IFEM HW05 Sol Ch12

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

1215

Solutions to Exercises

Homework Exercises for Chapter 12 - Variational Formulation of Plane Beam Element Solutions
EXERCISE 12.1 A Mathematica script for fe by analytical integration is shown in Figure E12.1.

ClearAll[EI,EIi,EIj,Le, , ]; Le= ; Be={{6* ,(3* -1)*Le,-6* ,(3* +1)*Le}}; EI=EI1*(1- )/2+EI2*(1+ )/2; Ke=1/(2*Le^3)*Integrate[EI*Transpose[Be].Be,{ ,-1,1}]; Ke=Simplify[Ke]; Print["Ke for variable xsec beam:\n", Ke//MatrixForm]; ClearAll[EI]; Ke=Simplify[Ke/.{EI2->EI,EI2->EI}]; Print["Ke for EI1=EI2=EI is ", Ke//MatrixForm];
Ke for variable xsec beam: 6 EI1 EI2 2 2 EI1 EI2
3 2

6 EI1 EI2
3

2 EI1 2 EI2
2

2 2 EI1 EI2
2

3 EI1 EI2 2 2 EI1 EI2


2

2 2 EI1 EI2
2

EI1 EI2 2 EI1 2 EI2


2

6 EI1 EI2
3

6 EI1 EI2
3

2 EI1 2 EI2
2

EI1 EI2

2 EI1 2 EI2
2

EI1 3 EI2

Figure E12.1. Script to solve Exercise 12.1

Transcribing the result:

Ke = 1
3

6( E I1 + E I2 )

symm

2(2 E I1 + E I2 ) (3 E I1 + E I2 ) 2

6( E I1 + E I2 ) 2(2 E I1 + E I2 ) 6( E I1 + E I2 )

2( E I1 + E I2 ) ( E I1 + E I2 ) 2 . 2(2 E I1 + 2 E I2 ) ( E I1 + 2 E I2 ) 2

(E12.10)

The print output of the check E I1 = E I2 = E I is omitted to save space, but it reproduces the matrix (12.20).
EXERCISE 12.2 A Mathematica script for fe by analytical integration is shown in Figure E12.2.

ClearAll[q,q1,q2, , ]; Le= ; Ne={2*(1- )^2*(2+ ), (1- )^2*(1+ )*Le, 2*(1+ )^2*(2- ),-(1+ )^2*(1- )*Le}/8; q=q1*(1- )/2+q2*(1+ )/2; fe=Simplify[ (Le/2)*Integrate[q*Ne,{ ,-1,1}] ]; Print["fe^T for lin varying load q:\n",fe]; ClearAll[q]; fe=Simplify[fe/.{q1->q,q2->q}]; Print["check for q1=q2=q: ",fe]; fe^T for lin varying load q: 1 1 1 1 { 20 (7 q1 + 3 q2) (3 q1 + 2 q2) 2 20 (3 q1 + 7 q2) 60 (2 q1 + 3 q2) 60
Figure E12.2. Script to solve Exercise 12.2

1215

Chapter 12: VARIATIONAL FORMULATION OF PLANE BEAM ELEMENT

1216

Transcribing the result: fe = [ 3(7q1 + 3q2 ) (3q1 + 2q2 ) 3(3q1 + 7q2 ) (2q1 + 3q2 ) ]T . (E12.11)

60

The output of the check qi = q j = q is omitted to save space, but it does reproduce (12.21).
EXERCISE 12.3 Following the hint, the shape function matrix is evaluated at x = a , or = 2a / 1:

fe = P NT

2a / 1

P
2

(a )2 (2a + )

a (a )2

a 2 (2a 3 )

a 2 (a )

(E12.12)

If a = 0 and a = the force vector reduces to [ P 0 0 0 ]T and [ 0 0 P 0 ]T , respectively, as expected.


EXERCISE 12.4 Figure E12.3 shows a Mathematica script that implements (E12.5) for a linearly varying

distributed moment m .
ClearAll[m,Le, , ]; Le= ; Ne={2*(1- )^2*(2+ ), (1- )^2*(1+ )*Le, 2*(1+ )^2*(2- ),-(1+ )^2*(1- )*Le}/8; m=m1*(1- )/2+m2*(1+ )/2; fe=Integrate[m*D[Ne, ],{ ,-1,1}]; fe=Simplify[fe]; Print["fe: ", fe]; fe: { 1 (m1 + m2) 2
1 12

(m1 m2)

1 2

(m1 + m2)

1 (m1 12

m2)

Figure E12.3. Script to solve Exercise 12.4

This gives 1 (E12.13) [ 6(m 1 + m 2 ) (m 1 m 2 ) 6(m 1 + m 2 ) (m 1 m 2 ) ]T . 12 There is a shortcut for uniform m ; that is, m 1 = m 2 = m . Moving m out of the integral (E12.5) gives fe =
1

fe = m
1

T N d = m NT

=1 =1

= m [ 1

0 ]T .

