2 + 𝑎3 𝑋 3 + ⋯
This is an admissible displacement field (that mean it satisfies the boundary
conditions).
Choosing one term:
𝑣(𝑥) = 𝑎2 𝑋 2
𝛱𝑃 =U+V
1 1 1
𝑈 = 𝑓𝑜𝑟𝑐𝑒. 𝑑𝑖𝑠𝑝. = ∬ ∫(𝜎. 𝐴). (𝜀. 𝑑𝑥) = ∬ ∫ 𝜎. 𝜀. 𝑑𝑣𝑜𝑙.
2 2 2
1 1 𝜎 2
1 1 𝑀. 𝑦 2
𝑈 = ∬ ∫ 𝜎. 𝜀. 𝑑𝑣𝑜𝑙. = ∬ ∫ . 𝑑𝑣𝑜𝑙. = ∬ ∫ ( ) . 𝑑𝑣𝑜𝑙.
2 2 𝐸 2 𝐸 𝐼
2
1 1 2 2
1 1 2 1 2𝐿 1 𝑑2𝑣
𝑈 = ∫ 2 𝑀 . 𝑑𝑥. ∬ 𝑦 . 𝑑𝐴 = ∫ 𝑀 . 𝑑𝑥 = ∫ (𝐸𝐼 2 ) . 𝑑𝑥
2 𝐸𝐼 2 𝐸𝐼 2 0 𝐸𝐼 𝑑𝑥
2
𝑣(𝑥) = 𝑎2 𝑋
𝑑𝑣 𝑑2𝑣
= 2𝑎2 . 𝑋 ; = 2𝑎2
𝑑𝑥 𝑑𝑥 2

2
1 2𝐿 𝑑2𝑣 1 2𝐿
𝑈 = ∫0 𝐸𝐼 ( 2 ) . 𝑑𝑥 = ∫0 𝐸𝐼(2𝑎2 )2 . 𝑑𝑥
2 𝑑𝑥 2
𝑉 = −𝑃. 𝑣(𝑙) − 𝑃. 𝑣(2𝑙)
𝑣(𝑙) = 𝑎2 𝐿2 ; 𝑣(2𝑙) = 4𝑎2 𝐿2
𝑉 = −𝑃. 𝑎2 𝐿2 − 𝑃. 4𝑎2 𝐿2 = −5𝑃. 𝑎2 𝐿2
1 2𝐿
𝛱𝑃 =U+V= ∫0 𝐸𝐼(2𝑎2 )2 . 𝑑𝑥 − 5𝑃. 𝑎2 𝐿2 = 4𝐸𝐼𝑎2 2 𝐿 − 5𝑃. 𝑎2 𝐿2
2
𝜕𝛱𝑃 5 𝑃𝐿
= 0 = 8𝐸𝐼. 𝑎2 𝐿 − 5𝑃. 𝐿2 ⇨ 𝑎2 =
𝜕𝑎2 8 𝐸𝐼
5 𝑃𝐿
Thus, 𝑣(𝑥) = 𝑋 2 (approximate displacement field)
8 𝐸𝐼

10
 5 𝑃𝐿 5 𝑃𝐿3
Max. displacement at x=2L⇨ 𝑣(2𝑙) = (2𝐿)2 =
8 𝐸𝐼 2 𝐸𝐼

𝑑2𝑣 10
Max. moment at x=0⇨ M = EI 2 = 𝐸𝐼(2𝑎2 ) = 𝑃𝐿
𝑑𝑥 8
Choosing two terms:
𝑣(𝑥) = 𝑎2 𝑋 2 + 𝑎3 𝑋 3
𝛱𝑃 =U+V
2
1 2𝐿 1 𝑑2𝑣
𝑈 = ∫0 (𝐸𝐼 𝑑𝑥 2 ) . 𝑑𝑥
2 𝐸𝐼
𝑣(𝑥) = 𝑎2 𝑋 2 + 𝑎3 𝑋 3
𝑑𝑣 𝑑2𝑣
= 2𝑎2 . 𝑋 + 3𝑎3 𝑋 2 ; = 2𝑎2 + 6𝑎3 𝑋
𝑑𝑥 𝑑𝑥 2

