Chapter 7

ROSE School, Pavia, Italy Instructor: Enrico Spacone
NONLINEAR STRUCTURAL ANALYSIS Teaching Assistant: K. Wijesundara

Spring Term 2009
The class notes© for this course were developed jointly by

Prof. E. Spacone at the University “G D’Annunzio” of Chieti-Pescara, Italy
and Prof. J. P. Conte at the University of California, San Diego, USA
7. Intro to Nonlinear Frame Analysis
1
INTRODUCTION
Three examples are presented hereafter

to introduce nonlinear problems and
nonlinear solution schemes
2
Example 1: Material Nonlinearity
P2,U2
P1 , U 1
P = KU
ELASTIC BEAM
Le = 3 m
3m
 P1 
 
P =  P2 
P 
 3
P3 , U 3  U1 
 
U = U2 
ELASTO-PLASTIC U 
HINGE  3
3
Example 1
2
1
U2
U1
U e=1 1 
2
e=1  
ID =  
0  U
4  3 
3
2 U3
0 
IDe=2 =  
U e= 2 3
1
4
ELEMENTS STRUCTURE
Example 1 U2
U1
2 U1 U2 U3
1
 12 6 12 6  U
 L3 − U1
L2b L3b L2b 
 b 
 6 4
−
6 2 
 L2 U2
e =1 Lb L2b Lb 
U K = EI b 
e =1

b
 − 12 −
6 12
− 2 
6 U3
 L3 L2b L3b Lb
 b 
 6 2
−
6 4 
4  L2 Lb L2b L b  U3
 b
3 U1 U2 U3
 12 6 6 
2  L3  U1
L2b L2b
 b 
 1 −1 6 4 2 
e= 2
= kh  K = EI b  2 U2
Lb 
e= 2 K 
U  −1 1  U3 L
 b
Lb

6 2 4 
 L2 + kh  U3
1 U3  b Lb Lb  5
Example 1
linear ϕ
Lb Hp:
curvature ϕ θ = ϕ Lpl
constant over Lpl
linear
non-
ELASTO-PLASTIC Lpl
M
Alternative 1
Order of magnitude of d θ
≤ Lpl ≤ d
plastic hinge length 2 6
Example 1
M M
My
ϕ−ϕy
ϕ
linear
ϕy
Lb
Consider plastic
deformations only
(θ −θy)= (ϕ−ϕ y)Lpl
linear
non-
PLASTIC 0-length
M
Alternative 2
θ−θy
7
Example 1
Lpl = 200 mm
EIel-b = 1013 N-mm2
EIel-h = 1013 N-mm2 kel-h = EIel-cp/Lpl = 5 x1010 N-mm
EIpl-h = 5,2x1010 N-mm2 kpl-h = EIpl-cp/Lpl = 2,6x108 N-mm
8
Example 1
60
(0.02, 50)
50
(0.0009, 45)
40
M (kN-m)
30
20
10
0
0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0,018 0,02
θ (rad)
kel-h = 5 x104 kN-m

EIel-b = 104 kN-m2
kpl-h = 2,6x102 kN-m
9
Example 1
P=1 kN P1 (kN)
λ3P = 17
λ2P = 15
λ1P = 7.5
Load History
P1 = λP U1
λ= {7.5, 15, 17}
10
Example 1
u2
u1
U2
U1
 U1 
nodal  
U = U2 
displ.s U 
 3 u4
u3
 P1 
nodal  
P =  P2 
forces P 
U3  3 u2
plastic
hinge θ h = u 2 − u1 = u 2
rotation
u1
11
Example 1
LOAD STEP 1: λ1P = 7.5 kN i=1 u2

u1
U2
U1 0 
0    0 
  0  0   
U = 0  ⇒ Ptr =   Ph =   ⇒ PR = 0 
0  0  0  0 
  0   
7.5
  u4
P= 0 
0  u3
 
U3 7.5
  u2
∆P = Punb = P − PR =  0 
0 
 
u1 12
Example 1
SIGN CONVENTION: When equilibrium is reached:

equal and opposite (from equilibrium)
P
7.5
7.5
Resisting forces P + PR = 0
at node
7.5 Resisting forces

at element end
P - PR = 0
Convention used here:

formally less correct
easier to represent
13
Example 1
LOAD STEP 1: λ1P = 7.5 kN

60
50
U2 i=1 40
M (kN-m)
30
U1
20
Initial stiffness
10
0
0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0,018 0,02
EIb = 104 kN-m2 θ (rad)
kh = kel-h = EIel-h/Lpl = 5 x104 kN-m
U3 Lb = 3 m
12EI b 6EI b 6EI b 
 L3 L2b L2b 
 b 
 6EI 4EI b 2EI b 
K = K el =  2 b 
 Lb Lb Lb 
 6EI b 2EI b 4EI b 
 2 + k el − h 
 Lb Lb Lb  14
Example 1
Kel
Graphic is purely indicative
Axes represent arrays!
λ1P = 7.5
Punb
15
Example 1

u1
U2
U1
 0.0081 
 
∆U = K -1 {∆P} =  −0.0038 
 −0.00045
 
 0.0081 
  u4
U = U + ∆U =  −0.0038  u3
 −0.00045
 
U3  0 
 0.0081  Uh =   u2
 −0.0038   −0.00045
 
Ub =   ⇓
 0  θh = −0.00045
 −0.00045
u1 16
Example 1
Kel
Determine resisting forces corresponding to

the current displacement vector U
(STRUCTURE STATE DETERMINATION)
7.5
U
U 17
Example 1 u2
u1
LOAD STEP 1: λ1P = 7.5 kN i=1
Elements’ resisting forces