(E12.14)

EXERCISE 12.5 Replacing m (x ) = C (a ) in the integral (E12.5) reduces it to

fe = C

d NT dx

=
x a

2C d NT d

=
2a / 1

C
2

6a (a )

( 3a )( a )

6a ( a )

a (3a 2 )
1 2

If a = 0 and a = , fe reduces to [ 0 C 0 0 ]T and [ 0 0 0 C ]T , respectively, as expected. If a = C [ 3/(2 ) 1/4 3/(2 ) 1/4 ]T , which is not that obvious.

(E12.15) we get

EXERCISE 12.6 The results of this exercise are collected on Table 12.1. The rules with 1, 2 and 3 points fail for monomial degrees 2, 4 and 6, respectively, and are exact for degrees up to one less. The property is extendible to any rule order: a Gauss rule with p points integrates exactly polynomials of degree 2 p 1 or less, as proven in texts on numerical analysis. Note that since the rules are symmetric about x = 0 they are exact for any odd function about x = 0 because the integral of any such function from 1 to 1 is zero. Consequently odd-degree monomials, such as or 3 , are integrated exactly by all rules.

1216

1217
Table 12.1. Results From Exercise 12.6 Monomial degree 1 2 3 4 5 6
2 3

Solutions to Exercises

0 Exact One-point rule (E12.6) Two-point rule (E12.7) Three-point rule (E12.8)

2 0 2 0 2 0 2 0

0 0 0

2 5 2 9 2 5

2 7

0
2 3 2 3

6 25

EXERCISE 12.7 A Mathematica script to compute Ke by two-point Gauss rule is shown in Figure E12.4.

This result reproduces exactly (E12.10) because the polynomials being integrated for linearly varying E I are cubic in (B, BT and E I are linear in ). From Exercise 12.6, the 2-point Gauss rule integrates exactly polynomials of order up to and including 3.
ClearAll[EI,EI1,EI2, , ]; Le= ; Be={{6* ,(3* -1)*Le,-6* ,(3* +1)*Le}}; EI=EI1*(1- )/2+EI2*(1+ )/2; Ke1=(1/(2*Le^3))*(EI*Transpose[Be].Be)/. -> Sqrt[3]/3; Ke2=(1/(2*Le^3))*(EI*Transpose[Be].Be)/. ->-Sqrt[3]/3; Ke=Simplify[Ke1+Ke2]; Print["2-Gauss-pt Ke for var xsec beam:\n ",Ke//MatrixForm]; 2-Gauss-pt Ke for var xsec beam:
6 EI1 EI2
3

2 2 EI1 EI2
2

6 EI1 EI2
3

2 EI1 2 EI2
2

2 2 EI1 EI2
2

3 EI1 EI2 2 2 EI1 EI2


2

2 2 EI1 EI2
2

EI1 EI2 2 EI1 2 EI2


2

6 EI1 EI2
3

6 EI1 EI2
3

2 EI1 2 EI2
2

EI1 EI2

2 EI1 2 EI2
2

EI1 3 EI2

Figure E12.4. Script to solve Exercise 12.7

EXERCISE 12.8 A Mathematica script to compute Ke by two-point Gauss rule is shown in Figure E12.5.

ClearAll[q,q1,q2, , ]; Le= ; Ne={2*(1- )^2*(2+ ), (1- )^2*(1+ )*Le, 2*(1+ )^2*(2- ),-(1+ )^2*(1- )*Le}/8; q=q1*(1- )/2+q2*(1+ )/2; fe1=(Le/2)*(Ne*q)/. -> Sqrt[3]/3; fe2=(Le/2)*(Ne*q)/. ->-Sqrt[3]/3; fe=Simplify[fe1+fe2]; Print["2-Gauss-pt fe^T for var load q:\n ",fe]; 2-pt Gauss for var load q: 1 1 { 36 (13 q1 + 5 q2) (2 q1 + q2) 36

1 36

(5 q1 + 13 q2)

1 36 (q1 + 2 q2)

Figure E12.5. Script to solve Exercise 12.8

This result is different from (E12.11) because the polynomials being integrated for linearly varying q are quartic in , and the 2-point Gauss rule is not exact for that order. But for uniform q the results would be the same.

1217

Chapter 12: VARIATIONAL FORMULATION OF PLANE BEAM ELEMENT

1218

EXERCISE 12.9 The derivation is presented in Przemieniecki [140], 5.6.

r = 4(1 4 2 )/(5 4 2 ). For = 0 the ratio is 4/5 and the FEM result is 20% in error. As grows the error increases, for example if = 1/4, r = 12/19 = 0.6317 and the error is over 36%.

EXERCISE 12.10 The FEM result is v2 = q L 4 (1 4 2 )2 /(96 E I ). The ratio to the exact solution is

1218

You might also like