2
1 2𝐿 𝑑2𝑣 1 2𝐿
𝑈 = ∫0 𝐸𝐼 ( 2 ) . 𝑑𝑥 = ∫0 𝐸𝐼(2𝑎2 + 6𝑎3 𝑋)2 . 𝑑𝑥
2 𝑑𝑥 2
𝑉 = −𝑃. 𝑣(𝑙) − 𝑃. 𝑣(2𝑙)
𝑣(𝑙) = 𝑎2 𝐿2 + 𝑎3 𝐿3 ; 𝑣(2𝑙) = 4𝑎2 𝐿2 + 8𝑎3 𝐿3
𝑉 = −𝑃. (𝑎2 𝐿2 + 𝑎3 𝐿3 ) − 𝑃(4𝑎2 𝐿2 + 8𝑎3 𝐿3 ) = −𝑃(5𝑎2 𝐿2 + 9𝑎3 𝐿3 )
1 2𝐿
𝛱𝑃 =U+V= ∫0 𝐸𝐼(4𝑎2 2 + 24𝑎2 𝑎3 𝑋 + 36𝑎3 2 𝑋 2 ). 𝑑𝑥 − 𝑃(5𝑎2 𝐿2 + 9𝑎3 𝐿3 )
2
𝐸𝐼
𝛱𝑃 = [ . (4𝑎2 2 𝑋 + 12𝑎2 𝑎3 𝑋 2 + 12𝑎3 2 𝑋 3 )]2𝑙 2 3
0 − 𝑃(5𝑎2 𝐿 + 9𝑎3 𝐿 )
2
𝐸𝐼
𝛱𝑃 = [ . (8𝑎2 2 𝐿 + 48𝑎2 𝑎3 𝐿2 + 96𝑎3 2 𝐿3 )] − 𝑃(5𝑎2 𝐿2 + 9𝑎3 𝐿3 )
2

𝜕𝛱𝑃 𝐸𝐼
= 0 = [ 2 . (16𝑎2 𝐿 + 48𝑎3 𝐿2 )] − 5𝑃. 𝐿2 ⇨ 4𝐸𝐼. (2𝑎2 𝐿 + 6𝑎3 𝐿2 ) − 5𝑃. 𝐿2 = 0 … … … (1)
𝜕𝑎2
𝜕𝛱𝑃 𝐸𝐼
= 0 = [ 2 . (48𝑎2 𝐿2 + 192𝑎3 𝐿3 )] − 9𝑃. 𝐿3 ⇨ 12𝐸𝐼. (2𝑎2 𝐿2 + 8𝑎3 𝐿3 ) − 9𝑃. 𝐿3 = 0 ….(2)
𝜕𝑎3

11 𝑃𝐿 𝑃
Solving the above equations: 𝑎2 = ; 𝑎3 = −
8 𝐸𝐼 4𝐸𝐼
11 𝑃𝐿 2 𝑃 3
Thus, 𝑣(𝑥) = 𝑋 − 𝑋 (approximate displacement field)
8 𝐸𝐼 4𝐸𝐼
11 𝑃𝐿 𝑃 7 𝑃𝐿3
Max. displacement at x=2L⇨ 𝑣(2𝑙) = (2𝐿)2 − (2𝐿)3 =
8 𝐸𝐼 4𝐸𝐼 2 𝐸𝐼

𝑑2𝑣 11 𝑃𝐿 6𝑃𝑋 11
Max.moment at x=0⇨ M = EI = 𝐸𝐼(2𝑎2 + 6𝑎3 𝑋) = 𝐸𝐼( − )= 𝑃𝐿
𝑑𝑥 2 4 𝐸𝐼 4𝐸𝐼 4

11
 Field Defl.@L Defl.@2L Mom.@0 Mom.@L Mom.@2L
𝑃𝐿3 𝑃𝐿3
(coff.× 𝐸𝐼 ) (coff.× 𝐸𝐼 ) (coff.× 𝑃𝐿) (coff.× 𝑃𝐿) (coff.× 𝑃𝐿)
One 0.625 2.5 1.25 1.25 1.25
term
Two 1.125 3.5 2.75 1.25 -0.25
terms
Exact 1.167 3.5 3 1 0

Note: There is the guarantee in getting more accurate solution on increasing the
number of terms and this is the main disadvantage of the Rayleigh-Ritz method.