(ELEMENT STATE DETERMINATION)
1) Column: linear elastic Pb = K b U b u4

u3
2) Plastic hinge Mh = min(kel-h θh , M* + kpl-h θh) = -22.5 kN-m
60
kh = kel-h u2
50
40
M (kN-m)
30
kel-h
20
22.5 u1
10
0
0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0,018 0,02
θ (rad)
18
0.00045
Example 1
F F=kel u
F=F*+kpl u
F* kpl
kel F = min(kel u , F* + kpl u)
F = max(kel u , F* + kpl u)
19
Example 1

0
U2 7.5
U1 7.5
 7.5 
 0 
   22.5 
Pb =   Ph =  
 −7.5  −22.5
 22.5 
7.5 22.5
  7.5
PR =  0 
0 
U3  
7.5 7.5 0  22.5
     
∆P = Punb = P − PR =  0  −  0  = 0  = 0
 0   0  0 
     
22.5 20
Example 1
7.5
In equililbrium
7.5
7.5 7.5
External Forces on nodes
Resisting Forces on nodes
Element Forces
7.5
22.5
22.5 7.5
7.5 22.5
22.5 7.5
In equililbrium 7.5 22.5
22.5
7.5
7.5
22.5
21
In equililbrium
Example 1
Kel
There is equilibrium
between applied and
resisting forces
→ Load increment
7.5 Punb = 0 Apply λ2P
∆U
22
Example 1
Kel
λ2P = 15
Punb
7.5
23
Example 1
15
Not in equililbrium!!
7.5
15 7.5
External Forces on nodes
Resisting Forces on nodes
Element Forces
7.5
22.5
22.5 7.5
7.5 22.5
22.5 7.5
In equililbrium 7.5 22.5
22.5
7.5
7.5
22.5
24
In equililbrium
Example 1
LOAD STEP 2: λ2P = 15 kN i=1 u2

u1
U2
U1 7.5 15
   
PR =  0  P=0
0  0
   
7.5
 
∆P = Punb = P − PR =  0 
0  u4
  u3
K = K el
U3  0.0081 
 
∆U = K -1 {∆P} =  −0.0038  u2
 −0.00045
 
u1 25
Example 1
Kel
15
Punb
7.5 (STRUCTURE STATE DETERMINATION)
∆U 26
Example 1

u1
U2
U1
 0.0162 
 
U = U + ∆U =  −0.00765
 −0.0009 
 
 0  u4
 0.0162  Uh =  
 −0.00765  − 0.0009  u3
 
Ub =   ⇓
 0 
U3 θh = −0.0009
 −0.0009  u2
u1 27
Example 1 u2
u1
LOAD STEP 2: λ2P = 15 kN i=1


u3
2) Plastic hinge Mh = min(kel-h θh , M* + kpl-h θh) = -45 kN-m
60
kh = kel-h u2
50
45
40
M (kN-m)
30
20 u1
10
kel-h
0
0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0,018 0,02
θ (rad)
28
0.0009
Example 1

0
U2 15
 15 
U1 15  0 
   45 
Pb =   Ph =  
 − 15   −45
 45 
15
 
PR =  0 
0 45
  15
15 15 0 
     
U3 ∆P = Punb = P − PR =  0  −  0  = 0 
 0   0  0  45
     
45 29
Example 1
Kel
There is equilibrium
15 Punb=0 between applied and
resisting forces
→ Load increment
Apply λ3P
7.5
30
Example 1

u1
U2
U1 15 17 
   
PR =  0  P=0
0 0
   
2
 
∆P = Punb = P − PR =  0 
0  u4
  u3
K = K el
U3  0.00216 
  u2
∆U = K -1 {∆P } =  −0.001 
 −0.00012 
 
u1 31
Example 1
Ke
λ3P = 17
Punb
15

7.5 (STATE DETERMINATION)
∆U 32
Example 1

u1
U2
U1
 0.01836 
 
U = U + ∆U =  −0.00867 
 
 −0.00102 
 0  u4
 0.01836  Uh =  
 −0.00867   − 0.00102  u3
 
Ub =   ⇓
U3  0  θh = −0.00102 u2
 −0.00102 
u1 33
Example 1 u2
u1


u3
2) Plastic hinge Mh = min(kel-h θh , M* + kpl-h θh) = -45,03 kN-m
60
kh = kpl-h u2
50 kpl-h
45,03
40
Momento (kN-m)
30
20 u1
10
0
0 0.005 0.01 0.015 0.02 0.025
34
0.00102 rotazione θcp
Example 1

0
U2 17
U1 17  17 
 0 
   45.0312 
Pb =   Ph =  
 −17   −45.0312 
 51 
 17 
 
PR =  0 
5.9688 
 
51
17
17   17   0 
     
∆P = Punb = P − PR =  0  −  0  =  0 
U3  0  5.9688  −5.9688
      45.03
There is no equilibrium
between applied and
resisting forces 45.03 35
Example 1

u1
U2
U1
K = K pl
 0.06887 
 
∆U = K -1 {∆P} =  −0.0229 
 −0.0229  u4
 
u3
 0.08723 
 
U3 U = U + ∆U =  −0.0316 
u2
 −0.0240 
 
u1 36
Example 1
17
Kep
15 Punb
7.5
∆U 37
Example 1 u2
u1


u3
2) Plastic hinge Mh = min(kel-h θh , M* + kpl-h θh) = -51 kN-m
60
51 Kh = kpl-h u2
50
40
Momento (kN-m)
30
20 u1
10
0
0 0.005 0.01 0.015 0.02 0.025
38
rotazione θcp
0.024
Example 1

