STERA3D Technical Manual

STERA 3D
STructural Earthquake Response Analysis 3D
Technical Manual
Version 7.0
Dr. Taiki SAITO
TOYOHOSHI UNIVERSITY OF TECHNOLOGY (TUT), JAPAN
1
UPDATE HISTORY
2008/07/08 STERA_3D Technical Manual Ver.2.0 is uploaded.

Masonry element is installed.
Viscous damper element is installed.
Chapter 7 and Chapter 8 are added.
Chapter 9 and Chapter 10 are added.
The definition of shear deformation for passive damper is added.
Explanation of floor element is added.
Connection panel is installed.
The freedom of walls in case they are connected in series is explained in Chapter 4.
The error for calculating yield rotation of nonlinear spring is fixed..
New definition of mass distribution is added in Chapter 5.
Element model for wall panel is added.
The equation of shear strength of wall element is modified.
Chapter for the release of unbalance force is added.
Missing part (2.10 Connection Panel) is supplemented. Appendix for Degrading
Tri-linear Slip Model and Modified Bi-linear Model are added.
Definition of effective slab width is added. Also, the Appendix for Degrading Tri-linear
Slip Model is modified. Operator splitting method is added for numerical integration
method.
Modified Bilinear Model for HDRB (High Damping Rubber Bearing) is added for
isolation devices.
2
Steel elements are added.
Column element with direct input is added.
K-brace (Chevron brace) is added.
SRC members are added for beam, column and wall.
For some isolation devices, strength reduction by dissipated energy is considered.
Hysteresis of Bouc-Wen Model is added for isolator and damper
Definition of External Spring is extended in three directions.
“5.5 Modal analysis”, “7.4 Calculation of ground displacement” are added.
Formulation of initial stiffness of nonlinear spring is fixed (Eqs. (3-1-34), (3-1-51))
“5.5 Modal analysis” is modified including participation factor, effective mass, etc.
“7.4 Calculation of ground displacement” is modified changing band-pass filter.
Ground springs are added.
“4.6 Mass matrix corresponding to independent degrees of freedom” is added.
2019/05/20 Radiation damping for ground springs is added.
2019/07/25 External force by Wind is added.
2019/10/08 Buckling hysteresis of a brace is added.
2020/03/16 Pile foundation is included for ground springs.
Air spring is added for an external spring.
For RC column and RC wall, the nonlinear bending springs independent in x and y
directions are introduced.
For Steel beam, the nonlinear shear spring for hysteresis damper is introduced.
Damage indices of members are introduced.
3
INDEX
1. Basic Condition
1.1 Coordinate
2. Constitutive Equation of Elements
2.1 Beam
2.2 Column
2.3 Wall
2.4 Brace
2.5 External Spring
2.6 Base Isolation
2.7 Masonry Wall
2.8 Passive Damper
2.9 Floor Element
2.10 Connection Panel
2.11 Ground Spring
3. Nonlinear Element Models
3.1 Beam
3.1.1 RC Beam
3.1.2 Steel Beam
3.1.3 SRC Beam
Appendix 3.1:
A-1. Hysteresis of Degrading Trilinear Slip Model
3.2 Column
3.2.1 RC Column
3.2.2 Steel Column
3.2.3 SRC Column
Appendix 3.2:
Plastic theory model for nonlinear interaction between shear and axial strength
3.3 Wall
3.3.1 RC Wall
3.3.2 Steel Wall (Brace)
3.3.3 SRC Wall (Brace)
3.4 External Spring
3.4.1 Lift up spring
3.4.2 Air spring
3.5 Base Isolation
Appendix 3.5:
A-1. Hysteresis of LRB (Lead Rubber Bearing)
4
A-2. Hysteresis of HDRB (High Damping Rubber Bearing)
A-3. Hysteresis of Lead Damper
A-4. Hysteresis of Elastic Sliding Bearing
A-5. Hysteresis of Bouc-Wen Model
3.6 Masonry Wall
3.7 Passive Damper
3.8 Ground Spring
3.8.1 Soil structure interaction
3.8.2 Cone model to calculate the static stiffness
3.8.3 Embedded foundation
3.8.4 Radiation damping
3.8.5 Complex stiffness with material damping
3.8.6 Impedance matrix
3.8.7 Pile foundation
3.8.8 Equivalent period and damping factor considering soil structure interaction
4. Freedom Vector
4.1 Node freedom
4.2 Freedom vector
4.3 Dependent freedom
4.4 Transformation matrix of dependent freedom
4.5 Stiffness matrix corresponding to independent degrees of freedom
4.6 Mass matrix corresponding to independent degrees of freedom
5. Equation of Motion
5.1 Mass matrix
5.2 Stiffness matrix
5.3 Modal analysis
5.4 Damping matrix
5.5 Input ground acceleration
5.6 External force by vibrator
5.7 External force by wind
5.8 Numerical integration method
5.9 Energy
6. Nonlinear Static Push-Over Analysis
6. 1 Lateral distribution of earthquake force
6. 2 Capacity Curve
7. Lumped Mass Model
7.1 Decomposition of shear and flexural deformation
7.2 Lumped mass model with shear and flexural stiffness
8. P-D effect
5
9. Unbalance force correction
10. Calculation of ground displacement
11. Damage Index
6
1. Basic Condition
1.1 Coordinate
(1) Global Coordinate
The global coordinate is defined as the left-hand coordinate as shown in Figure 1-1-1.
Z 3
6
X 1
4
5
Y
2
Figure 1-1-1 Global coordinate
Z 3 Z Z
6
X 1 7 X X
4
5 8
Y Y Y
2
(a) lateral and rotational displacement (b) shear displacement
7
(2) Local Coordinate
The local coordinate is defined for each element. The displacement freedoms and force freedoms are named
with subscripts indicating the coordinate direction and node name. For example, the local coordinate of a
beam element in Figure 1-2 is defined to have its x-axis in the same direction of the element axis. Also the
displacement and force freedoms of a beam element are expressed as shown in Figure 1-1-2.
z
u zA u zB
θ yA θ yB x
A B
Displacement freedoms
y
Q zA Q zB
Local coordinate
M yA M yB
A B
Force freedoms
Figure 1-1-2 Local coordinate of a beam element
8
2. Constitutive Equation of Elements
2.1 Beam
τ ' yA
δx
M ' yA M ' yB
A B
Nx
τ ' yB
elastic element
θ ' yA φ yA
M ' yA M ' yB M ' yA M ' yB
A B A B
N 'x θ ' yB φ yB
nonlinear bending springs
l'
θ ' yA  τ ' yA  φ yA  η yA  η yA
        M ' yA M ' yB
θ ' yB  = τ ' yB  + φ yB  + η yB  A η yB B
 δ '  δ '   0   0 
 x  x    
nonlinear shear springs
Figure 2-1-1 Element model for beam
Force-displacement relationship for elastic element

The relationship between the displacement vector and force vector of the elastic element in Figure 2-1-1 is
expressed as follows:
 l' l' 
 3EI − 0 
6 EI y
τ ' yA   y
  M ' yA 
   l' l'  
τ ' yB  = − 0  M ' yB  (2-1-1)

 δ '   6 EI y 3EI y
 N ' 
 x
 0 l'  x 
0
 EA 
where, E , I y , A and l ' are the modulus of elasticity, the moment of inertia of the cross-sectional area
along y-axis, the cross-sectional area and the length of the element. The rotational displacement vector of
the nonlinear bending springs is,
φ yA   f yA 0   M ' yA 
 =   (2-1-2)
φ yB   0 f yB  M ' yB 
where, f yA and f yB are the flexural stiffness of nonlinear bending springs at both ends of the element.
9
The force-deformation relationship of shear spring is
Qz = k z sz or sz = (1 k z ) Qz
From the relationship between shear force and moment,
 M' yA 
Qx = [1 l ' 1 l ']  
 M' yB 
The end rotational displacement due to shear deformation is obtained as,
 1 1 
η yA  1 l' 1 l' 1 l'  M' yA   k sz l'2 ksz l'2   M' yA 
=
  = k z ) Qz   (1 k z ) [1 l' 1 l'] =
 sz 1 l' (1 =    
η
   
yB 1 l '   1
 l '  M' yB   1 1   M' yB 
 k l'2 ksz l'2 
 sz
(2-1-3)
where, k sz is the shear stiffness of the nonlinear shear spring. Then, the displacement vector of the beam
element is obtained as the sum of the above three displacement vectors.
θ ' yA  τ ' yA  φ yA  η yA   M ' yA 

         
θ ' yB  = τ ' yB  + φ yB  + η yB  = [ f B ]M ' yB  (2-1-4)
 δ '  δ '   0   0   N' 
 x  x      x 
where,
 l' 1 l' 1 
 f yA + 3EI + k l'2 − + 2
0 
6 EI y k sz l'
 y sz

 l' 1 
=[ fB ]  f yB + + 2
0  (2-1-5)
 3EI y k sz l' 
 l' 
 sym. 
 EA 
[ f B ] is the flexural stiffness matrix of the beam element. By taking the inverse matrix of [ f B ] , the
constitutive equation of the beam element is obtained as,
 M ' yA  θ ' yA  θ ' yA 

  −1    
 M ' yB  = [ f B ] θ ' yB  = [k B ]θ ' yB  (2-1-6)
 N'  δ'  δ' 
 x   x  x
where, [k B ] is the stiffness matrix of the beam element.
10
Including rigid parts and node movement
Including rigid parts and node movement as shown in Figure 2-1-2, the rotational displacement vector is,
θ ' yA  θ yA − τ  (u zB − λ B l 'θ yB ) − (u zA + λ Al 'θ yA )

 = , τ =
θ ' yB  θ yB − τ  l'
u 
 1 1  1 1   zA 
θ +
 yA l ' zAu + λ θ − u + λ θ
B yB   − 1 + λ λ B  u
A yA zB A
 zB 
= l'
 = l' l'
   (2-1-7)
1 1 1 1 θ
θ yB + u zA + λ Aθ yA − u zB + λ Bθ yB   − λA 1 + λ B   yA 
 l' l'  l' l'  θ yB 
 
θ yB
θ ' yA θ yA τ
θ yB
θ yA τ θ ' yB
u zB
Z
u zB − λ B l 'θ yB
u zA u zA + λ A l 'θ yA
A B X
λ Al ' l' λB l '
Figure 2-1-2 Including rigid parts and node movement
From node axial displacements, relative axial displacement is,

δ ' x = δ xB − δ xA (2-1-8)
Therefore
θ ' yA  θ ' yA 
θ ' yA  1 0 0 0    

θ '
 
= 0 1 0 0  θ ' yB  = [n ]θ ' yB  (2-1-9)
 yB     B  
 δ '  0 0 − 1 1  δ xA   δ xA 
 x   δ   δ xB 
 xB 
Combining Equations (2-1-7) and (2-1-9),
u zA  u zA 
1 1   u 
θ ' yA   l ' − l ' 1 + λ A λB 0 0 u zB
   zB 
θ '   1 1  θ  θ 
 yB  
  = l' − l' λ A 1 + λB 0 0  yA  = [Λ B ] yA  (2-1-10)
δ   θ yB θ yB 
 xA  0 0
 δ xB   0 0 1 0   δ xA 
δ xA
0 0
 0 0 0 1    
δ xB  δ xB 
11
Out of plane deformation of beam
If we consider out-of-plane deformation of beam in case of flexible floor, as shown in Figure 2-1-4, the
rotational displacement vector is,
θ ' zA  θ zA − τ  (u yA − λ Al 'θ zA ) − (u yB + λ B l 'θ zB )

 = , τ =
θ ' zB  θ zB − τ  l'
u 
 1 1   1 1  yA
θ zA − l ' u yA + λ Aθ zA + l ' u yB + λBθ zB  − l ' 1 + λA λB  u 
l'  yB 
= = 1   (2-1-11)
1 1 1
θ zB − u yA + λ Aθ zA + u yB + λBθ zB  − λ A 1 + λB  θ zA 
 l' l'   l' l'  θ zB 
λ Al ' l' λB l '

Y
u yA θ zA u yB
θ ' zA τ
θ zB
τ θ zB θ ' zB
θ zA
Figure 2-1-3 Beam displacement with rigid connection (X-Y plane)
We can summarize for both ends as
u 
 1 1   zA 
− l ' 1 + λA λB  u zB 
l'
θ ' yA   1 1
 θ 
θ '   − λ A 1 + λB   yA 
 yB   l ' l'  θ yB 
θ 'zA   1 1
 
  =  − 1 + λA λB  u yA  (2-1-12)
θ '
 zB   l' l'  u yB 
 δ xA   1 1  
  − λ A 1 + λB  θ zA 
l' l'
 δ xB    θ zB 
 1  
  δ xA
 1  
δ xB 
12
From global node displacement to element node displacement
Transformation from global node displacements to element node displacements is,
u zA 
u 
 zB   u1 
θ yA  u 
 2
  = [TixB ]  (2-1-13)
θ yB  
δ xA  un 
 
δ xB 
The component of the transformation matrix, [TixB ] , is discussed in Chapter 4 (Freedom Vector).
From global node displacement to element face displacement

Transformation from the global node displacement to the element face displacement is,
 u1   u1 
θ ' yA  u  u 
   2  2
θ ' yB  = [nB ][Λ B ][TixB ]  = [TxB ]  (2-1-14)
δ '   
 x un  un 
Transformation matrix for nonlinear spring displacement

The nonlinear spring displacement vector is obtained from the element face displacement as,
 
φ yA   f yA 
0 0  M' yA 
    
φ yB  =  0 f yB 0   M' yB  (2-1-15)
s   1 1

N' 
 z  0  x 
 ksz l' ksz l' 
13
In case of Y-direction beam
z Z
y X
x Y
Local coordinate of Y-beam Global coordinate
Figure 2-1-4 Relation between local coordinate and global coordinate
In case of Y-direction beam, the axial direction of the beam element coincides to the Y-axis in the global
coordinate, transformation of the sign of the vector components of the element coordinate is,
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-1-16)
z  
 Y − Beam  0 0 1  Z  Global
Therefore
u zA  1  u zA  u zA 
u    u 
 zB 
 1 0  u zB   zB 

θ yA   −1  θ xA  θ xA 
  =   = [s B ]  (2-1-17)
θ yB   −1  θ xB  θ xB 
δ xA   0 1  δ yA  δ yA 
      
δ xB Y − Beam  1 δ yB  Global δ yB  Global
Transformation from the global node displacement to the element node displacement is,
u zA 
u 
 zB   u1 
θ xA  u 
  = TiyB [ ]  2
  (2-1-18)
θ xB  
δ yA  u n 
 
δ yB 
 u1   u1 
θ ' yA  u  u 
 
[ ]
θ ' yB  = [n B ][Λ B ][s B ] TiyB
 2
[ ]
  = T yB
 2
  (2-1-19)
δ '   
 x u n  u n 
14
Constitutive equation
Finally, the constitutive equation of the X-beam is,
 P1   u1   u1 
P  u  u 
 2  2  2
  = [TxB ]T
[k B ][T ]
xB   = [K ]
xB   (2-1-20)
     
 Pn  un  un 
For Y-beam,
 P1   u1   u1 
P  u  u 
 2
  = TyB [ ] T
[ ] [ ]
[k B ] TyB  2  = K yB  2
  (2-1-21)
  
 Pn  un  un 
15
2.2 Column
Element model for column is defined as a line element with nonlinear bending springs at both ends and two
nonlinear shear springs in the middle of the element in x and y directions as shown in Figure 2-2-1.
X-Z plane Y-Z plane
M ' xB M ' xB
M ' yB
M ' yB
B B B
θ ' xB
θ ' yB
N 'z Z
l' = +
T 'z
θ ' yA θ ' xA X
A A A
M ' yA Y
M ' yA
M ' xA M ' xA
Figure 2-2-1 Element model for column

In the same way as the beam element, the relationship between the displacement vector and force vector of
the elastic element is,
 l' l' 
−
τ ' yA   3EI y 6 EI y   M ' yA 
 =  in X-Z plane (2-2-1)
τ ' yB  − l ' l '  M ' yB 
 6 EI y
 3EI y 
 l' l' 
−
τ ' xA   3EI x 6 EI x   M ' xA 
 = 
l '  M ' xB 
in Y-Z plane (2-2-2)
τ ' xB  − l '
 6 EI x 3EI x 
The axial displacement is,
l'
δ ' 'z = N 'z (2-2-3)
EA
The torsion angle by torque force is,
l'
θ 'z = T 'z (2-2-4)
GI z
where, G and I z are the shear modulus and the pole moment of inertia of the cross-sectional area.
16
Force-displacement relationship for nonlinear bending springs
Case 1: In the case that bending springs in x and y directions are independently defined
X-Z plane Y-Z plane
M ' yB M ' xB
B B
φ yB φxB
φ yA φxA
A A
M ' yA
M ' xA
The rotational displacement vector of the nonlinear bending spring is defined independently,
φ yA = f yA M' yA , φxA = f xA M'xA at end A (2-2-5)
φ yB = f yB M' yB , φxB = f xB M'xB at end B (2-2-6)
where, f yA , f xA , f yB , and f xB are the flexural stiffness of nonlinear bending springs at both ends of
the element.
It can be expressed as
φ yA  M ' yA   f yA 
 
[ ]
φ xA  = f pA
 
 M ' xA  ,  f pA  = 
 f xA 
 at end A (2-2-7)
ε   N'   0 
 zA   zA 
φ yB   M ' yB   f yB 
   
φxB  =  f pB   M 'xB  ,  f pB  =  f xB  at end B (2-2-8)
 
ε  N'   0 
 zB   zB 
17
Case 2: In the case that nonlinear interaction between moment and axial components is considered
Nonlinear interaction M x − M y − N z is considered in the nonlinear bending springs,
φ yA  M ' yA 
N ' zB , ε zB
 
[ ]
φ xA  = f pA
 
 M ' xA  at end A (2-2-9)
ε   N'  M ' xB , φ xB
 zA   zA 
M ' yB , φ yB
φ yB  M ' yB 
 
[ ]
φ xB  = f pB
 
 M ' xB  at end B (2-2-10)
B
ε   N' 
 zB   zB 
where, [ f pA ] and [ f pB ] are the flexural

stiffness matrices of the nonlinear bending springs.
Therefore, the force-displacement relationship of A
nonlinear bending springs is,
M ' yA , φ yA
φ yA   M ' yA 
φ  M '  M ' xA , φ xA
 xA   xA 
[ ]
ε zA   f pA 0   N ' zA 
N ' zA , ε zA
 =
φ yB   0 [ ] 
f pB  M ' yB 
 (2-2-11)
φ xB   M ' xB  Figure 2-2-3 Nonlinear bending springs

   
ε zB   N ' zB 
Rearrange the order of the components of the displacement vector and change the node axial displacements
into the relative axial displacement,
φ yA  φ yA 
φ yA  1 0 0 0 0 0   φ 
φ  0 φ xA
 yB   0 0 1 0 0    xA 
 ε zA  ε zA 
φ xA  = 0 1 0 0 0 0   = n p   [ ] (2-2-12)
φ  0  φ yB φ yB 
0 0 0 1 0  
 xB   φ xB  φ xB 
 ε z  0 0 − 1 0 0 1    
ε zB  ε zB 
The force-displacement relationship is then expressed as,
φ yA   M ' yA   M ' yA 
φ  M '  M ' 
[ ][ ] [ ][ ]
 yB   f pA 0   yB   yB 
φ xA  = n p 
T  
[ ]
n  M '  = f p  M ' xA 
f pB  p  xA 
(2-2-13)
φ   0 M' M ' 
  xB
 xB   xB 
 ε z   N ' z   N ' z 
18
Force-displacement relationship for nonlinear shear springs
Case 1: In the case that shear springs in x and y directions are independently defined
X-Z plane Y-Z plane
M ' yB M ' xB
B B
η yB
η xB
ksx
Qx , sx ksy
Qy , s y
η yA η xA
A A
M ' yA
M ' xA
Q'x   ksx 0   sx   sx  1 ksx 0  Q'x 

 =  ,  =   (2-2-14)
Q' y   0 ksy   s y   s y   0 1 ksy  Q' y 
 M' yA 
Q'x  1 l ' 1 l ' 0 0   M' yB 
 =   (2-2-15)
Q' y   0 0 1 l ' 1 l '  M'xA 
 M'xB 
η yA  1 l ' 0  1 l ' 0  1 ( ksx l') 0 

 
η 
 yB 
1 l ' 0   s 
=  x
1 l ' 0  1 k
   sx=
0  Q'x  1 ( ksx l') 0  Q' 
x
= 
 0 1 l '  s y 
 
 0 1 l '  0 1 ksy  Q' y   0 1 ( ksy l') Q' y 
 
η xA    
   
η 'xB   0 1 l '  0 1 l '  0
 1 ( ksy l') 
1 ( ksxl') 0   M' yA   M' yA 
 
1 ( ksxl') 0  1 l ' 1 l ' 0 
0   M' yB    M' 
 yB 
=
0 1 ( ksy l' )   0    = [ f s1 ]  
  0 1 l ' 1 l '  M'xA   M'xA 

 0 1 ( k sy l')   M'xB   M'xB 
(2-2-16)
19
 1 1 
 k l '2 0 0 
ksx l '2
 sx 
 1 1 
 k l '2 0 0 
ksx l '2
[ f s1 ] =  
sx
where (2-2-17)
1 1 
 0 0 
 ksy l '2 ksy l '2 
 1 1 
 0 0 
 ksy l '2 ksy l '2 
Case 2: In the case that nonlinear interaction between shear and axial components is considered
N 'zB
M 'xB
M ' yB
B
N z , ε sz
Qx , sx
Qy , s y
A
M ' yA
M 'xA
N 'zA
Figure 2-2-5 Nonlinear shear springs
Q'x   sx  l '  η y  η y 
         
Q' y  =
ksp   s y   ksp    η x   k 'sp  η x 
= l '=   (2-2-18)
    1 ε sz   
 N' z  ε sz   ε sz 
η y  Q'x 
    −1
=η x  =
f sp  Q' y  ,  f sp   k 'sp  (2-2-19)
ε   N' 
 sz   z
20
 M' yA   M' yA 
Q'x  1 l ' 1 l ' 0 0 0   M' yB   M' 
 yB 
  =  
Q' y   0 0 1 l ' 1 l ' 0   M'xA  [ Ls ]  M'xA  (2-2-20)
 N'   0 0 0 0 1   M'xB   M' 
 z    xB 
 N'z   N'z 
η yA  1  1   M' yA   M' yA 
  1  η 1     
η yB     y     M' yB   M' yB 
η xA  1  η x  sp  [ Ls ]  M' xA  [ f s 2 ]  M'xA 
= =  1   f=  (2-2-21)
η        M'   M' 
 xB   1  ε sz   1   xB   xB 
 ε sz   1  1  N'z   N'z 
where
1   f11 f12 f13 
1  f f12 f13  f f12 f13  1 l ' 1 l ' 0 0 0
   11  11
 1  f 21 f 22 
f 23  [ Ls ] =  f 21 f 22 
f 23   0 0 1 l ' 1 l ' 0 
   
 1   f31 f32 f33   f 21 f 22 f 23   0 0 0 0 1 
 1  f31 f32 f33 
(2-2-22)
 f11 l ' f11 l ' f12 l' f12 l' f13 
 f l' f l' f l' f12 l' f13 
 11 11 12
=  f 21 l ' f 21 l ' f 22 l' f 22 l' f 23 

 
 f 21 l ' f 21 l ' f 22 l' f 22 l' f 23 
 f31 l ' f31 l ' f32 l' f32 l' f33 
Both cases can be written in
η yA   M' yA 
η   M' 
 yB   yB  [ f ] 0
η xA  = [ f s ]  M'xA  , [ f s ] =  s1  (case 1), [ fs ] = [ fs2 ] (case 2) (2-2-23)
η   M'   0 0
 xB   xB 
 esz   N'z 
The displacement vector of the column element is obtained as the sum of the displacement vectors of
elastic element, nonlinear shear springs and nonlinear bending springs,
21
θ ' yA  τ ' yA  ϕ yA  η yA   M' yA 
θ '  τ '  ϕ  η   M' 
 yB   yB   yB   yB   yB 
θ 'xA  τ 'xA  ϕ xA  η   M' 

θ '
= 
τ '
 + 
ϕ
+  xA 
η
= [ fC ]  M' xA  (2-2-24)
 xB   xB   xB   xB   xB 
 δ 'z  δ ''z   εz   ε sz   N 'z 
         
 θ 'z   θ 'z elastic element  0 bending spring  0 shear spring  T 'z 
The flexural matrix [ f C ] is;
 l' l' 
 3EI − 
6 EI y
 y

 l' 
0
 3EI y 
 
 l' l' 
−
[ f C ] =  3EI x 6 EI x  +
l' 
 
 3EI x 
 l' 
 EA 
 l' 
 sym. 
 GI z  elestic element
 f p11 f p12 f p13 f p14 f p15 0


 f p 22 f p 23 f p 24 f p 25 0
 f p 33 f p 34 f p 35 0
  +
 f p 44 f p 45 0
 f p 55 0
 
 sym. 0 bending spring
[ f s ] 0 
  (2-2-25)
 0 0  shear spring
By taking the inverse matrix of [ f C ] , the constitutive equation of the column element is obtained as,
 M ' yA  θ ' yA  θ ' yA 

M '  θ '  θ ' 
 yB   yB   yB 
 M ' xA  θ ' xA 
−1  θ ' xA 
  = [ fC ]   = [k C ]  (2-2-26)
 M ' xB  θ ' xB  θ ' xB 
 N 'z   δ 'z   δ 'z 
     
 T ' z   θ ' z   θ ' z 
22
Change relative axial displacement and torsion displacement into node displacement,
θ ' yA  θ ' yA 
  θ ' 
θ ' yA  1  θ ' yB   yB 
θ '   1 0  θ '  θ ' xA 
 yB     xA   
θ ' xA   1  θ ' xB  θ ' xB 
 =   = [nC ]  (2-2-27)
θ ' xB   1  δ zA   δ zA 
 δ 'z   0 −1 1   δ zB   δ zB 
       
 θ 'z   − 1 1  θ zA   θ zA 
 θ zB   θ zB 
Including rigid parts and node movement,
u xA  u xA 
 1 1 
 u xB  u 
− 1 + λA λB     xB 
θ ' yA   l ' l'
 θ yA  θ yA 
θ '  − 1 1
λA 1 + λB 
0  θ yB   
 yB   l ' l'    θ yB 
θ ' xA   1 1  u yA  u yA 
   − 1 + λA λB    
l' l'  u
θ ' xB   u
   = [Λ C ] yB 
 yB 
 = 1 1
 δ zA   − λA 1 + λB  θ xA  θ xA 
l' l'  θ 
 δ zB   1 θ xB 
     xB   
 θ zA   1  δ zA  δ zA 
θ     δ 
 zB   0 1  δ zB   zB 
 
1 θ zA  θ zA 

θ  θ 
 zB   zB 
(2-2-28)
B u xB
u yB
λB l '
θ yB
θ xB θ ' yB
l'
θ 'xB
θ ' yA Z
θ 'xA
θ yA
λ Al ' X
θ xA
A u xA
u yA
Y
Figure 2-2-6 Including rigid parts and node movement
23
Transformation from global node displacement to element node displacement is;
u xA 
u 
 xB 
θ yA 
 
θ yB 
u yA   u1 
  u 
u yB   2
  = [T ]
iC   (2-2-29)
θ xA  
θ xB  un 
 
δ zA 
δ zB 
 
θ zA 
θ 
 zB 
The component of the transformation matrix, [TiC ] , is discussed in Chapter 4 (Freedom Vector).

θ ' yA 
θ ' 
 yB   u1   u1 
θ ' xA    u 
u 2   2
  = [n C ][Λ C ][T ]
iC   = [T ]
C   (2-2-30)
θ ' xB    
 δ 'z  u n  u n 
 
 θ ' z 
Finally, the constitutive equation of the column is;
 P1   u1 
P  u 
 2  2
  = [K C ]  (2-2-31)
 
 Pn  un 
where,
[K C ] = [TC ]T [kC ][TC ] (2-2-32)
24
The nonlinear spring displacement vector is obtained from
ϕ yA   M ' yA   M' yA   M' yA 

ϕ  M '   M'   M' 
 xA   xA   yB   yB 
 ε zA    f pA  0 N '    f pA  0   M'   M' 
    zA       n p  0   xA  [ f mC ]  M'xA 
T
  = =  
ϕ yB   0  f pB    M ' yB   0  f pB     M'
 xB   xB 
   
ϕ xB   M 'xB   N 'z   N 'z 
       
 ε zB   N 'zB   T 'z   T 'z 
(2-2-33)
and
 M' yA   M' yA 
 M' yA   M'   M' 
   yB   yB 
η y  Q'x   M' yB 
       M'   M' 
f sp  Q' y   f sp  [ =
η x  =
= Ls ]  M'xA   f sp  [ Ls ] [ I ] =
0   xA  [ f sC ]  M' xA 
ε     M'   M'xB   xB 
 sz   N' z   xB
  N 'z   N 'z 
 N'z     
 T 'z   T 'z 
(2-2-34)
25
2.3 Wall
Element model for wall is defined as a line element with nonlinear bending springs at both ends and three
nonlinear shear springs; one is in the middle of the wall panel and others are in the side columns as shown
in Figure 2-3-1.
M ' xB1 M ' yBc M ' xB 2
θ ' xB θ ' xB Z
θ ' yB
l' X
θ ' xA θ ' yA θ ' xA
Y
M ' xA1 M ' yAc M ' xA2
Figure 2-3-1 Element model for wall

In the same way as the beam element, the relationship between the displacement vector and force vector of
 l' l' 
−
τ ' yAc   3EI c 6 EI c   M ' yAc 
 = 
l '   M ' yBc 
in wall panel (2-3-1)
τ ' yBc   − l '
 6 EI c 3EI c 
 l' l' 
−
τ ' xA1   3EI 1 6 EI 1   M ' xA1 
 = 
l '   M ' xB1 
in side column 1 (2-3-2)
τ ' xB1  − l '
 6 EI 1 3EI 1 
 l' l' 
−
τ ' xA2   3EI 2 6 EI 2   M ' xA2 
 = 
l '   M ' xB 2 
in side column 2 (2-3-3)
τ ' xB 2  − l '
 6 EI 2 3EI 2 
The axial displacement is,
l'
δ ' ' zc = N ' zc (2-3-4)
EA
26
Force-displacement relationship for nonlinear bending springs
Nonlinear interaction M x − M y − N z is considered in the nonlinear bending springs,
N ' zBc , ε zBc

M ' yBc , φ yBc
M ' xB1 , φ xB1 M ' xB 2 , φ xB 2
B
A
M ' xA1 , φ xA1 M ' xA2 , φ xA2
M ' yAc , φ yAc
N zAc , ε zAc
Figure 2-3-2 Nonlinear bending springs
φ yAc   M ' yAc 

φ  M ' 
 xA1 
 [ ]
 = f pA
 xA1 
  at end A (2-3-5)
φxA2  M 'xA2 
ε zAc   N ' zAc 
φ yBc   M ' yBc 

φ  M ' 
 xB1 

φ
[ ]
 = f pA
 xB1 

M '
 at end B (2-3-6)
 xB 2   xB 2 
ε zBc   N ' zBc 
where, [ f pA ] and [ f pB ] are the flexural stiffness matrices of the nonlinear bending springs. Therefore,
the force-displacement relationship of nonlinear bending springs is,
27
φ  M ' 
 xA1   xA1 
φ xA2   M ' xA2 
 
[ ]
ε zAc   f pA

0   N ' zAc 

=

φ yBc   0 [ ]  
f pB   M ' yBc 
(2-3-7)
φ xB1   M ' xB1 

   
φ xB 2  M ' xB 2 
ε   N' 
 zBc   zBc 
Rearrange the order of the components of the displacement vector and change the node axial displacements
into the relative axial displacement,
φ yAc  φ yAc 
φ yAc  1   φ 
φ    φxA1   xA1 
 yBc   1
 φxA2  φxA2 
φxA1   1    
 ε zAc  ε zAc 
  
φxB1  =  1  φ  = n p φ  [ ] (2-3-8)
φ   1   yBc   yBc 
 xA2   
 φxB1  φxB1 
φxB 2   1    
ε    φxB 2  φxB 2 
 zc   −1 1   ε 
ε zBc   zBc 
The force-displacement relationship in Equation (2-3-7) is then expressed as,
φ yAc   M ' yAc   M ' yAc 

φ  M '  M ' 
 yBc   yBc   yBc 
φ xA1   M ' xA1   M ' xA1 
 
[ ][ ] [ ][ ]
 f pA
φ xB1  = n p 
0 

T  
[ ]  
n p  M ' xB1  = f p  M ' xB1  (2-3-9)
φ   0 f pB  M '  M ' 
 xA 2
  xA2   xA2 
φ xB 2  M ' xB 2  M ' xB 2 
ε   N'   N' 
 zc   zc   zc 
Force-displacement relationship for nonlinear shear springs

The force-deformation relationship of shear spring in the center is
Q'xc = ksc sxc , sxc = (1 k sc ) Q'xc
 M' yAc 
Q'xc = [1 l ' 1 l ']  
 M' yBc 
28
 1 1 
η yAc  1 l ' 1 l '  k l '2 ksc l'2   M'xAc 
=

η
= s
   xc   (1 k
= sc ) Q' xc
 sc   in wall panel (2-3-10)
 yBc  1 l ' 1 l '  1 1   M'xBc 
 k l'2 ksc l'2 
 sc
 1 1 
η xA1   ks1l' ks1l'2   M'xA1 
2
 =   in side column 1 (2-3-11)