Exampe2: Using Rayleigh-Ritz method to determine the expressions for deflection

and bending moments in a simply supported beam subjected to uniformly distributed
load over entire span. Find the deflection and moment at midspan and compare with
𝑚𝜋𝑥
exact solutions. Assume the Fourier series 𝑦 = ∑𝑚 𝑖=1,3,… 𝑎𝑖 sin 𝑙 is the deflection
function.

Solution: let us consider only two terms in the series i.e. let
𝜋𝑥 3𝜋𝑥
𝑦 = 𝑎1 sin + 𝑎2 sin
𝑙 𝑙
𝛱𝑃 =U+V
2
1 𝑙 1 𝑑2𝑦
𝑈 = ∫0 (𝐸𝐼 2 ) . 𝑑𝑥
2 𝐸𝐼 𝑑𝑥
𝜋𝑥 3𝜋𝑥
𝑦(𝑥) = 𝑎1 sin + 𝑎2 sin
𝑙 𝑙
𝑑𝑦 𝜋 𝜋𝑥 3𝜋 3𝜋𝑥 𝑑2𝑦 𝜋 𝜋𝑥 3𝜋 3𝜋𝑥
= −𝑎1 cos − 𝑎2 cos ; = −𝑎1 ( )2 sin − 𝑎2 ( )2 sin
𝑑𝑥 𝑙 𝑙 𝑙 𝑑𝑥 2 𝑙 𝑙 𝑙 𝑙 𝑙
2
1 𝑙 𝑑2𝑦 1 𝑙 𝜋 2 𝜋𝑥 3𝜋 2 3𝜋𝑥 2
𝑈= ∫ 𝐸𝐼 ( ) . 𝑑𝑥 = ∫ 𝐸𝐼 (−𝑎 1 ( ) sin − 𝑎2 ( ) sin ) . 𝑑𝑥
2 0 𝑑𝑥 2 2 0 𝑙 𝑙 𝑙 𝑙
1 𝑙 𝜋4 𝜋𝑥 3𝜋𝑥 2
𝑈= ∫ 𝐸𝐼 (𝑎 1 sin + 9𝑎 2 sin ) . 𝑑𝑥
2 0 𝑙4 𝑙 𝑙
𝑙 𝑙 𝜋𝑥 3𝜋𝑥
𝑉= − ∫0 𝑤. 𝑦 𝑑𝑥 =− ∫0 𝑤. (𝑎1 sin + 𝑎2 sin )𝑑𝑥
𝑙 𝑙

12
 1 𝑙 𝜋4 𝜋𝑥 3𝜋𝑥 2 𝑙 𝜋𝑥 3𝜋𝑥
𝛱𝑃 =U+V=2 ∫0 𝐸𝐼 (𝑎1 sin + 9𝑎2 sin ) . 𝑑𝑥 − ∫0 𝑤. (𝑎1 sin + 𝑎2 sin )𝑑𝑥
𝑙4 𝑙 𝑙 𝑙 𝑙

𝜋4 2𝑤𝑙 𝑎2
𝛱𝑃 = 𝐸𝐼 (𝑎1 2 + 81𝑎2 2 ) − (𝑎1 + )
4𝑙 3 𝜋 3
𝜕𝛱𝑃 𝜋4 2𝑤𝑙 4𝑤𝑙 4
= 0 = 𝐸𝐼 . (2𝑎1 ) − ⇨ 𝑎1 =
𝜕𝑎1 4𝑙 3 𝜋 𝐸𝐼𝜋5
𝜕𝛱𝑃 𝜋4 2𝑤𝑙 4𝑤𝑙 4
= 0 = 𝐸𝐼 3
. (81 × 2𝑎2 ) − ⇨ 𝑎2 =
𝜕𝑎2 4𝑙 3𝜋 243𝐸𝐼𝜋5