0
U2 17
U1 17
 17 
 0 
   51 
Pb =   Ph =  
 − 17   −51
 51 
17  51
  -17
PR =  0 
0
 
U3
17  17  0  -51
     
∆P = Punb = P − PR =  0  −  0  = 0 
 0   0  0 
     
51 39
Example 1
17 Punb=0
15
7.5
40
Example 1
(87.23,17)
(16.2,15)
(8.1,7.5)
41
Example 1
(87.23,17)
(16.2,15)
(8.1,7.5)
42
Example 1
The load path is not known: A load history is the applied
P=1 kN
P1 (kN)
λ8P = 16
λ7P = 14
λ6P = 12
λ5P = 10
λ4P = 8
λ3P = 6
Load history λ2P = 4
P1 = λP λ1P = 2
λ= {2, 4, 6, 8, 10 ,12, 14, 16, …} U1 43
Example 1
P=1 kN
22
20
18
16
14
P 1 (kN)
12
10
8
6
4
2
0
0 50 100 150 200
U1 (mm)
Load history
P1 = λP
λ= {0, 20}
44
Example 1
P=1 kN
22
20
18
16
14
P 1 (kN)
12
10
8
6
4
2
0
0 50 100 150 200
U1 (mm)
Load history
P1 = λP
λ= {0, 10, 20}
45
Example 1
P=1 kN
22
20
18
16
14
P 1 (kN)
12
10
8
6
4
2
0
0 50 100 150 200
U1 (mm)
Load history
P1 = λP
λ= {0, 5, 10, 15, 20}
46
Example 1
P=1 kN
22
20
18
16
14
P 1 (kN)
12
10
8
6
4
2
0
0 50 100 150 200
U1 (mm)
Load history
P1 = λP
λ= {0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20}
47
Example 2
PLASTIC HINGE
Phenomenological
P, U
nonlinear M-θ model
u2
u1
rotation = u2-u1
48
Example 2
SYSTEM RESPONSE
(closed form solution)
18
P, U
16
14
12
P1 (kN)
10
0
0 10 20 30 40 50 60 70 80 90
U1 (mm)
49
Example 2

u1
U2
U1 0 
0    0 
  0  0   
U = 0  ⇒ Pb =   Ph =   ⇒ PR = 0 
0  0  0  0 
  0   
7.5
  u4
P= 0 
0  u3
 
U3 7.5
  u2
∆P = Punb = P − PR =  0 
0 
 
u1 50
Example 2
60
LOAD STEP 1: λ1P = 7.5 kN 50
U2 i=1 40
M (kN-m)
U1 30
Initial stiffness 20
10
EIb = 104 kN-m2 0

0 0,005 0,01 0,015 0,02 0,025
θcp
kh = kel-h = EIel-h/Lpl = 5 x104 kN-m
U3 Lb = 3 m
12EI b 6EI b 6EI b 
 L3 L2b L2b 
 b 
 6EI 4EI b 2EI b 
K 0 = K el =  2 el 
 Lb Lb Lb 
 6EI b 2EI b 4EI b 
 2 + k el − h 
 Lb Lb Lb  51
Example 2
LOAD STEP 1: λ1P = 7.5 kN u2

u1
U2 i=1
U1
 0.0081 
 
∆U = K -10 {∆P} =  −0.0038 
 −0.00045
 
 0.0081 
  u4
U = U + ∆U =  −0.0038  u3
 −0.00045
 
U3  0 
 0.0081  Uh =   u2
 −0.0038   − 0.00045 
 
Ub =   ⇓
 0  θh = −0.00045
 −0.00045
u1 52
Example 2

u2
Elements’ resisting forces u1
Column: linear elastic Pb = K b U b
Plastic hinge Mh = -15.07 kN-m
60
50 u4
u3
40
Momento (kN-m)
30
20
u2
15.07
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.00045 u1 53
Example 2
PLASTIC HINGE 1: λ1P = 7.5 kN i=2

0
U2 7.5
 7.5 
U1 7.5  0 
   15.066586 
Pb =   Ph =  
 − 7.5   − 15.066586 
 22.5 
 7.5 
 
PR =  0 
7.433414  22.5
 
7.5  7.5   0  7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  7.433414   −7.433414 
     
15.07
There is no equilibrium between
applied and resisting forces
Apply Punb
15.07 54
Example 2

u1
U2 i=2
U1
 0.00044 
 
∆U = K -10 {∆P} =  −0.00015
 −0.00015
 
 0.00854  u4
 
U = U + ∆U =  −0.00397  u3
 −0.000599 
 
U3  0 
 0.00854  Uh =   u2
 −0.00397   − 0.000599 
 
Ub =   ⇓
 0  θh = −0.000599
 −0.000599 
u1 55
Example 2

u2
60
50 u4
u3
40
Momento (kN-m)
30
u2
19.43 20
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.000599 u1 56
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   19.430188 
Pb =   Ph =  
 − 7.5   − 19.430188 
 22.5 
 7.5 
 
PR =  0 
3.069812 
  22.5
7.5  7.5   0  -7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  3.069812   −3.069812 
     
-19.43
Apply Punb
19.43
i =2
Note that Punb < Punb
i =1 57
Example 2

u1
U2
U1
 0.00018 
 
∆U = K -10 {∆P} =  −0.00006 
 −0.00006 
 
 0.00873 
  u4
U = U + ∆U =  −0.00403  u3
 −0.00066 
 
U3  0 
 0.00873  Uh =   u2
 −0.00403  − 0.00066 
 
Ub =   ⇓
 0  θh = −0.00066
 −0.00066 
u1 58
Example 2

u2
60
50 u4
u3
40
Momento (kN-m)
30
21.06 20
u2
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.00066 u1 59
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   21.060194 
Pb =   Ph =  
 − 7.5   − 21.060194 
 22.5 
 7.5 
 
PR =  0 
1.439806 
  22.5
7.5  7.5   0  7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  1.439806   −1.439806 
     
21.06
Apply Punb
21.06
i =3
Note that Punb < Punb
i =2 60
Example 2

u1
U2 i=15
U1
 0.00000000007 
 
∆U = K -10 {∆P} =  −0.00000000002 
 −0.00000000002 
 
 0.0089 
  u4
U = U + ∆U =  −0.00409  u3
 −0.00072 
 
U3  0 
 0.0089  Uh =   u2
 −0.00409   − 0.00072 
 
Ub =   ⇓
 0  θh = −0.00072
 −0.00072 
u1 61
Example 2

u2
Column: linear elastic Ph = K h U h
60
50 u4
u3
40
Momento (kN-m)
30
22.5 20
u2
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.00072 u1 62
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   22.499999 
Pb =   Ph =  
 −7.5   −22.499999 
 22.5 
 7.5 
 