η xB1   1 1   M'xB1 
 k l' 2 ks1l'2 
 s1
 1 1 
η xA 2   ks 2l' ks 2l'2   M'xA 2 
2
  =   in side column 2 (2-3-12)

η xB 2   1 1   M'xB 2 
 k l'2 ks 2l'2 
 s2
where, k sc , k s1 and k s 2 are the shear stiffness of the nonlinear shear springs.
θ ' yAc  τ ' yAc  φ yAc  η yAc   M ' yAc 

θ '  τ '  φ  η  M ' 
 yBc   yBc   yBc   yBc   yBc 
θ ' xA1  τ ' xA1  φxA1  η xA1   M ' xA1 
         
θ ' xB1  = τ ' xB1  + φxB1  + η xB1  = [ fW ] M ' xB1  (2-3-13)
θ '  τ '  φ  η  M ' 
 xA2   xA2   xA2   xA2   xA2 
θ ' xB 2  τ ' xB 2  φxB 2  η xB 2  M ' xB 2 
 δ '  δ ' '  ε     N' 
 zc   zc elastic element  zc bending spring  0 shear spring  zc 
The flexural matrix [ f W ] is;
29
 l' l' 
 3EI − 
6 EI c
 c

 l' 
 3EI c 
 l' l' 
 − 
 3EI 1 6 EI 1 
 l' 
[ fW ] =   +
 3EI 1 
 l' l' 
−
 3EI 2 6 EI 2 
 l' 
 sym. 
 3EI 2 
 l' 
 
 EAc  elestic element
 f p11  f p17 
 
    +
 f p 71 
 f p 77 
bending spring
 1 1 
 k l '2 ksc l'2 
 sc 
 1 
 ksc l'2 
 
 1 1 
 
 ks1l'2 ks1l'2 
 1  (2-3-14)
 
 ks1l'2 
 1 1 
 
 ks 2l'2 ks 2l'2 
 1 
 sym. 
 ks 2l'2 
 0  shear spring

By taking the inverse matrix of [ fW ] , the constitutive equation of the column element is obtained as,
 M ' yAc  θ ' yAc  θ ' yAc 

M '  θ '  θ ' 
 yBc   yBc   yBc 
 M ' xA1  θ ' xA1  θ ' xA1 
  −1    
 M ' xB1  = [ f W ] θ ' xB1  = [kW ]θ ' xB1  (2-3-15)
M '  θ '  θ ' 
 xA2   xA2   xA2 
M '
 xB 2  θ '
 xB 2  θ ' xB 2 
 N'  δ'  δ' 
 zc   zc   zc 
30
Change relative axial displacement and torsion displacement into node displacement,
θ ' yAc  θ ' yAc 

θ ' yAc  1   θ ' 
θ '   1  θ ' yBc   yBc 
 yBc    θ ' xA1  θ ' xA1 
θ ' xA1   1    
    θ ' xB1  θ ' xB1 
θ ' xB1  =  1  θ '  = [nW ]θ '  (2-3-16)
θ '   1   xA2   xA2 
 xA2   θ '
 xB 2  θ ' xB 2 
θ ' xB 2   1 
 δ'   
δ'     zAc  δ ' zAc 
 zc   − 1 1   δ ' 
δ ' zBc   zBc 
Including rigid parts and node movement,

u xAc  u xAc 
u  u 
 1 1   xBc   xBc 
− l ' 1+ λA λB  θ yAc  θ yAc 
l'
    
θ ' yAc  − 1 1
λA 1+ λB  θ yBc  θ yBc 
θ '   l ' l'   u yA1   u yA1 
 yBc   1 1    
θ ' xA1   − 1+ λA λB  u yB1  u yB1 
  l' l'
θ ' xB1   1 1  θ  θ 
 = − λA 1+ λB   xA1  = [Λ W ] xA1 
θ ' xA2   l' l'  θ xB1  θ xB1 
θ ' xB 2   1 1  u  u yA2 
− 1+ λA λB
   l' l'   yA2   
δ ' zAc   1 1  u yB 2  u yB 2 
δ '   − λA 1+ λB   θ 
 zBc   l' l'  θ xA2  xA2 
 1  θ 
 xB 2  θ xB 2 

 1 δ  δ 
 zAc   zAc 
δ zBc  δ zBc 
(2-3-17)
Transformation from the center displacements to the node displacements is,
θ yc
δ z 2 − δ z1
θ yc =
δ z1 δ zc δ z2 w
δ +δ
δ zc = z1 z 2
2
w
Figure 2-3-3 Relationship between center and node displacements
31
1 
 u xAc     u xA1   u xA1 
u   1 δ  δ 

 xBc   1 1   zA1   zA1 
θ yAc   − w w δ
 zA 2  δ zA 2 
   1 1  u   
θ yBc   −   xB1   u xB1 
 u yA1   w w  δ zB1  δ zB1 
   1  δ   
 u yB1   1   zB 2  δ zB 2 
θ     u yA1   u 
 = [DW ]
yA1

xA1
= 1  
θ
 xB1   u
  yB1  u
 yB1 
1
u yA 2    θ xA1  θ xA1 
   1    
u yB 2   1  θ xB1  θ xB1 
θ    u  u 
 xA 2   1   yA 2   yA2 
θ xB 2    u yB 2  u yB 2 
δ   1   θ 
θ
 zAc   0.5 0.5   xA2   xA2 
δ zBc    θ xB 2  θ xB 2 
 0.5 0.5  (2-3-18)
Transformation from the global node displacements to the element node displacements is;
 u xA1 
δ 
 zA1 
δ zA 2 
 
 u xB1 
δ zB1 
 
δ zB 2   u1 
 u  u 
 2
 = [TixW ] 
yA1
 (2-3-19)
u yB1  
θ xA1  u n 
 
θ xB1 
u 
 yA2 
u yB 2 
θ 
 xA2 
θ xB 2 
The component of the transformation matrix, [TixW ] , is discussed in Chapter 4 (Freedom Vector).

32
θ ' yAc 
θ ' 
 yBc   u1   u1 
θ ' xA1  u  u 
   2  2
θ ' xB1  = [nW ][Λ W ][DW ][TixW ]  = [TxW ]  (2-3-20)
θ '   
 xA 2
 u n  u n 
θ ' xB 2 
δ' 
 zc 
In case of Y-direction wall
z Z
y X
x Y
Local coordinate of Y-wall Global coordinate
In case of Y-direction wall, the wall panel direction coincides to the Y-axis in the global coordinate,
transformation of the sign of the vector components of the element coordinate is,
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-3-21)
z  
 Y −Wall  0 0 1  Z  Global
Therefore
33
 u xA1  1   u yA1   u yA1 
δ   1  δ  δ 
 zA1     zA1   zA1 
δ zA 2   1  δ zA 2  δ zA 2 
    u   
 u xB1   1   yB1  u yB1 
δ zB1   1  δ zB1  δ zB1 
      
δ zB 2   1  δ zB 2  δ zB 2 
 u   −1   u xA1  u 
 xA1 

yA1
 =   u  = [ε ]
W  
u yB1   −1   xB1   u xB1 
θ xA1   1  θ  θ yA1 
     yA1   
θ xB1   1  θ yB1  θ yB1 
u    u  u xA2 
 −1   xA2 
 yA2   
u yB 2   −1  u xB 2  u xB 2 
  θ 
θ 
 1  θ yA2 
 xA2     yA2 
θ   θ
1  yB 2  Global θ yB 2  Global
 xB 2  Y −Wall 
(2-3-22)
Transformation from the global node displacement to the element node displacement is;
 u yA1 
δ 
 zA1 
δ zA 2 
 
u yB1 
δ zB1 
 
δ zB 2   u1 
u  u 
 xA1 

u
[ ]
 = TiyW
 2
  (2-3-23)
 xB 1  
θ yA1  u n 
 
θ yB1 
u xA2 
 
u xB 2 
θ 
 yA2 
θ yB 2 
34
θ ' yAc 
θ ' 
 yBc   u1   u1 
θ ' xA1  u  u 
 
[n ][ ][D ][ε ][T ] 2
θ ' xB1  = W Λ W W W ixW   = T yW [ ]
 2
  (2-3-24)
θ '   
 xA 2
 u n  u n 
θ ' xB 2 
δ' 
 zc 
Finally, the constitutive equation of the wall is;
 P1   u1 
P  u 
 2  2
  = [K xW ]  (2-3-25)
 
 Pn  u n 
where,
[K xW ] = [TxW ]T [kW ][TxW ] (2-3-26)
For Y-wall,
 P1   u1 
P  u 
 2
[
  = K yW ]
 2
  (2-3-27)
 
 Pn  u n 
where,
[K ] = [T ] [k ][T ]
yW yW
T
W yW (2-3-28)

The nonlinear spring displacement vector is obtained from
35
ϕ yAc   M ' yAc   M ' yAc 
ϕ  M '  M ' 
 xA1   xA1   xA1 
ϕ xA 2   M 'xA 2   M 'xA 2 
    f pA  0   N 'zAc   
 ε zAc     N 'zAc 
  =   [ f mW ]  M ' 
ϕ yBc   0

 f pB    M ' yBc 
  yBc 
ϕ xB1   M 'xB1   M 'xB1 
     
ϕ xB 2   M 'xB 2   M 'xB 2 
ε  N'  N' 
 zBc   zBc   zBc 
 M ' yAc 
M ' 
 xA1 
M ' 
 sxc  1 ksc   Q'xc  1 ksc  1 l ' 0 1 l' 0   xA 2 
  =      N' 
 s y1   1 ks1   Q' y1   1 ks1  1 l' 0 1 l' 0   zAc 
  M'
sy 2   1 ks 2  Q' y 2   1 k s 2   1 l' 0 1 l ' 0   yBc 
 M 'xB1 
 
 M 'xB 2 
N' 
 zBc 
 
 
  f pA  0 
ϕ yAc   
ϕ   
 xA1     M ' yAc   M ' yAc 
ϕ xA 2    M '  M ' 
     xA1   xA1 
 ε zAc     M 'xA 2   M 'xA 2 
ϕ yBc   0  f pA 
   
     N 'zAc  =  f   N 'zAc 
ϕ xB1  =    M ' yBc   pW   M ' yBc 
ϕ      
 xB 2   
 M 'xB1   M 'xB1 
 ε zBc   1 0
1 
0 M '   
 s   ksc l' ksc l'  xB 2   M 'xB 2 

 xc  
1 1   N 'zBc  N' 
 zBc 
 s y1   0 0
   k s1l' ks1l' 
 sy2   
1 1
 0 0
 k s 2 l' k s 2 l' 
(2-3-29)
36
Furthermore, in the same way as Equation (2-3-8),
 M ' yAc 
M '   M ' yAc 
 xA1  M ' 
 M ' xA2   yBc 
   M ' xA1 
 N ' zAc 
  = np [ ] T  
 M ' xB1  (2-3-30)
 M ' yBc  M ' 
 M ' xB1   xA2 
  M ' xB 2 
M ' xB 2   N' 
 N'   zc 
 zBc 
Therefore, the nonlinear spring displacement vector is obtained from the element face displacement as,
ϕ yAc 
ϕ 
 xA1 
ϕ xA 2   M ' yAc 
  M ' 
 ε zAc   yBc 
ϕ yBc   M 'xA1 
  T  
ϕ xB1  =  f pW   n p   M 'xB1  (2-3-31)
ϕ  M ' 
 xB 2   xA 2 
 ε zBc   M 'xB 2 
η   N' 
 yc   zc 
 η x1 
 
 ηx2 
37
In case of direct input wall
Direct input wall model is defined as a line element with a nonlinear shear spring and a nonlinear bending
spring in the middle of the element as shown in Figure 2-3-1.
M ' yBc
kn Z
ks h X
kb
Y
M ' yAc
This model can be used as an alternative model so called the lumped mass model representing the restoring
force characteristics of each layer in the analysis of high-rise building as shown below. The detail of the
model is described in Chapter 7.1
Figure 2-3-6 Lumped mass model of high-rise building
38
STERA_3D adopts the formulation to have nonlinear shear and bending springs at the middle of the
element.
δ sx
B u xBc
Z
θ yBc ks
h φ yb θ yAc X
u xAc kb Y
A
Figure 2-3-7 Nonlinear bending and shear springs
Force-displacement relationship
The relationship between the displacement and force of the springs is,
 Qxs   k s 0 0  δ xs  δ xs 
   
= M yb 
=
 0 kb 0  φ yb  [ kW ] φ yb  (2-3-32)
N     
 zn   0 0 kn  ε zn  ε zn 
where, k s : the stiffness of the nonlinear shear spring

kb : the stiffness of the nonlinear bending spring
kn : the stiffness of the linear axial spring
Including node movement

The relationship between the shear spring displacement and nodal displacement is,
δ xs = ( u xBc − u xAc ) + hθ yBc

φ=
yb θ yBc − θ yAc (2-3-33)
ε=
zn δ zBc − δ zAc
In a matrix form
u xAc  u xAc 
u  u 
δ xs   −1 1 0 h 0 0   xBc   xBc 
   θ yAc 
 θ yAc 
 
φ yb  =
 0 0 −1 1 0 0 
 θ  [
= Λ W ] 
θ
 (2-3-34)
ε   0 0 0 0 −1 1   yBc   yBc 
 zn    δ  δ zAc 
zAc
   
δ zBc 
  δ zBc 
 
39
Transformation from the center displacements to the node displacements is,
θ yc
δ z 2 − δ z1
θ yc =
δ z1 δ zc δ z2 w
δ +δ
δ zc = z1 z 2
2
w
Figure 2-3-8 Relationship between center and node displacements
1 
u xAc   1   u xA1   u xA1 
u     δ 
 xBc   1 1   δ zA1   zA1 
θ yAc   −  δ  δ zA 2 
w w
  =   zA 2  [ DW ]  u  (2-3-35)
θ yBc  
−
1 1   u xB1   xB1 
δ zAc   w w  δ zB1  δ zB1 
      
0.5 0.5
δ zBc    δ zB 2  δ zB 2 
 0.5 0.5
Transformation from the global node displacements to the element node displacements is;
 u xA1 
δ 
 zA1   u1 
δ zA 2   
u 2 
  [ ixW ]  
= T (2-3-36)
 u xB1  
δ zB1  un 
 
δ zB 2 
The component of the transformation matrix, [TixW ] , is discussed in Chapter 4 (Freedom Vector).

Transformation from the global node displacement to the spring displacement is,
 u1   u1 
δ xs  u   
   2 u 2 
 yb  [ W ][ W ][ ixW ]   [ xW ]  
φ =
Λ D T T
= (2-3-37)
ε   
 zn
un  un 
40
z Z
y X
x Y
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-3-38)
z  
 Y −Wall  0 0 1  Z  Global
Therefore
 u xA1  1   u yA1   u yA1 

δ   1  δ  δ 
 zA1     zA1   zA1 
δ zA 2   1  δ zA 2  δ 
  =
1
  [εW ] uzA2  (2-3-39)
 u xB1     u yB1   yB1 
δ zB1   1  δ zB1  δ zB1 
      
δ zB 2 Y −Wall  1 δ zB 2 Global δ zB 2 Global
 u yA1 
δ 
 zA1   u1 
δ zA 2  u 
 2
  = TiyW    (2-3-40)
 u yB1  
δ zB1  un 
 
δ zB 2 
 u1   u1   u1 
δ ys  u  u  u 
   2  2  2
 xb  [ W ][ W ][ W ][ ixW ]   [ W ][ W ][ ixW ]    yW   
φ =
Λ D ε T =
Λ D T 
=T  (2-3-41)
ε  
  
  
 
 zn  un  un  un 
41
 P1   u1 
P  u 
 2  2
  = [K xW ]  (2-3-42)
 
 Pn  u n 
where,
[K xW ] = [TxW ]T [kW ][TxW ] (2-3-43)
For Y-wall,
 P1   u1 
P  u 
 2
[
  = K yW ]
 2
  (2-3-44)
 
 Pn  u n 
where,
[K ] = [T ] [k ][T ]
yW yW
T
W yW (2-3-45)
42
2.4 Brace
Element model for Brace is defined as a truss element with a nonlinear axial spring and pin-supported at
both ends as shown in Figure 2-6-1.
3 B 4
N', δ '
Z
h
1 2 X
A
w Y
Figure 2-4-1 Element model for brace
~ ~ ~ ~
f x4 , u x4 f x3 , u x3
4 3
N1 , δ 1
x N2, δ2
y x
~ ~ y ~ ~
f x1 , u x1 f x2 , u x2
1 2
(Brace 1) (Brace 2)
Figure 2-4-2 Local coordinate
The relationship between axial deformation and axial force of the truss element is,
N 1 = k1δ 1 (2-4-1)
N 2 = k 2δ 2 (2-4-2)
Replacing with the nodal force and displacement in local coordinate along the element,
~ ~
N 1 = − f1x = f 4 x , δ 1 = u~4 x − u~1x (2-4-3)
~ ~
N 2 = − f 2 x = f 3 x , δ 1 = u~3 x − u~2 x (2-4-4)
43
In a matrix form,
 u~1x 
u~ 
 1y 
~
u1x  u~2 x 
  ~ 
δ 1  − 1 0 0 1 u~2 x  − 1 0 0 0 0 0 1 0 u 2 y  ~
 =
δ 0 − 1 1 0  u~  =  0 0 − 1 0 1 0 0 0 u~  = [nb ]{u } (2-4-5)
 2   3x    3x 
u~4 x  u~3 y 
~ 
u 4 x 
u~4 y 
 
~
 f x1  − 1 0 
~  
 f z1   0 0 
~ ~
 f 1x  − 1 0  f x 2   0 − 1
~     ~   
 f 2 x   0 − 1  N 1 
~  =  → f = ~  =  {}
~  f z2   0 0   N1 
  
T N 
= [nb ]  1  (2-4-6)
 f 3x   0 1 N 2   f x3   0 1   N 2  N 2 
~
f   1   ~
f   0
 4x  0 0
 ~z 3  
 f x4   1 0
~ 
f   0 0 
 z4  
From Figure 2-4-3, the relation between the nodal forces in local coordinate and those of global coordinate
is,
~
f x1 = f x1 cos θ + f z1 sin θ
~ for Brace 1 (2-4-7)
f y1 = − f x1 sin θ + f z1 cos θ
and
~
f x 2 = − f x 2 cos θ + f z 2 sin θ
~ for Brace 2 (2-4-8)
f y 2 = − f x 2 sin θ − f z 2 cos θ
Eq. (2-4-8) can be also obtained from the Eq. (2-4-7) by replacing θ by (π − θ ) and using the
formulas sin (π − θ ) = sin θ , cos(π − θ ) = − cos θ .
44
f z4 ~ ~ f z3
f x4 f x3
4 3
f x4 f x3
h
l = w2 + h 2
f z1 ~ f z2
~ f 2x
f x1
1 2
f x2
f x1
f z1 f z2
f z1 cos θ f z1 sin θ
θ f x1 cos θ f z 2 sin θ θ
f z 2 cos θ
− f x 2 cos θ
θ f x1 θ
− f x2
f x1 sin θ − f x 2 sin θ
(Brace 1) (Brace 2)
Figure 2-4-3 Coordinate transformation
In a matrix form,
~
 f x1  c s   f x1   f x1 
~  − s c  f  f 
 f z1     z1   z1 
~
 f x2   f x2 
 −c s  f x2 
~      
 f z2  −s −c  f z2  f 
~  =  
  = [Cb ] z 2  (2-4-9)
  f x3
 f x3  
−c s
   f x3 
~  f z3 
f   −s −c  f z3 
 ~z 3     
 f x4 
 c s f x4   f x4 
~  
f  − s c   f z 4 Global f 
 z 4  Local   z4 
where
w h
c = cos θ = , s = sin θ =
l l
Since [C b ][C b ] = I ,
T
[Cb ] is an orthogonal matrix, therefore,
[Cb ]−1 = [Cb ]T (2-4-10)
45
In a similar manner, from Figure 2-4-4, the relation between the nodal displacements in local coordinate
and those of global coordinate can be obtained as,
u x1 = u~x1 cos θ − u~z1 sin θ
for Brace 1 (2-4-11)
u = u~ sin θ + u~ cos θ
z1 x1 z1
and
u x 2 = −u~x 2 cos θ − u~z 2 sin θ
for Brace 2 (2-4-12)
u = u~ sin θ − u~ cos θ
z2 x2 z2
Eq. (2-4-12) can be also obtained from the Eq. (2-4-11) by replacing θ by (π − θ ) and using the
formulas sin (π − θ ) = sin θ , cos(π − θ ) = − cos θ .
uz4 u z3
u~4 u~3
4 3
u x4 u x3
h
l = w2 + h 2
u~2 uz2
u z1 u~1
1 u x2
u x1 2
w
u z 2 = u~x 2 sin θ − u~z 2 cos θ
u~x 2
u z1 = u~x1 sin θ + u~z1 cos θ
u x 2 = −u~x 2 cos θ − u~z 2 sin θ

u~x1 θ
u~z1
θ
u x1 = u~x1 cos θ − u~z1 sin θ u~z 2
Figure 2-4-4 Coordinate transformation
In a matrix form,
u x1  c − s  u~x1  u~x1 
u  s c   u~   u~ 
 z1     z1   z1 
u x 2   −c −s  u~  u~x 2 
x2
    ~  ~ 
u z 2  s −c u z 2  T u z 2 
=   = [C b ]  ~ 
 
u   u~x 3 
 x 3

− c − s
~  u x 3 
u z 3   s −c  u z 3  u~z 3 
    ~ 
u x 4 
 c − s  u~x 4  u x 4 
  ~
u   
s c  u z 4  Global u~ 
 z 4  Local   z4 
46
The stiffness matrix of brace element is,
{N } = [k~ ]{δ }
 N 1   k1 0  δ 1 
 =   or (2-4-13)
N 2   0 k 2  δ 2 
Where
{δ } = [nb ]{u~} = [nb ][( Cb ]T ) {u} = [nb ][Cb ]{u}

−1
(2-4-14)
{ f } = [Cb ]−1 {~f } = [Cb ]T [nb ]T {N } (2-4-15)

 u1 
u 
{u} = [TixBr ] 2  (2-4-18)

u n 
The component of the transformation matrix, [TixBr ] , is discussed in Chapter 4 (Freedom Vector).

 u1   u1 
u  u 
{δ } = [nb ][Cb ]{u} = [nb ][Cb ][TixBr ]  = [TxBr ] 2 
 2
(2-4-19)
 
u n  u n 
Finally, the constitutive equation of the brace is;
 P1   u1 
P  u 
 2  2
  = [K xBr ]  (2-4-20)
 
 Pn  u n 
where,
[K xBr ] = [TxBr ]T [k Br ][TxBr ] (2-4-21)
47
In case of Y-direction brace
z Z
y X
x Y
In case of Y-direction brace, transformation of the sign of the vector components of the element coordinate
is,
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-4-20)
z  
 Y − Beam  0 0 1  Z  Global
Therefore
u x1  1   u y1   u y1 
u   1 u  u 
 z1     z1   z1 
u x 2   1  u y2
 u y 2 
      
u z 2   1  u z 2  u 
  =   =  z2  (2-4-21)
 1  u y3
u x 3     u y 3 
u z 3   1  u z 3  u z 3 
     
u x 4 
 1  u y 4  u y 4 
 
u 
 z 4 Y − Brace  1 u z 4  Global u z 4  Global
 
 u1 
u 
{u} = [TiyBr ] 2
  (2-4-22)

u n 
 u1 
u 
{δ } = [T yBr ]
 2
 , [T ] = [n ][C ][T ]
yBr b b iyBr (2-4-23)

u n 
48
The constitutive equation of the Y-direction brace is;
 P1   u1 
P  u 
 2
[
  = K yBr ]
 2
  (2-4-24)
 
 Pn  u n 
where,
[K ] = [T ] [k ][T ]
yBr yBr
T
Br yBr (2-4-25)
49
In case of K-brace (or Cheveron brace)
3 6 4
1 5 2
3 6 6 4
l'
2
 w
l'=   + h2
2
1 5 5 2
For the left half part, as we defined before for the ordinary brace, the stiffness equation of brace element is,
{ f }L = [k ]L {u}L [k ]L = [Cb ]T [nb ]T [k~ ]L [nb ][Cb ] (2-4-26)
where
{ f }L = { f x1 f z1 f x5 f z5 f x3 f z3 f x6 f z6 }
T
{u}L = {u x1 u z1 u x 5 uz5 u x3 uz3 u x6 uz6 }

T
[k~]L
k1, L
=
0 
k 2, L 
, [nb ] = 
− 1 0 0 0 0 0 1 0

 0  0 0 − 1 0 1 0 0 0
c s 
− s c 
 
 −c s 
 
[Cb ] = 
 −s −c , c = cosθ =
(w / 2) , s = sin θ =
h
−c s  l' l'
 
 −s −c 
 c s
 
 − s c 
50
For the right half part, in the same way, the stiffness equation of brace element is,
{ f }R = [k ]R {u}R [k ]R = [Cb ]T [nb ]T [k~ ]R [nb ][Cb ] (2-4-27)
where
{ f }R = { f x5 f z5 f x2 f z2 f x6 f z6 f x4 f z4}
T
{u}R = {u x5 u z5 ux2 uz2 u x6 uz6 ux4 uz4}

T
We can express the nodal displacement vector as,
u x1  u x1 
u  u 
 z1   z1 
u x1  u x 2  u x 2 
u     
 z1  u z 2  u z 2 
u x 5  [1] 
 u x3  u x 3 
    u   
u z 5  [1] u
{u}L = =   z 3  = [DL ] z 3 
u x 3   [1]  u x 4  u x 4 
u z 3    
[1] u z 4 u z 4 
     
u x 6  u x 5  u x 5 
u  u  u 
 z6   z5   z5 
u x 6  u x 6 
u  u 
 z6   z6 
u x1  u x1 
u  u 
 z1   z1 
u x 5  u x 2  u x 2 
u     
 z5  u z 2  u z 2 
u x 2   [1] 
 u x3  u x 3 
    u   
u z 2  [1] u
{u}R = =   z 3  = [DR ] z 3 
u x 6   [1] u x 4  u x 4 
u z 6   [1]
 
 uz4 u z 4 
     
u x 4  u x 5  u x 5 
u  u  u 
 z4   z5   z5 
u x 6  u x 6 
u  u 
 z6   z6 
51
We assume the displacements of intermediate nodes, 5 and 6, are calculated from those of end nodes as
follows,
u z 2, 4
u z 5, 6
1 1
u x 5 = (u x1 + u x 2 ), u z 5 = (u z1 + u z 2 ) u z1,3 u x 2, 4
2 2 2,4
1 1 u x 5, 6
u x 6 = (u x 3 + u x 4 ), u z 6 = (u z 3 + u z 4 ) u x1,3
5,6
2 2
1,3
In a matrix form
u x1   u x1 
u  u 
 z1   z1 
u x 5  1 / 2 0 1 / 2 0 0 0 0 0  u x 2  u x 2 
u   0 1/ 2 0 1/ 2 0     
 z5  0 0 0 u z 2  u 
  =     = [hCh ] z 2  (2-4-28)
u x 6   0 0 0 0 1 / 2 0 1 / 2 0  u x 3  u x 3 
  u z 3 
u z 6   0 0 0 0 0 1 / 2 0 1 / 2  u z 3 
Local
   
u x 4  u x 4 
u  u 
 z 4  Local  z 4  Local
Therefore,
u x1 
u 
 z1 
u x 2  u x1  u x1 
    u 
u z 2   u z1   z1 
u x 3  u x 2  u x 2 
     
u z 3   [I ]  u z 2  u 
 =    = [TCh ] z 2  (2-4-29)
u x 4  [hCh ] u x 3  u x 3 
u z 4  u z 3  u z 3 
     
u x 5  u x 4  u x 4 
u  u  u 
 z5   z 4  Local  z 4  Local
u x 6 
u 
 z6 
52
Therefore,
u x1  u x1 
u  u 
 z1   z1 
u x1  u x 2   u x1  u x 5  u x 2  u x1 
u    u  u    u 
 z1  u z 2   z1   z5  u z 2   z1 
u x 5  u x 3  u x 2  u x 2  u x 3  u x 2 
           
u z 5  u z 3  u z 2  u z 2  u z 3  u z 2 
{u}L =   = [DL ]  = [DL ][TCh ]  , {u}R =   = [DR ]  = [DR ][TCh ] 
u
 x 3 u
 x 4 u
  x 3 u x 6  u x 4  u x 3 
u z 3  u z 4  u z 3  u z 6  u z 4  u z 3 
           
u x 6  u x 5  u x 4  u x 4  u x 5  u x 4 
u  u  u  u  u  u 
 z6   z5   z 4  Local  z4   z5   z 4  Local
u x 6  u x 6 
u  u 
 z6   z6 
(2-4-30)
Finally the force-displacement relationship of Cheveron brace is,
[k ]L = [Cb ]T [nb ]T [k~ ]L [nb ][Cb ] (2-4-31)
 f x1  u x1 
f  u 
 z1   z1 
 f x2  u x 2 
   
 f z2 
  ( T T T T
)
u 
= [TCh ] [DL ] [k ]L [DL ][TCh ] + [TCh ] [DR ] [k ]R [DR ][TCh ]  z 2  (2-4-32)
 f x3  u x 3 
 f z3  u z 3 
   
 f x4  u x 4 
f  u 
 z 4  Local  z4 
53
2.5 External Spring
δ zB
δ yA Z
B
A B A
δ xA δ xB X
B
Y
δ yB A
δ zA
Figure 2-5-1 Element model for external spring
Force-displacement relationship for the element

The relationship between the displacement vector and force vector of the elastic element in Figure 2-5-1 is
expressed as follows:
{N 'i } = [k E ]{δ 'i }, i = x, y , z (2-5-1)
δ ' x = u xB − u xA
δ ' y = u yB − u yA (2-5-2)
δ ' z = δ zB − δ zA
Therefore
u  u 
 xB   xB 
u  u 
{δ ' x } = [− 1 1 0 0 0 0] yA  = [n xE ] yA  (2-5-3)
u yB  u yB 
δ zA  δ zA 
   
δ zB  δ zB 
u  u 
 xB   xB 
u yA 
{δ ' y } = [0 0 − 1 1 0 0]uu yA  = n yE [ ]   (2-5-4)
 yB  u yB 
   
δ zB  δ zB 
54
u  u 
 xB   xB 
u yA  u 
{δ ' z } = [0 0 0 0 − 1 1]  = [n zE ] yA  (2-5-5)
u yB  u yB 
   
δ zB  δ zB 
u xA 
u 
 xB   u1 
u yA  u 
 2
  = [TiE ]  (2-5-6)
u yB  
δ zA  u n 
 
δ zB 
The component of the transformation matrix, [TiE ] , is discussed in Chapter 4 (Freedom Vector).

 u1   u1 
u  u 
{δ 'i } = [niE ][TiE ]  = [TE ] 2 , i = x, y, z
 2
(2-5-7)
 
u n  u n 
The constitutive equation of the external spring is;
 P1   u1 
P  u 
 2  2
  = [K E ]  (2-5-3)
 
 Pn  u n 
where,
[K E ] = [TE ]T [k E ][TE ] (2-5-4)
55
2.6 Base Isolation
δ zB
B Z
u xB
u yB
l X
A u xB
Y
u yA
δ zA
Figure 2-6-1 Element model for base isolation

The relationship between the displacement vector and force vector of the element is expressed as follows:
Q' x 
  = k pBI [ ]δδ '' 
x
(2-6-1)
Q' y   y 
Including the axial stiffness,
Q ' x 
 
[
 k pBI ] 0  δ ' x 
 
δ ' x 
 
Q ' y  =  EA  δ ' y  = [k BI ]δ ' y  (2-6-2)
δ '   0 
 z  l'  
δ ' z 
δ ' 
 z
From node displacements, relative displacements are;
δ ' x = u xB − u xA
δ ' y = u yB − u yA (2-6-3)
δ ' z = δ zB − δ zA
Therefore
u  u 
δ ' x  − 1 1  
xB  xB 
    u yA  = [n ]u yA 
 
δ ' y  =  −1 1  u  BI   (2-6-4)
δ '    yB   u yB 
 z  − 1 1
   
δ zB  δ zB 
56
u xA 
u 
 xB   u1 
u yA  u 
 2
  = [TiBI ]  (2-6-5)
u yB  
δ zA  u n 
 
δ zB 
The component of the transformation matrix, [TiBI ] , is discussed in Chapter 4 (Freedom Vector).

 u1   u1 
δ ' x  u  u 
   2  2
δ ' y  = [n BI ][TiBI ]  = [TBI ]  (2-6-6)
δ '   
 z u n  u n 
The constitutive equation of the Base isolation is;
 P1   u1 
P  u 
 2  2
  = [K BI ]  (2-6-7)
 
 Pn  u n 
where,
[K BI ] = [TBI ]T [k BI ][TBI ] (2-6-8)
57
2.7 Masonry Wall
Element model for Masonry wall is defined as a line element with a nonlinear shear spring and a vertical
spring in the middle of the wall panel as shown in Figure 2-6-1.
B1 B B2
N ' z1 , ε ' z1 N 'z 2 , ε 'z 2

l
Q' xc , γ ' xc
A1 A2
A
Figure 2-7-1 Element model for masonry wall
The relationship between the shear deformation and shear force of the nonlinear shear spring is,
Q' xc = k sx γ ' xc (2-7-1)
For axial spring,

N ' z1 = k z ε ' z1 , N ' z 2 = k zε ' z 2 (2-7-2)
In a matrix form,
 Q' xc  k sx 0 0  γ ' xc  γ ' xc 

       
 N ' z1  =  0 kz 0  ε ' z1  = [k N ]ε ' z1  (2-7-3)
N '   0
 z2   0 k z  ε ' z 2  ε ' 
 z2 

The shear angle of the frame with four nodes, A1, A2, B1, B2, is defined as,
∂δ z ∂u x
τ= + (2-7-4)
∂x ∂z
where,
∂δ z 1  δ zA 2 − δ zA1 δ zB 2 − δ zB1 
≈  +  (2-7-5)
∂x 2 w w 
58
∂u z 1  u xB1 − u xA1 u xB 2 − u xA2 
≈  +  (2-7-6)
∂z 2 l l 
The shear deformation, γ ' xc , is then,
γ ' xc = τ l =
l
(δ zA2 − δ zA1 + δ zB 2 − δ zB1 ) + 1 (u xB1 − u xA1 + u xB 2 − u xA2 ) (2-7-7)
2w 2
The axial deformation, ε ' z1 , ε ' z 2 , is,
ε 'z1 = δ zB1 − δ zA1 , ε 'z 2 = δ zB 2 − δ zA 2 (2-7-8)
In a matrix form,
 u xA1   u xA1 
δ  δ 
 zA1   zA1 
 l l l l  u xA2  u xA2 
γ ' xc  − 0.5 − 0.5 w − 0.5 0.5
w
0.5 − 0.5
w
0.5 0.5  
w δ    
   δ 
 ε ' z1  =  0 −1 0 0 0 1 0 0   zA 2  = [DN ] zA 2 
 u
ε '   0 0 0 −1 0 0 0 1   xB1  u xB1 
 z2    δ zB1 
  δ zB1   
u xB 2  u xB 2 
δ  δ 
 zB 2   zB 2 
(2-7-9)
 u xA1 
δ 
 zA1 
u xA2   u1 
  u 
δ zA 2   2
  = [T ]
ixN   (2-7-10)
u xB1  
δ zB1  u n 
 
u xB 2 
δ 
 zB 2 
The component of the transformation matrix, [TixN ] , is discussed in Chapter 4 (Freedom Vector).