4𝑤𝑙 4 𝜋𝑥 4𝑤𝑙 4 3𝜋𝑥

Thus, 𝑦(𝑥) = sin + sin (approximate displacement field)
𝐸𝐼𝜋5 𝑙 243𝐸𝐼𝜋5 𝑙
𝑙 4𝑤𝑙 4 𝜋 4𝑤𝑙 4 3𝜋 𝑤𝑙 4
Max. displacement at x= ⇨ 𝑦𝑚𝑎𝑥 = sin + sin =
2 𝐸𝐼𝜋5 2 243𝐸𝐼𝜋5 2 76.82𝐸𝐼
We know the exact solution is:
5𝑤𝑙 4 𝑤𝑙 4
𝑦= ==
384𝐸𝐼 76.8𝐸𝐼
Thus the deflection is almost exact.
𝑙 𝑑2𝑦 𝜋 𝜋𝑥 3𝜋 3𝜋𝑥
Max. moment at x= ⇨ M = EI = 𝐸𝐼(−𝑎1 ( )2 sin − 𝑎2 ( )2 sin )
2 𝑑𝑥 2 𝑙 𝑙 𝑙 𝑙
4𝑤𝑙 2 𝜋 36𝑤𝑙 2 3𝜋 𝑤𝑙 2
M = 𝐸𝐼 (− sin − sin ) = (8.05 )
𝐸𝐼𝜋3 2 243𝐸𝐼𝜋3 2
𝑤𝑙 2
We know the exact solution is:
8
Note: By taking more terms in Fourier series more accurate results can be obtained.

Exampe3: Using Rayleigh-Ritz method to determine the expressions for

displacement and stress in a fixed bar subjected to axial force P as shown in figure
below. Draw the displacement and stress variation diagram. Take three terms in
displacement function.

13
Solution: Let the displacement at load point be 𝑢1 . Then the strain energy of the bar

𝛱𝑃 =U+V
1 1 1
𝑈 = 𝑓𝑜𝑟𝑐𝑒. 𝑑𝑖𝑠𝑝. = ∬ ∫(𝜎. 𝐴). (𝜀. 𝑑𝑥) = ∬ ∫ 𝜎. 𝜀. 𝑑𝑣𝑜𝑙.
2 2 2
1 1 2
1 𝑑𝑢 2
𝑈 = ∬ ∫ 𝐸𝜀. 𝜀. 𝑑𝑣𝑜𝑙. = ∬ ∫ 𝐸𝜀 . 𝑑𝑣𝑜𝑙. = ∬ ∫ 𝐸 ( ) . 𝑑𝑣𝑜𝑙.
2 2 2 𝑑𝑥

1 𝑑𝑢 2 1 𝑑𝑢 2
𝑈 = ∫ 𝐸 ( ) . 𝑑𝑥. ∬ 𝑑𝐴 = ∫ 𝐸𝐴 ( ) . 𝑑𝑥
2 𝑑𝑥 2 𝑑𝑥
Let the displacement at any point be given by,
𝑢(𝑥) = 𝑎0 + 𝑎1 𝑋 + 𝑎2 𝑋 2 + 𝑎3 𝑋 3 + ⋯ (Polynomial)
Choose three terms only:
𝑢(𝑥) = 𝑎0 + 𝑎1 𝑋 + 𝑎2 𝑋 2