PR =  0 
.00000113
 
22.5
7.5  7.5   0 
      7.5
Punb = P − PR =  0  −  0 = 0 
 0  .00000113  −.00000113
     
U3
22.5
Small enough!!
There is equilibrium between

i =15
Note that Punb ≈0 Apply λ2P
22.5 63
Example 2
Convergence was very slow because initial stiffness was used
8,00
7,00
K-initial
6,00
Punb(3) (kN--m)
5,00
4,00
3,00
2,00
1,00
0,00
1 3 5 7 9 11 13 15
iteration (i)
64
Example 2
The tangent stiffness does not change:

Advantage: K is inverted only once
Disadvantage: convergence is slow
What if the stiffness is updated at every

step (tangent stiffness)?
65
Example 2
Tangent stiffness (Newton-Raphson)
It should be much faster!
66
Example 2

u1
U2
U1 0 
0    0 
  0  0   
U = 0  ⇒ Pb =   Ph =   ⇒ PR = 0 
0  0  0  0 
  0   
7.5
  u4
P= 0 
0  u3
 
U3 7.5
  u2
∆P = Punb = P − PR =  0 
0 
 
u1 67
Example 2

u1
U2 i=1
U1
 0.0081 
 
∆U = K -1tan {∆P} =  −0.0038  for i = 1, K tan = K 0
 −0.00045
 
 0.0081  u4
 
U = U + ∆U =  −0.0038  u3
 −0.00045
 
U3  0 
 0.0081  Uh =   u2
 −0.0038   − 0.00045 
 
Ub =   ⇓
 0  θh = −0.00045
 −0.00045
u1 68
Example 2

u2

Kh,tan= 3.14 104 kN-m
60
50 u4
u3
40
Momento (kN-m)
30
20
u2
15.07
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.00045 u1 69
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   15.066586 
Pb =   Ph =  
 − 7.5   − 15.066586 
 22.5 
 7.5 
 
PR =  0 
7.433414 
  22.5
7.5  7.5   0  7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  7.433414   −7.433414 
     
15.07

Apply Punb
15.07 70
Example 2

u1
U2 i=2
U1
 0.00071 
 
∆U = K -1tan {∆P} =  −0.00024 
 −0.00024 
 
 0.00881 
  u4
U = U + ∆U =  −0.00406  u3
 −0.000687 
 
U3  0 
 0.00881  Uh =   u2
 −0.00406   − 0.000687 
 
Ub =   ⇓
 0  θh = −0.000687
 −0.000687 
u1 71
Example 2

u2

Kh,tan= 2.5 104 kN-m
60
50 u4
u3
40
Momento (kN-m)
30
21.74 20
u2
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.000687 u1 72
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   21.743466 
Pb =   Ph =  
 − 7.5   − 21.743466 
 22.5 
 7.5 
 
PR =  0 
0.756533
  22.5
7.5  7.5   0  7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  0.756533  −0.756533
     
21.74
Apply Punb
i =2 i =1
21.74 73
Note that Punb ≪ Punb
Example 2

u1
U2 i=3
U1
 0.00009 
 
∆U = K -1tan {∆P} =  −0.00003
 −0.00003
 
 0.0089 
  u4
U = U + ∆U =  −0.00409  u3
 −0.000717 
 
U3  0 
 0.0089  Uh =   u2
 −0.00400   − 0.000717 
 
Ub =   ⇓
 0  θh = −0.000717
 −0.000717 
u1 74
Example 2

u2

Kh,tan= 2.43 1010 N-mm
60
50 u4
u3
40
Momento (kN-m)
30
22.49 20
u2
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.000717 u1 75
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   22.488482 
Pb =   Ph =  
 − 7.5   − 22.488482 
 22.5 
 −7.5 
 
PR =  0 
 −0.011518 
  22.5
7.5  7.5   0  7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  0.011518   −0.011518 
     
22.49
Apply Punb
i =3
Note that Punb i =2
≪ Punb 22.49 76
Example 2

u1
U2 i=4
U1
 0.0000014 
 
∆U = K -1tan {∆P} =  −0.000000474 
 −0.000000474 
 
 0.0089  u4
 
U = U + ∆U =  −0.00409  u3
 −0.00072 
 
U3  0 
 0.0089  Uh =   u2
 −0.00409   − 0.00072 
 
Ub =   ⇓
 0  θh = −0.00072
 −0.00072 
u1 77
Example 2

u2

Kh,tan= 2.4289 104 kN-m
60
50 u4
u3
40
Momento (kN-m)
30
22.5 20
u2
10
0
0 0.005 0.01 0.015 0.02 0.025
rotazione θcp
0.00072 u1 78
Example 2

0
U2 7.5
 7.5 
U1 7.5  0 
   22.499999 
Pb =   Ph =  
 − 7.5   − 22.499999 
 22.5 
 7.5 
 
PR =  0 
0.0000028 
  22.5
7.5  7.5   0  7.5
     
∆P = Punb = P − PR =  0  −  0 = 0 
U3  0  0.0000028   −0.0000028 
     
22.5
Small enough!!
There is equilibrium between
i =4 i =3
Note that P
unb ≪ P unb applied and resisting forces
Apply λ2P
i =4
Punb ≈0 22.5 79
Example 2
COMPARISON BETWEEN CONVERGENCE SPEEDS
8,00
7,00
K-initial
6,00 K-tangent
(kN-m)
5,00
Punb(3) (kN
4,00
3,00
2,00
1,00
0,00
1 3 5 7 9 11 13 15
iteration
80
Example 3
NONLINEAR GEOMETRY: P-∆ EFFECT
P P∆ << HL P
Equilibrium
H H ∆
in the undeformed configuration
Mbase = HL
L otherwise
Equilibrium
in the deformed configuration
Mbase = HL + P∆
81
Example 3
60
50
Mhinge (kN-m)
40
30 NL law: Example 2
NL law: Example 2
20
10
0
0 0,005 0,01 0,015 0,02
rotation θcp
Nonlinear hinge law: from Example 2