59
 u1   u1 
γ ' xc  u  u 
   2  2
ε '
 z1  = [D N ][T ]
ixN   = [T ]
xN   (2-7-11)
ε '   
 z2  u n  u n 
z Z
y X
x Y
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-7-12)
z  
 Y − Beam  0 0 1  Z  Global
Therefore
 u xA1  1   u yA1   u yA1 

δ   1  δ  δ 
 zA1     zA1   zA1 
u xA2   1  u yA2  u yA2 
      
δ zA 2   1  δ zA2  δ zA 2 
= =  (2-7-13)

u
  1  u yB1  u yB1 
 xB1      
δ zB1   1  δ zB1  δ zB1 
     
u xB 2 
 1  u yB 2  u yB 2 
 
δ 
 zB 2 Y −Wall  1 δ zB 2  Global δ zB 2  Global
60
 u yA1 
δ 
 zA1 
u yA2   u1 
  u 
δ zA 2 

u
[ ]
 = TiyN
 2
  (2-7-14)
 yB1  
δ zB1  u n 
 
u yB 2 
δ 
 zB 2 
 u1   u1 
γ ' xc  u  u 
 
ε ' z1  = [D N ] TiyN[ ]  2
[ ]
  = T yN
 2
  (2-7-15)
ε '   
 z2  u n  u n 
 P1   u1 
P  u 
 2  2
  = [K xN ]  (2-7-16)
 
 Pn  u n 
where,
[K xN ] = [TxN ]T [k N ][TxN ] (2-7-17)
For Y-wall,
 P1   u1 
P  u 
 2
[ ]
  = K yN
 2
  (2-7-18)
 
 Pn  u n 
where,
[K ] = [T ] [k ][T ]
yN yN
T
N yN (2-7-19)
61
2.8 Passive Damper
Element model for passive damper with a shear spring is defined as a line element with a nonlinear shear
spring as shown in Figure 2-8-1.
B1 B B2
l
Q' xc , γ ' xc
A1 A2
A
Figure 2-8-1 Element model for passive damper
The relationship between the shear deformation and shear force of the nonlinear shear spring is,
Q' xc = k sx γ ' xc (2-8-1)

The shear angle of the frame with four nodes, A1, A2, B1, B2, is defined as,
∂δ z ∂u x
τ= + (2-8-2)
∂x ∂z
where,
∂δ z 1  δ zA 2 − δ zA1 δ zB 2 − δ zB1 
≈  +  (2-8-3)
∂x 2 w w 
∂u z 1  u xB1 − u xA1 u xB 2 − u xA2 

≈  +  (2-8-4)
∂z 2 l l 
The shear deformation, γ ' xc , is then,
γ ' xc = τ l =
l
(δ zA2 − δ zA1 + δ zB 2 − δ zB1 ) + 1 (u xB1 − u xA1 + u xB 2 − u xA2 ) (2-8-5)
2w 2
62
The axial deformation, ε 'z1 , ε 'z 2 , is,
ε 'z1 = δ zB1 − δ zA1 , ε 'z 2 = δ zB 2 − δ zA 2 (2-8-6)
In a matrix form,
 u xA1   u xA1 
δ  δ 
 zA1   zA1 
 l l l l  u  u xA2 
γ ' xc  − 0.5 − 0.5 w − 0.5 0.5
w
0.5 − 0.5
w
0.5 0.5   xA2 
w δ zA 2   
   δ zA 2 
 ε ' z1  =  0 −1 0 0 0 1 0 0  = [DD ] 
 u xB1  u xB1 
ε '   0 0 0 −1 0 0 0 1    
 z2    δ zB1 
  δ zB1   
u xB 2  u xB 2 
δ  δ 
 zB 2   zB 2 
(2-8-7)
 u xA1 
δ 
 zA1 
u xA2   u1 
  u 
δ zA 2   2
  = [TixD ]  (2-8-8)
 u xB1  
δ zB1  un 
 
u xB 2 
δ 
 zB 2 
The component of the transformation matrix, [TixD ] , is discussed in Chapter 4 (Freedom Vector).

 u1   u1 
γ ' xc  u  u 
   2  2
ε '
 z1  = [DD ][T ]
ixD   = [T ]
xD   (2-8-9)
ε '   
 z2  un  un 
63
In case of Y-direction damper
z Z
y X
x Y
In case of Y-direction damper, the damper direction coincides to the Y-axis in the global coordinate,
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-8-10)
z  
 Y − Beam  0 0 1  Z  Global
Therefore
 u xA1  1   u yA1   u yA1 

δ   1  δ  δ 
 zA1     zA1   zA1 
u xA2   1  u yA2  u yA2 
      
δ zA 2   1  δ zA2  δ zA 2 
= =  (2-8-11)

u
  1  u yB1  u yB1 
 xB1      
δ zB1   1  δ zB1  δ zB1 
     
u xB 2 
 1  u yB 2  u yB 2 
 
δ 
 zB 2 Y −Wall  1 δ zB 2  Global δ zB 2  Global
 u yA1 
δ 
 zA1 
u yA2   u1 
  u 
δ zA 2 

u
[ ]
 = TiyD
 2
  (2-8-12)
 yB1  
δ zB1  un 
 
u yB 2 
δ 
 zB 2 
64
 u1   u1 
γ ' xc  u  u 
 
ε ' z1  = [DD ] TiyD[ ]  2
[ ]
  = TyD
 2
  (2-8-13)
ε '   
 z2  un  un 
Finally, the constitutive equation of the damper is;
 P1   u1 
P  u 
 2  2
  = [K ]
xD   (2-8-14)
 
 Pn  un 
where,
[K xD ] = [TxD ]T [kD ][TxD ] (2-8-15)
For Y-damper,
 P1   u1 
P  u 
 2
  = K yD[ ]  2
  (2-8-16)
 
 Pn  un 
where,
[K ] = [T ] [k ][T ]
yD yD
T
D yD (2-8-17)
65
Appendix : Calculation of shear component
For “Masonry Wall” and “Passive Damper”, the shear deformation is defined as follows:
1) Shear deformation in one direction
Shear strain is τ = Δl / l ≈θ
2) Shear deformation in two directions
Shear strain is τ = θ1 + θ2 = Δlx / lｙ+Δly / lx
∂u x ∂u y
If we discuss small element τ= +  Eq. (2-7-4) and Eq. (2-8-2)
∂y ∂x
66
This definition is necessary to remove rotational component. To explain this, suppose there is only
rotational (or bending) deformation,
-θ
From the above definition, shear angle will be

τ= θ + (- θ) = 0
For example, in the upper story of the building under horizontal deformation, the bending
component is dominant and the shear component is small. Therefore, the brace damper
doesn’t work in the upper story.
67
3) In case of damper element
We define the shear angle in one direction as follows:
θ1 θ2
We adopt the average angle,
θ = 1/2 ( θ1 + θ2 )  Eq. (2-7-5) and Eq. (2-8-3)
In the same way, the shear angle in another direction is
θ'2
θ'1
θ’ = 1/2 ( θ’1 + θ’2 )  Eq. (2-7-6) and Eq. (2-8-4)
68
2.9 Floor Element
In the default setting, STERA 3D adopts “rigid floor”. However, elastic deformation of a floor diaphragm
in-plane can be considered by the option menu selecting “flexible floor”. The stiffness matrix of the floor
element is constructed using a two dimensional isoparametric element.
Figure 2-9-1 4-nodes isoparametric element
The stiffness matrix with 4-nodes isoparametric is expressed as,
 P1  u1 
Q  v 
 1  1
 P2  u 2 
   
Q2  v 2 
  = [K F ] 
 P3  u 3 
Q3   v3 
   
 P4  u 4 
Q  v 
 4  4
F =K u (2-9-1)
The coordinate transfer function {x, y} is expressed using the interpolation functions as follows:
4
1 1 1 1
x(r , s ) = ∑ hi (r , s ) xi = (1 + r )(1 + s ) x1 + (1 − r )(1 + s) x 2 + (1 − r )(1 − s ) x3 + (1 + r )(1 − s ) x 4
i =1 4 4 4 4
4
1 1 1 1
y (r , s ) = ∑ hi (r , s ) y i = (1 + r )(1 + s ) y1 + (1 − r )(1 + s ) y 2 + (1 − r )(1 − s ) y 3 + (1 + r )(1 − s ) y 4
i =1 4 4 4 4
(2-9-2)
69
The deformation function {u, v} is also expressed using the same interpolation functions.
4
1 1 1 1
u (r , s ) = ∑ hi (r , s )u i = (1 + r )(1 + s )u1 + (1 − r )(1 + s )u 2 + (1 − r )(1 − s )u 3 + (1 + r )(1 − s )u 4
i =1 4 4 4 4
4
1 1 1 1
v(r , s ) = ∑ hi (r , s )v i = (1 + r )(1 + s )v1 + (1 − r )(1 + s )v 2 + (1 − r )(1 − s )v 3 + (1 + r )(1 − s )v 4
i =1 4 4 4 4
(2-9-3)
Stiffness matrix can be obtained from the “Principle of Virtual Work Method,” which is expressed in the
following form:
∫ε σ dv = u T F
T
(2-9-4)
V
where, ε is a virtual strain vector, σ is a stress vector, u is a virtual displacement vector and F is a
load vector, respectively.
In case of the plane problem, the strain ε vector is defined as,

 ∂u 
 
 ε x   ∂x 
   ∂v 
εy  =  (2-9-5)
 γ   ∂y 
 xy   ∂u ∂v 
 ∂y + ∂x 
 
Substituting equation (2-9-3) into equation (2-9-5), the strain vector is calculated from the nodal
displacement vector as,
 ∂u   4
∂hi 
   ∑
i =1 ∂x
ui 

 ε x   ∂x  
   ∂v   4
∂hi 
εy  =  = ∑ vi 
 γ   ∂y   i =1 ∂y 
 xy   ∂u ∂v   4 ∂hi 4
∂hi 
 ∂y + ∂x   ∑ ∂y u i + ∑ ∂x vi 
   i =1 i =1 
 u1 
 
v
 ∂h1 ∂h2 ∂h3 ∂h4  1 
 0 0 0 0  u 
 ∂x ∂x ∂x ∂x  2 
 ∂h1 ∂h2 ∂h3 ∂h4  v 2 
= 0 0 0 0  u 
 ∂y ∂y ∂y ∂y  3 
 ∂h1 ∂h1 ∂h2 ∂h2 ∂h3 ∂h3 ∂h4 ∂h4  v3 
 ∂y ∂x ∂y ∂x ∂y ∂x ∂y ∂x  
  u4 

v 
 4
ε= B u (2-9-6)
70
In the plane stress problem, the stress-strain relationship is expressed as,
σ x    
  1 ν 0  ε x 
E ν 1  ε 
σ y  = 
0
 y 
(2-9-7)
τ  1 − ν 0 0
1 −ν
 γ xy 
 xy   2 
σ= C ε
Substituting equation (2-9-6) into equation (2-9-7),
σ= C B u (2-9-8)
From the Principle of Virtual Work Method,
T  
∫ (Bu ) (CBu )dv = u  B T CBdxdy u = u T F
∫
T
(2-9-9)
 
V  V ( x, y ) 
Therefore, the stiffness equation is obtained as,
F = Ku , K = ∫ B T CBdv (2-9-10)
V
If we assume the constant thickness of the plate (= t), using the relation dv = tdxdy ,
∫B
T
K =t CBdxdy (2-9-11)
V ( x, y )
Since this integration is defined in x-y coordinate, we must transfer the coordinate into r-s coordinate to use
the numerical integration method. Introducing the Jacobian matrix,
 ∂x ∂y 
 
J =  ∂r ∂r  ; Jacobian Matrix (2-9-12)
 ∂x ∂y 
 
 ∂s ∂s 
the above integration is expressed in r-s coordinate as,
1 1
∂ ( x, y )
K = t ∫ ∫ B( x(r , s ), y (r , s )) CB( x(r , s ), y (r , s ))
T
drds (2-9-13)
−1 −1
∂ (r , s)
where
∂x ∂y
∂ ( x, y )
= det J = ∂r ∂r (2-9-14)
∂ (r , s) ∂x ∂y
∂s ∂s
71
Evaluation of Jacobian Matrix
∂y   ∂hi x ∂hi 
4 4
 ∂x
  ∑
∂r  =  i =1 ∂r
i ∑ ∂r yi 
J =  ∂r i =1  (2-9-15)
 ∂x ∂y   4 ∂hi ∂hi 
4
∂s   ∑ ∑
  xi yi 
 ∂s i =1 ∂s i =1 ∂s 
Evaluation of the matrix B
 ∂h1 ∂h2 ∂h3 ∂h4 

 0 0 0 0 
 ∂x ∂x ∂x ∂x 
 ∂h1 ∂h2 ∂h3 ∂h4 
B= 0 0 0 0  (2-9-16)
 ∂y ∂y ∂y ∂y 
 ∂h1 ∂h1 ∂h2 ∂h2 ∂h3 ∂h3 ∂h4 ∂h4 
 ∂y
 ∂x ∂y ∂x ∂y ∂x ∂y ∂x 
∂h1 ∂h ∂h ∂h
The derivatives ,  , 4 , 1 , , 4 are calculated as,
∂x ∂x ∂y ∂y
∂h1 ∂h1 ∂r ∂h1 ∂s ∂h ∂h ∂r ∂h4 ∂s
= + ,  , 4 = 4 + ,
∂x ∂r ∂x ∂s ∂x ∂x ∂r ∂x ∂s ∂x
∂h1 ∂h1 ∂r ∂h1 ∂s ∂ h ∂ h ∂r ∂h4 ∂s
= + ,  , 4 = 4 +
∂y ∂r ∂y ∂s ∂y ∂y ∂r ∂y ∂s ∂y
In a matrix form,
 ∂h1 ∂h2 ∂h3 ∂h4   ∂r ∂s  ∂h1 ∂h2 ∂h3 ∂h4 

    
 ∂x ∂x ∂x ∂x  =  ∂x ∂x  ∂r ∂r ∂r ∂r 
 ∂h1 ∂h2 ∂h3 ∂h4   ∂r ∂s  ∂h1 ∂h2 ∂h3 ∂h4 
 ∂y
 ∂y ∂y ∂y   ∂y
  ∂y  ∂s ∂s ∂s ∂s


 ∂h1 ∂h2 ∂h3 ∂h4 

 
= J −1  ∂r ∂r ∂r ∂r  (2-9-17)
 ∂h1 ∂h2 ∂h3 ∂h4 
 
 ∂s ∂s ∂s ∂s 
Evaluation of partial derivatives of the interpolation functions
∂h1 1 ∂h1 1
= (1 + s ) = (1 + r )
∂r 4 ∂s 4
∂h2 1 ∂h2 1
= − (1 + s ) = (1 − r )
∂r 4 ∂s 4
, (2-9-18)
∂h3 1 ∂h3 1
= − (1 − s ) = − (1 − r )
∂r 4 ∂s 4
∂h4 1 ∂h4 1
= (1 − s ) = − (1 + s )
∂r 4 ∂s 4
72
The 3 points Gauss Integration Formula is defined as:
∫ f (t )dt = 0.5556 f (−0.7746) + 0.8889 f (0) + 0.5556 f (0.7746)

−1
(2-9-19)
= α 1 f (t1 ) + α 2 f (t 2 ) + α 3 f (t 3 )
where, α 1 = 0.5556, α 2 = 0.8889, α 3 = 0.5556

t1 = −0.7746, t 2 = 0, t 3 = 0.7746
f(0.7746)
f(t)
f(-0.7746)
f(0)
t
-1 -0.7746 0 +0.7746 +1
The stiffness matrix is then calculated numerically as follows:
1 1
∂ ( x, y )
K = t ∫ ∫ B( x(r , s ), y (r , s )) CB( x(r , s ), y (r , s ))
T
drds
−1 −1
∂ (r , s)
1 1
= t ∫ ∫ F (r , s )drds (2-9-20)
−1 −1
3 3
= t ∑∑ α iα j F (ri , s j )
i =1 j =1
where
∂ ( x, y )
F (r , s ) = B(x(r , s ), y (r , s )) CB( x(r , s ), y (r , s ))
T
∂ (r , s)
α 1 = 0.5556, α 2 = 0.8889, α 3 = 0.5556

r1 = s1 = −0.7746, r2 = s 2 = 0, r3 = s3 = 0.7746
73
Transformation from global node displacements to element node displacements is,
u1 
v 
 1
u 2   u1 
  u 
v 2   2
  = [T ]
iF   (2-9-21)
u 3  
 v3  u n 
 
u 4 
v 
 4
The component of the transformation matrix, [TiF ] , is discussed in Chapter 4 (Freedom Vector).
74
2.10 Connection Panel
1) General case
In the default setting, STERA3D assumes the rigid connection zone between column and beam. You can
consider shear deformation of the connection area (we call “connection panel”) by the “Connection
member” menu.
Figure 2-10-1 Connection area
Shear deformation of the connection panel, γ, is defined as shown in Figure 2-10-2.
uC= -0.5γAh θC= -0.5γA
C C
h
0.5γA 0.5γA
A B θB= 0.5γA A 0.5γA vB= -0.5γAw

h B
w w
Figure 2-10-2 Definition of shear deformation
Differences of displacement at node B and C are;
 ∆u B   0   ∆u C  − 0.5γ A h 
       
Node B:  ∆v B  ≈ − 0.5γ A w , Node C:  ∆vC  ≈  0  (2-10-1)
∆θ   0.5γ  ∆θ   − 0.5γ 
 B  A   C  A 
75
First we consider nodal movement without shear deformation of the connection panel. As shown in Figure
2-10-3, the displacement at node B and node C will be;
u B   u A  u C  u A − θ A h 
       
Node B: v B  ≈ v A + θ A w , Node C: vC  ≈  v A  (2-10-2)
θ   θ  θ   θ 
 B  A   C  A 
Then, we consider shear deformation of the connection as shown in Figure 2-10-4. By adding Equation
(2-10-1) to (2-10-2), the displacement at node B and node C will be;
− θ Ah
h
θ Aw
θA
C w
vA
uA
A B
Figure 2-10-2 Nodal movement without shear deformation of the panel
A B
Figure 2-10-4 Nodal movement with shear deformation of the panel
76
Node B:
u 
u B   u A   0   uA  1 0 0 0  A 
          v A 
v B  ≈ v A + θ A w + − 0.5γ A w = v A + θ A w − 0.5γ A w = 0 1 w − 0.5w  
θ
θ   θ
 B 
  0.5γ   θ + 0.5γ
  A  
 0 0 1
  0.5   A 
A A A
γ A 
(2-10-3)
Node C:
u A 
u C  u A − θ A h  − 0.5γ A h  u A − θ A h − 0.5γ A h  1 0 h − 0.5h   
         v 
vC  ≈  v A  +  0 = vA  = 0 1 0 0   A 
θ   θ   − 0.5γ   θ − 0.5γ  0 0 1 − 0.5  θ A 
 C  A   A   A A    γ 
 A
(2-10-4)
2) Beam element
In case of rigid connection, as described in Equation (2-1-7), the nodal displacement is expressed as,
θ ' yA  θ yA − τ  (u zB − λ B l 'θ yB ) − (u zA + λ Al 'θ yA )

 = , τ =
θ ' yB  θ yB − τ  l'
u 
 1 1  1 1   zA 
θ +
 yA l ' zAu + λ θ − u + λ θ
B yB   − 1 + λ λ B  u
A yA zB A
 zB 
= l'
 = l' l'
   (2-10-5)
1 1 1 1 θ
θ yB + u zA + λ Aθ yA − u zB + λ Bθ yB   − λA 1 + λ B   yA 
 l' l'  l' l'  θ yB 
 
θ yB
θ ' yA θ yA τ
θ yB
θ yA τ θ ' yB
u zB
Z
u zB − λ B l 'θ yB
u zA u zA + λ A l 'θ yA
A B X
λ Al ' l' λB l '
Figure 2-10-5 Beam displacement with rigid connection
77
If we consider shear deformation of connection panel, from Figure 2-10-6,
θ ' yA  θ yA + 0.5γ yA − τ  (u zB − λB l ' (θ yB − 0.5γ yB )) − (u zA + λ Al ' (θ yA − 0.5γ yA ))

 = , τ =
θ ' yB  θ yB + 0.5γ yB − τ  l'
 1 1 
θ yA + l ' u zA + λ Aθ yA − l ' u zB + λ Bθ yB + 0.5γ yA − 0.5λ Aγ yA − 0.5λ B γ yB 
= 
1 1
θ yB + u zA + λ Aθ yA − u zB + λ Bθ yB + 0.5γ yB − 0.5λ Aγ yA − 0.5λ B γ yB 
 l' l' 
u zA 
u 
1 1   zB 
l' − 1 + λA λB 0.5 − 0.5λ A − 0.5λ B  θ 
l '  yA 
=  θ  (2-10-6)
1 1
 − λ A 1 + λB − 0.5λ A 0.5 − 0.5λ B   yB 
l' l'  γ yA 
 
γ yB 
θ yB
τ
θ ' yA θ yA + 0.5γ yA
θ yB + 0.5γ yB θ ' yB
τ
θ yA u zB
u zB − λ B l ' (θ yB − 0.5γ yB )
u zA u zA + λ Al ' (θ yA − 0.5γ yA )
A B
Z
λ Al ' l' λB l '

X
Figure 2-10-6 Beam displacement with shear deformation of connection panel
78
The transformation matrices for beam element are;
Including connection panel and node movement
u zA  u zA 
u  u 
1 1   zB   zB 
θ ' yA   l ' − l ' 1 + λ A λB 0.5 − 0.5λ A − 0.5λ B  θ yA
  θ yA 
θ '   1 
 θ   
1 θ
 yB  
  = l' − l' λA 1 + λB − 0.5λ A 0.5 − 0.5λ B    = [Λ B ] yB 
 yB 
δ   γ yA γ yA 
  1   
xA
 δ xB   γ yB γ yB 

 1    
δ xA  δ xA 
δ  δ 
 xB   xB 
(2-10-10)
u zA 
u 
 zB 
θ yA   u1 
  u 
θ yB   2
  = [TixB ]  (2-10-11)
γ yA  
γ yB  u n 
 
δ xA 
δ 
 xB 

 u1   u1 
θ ' yA  u  u 
   2  2
θ '
 yB  = [n B ][Λ B ][T ]
ixB   = [T ]
xB   (2-10-12)
δ '    
 x un  un 
79
In case of Y-direction beam
x  0 1 0  X 
   
 y = − 1 0 0  Y  (2-10-13)
z  
 Y − Beam  0 0 1  Z  Global
u zA  1  u zA  u zA 
u   1  u  u 
 zB     zB   zB 
θ yA   −1  θ xA  θ xA 
      
θ yB  − 1 θ  θ xB 
=  = [s B ] 
xB
    (2-10-14)
  γ xA
γ yA  
−1
  γ xA 
γ yB   −1  γ xB  γ xB 
     
δ xA 
 1  δ yA  δ yA 
 
δ  1 δ yB  Global δ yB 
 xB Y − Beam    Global
u zA 
u 
 zB 
θ xA   u1 
  u 
θ xB 
  = TiyB [ ]  2
  (2-10-15)
γ xA  
γ xB  u n 
 
δ yA 
δ yB 
 
 u1   u1 
θ ' yA  u  u 
 
[ ]
θ ' yB  = [n B ][Λ B ][s B ] TiyB
 2
[ ]
  = T yB
 2
  (2-10-16)
δ '   
 x u n  u n 
80
3) Column element
In case of rigid connection, as described in Equation (2-2-16), the nodal displacement in X-Z plane is
expressed as,
θ ' yA  θ yA − τ  (u xA − λ Al 'θ yA ) − (u xB + λB l 'θ yB )

 = , τ =
θ ' yB  θ yB − τ  l'
u 
 1 1   1 1  xA
θ − u + λ θ + u + λ θ  − l ' 1 + λA λ B  u 
yA
l '
xA A yA
l '
xB B yB
l'  xB 
= = 1    (2-10-17)
1 1 1
θ yB − u xA + λ Aθ yA + u xB + λ Bθ yB  − λ A 1 + λ B  θ yA 
 l' l'   l' l'  θ yB 
 
B u xB
λB l ' θ yB
u xB + λ B l 'θ yB θ ' yB
θ yB
τ
l' Z
τ
θ yA
u xA − λ A l 'θ yA θ ' yA X
θ yA
λ Al '
Y
A u xA
Figure 2-9-7 Column displacement with rigid connection (X-Z plane)
81
θ ' yA  θ yA − 0.5γ yA − τ  (u xA − λ Al ' (θ yA + 0.5γ yA )) − (u xB + λB l ' (θ yB + 0.5γ yB ))

 = , τ =
θ ' yB  θ yB − 0.5γ yB − τ  l'
 1 1 
θ yA − l ' u xA + λ Aθ yA + l ' u xB + λ Bθ yB − 0.5γ yA + 0.5λ Aγ yA + 0.5λ B γ yB 
= 
1 1
θ yB − u xA + λ Aθ yA + u xB + λ Bθ yB − 0.5γ yB + 0.5λ Aγ yA + 0.5λ B γ yB 
 l' l' 
u xA 
u 
 1 1   xB 
− l ' 1 + λA λB − 0.5 + 0.5λ A 0.5λ B  θ yA 
= l' (2-10-18)
1 1  θ 
− λ A 1 + λB 0.5λ A − 0.5 + 0.5λ B   yB 
 l' l'  γ yA 
 
γ yB 
B u xB
λB l ' θ yB
u xB + λ B l ' (θ yB + 0.5γ yB )
θ ' yB
Z θ yB − 0.5γ yB
τ
l'
X
τ
θ yA − 0.5γ yA
θ ' yA
u xA − λ Al ' (θ yA + 0.5γ yA )
θ yA
λ Al '
u xA
A
Figure 2-9-8 Column displacement with shear deformation of connection panel (X-Z plane)
82
In the same manner, assuming rigid connection, the nodal displacement of column in Y-Z plane is
expressed as,
θ ' xA  θ xA − τ  (u yB − λ Al 'θ xB ) − (u yA + λ Al 'θ xA )

 = , τ =
θ ' xB  θ xB − τ  l'
u
 1 1  1 1   yA 
θ xA + l ' u yA + λ Aθ xA − l ' u yB + λ Bθ xB   l ' − l ' 1 + λ A λB  u
 yB 
= =
 1   (2-10-19)
1 1 1
θ xB + u yA + λ Aθ xA − u yB + λ Bθ xB   − λ A 1 + λ B  θ xA 
 l' l'  l' l'  θ xB 
λB l ' u yB
u yB − λ B l 'θ xB
θ xB
θ 'xB
l' θ xB
τ Z
τ
θ xA u yA + λ Al 'θ xA
X
θ 'xA
λ Al '
θ xA A Y
u yA
Figure 2-9-9 Column displacement with rigid connection (Y-Z plane)
83
θ ' xA  θ xA − 0.5γ xA −τ  (u yB − λB l ' (θ xB + 0.5γ xB )) − (u yA + λ Al ' (θ xA + 0.5γ xA ))

 = , τ =
θ ' xB  θ xB − 0.5γ xB −τ  l'
 1 1 
θ xA + l ' u yA + λ Aθ xA − u yB + λ Bθ xB − 0.5γ xA + 0.5λ Aγ xA + 0.5λ B γ xB 
l'
= 
1 1
θ xB + u yA + λ Aθ xA − u yB + λ Bθ xB − 0.5γ xB + 0.5λ Aγ xA + 0.5λ B γ xB 
 l' l' 
u yA 
u 
1 1   yB 
l' − 1 + λA λB − 0.5 + 0.5λ A 0.5λ B  θ xA 
= l ' (2-10-20)
1 1  
 − λA 1 + λB 0.5λ A − 0.5 + 0.5λ B  θ xB 
l' l'  γ xA 
 
γ xB 
u yB B
θ xB λB l '
u yB − λ B l ' (θ xB + 0.5γ xB )
θ ' xB
τ θ xB − 0.5γ xB
Z
l'
τ
Y
θ xA − 0.5γ xA
θ ' xA
u yA + λ Al ' (θ xA + 0.5γ xA )
θ xA
λ Al '
u yA
A
Figure 2-9-10 Column displacement with shear deformation of connection panel (Y-Z plane)
84
The transformation matrices for column element are;
Including connection panel and node movement
u xA 
u 
 xB 
θ yA 
 1 1 1 λA λB  θ yB 
− 1 + λA λB − +  
θ ' yA   l ' l' 2 2 2
 γ yA 
θ '  − 1 1 λA 1 λB
λ A 1 + λB − +  γ 
 yB   l ' l' 2 2 2   yB 
θ ' xA   1 1 1 λA λB  u yA 
   − 1 + λA λB − +  u 
θ ' xB   l' l' 2 2 2
 = 1 1 λA 1 λB   yB 
 δ zA   − λ A 1 + λB − +  θ xA 
l' l' 2 2 2  
 δ zB  
  
1  θ xB 
 θ zA   1  γ xA 
θ    
 zB   1  γ xB 

 1 δ zA 
δ 
 zB 
θ zA 
 
θ zB 
u xA 
u 
 xB 
θ yA 
 
θ yB 
γ yA 
 
γ yB 
u 
 yA 
u yB 
= [Λ C ] 
θ xA 
θ xB 
 
γ xA 
γ 
 xB 
δ zA 
δ 
 zB 
θ zA  (2-10-21)
 
θ zB 
85
u xA 
u 
 xB 
θ yA 
 
θ yB 
γ yA 
 
γ yB 
u   u1 
 yA  u 
u yB   2
  = [TiC ] 
θ xA  
θ xB  u n 
 
γ xA 
γ 
 xB 
δ zA 
δ 
 zB 
θ zA 
  (2-10-22)
θ zB 

θ ' yA 
θ ' 
 yB   u1   u1 
θ ' xA    u 
u 2   2
  = [n C ][Λ C ][T ]
iC   = [T ]
C   (2-10-23)
θ ' xB    
 δ 'z  u n  u n 
 
 θ ' z 
86
4) Force-displacement relationship for the connection
γx
γy
Figure 2-9-11 Shear deformation of connection area
The relationship between the displacement vector and force vector of the element is expressed as follows:
M Px  k Px 0  γ x 
 =   (2-10-24)
M Py   0 k Py  γ y 
where, initial stiffness of connection area is,
k Px = k Py = GV (2-10-25)
where, G is the shear modulus and V is the volume of the connection.

 u1 
γ x  u 
 2
  = [T ]
P   (2-10-26)
γ
 y 
u n 
The component of the transformation matrix, [TP ] , is discussed in Chapter 4 (Freedom Vector).
The constitutive equation of the external spring is;
 P1   u1 
P  u 
 2  2
  = [K P ]  (2-10-27)
 
 Pn  u n 
where,
[K P ] = [TP ]T [k P ][TP ] (2-10-28)
87
2.11 Ground Spring
Foundation
K Hx M yG , θ yG
PxG , u xG
G
CSx K Ry CRy
Foundation
K Hy
CSy
M xG , θ xG
G
CRx Z
PyG , u yG
K Rx
X
Figure 2-11-1 Element model for ground spring

The relationship between the displacement vector and force vector of the ground springs attached at the
center of gravity of the foundation in Figure 2-11-1 is expressed as follows:
Sway and rocking in X-direction
 PxG   K Hx 0  u xG  CHx 0  u xG 

=    +   (2-11-1)
 M yG   0 K Ry  θ yG   0 CRy  θyG 
Sway and rocking in Y-direction
 PyG   K Hy 0  u yG  CHy 0  u yG 