This function has to satisfy the boundary conditions

(i) At x = 0, u = 0
(ii) At x = L , u = 0
From Boundary condition (i), we get
𝑎0 = 0
From Boundary condition (ii), we get
0 = 𝑎1 𝑙 + 𝑎2 𝑙 2 ⇨ 𝑎1 = −𝑎2 𝑙
𝑢(𝑥) = −𝑎2 𝑙𝑋 + 𝑎2 𝑋 2
𝑑𝑢
= −𝑎2 𝑙 + 2𝑎2 𝑋
𝑑𝑥
1 𝑑𝑢 2 1
𝑈 = ∫ 𝐸𝐴 ( ) . 𝑑𝑥 = ∫ 𝐸𝐴(−𝑎2 𝑙 + 2𝑎2 𝑋 )2 . 𝑑𝑥
2 𝑑𝑥 2

1 𝑙 𝐸𝐴𝑎2 2 4 𝑙 𝐸𝐴𝑎2 2 3
𝑈 = ∫0 𝐸𝐴𝑎2 2 (𝑙 2 − 4𝑙𝑋 + 4𝑥 2 ). 𝑑𝑥= [𝑙 2 𝑋 − 2𝑙𝑋 2 + 𝑥 3 ] = 𝑙
2 2 3 0 6
𝑉 = −𝑃. 𝑢(𝑙/2)
𝑎 𝑎 𝑎
𝑢(𝑙/2) = − 2 𝑙 2 + 2 𝑙 2 = − 2 𝑙 2
2 4 4
𝑎2 2 𝑎
𝑉 = −𝑃. (− 𝑙 ) = 2 𝑃𝑙 2
4 4
𝐸𝐴𝑎2 2 3 𝑎2
𝛱𝑃 =U+V = 𝑙 + 𝑃𝑙 2
6 4
𝜕𝛱𝑃 𝐸𝐴𝑎2 3 1 3 𝑃
=0= 𝑙 + 𝑃𝑙 2 ⇨ 𝑎2 = −
𝜕𝑎2 3 4 4 𝐸𝐴𝑙

14
 3 𝑃 3 𝑃
𝑢(𝑥) = 𝑙𝑋 − 𝑋2
4 𝐸𝐴𝑙 4 𝐸𝐴𝑙
𝑑𝑢 3𝑃 6 𝑃
𝜎 = 𝐸𝜀 = 𝐸 = 𝐸(−𝑎2 𝑙 + 2𝑎2 𝑋) = − X
𝑑𝑥 4𝐴 4 𝐸𝐴𝑙

The variation of displacement and stresses are shown in the figure below:

(a)Variation of displacement (u)

(b)Variation of stress (𝜎)

15
Comments on Rayleigh-Ritz Method:
- In this method the approximating functions (displacement fields) must satisfy
the boundary conditions and should be easy to use.
- Polynomials are normally used. Sometimes sine-cosine terms are also used.
- Results can be obtained for complex problems. But for complex problems it
is difficult to say whether the results obtained are accurate enough to use. The
doubt will arise due to the following two reasons:
(a) Whether this is the only function which can be used
(b) How many terms in the function are to be used.
- The best way to ensure the accuracy is to get result using a certain number of
terms and then use additional terms to get the results. If the difference is
negligible, we can conclude that the satisfactory result is obtained. However
it may be noted that the lowest terms in the series should not be omitted in the
approximating functions.
- The Rayleigh-Ritz method is rarely exact but becomes more accurate as more
d.o.f. (degrees of freedom) are used.
- The approximately displacement fields should be admissible such as 𝑓𝑖 (𝑥, 𝑦, 𝑧)
of the elastic body times an amplitude 𝑎𝑖 . For two dimensional problems:
𝑛

𝑢(𝑥, 𝑦) = ∑ 𝑎𝑖 . 𝑓𝑖 (𝑥, 𝑦) = 𝑎1 . 𝑓1 (𝑥, 𝑦) + 𝑎2 . 𝑓2 (𝑥, 𝑦) + ⋯.

𝑖=1

𝑣(𝑥, 𝑦) = ∑ 𝑏𝑗 . ℎ𝑗 (𝑥, 𝑦) = 𝑏1 . ℎ1 (𝑥, 𝑦) + 𝑏2 . ℎ2 (𝑥, 𝑦) + ⋯.

𝑗=1

16