82
Example 3
P
y ∆ M ( x ) = -Pv - Hx
H
EIv" = -Pv - Hx
θh P Hx
v" + v=-
x EI EI
P P H
v v ( x ) = C1 sin x + C2 cos x− x
L EI EI P
v ( 0 ) = 0 ⇒ C2 = 0
H P +θ h
v' ( L ) = θ h ⇒ C1 =
P P
cos L
EI EI
x
83
Example 3
EQUILIBRIUM IN THE DEFORMED CONFIGURATION
P H P +θh P H
v( x)= sin x− x
y ∆ H
P
cos
P
L
EI P
EI EI
θcp P
tan L
x ∆ = v ( L) =
H EI + θ h tan P L − H L
P P P EI P
v EI EI
L
from equilibrium M h = HL + P∆
M h − HL
∆=
P
Get closed form solution ∆

x
84
Example 3
EQUILIBRIUM IN THE DEFORMED CONFIGURATION
P
tan L
Mh = H EI + P θ h tan P L
P P EI
EI EI
P
H = Mh EI − Pθ h
P
tan L
EI
M - HL
∆= h
P
Assegna θ h ⇒ M h ⇒ H ⇒ ∆
85
Example 3
20
18
P 16
14
H
H [kN] 12
10
8
Cerniera EPI P=0
6
Cerniera NL P=0
Cerniera NL P=25 kN
4
Cerniera NL P=100 kN
2 Cerniera NL P=250 kN
Cerniera NL P=500 kN
0
0 20 40 60 80 100
∆ [mm]
π 2 EI
Pcr-el = 2
= 2740 kN 86
4L
Nonlinear geometry in Italian Seismic Codes
4.11.1.1 ... Second order effests

P (forces from all Secondo order effects
dr θ < 0,1
stories above) are neglected
V (total floor load) Horizontal seismic action

0,1< θ < 0,2 Effects are incremented by
1/(1-θ)
h
θ = P dr/V h 0,2 < θ < 0,3 No comment
V
θ > 0,3 Not allowed
87
Conclusions
1st order linear elastic analysis αP2 ∆

αP1
α
2nd order linear elastic analysis
Bifurcation
1s order inelastic analysis
2nd order inelastic analysis
∆
88
Conclusions
Elastic Analysis – Materials are all linear elastic Material

Inelastic Analysis – Materials are inelastic
First order analysis – Equilibrium in the underformed

configuration Geometry
Second order analysis – Equilibrium in the deformed
configuration (large displacements, small, moderate or
finite deformations)
Structural collapse is typically associated with loads that lead materials

into the inelastic range, and with displacements that lead to structural
instability at collapse
89
Conclusions
Elastic Analysis – Materials are all linear elastic Material

Inelastic Analysis – Materials are inelastic
First order analysis – Equilibrium in the underformed

configuration Geometry
Second order analysis – Equilibrium in the deformed
configuration (large displacements, small, moderate or
finite deformations)
Structural collapse is typically associated with loads that lead materials

into the inelastic range, and with displacements that lead to structural
instability at collapse
90
Nonlinear Quasi-Static Analysis of Structures
91
• The nonlinear behavior of the structure derives from

the nonlinear behavior of the elements
Pe
Ue 92
Curve is traced (computed) by points
Pe
Ue 93
Curve is traced (computed) by points
Pe
Ue 94
• Given the elements’ behavior, the structure response

(P-U curve) is nonlinear and computed only at a
discrete set of load levels.
• A reference external force vector (also called load
pattern) is defined for the analysis.
3 
1

 


0 

2 
2
Pref =  
 
1

0

 
 3


 


 
0 

95
• The external load is applied in increments (load steps)
P
λ4 Pref
λ3 Pref
λ2 Pref
λ1 Pref
U
96
• Generic load step “n”

P
λn Pref
Given the external load
Pn = λ n Pref
Find the displacements Un
for which
R n ≈ Pn
Un
U
97
• Incremental application of the external load
for n = 1, load_steps
Pn = λ n Pref
find Un for which R n ≈ Pn
end
98
• Discrete levels of applied load and corresponding
displacements
P
λ4 Pref
λ3 Pref
λ2 Pref
λ1 Pref
U1 U2 U3 U4
U
99
• Does a solution always exist? No, it depends ...
P
λ5 Pref ?
λ4 Pref
λ3 Pref
λ2 Pref
λ1 Pref
U 100
• If possible, a displacement-controlled analysis strategy is
used
P
λ1Uref λ3Uref λ5Uref λ7Uref U

λ2Uref λ4Uref λ6Uref 101
• How many load steps?
few ok Too many?