=    +   (2-11-2)
 M xG   0 K Rx  θ xG   0 CRx  θxG 
88
Therefore
 PxG   K Hx 0  u xG  CHx 0  u xG 

P   K Hy  u   CHy  u 
 yG     yG  +    yG 
 
 M yG   K Ry  θ yG   CRy  θyG 
   
 M xG   0 K Rx  θ xG   0 CRx  θxG 
(2-11-3
u xG  u xG 
u  u 
 yG   
= [ kG ]   + [ cG ]  yG 
θ yG  θ yG 
θ xG  θxG 
u xG   u1 
u  u 
 yG   2
  = [TG ]   (2-11-4)
θ yG  
θ xG  un 
The component of the transformation matrix, [TG ] , is discussed in Chapter 4 (Freedom Vector).
The constitutive equation of the ground spring is;
 P1   u1   u1 
P  u  u 
 2
=
  [ G ]    [ G ]  2 
K
 2
+ C (2-11-5)
    
 Pn  un  un 
where,
[ KG ] [T=
G ] [ kG ][TG ] , [ CG ] [TG ] [cG ][TG ]
T T
(2-11-6)
89
3. Nonlinear Element Models
Notation
at : Area of rebar in the tension side of the section
As : Total area of rebar in the section
σy : Strength of rebar
σB : Compression strength of concrete
σ wy : Strength of shear reinforcement
D : Depth of the section
d : Effective depth of the section.
b : Width of the beam
j : Distance between the centers of stress in the section ( = (7 / 8)d ).
Ze : Section modulus including the slab effect.
n : Ratio of Young’s modulus (= E s / Ec )
pt : Tensile reinforcement ratio
pw : Shear reinforcement ratio
Ie : Moment of inertia of section considering the slab effect
Mc : Crack moment
My : Yield moment
M/(QD) : Shear span-to-depth ratio
θc : Crack rotation of the beam end
θy : Yield rotation of the beam end
φc : Crack rotation of the nonlinear bending spring
φy : Yield rotation of the nonlinear bending spring
k0 : Initial stiffness
ky : Tangential stiffness at the yield point
k y2 : Stiffness after the yield point in the nonlinear bending spring
k y3 : Stiffness after the ultimate point in the nonlinear shear spring
αy : Stiffness degradation factor at the yield point
Qc : Crack shear force
Qy : Yield shear force
Qu : Ultimate shear force
xs : Distance between the corner springs in the Multi-spring model
90
γc : Crack shear deformation
γy : Yield shear deformation
γu : Ultimate shear deformation
91
3.1.1 Beam
3.1.1 RC Beam
a) Section properties
S
as
d1
t d1
a1
D
d2 a2 d2
B
B : Width of beam,
D : Height of beam,
S : Effective width of slab,
t : Thickness of slab
d1 : Distance to the center of top main rebars,
d2 : Distance to the center of bottom main rebars,
a1 : Area of top main rebars,
a2 : Area of bottom main rebars
as : Area of rebars in slab
Figure 3-1-1 RC Beam Section
Area of section to calculate axial deformation
AN = BD + (S − B )t + (n E − 1)(a1 + a 2 + a S ) (3-1-1)
where,
nE = E s / Ec : Ratio of Young’s modulus between steel (Es) and concrete (Ec)
Area of section to calculate shear deformation
AS = BD (3-1-2)
Moment of inertia around the center of the section
2 2
BD 3 ( S − B )t 3  D  t 
Ie = + + BD g −  + (S − B )t  D − − g  +
12 12  2  2 
2
 t 
(n E − 1)a1 (d 1 − g ) + (n E − 1)a 2 (D − d 2 − g ) + (n E − 1)a S  D − − g 
2 2
(3-1-3)
 2 
where, g is the center of beam section calculated by
BD 2 / 2 + ( S − B ) t ( D − t / 2 ) + ( nE − 1) {a2 d 2 + a1 ( D − d1 ) + aS ( D − t / 2 )}
g= (3-1-4)
AN
92
b) Nonlinear bending spring
τA
δx
MA MB
A B
τB
Nx
elastic element
θA φA
MA MB MA MB
A B A B
Nx θB φB

l
θ A  τ A  φ A  η A  ηA
        MA MB
θ B  = τ B  + φB  + η B  A ηB B
δ  δ   0   0 
 x  x    
Hysteresis model of a nonlinear bending spring is defined as the moment-rotation relationship under the
anti-symmetry loading in Figure 3-1-3. The initial stiffness of the nonlinear spring is supposed to be infinite,
however, in numerical calculation, a large enough value is used for the stiffness.
φ θ
M τ τ M
A B
θ φ
Moment distribution
M M M
My My
6 EI = +
Mc k0 = Mc
l
k 0 α y k0 k0 kn ≈ ∞
θc θy θ τ φc φ y φ
Elastic element Nonlinear bending spring
Figure 3-1-3 Moment – rotation relationship at bending spring
93
Crack moment force
For reinforced concrete elements, the crack moment, M c is calculated as,
M c1 = 0.56 σ B Z e1 , Z e1 = I e / g when tension in top main rebars (3-1-5)
M c 2 = 0.56 σ B Z e 2 , Z e 2 = I e / (D − g ) when tension in bottom main rebars (3-1-6)
where,
σB : Compression strength of concrete (N/mm2)
Z e1 , Z e 2 : Section modulus
Yield moment force

The yield moment, M y is calculated as,
M y1 = 0.9a1σ y (D − d1 ) + 0.9a S σ y (D − t / 2) when tension in top main rebars (3-1-7)
M y 2 = 0.9a 2σ y (D − d 2 ) when tension in bottom main rebars (3-1-8)
where,
σy : Strength of rebar (N/mm2)
Yield rotation
The tangential stiffness at the yield point, k y , is obtained from the following equation,:
6 Ec I e
k y = α y k0 , k0 = (3-1-9)
l
where,
α y is the stiffness degradation factor at the yield point, which is obtained from the following
empirical formulas:
α y = (0.043 + 1.63npt + 0.043a / D )(d / D )2 , (a / D ≤ 2) (3-1-10)
α y = (− 0.0836 + 0.159a / D )(d / D )2 , (a / D > 2) (3-1-11)

where,

pt = (a1 + aS ) / (BD ) (when tension in top main rebars)
pt = (as ) / (BD ) (when tension in bottom main rebars)
a/D : ≈ Shear span-to-depth ratio (= l /( 2 D ) )
d : Effective depth
d = D − d1 (when tension in top main rebars)
d = D − d2 (when tension in bottom main rebars)
94
α y is modified in case of tension in top main rebars as
I
α y ' = α y e0 (3-1-12)
Ie
BD 3
where I e0 = : the moment of inertia of square section without slab
12
The yield rotation of the nonlinear bending beam, φ y , is then obtained from,
 1 My
φ y =  − 1
 k
(3-1-13)
αy  0
In general, the relation between the rotation of bending spring and that of nonlinear bending spring is
My
φ =θ − (3-1-14)
k0
Crack rotation
From Figure 3-1-2, the crack rotation of the nonlinear bending beam, φ c , is supposed to be zero value,
however, in STERA_3D program, it is assumed as,
φ c = 0.001 φ y (3-1-15)
95
Effective width of slab
Ls
B Lb
Sb
Sb B
Figure 3-1-4 Effective slab area for flexural capacity of beam
In general, effective width of slab for the flexural behavior of a beam is assumed as,
S b = 0.1 Lb ≈ D (3-1-16)
where, Lb : Length of beam

D : Height of beam
However, recent studies suggest the contribution of full length of slab to the flexural capacity, M y , of a
beam. Therefore, STERA3D adopts two types of effective widths:
1) For calculating section are and moment of inertia

S b = 0.1 Lb ≈ D
2) For calculating the yield moment, M y , in Equation (3-1-8),
S b = η s Ls (3-1-17)
where, Ls : Length of span

ηs : Effective slab ratio (0.1 ~ 0.5) , the default value is 0.1.
96
Hysteresis model
To consider the difference of the flexural capacity between positive and negative side of the beam, a
degrading tri-linear slip model is developed based on the Takeda Model for the hysteresis model of the
bending springs of the beam.
M M
kp
My
My
Mc Mc
ks kr
φ φ
φx φy φm
α β
 My  My  φy  Mm  φy
kp =η   kr =   k s =   (3-1-18)
ϕ φ φ
 y   y  m  φm − φ x  φm
Figure 3-1-5 Degrading Tri-linear Slip Model

(α=0.5, β=0.0 and η=0.001 as default values)
The strength degradation under cyclic loading is considered by elongating the target displacement, φ m , to
be φ ' m as shown in the following Figure:
M
kp
Mm
My
ks kr
φ
n φm n φy φ φy φm φ 'm
x
 φy 
φ ' m = 1 + γ
n
φ m (3-1-19)

 n φm 
Figure 3-1-6 Introducing strength degradation (γ=0.0 as default value)
97
Relationship between curvature and rotation
εc
θ EI
M M
A B
θ M
D
φ
l
Figure 3-1-7 Rotation angle and curvature at beam ends
Let’s think about the relationship between curvature and rotation at the end of a beam.
In the above loading condition, the relationship between moment and rotation is
6 EI
M= θ (3-1-20)
l
On the other hand, the relationship between moment and curvature is
M
φ= (3-1-21)
EI
Therefore,
6
φ= θ (3-1-22)
l
Assuming the neutral axis is in the middle of the section, the relationship between curvature and
compression strain at the section end is
εc
φ= (3-1-23)
D/2
Therefore, the relationship between rotation and compressive strain is
l l
θ= φ= εc (3-1-24)
6 3D
l
Assuming D ≈ , then
9
θ = 3ε c (3-1-25)
If ε c reaches 0.003, θ is around 0.01 (=1/100).

It corresponds to the yielding rotation of a beam.
98
c) Nonlinear shear spring
Hysteresis model of nonlinear shear spring is defined as the shear force – shear rotation relationship using
an origin-oriented poly-linear model.
ηA s
MA MA + MB
MB Q=
A ηB B l
s
η=
A η=
B
l l
Q Q
k y 3 = 0.001k 0
Qy
k y3
Qu
Qc s
k0 = GA l
k0
sc sy su s
Figure 3-1-8 Force–deformation relationship of shear spring
Yield shear force
The yield shear force, Q y is calculated as,
 0.053 pt 0.23 (σ B + 18) 

Qy =  + 0.85 p w ⋅ σ wy b ⋅ j (3-1-26)
 M /(QD) + 0.12 
where,

Crack shear force

The crack shear force is, Qc , is assumed as,
Qy
Qc = (3-1-27)
3
99
Ultimate shear force
The ultimate shear force is, Qu , is assumed as,
Qy + k y 3 ( su − s y )
Qu = (3-1-28)
NOTE)
In STERA_3D, the stiffness after yielding is temporary assumed to be positive to avoid instability in
numerical analysis.
Q
Qu
Qy
k y 3 = 0.001k 0
Qc
k0 = GA l
sc sy su s
Figure 3-1-9 Stiffness after yielding
Crack shear deformation

The crack shear deformation is obtained as,
Qc
sc = γ c l , γc = (3-1-29)
GA
Yield shear displacement
The yield shear deformation is assumed as,
1
sy = γ yl , γ y = (3-1-30)
250
Ultimate shear displacement
The ultimate shear deformation is assumed as,
1
su = γ u l , γ u = (3-1-31)
100
100
d) Modification of initial stiffness of nonlinear springs
In numerical calculation, a large dummy value is used for the initial stiffness of the nonlinear spring to
represent rigid condition. This large stiffness may cause an error for estimating the force from the
displacement. One possible way to solve the problem is to reduce the initial stiffness of the nonlinear spring
to be a certain value reasonable for calculation, and on the other hand, increase the stiffness of the elastic
element so that the total initial stiffness of the beam element does not change from the original one. This
idea is proposed by K-N Li (2004) for MS model.
φ θ
M τ τ M
A B
θ φ

M M M
My My
6 EI = +
Mc k0 = Mc
l
k 0 α y k0 k0 kn ≈ ∞
My My  1 M
θy = = τ= + φy =  − 1 y
α y k0 k0 α  k
 y  0
Increase Reduce
stiffness stiffness
M M
My
+ Mc
k = k0 / γ 1
*
0
kn* = EI / p1
τ* φ c* φ y* φ*
My γ 1M y  1 M
θy = = τ* = + φ y* =  − γ1  y
α y k0 k0 α  k
 y  0
Figure 3-1-10 Modification of moment – rotation relationship
101
The idea is realized using flexibility reduction factors, γ 1 (< 1), γ 2 (< 1) , in the relationship between the
displacement vector and force vector of the elastic element in Equation (2-1-1) as,
 l' l' 
γ 1 3EI −
6 EI y
0 
τ ' yA   y   M ' yA 
   l' l'  
τ ' yB  =  − γ2 0  M ' yB  (3-1-32)

 δ '   6 EI y 3EI y
 N ' 
 x l'  x 
 0 0
 EA 
l' l' l' l'

It must be γ1 > or γ 1 > 0.5 and. γ 2 > or γ 2 > 0.5 .
3EI y 6 EI y 3EI y 6 EI y
Also the initial flexibility matrix of the nonlinear spring can be expressed as follows, introducing the
parameters, p1 , p 2 to increase the initial flexibility.
φ yA   p1 EI 0   M ' yA 
 =   (3-1-33)
φ yB   0 p2 EI  M ' yB 
When p1 → 0, p 2 → 0 , it represents the infinite stiffness for rigid condition. Accordingly, the crack and
yield rotation will be modified as,
Mc  1 M
φ c* = p1 , φy =  − γ1  y
*
(3-1-34)
EI α  k
 y  0
My
φ = θ − γ1 (3-1-35)
k0
Making the modified flexibility matrix to be identical to the original one,
 l' l'   p1 l' l' 

 3EI −
6 EI y
0   EI + γ 1 3EI −
6 EI y
0 
 y   y 
 l'  p2 l' 
0  = +γ2 0  (3-1-36)
 3EI y  EI 3EI y
   
 sym. l'   l' 
sym.
 EA  original  EA  mod ified
This gives the flexivility reduction factors as:

3 3
γ1 =1− p1 , γ 2 = 1 − p 2 (3-1-37)
l' l'
From the conditions γ 1 > 0.5 and γ 2 > 0.5 ,
l' l'
p1 < , p2 < (3-1-38)
6 6
l'
K-N Li (2004) calls these parameters, p1 , p 2 , as “plastic zones” and recommends to be p1 = p 2 = .
10
Them the reduction factors will be γ 1 = γ 2 = 0.7 .
102
e) Modification of stiffness degradation factor at the yield point
(The following modification of the stiffness degradation factor, α y , is suggested by Prof. Okano at Chiba
University.)
From Equations (3-1-32) and (3-1-34), the yield rotation of the member θ y under anti-symmetric loading
condition, M A = M B = M y , is calculated as,
(2γ − 1)M y  1 M  1 M
θy = + −γ  y =  + γ − 1 y (3-1-39)
α  k   k
k0  y  0 αy  0
where γ1 = γ 2 = γ .
The yield rotation θ y in Equation (3-1-39) is different from the formula in Figure 3-1-10 since the factor
γ is multiplied to only diagonal elements of flexural matrix in Equation (3-1-32).
The stiffness degradation factor is then obtained as,
1  1 
= + γ − 1 (3-1-40)
α ' y  α y 

To realize the designated value of stiffness degradation factor, α y should be modified as,
 1 
αy = 1  +1− γ  (3-1-41)
α' 
 y 
For example, to realize the stiffness degradation factor α ' y = 0.4, assuming γ = 0.7 , the modified α y is
 1 
αy =1  + 1 − 0.7  = 0.357
 0.4 
This modification is done automatically in STERA_3D.
103
f) Modification of considering rigid zone ratio
A beam-column connection can be idealized as a rigid zone. In case of a beam element, the default length
of the rigid zone is set to be a half of the column width, and the nonlinear bending spring of the beam
element is arranged at the position of the column face.
On the other hand, if elastic deformation of the connection is considered by reducing the length of rigid
zone, the position of the nonlinear bending spring will be inside the connection area. In this case, when the
nonlinear bending spring is yielding, the moment value at the position of the column face is smaller than
the yield moment.
Column Column
My
A B
My
dA l dB
A B
ηd A l' ηd B
My' My
Figure 3-1-11 Reduction of rigid zone and modification of yield moment
To make the moment at the column face to be the same as yield moment, the yield moment of the nonlinear
bending spring is increased as,
l / 2 + (1 − η )d A
M y '= M y = ξM y
l/2
(3-1-42)
d
ξ = 1 + 2(1 − η ) A
l
For example, when l = 540cm, d A = 30cm, η = 0.75 ,
ξ = 1 + 2 × (0.25) × 30 / 540 = 1.027 (3-1-43)
104
3.1.2 Steel Beam
tf
H
tw
B : Width, H : Height, tw, tf : Thickness
Figure 3-1-12 Steel Beam Section
AN = 2 Bt f + (H − 2t f )t w (3-1-44)
Area of section to calculate shear deformation ( )
AS = (H − 2t f )t w (3-1-45)

BH 3 − ( B − t w )( H − 2t f ) 3
Iy = : along strong axis (3-1-46)
12
3 3
2t f B + ( H − 2t f )t w
Iz = : along weak axis (3-1-47)
12
Moment of inertia for torsion
3 3
2 Bt f + ( H − 2t f ) t w
J= (3-1-48)
3
105
θA τA
δx
MA MB
MA MB A B
A B τB
θB Nx
Nx
elastic element
l φA
θ A  τ A  φ A  MA MB
      A B
θ B  = τ B  + φ B  φB
δ  δ   0 
 x  x  
Hysteresis model of a nonlinear bending spring is defined as the moment-rotation relationship under the
anti-symmetry loading as shown in Figure 3-1-14. The initial stiffness of the nonlinear spring is supposed
to be infinite, however, in numerical calculation, a large enough value is used for the stiffness.
φ θ
M τ τ M
A B
θ φ
Moment distribution
M M M
My My
6 EI = +
k0 =
l kn ≈ ∞
k0 k0
θy θ τ φy φ
106
Yield moment force

 1 
M y =  Bt f ( H − t f ) + t w ( H − 2t f ) 2 σ y (3-1-49)
 4 
where,
σy : Strength of steel (N/mm2)
σy
My
Figure 3-1-15
Yield rotation
From Figure 3-1-14, the yield rotation of the nonlinear bending beam, φ y , is supposed to be zero value,
however, in STERA_3D program, it is assumed as,
φ y = 0.001 θ y (3-1-50)
where
6 EI
θ y = M y / k0 , k0 =
l
Hysteresis model
A bi-linear model is assumed for the hysteresis model.
My
φy
Figure 3-1-16 Hysteresis of steel
107
c) Modification of initial stiffness of nonlinear springs
In numerical calculation, a large dummy value is used for the initial stiffness of the nonlinear spring to
represent rigid condition. This large stiffness may cause an error for estimating the force from the
displacement. One possible way to solve the problem is to reduce the initial stiffness of the nonlinear spring
to be a certain value reasonable for calculation, and on the other hand, increase the stiffness of the elastic
element so that the total initial stiffness of the beam element does not change from the original one. This
idea is proposed by K-N Li (2004) for MS model, and can be used for nonlinear spring model also.
φ θ
M τ τ M
A B
θ φ
M M M
My My
6 EI = +
k0 =
l kn ≈ ∞
k0 k0
θy θ τ φy φ
My My
θy = = τ= + φy ≈ 0
k0 k0
Increase Reduce
stiffness stiffness
M M
My
+
k0* = k0 / γ 1 kn* = EI / p1
τ* φ y* φ*
My γ 1M y My
θy = = τ* = + φ y* = (1 − γ 1 )
k0 k0 k0
108
The idea is realized using flexibility reduction factors, γ 1 (< 1), γ 2 (< 1) , in the relationship between the
displacement vector and force vector of the elastic element in Equation (2-1-1) as,
 l' l' 
γ 1 3EI −
6 EI y
0 
τ ' yA   y   M ' yA 
   l' l'  
τ ' yB  =  − γ2 0  M ' yB  (3-1-51)

 δ '   6 EI y 3EI y
 N ' 
 x l'  x 
 0 0
 EA 
l' l' l' l'

It must be γ1 > or γ 1 > 0.5 and. γ 2 > or γ 2 > 0.5 .
3EI y 6 EI y 3EI y 6 EI y
Also the initial flexibility matrix of the nonlinear spring can be expressed as follows, introducing the
parameters, p1 , p 2 to increase the initial flexibility.
φ yA   p1 EI 0   M ' yA 
 =   (3-1-52)
φ yB   0 p2 EI  M ' yB 
When p1 → 0, p 2 → 0 , it represents the infinite stiffness for rigid condition. Accordingly, the yield
rotation will be modified as,
My
φ y* = p1 (3-1-53)
EI
My
φ = θ − γ1 (3-1-54)
k0
 l' l'   p1 l' l' 

 3EI −
6 EI y
0   EI + γ 1 3EI −
6 EI y
0 
 y   y 
 l'  p2 l' 
0  = +γ2 0  (3-1-55)
 3EI y  EI 3EI y
   
 sym. l'   l' 
sym.
 EA  original  EA  mod ified

3 3
γ1 =1− p1 , γ 2 = 1 − p 2 (3-1-56)
l' l'
From the conditions γ 1 > 0.5 and γ 2 > 0.5 ,
l' l'
p1 < , p2 < (3-1-57)
6 6
l'
K-N Li (2004) calls these parameters, p1 , p 2 , as “plastic zones” and recommends to be p1 = p 2 = .
10
Then, the reduction factors will be γ 1 = γ 2 = 0.7 .
109
d) Shear spring for damper using low yield strength steel
ηA
s
MA MB
A ηB B
MA + MB s
Q= , η=
A η=
B
l l
Nonlinear relationship between shear force Q and deformation s is defined.
110
3.1.3 SRC Beam
S
as
d1
t d1
a1
D
d2 a2 d2
B
b1
B : Width of beam,
D : Height of beam,
S : Effective width of slab, tf
t : Thickness of slab
d1 : Distance to the center of top main rebars, tw h1
d2 : Distance to the center of bottom main rebars,
a1 : Area of top main rebars,
a2 : Area of bottom main rebars
as : Area of rebars in slab
b1 : Width of steel
h1 : Height of steel
tw : Thickness of web
tf : Thickness of flange
Figure 3-1-18 SRC Beam Section
AN = BD + (S − B )t + (nE − 1)(a1 + a2 + a S + a ST )) (3-1-58)
where,
aST = 2(b1 − tw )t f + h1 tw :Area of steel

AS = BD (3-1-59)
2 2
BD 3 ( S − B )t 3  D  t 
Ie = + + BD g −  + (S − B )t  D − − g  +
12 12  2  2 
2
(nE − 1)a1 (d1 − g )2 + (nE − 1)a2 (D − d 2 − g )2 + (nE − 1)aS  D − t 

− g +
 2 
3 3
b1h1 − (b1 − t w )(h1 − 2t f )
(nE − 1) (3-1-60)
12
111
where, g is the center of beam section calculated by
BD 2 / 2 + ( S − B)t (D − t / 2 ) + (nE − 1)(a1d1 + a2 (D − d 2 ) + aS (D − t / 2 ) + aST D / 2 )
g=
AN
(3-1-61)
Hysteresis model of a nonlinear bending spring is the same as RC beam.
Crack moment force
M c1 = 0.56 σ B Z e1 , Z e1 = I e / g when tension in top main rebars (3-1-62)
M c 2 = 0.56 σ B Z e 2 , Z e 2 = I e / (D − g ) when tension in bottom main rebars (3-1-63)
where,
σB : Compression strength of concrete (N/mm2)
Z e1 , Z e 2 : Section modulus
Yield moment force

M y = M y1, 2, RC + M y ,S (3-1-64)
where
M y1, 2, RC : Yield moment of reinforced concrete (3-1-65)
M y1, RC = 0.9a1σ y (D − d1 ) + 0.9aS σ y (D − t / 2 ) when tension in top main rebars
M y 2, RC = 0.9a2σ y (D − d 2 ) when tension in bottom main rebars
where,
σy : Strength of rebar (N/mm2)
 1 
M y ,S = b1t f (h1 − t f ) + t w (h1 − 2t f ) 2 σ y ,S : Yield moment of steel (3-1-66)
 4 
where,
σ y, S : Strength of steel (N/mm2)
112
Appendix 3.1:
A-1. Hysteresis of Degrading Trilinear Slip Model
In OPTION menu in Beam Editor, you can control the shape of hysteresis loop.
R s : Effective Slab Ratio

As described in Eq. (3-1-8), when tension in slab side, the yield moment of beam , M y1 , is
M y1 = 0.9a 2σ y (D − d 2 ) + 0.9a s σ y (D − t / 2)
where, a s is the area of rebars in effective width of slab, S b , which is defined as Eq.(3-1-17),
S b = η s Ls
η s ( R s in the menu) is the effective slab ratio , the default value is 0.1.
Depending on the effective slab ratio R s , the yiled moment M y and the yield rotaion R y will change together
as shown in the Figure below, since the tangential stiffness at the yield point, K y , is assumed to be the
same.
Ru : Ultimate rotation angle to define the maximum moment before degradation. The default value is 1/50.
K p : The stiffness after the yield rotation angle, R y .
K u : The stiffness after the ultimate rotation angle, Ru .
It can be the negative value to consider strength degradation, however, the default value of the ratio K u / K y is
1/1000 without degradation.
113
M
kp
My
Effective slab ratio Rs
ku
( η s in Eq.(3-1-17))
Stiffness over Ru
(could be negative)
ky
R
Ry Ru
Ultimate rotation angle
(recommended over 1/50)
R1 : stiffness degrading ratio in the trilinear hysteresis is 0.5. (0: no degradation)

R2 : slip stiffness ratio in the trilinear hysteresis is 0.0 (0: no slip).
R3 : strength degrading ratio in the trilinear hysteresis is 0.0.
Those parameters control the shape of hysteresis loop as descrived in Eqs. (3-1-18) and (3-1-19). That is,
Mm My  φy
α
My kr =   (α = R1 )
φ φ
 y  m
n φm n φy ks kr  Mm  φy
β
φ k s =   ( β = R2 )
φx φy φm φ 'm  φm − φ x  φm
 φy 
φ ' m = 1 + γ
n
φ m (γ = R3 )

 n φm 
More detail rule in the hysteresis loop is described in the following sections:
114
1. Elastic range
fy(1)
fc(1)
s0(1)
1
s0(2)
fc(2)
Initial stiffness, s0, is calculated from

2 s0(1) = fc(1) / dc(1)
fy(2)
2. From crack point to yield point
sy
4
sc
2 (drm(1), frm(1))
unloading point
3
s0
1
3
towards the maximum point of the other side
(drm(2), frm(2))
unloading point
115
3. Loading on the primary curve after yielding
(drm(1),frm(1))
fu 4
fy
fdm 5
ssd
xdm xd0
xd0 du xdm
(du,fu)
(drm,frm) (du,fu)
ssd fdm
su
5
ssd
4 xdm
(drm(2),frm(2))
If the displacement is less than the degrading point (du, fu),
the unloading point will the target point (drm, frm)
 fy  dy α
The stiffness of unloading. ssd, will be calculated from ssd =   , where αis the parameter to control the stiffness
d  drm
 y 
degradation depending on the ductility factor, (drm/dy). The default value of is α = 0.5
(du,fu) (drm,frm)
fdm
If the displacement is over the degrading point,
su
intersection of the LINE1 (degrading line) ssd
and LINE2 (unloading line) will be the target
LINE1
point. LINE2
xdm
If the force is lower than 0.1fy, intersection of

the LINE1 (degrading line) and LINE2 (lower (drm,frm)
boundary) will be the target point.
ssd
LINE1
0.1fy
Lower boundary
LINE2
xdm
116
4. Crossing zero force line
(drm,frm)
4
fy
5
ssd
xd0
Target point of the other side, drm(2), will be increased

according to the ductility factor, (drm(1)/dy(1)), as follows:
 drm(1) 
drm(2) = 1 + γ drm(2)
 dy (1) 
where γ is the parameter to control the strength degradation.
The default value is γ=0.0 (no degradation).
(drm(2),frm(2)) (drm(2),frm(2))
srm
(drm,frm)
(dsl,fsl)
6 5
ssl
xd0
ssl
(dsl,fsl) 6
5
srm
(drm(2),frm(2))
117
5. Calculation of slip point
srm
(dsl,fsl)
5
xd0
(dsl,fsl) 6
ssl
srm
After crossing zero force line, the stiffness will be

calculated as:
β
frm(2) dy
(drm(2),frm(2)) ssl =
drm(2) − xd 0 drm
where β is the parameter to control the slip ratio. For the
default value (β=0.0), it will be no slip and towards the
maximum point.
LINE1 xd0
6
(dsl,fsl) ssl
LINE2
8
srm
The stiffness will change at the intersection of

LINE1 and LINE2. After the intersection point,
(dsl, fsl), Level 8 will be towards the maximum
(drm(2),frm(2)) point.
ssd ssd
7 xd0 srm
ssd 6
ssl
10
8
ssd srm Stiffness unloading from Level 6 (Level 7) and

stiffness unloading from Level 8 (Level 10) is the
same as the unloading stiffness, ssd ( Level 5).
(drm(2),frm(2))
5
118
dslp = s3*xd0/(s3-s2) 4
ssd
5
ssd
xd0
ssl 5
srm 6
11
9 ssd
5 8
7 xd0
6
After crossing zero force line from Lavel 7, Level 9 will be towards the
maximum point. Stiffness unloading from Level 9 (Level 11) is the same as
the stiffness of ssd (Level 5).
5
9
xd0
ssl 5
srm 6
11
5 8
7 xd0
6
119
3.2 Column
3.2.1 RC Column
d1
a2
ac
x
a1
D
d2
y
B
B : Width of beam,
D : Height of beam,
d1 : Distance to the center of x-direction main rebars,
d2 : Distance to the center of y-direction main rebars,
a1 : Area of x-side main rebars,
a2 : Area of y-side main rebars,
ac : Area of corner main rebars
Figure 3-2-1 RC Column Section
AN  BD  n E  1a1  a 2  a c  (3-2-1)
AS  BD /  ,   1.2 (3-2-2)
2
DB 3 B 
Iy   n E  1a c  a1   d1  (3-2-3)
12 2 
2
BD 3 D 
Ix   n E  1a c  a 2   d 2  (3-2-4)
12 2 
120
Hysteresis model of nonlinear bending spring is defined as the moment-rotation relationship under the
anti-symmetry loading in Figure 3-2-2. The initial stiffness of the nonlinear spring is supposed to be infinite,
however, in numerical calculation, a large enough value is used for the stiffness.
 