P P
U U
102
• Generic load step n+1: starting from the converged step/state n, the
external load λn+1Pref is applied
Pn+1
( Punb )n+1
i=0
Pn ≈ R n External loads: Pn +1 = λn +1 Pref

Resisting forces: R n ≈ Pn
( Punb )n+1 = Pn+1 − R n
i=0
Unbalanced forces:
Un
U
103
• Generic load step n+1
Pn+1 ≈ R n+1
Find the displacements U n+1 to
which correspond the internal
Rn (resisting) forces R n+1 such that
equilibrium between external and
internal forces is satisfied, i.e.
Pn +1 ≈ R n +1
U
Un U n+1
104
• Application of the unbalanced forces after linearization of the structural response
behavior about the converged state at the end of the previous load step (i.e., first-
order multi-dimensional Taylor series expansion of Punb):
i =0
 ∂P 
Punb ( U in=+11 ) ≡ ( Punb )n +1 ≅ ( Punb )n +1 +  unb  ⋅ δ U in=1 = 0
i =1 i =0
 ∂U n +1
: Linearized incremental equation of equilibrium about U = U in=+01 ≡ U n
P N-R Iteration i = 1
K iT,=0n +1 = K T, n
1
Pn +1
( Punb )n +1 = ∆Pni=+11
i =0
R n = R in=+01
δ Uin=1
U
i =0
Un = U n +1
105
i=0
 ∂P 
Punb ( U in=+11 ) ≡ ( Punb )n +1 ≅ ( Punb )n +1 +  unb  ⋅ δ U in=1 = 0
i =1 i =0
 ∂U  n +1
n+1 = U n (displ. at last converged load step)
where U i=0
( Punb )n +1 = Pn +1 − R in=+01 = Pn +1 − R n
i=0
i =0 i =0
 ∂P  ∂R ∂R
⇒  unb  = − n +1 = − n = K T, n = K iT,=0n +1
 ∂U  n +1 ∂U ∂U
: Structure tangent stiffness matrix at U = U in=+01 = U n
⇒ Punb ( U in=+11 ) ≅ ( Punb )n +1 − K iT,=0n +1 ⋅ δ U in=1 = 0

i =0
K iT,=0n +1 ⋅ δ U in=1 = ( Punb )n +1 ≡ ∆Pni =+11

i =0
−1
δ U in=1 = K iT,=0n +1  ⋅ ∆Pni=+11
106
• Updating of displacement vectors
Pn+1
Last incremental displacements:
δ U i=1
n
Total incremental displacements
(from last converged state):
Rn
∆U i=1
n = δ U i=1
n
∆Ui=1
n
Current total displacements:
δU i=1
n
U i=1
n +1 = U n + ∆U i=1
n
i=1
U
Un U n+1
107
• STRUCTURE STATE DETERMINATION: i = 1
Computation of internal resisting forces and tangent stiffness matrix
P IN
i =1
K T, n +1 =?
1 U in=+11 ,...
=1
n +1 = ?
R in+
Structure
K iT,=1n +1
R in=+11
OUT
i=1
U 108
U n+1
FOR ALL THE ELEMENTS
• Determine the element end displacements
e
P
(U )
e i =1
= E {U in=+11}
e
n +1
e
Operator that extracts the element end
E displacements from the structure nodal
displacement vector (using the element ID array)
Ue
(U )
e i =1
n +1
109
• ELEMENT STATE DETERMINATION: i = 1
Computation of element end forces and tangent stiffness matrix
(K )
e
P e i =1
T n +1 IN
1
(U )e i =1
n +1
,...
(P )
e i =1
n +1
Element
(K )
e i =1
T n +1
(P )
e i =1
n +1
e
U
(U )e i =1
n +1
OUT 110
• Assembly of structure tangent stiffness matrix and internal
resisting force vector: i = 1
P K iT,= 1n +1
1
R in=+11
= A (K )
Nel
i =1 e i =1
K T, n +1 T n +1
e =1
= A(P )
Nel
i =1 e i =1
R n +1
e =1 n +1
i=1
U
U n+1 111
• Check for convergence at i = 1: How small is Punb?
( Punb )n+1 = Pn+1 − R in=+11

i =1
P
Pn+1
=1
R ii=
n +1
( Punb )n +1
i =1
large
continue N-R
iterations
with i=2
U 112
• Application of unbalanced forces after linearizing the structural response
behavior about the trial state at the end of the first N-R iteration:
i =1
 ∂P 
Punb ( U in=+21 ) ≡ ( Punb )n +1 ≅ ( Punb )n +1 +  unb  ⋅ δ U in= 2 = 0
i=2 i =1
 ∂U  n +1
: Linearized incremental equation of equilibrium about U = U in=+11
N-R Iteration i = 2
P
=1
K iT,n
( Punb )n +1 = ∆Pni=+12
i =1 +1
Pn +1 1
R in=+11
δ Uin=1 δ U in=2
i =1 U
Un U n +1
113
i =1
 ∂P 
Punb ( U in=+21 ) ≡ ( Punb )n +1 ≅ ( Punb )n +1 +  unb  ⋅ δ U in= 2 = 0
i=2 i =1
 ∂U  n +1
where U i=1
n+1 (trial displ. at the end of the first N-R iteration)
( Punb )n +1 = Pn +1 − R in=+11
i =1
i =0 i =1
 ∂P  ∂R
⇒  unb  = − n +1 = K iT,=1n +1
 ∂U  n +1 ∂U
: Structure tangent stiffness matrix at U = U in=+11
⇒ Punb ( U in=+21 ) ≅ ( Punb )n +1 − K iT,=1n +1 ⋅ δ U in= 2 = 0

i =1
K iT,=1n +1 ⋅ δ U in= 2 = ( Punb )n +1 ≡ ∆Pni =+12

i =1
−1
δ U in=2 = K iT,=1n +1  ⋅ ∆Pni=+12
114
• Updating of displacement vectors
Pn+1
Last incremental displacements:
δ U i=n 2
Total incremental displacements
(from last converged state):
∆U i= 2
= ∆U i=1
+ δ U i= 2
∆Ui= 2 n n n
n
Current total displacements:
δU i=1
n
δU i= 2
n U i= 2
= U + ∆U i= 2
n +1 n n
i=1 i= 2
U
Un U n+1 U n+1
115
• STRUCTURE STATE DETERMINATION: i = 2
Computation of internal resisting forces and tangent stiffness matrix
P IN
K iT,=1n +1 Uin=+21 ,...