 
A
Moment distribution
M
M M M
My My
6 EI = +
Mc k0  Mc
l
k0 k y   y k0 k0 kn  
c y   c  y 
The crack moment, M c is calculated as,

ND
M c  0.56  B Z e  (3-2-5)
6
where,
B : Compression strength of concrete (N/mm2)
Ze : Section modulus
N : Axial load
121
The yield moment, M y is calculated from the following formula under the axial load, N
if (0  N  N b )
 N 
M y  0.8at  y D  0.5 ND1   (3-2-6)
 bD B 
if ( N b  N  N max )
 N N 

M y  0.8a t  y D  0.12bD 2 B  max   (3-2-7)
 N max  N b 
where, N b is the balance axial force,

N b  0.4bD B (3-2-8)
and N max is the maximum axial force,
N max  bD B  As y (3-2-9)
The tangential stiffness at the yield point, k y , is obtained from the following equation,
6 EI
k y  y K0 K0  (3-2-10)
l
where,
 y is the stiffness degradation factor at the yield point, which is obtained from the following
empirical formulas:
 y   0.043  1.64npt  0.043a / D  0.3250  d / D  , 2  a / D

2
(3-2-11)
 y   0.0836  0.159a / D  0.1690  d / D  , 1  a / D  2 

2
(3-2-12)
where,

p t  a c  a1  / 2 BD  (when tension in x-main rebars)
p t  a c  a 2  / 2 BD  (when tension in y-main rebars)
a/D : ≈ Shear span-to-depth ratio (= l /( 2 D ) )
d : effective depth
d = D-d1 (when tension in bottom main rebars)
d = D-d2 (when tension in upper main rebars)
N
0  : Axial load ratio
bD B
122
The yield rotation of the nonlinear bending beam,  y , is then obtained from,
 1 My
 y    1
K
(3-2-13)
 y  0
Reference:
AIJ Standard for Structural Calculation of Reinforced Concrete Structures, Architectural Institute of Japan,
2018 (in Japanese)
123
Case 1: In the case that bending springs in x and y directions are independently defined
X-Z plane Y-Z plane
M ' yB M ' xB
B B
 yB xB
 yA xA
A A
M ' yA
M ' xA
 yA  f yA M' yA , xA  f xA M'xA at end A (3-2-14)
 yB  f yB M' yB , xB  f xB M'xB at end B (3-2-15)
where, f yA , f xA , f yB , and f xB are the flexural stiffness of nonlinear bending springs at both ends of
the element, and
f yA  1 k yA , f xA  1 k xA , f yB  1 k yB , f xB  1 k xB (3-2-16)
The rotational displacement vector of the nonlinear bending springs will be
 yA   M ' yA   f yA   M ' yA 
  M '   f xA M ' 
 xA   xA     xA 
 
 zA   f pA 0   N ' zA   0   N ' zA 
  
 yB   0   
f pB  M ' yB   f yB

 M ' yB 
 (3-2-17)
 xB   M ' xB   f xB   M ' xB 
      
 zB   N ' zB   0  N ' zB 
124
The hysteresis model for M   relationship is the degrading tri-linear slip model as used for the
hysteresis model of the bending springs of the RC beam.
M M
kp
My
My
Mc Mc
ks kr
 
x y m
 
M  My  y  Mm  y
kp   y  kr    k s    (3-2-18)
 y  
   y  m  m   x  m

125
To consider nonlinear interaction among M x  M y  N z , the nonlinear bending spring at the member
end is constructed from the nonlinear vertical springs arranged in the member section as shown in Figure
3-2-4.
N zB ,  zB
M xB ,  xB z
M yB ,  yB
B y
x  z  i xi
x
yi
i
A
y
M yA ,  yA
M xA ,  xA
N zA ,  zA
Displacement of the i-th nonlinear axial spring is,
 i   z  y i x  xi y (3-2-18)
Equilibrium condition in the nonlinear section is,
M ' y   k i  i xi   k i ( z  yi x  xi y ) x i

i i
M ' x    k i  i yi    k i ( z  yi x  xi y ) y i (3-2-19)

i i
N ' z   k i  i   k i ( z  yi x  xi y )

i i
In a matrix form
 
 M ' y  
k i xi2   k i xi y i  k x   i i
y   y 
  i 
x   k p  x 
 i i
  
M 'x    k y i
2
i   k y   i i (3-2-20)
 N'      
 k  
i i
 z   sym. z  z
i
 i
126
Therefore
 y  M ' y  M ' y 
x   k p   M ' x    f p  M ' x 
  1    
(3-2-21)
   N'   N' 
 z  z  z
For both ends
 yA   M ' yA 
  M ' 
 xA   xA 
 zA   f pA  0   N ' zA 
 
 yB   0  f pB  M ' yB  (3-2-22)
xB   M ' xB 
   
 zB   N ' zB 
127
c) Nonlinear vertical springs
The nonlinear bending spring is constructed from the nonlinear vertical springs arranged in the member
section as shown in Figure 3-2-5. This model is called “Multi-spring model” proposed by S. S. Lai, G. T.
Will and S. Otani (1984) and modified by K-N. Li (1988). The section is devided in 5 areas; where 4 corner
areas have steel springs and concrete springs and the center area has one concrete spring.
The strength and the location of nonlinear springs are obtained from the equilibrium condition under
the balance axial force, N b .
A1 1 2
A2
x x
5
3 4 Concrete spring
Steel spring
y y
(a) Original column section (b) Multi-spring model
(tension) (tension)
s fy
c d y
s dy
c f y
(compression) (compression)
(c) Hysteresis of steel spring (d) Hysteresis of concrete spring
Figure 3-2-6 Nonlinear vertical springs
Strength of steel spring

The strength of the steel spring is one-forth of total strength of rebars in the section, i.e.,
As y
s fy  (3-2-23)
4
where,
As : Total area of rebar in the section
y ： Strength of rebar
128
Strength of concrete spring
As shown in Figure 3-2-6, the strength of the corner concrete spring is obtained from the equilibrium
condition in the vertical direction under the balance axial force, N b  0.4bD B , that is,
Nb
c f y1   0.2bD B (3-2-24)
2
Therefore, the area of the corner concrete, A1 , is,
c fy
A1  (3-2-25)
0.85 B 
 Nb
s f y  c f y1 My
s f y  c f y1
xs x
 s f y1
 s f y1
Figure 3-2-7 Equilibrium condition in the column section
The area of the center concrete, A2 , is the rest of the area of the section,
A2  bD  4 A1  0 (3-2-26)
The strength of the center concrete spring is then obtained as,
c f y 2  0.85k B A2 (3-2-27)
where, k is the confined effect k  1.3 of the concrete.
Location of vertical springs

The distance between the corner springs, x s , is obtained from the equilibrium condition regarding the
moment force in Figure 3-2-6,
M y  x s 2 s f y  c f y1   x s 2 s f y  0.5 N b  (3-2-28)
Therefore,
My
xs  (3-2-29)
2 s f y  0 .5 N b
Note that M y is calculated from Equation (3-2-6) for the balance axial force, N  Nb .
129
Example)
To verify the efficiency of the Multi-Spring model for the column element, the M-N relationship is
compared between MS-model and theory using one column element. The column section is shown in the
Figure below:
N=1000kN
300cm
50cm
Figure 3-2-8
Firstly, the strengths and locations of vertical springs are calculated as
a t  15 .484 ( cm 2 )  y  1 .1 f y  32 .45 ( kN / cm 2 )  B  2 .4 ( kN / cm 2 )
N b  0.4bD B  2400 ( kN ) N max  bD B  As y  6502 ( kN )
s f y  251.2 (kN ) f
c y1  1200 (kN ) f
c y2  390 (kN ) x s  30 (cm)
In the range (0  N  N b ) , the Multi-Spring model gives
 
M y  2 s f y  0 .5 N x s
which is plotted as the solid line in Figure 3-2-8. The results of Multi-Spring model give smaller values
than theoretical results in the range 0 < N < Nb.
130
M-N relationship
6000
Theory
5000 Multi-Spring
4000
N (kN)
3000
2000
Under-estimate
1000
0
0 10000 20000 30000 40000 50000 60000
M (kN*cm)
Figure 3-2-9 Comparison of M-N relationship
K-N. Li (1988) proposed to use the following formulation for deciding the location of vertical springs
instead of Equation (3-2-29), as follows:
M y0
xs  (3-2-30)
2 s f y  0.5 N 0
where, N 0 is the axial force from the dead loads and the live loads acting on the column ( N 0  N b ),
and M y 0 is the yield moment under the axial force N 0 , that is:
 N0 
M y 0  0.8a t  y D  0.5 N 0 D1  
 bD B 
For the example column, assuming N 0 = 1000 (kN),
x s  35.8 (cm)
The yield moment is plotted as the solid line in Figure 3-2-9. It improves the results of Multi-Spring model.
131
M-N relationship
6000
Theory
5000 Multi-Spring
4000
N (kN)
3000
2000
1000
0
0 20000 40000 60000 80000
M (kN*cm)
Figure 3-2-10 Comparison of M-N relationship
132
Yield displacement of vertical spring
 N0
 2 fc M y0 Compression
y s dy c f y sf y
dc fc
2 s f y
xs dc c d y s d y
Figure 3-2-11 Equilibrium condition under the axial force N0
From the equilibrium condition under the axial force N 0 as shown in the above Figure, the yield
displacement of the tension side steel spring, s d y , is obtained as follows:
s dy  d c   y xs
fc
dc  sdy
s f y c f y
N 0 2 s f y
fc  (3-2-31)
2
 y xs
sdy 
N 0 2 s f y
1
2 s f y 2c f y
The yield displacement of concrete spring, c d y , is assumed to be the same as that of the steel spring,
c d y s d y (3-2-32)
133
d) Nonlinear shear spring
d-1) Force-deformation relationship
There are two nonlinear shear springs in x and y directions. Hysteresis model of the nonlinear shear springs
is the same as that in the beam element.
Yield shear force
 0.053 pt 0.23 ( B  18) 

Qy    0.85 p w   wy  0.1 0 b  j (3-2-33)
 M /(QD)  0.12 
where,

B : Compression strength of concrete
M/(QD) : ≈ Shear span-to-depth ratio (= l /(2 D) )
 wy : Strength of shear reinforcement
0 : Axial stress of the column
j : Distance between the centers of stress in the section (  7 / 8d ).
Crack shear force
Qc  0.3Qy (3-2-34)

The ultimate shear force is, Qu , is assumed as,
Qu  Qy  k y 3  su  s y  (3-2-35)
Qy
k y3
Qu
Qc 0.001k0
k0  GA l
sc sy su s
Figure 3-2-12 Shear force - deformation relationship
134
Qc
sc   cl , c  (3-2-36)
GA
1
s y   yl ,  y  (3-2-37)
250
1
su   u l ,  u  (3-2-38)
100
The poly-linear slip model (see Appendix) is adopted for the hysteresis of the shear spring.
Figure 3-2-13 Poly-linear slip model for shear spring
The parameters on the backbone curve can be changed in the Option Menu of Column element. The default
values are given as follows:
135
d-2) Shear spring model 1
Case 1: In the case that shear springs in x and y directions are independently defined
Qx
Qy
sx z
sy
x
 Qx y
 Qy
Figure 3-2-14 Nonlinear shear springs in column
Qx   k sx 0   sx 
    (3-2-39)
Qy   0 ksy   s y 
136
d-3) Shear spring model 2
Case 2: In the case that nonlinear interaction between shear and axial components is considered
N 'zB
M 'xB
M ' yB
B
N z ,  sz
Qx , sx
Qy , s y
A
M ' yA
M 'xA
N 'zA
Figure 3-2-15 Nonlinear shear springs
 Qx   sx 
   
Qy    k sp   s y  (3-2-40)
N   
 z  sz 
The stiffness matrix  k sp  is obtained by the Plastic Theory as explained in the Appendix (not
implemented).
137
e) Modification of initial stiffness of nonlinear springs
The same modification can be done for the nonlinear springs of column element as described for those of
beam element by reducing the initial stiffness of the nonlinear spring and increasing the stiffness of the
elastic element as shown in the following figure:
 


 
A
Moment distribution
M
M M M
My My
6 EI = +
Mc k0  Mc
l
k 0  y k0 k0 kn  
c y   c  y 
Increase Reduce
stiffness stiffness
M M
My
+ Mc
k *
0
kn*
 c*  y* 
138
Introducint the concept of “plastic zones”, the initial stiffness of the i-th multi-spring can be expressed as,
Ei Ai
k 0i  (3-2-41)
pz
where Ei : the material young’s modulus, Ai : the spring governed area, and pz : the length of assumed
plastic zone. When p z  0 , it represents the infinite stiffness for rigid condition.
From Equation (3-2-20), when we consider the flexural flexibility in x-z plane, the flexibility matrix for the
nonlinear MS section is,
1
 y   k i
0 xi2 0   pz
 M ' y   
E A x i i
2
i 0 
 M ' y 
 
i i
    (3-2-42)
 z  0 1  k 0i   N ' z   0 p z  Ei A  N ' z 
 i   i 
Also, introducing the flexibility reduction factors,  0  0,  1  0,  2  0 , the flexibility matrix of
 l' l' 
 1 3EI 
6 EI y 
 y

l ' l'  (3-2-43)
 f C     2

6 EI y 3EI y
 
 l' 
0
 EA 
 
 p z1 l' l' 
 l' l'   1  0
 E A x2 

 3EI  0  3EI y 6 EI y
6 EI y i i i

 y
 
i

l' pz2 l'
 0   2 0 


3EI y 

  Ei Ai xi
2
3EI y 
l'   i 
 sym.  p z1 pz2 l' 
 EA  original sym.  0


i Ei A i Ei A EA 
mod ified
(3-2-44)
Since Ax
i
2
i i  I y , this gives the flexivility reduction factors as:
3 3 1
1  1 p z 1 ,  2  1  p z 2 ,  0  1   p z1  p z 2  (3-2-45)
l' l' l'
l'
Adopting p z1  p z 2  as discussed for beam element, the reduction factors will be:
10
 1   2  0.7 ,  0  0.8 (3-2-46)
139
References
1) S. S. Lai, G. T. Will, and S. Otani (1984), “Model for Inelastic Biaxial Bending of Concrete Members,”
Journal of Structural Division, ASCE, Vol. 110, ST1, 1984, pp.2563-2584.
2) K-N. Li (1988), “Nonlinear Earthquake Response of Reinforced Concrete Space Frames,” the
dissertation for the degree of Doctor in University of Tokyo (in Japanese), 1988.12.
3) K-N. Li (2004), CANNY, Technical Manual.
140
3.2.2 Steel Column
tf t2
H t1
tw t1
B B D
B : Width, H : Height, tw, tf, t1, t2, t: Thickness
Figure 3-2-17 Steel Column Section

AN  total area of section (3-2-50)
AS  ( ) (3-2-51)
AS  0.5 AN
Figure 3-2-18 Area of section for shear

1) H section
BH 3  ( B  t w )( H  2t f ) 3
I : along strong axis (3-2-52)
12
3
2t f B  ( H  2t f )t w
3
I : along weak axis (3-2-53)

12
2) Box section
BH 3  ( B  2t1 )( H  2t 2 ) 3
I (3-2-54)
12
141
3) Circle section
I

64
D 4
 D  2t 
4
 (3-2-55)
Moment of inertia for torsion

1) H section
3 3
2 Bt f  ( H  2t f ) t w
J (3-2-56)
3
4) Box section
2t1t 2 ( B  t1 ) 2 ( H  t 2 ) 2
J 2 2
(3-2-57)
BHt1t 2  t1  t1
5) Circle section
J

32
D 4
 D  t 
4
 (3-2-58)
142
To consider nonlinear interaction among M x  M y  N z , the nonlinear bending spring at the member
end is constructed from the nonlinear vertical springs arranged in the member section as shown in the
following Figure.
N zB ,  zB
M xB ,  xB z
M yB ,  yB
B y
x  z  i xi
x
yi
i
A
y
M yA ,  yA
M xA ,  xA
N zA ,  zA
 i   z  y i x  xi y (3-2-59)
Equilibrium condition in the nonlinear section is,
M ' y   k i  i xi   k i ( z  yi x  xi y ) x i

i i
M ' x    k i  i yi    k i ( z  yi x  xi y ) y i (3-2-60)

i i
N ' z   k i  i   k i ( z  yi x  xi y )

i i
In a matrix form
 
 M ' y  
k i xi2   k i xi y i  k x   i i
y   y 
  i 
x   k p  x 
 i i
  
M 'x    k y i
2
i   k y   i i (3-2-61)
 N'      
 k  
i i
 z   sym. i
z  z
 i
Therefore
143
 y  M ' y  M ' y 
x   k p   M ' x    f p  M ' x 
  1    
(3-2-62)
   N'   N' 
 z  z  z
For both ends
 yA   M ' yA 
  M ' 
 xA   xA 
 zA   f pA  0   N ' zA 
 
 yB   0  f pB  M ' yB  (3-2-63)
xB   M ' xB 
   
 zB   N ' zB 
Hysteresis model of nonlinear bending spring is defined as the moment-rotation relationship under the
anti-symmetry loading as shown in Figure 3-2-20. The initial stiffness of the nonlinear spring is supposed
to be infinite, however, in numerical calculation, a large enough value is used for the stiffness.
 


 
A
Moment distribution
M
M M M
My My
6 EI = +
k0 
l kp  
k0 k0
y   y 
144
Yield moment force (full plastic moment)
1) I shape
tf
H
tw
a) When the neutral axis is inside the web, i.e., N < Aw y = t w ( H - 2t f ) y
M y  M y 0  y 0 t w y
2
(3-2-64) Bt f  y
1
t w ( H  2t f ) y
2
where
 1  B
M y 0   Bt f ( H  t f )  t w ( H  2t f ) 2  y
 4 
N M y0
y0 
2t w y H
b) When the neutral axis is inside the flange, i.e., N  Aw y  t w ( H - 2t f ) y
H  H 
M y  B  y 0   y 0  y (3-2-65) H 
2  2  B  y 0  y
2 
where
1  N  N y 
y0  H
2  B y 


N y  2 Bt f  t w H  2t f   y
My
145
2) H shape
a) When the neutral axis is inside the web, i.e., N  Aw y  t w H y
1
M y  M y 0  y 0 H y
2
(3-2-66) Bt f  y
2
where 1
t w ( H  2t f ) y
2
1 1 2 
M y 0   B 2 t f  t w ( H  2t f ) y
2 4  H
M y0
N
y0 
2 H y
B
b) When the neutral axis is inside the web, i.e., N  Aw y  t w H y
B  B  1 
M y  2t f   y 0   y 0  y (3-2-67) t f  B  y 0  y
2  2   2 
where
1  N  N y 
y0   B
2  2t f  y 

H
My
 
N y  2 Bt f  t w H  2t f   y
3) Box shape
1
t w ( H  2t f ) y
t2 Bt f  y 2
H M y0
t1
B
B H
a) Moment around x-axis
M y  M y ( I shape by changing t w  2t1 , t f  t 2 ) (3-2-68)
b) Moment around y-axis
M y  M y ( I shape by changing t w  2t 2 , t f  t1 , B  H ) (3-2-69)
146
4) Circle shape
t1
 N 
M y  M y 0 cos  (3-2-70)
 2N 
 y 
where
M y 0  D  t1  t1 y
2
N y   D  t1 t1 y
Yield rotation
The yield rotation is
6 EI
 y  M y / k0 , k0  (3-2-71)
l
147
section as shown in Figure 3-2-21. This model is called “fiber model”. The section is devided in several
areas which have steel springs.
30°
(H-2tf)/5 tf t2
H t1
tw t1
B B
B : Width, H : Height, tw, tf, t1, t2, t: Thickness
(tension)
f y ,i
d y ,i
(compression)
Hysteresis of steel spring
Strength of steel spring

The strength of the i-th steel spring is,
f y ,i  Ai y (3-2-72)
where, Ai : the spring governed area,  y : the strength of steel
Yield displacement of steel spring

The yield displacement of the i-th steel spring is,
d y ,i  f y ,i / k 0i , k 0i  E s Ai (3-2-73)
where Es : the young’s modulus of steel
148
The same modification can be done for the nonlinear springs of column element as described for those of
beam element by reducing the initial stiffness of the nonlinear spring and increasing the stiffness of the
elastic element as shown in the following figure:
 


 
A
Moment distribution
M
M M M
My
6 EI = +
k0 
l kp  
k0 k0
y   y 
Increase Reduce
stiffness stiffness
M M
My
+
k 0*
  y* 
Ei Ai
k 0i  (3-2-74)
pz
149
plastic zone. When p z  0 , it represents the infinite stiffness for rigid condition.
From Equation (3-2-20), when we consider the flexural flexibility in x-z plane, the flexibility matrix for the
nonlinear MS section is,
1
 y   k i
0 xi2 0   pz
 M ' y   
E A x i i
2
i 0
 M ' y 
 
i i
    (3-2-75)
 z  0 1  k 0i   N ' z   0 p z  Ei A  N ' z 
 i   i 
Also, introducing the flexibility reduction factors,  0  0,  1  0,  2  0 , the flexibility matrix of
 l' l' 
 1 3EI 
6 EI y 
 y

l ' l' 
 f C     2

(3-2-76)
6 EI y 3EI y
 
 l' 
0
 EA 
 
 p z1 l' l' 
 l' l'   1  0
 E A x2 

 3EI  0  3 EI 6 EI y
6 EI y i i i y

 y
 
i

l' pz2 l'
 0   2 0 


3EI y 

  i i i
E A x 2
3 EI y 
l'   i 
 sym.  p z1 pz2 l' 
 EA  original  sym.  0
 i Ei A i Ei A EA 
mod ified
(3-2-77)

3 3 1
1  1 p z 1 ,  2  1  p z 2 ,  0  1   p z1  p z 2  (3-2-78)
l' l' l'
l'
Adopting p z1  p z 2  as discussed for beam element, the reduction factors will be:
10
 1   2  0.7 ,  0  0.8 (3-2-79)
150
3.2.2 Column with direct input
X-Z plane Y-Z plane

M ' xB M ' xB
M ' yB
M ' yB
B B B
 ' xB
 ' yB
N 'z Z
l' = +
T 'z
 ' yA  ' xA X
A A A
M ' yA Y
M ' yA
M ' xA M ' xA
In case of direct input for Moment-Rotation relationship, we neglect nonlinear interaction among
M x  M y  N z and define the flexural stiffness of nonlinear bending springs in X and Y directions
independently. The rotational displacement vector of the nonlinear bending springs will be
 yA   M ' yA   f yA   M ' yA 
  M '   f xA M ' 
 xA   xA     xA 
 
 zA   f pA 0   N ' zA   0   N ' zA 
  
 yB   0   
f pB  M ' yB   f yB

 M ' yB 
 (3-2-80)
 xB   M ' xB   f xB   M ' xB 
      
 zB   N ' zB   0  N ' zB 
 ' yA   ' yA   yA   yA   M ' yA 

 '   '      M ' 
 yB   yB   yB   yB   yB 
 'xA   ' xA       M ' 
     xA    xA    f C  xA  (3-2-81)
 'xB   ' xB   xB   xB   M ' xB 
  ' z   ' ' z  z   0   N 'z 
         
  'z    ' z elastic element  0 bending spring  0  shear spring  T ' z 
151
The flexural matrix [ f C ] is;
 l' 1 l' 1 
 f yA  3EI  k l'2  
6 EI y k sx l'2 
 y sx

 l' 1 
 f xA   0 
 3EI y k sx l'2 
 l' 1 l' 1 
 f yB     
3EI x k sy l'2 6 EI x ksy l'2
 C  
f  

 l' 1 
f xB  
 3EI x ksy l'2 
 
 l' 
 EA 
 l' 
 sym.
 GI z 
(3-2-82)
152
3.2.3 SRC Column
d1
x
a2
ac
a1
y D
d2
B
B : Width of beam,
D : Height of beam,
d1 : Distance to the center of x-direction main rebars,
d2 : Distance to the center of y-direction main rebars,
a1 : Area of x-side main rebars,
a2 : Area of y-side main rebars,
ac : Area of corner main rebars
b1 : Width of steel
h1 : Height of steel
tw : Thickness of web
tf : Thickness of flange
Figure 3-2-24 RC Column Section

AN  BD  nE  1a1  a2  ac  aST  (3-2-83)
where,
nE  E s / Ec : Ratio of Young’s modulus between steel (Es) and concrete (Ec)
aST  n f b1  t w  t f  nw h1 t w :Area of steel
n f  2, nw  1 : Type1, Type2,
n f  4, nw  2 : Type3
n f  3, nw  2 : Type4, Type5
AS  BD /  ,   1.2 (3-2-84)
153
DB 3  B 
2

Iy   n E  1 a c  a1   d 1   I S , y  (3-2-85)
12  2  
 
BD 3  D 
2

Ix   n E  1 a c  a 2   d 2   I S , x 
 (3-2-86)
12  2  
 
where
IS : Moment of inertia of steel
I S,x I S,y
II 
1
12

b1h1  b1  t w h1  2t f 
3 3
 IH 
1
12
 3
2t f b1  (h1  2t f )t w
3

Type 1
IH II
Type 2
II  IH II  IH
Type 3
2
II  IH h 
I I  I H  AH  1 
Type 4 2
2
h  II  IH
I I  I H  AH  1 
Type 5 2

Hysteresis model of a nonlinear bending spring is the same as RC beam.
Crack moment force

ND
M c  0.56  B Z e  (3-2-87)
6
154
Yield moment force
M y  M y , RC  M y , S (3-2-88)
where
M y , RC : Yield moment of reinforced concrete
 Nb 
M y , RC  0.8a t  y D  0.5 N b D1   (3-2-89)
 bD B 
M y,S : Yield moment of steel
M y,S , x M y,S , y
M yI M yH
Type 1
M yH M yI
Type 2
M yI  M yH M yI  M yH
Type 3
M yI  M yH M yI  M yT
Type 4
M yI  M yT M yI  M yH
Type 5
 1 
M yI  b1t f (h1  t f )  t w (h1  2t f ) 2  y ,S
 4 
1 2 1 2 
M yH   b1 t f  t w (h1  2t f ) y , S
2 4 
 1 
M yT  b1t f (h1  t f )  t w (h1  t f ) 2  y , S
 2 
b1 M yH h1 b1
M yI M yT
h1 b1 h1
155
Appendix 3.2:
A-1. Hysteresis of Steel and Concrete Springs of Multi-Spring Models for RC elements
a) Steel spring
For the steel spring, the maximum-oriented model is adopted for the hysteresis before yielding, and the
tri-linear model is adopted after yielding as shown in Figure 3-2-15.
φ sf y
−φ sf y
(a) before yielding point (b) after yielding point
s fy
ν sf y
E i Ai
k 0i = s
pz
s d *y
Figure A-1-1 Normal tri-linear model for steel spring
The hysteresis of steel spring has the degradation point at the forces, ν sf y and φ s f y , where ν and φ
are the arbitrary parameters (ν < 1, φ < 1) . The STERA_3D Program adopts the values as:
ν = 1 / 3, φ = 0.5 (3-2-47)
*
Then, the yield deformation, s d y, may be obtained by Equations (3-2-31) and (3-2-13) considering the
reduction factor γ.
φ y* x s
sdy = (3-2-48)
N 0 +2 s f y
1+
2 s f y +2c f y
 1 1My
φ y* =  −  (3-2-49)
 α y γ  k0
 
156
b) Concrete spring
The hysteresis of concrete spring is also defined as tri-linear hysteresis model as shown in Figure 3-2-16.
After compression yielding, strength degradation is considered by reducing the strength of the target point
in reloading stage.
dc
fc
fy
(a) hysteresis rule after compression crack point
dc
fc
fy
dm dc
fc
fy
(b) hysteresis rule after compression yield point
 dp 
dm' = dm1 + r 
 dy  dp
dm’ dm dc
fc
fy
(c) strength degradation rule
Figure A-1-2 Tri-linear hysteresis model for concrete spring
157
A-2. Hysteresis of Poly-linear Slip Model for Shear Springs for RC and Masonry members
Reference: FRAME-D manual, Tohoku University, 1983 (in Japanese)
The poly-linear slip model is defined as the following hysteresis model.
Force
FY
(DRX, FRX)
SY
FC SU
SF
S2 S1
SC
-DU -DY 0.15 FRX
DISP
0.15 FRN DY DU
-FC
(DRN, FRN)
-FY SE
(DRX, FRX)
FRX + FRN
SE =
DRX + DRN
S1= 2 × SE S1
S1
2 1.2 × SE
S=
(DL2, FL2) (DL1, FL1)
FL
= 1 FL
= 2 0.15 × FRX
FL
= 3 FL
= 4 0.15 × FRN
(DL4, FL4) (DL3, FL3)
S1
S1
(DRN, FRN)
Figure A-2- Poly-linear slip hysteresis model for shear spring
158
3.3 Wall
3.3.1 RC Wall
l w1
lw
lw2
t
x
y
l w , l w1 , l w 2 : Width of wall,
t : Depth of wall,
C1, C2 : Side columns,
aw : Area of rebars in a wall panel
Figure 3-3-1 Wall Section
AN = AN ,C1 + AN ,C 2 + t l w 2 + (n E − 1)(a w ) (3-3-1)
where,
AN ,C1 , AN ,C 2 : Area of section of side columns for axial deformation

AS = AS ,C1 + AS ,C 2 + t l w 2 / κ , κ = 1.2 (3-3-2)
where,
AS ,C1 , AS ,C 2 : Area of section of side columns for shear deformation

2 2
l t3 l  l 
I y = I y ,C 1 + I y ,C 2 + w 2 + AN ,C1  w1  + AN ,C 2  w1  (3-3-3)
12  2   2 
where,
I y ,C 1 , I y ,C 2 : Moment of inertia of side columns
159
For the out of wall direction, each side columns behave independently in the same way as the column
element. Therefore, we discuss here only the hysteresis model in the wall panel direction. Hysteresis model
of nonlinear bending spring is defined as the moment-rotation relationship under the symmetry loading in
Figure 3-3-5. The initial stiffness of the nonlinear spring is supposed to be infinite, however, in numerical
calculation, a large enough value is used for the stiffness.
−M
φ τ
θ
θ
φ τ
A
Moment distribution
M
M M M
My My
2 EI
Mc k0 = = + Mc
l
k0 k y = α y k0 k0 kp ≈ ∞
The yield moment, M y is obtained from the equilibrium condition in Figure 3-3-6 as,
M y = a sσ y l w + 0.5a wσ wy l w + 0.5Nl w (3-3-4)
where,
as : Total area of rebar in the side column
σy : Strength of rebar in the side column
aw : Total area of vertical rebar in the wall panel
σ wy : Strength of rebar in the wall panel
N : Axial load from the dead load
160
N
a sσ y
My
a wσ wy
lw
Figure 3-3-3 Equilibrium condition under yielding moment
The crack moment, M c is assumed to be,
M c = 0.3M y (3-3-5)
The tangential stiffness at the yield point, k y , is obtained from the following equation:
k y = 0.2 K 0 (3-3-6)
The yield rotation of the nonlinear bending beam, φ y , is then obtained from,
 1 My
φ y =  − 1 (3-3-7)
K
α y  0
where, the stiffness degradation factor, α y , is assumed as,
α y = 0.02 (3-3-8)
161
Case 1: In the case that bending springs are independently defined
M 'xB1 M ' yBc M 'xB 2

z
B1 Bc B2
φxB1 φ yBc φxB 2 x
y
φxA1 φ yAc φxA 2
A1 Ac A2
M ' yAc
M 'xA1 M 'xA 2
Figure 3-3-4 Nonlinear bending springs in the wall
φxA1 = f xA1M'xA1 , φxB1 = f xB1M'xB1 in y-direction at Side Column 1

φxA 2 = f xA 2 M'xA 2 , φxB 2 = f xB 2 M'xB 2 in y-direction at Side Column 2 (3-3-9)
φ yAc = f yAc M' yAc , φ yBc = f yBc M' yBc in x-direction at center Wall panel
where, f xA1 , f xB1 , f xA 2 , f xB 2 , and f yAc , f yBc are the flexural stiffness of nonlinear bending springs
at side columns and the center wall panel of the element, and
f xA1 = 1 k xA1 , f xB1 = 1 k xB1
f xA 2 = 1 k xA 2 , f xB 2 = 1 k xB 2 (3-3-10)
f yAc = 1 k yAc , f xB1 = 1 k xB1
ϕ yAc   M ' yAc   f yAc   M ' yAc 

ϕ  M '   f xA1  M ' 
 xA1   xA1     xA1 
ϕ xA 2   M 'xA 2   f xA 2   M 'xA 2 
    f pA  0  N 'zAc 
   
 ε zAc  =   0   N 'zAc 
      M ' yBc 
ϕ yBc   0

 f pB    M ' yBc 
 
f yBc
 
ϕ xB1   M 'xB1   f xB1   M ' xB1

      M' 

ϕ xB 2   M 'xB 2  
f xB 2
  xB 2 
 
 ε zBc 

 N 'zBc 
  0   N 'zBc 
(3-3-11)
162
The hysteresis model for M − φ relationship is the degrading tri-linear slip model as used for the
hysteresis model of the bending springs of the RC beam.
M M
kp
My
My
Mc Mc
ks kr
φ φ
φx φy φm
α β
 My  My  φy  Mm  φy
kp =η   kr =   k s =   (3-3-12)
ϕ φ φ
 y   y  m  φm − φ x  φm

163
To consider nonlinear interaction among M x − M y − N z , the nonlinear bending spring at the member
end is constructed from the nonlinear vertical springs arranged in the member section as shown in Figure
3-3-2.
z
φx 2
φx1 φ yc
ε zc εi
yi x
xi
ε i = ε zc + xiφ yc in a wall panel

ε i = ε zc − y iφ x1 + xiφ yc in a side column 1 (3-3-13)
ε i = ε zc − y iφ x 2 + xiφ yc in a side column 2
N ' zc , ε zc
M ' yc , φ yc
Figure 3-3-7 Equilibrium condition in the wall panel direction
164
In the wall panel direction, all vertical springs in the nonlinear section are assumed to work against the
moment and the axial force. The equilibrium conditions are,
Nc N1 N2
M ' yc = ∑ k i ε i xi + ∑ k i ε i xi + ∑ k i ε i xi
i i i
Nc N1 N2
= ∑ k i (ε zc + xiφ yc ) x i + ∑ k i (ε zc − y iφ x1 + xiφ yc ) xi + ∑ k i (ε zc − y iφ x 2 + xiφ yc ) xi
i i i
φ yc 
 
 Nc + N 1+ N 2 N1 N2 Nc + N 1+ N 2
 φ x1 
=  ∑ k i xi2 − ∑ k i xi y i − ∑ k i xi y i ∑i k i xi  φ 
 i i i  x2 
ε zc 
(3-3-14)
Nc N1 N2
N ' zc = ∑ k i ε i + ∑ k i ε i + ∑ k i ε i
i i i
Nc N1 N2
= ∑ k i (ε zc + xiφ yc ) + ∑ k i (ε zc − y iφ x1 + xiφ yc ) + ∑ k i (ε zc − y iφ x 2 + xiφ yc )
i i i
φ yc 
Nc + N 1+ N 2 φ 
 Nc + N 1+ N 2 N1 N2
 
=  ∑ k i xi − ∑ k i yi − ∑ k i yi ∑i k i  φ x1 
 i i i  x2 
ε zc 
 
(3-3-15)
where, Nc, N1 and N2 are the number of vertical springs in a wall panel, side column 1 and side column 2,
respectively.
M ' x1 , φ x1 M 'x2 , φx2
side column 1 side column 2
Figure 3-3-8 Equilibrium condition in the out of wall direction
165
In the out of wall direction, we establish the equilibrium condition for each side column independently. The
equilibrium condition for the side column 1 is,
N1
M ' x1 = −∑ k i ε i y i
i
N1
= −∑ k i (ε zc − y iφ x1 + xiφ yc ) y i
i
φ yc 
 
 N1 N1 N1
 φ x1 
= − ∑ k i xi y i ∑k y i
2
i 0 − ∑ k i yi   
 i i i  φ x 2 
ε zc 
(3-3-16)
Also, for the side column 2,
N2
M ' x 2 = −∑ k i ε i y i
i
N2
= −∑ k i (ε zc − y iφ x1 + xiφ yc ) y i
i
φ yc 
 
 N2 N2 N2
 φ x1 
= − ∑ k i xi y i 0 ∑k y i
2
i − ∑ k i yi   
 i i i  φ x 2 
ε zc 
(3-3-17)
In a matrix form
 Nc + N 1+ N 2 2 N1 N2 Nc + N 1+ N 2

 ∑ k i xi − ∑ k i xi y i − ∑ k i xi y i ∑k x  i i
 M ' yc   N 1
i i
N1
i i
N1  φ  φ yc 
 M '  − ∑ k i xi y i − ∑ k i yi   
yc
 x1   i ∑k y i
2
i 0
 φ x1 
φ 
  =  N2
i
N2 N2
i
[ ]  x1 
 φ  = k p φ 
M ' x 2   − ∑ k i xi y i 0 ∑k y i
2
i − ∑ k i yi   x 2   x2 
 N ' zc   i i i  
 ε zc  ε zc 
 Nc + N 1+ N 2 N1 N2 Nc + N 1+ N 2

 ∑ k i xi − ∑ k i yi − ∑ k i yi ∑i k i 
 i i i
(3-3-18)
Therefore
φ yc   M ' yc   M ' yc 
φ  M '  M ' 
 x1 
  = kp [ ] −1 

x1 
 = f [ ]
 x1 
p   (3-3-19)
φ x 2  M ' x 2  M ' x 2 
ε zc   N ' zc   N ' zc 
For both ends
166
φ  M ' 
 xA1   xA1 
φ xA2   M ' xA2 
 
[ ]
ε zAc   f pA

0   N ' zAc 

=

φ yBc   0 [ ]  
f pB   M ' yBc 
(3-3-20)
φ xB1   M ' xB1 

   
φ xB 2  M ' xB 2 
ε   N' 
 zBc   zBc 
167
section as shown in Figure 3-3-6. This model is based on the concept of “Multi-spring model” and
modified for the wall element by Saito et.al. The vertical springs in the side columns are determined
independently in the same way as the Multi-spring models of columns. The wall panel section is devided in
5 areas, and a steel springs and a concrete spring are arranged at the center of each area.
l w1
lw
lw2
(a) Original column section

1 2 6 7
5 x 10
11 12 13 14 15
3 4 8 9
y Concrete spring
Steel spring
(b) Multi-spring model
(tension) (tension)
s fy
−c d y
s dy
(compression) (compression)
(c) Hysteresis of steel spring (d) Hysteresis of concrete spring
168
Strength of steel spring in wall panel
The strength of the steel spring in the wall panel is one-fifth of total strength of rebars in the section,
a wσ wy
s fy = (3-3-21)
5
where,
aw : Total area of vertical rebar in the wall panel
σ wy : Strength of rebar in the wall panel
Strength of concrete spring in wall panel
The strength of the concrete spring in the wall panel is one-fifth of total strength of concrete in the section,
0.85 A pσ B
c fy = (3-3-22)
5
where,
Ap : Total area of wall panel section
σB ： Compression strength of concrete
Yield displacement of vertical spring in wall panel

The yield displacements of steel and concrete springs in the wall panel are assumed to be the same as those
of the springs in the side columns.
d) Nonlinear shear spring

There are three nonlinear shear springs in x direction in wall panel and y direction in side columns.
Hysteresis model of the nonlinear shear springs is the same as that in the beam element in Figure 3-1-4.
Yield shear force
 0.053 pt 0.23 (σ B + 18) 

Qy =  + 0.85 p w ⋅ σ wy + 0.1σ 0 b ⋅ j (3-3-23)
 M /(QD) + 0.12 
where,

M/(QD) : ≈ Shear span-to-depth ratio (= l /( 2 D ) )
σ0 : Axial stress of the column
169
Crack shear force
Qy
Qc = (3-3-24)
3
The crack shear force is, Qu , is assumed as,
Qu = Qc (3-3-25)

Qc
sc = γ c l , γc = (3-3-26)
GA
1
sy = γ yl , γ y = (3-3-27)
250
1
su = γ u l , γ u = (3-3-28)
100
Q y1 Q xc Qy2
− Q y1 − Q xc − Qy2
Figure 3-3-10 Nonlinear shear springs in the wall
170
Q Q
k y 3 = 0.001k 0
Qy
k y3
Qu
Qc s
k0 = GA l
k0
sc sy su s
Figure 3-3-11 Force–deformation relationship of shear spring
171
e) Modification of initial stiffness of nonlinear springs
The same modification can be done for the nonlinear springs of wall element as described for those of
beam and column elements by reducing the initial stiffness of the nonlinear spring and increasing the
stiffness of the elastic element as shown in the following figure:
−M
φ τ
θ
θ
φ τ
A
Moment distribution
M
M M M
My My
2 EI
Mc k0 = = + Mc
l
k0 k y = α y k0 k0 kp ≈ ∞
Increase Reduce
stiffness stiffness
M M
My
+ Mc
k *
0
k *p
τ φ c* φ y* φ
172
Ei Ai
k 0i = (3-3-29)
pz
plastic zone. When p z → 0 , it represents the infinite stiffness for rigid condition.
In the same manner of beam and column elements, introducing the flexibility reduction factors,
γ 0 (< 0), γ 1 (< 0), γ 2 (< 0) , the flexibility matrix of the elastic element is,
 l' l' 
γ 1 3EI −
6 EI c 
 c

 l' 
γ2
 3EI c 
 l' l' 
 γ1 − 
 3EI 1 6 EI 1 
 l' 
[ fW ] =  γ2  (3-3-26)
 3EI 1 
 l' l' 
γ1 −
 3EI 2 6 EI 2 
 l' 
 sym. γ2 
 3EI 2 
 l' 
 γ0 
 EAc 
l'
Also, adopting p z = as discussed for beam and column elements, the reduction factors will be:
10
γ 1 = γ 2 = 0.7 , γ 0 = 0.8 (3-3-27)
173
f) Reduction factor of shear stiffness
If shear cracking occurs in the reinforced concrete wall, the shear stiffness decreases. The following graph
shows the test results of the relationship between the stiffness reduction factor β and the lateral drift
−3
angle R × 10 (referred from “Standard for Structural Calculation of Reinforced Concrete Structure”,
Architectural Institute of Japan).
For example, if the lateral drift angle is over than 1/1000, the reduction factor becomes less than 0.2.
Therefore, STERA_3D assumes the “Reduction Factor for Stiffness” is 0.2 in the default setting for the
option of the RC wall element.
174
3.3.1 Direct Wall
Direct Wall identifies the force-displacement points in the back-bone curves of the nonlinear shear spring
and the nonlinear bending spring.
M ' yBc
h X
M ' yAc
Different types of hysteresis model are prepared for the force-deformation relationship of the spring.
qi qi
ky ky
fy fy
fc k1 fc k1
k0 ui k0 ui
(a) Normal-trilinear (b) Degrading-trilinear
Figure 3-3-14 Hysteresis model of the shear and bending springs
175
3.3.2 Steel Wall (Brace)
a) Buckling of brace
3 B 4
N', δ '
Z
h
1 2 X
A
w Y
Figure 3-3-15 Element model for brace
Under the compression load, the stress of buckling failure is calculated theoretically as
π 2E
σE = 2 ,
λ
L
where λ = : slenderness ratio
i
If σ E > σ y (strength of steel), the compression failure will occur before buckling.
Figure 3-3-16 Relationship between buckling stress and slenderness ratio
176
The AIJ (Architectural Institute of Japan) guideline adopts the following equation for the stress of buckling.
σ cr= {1 − 0.4 ( λ λ ) }σp

2
y , for λ ≤ λ p (3-3-30)
0.6
σ cr = σy, for λ > λ p (3-3-31)
(λ λ ) p
π 2E
where λp = : Critical slenderness ratio
0.6σ y
b) Hysteresis model
The hysteresis model proposed by Wakabayashi et. al. is adopted in STERA_3D (hereinafter referred to as
Wakabayashi model). The model consists of four Stages A, B, C and D.
σ [N/mm2] σ [N/mm2]
300.00 300.00
Stage A
250.00 250.00
200.00 200.00
150.00 λ=60 λ=60
150.00
100.00 100.00
50.00 50.00 Stage D
0.00 0.00
-50.00 -50.00
-100.00 -100.00
-150.00 -150.00
-200.00 -200.00
-250.00
ε -250.00 ε
-0.004 -0.002 0.000 0.002 0.004 -0.004 -0.002 0.000 0.002 0.004
Stage A: tension failure with constant strength Stage D: unloading stage
σ [N/mm2] σ [N/mm2]
300.00 300.00
250.00 250.00
200.00 200.00
λ=60 150.00 λ=60
150.00
Stage C
100.00 100.00
Stage D
50.00 50.00
0.00 0.00
-50.00 -50.00
-100.00 -100.00
-150.00 Stage B -150.00
-200.00 -200.00
-250.00
ε -250.00 ε
-0.004 -0.002 0.000 0.002 0.004 -0.004 -0.002 0.000 0.002 0.004
Stage B: buckling failure and strength reduction Stage C: tension stage after buckling
177
The compression curve (Stage B) and the tension curve (Stage C) are defined using the nondimensional
strength and deformation as,
n = N / N 0 : nondimensional strength
δ =∆ / ∆ 0 : nondimensional deformation
where N : axial load, N 0 = Aσ y : axial strength (A: area, σ y : yielding stress of steel)
∆ : displacement, ∆=
0 Lε=
y Lσ y E : yield deformation
Both curves are assumed to be the following form
n 1 ( aδ + b )
r
=
where a, b : parameters of the function of nondimensional Euler load

= nE σ=
E /σy π 2 E ( λ 2σ y )
b-1) Compression Curve

Compression curve (Stage B) is defined from the following empirical formula,
n 1 ( p1δ + p2 )
12
=
10 nE − 1
where p1
= , = p2 4 nE + 0.6
3
Compression strength nc is also on this curve, therefore,
nc 1 ( p1δ c + p2 )
12
=
or p1nc 2δ c + p2 nc − 1 =0
Nc E ( ∆c / L ) A ∆c
Since nc
= = = = δ
N0 E (∆ / L) A ∆
Finally nc is obtained by solving
p1nc 3 + p2 nc − 1 =0
b-2) Tension Curve

Tension curve (Stage C) is defined from the following empirical formula,
n 1 ( p3δ + 1)
32
=
1
where p3 =
3.1 nE + 1.4
178
b-3) Movement of Tension Curve
Movement of tension curve x is defined as follows:
=x ln ( q1δ a + 1) − q2 s δa s x
3 − 1 nE
where
= q1 = , q2 0.115 nE + 0.36
10
b-4) Movement of Compression Curve

Movement of compression curve y is defined to satisfy the following relationship
y δ
= b δb0
y0 δ b 0
δb
y
y0
179
b-5) Movement of Compression Curve
The point shifting from the unloading Stage D to Stage C is obtained by assuming that the
plastic tension deformation δ t is proportional to the plastic compression deformation δ c as
δ t = q3δ c
δt
where
= q3 0.3 nE + 0.24
δc
Example
λ = 60
Starting from compression Starting from tension

λ = 120
Starting from compression Starting from tension
References
M. Shibata, T. Nakayama and M. Wakabayashi, "Mathematical Expression of Hysteretic Behavior of
Braces", Research Report, Architectural Institute of Japan, No. 316, pp.18-24, 1982.6 (in Japanese)
180
3.3.3 SRC Wall (Brace)
Figure 3-3-17 Element model for SRC wall (RC wall with steel brace)
b) Nonlinear shear spring
Yield shear force
Qy = Qy , RC + Qy , S (3-3-32)
where
Qy , RC : Yield shear force of reinforced concrete
 0.053 pt 0.23 (σ B + 18) 

Qy , RC =  + 0.85 pw ⋅ σ wy + 0.1σ 0 b ⋅ j (3-3-33)
 M /(QD) + 0.12 
Qy , S : Yield shear force of steel
Qy , S = AS σ y , S cos R (3-3-34)
where,
AS : Area of steel (mm2)
σ y, S : Strength of steel (N/mm2)
R : Angle of steel
181
3.4 External Spring
δ zB
δ yA Z
B
A B A
δ xA δ xB X
B
Y
δ yB A
δ zA
Figure 3-4-1 Element model for external spring
3.4.1 Lift up spring
N
tension
δ
K0
compression
ground
Figure 3-4-2 Hysteresis model of the external spring
In STERA_3D, if there is no building element at one end of the external spring, this end is considered fixed. Such
spring is used to express the stiffness the ground attached to the building. In such a case, as the relationship
between axial force and deformation of the spring, the linear stiffness is defined only in compression side and zero
stiffness in the tension side as shown in Figure 3-4-2, assuming that the building detaches from the ground.
182
3.4.2 Air spring
Reference:
1) Marin Presthus, “Derivation of Air Spring Model Parameters for Train Simulation”, Master of Science
Programme, Department of Applied Physics and Mechanical Engineering, Luleå University of Technology,
Sweden, 2002
Figure 3-4-3 Air spring ( V : volume, p : relative pressure, A : area)
An effective area Ae is introduced to express the volume change of air bag ∆Vb as
∆Vb =
Ae z (3-4-1)
When the initial pressure of air spring is p0 , after the deflection, the pressure will change as
p=
b p0 + ∆pb for air bag (3-4-2a)
p=
r p0 + ∆pr for reservoir (3-4-2b)
The volume will also change as
Vb = Vb 0 − zAe + zs As for air bag (3-4-3a)
V=
r Vr 0 − zs As for reservoir (3-4-3b)
where
zs : the movement of air mass through orifice
As : area of surge pipe
The pressure and the volume of the isentropic process can be described by
p1 ⋅ V1n = p2 ⋅ V2n (3-4-4)
where
p1 , V1 : initial pressure and volume
p2 , V2 : final pressure and volume
183
n : ratio of specific heat = 1.4 for Air
Applying the above equation to the air bag
( p0 + ∆pb ) ⋅ (Vb 0 − zAe + zs As )=

n
p0 ⋅ Vbn0 (3-4-5a)
n
 − zA + z A 
( p0 + ∆pb ) ⋅ 1 + e s s  = p0 (3-4-5b)
 Vb 0 
by using Taylor expansion (1 + x ) ≈ 1 + nx

n
( x  1)
 ∆pb   n ( − zAe + zs As ) 
1 +  ⋅ 1 + =1 (3-4-5c)
 p0   Vb 0 
 ∆pb   − zAe + zs As 
Assuming   ≈0
 p0   Vb 0 
∆pb n ( zAe − zs As )
≈ (3-4-5d)
p0 Vb 0
Using the same procedure for the reservoir
( p0 + ∆pr ) ⋅ (Vr 0 − zs As )=
n
p0 ⋅ Vrn0 (3-4-6a)
∆pr nzs As
≈ (3-4-6b)
p0 Vr 0
From the Bernoulli equation, the difference of the pressure between the left and right of the pipe speeds up a
portion of gas through the orifice. The force balance in the pipe is given by
As ( ∆pb − ∆pr ) = Cs zs β (3-4-7a)
where
β : viscous damping parameter determined by experiment
Substituting Eq. (3-4-5d) and (3-4-6b),
 zA − z A z A 
p0 As n  e s s − s s =Cs zs β (3-4-7b)
 Vb 0 Vr 0 
np0 As Ae  Vb 0 As  1 1  
z−  + Cs zs β
 zs  = (3-4-7c)
Vb 0  Ae  Vb 0 Vr 0  
184
The force balance for the piston can be expressed as
=Fz Ae ( pb − patm ) (3-4-8)

where
patm : atmospheric pressure
Substituting Eq. (3-4-2a),
( p0 + ∆pb − patm ) Ae
F
=z
∆pb Ae + ( p0 − patm ) Ae
=
n ( zAe − zs As )
= p0 Ae + ( p0 − patm ) Ae
Vb 0
np0 Ae 2  As 
=  z − zs  + ( p0 − patm ) Ae (3-4-9)
Vb 0  Ae 
From Eq. (3-4-7c)
np0 As Ae  As  Vb 0 + Vr 0  
 z −  Cs zs β
 zs  = (3-4-10)
Vb 0  Ae  Vr 0  
Vr 0
λ=
Vb 0 + Vr 0
np0 Ae 2  As  Ae
z− zs  = Cs zs β (3-4-11)
Vb 0  Ae λ  As
np0 Ae 2 λ  1 As 
=Fz  z− zs  + ( p0 − patm ) Ae (3-4-12)
Vb 0  λ Ae λ 
np0 Ae 2 λ np0 Ae 2 Vr 0 V np0 Ae 2

=Kv = = Ke r 0 , Ke =
Vb 0 Vb 0 + Vr 0 Vb 0 Vb 0 Vb 0 + Vr 0
A
Introducing a new variable y = s zs
Ae λ
β β
A  Aλ A  Aλ
K v ( z − y ) = λ e  e  Cs y β = Cβ ⋅ y β , Cβ = λ e  e  Cs (3-4-13)
As  As  As  As 
1  1 
Fz = K v  z − y  + ( p0 − patm ) Ae = K v ( z − y ) + K v  − 1 z + ( p0 − patm ) Ae (3-4-14)
λ  λ 
Therefore Fz
K v ( z − y ) = Cβ ⋅ y β (3-4-15)
Kv z
Ke
F=
z K v ( z − y ) + K e z + ( p0 − patm ) Ae (3-4-16) Cβ y
185
Incremental form of equation is
Fz ( n +1)= K v ( z( n +1) − y( n +1) ) + K e z( n +1) (3-4-17)
z( n +=
1) z( n ) + ∆t ⋅ z ( tn +1 )
y( n +=
1) y( n ) + ∆t ⋅ y ( tn +1 )
Then
β
 y( n +1) − y( n ) 
Cβ  =  K v ( z( n +1) − y( n +1) ) (3-4-18)
 ∆ t 
The solution of Eq. (3-4-18) is obtained by solving the following equation:

β
 y( n +1) − y( n ) 
f ( y( n +1) ) Cβ 
=  − K v ( z( n +=
1) − y( n +1) ) 0 (3-4-19)
 ∆t 
Its derivative regarding y( n +1) is
β −1
β Cβ  y( n +1) − y( n ) 
=f ' ( y( n +1) )   + Kv (3-4-20)
∆t  ∆t 
A Newton-Raphson method is applied to solve the nonlinear equation f ( y( n +1) ) = 0
new
f ( y( n +1) )
y= y( n +1) old −
f ' ( y( n +1) )
( n +1)
where the prime f ' ( y( n +1) ) denotes derivative with respect to y( n +1) ,
f ( y)
y
y( n +1) new y( n +1) old
186
3.5 Base Isolation
The element model of base isolation consists of shear springs arranged in x-y plane changing its direction with
equal angle interval as shown in Figure 3-5-1. This model is called MSS (Multi-Shear Spring) model developed by
Wada et al.
x x qi , u i
y θi
x
Figure 3-5-1 Element model of base isolation
a) Nonlinear shear spring
The hysteresis model of each nonlinear shear spring is defined as a bi-linear model as shown in Figure 3-5-2. The
force and displacement vectors of i-th shear spring are expressed as,
qi , x  cos θ i 
 =  qi (3-5-1)
qi , y   sin θ i 
u x 
u i = [cos θ isin θ i ]  (3-5-2)
u y 
From the relationship, qi = k i u i , the constitutive equation of i-th shear spring is,
qi , x  cos θ i  u x   cos 2 θ i cos θ i sin θ i  u x 

  = ki  [cos θ i sin θ i ]  =    (3-5-3)
q i , y   sin θ i  u y  cos θ i sin θ i sin 2 θ i  u y 
qi
fy ky qi , y
qi
qi , x
k0 θi
ui x
dy
Figure 3-5-2 Hysteresis model of the shear spring
187
From the sum of all nonlinear shear springs in the element, the constitutive equation of the base isolation element
is,
Q x   N  cos 2 θ i cos θ i sin θ i  u x 

  =  ∑ k i     (3-5-4)
Q y   i =1 cos θ i sin θ i sin 2 θ i  u y 
where, N is the number of shear springs in an element. In STERA_3D, N=6 is selected.
First and second stiffness
We assume that all nonlinear shear springs in an element have the same stiffness and strength. The initial stiffness
of the base isolation element, K 0 , is obtained from Equation (3-5-4) by substituting u x = 1, u y = 0 .
 N 
K 0 =  ∑ cos 2 θ i k 0 (3-5-5)
 i =1 
Therefore, the initial stiffness of each shear spring is,
K0
k0 = N
(3-5-6)
∑ cos
i =1
2
θi
The same relationship is established for the second stiffness after yielding,
Ky
ky = N
(3-5-7)
∑ cos
i =1
2
θi
where, K y and k y are the second stiffness after yielding for the base isolation element and the nonlinear shear
spring, respectively.
Yield shear force

The yield shear force of the base isolation element, Q y , is obtained assuming that all the nonlinear shear springs
reach their yielding points except the spring perpendicular to the loading direction, and the increase of the force
after yielding is negligible (Figure 3-5-3). That is,
 N 
Q y =  ∑ cos θ i  f y (3-5-8)
 i =1 
Therefore, the yield shear force of each shear spring is,
Qy
fy = N
(3-5-9)
∑ cosθ
i =1
i
188
fy
fy
Qy θi fy
fy
fy
Figure 3-5-3 Assumption of yield shear force
189
Appendix 3.5:
A-1. Hysteresis of LRB (Lead Rubber Bearing)
LRB (Lead Rubber Bearing) is composed by rubber layers, steel plates and a lead plug core.
Figure A1-1. Lead Rubber Bearing (from Bridgestone Catalog)
1) Bi-Linear Model
The bi-linear hysteresis of LRB is defined as a combination of an elastic model and elasto-plastic model as shown
Figure A1-2.
ELASTIC PLASTIC ELASTO – PLASTIC
K2
Fy
Qd
K1
+ Kp = Keq
Qo Kr
Dy Dy Dy
Qd
RUBBER LEAD
Figure A1-2. Bi-linear model
The elastic stiffness, Kr, from the rubber is calculated as,

Ar
K r = Gr (A1-1)
Hr
190
where Gr is the shear modulus of the rubber, Ar is the cross section area of the rubber and Hr is the total height of
the rubber.
The elastic stiffness, Kp, from the lead plug is calculated as,
Ap
K p = Gp (A1-2)
Hp
where Gp is the shear modulus of lead, Ap is the cross section area of lead plug and Hp is the total height of the plug.
The initial elastic stiffness, K1, and the secondary stiffness, K2, of the bi-linear model are then obtained as,
K1 = K r + K p
(A1-3)
K2 = Kr
The yielding deformation, Dy, is determined from the characteristics of the lead plug. The yielding force, Fy, is
calculated as,
F y = (K r + K p )D y (A1-4)
2) Modified Bi-linear Model
Hysteresis of a lead rubber bearing has a characteristic of stiffness degrading according to the strain level as shown
in Figure A1-3.
Force, F Skelton curve
Deformation, δ
Figure A1-3. Hysteresis of a lead rubber bearing
The secondary stiffness of a lead rubber bearing, Kd, is expressed as,
K d (γ ) = C Kd (γ )(K r + K p ) (A1-5)
where γ is a strain ratio ( γ = δ / H r ) and C Kd (γ ) is a modification factor of the secondary stiffness, which
takes into consideration the strain dependency. Also, the intercept force is defined as,
191
Qd (γ ) = C Qd (γ )σ p A p (A1-6)
where C Qd (γ ) is a modification factor of the yielding shear force and σ p is the yielding shear stress of lead.
The force is then expressed by:
F (γ ) = K d (γ ) δ + Qd (γ ) (A1-7)
Figure A1-4 Hysteresis loop model of lead rubber bearing
The modification factors, C Kd (γ ) and C Qd (γ ) , are represented by the following formulas under 15 degrees
Celsius.
0.779γ −0.43 , γ < 0.25


C Kd (γ ) =  γ − 0.25 , 0.25 ≤ γ < 1.0 (A1-8)
 γ − 0 . 12
, 1.0 ≤ γ < 2.5

 2.036γ 0.41 , γ < 0.1


C Qd (γ ) = 1.106γ 0.145 , 0.1 ≤ γ < 0.5 (A1-9)
 1 , 0.5 ≤ γ

C Kd (γ ) C Qd (γ ) =
6 1.2
5 1
4 0.8
3 0.6
2 0.4
1 0.2
γ γ
0 0
0 1 2 3 0 1 2 3
Figure A1-5. Modification factors
192
Under a different temperature, t, the secondary stiffness and the yielding shear force are to be corrected by the
following formulas:
K d (t ) = K d (t 0 ) exp(− 0.00271 (t − t 0 )) (A1-10)

Qd (t ) = Qd (t 0 ) exp(− 0.00879 (t − t 0 )) (A1-11)
where t0 = 15 degree Celsius.
The primary stiffness of the lead rubber bearing, Ku, in Figure A1-4 is determined from the secondary stiffness, Kd,
as,
Ku = β Kd (A1-12)
where 10 ≤ β ≤ 15 .
Following the suggestion in the manual of CANNY (K. Li, 2004), the hysteresis rules are:
a) Elastic range
Under the strain level less than γe, the hysteresis is assumed to be linear with the secant stiffness at the strain, γe,
that is:
K 0 = Fe / γ e (A1-13)
Fe = K d (γ e ) δ e + Qd (γ e ), δ e = γ e H r (A1-14)
The value, γ e = 0.01 , is adopted in STERA3D.
b) Loading on the skeleton curve after elastic range
Under the loading on the skeleton curve after elastic range, tangent stiffness is used to estimate the response at
the next step:
K = dF (γ ) / dγ (A1-15)
Reference:
Response Control and Seismic Isolation of Buildings, Edited by Masahiko Higashino and Shin Okamoto, SPON
PRESS, October 17, 2006.
Canny Technical Manual, Kangning Li, August 2004
193
2) Consideration of strength reduction by dissipated energy
Reference
1) Masanori Iiba, et.al., “Research on Characteristics of Isolators and Dampers under Multi-cyclic Earthquake Motions
and Effects on Response of Seismically Isolated Buildings”, Building Research Institute, National Research and
Development Agency, Building Research Data, No. 170, April 2016 (in Japanese).
2) Haruyuki Kitamura and Miyuki Omiya, “Design method for long period ground motion - Points to note when
dealing with long-period ground motion”, The Kenchiku Gijyutsu, No. 815, pp.116-125, 2017.12 (in Japanese)
From Reference 1), the yield shear stress of lead plug, τ , is expressed as
τ= {
τ 0 1 − (T TL )
αT
}, 0.4 + 0.25 (T TL )
αT = (A1-16)
Where,
τ 0 : Design value of the yield shear stress of lead plug = 15.0 (N/mm2)
T : Average temperature of lead plug
TL : Melting point of lead plug = 327.5 (oC)
For example, when T =20 (oC), τ is calculated to be 10.3 (N/mm2).
Reference 2) suggested another formula as
Qd ' ( γ ) = µ Qd ( γ ) (A1-17)
 1 W pb 
µ=
−0.06 + 1.25exp  −  (A1-18)
 360 V pb
 
where
Qd ( γ ) : Intercept force without reduction
µ : Reduction factor
W pb : Dissipated energy
π
V pb = D pb 2 hpb : Volume of lead plug
4
hpb = ntr + ( n − 1) ts : Height of lead plug
n : number of rubber layer, tr : thickness of rubber layer, ts : thickness of steel plate
Also, the following formula is sometimes used
8.33   1 W pb  
µ= −0.06 + 1.25exp  − f ( D pb )   (A1-19)
7.97   360 V pb  
where
f ( D pb ) = 0.16 D pb 0.31 : Correction value by the diameter of the lead plug, D pb (mm)
194
STERA_3D adopts Equation (A1-19). The reduction factor µ is plotted as a function of energy dissipation as
follows.
Figure A1-6. Strength reduction factor by energy dissipation

Example)
Bridgestone Product: LH060G4_C
Diameter (mm) 600
Lead plug diameter (mm) 100
2 2
Effective area (×10 mm ) 2749
Thickness of one rubber layer (mm) 4
Number of rubber layers 50
Total rubber thickness (mm) 200
Total height (mm) 407.9
Shear modulus of rubber Gr (N/mm2) 0.385
Apparent shear modulus of lead αp (N/mm2) 0.583
Yield shear stress of lead sy (N/mm2) 7.967
(shear properties at shear strain = 100%)
Initial stiffness K1 (×103kN/m) 7.18 (=13×K2)
Post yield stiffness K2 (×103kN/m) 0.552*1)
Characteristic strength Qd (kN) 62.6*1)
*1)
Shear stiffness of laminated rubber Kr = Gr Ar / H (×103kN/m) 0.529
Additional shear stiffness by lead plug : Kp = αp Ap /H (×103kN/m) 0.023
3
Total stiffness K2 = Kr+Kp (×10 kN/m) 0.552
Yield strength of lead Qd = sy Ap (kN) 62.573
195
250 250
200 200
150 150
100 100
50 50
0 0
-30 -20 -10 -50 0 10 20 30 -30 -20 -10 -50 0 10 20 30
-100 -100
-150 -150
-200 -200
-250 -250
(a) Bi-linear (b) Modified bi-linear

Figure A1-7. Comparison of hysteresis loops
196
A-2. Hysteresis of HDRB (High Damping Rubber Bearing)
HDRB (High Damping Rubber Bearing) is composed by rubber layers and steel plates. By adding special
ingredient in the natural rubber, rubber itself demonstrates damping characteristics.
Figure A2-1. High Damping Rubber Bearing (from Bridgestone Catalog)
1) Modified Bi-linear Model
The hysteresis of HRB is defined as a modified bilinear model as shown Figure A2-2.
Figure A2-2. Bi-linear model
The initial stiffness, K 1 , from the rubber is calculated as,
K1 (γ ) = G1 (γ )× A / H r
u (γ ) − π × heq (γ ) / 2 + u (γ )× π × heq (γ ) / 2
G1 (γ ) = × Geq (γ ) (A2-1)
u (γ ) − π × heq (γ ) / 2
197
where γ : shear strain ( γ = δ / H r )
Ar : cross section area of the rubber
Hr : total height of the rubber.
Geq (γ ) : Equivalent shear modulus

Geq (γ ) = α 0 + α1γ + α 2γ 2 + α 3γ 3 +  + α nγ n (A2-2)
heq (γ ) : Equivalent damping factor
heq (γ ) = β 0 + β1γ + β 2γ 2 + β 3γ 3 +  + β nγ n (A2-3)
u (γ ) : Intercept force
u (γ ) = µ 0 + µ1γ + µ 2γ 2 + µ3γ 3 +  + µ nγ n (A2-4)
Example)
Diameter: φ1500
Thickness of rubber: 7.5mm × 20 layers = 150mm
S1 = 49.7
S 2 = 10.0
2
Nominal compression stress: 10 N / mm
Strain γ Coefficient of each order

0 1st 2nd 3rd 4th
G eq (γ ) 0.1 ~ 1.5 1.1503 -2.5382 3.3047 -2.0356 0.4728
( N / mm 2 ) 1.5 ~ 2.5 3.7412 -6.8745 5.1256 -1.6946 0.2092

2.5 ~ 3.0 0.1749 0.0261 0.0071
heq (γ ) 0.1 ~ 1.5 0.135 0.0903 -0.13 0.1067 -0.032
1.5 ~ 2.5 -0.6239 1.5853 -1.1493 0.3627 -0.0427
2.5 ~ 3.0 -0.05016 0.1762 -0.0376
u (γ ) 0.1 ~ 1.5 0.2989
1.5 ~ 3.0 0.3685 -0.0464
198
Hysteresis of a high damping rubber bearing has a characteristic of stiffness degrading according to the strain level
as shown in Figure A2-3.
Force, F Skelton curve
Deformation, δ
Figure A2-3. Hysteresis of a high damping rubber bearing
The secondary stiffness of a lead rubber bearing, K 2 , is expressed as,

K 2 (γ ) = G2 (γ )× A / H r (A2-5)
G2 (γ ) = (1 − u )× Geq (γ ) (A2-6)
The shear force is defined as,

PM (γ ) = K h (γ )× X M (A2-7)
K h (γ ) = Geq (γ )× A / H r (A2-8)
where X M : the maximum deformation
Also intercept force is defined as,
Pd (γ ) = u (γ )× PM (γ ) (A2-9)
The hysteresis rules are:
a) Elastic range
Under the strain level less than γ = 0.01 , the hysteresis is assumed to be linear with the secant stiffness at the
strain, that is:
h (γ
K 0 K=
= 0.01) (A2-10)
b) Loading on the skeleton curve after elastic range

Under the loading on the skeleton curve after elastic range, tangent stiffness is used to estimate the response at
the next step:
K = dQ(γ ) / dγ (A2-11)
199
2) Consideration of strength reduction by dissipated energy
Reference
3) Takuya Nishimura et al., “Practical Response Evaluation Method for Seismic Isolation System against Long Period
Earthquake Motions - Part2- High-Damping Rubber Bearing and Lead Damper”, AIJ Annual Convention,
Architectural Institute of Japan, 2013, pp.767-768 (in Japanese)
In the above reference, the reduction factors of equivalent stiffness and equivalent damping are proposed as,
C K = −0.0073 ⋅ (E V ) + 1.0　　 (
E V ≤ 10.0 N mm 2 ) (A2-12)
C K = −0.0025 ⋅ (E V ) + 0.952　　 (
E V ＞10.0 N mm 2 )
(
C h = −0.0039 ⋅ (E V ) + 1.0　　E V ≤ 10.0 N mm 2 ) (A2-13)
(
C h = −0.0016 ⋅ (E V ) + 0.977 　　E V ＞10.0 N mm 2 )
where E: dissipated energy, V: volume of rubber
To consider the strength reduction by energy dissipation, STERA_3D modifies the equivalent shear modulus and
the equivalent damping factor as,
Geq ( γ ) C=
= K Geq ( γ 0.01) (A2-14)
heq ( γ ) C=
= h heq ( γ 0.01) (A2-15)
200
A-3. Hysteresis of Lead Damper
Reference
1) Takuya Nishimura et al., “Experimental Study on Multi-cyclic Characteristics of Devices for Seismic Isolation
against Long Period Earthquake Motions: Part 7- Lead Damper”, AIJ Annual Convention, Architectural Institute of
Japan, 2011, pp.667-668 (in Japanese)
2) Takuya Nishimura et al., “Study on Multi-cyclic Modeling of Devices and Response Evaluation for Seismic
Isolation against Long Period Earthquake Motions: Part 5-Modeling of Lead Damper and Seismic Response
Analyses”, AIJ Annual Convention, Architectural Institute of Japan, 2012.9, pp.383-384 (in Japanese)
Cast Lead
Steel Flange Plate
Figure A3-1. Lead damper
1.2
① d=100mm d=200mm d=400mm
1
0.8
0.6 ② ③
0.4
0.2
0
④
0 2000 4000 6000 8000 10000
Dissipated energy E (kNm)