1
i=2
R n +1 =?
Structure
K iT,= 2n +1
R in=+21
OUT
i= 2 U
U n
116
FOR ALL THE ELEMENTS
• Determine the element end displacements
e
P
(U )
e i=2
= E {U in=+21}
e
n +1
Ue
(U )
e i =1
n +1
117
• ELEMENT STATE DETERMINATION: i = 2
Computation of element end forces and tangent stiffness matrix
P e (K )e i =2
T n +1 IN
1
(U )
e i=2
,...
(P )
e i=2
n +1
n+
n +1
Element
(K )
e i=2
T n +1
(P )
e i=2
n +1
Ue
(U )
e i =2
n +1
OUT
118
• Assembly of structure tangent stiffness matrix and internal resisting force
vector: i = 2
P
K iT,= 2n +1
1
R in=+21
= A (K )
Nel
i =2 e i =2
K T, n +1 T n +1
e =1
= A (P )
Nel
i =2 e i =2
R n +1
e =1 n +1
i= 2
U
U n+1
119
• Check for convergence at i = 2: How small is Punb?
P
( Punb )n+1 = Pn+1 − R in=+21
i =2
Pn+1
R in=+21
( Punb )n +1
i=2
"small"
go to the next
load step (n+2)
U
U n +1 = U i=2
n +1 120
• Summary of load step n+1
P
Pn+1
Newton-Raphson iterations
make use of the tangent
Pn ≈ R n
stiffness matrix
Un U n+1
U
121
• Summary of N-R iterations at load step (n+1) :
for i = 1, max_numb_iterations
∆Pni ++11 = ( Punb ) n +1 = Pn +1 − R in +1
i
−1
Compute K  i
T, n+1
−1
δ U n = K T, n+1  ⋅ ∆Pni ++11
i +1
 i

∆U in+1 = ∆U in + δ U in+1 , U in++11 = U n + ∆U in+1
for e = 1, all elements (Element State Determination)
(U ) = E {U in++11}
i +1 e
e
n +1
(K ) (P )
i+1 i+1
e e
Compute T n+1 ,
n+1
end
= A (K )
e i +1
= A (P )
Nel Nel
i +1 i+1 e i+1
K T, n +1 T n +1 , R n+1 : Structure State Determination
e =1 e =1 n+1
( Punb )n +1 = Pn +1 − R i+1
i+1
n+1
If Punb ≅ 0, exit loop and increment n

end 122
• Graphical Summary :
P
P3
P2
Newton-Raphson iterations
using the tangent stiffness
matrix
P1
U
U1 U2 U3 123
• Graphical Summary :
P
P3
P2
P1
U
U1 U2 U3 124
• What do I see of the response?
P
P3
P2
P1
U
U1 U2 U3 125
• How many load steps to select?
126
• Number of load steps
P
FEW LOAD STEPS

Analysis:
Fast
Not very precise
U
127
P
MANY STEPS
Analysis:
Slower
More precise
U
128
150
100
MOMENT (kip-in)
50
APPLIED CURVATURE INCREMENTS

-50
∆φ = 0.1E-3 rad/in
∆φ = 0.3E-3 rad/in
-100
∆φ = 0.6E-3 rad/in
-2 -1 -1 0 1 1 2 2
CURVATURE (10-3 rad/in)
Example of section analysis: imposed curvature 129

• How do we define the load history?
130
• load history
There exist sveral ways to define the load history
Example:
Definition of reference load vector Pref
0 
3 1 
 
0 
2  
Pref = 2 
0 
1  
3
0 
  131
• load history
Definition of loading function λref
In the pseudo-time axis τ, a ∆τ is selected that can change
during the loading history
λref At load step n the loading functin and the
applied load vector are:
 n 
λrefn = λref ( τn ) = λref  ∑ ∆τl 
 l =1 
Pn = λref n Pref
τ is the pseudo-time
τ 132
• Which stiffness matrix to use?
133
• Stiffness matrix
• Tangent (Newton – Raphson)
P3
Faster
P2
Unstable near peaks
P1
Iteration with new K
U
134
U1 U2 U3
• Initial Stiffness
P3
Slower
P2
More stable than NR
P1
U
135
U1 U2 U3
• Modified Newton – Raphson
P3
Compromise
between two
P2 previous cases
P1
U
U1 U2 U3 136
• Total secant
P3
P2
P1
U
137
U1 U2 U3
• Incremental secant
P3
P2
P1
U
138
U1 U2 U3
• Modified secant
P3
P2
P1
U
139
U1 U2 U3
• Summary of procedure at load step n+1 (applied load Pn+1)
for i=1,max_numb_iterations
−1
∆Pn+1
i
= Pn +1 − R in+1
−1
eventually compute  K * 
−1
dUin+1 =  K *  ∆Pn+1
i
−1
∆Uin+1 = ∆Uin+1 + dUin+1 , Uin+1 = U n + ∆Uin+1
for e=1,all elements

(U )
e
Structure e i
= E U i
n +1 n +1
state determination
Given : U i
n+1
compute (K ) , (P )
e i
n +1
e i
n +1
i
end
Find : R n +1
= A (K ) = A (Pe )
nEL nEL
i e i i i
K n +1 R n +1
e n +1 e n +1
Punb = Pn+1 − R in +1
Convergence
if P unb
≈ 0 exit loop and increment n
check
end 140
• Summary of revised procedure at load step n+1
for i=1,max_numb_iterations
for e=1,all elements
(U )
e
e i −1
= E Uin−+11
n +1
Structure
(K e ) , ( Pe )
i −1 i −1
compute
state determination end
n +1 n +1
= A (K ) = A ( Pe )
nEL nEL
i −1 e i −1 i −1 i −1
K n +1 R n +1
e n +1 e n +1
Punb = Pn+1 − R in−+11