Figure A3-2. Relationship between dissipated energy and strength reduction factor
In the above references, from the cyclic loading test of a lead damper with the different horizontal displacement
amplitudes, three line graphs are obtained for the relationship between the dissipated hysteresis energy and the
horizontal strength reduction ratio. The breaking points of the line are proposed as follows to match the test results.
201
a) The first point of strength reduction, (R1 , E1 )
R1 = 1.0
(0.1 ≤ d ≤ 0.4) , 195 ( d ＜0.1), 　892 ( d ＞0.4)　 (kNm)
E1 = −37 + 2322 d 　　 (A3-1)
b) The second point of strength reduction, (R2 , E 2 )

(0.1 ≤ d ≤ 0.4) , 0.680 ( d ＜0.1), 0.860 ( d ＞0.4)
R2 = 0.62 + 0.60 d 　　
E2 　= 2,205 (kNm) (A3-2)
c) The third point of strength reduction, (R3 , E3 )

(0.1 ≤ d ≤ 0.4) , 0.428 ( d ＜0.1), 0.585 ( d ＞0.4)
R3 = 0.375 + 0.525 d 　　
E3 　= 8,000 (kNm) (A3-3)
d) The fourth point of strength reduction, (R4 , E 4 )

R4 = 0
(0.15 ≤ d ≤ 0.4) , 9854 ( d ＜0.15), 　8859 ( d ＞0.4)　 (kNm)
E 4 　= 9683 − 2060 d 　　 (A3-4)
The hysteresis of the lead damper is defined as a bilinear model. To consider the strength reduction by energy
dissipation, STERA_3D adopts the line of d = 0.2 (m) for random amplitude. The strength of a lead damper, Qd ,
is then expressed as,
Qd = R Qd 0 (A3-5)
where, R : Strength reduction factor

Qd : Initial strength of a lead damper
Force, F
Skelton curve
Deformation, δ
202
A-4. Hysteresis of Elastic Sliding Bearing
Reference
1) Shigeo Minewaki et al., “Study on Multi-cyclic Modeling of Devices and Response Evaluation for Seismic Isolation
against Long Period Earthquake Motions : Part 2-Modeling of Low Friction Bearing and Viscous Damper”, AIJ
Annual Convention, Architectural Institute of Japan, 2012, pp.377-378 (in Japanese)
Figure A4-1. Elastic Sliding Bearing
In the above reference, the dynamic friction coefficient changes according to the temperature of the sliding plate as,
µ 0 = −7.5 × 10 −5 ⋅ T + 0.0145 (A4-1)
The change of the friction coefficient is expressed as a function of the increment of temperature as
∆µ = 0.03 ⋅ (∆T + 1)
−0.06
− 0.03 (A4-2)
On the other hand, the increment of temperature has the following relationship with the dissipated energy
E d (kNmm),
0.9
∆T = 0.00019 ⋅ E d (A4-3)
Therefore, the dynamic friction coefficient is obtained from the dissipated energy,
µ = µ 0 (T ) + ∆µ (E d ) (A4-4)
The hysteresis of the elastic sliding bearing is defined as a bilinear model. In STERA_3D, the initial friction
coefficient is temporary assumed as µ 0 = 0.029 from the catalog of a manufacture. The strength reduction by
energy dissipation will be expressed as,
Qd = (µ 0 + ∆µ (E d )) (Qd 0 µ 0 ) (A4-5)
where, Qd 0 : Initial strength of an elastic sliding bearing
203
A-5. Hysteresis of Bouc-Wen Model
1 1
0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
-1 -1
β=0.5, γ=0.5 β=0.1, γ=0.9
Reference
1) Terje Haukaas and Armen Der Kiureghian, “Finite Element Reliability and Sensitivity Methods for
Performance-Based Earthquake Engineering”, PEER 2003/14, APRIL 2004
2) Wen, Y.-K. (1976) “Method for random vibration of hysteretic systems." Journal of Engineering
Mechanics Division, 102(EM2), 249-263.
3) Baber, T. T. and Noori, M. N. (1985). “Random vibration of degrading, pinching systems." Journal
of Engineering Mechanics, 111(8), 1010-1026.
a) Basic formulation
The basic formula of Bouc-Wen model is
f = α k 0 x + (1 − α ) k 0 z (A5-1)
z =
{
Ax − β x z
N −1
z + γ x z
N
}ν (A5-2)
η
where, β , γ , and N are parameters that control the shape of the hysteresis loop, while A, ν , and η
are variables that control the material degradation.
From the yield deformation, δ y , the parameters β , γ are expressed as,
β = β 0 δ yN and γ = γ 0 δ yN (A5-3)
The model can be written as,
A− z
N
{β sgn (xz ) + γ }ν ∂z ∂x
z = x = (A5-4)
η ∂x ∂t
204
This leads to the following expression for the continuum tangent
∂f ∂z A− z
N
{β sgn (xz ) + γ }ν
k= = α k 0 + (1 − α ) k 0 = α k 0 + (1 − α ) k 0 (A5-5)
∂x ∂x η
The evolution of material degradation is governed by the following choice of equations (Baber and
Noori 1985):
A = A0 − δ A e, ν = 1 + δν e, η = 1 + δ η e (A5-6)
where e is defined by the rate equation
e = (1 − α ) k 0 z x (A5-7)
and A0 , δ A , δ ν , and δ η are user-defined parameters.
b) Incremental form for numerical analysis

Incremental form of Eq.(A5-1) is
f ( n +1) = α k 0 x( n +1) + (1 − α ) k 0 z ( n +1) (A5-8)
By a backward Euler solution,
z ( n +1) = z ( n ) + ∆t z (t n +1 )
(A5-9)
x( n +1) = x( n ) + ∆t x (t n +1 )
Applied to Eq. (A5-4),
N
  x( n +1) − x( n )  
A( n +1) − z ( n +1) β sgn  z ( n +1)  + γ  ν ( n +1)
  ∆t   x( n +1) − x( n )
z ( n +1) = z ( n ) + ∆t (A5-10)
η ( n +1) ∆t
where
A( n +1) = A0 − δ A e( n +1) , ν ( n +1) = 1 + δν e( n +1) , η ( n +1) = 1 + δ η e( n +1) (A5-11)
x( n +1) − x( n )
e( n +1) = e( n ) + ∆t (1 − α ) k0 z( n +1)
∆t (A5-12)
= e( n ) + (1 − α ) k0 z( n +1) (x(n +1) − x(n) )
Since
 x( n +1) − x( n ) 
sgn  z ( n +1)  = sgn{(x( n +1) − x( n ) )z ( n +1) } (A5-13)
 ∆t 
205
Φ
f (z ( n +1) ) = z ( n +1) − z ( n ) − (x ( n +1) − x( n ) ) = 0 (A5-14)
η ( n +1)
N
Φ = A( n +1) − z ( n +1) Ψ ν ( n +1) (A5-15)
Ψ = β sgn{(x( n +1) − x( n ) )z ( n +1) }+ γ (A5-16)
A Newton-Raphson method is applied to solve the nonlinear equation f z ( n +1) = 0 , ( )

f (z ( n +1) )
z (new (A5-17)
old
n +1) = z ( n +1) −
f ' (z ( n +1) )
where the prime f ' z ( n +1) ( ) denotes derivative with

respect to z ( n +1) ,
f (z ( n +1) )
z
new
z ( n +1) z old
( n +1)
Evaluation of the function derivatives is summarized below.
(
Original f z ( n +1) ) Function derivatives f ' z ( n +1) ( )
x( n +1) − x( n ) x( n +1) − x( n )
e( n +1) = e( n ) + ∆t (1 − α ) k 0 z ( n +1) e' ( n +1) = ∆t (1 − α ) k 0
∆t ∆t
A( n +1) = A0 − δ A e( n +1) A' ( n +1) = −δ A e' ( n +1)
ν ( n +1) = 1 + δν e( n +1) ν ' ( n +1) = δν e' ( n +1)
η ( n +1) = 1 + δ η e( n +1) η ' ( n +1) = δ η e' ( n +1)
N
Φ = A( n +1) − z ( n +1) Ψ ν ( n +1) Φ ' = A' ( n +1) − N z ( n +1)
N −1
( )
sgn z ( n +1) Ψ ν ( n +1)
N
− z ( n +1) Ψν ' ( n +1)
Φ Φ 'η( n +1) − Φη '( n +1)
f (z ( n +1) ) = z ( n +1) − z ( n ) − (x ( n +1) − x( n ) ) f ' (z( n +1) ) = 1 − (x − x( n ) )
η ( n +1) η 2 ( n +1) ( n +1)
(A5-18)
206
The procedure can now be summarized as follows:
1. While ( z( n+1) − z( n+1) > tol )

new old
(a) Evaluate function
e( n +1) = e( n ) + (1 − α ) k0 z( n +1) (x( n +1) − x( n ) )
A( n +1) = A0 − δ A e( n +1) , ν ( n +1) = 1 + δν e( n +1) , η ( n +1) = 1 + δ η e( n +1)
Ψ = β sgn{(x( n +1) − x( n ) )z ( n +1) }+ γ
N
Φ = A( n +1) − z ( n +1) Ψ ν ( n +1)
Φ
f (z( n +1) ) = z( n +1) − z( n ) − (x( n +1) − x( n ) ) (A5-19)
η( n +1)
(b) Evaluate function derivatives
x( n +1) − x( n )
e' ( n +1) = ∆t (1 − α ) k 0
∆t
A' ( n +1) = −δ A e' ( n +1)
ν ' ( n +1) = δν e' ( n +1)
η ' ( n +1) = δ η e' ( n +1)
sgn (z( n +1) )Ψ ν ( n +1) − z( n +1) Ψν '( n +1)

N −1 N
Φ ' = A'( n +1) − N z( n +1)
Φ 'η( n +1) − Φη '( n +1)

f ' (z( n +1) ) = 1 − (x ( n +1) − x( n ) ) (A5-20)
η 2 ( n +1)
(c) Obtain trial value in the Newton-Raphson scheme
f (z( n+1) )
z(new
n +1) = z ( n +1) − (A5-21)
f ' (z( n+1) )
(d) Update z( n +1)
z(old
n +1) = z ( n +1) and z( n+1) = z(new
n +1) (A5-22)
207
c) Tangent stiffness
The tangent stiffness is necessary to compute the nonlinear structural analysis.
From the incremental forms:
f ( n +1) = α k 0 x( n +1) + (1 − α ) k 0 z ( n +1)
N
  x( n +1) − x( n )  
A( n +1) − z ( n +1) β sgn  z ( n +1)  + γ  ν ( n +1)
  ∆t   x( n +1) − x( n )
z ( n +1) = z ( n ) + ∆t
η ( n +1) ∆t
The tangent stiffness is calculated as (T. Haukaas and A. D. Kiureghian, 2004);
∂f ( n+1) ∂z ( n+1)
k ( n+1) = = α k 0 + (1 − α ) k 0 (A5-23)
∂x( n+1) ∂x( n+1)
∂z( n+1) b4
= (A5-24)
∂x( n+1) b5
where
Ψ = β sgn{(x( n+1) − x( n ) )z ( n+1) }+ γ
N
Φ = A( n +1) − z ( n +1) Ψ ν ( n +1)
b1 = (1 − α ) k 0 z ( n+1)
b2 = (1 − α ) k 0 (x( n+1) − x( n ) )
b3 =
(x ( n +1) − x( n ) )
η ( n+1)
Φ Φ
b4 = −b3δ Ab1 − b3 z( n+1) Ψδν b1 −
N
(x ( n +1) − x( n ) )δη b1 +
η 2
( n +1) η ( n+1)
sgn (z( n+1) )Ψν ( n+1)

N −1
b5 = 1 + b3δ Ab2 + b3 N z( n+1)
Φ
N
+ b3 z( n+1) Ψδν b2 + (x ( n +1) − x( n ) )δη b2
η (2n+1)
208
3.6 Masonry Wall
B1 B B2
N ' z1 , ε ' z1 N 'z 2 , ε 'z 2

l
Q' xc , ε ' xc
A1 A2
A
Figure 3-6-1 Element model for masonry wall
a) Nonlinear shear spring
Hysteresis model of the nonlinear shear spring is defined as the poly-linear slip model as shown in Figure
3-6-2.
Q
Q
Qy
Qc αk 0
γ
Qu
k0
γc γy γu γ
Figure 3-6-2 Hysteresis model of the nonlinear shear spring
The characteristic values, Qc , Q y , Qu are obtained based on the formulation described in the reference
(Paulay and Priestley, 1992).
209
The procedure to obtain the shear strength is shown below:
(1) Compression strength of masonry prism
Compression strength of diagonal strut is

R = Z t f 'm (3-6-1)
where,
f 'm : Compression strength of the masonry prism
Z : Width of the diagonal strut (Z = 0.25 d, d is diagonal length)
t : Thickness of wall
The compression strength of the masonry prism ( f ' m ) is determined by the following equation (Paulay and
Priestley, 1992),
f ' cb ( f 'tb +αf ' j )
f 'm = (3-6-2)
U u ( f 'tb +αf ' cb )
j
α= (3-6-3)
4.1hb
where,
f 'cb : Compressive strength of the brick
f 'tb : Tensile strength of the brick (= 0.1 f ' cb )
f 'j : Compressive strength of the mortar
j : Mortar joint thickness
hb : Height of masonry unit
Uu : Stress non-uniformity coefficient (=1.5)
The shear strength is then obtained as,

Vc = R cos θ = Z t f ' m cos θ (3-6-4)
210
(2) Shear strength by sliding shear failure
The maximum shear stress is obtained from the Mohr-Coulomb criterion:
τ f = τ 0 + µσ 0 = τ 0 + tan φ σ 0 (3-6-5)
where,
τ0 : Cohesive capacity of the mortar beds (=0.04 f ' m ) (Paulay and Priestly, 1992)
µ : Sliding friction coefficient along the bed joint
µ = 0.654 + 0.000515 f ' j (Chen et.al, 2003)
σ0 : Compression stress ( = W / Aw = R sin θ / Aw )
The shear strength is
 W 
V f = τ f AW = τ 0 + µ  AW = τ 0 Aw + µW (3-6-6)
 AW 
Substituting V f = R cos θ , W = R sin θ

where θ is an angle subtended by diagonal strut to horizontal plane
R cos θ = τ 0 Aw + µR sin θ
R cos θ (1 − µ tan θ ) = τ 0 Aw (3-6-7)
τ 0 Aw
R cos θ =
1 − µ tan θ
Therefore,
τ 0 Aw
Vf = (3-6-8)
1 − µ tan θ
211
(3) Characteristic values of nonlinear skeleton
The shear resistance, Q y , is calculated to be the minimum value between the shear strength by sliding
shear failure, V f , and the shear strength of diagonal compression failure, Vc , that is,
Q y = min(V f , Vc ) (3-6-9)
The shear displacement at the maximum resistance, γ y , is obtained as (Madan et al.,1997),
ε 'm d m
γy = (3-6-10)
cos θ
where,
ε 'm : Compression strain at the maximum compression stress
( ε ' m =0.0018, Hossein and Kabeyasawa, 2004)
Initial elastic stiffness is assumed as (Madan et al., 1997)
k 0 = 2Q y / γ y (3-6-11)
From Figure 3-6-2, the shear resistance at crack, Qc , is obtained as,

Q y − αk 0 γ y
Qc = (3-6-12)
1−α
where, α is the stiffness ratio of the second stiffness and assumed to be 0.2.
Shear displacement at crack is then obtained as,

γ c = Qc / k 0 (3-6-13)
Shear resistance and displacement at the ultimate stage are assumed as (Hossein & Kabeyasawa, 2004)
Qu = 0.3Q y (3-6-14)
γ u = 3.5(0.01hm − γ y ) (3-6-15)
where, hm is the height of masonry wall.
References:
1) T. Pauley, M.J.N. Priestley, 1992, Seismic Design of Reinforced Concrete and Masonry building, JOHN
WILEY & SONS, INC.
2) Hossein Mostafaei, Toshimi Kabeyasawa, 2004, Effect of Infill Walls on the Seismic Response of
Reinforced Concrete Buildings Subjected to the 2003 Bam Earthquake Strong Motion : A Case Study of
Bam Telephone Centre, Bulletin Earthquake Research Institute, The university of Tokyo
3) A. Madan,A.M. Reinhorn, ,J. B. Mandar, R.E. Valles, 1997, Modeling of Masonry Infill Panels for
Structural Analysis, Journal of Structural Division, ASCE, Vol.114, No.8, pp.1827-1849
212
b) Vertical springs
For the moment, the vertical springs of the element model in Figure 3-6-1 are assumed to be elastic springs.
N ' z1 = k z ε ' z1 , N 'z2 = k zε 'z2 (3-6-16)
k z = E m (tl w ) / 2 (3-6-17)
where,
Em : Modulus of elasticity of masonry prism (=550 f ' m , FEMA 356, 2000)
t : Thickness of masonry wall
lw : Width of masonry wall
213
3.7 Passive Damper
a) Hysteresis damper
Hysteresis damper is modeled as a shear spring as shown in Figure 3-7-1.
B1 B B2
l
Q' xc , ε ' xc
A1 A2
A
Different types of hysteresis model are prepared for the force-deformation relationship of the spring.
(1) Bi-linear Model

f f f
k2
k2
k1 k1 − k2
x x ＋ x
xy xm =
(2) Normal-trilinear Model

f= f1 + f 2 f1 f2
x x x
= +
214
(3) Degrading Tri-linear model
p My
P
n θm kr
ks θs θm
p θy p p
My
p pθm
= µα , k y
kr k y = = , µ
p θy pθ y
β Mm
= m µ ,
ks k= km n
M
n m
( nθ m − pθ s )
(4) Bouc-Wen model

y y y
A =1 A =1 A =1
α = 0.01 α = 0.01 α = 0.01
x x x
( β 0.5,
= = γ 0.5,
= N 10 ) ( β 0.5,
= = γ 0.5,
= N 2) ( β 0.1,
= = γ 0.9,
= N 2)
(5) Nonlinear Spring model
k2
k1 x
xy xm
Figure 3-7-2 Hysteresis model of the shear spring
215
b) Viscous damper
Viscous damper is modeled as a shear spring as shown in Figure 3-7-3.
B1 B B2
A1 A2
A
(1) Algorithm for oil damper devise
Figure 3-7-4 shows the Maxwell model with an elastic spring with stiffness, K d , and a dashpot with
damping coefficient, C.
Fij, uij
Node i Node j
Fk, uk Fc, uｃ
Figure 3-7-4 Maxwell model
Since the elastic spring and the dashpot are connected in a series,
Fk = Fc = Fij (3-7-1)
where, Fk : force of the elastic spring
Fc : force of the dashpot
Fij : force between i-j nodes
216
The force of the elastic spring, Fk , is obtained as,
Fk = K d uk = K d (uij − uc ) (3-7-2)
where, u k : relative displacement of the elastic spring
u c : relative displacement of the dashpot
uij : relative displacement between i-j nodes
For an oil damper, the force-velocity relationship of the dashpot is defined as shown in Figure 3-7-5.
Fc
relief point
.
uc
Figure 3-7-5 Dashpot element

The force of the dashpot after the relief point is,
Fc = C 2 u c + Qc (3-7-3)
Substituting Equations (3-7-2) and (3-7-3) into (3-7-1)

K d (u ij − u c ) = C 2 u c + Qc (3-7-4)
When the time interval Δt is small enough, the velocity at time t can be expressed as,
∆u c (t )
u c (t ) = (3-7-5)
∆t
∆u c (t ) = u c (t ) − u c (t − ∆t ) (3-7-6)
Substituting above equations into Equation (3-7-4),

( )
K d u ij (t ) − u c (t − ∆t ) − Qc
∆u c (t ) = (3-7-7)
C2
+ Kd
∆t
The algorithm to obtain the force Fij (t ) from u ij (t ) is as follows:

1) Evaluate ∆u c (t ) from Equation (3-7-7)
2) Evaluate u c (t ) from Equation (3-7-6)
3) Evaluate Fij (t ) from Equation (3-7-2)
217
Before the relief point of the dashpot, Equation (3-7-7) will be obtained by changing C2 → C1 , Qc = 0
as
(
K d u ij (t ) − u c (t − ∆t ) )
∆u c (t ) = (3-7-8)
C1
+ Kd
∆t
When the velocity of the dashpot is over the negative relief point, Equation (3-7-7) will be obtained by
changing Qc → −Qc ,
( )
K d u ij (t ) − u c (t − ∆t ) + Qc
∆u c (t ) = (3-7-9)
C2
+ Kd
∆t
In case there is no elastic spring,
Fij, uij
Node i Node j
Fc, uｃ
Figure 3-7-6 Dashpot element without elastic spring
u ij (t ) = u c (t )
Fuj = Fc = C 2 u c + Qc
∆u c (t ) ∆u ij (t )
u c (t ) = =
∆t ∆t
Therefore,
∆u ij (t )
Fij (t ) = C 2 + Qc (3-7-10)
∆t
Before the relief point of the dashpot,

∆u ij (t )
Fij (t ) = C1 (3-7-11)
∆t
When the velocity of the dashpot is over the negative relief point,
∆u ij (t )
Fij (t ) = C 2 − Qc (3-7-12)
∆t
218
(2) Algorithm for viscous damper devise
Figure 3-7-7 shows the Maxwell model with an elastic spring with stiffness, K d , and a dashpot with
damping coefficient, C.
Fij, uij
Node i Node j
Fk, uk Fc, uｃ
Figure 3-7-7 Maxwell model
Since the elastic spring and the dashpot are connected in a series,
Fk = Fc = Fij (3-7-13)
where, Fk : force of the elastic spring
Fc : force of the dashpot
Fij : force between i-j nodes
The force of the elastic spring, Fk , is obtained as,

Fk = K d uk = K d (uij − uc ) (3-7-14)
where, u k : relative displacement of the elastic spring
u c : relative displacement of the dashpot
uij : relative displacement between i-j nodes
For a viscous damper, the force-velocity relationship of the dashpot is defined as shown in Figure 3-7-8,
Figure 3-7-8 Dashpot element

That is,
Fc = C sgn ( uc ( t ) ) uc ( t )
α
(3-7-15)
219
From Equations (3-7-13) and (3-7-14)
Fij (t )
+ u c (t ) = u ij (t ) (3-7-16)
Kd
Taking time differential and substituting Equation (3-7-15) give
1/ α
Fij (t )  Fij (t ) 
Kd
(
+ sgn Fij (t )  )
 C


= u ij (t ) (3-7-17)
 
The numerical integration method, Runge-Kutta Method, can be used to solve the Equation (3-7-17).
In general, the solution of the differential equation, y (t ) = f ( y, t ) , is obtained by Rungu-Kuttta Method as

follows:
1
y n +1 = y n + (k 0 + 2k1 + 2k 2 + k 3 ) (3-7-18)
6
k 0 = f ( y n , t n )∆t
k1 = f ( y n + k 0 / 2, t n + ∆t / 2)∆t
k 2 = f ( y n + k1 / 2, t n + ∆t / 2)∆t
k 3 = f ( y n + k 2 , t n + ∆t )∆t
Equation (3-7-17) can be written as

1/ α
  Fij (t )  

 (
Fij (t ) =  u ij (t ) − sgn Fij (t )  )
 C   d
 K (3-7-19)
    

Applying Runge-Kutta Method gives the following algorithm,
1
Fij (t n +1 ) = Fij (t n ) + (k 0 (t n ) + 2k1 (t n ) + 2k 2 (t n ) + k 3 (t n ) ) (3-7-20)
6
1/ α
  Fij (t n )  

(
k 0 =  u ij (t n ) − sgn Fij (t n )  )
 C   d
 K ∆t
    

1/ α
  Fij (t n ) + k 0 / 2  

(
k1 =  u ij (t n + ∆t / 2) − sgn Fij (t n ) + k 0 / 2 

) C

  d
K ∆t
    

1/ α
  Fij (t n ) + k1 / 2  

(
k 2 =  u ij (t n + ∆t / 2) − sgn Fij (t n ) + k1 / 2 

)C

  d
K ∆t
    

1/ α
  Fij (t n ) + k 2  
 
(
k 3 =  u ij (t n + ∆t ) − sgn Fij (t n ) + k 2 )
 C

  K d ∆t
   
 
220
In this algorithm, it is assumed as,
u ij (t n ) + u ij (t n + ∆t )
u ij (t n + ∆t / 2) = (3-7-21)
2
In case there is no elastic spring,
Fij, uij
Node i Node j
Fc, uｃ
Figure 3-7-6 Dashpot element without elastic spring
u ij (t ) = u c (t ) (3-7-22)
C sgn ( uc ( t ) ) uc ( t )

α
F=
uj F=
c (3-7-23)
∆u c (t ) ∆u ij (t )
u c (t ) = = (3-7-24)
∆t ∆t
Therefore,
α
 ∆uij (t )  ∆uij (t )
Fuj ( t ) = C sgn   (3-7-25)
 ∆t  ∆t
221
3.8 Ground Spring
3.8.1 Soil structure interaction
a) When building and foundation on ground is subjected to an earthquake excitation, the system can be
divided into two parts: b-1) building and foundation with interaction forces and b-2) ground with zero-mass
foundation subjected to the reaction of interaction forces and an earthquake excitation, which can be
divided further into c-1) zero-mass foundation subjected to an earthquake excitation (kinematic
interaction) and c-2) zero-mass foundation subjected to the reaction of interaction forces (inertia
interaction).
b-1) Building and foundation
a) Building and foundation
c-1) Kinematic interaction

Ground
G
Input ground motion b-2) Ground with zero-mass foundation
G
c-2) Inertia interaction
222
In case of c-2), the force-displacement relationship is written as,
 PG   K H K HR  uG 
 =   (3-8-1)
 M G   K HS K R  θG 
where PG , M G are sway and rocking forces corresponding to the interaction forces between the
superstructure (building-foundation) and the ground, uG , θG are sway and rocking displacements. This
stiffness matrix is called “dynamic impedance matrix”.
If we neglect the coupling between sway and rocking degrees of freedom, the dynamic impedance matrix is
evaluated separately from the d-1) sway impedance K H and d-2) rocking impedance K R as follows:
 PG   K H 0  uG 
 =   (3-8-2)
 M G   0 K R  θG 
d-1) Sway
PG , uG
c-2) Inertia interaction

M G , θG P=
G K H ⋅ uG
PG , uG
G
d-2) Rocking
M G , θG
 PG   K H K HR  uG  G
 =  
 M G   K HS K R  θG 
M=
G K R ⋅θG
223
This corresponds to the Sway-Rocking model as shown below:
c-2) Inertia interaction c-2) Sway-Rocking model
M G , θG M G , θG
PG , uG PG , uG
KH
G G
KR
Finally, the soil-structure interaction is

implemented adding the sway and rocking
springs at the bottom of superstructure.
KH
KR
Input ground motion
It is important to note that the input ground motion to an embedded foundation is smaller than the input
ground motion in the free field due to the influence of the embedding of the foundation. This effect is called
“kinematic interaction”.
Free filed Embedded foundation
224
3.8.2 Cone model to calculate the static stiffness
The cone model is proposed by Wolf [1994] for determining the dynamic stiffness of a foundation on the
ground. The foundation is assumed as an equivalent rigid cylinder and only vertically incident shear wave
is considered. In case of the stratified ground, a simplified formulation is proposed by IIba et.al. [2000]
without considering the reflection and refraction coefficients at the boundary of the soil layer to obtain the
static stiffness. The following formulation is adopted in the STERA_3D software.
Reference:
1) John P Wolf, Foundation Vibration Analysis Using Simple Physical Models, Prentice Hall, 1994
2) Iiba M., Miura K and Koyamada K, "Simplified Method for Static Soil Stiffness of Surface Foundation",
Proceedings of AIJ Annual Meeting, 303-304, AIJ, 2000. (in Japanese)
a) Sway spring
Consider a semi-infinite cone whose area increases in the depth direction. First, we show the calculation
method of the horizontal ground spring (sway spring) for the rectangular foundation 2b × 2c (ground
surface foundation or embedded foundation). The equivalent radius of a circle having the same area is
bc
obtained as r0 = 2 .
π
2r0
The forces of the minute portion at the distance z from the apex of the cone are:
・Shear force at the upper surface
∂u
=Q π=
r 2Gγ π r 2G (3-8-3)
∂z
・Shear force at the lower surface
2 2
dQ   dz   ∂  ∂u   dz   ∂u ∂ 2u 
Q+ π  1 +  r  G  u + dz  =
dz = π 1 +  r 2G  + 2 dz 
dz  z   ∂z  ∂z   z   ∂z ∂z 
(3-8-4)
Considering the static case ignoring the inertial force acting on the minute part, from the balancing of
forces,
225
 dQ 
 Q + 0
dz  − Q =
 dz 
2
 dz  2  ∂u ∂ 2u  ∂u
 π 1 +  r G  + 2 dz  − π r G
2
0
=
 z   ∂z ∂z  ∂z

2
dz   ∂u ∂ 2u  ∂u ∂ 2u  dz  dz 2   ∂u ∂ 2u 
 1 + + dz − = dz + 2 + + dz= 0

  
z   ∂z ∂z 2  ∂z ∂z 2  z  z    ∂z ∂z 2 
 
 Ignoring high-order small amount terms
∂ 2u 2 ∂u
+ 0
= (3-8-5)
∂z 2 z ∂z
The solution to this equation can be expressed as follows:

A
u= +B (3-8-6)
z
where A and B as undetermined coefficients.
Assuming that the displacement on the ground surface is U and the displacement at the depth d is 0 as
boundary conditions,
A A
U =+ B, 0 =+ B (3-8-7)
l d
From this, the coefficient A is
A=
(l + d ) l U (3-8-8)
d
Let Q0 be the shear force of the ground surface
∂u  A  l+d 
π r02G
Q0 = = π r02G  − 2  =
−  π r02G U (3-8-9)
∂z  l   ld 
Therefore, the horizontal spring K H on the ground surface is
Q l+d
KH = π r02G
− 0 = (3-8-10)
U ld
Assuming that d is infinite,
π r02G
KH = (3-8-11)
l
The horizontal spring of the circular rigid foundation on semi-infinite uniform ground is obtained
theoretically from the following formula.
8Gr0
KH = (3-8-12)
2 −ν
If the two springs are set to be equal, the distance l from the apex of the cone to the ground surface is
obtained as follows:
226
8Gr0 π r02G
= →=l
( 2 −ν ) π r (3-8-13)
0
2 −ν l 8
In case of the stratified ground, consider a truncated cone of thickness di from the i-th layer of stratified
ground and zi be the coordinate of the bottom of the i-th layer. The radius of the truncated cone ri at
depth zi is then calculated as follows from the geometric relationship.
zi
ri = r0 (3-8-14)
z0
The horizontal spring on the upper surface of this truncated cone is
2
z +d z  zi π r02G1  Gi  zi zi −1
=K π=
r Gi i −1 i π  i −1 r0  Gi=
i 2
  (3-8-15)
zi −1 ( zi − zi −1 ) z0  G1  z0 ( zi − zi −1 )
H i −1
zi −1di  z0 
The horizontal spring K hb at the base bottom position is obtained as a synthetic spring in which
horizontal springs of each layer are connected in series.
1 n −1
1
=∑ i (3-8-16)
K hb i =0 K H
However, in the bottom layer,
π r02G1  Gn  zn zn −1 π r02G1  Gn  zn −1
=K n −1
  →   ( zn → ∞ ) (3-8-17)
 G1  z0 ( zn − zn −1 )
H
z0 z0  G1  z0
Finally, the horizontal ground spring K hb is obtained as,

K hb = β h K1h (3-8-18)
where
1 π r02G1 π r02G1 8π r0G1
=βh , = K = =
n
1
1h
z0 l ( 2 −ν )
∑  
i =1  α i 
G  zi zi −1 G  z
αi =  i  ( i = 1, 2, , n − 1) , α n =  n  n−1
 G1  z0 ( zi − zi −1 )  G1  z0
2 −ν 1
z0 = π r0
8
227
b) Rocking spring
Rotational spring can be obtained as follows, similar to the method for determining horizontal spring. For
the rectangular foundation 2b × 2c (ground surface foundation or embedded foundation, 2b is the
length in rotational direction), the equivalent radius of a circle having the same moment of inertia is
( 2b ) ( 2c ) .
3
obtained as rr 0 = 4
3π
π
) The moment of inertia of a circle Ic = rr40
4
( 2b ) ( 2c )
3
The moment of inertia of a rectangular Ib =