Convergence
if Punb ≈ 0 exit loop and increment n
check
−1
∆Pn+1
i
= Pn +1 − R in+1
−1
eventually compute  K * 
−1
dUin+1 =  K *  ∆Pn+1
i
−1
∆Uin+1 = ∆Uin+1 + dUin+1 , Uin+1 = U n + ∆Uin+1
end
141
• What do we mean by “eventually compute”?
• K is not inverted, but it is triangularized (this operation is

cheaper than the inversion)
• “Inverting” K remains the slow step of the procedure, thus
sometimes we prefer not to do it at every step (modified NR
rather than NR or other methods)
• Initial stiffness: only K0 is computed and inverted
(at the very first step)
K * = K i=0
n=1 = K 0
• Modified NR: K is computed and inverted only at first

iteration of load step) when i=1
K * = K i=0
n
142
• How is convergence checked?
143
• Possible convergence criteria
E = ∆U ⋅ ∆P
∆U < tolU ∆U E2
E1
E0
does not work well for “high” K
E < tolE
∆P
or
∆P < tolP Ei
< tolE but E0 must not be
Ei =0 too small
144
Does not work well for “small” K
• Convergence
How many iterations in a load step?

Programs have a default value and/or ask user to
fix it (for example i≤10)
If the convergence criteria is satifsied for i≤10 the
load step is complete and the load is incremened
to the next step
If the convergence criteria is not satifsied for
i=10, either the program stops, or it advances and
it flags out the problem
145
• Convergence
CONVERGENCE? YES→ Go to next load step

n=n+1
CONVERGENCE? NO
Program stops. The analysis restarts at the last
converged step using a smaller step
Some programs use automatic (time) step subdivision,
but there is no guarantee of convergence
Other programs continue and flag out problem
Program may diverge at following load steps
Program may converge at following steps (need to check
whether convergence takes place on spurious branch)
146
• Convergence “delayed” to following step
Pn+1
Pn
U
147
• Divergence
Pn+1
Pn
U
148
Importance of Nonlinear Program

Not all programs use the same procedures or
convergence rules
I prefer programs that require user to decide
what to do
User must be knowledgeable
More work for user (first times)
Smaller danger of errors
Black box programs for nonlinear anlayses should

be avoided
149
Summary of Quasi-Static Analysis of
Structures
Search for the displacement field which satisfies the static

equilibrium between external and internal resisting forces:
R(t) = P(t)
Linear Structure K ⋅ U(t) = P(t)
without iterations
R ( U ( τ ) ,0 ≤ τ ≤ t ) = P ( t )
Nonlinear Structure
with iterations
150

Chapter 7

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Chapter 7

Caricato da

Copyright:

Formati disponibili

ROSE School, Pavia, Italy Instructor: Enrico Spacone

NONLINEAR STRUCTURAL ANALYSIS Teaching Assistant: K. Wijesundara

The class notes© for this course were developed jointly by

7. Intro to Nonlinear Frame Analysis

Three examples are presented hereafter

kel-h = 5 x104 kN-m

LOAD STEP 1: λ1P = 7.5 kN i=1 u2

SIGN CONVENTION: When equilibrium is reached:

7.5 Resisting forces

Convention used here:

LOAD STEP 1: λ1P = 7.5 kN

EIb = 104 kN-m2 θ (rad)

kh = kel-h = EIel-h/Lpl = 5 x104 kN-m

LOAD STEP 1: λ1P = 7.5 kN i=1 u2

Determine resisting forces corresponding to

LOAD STEP 1: λ1P = 7.5 kN i=1

Elements’ resisting forces

1) Column: linear elastic Pb = K b U b u4

kel F = min(kel u , F* + kpl u)

LOAD STEP 1: λ1P = 7.5 kN i=1

LOAD STEP 2: λ2P = 15 kN i=1 u2

LOAD STEP 2: λ2P = 15 kN i=1 u2

LOAD STEP 2: λ2P = 15 kN i=1

Elements’ resisting forces

1) Column: linear elastic Pb = K b U b u4

LOAD STEP 2: λ2P = 15 kN i=1

LOAD STEP 3: λ3P = 17 kN i=1 u2

Determine resisting forces corresponding to

LOAD STEP 3: λ3P = 17 kN i=1 u2

LOAD STEP 3: λ3P = 17 kN i=1

Elements’ resisting forces

1) Column: linear elastic Pb = K b U b u4

LOAD STEP 3: λ3P = 17 kN i=1

LOAD STEP 3: λ3P = 17 kN i=2 u2

LOAD STEP 3: λ3P = 17 kN i=2

Elements’ resisting forces

1) Column: linear elastic Pb = K b U b u4

LOAD STEP 1: λ3P = 17 kN i=2

The load path is not known: A load history is the applied

LOAD STEP 1: λ1P = 7.5 kN i=1 u2

LOAD STEP 1: λ1P = 7.5 kN 50

EIb = 104 kN-m2 0

kh = kel-h = EIel-h/Lpl = 5 x104 kN-m

LOAD STEP 1: λ1P = 7.5 kN u2

LOAD STEP 1: λ1P = 7.5 kN i=1

Column: linear elastic Pb = K b U b

Plastic hinge Mh = -15.07 kN-m

PLASTIC HINGE 1: λ1P = 7.5 kN i=2

LOAD STEP 1: λ1P = 7.5 kN u2

LOAD STEP 1: λ1P = 7.5 kN i=2

Column: linear elastic Pb = K b U b

Plastic hinge Mh = -19.43 kN-m

LOAD STEP 1: λ1P = 7.5 kN i=2

LOAD STEP 1: λ1P = 7.5 kN i=3 u2

LOAD STEP 1: λ1P = 7.5 kN i=3

Column: linear elastic Pb = K b U b

Plastic hinge Mh = -21.06 kN-m

LOAD STEP 1: λ1P = 7.5 kN i=3

LOAD STEP 1: λ1P = 7.5 kN u2

LOAD STEP 1: λ1P = 7.5 kN i=15

Column: linear elastic Ph = K h U h

Plastic hinge Mh = -22.5 kN-m

LOAD STEP 1: λ1P = 7.5 kN i=15

There is equilibrium between