12
2r0
∂M
M+ dz
∂z
The forces of the minute portion at the distance z from the apex of the cone are:
・Moment at the upper surface
∂θ π r 4 ∂θ
M=
− EI E r0
= (3-8-19)
∂z 4 ∂z
・Moment at the lower surface
4 4
dM π   dz   ∂  ∂θ  π  dz  4  ∂θ ∂ 2θ 
M+ −  1 +  rr 0  E  θ +
dz = − 1 +  rr 0 E 
dz  = + 2 dz 
dz 4  z   ∂z  ∂z  4 z   ∂z ∂z 
(3-8-20)
Considering the static case ignoring the inertial force acting on the minute part, from the balancing of
forces,
 dM 
 M + 0
dz  − M =
 dz 
4
dz  4  ∂θ ∂ 2θ  rr40 ∂u
π
 − 1 +  rr 0 E  + 2 dz  + E 0
=
4 z   ∂z ∂z  4 ∂z
 Ignoring high-order small amount terms
∂ 2θ 4 ∂θ
+ 0
= (3-8-21)
∂z 2 z ∂z
The solution to this equation can be expressed as follows:
228
A
θ
= +B (3-8-22)
z3
where A and B as undetermined coefficients.
Assuming that the rotational displacement on the ground surface is Θ and the displacement at the depth d
is 0 as boundary conditions,
A A
Θ= + B, 0= +B (3-8-23)
( lr + d )
3
lr3
From this, the coefficient A is
( lr + d ) lr 3 Θ
3
=A (3-8-24)
( l r + d ) − lr 3
3
Let M 0 be the shear force of the ground surface
3 ( lr + d )
3
π r 4  3 A  π rr40
π rr40
∂u
M0 =
− E − r0 E  − 4  =
= E Θ (3-8-25)
4 ∂z 4  lr  4 {
( lr + d ) − lr3 lr
3
}
Therefore, the rotational spring K R on the ground surface is
3 ( lr + d )
3
M 0 π rr40
K
= = E (3-8-26)
R
Θ 4 { ( lr + d ) − lr3 lr
3
}
Assuming that d is infinite,
3π rr40 E
KR = (3-8-27)
4lr
The horizontal spring of the circular rigid foundation on semi-infinite uniform ground is obtained
theoretically from the following formula.
8Grr30
KR = (3-8-28)
3 (1 −ν )
If the two springs are set to be equal, the distance lr from the apex of the cone to the ground surface is
obtained as follows:
8Grr30
=
3π rr40 E 3π rr40 E
= 2G (1 +ν ) → =
lr
9 1 −ν 2
π rr 0
( ) (3-8-29)
3 (1 −ν ) 4lr 4lr 16
229
In case of the stratified ground, consider a truncated cone of thickness di from the i-th layer of stratified
ground and zri be the coordinate of the bottom of the i-th layer. The radius of the truncated cone rri at
depth zri is then calculated as follows from the geometric relationship.
zri
rri = rr 0 (3-8-30)
zr 0
The rotational spring on the upper surface of this truncated cone is
3 ( zri −1 + di )
3
i π rri4−1 3π rr40 E1  Ei  zri3 zri3 −1
K = Ei   3 3 (3-8-31)
{ } ( )
R
4 ( zri −1 + di ) − zri3 −1 zri −1 4 zr 0  E1  zr 0 zri − zri −1
3 3
The rotational spring K rb at the base bottom position is obtained as a synthetic spring in which rotational
springs of each layer are connected in series.
1 n −1
1
=∑ i (3-8-32)
K rb i =0 K R
However, in the bottom layer,
3π rr40 E1  En  zrn3 zrn3 −1 3π rr40 E1  En  zrn3 −1

K Rn −1   3 3 →   ( zn → ∞ ) (3-8-33)
4 zr 0  E1 (
 zr 0 zrn − zrn −1
3
) 4 zr 0  E1  zr30
Finally, the horizontal ground spring K rb is obtained as,

K rb = β r K1r (3-8-34)
ここに、
1 3π rr40 E1 3π rr40 E1 4 rr30 E1
=βr =, K1r = =
n
 1  4 zr 0 4lr 3 1 −ν 12
∑ 
i =1  α ri


3
E  z3 z3  E  z 
α i =  i  3 ri3 ri −1 3 ( i = 1, 2,, n − 1) , α n =  n   n−1 
(
 E1  zr 0 zri − zri −1 )  E1   z0 
zr 0 =
(
9 1 −ν 12 )πr
r0
16
230
3.8.3 Embedded foundation
K hb , K rb K he , K re
In case of embedded spread foundation, the resistances at the side of the foundation K he , K re can be
expected in addition to the resistances K hb , K rb at the base of the foundation. That is,
K
= h K hb + K he
(3-8-35)
K
= r K rb + K re
where
De Ghe
K he = ξ he K hb (3-8-36)
r0 Ghb
 D  De   Ghe
3
=K re ξ re K rb 2.3 + 0.58   
e
(3-8-37)
 rr 0  rr 0   Ghb
m
∑G H i i
( 2 −ν ) K hb
=Ghe =
i =1
m
, Ghb (3-8-38)
8r0
∑ Hi
i =1
De is the depth of the foundation. ξ he and ξ re are the earth pressure reduction coefficients of
horizontal and rotational directions at the side of the foundation and they are set to 0.5 when considering
only the side receiving the reaction force from ground at the time of the earthquake. m is the number of
soil layers from the surface to the bottom at the side the foundation where the earth pressure acts. ν is the
average Poisson's ratio of the ground under the foundation base. The damping at the embedded part is not
considered.
231
3.8.4 Radiation damping
The static stiffness obtained by the cone model alone can not express the radiation damping that the
energy of ground shaking spreads to a distance.
To evaluate the radiation damping, we consider a semi-infinite earth column with the same area of the
foundation where a shear wave travels downward when the foundation sways harmonically in a horizontal
direction.
The wave travels in the earth column can be expressed as the solution of the wave equation.
∂ 2u 2
2 ∂ u G
= V=
s , Vs (3-8-39)
∂t 2 ∂2 z ρ
where G is the shear modulus of the soil, ρ is the density of the soil, and Vs is the shear wave
velocity.
ipt
When the foundation sways harmonically as ue , the solution of the wave equation is
u ( z , t ) = ue
ip ( t − z /Vs )
(3-8-40)
The shear force at the bottom of the foundation is,

∂u GA GA du du
Q=
−GA z =0 = iupeipt = ( ρVs A)
= (3-8-41)
∂z Vs Vs dt dt
where A is the area of the foundation. Therefore, the damping force by the radiation is equivalent as the
viscous damping of a dashpot with a damping coefficient
CH = ρVs A (3-8-42)
The radiation damping of a rocking motion is expressed as the similar formula

CR = ρ (ηVs ) I (3-8-43)
π r4
where I= is the second moment of inertia for a circular foundation with the radius r
4
3.4
η= is the coefficient for vertical wave velocity, where ν is the Poisson’s ratio
π (1 −ν )
Peipt ipt Peipt

ue
Q
C = ρVA
G, ν , ρ
232
In case of the stratified ground, we can use the following formula for the radiation damping
CH = ρeVe A (3-8-44)
3.4
CR = ρe (ηeVe ) I , ηe = (3-8-45)
π (1 −ν e )
where ρe is the average density, Ve is the average shear wave velocity and ν e is the average shear
modulus defined by the weighted average by depth of layers under the basement as
n n n
∑ ρi d i ∑Vi di ∑ν d i i
ρe = i =1
n
, Ve = i =1
n
, νe = i =1
n
(3-8-46)
∑d
i =1
i ∑d
i =1
i ∑d
i =1
i
3.8.5 Complex stiffness with material damping
The damping effect of the soil material can be considered by setting the shear modulus to the following
complex shear modulus.
G * G (1 + 2ih )
= (3-8-47)
where h is the damping factor of the soil. As a result, the dynamic stiffness obtained from the cone model
becomes also complex value as,
K H * =K H + iK H ' =K H (1 + 2ihH ) : sway spring

K R* =K R + iK R ' =K R (1 + 2ihR ) : rocking spring (3-8-48)
Furthermore, the damping coefficient is obtained from the imaginary part of the complex stiffness under the
periodic vibration of the circular frequency ω.
K ⋅ x + C ⋅ x  assuming x = ae iωt  ( K + iωC ) x

From the equivalent condition,
( K + iωC ) x =
( K + iK ') x =
K (1 + 2ih ) x
K ' 2hK
Therefore, C
= = (3-8-49)
ω ω
STERA_3D calculates the circular frequency ω as

2π
ω1 = (3-8-50)
T1
where T1 is the first natural period of the structure with the ground spring (real part).
233
3.8.6 Impedance matrix
It is known that radiation damping is likely to occur in a frequency band higher than the dominant
frequency of the ground ( f G ), and the effect is greater at higher frequencies. Therefore, the damping is
evaluated separately for a lower frequency side and a higher frequency side than the dominant frequency.
a) In case of f ≤ f G (ω ≤ ωG ) for Sway spring and f ≤ 2 f G (ω ≤ 2ωG ) for Rocking spring
Considering material damping only,

K ' 2hK
=PG K H uG + CH uG , C
= = (3-8-51)
ω ω
M G K R θG + CRθG
= (3-8-52)
where
K H , CH : stiffness and damping of sway spring
K R , CR : stiffness and damping of rocking spring
b) In case of f > f G (ω > ωG ) for Sway spring and f > 2 f G (ω > 2ωG ) for Rocking spring
Considering both material damping and radiation damping,
PG= K H uG + ( CH + CH ') uG (3-8-53)

M = K θ + ( C + C ') θ
G R G R R G (3-8-54)
where
CH ', CR ' : radiation damping for sway and rocking
To avoid the discontinuous of damping, we modify the formula as

f − fG
PG= K H uG + ( CH + ζ H CH ') uG , ζ H= (3-8-55)
f
f − 2 fG
M G= K R θG + ( CR + ζ R CR ')θG , ζ R= (3-8-56)
f
In a matrix form
 PG   K H 0  uG  CH + ζ H CH ' 0  uG 

=     +    (3-8-57)
M G   0 K R  θG   0 CR + ζ R CR ' θG 
234
3.8.7 Pile foundation
Now we discuss the Sway and Rocking springs for the foundation with piles.
1) Sway
PG , uG
G
PG , uG
Inertia interaction
M G , θG
PG , uG
G
2) Rocking
M G , θG
a) Vertical stiffness of a single pile

The vertical stiffness of a single pile is obtained from the follow formula:
KV = EAα
( ) ( ),
k B 1 + e −2α L + EAα 1 − e −2α L
α=
kS
(3-8-58)
kB (1 − e ) + EAα (1 + e )
−2α L −2α L
EA
where,
E : Young’s Modulus of the pile, A : Area of the pile, L : Length of the pile
kS : Vertical spring of the soil surrounding the pile, k B : Vertical spring at the bottom of the pile
235
)
P0
u0
Pz
uz
uz
kS u z dz dz
u z + du z
Pz + dPz
uL
PL
a-1) Equilibrium condition of the vertical forces in a pile
The equilibrium condition of the vertical forces in a small segment is
dP k=
= S u z dz 0 (3-8-59)
The axial strain in the segment is obtained as
du z P
= − z (3-8-60)
dz EA
Therefore
dP d 2u z
= EA= kS uz (3-8-61)
dz dz 2
The solution of this second order differential equation is
kS
u z c1eα z + c2 e −α z , α =
= (3-8-62)
EA
Also
= (
Pz EAα c2 e −α z − c1eα z ) (3-8-63)
Setting the boundary conditions as Pz = P0 at z = 0 and u z = u L at z = L ,
=P0 EAα ( c2 − c1 ) (3-8-64)
uL c1eα L + c2 e −α L
= (3-8-65)
Therefore, the coefficients c1 and c2 are obtained as
236
EAα uL − P0 e −α L EAα uL + P0 eα L
c1 = , c2 = (3-8-66)
(
EAα eα L + e −α L ) (
EAα eα L + e −α L )
The force at the bottom of the pile PL is
=PL EAα c2 e −α L − c1eα L ( ) (3-8-67)
From the relationship uL = PL k B ,
2 K P P0
PL = (3-8-68)
KP e ( αL
+e −α L
) + EAα eα L − e −α L ( )
and
2 P0
uL = (3-8-69)
KP e ( αL
+e −α L
) + EAα eα L − e −α L ( )
The displacement at the head of the pile is
u0= c1 + c2 (3-8-68)
Therefore, the stiffness of the vertical spring at the head of the pile is
K=
P0 EAα ( c2 − c1 )
= =
EAα P0 eα L + P0 e −α L ( )
0
u0 c1 + c2 2 EAα uL − P0 e −α L + P0 eα L
=
(
EAα eα L + e −α L )
2
2 EAα + eα L − e −α L ( )
kB e ( αL
+e −α L
) αL
+ EAα e − e −α L
( ) (3-8-69)
EAα e( αL
+e −α L
){k ( e + e ) + EAα ( e − e )}
B
αL −α L αL −α L
=
(
4 EAα + eα L − e −α L ){k ( e + e ) + EAα ( e − e )}
B
αL −α L αL −α L
= EAα
kB (e αL
+ e −α L ) + EAα ( e − e ) αL −α L
kB (e αL
− e −α L ) + EAα ( e + e ) αL −α L
237
a-2) Vertical spring of the soil surrounding the pile
The vertical spring of the soil surrounding the pile k S is obtained as the friction resistance of soil
surrounding the soil (Randolph and Wroth, 1978).
(a) Concentric cylinder around loaded pile (b) Stresses in soil element
Reference: Randolph M.F and Wroth C.P, “Analysis and deformation of vertically loaded piles”, Journal of
Geotechnical Engineering 104(12): 1465-1487. 1978.
From the equilibrium condition of vertical forces
 ∂τ   ∂σ y   dr   dr 
 τ +  ( r + dr ) d θ dy − τ rd θ dy +  σ y + dy   r +  dθ dr − σ y  r +  dθ dr =
0
 ∂r   ∂y  2   2 
(3-8-70)
Neglecting higher order
∂ (τ r ) ∂σ y
+r 0
= (3-8-71)
∂r ∂y
Assuming the stress change along the depth ∂σ y ∂y is negligible, the second term will be zero. Then,
∂ (τ r )
=0 (3-8-72)
∂r
Integrating from the pile radius r0 to r ,
r
∫ d (τ r ) =⇒
r0
0 τ ( r ) r − τ ( r0 ) r0 =
0
τ ( r0 ) r0 τ 0 r0
τ (r )
= = (3-8-73)
r r
Assuming the deformation along the radius du is smaller than the deformation along the depth dw , the
shear stain is
∂u ∂w dw τ ( r ) τ r
γ= + ≈ = = 00 (3-8-74)
∂z ∂r dr G ( r ) rG ( r )
238
The vertical shear deformation is obtained by integrating from r0 to rm ,
rm 1 τ 0 r0  rm 
= 0 r0 ∫r
wS τ= dr ln   (3-8-75)
0 rG G  r0 
Randolf and Worth proposed the following empirical formula for the radius rm
=rm 2.5 L (1 −ν ) (3-8-76)
The vertical force around the pile is calculated as
 2π G 
=P (=
2π r0 )τ 0   wS (3-8-77)
 ln ( rm r0 ) 
Therefore, the vertical spring of the soil surrounding the pile k S is
2π Ge
kS = , rm 2.5 L (1 −ν e )
= (3-8-78)
ln ( rm r0 )
where,
1 n 1 n
Ge = ∑ ii
L i =1
( G d ) : average shear modulus, ν e = ∑ (ν i di ) : average Poisson ratio
L i =1
a-3) Vertical spring at the bottom of the pile

The vertical spring at the bottom of the pile k B is obtained as a static impedance of circular
foundation as,
3π π GB r0
kB = (3-8-79)
8 (1 −ν B )
where,
GB : shear modulus of the soil at the bottom of the pile
ν B : Poisson ratio of the soil at the bottom of the pile
239
b) Horizontal stiffness of a single pile
b-1) Horizontal stiffness of a single pile
H
x y
p ( x)
The flexural deformation of the infinite pile under horizontal load at the top of the pile is
d4y
EI + p ( x) =
0 (3-8-80)
dx 4
where p ( x ) is the reaction force of the soil.
Assuming
p ( x ) = kh By (3-8-81)
where B is the width of the pile.

The solution is expressed as
y e β x ( A1 sin β x + B1 cos β x ) + e − β x ( C1 sin β x + D1 cos β x )

=
kh B (3-8-82)
β=4
4 EI
Since the deformation in infinite depth is zero, that is x → ∞, y =0 ,
A=
1 B=
1 0 (3-8-83)
In case of fixed pile head,
dy
θ (0) = − β C1 + β D1 =
= 0  C1 = D1 (3-8-84)
dx x =0
The horizontal force at the pile head is
H = −Q ( 0 ) (3-8-85)
Therefore,
Q (0)d3y H H
=
− 3 −4C1β 3 =
= −  C1 = (3-8-86)
EI dx x =0
EI 4 β 3 EI
240
The horizontal deformation of the pile is
H
=y e − β x ( sin β x + cos β x ) (3-8-87)
4 β 3 EI
The deformation of the pile head is
H
y= − (3-8-88)
4 β 3 EI
Therefore, the horizontal stiffness is
( 4 EI ) ( kh B )
1/4 3/4
= β 3 EI
K h 4= (3-8-89)
Francis (1964) proposed the following formula for the horizontal ground spring per unit length of a
single pile:
1/12
1.3ES  ES B 4 
k=
fS k=h B   (3-8-90)
1 −ν S 2  EP I P 
where
EP : Young’s modulus of a pile, I P : Moment of inertia of a pile
ES : Young’s modulus of soil, ν P : Poisson ratio of soil
This formula is based on the study by Biot (1937) with respect to the ground spring against bending of
an infinite beam on ground and is modified by Visic (1961). Francis extended this concept to the pile
considered that there is ground on both sides of the beam and doubled the ground stiffness.
Reference:
1) Francis A. J, Analysis of Pile Groups with Flexural Resistance, Journal of the Soil Mechanics and
Foundations Division, 1964, Vol. 90, Issue 3, Pg. 1-32
2) Biot, M. A. Bending of an infinite beam on an elastic foundation. J. Appl. Mech., 1937, 4, 1, Al-A7
3) Vesic A.B, Bending of beams resting on isotropic elastic solid, Journal of the Engineering Mechanics
Division, 1961, Vol. 87, Issue 2, Pg. 35-54
241
b-2) Horizontal damping of a single pile
Gazetas proposed the following formula for the horizontal damping per unit length of a single pile:
=cgS 2 ρ S B (VL + VS ) (3-8-91)

S wave
where
3.4VS
VL = : Lysmer analog wave
π (1 −ν S )
This damping expresses the radiation damping P wave
in both directions of the pile. Pile
Reference: Gazetas, G. and Dobry, R, Horizontal Response of Piles in Layered Soils, J. Geo tech. Engrg.
Div.,ASCE, Vol.110, pp.20-40, 1984
b-3) Ground spring and damping coefficient between multiple layers

The ground spring and damping coefficient between multiple layers can be calculated by multiplying
the layer thickness of each layer and averaging as
= (
k ' fSi 0.5 k fS ( i −1) H i −1 + k fSi H i ) (3-8-92)
= (
c 'gSi 0.5 cgS ( i −1) H i −1 + cgSi H i ) (3-8-93)
242
c) Impedance of group piles
In case of group piles, the impedance of the foundation can not be obtained from the simple addition
of the impedances of individual piles because of the interaction of piles.
c-1) Group effect in horizontal direction (stiffness)
The horizontal stiffness of group piles is obtained from the horizontal stiffness of a single pile as,
K HG = N P β H K HS (3-8-94)
where
K HG : horizontal stiffness of group piles, K HS : horizontal stiffness of a single pile

N P : number of piles, β H : coefficient of group effect
The following formula is adopted for STERA_3D as the coefficient of group effect for horizontal ( x )
direction,
−0.54
( N y 2)
−0.74( S B )
−0.43
−0.59( S B )
β Hx = 0.4 ( S B ) ( N x 2)
0.3
(3-8-95)
where
S : distance between piles in x-direction, B : diameter of pile,

N x , N y : number of piles in x-direction and y-direction
The horizontal stiffness of a single pile is obtained from Eq. (3-8-89) as
K HS = ( 4 EP I P ) kS 3/4
1/4
(3-8-96)
where
kS : the stiffness coefficient of a single pile under homogenous ground

The horizontal stiffness of group piles
N P β H ( 4 EP I=
P) ( 4 N P EP I P )
1/4 1/4
= P β H K HS
K HG N= k S 3/4 kG 3/4 (3-8-97)
kG = N P β H 4/3 k S (3-8-98)
243
For the horizontal damping, the group effect is assumed negligible, and the horizontal damping of
group piles is obtained as,
cHG = N P cHS (3-8-99)
where
cHG : damping coefficient of group piles, cHS : damping coefficient of a single pile
In evaluating the horizontal ground stiffness of the group pile K HG in layered ground, it is necessary
to determine the value of the stiffness coefficient kG which represents the average stiffness coefficient in
layered ground. The following iterative procedure is used to calculate kG .
P
u0 = u ( 0 )
ui
ui +1
Step. 1 Set the initial value of kG as
kG = average of kGi in the surface layer (< 5B )
where kGi = N P β H
4/3
k Si : horizontal stiffness of group pile at i-th layer from Eq.(3-8-89)
Step. 2 The flexural deformation of a pile under the horizontal load P at the top is approximated by
P kG
=u e − β x ( sin β x + cos β x ) , β=4 (3-8-100)
4 N P EP I P β 3 4 EI
The horizontal stiffness at the top can be calculated by,
K HG 2 =
∑k u
Gi i
(3-8-101)
u0
Step. 3 Update kG as
kG = ( K HG 2 ) ( 4 N P EP I P )
4/3 1/4
(3-8-102)
Step. 4 Go back to Step 1 until K HG 2 ≈ K HG .
244
c-2) Group effect in horizontal direction (damping)
The damping effect of the soil material is considered as
kGi* kGi (1 + 2ihGi )

= (3-8-103)
where hGi is the damping factor of the soil in i-th layer. The horizontal damping at the top of group piles
can be calculated by,
hHG =
∑h u
Gi i
(3-8-104)
u0
Therefore, the imaginary part of the horizontal stiffness is

K HG ' ≈ 2hHG K HG
In the same way, the horizontal radiation damping at the top of group piles can be calculated by
cHG =
∑c u
Gi i
(3-8-105)
u0
where cGi = N p cSi
c-3) Group effect in rocking direction (stiffness)
The group effect in rotational direction is assumed negligible and the coefficient of group effect is
one. Therefore, the rotational stiffness is calculated from the vertical stiffness of individual pile as
m
K RGx = ∑ KVi yi2 : around x-axis (3-8-106)
im
=1
K RGy = ∑ KVi xi2 : around x-axis (3-8-107)
i =1
where
xi , yi : distance from the center of ration in x, y directions
c-4) Group effect in rocking direction (damping)

In case of rocking direction, the damping effect of the soil at the bottom of the pile is considered
dominant.
hRG = hb (3-8-108)
where hb : damping factor of the soil at the bottom of the pile
Therefore, the imaginary part of the rocking stiffness is
K RG ' ≈ 2hRG K RG (3-8-109)
245
3.8.8 Equivalent period and damping factor considering soil structure interaction
a) Equivalent period
δB δS δB δR δS δB
F m F m F
m
K B , CB H
K S , CS
K R , CR
Force and deformation
δB = F KB δ S = F KS θR H ( M KR )
δ R H=
=
δ δ=
= B F KB δ = δ B + δ S = F (1 K B +=
1 KS ) ( FH K R ) F ( K R H 2 )
H=
δ = δB + δS + δR
(
= F 1 KB +1 KS +1 ( KR H 2 ) )
Stiffness
K = KB 1
K= 1
1 KB +1 KS K=
1 KB +1 KS +1 ( KR H 2 )
Period (mass of foundation is ignored)
m m mH 2
TB = 2π TS = 2π TR = 2π
KB KS KR
m m m m m m m mH 2
= π
T 2= TB = π
T 2= 2π + T 2π
= = 2π + +
K K KB KS K KB KS KR
= TB2 + TS2 = TB2 + TS2 + TR2
246
b) Equivalent damping
b-1) Equivalent damping for material damping

Force including damping force is
F Cδ + K δ
=
For a harmonic excitation δ = aeiωt
δ = iωδ
Then
 C
K (1 + 2hi ) δ
K 1 + iω  δ =
F=
 K
where h is the damping ratio
ωC
h=
2K
Defining the viscous damping ratio separately for each dashpot,
ωB CB ω S CS ωR CR
hB = , hS = , hR =
2K B 2KS 2K R
This is the case to define the damping force to be independent to the frequency of excitation.
This type of damping is called “material damping”.
Total complex stiffness will be

1 1 1 1
= + +
K (1 + 2hi ) K B (1 + 2hB i ) K S (1 + 2hS i ) K R H (1 + 2hR i )
2
Using the relationship

1 1 − 2hi 1 − 2hi
= = ≈ 1 − 2hi
1 + 2hi (1 + 2hi )(1 − 2hi ) 1 + 4h2
Then
1 1 1 1
(1 − 2hi ) = (1 − 2hBi ) + (1 − 2hS i ) + (1 − 2hRi )
K KB KS KR H 2
From the real part
1 1 1 1
= + +
K KB KS KR H 2
From the imaginary part
2 2 2
K K K T  T  T 
h= hB + hS + h =  B  hB +  S  hS +  R  hR
2 R
KB KS KR H T  T  T 
247
b-2) Equivalent damping for viscous damping
Force including damping force is
C K 
F = Cδ + K δ = m  δ + δ  = m 2hωδ + ω 2δ
m m 
( )
For a harmonic excitation δ = aeipt
  p 
m (ω 2 + 2hω pi ) δ =
F= mω 2 1 + 2h   i  δ
 ω  
This is the case to define the damping force to be dependent to the frequency of excitation.
This type of damping is called “viscous damping”.
Total complex stiffness will be

1 1 1 1
= + +
  p    p     p     p  
mω 2 1 + 2h   i  mωB 2 1 + 2hB  i  mωS 2 1 + 2hS  i  mωR 2 1 + 2hR  i 
 ω     ωB     ωS     ωR  
Using the relationship
1  p
≈ 1 − 2h   i
 p ω 
1 + 2h   i
ω 
Then
1   p  1   p   1   p   1   p  
2 
1 − 2h   i  = 2 
1 − 2 h   i  + 2 
1 − 2 h   i  + 2 
1 − 2 h   i 
 ω   mωB   
B S R
mω   ωB   mωS   ωS   mωR   ωR  
From the real part

1 1 1 1 1 1 1 1
= + +  = + +
ω 2
ωB 2
ωS 2
ωR 2
K KB KS KR H 2
From the imaginary part
 p K  p  K  p  K  p 
h  = hB  + hS   + h
2 R 
 ω  K B  ω B  K S  ωS  K R H  ωR 
In case of the resonance frequency, p =ω
3 3 3
ω  ω  ω 
h=   hB +   hS +   hR
 ωB   ωS   ωR 
or
3 3 3
T  T  T 
h =  B  hB +  S  hS +  R  hR
T  T  T 
248
4. Freedom Vector
4.1 Node freedom
Each node has six degrees of freedom and the freedom number is defined as shown in the figure below.
Z 3 Z Z
6
X 1 7 X X
4
5 8
Y Y Y
2
(a) lateral and rotational displacement (b) shear deformation of connection
4.2 Freedom vector
The freedom vector is defined to indicate the number of all freedoms of the structure, where the restrained
freedom is set to be zero. For the structure in the figure below, the freedom vector has zero components for
the fixed nodes (Nodes 1-4) and eight components for other nodes (Nodes 5-8). Therefore, the total number
of freedom of the structure is 8×4 = 32.
Node 1  0 
| 
 
Node 4 0
 
1
Node 5 
5 6
 
8
9
  7 8
Node 6

16 
  1 2
17 
Node 7  

24 3 4
 
25
Node 8 
 
32  shear deformation of connection
Figure 4-2-1 Example of the freedom vector
249
4.3 Dependent freedom
(1) Rigid floor assumption
In the default setting, the floor diaphragm is assumed to be rigid for the in-plane deformation. Therefore,
the in-plane freedoms at the nodes in a floor are represented by the freedoms at the center of gravity of the
same floor.
6
1 4, 7
5, 8
2
(a) In-place freedoms (b) Out-of-plane freedoms
Figure 4-3-1 In-plane and out-of-plane freedom
For example, the in-plane freedoms at the node A in Figure 4-3-2 are expressed by the in-plane freedoms at
the center of gravity G as follows:
u xA  1 0 l yA  u xG 
    
u yA  = 0 1 − l xA  u yG  (4-3-1)
θ  0 0
 zA   1  θ zG 
G: center of gravity
θ zG
G G u xG
l xA
l yA
u yG θ zA
A
A u xA
u yA
Figure 4-3-2 Rigid floor assumption
250
In case of the structure in the figure below, in addition to the original nodes, new nodes for the center of
gravity are defined as “Node 5” and “Node 10”. Under the rigid floor assumption, the freedom vector has
zero components for the in-plane freedoms at the nodes except the center of gravity. Therefore, the total
number of independent freedom is 23.
0
Node 1-5 :
0
0
0
1
2
Node 6 3
0
4
5
0
0
6
Node 7 7
8
0
9 6 7
10
0 10
0
11 8 9
Node 8 12
13
0
14 1 2
15
0 5
0 3
16 4
17
Node 9 18
0
19
20 shear deformation of connection
21
22
0
0
Node 10 0
23
0
0
Freedom vector
Figure 4-3-3 Example of the freedom vector with rigid floor assumption
251
(2) Including wall element
The wall element model has rigid beams at the top and bottom of the wall, therefore, as shown in Figure
4-3-4, the rotation angles in the wall panel plane, θ y1 and θ y 2 , are dependent to the vertical
displacements, δ z1 and δ z 2 . Also, the horizontal displacement in the wall panel plane, u x 2 , is
dependent to the displacement, u x1 . The connection is assumed to be rigid.
θ y2
θ y1
δ z 2 − δ z1
θ y1 = θ y 2 =
δ z2 w
δ z1
u x1 u x2 u x1 = u x 2
Figure 4-3-4 Relationship between node displacements for a wall element (X-wall)
In a matrix form;
 u x1  1 0 0  u x 2 
    
θ y1  = 0 − 1 / w 1 / w δ z1  (4-3-2)
θ  0 − 1 / w 1 / w δ 
 y2    z 2 
In case of Y-direction wall, the relationship can be written as;
θ x1
u y1  1 0 0  u y 2 
δ z1     
θ x1  = 0 1 / w − 1 / w δ z1  (4-3-3)
θ  0 1 / w − 1 / w δ 
1  x2    z2 
θ x2
u y1
δ z2
2 w
u y2
Figure 4-3-5 Relationship between node displacements for a wall element (Y-wall)
252
For example, in case of the structure in the figure below, by eliminating dependent freedoms, the total
number of freedom becomes 17.
0
Node 1-5 :
0
0
0
1
Node 6 2
0
0
0
0
0
0
3
Node 7 4
0
0 6 7
0
0 10
0
0 8 9
5
Node 8 6
7
0 1 2
8
9 5
0
0 3 4
10
Node 9 11
12
0
13
14 shear deformation of connection
15
16
0
0
Node 10 0
17
0
0
Freedom vector
Figure 4-3-6 Example of the freedom vector with a wall element
253
(3) Series of walls
In case several walls are joined together in series, it is assumed that all walls are connected by rigid beams
at the top and bottom. Therefore, as shown in Figure 4-3-7, the rotation angles in the wall panel plane, θ y1
and θ y 2 , are dependent to the vertical displacements, δ z1 and δ z 2 . Also, the horizontal displacement in
the wall panel plane, u x 2 , is dependent to the displacement, u x1 . The connection is assumed to be rigid.
θ yN
θ y2
θ y1
δ z1 δ z2 δ zN
u x1 u x2 u xN
w1
δ zN − δ z1 N −1
θ y1 = θ y 2 =  = θ yN = , L = ∑ wi
L k =1
i −1
δ zi = δ z1 + θ yi Li = (1 − Li / L )δ z1 + (Li / L )δ zN , Li = ∑ wi
k =1
u x1 = u x 2 =  = u xN
Figure 4-3-7 Series of wall connected by a rigid beam (X-wall)
In a matrix form;
δ 
θ yi = [− 1 / L 1 / L ] z1  (4-3-4)
δ zN 
δ 
δ zi = [1 − Li / L Li / L ] z1  (4-3-5)
δ zN 
254
In case of Y-direction wall, the relationship can be written as;
θ x1
δ z1
θ x2
1
u y1
δ z2
2 w1
u y2
δ zN
N δ z1 − δ zN N −1
θ x1 = θ x 2 =  = θ xN = , L = ∑ wi
u yN L k =1
i −1
δ zi = δ z1 − θ xi Li = (1 − Li / L )δ z1 + (Li / L )δ zN , Li = ∑ wi
k =1
u y1 = u y 2 =  = u yN
Figure 4-3-8 Series of wall connected by a rigid beam (Y-wall)
In a matrix form;
δ 
θ xi = [1 / L − 1 / L ] z1  (4-3-6)
δ zN 
δ 
δ zi = [1 − Li / L Li / L ] z1  (4-3-7)
δ zN 
255
(4) Including ground springs
In case there are ground springs (sway and rocking springs) at the basement of the building, the floor
diaphragm of the basement is assumed to be rigid for both in-plane and out-of-plane deformation and the
freedoms other than sway and rocking freedoms are restricted at the center of gravity.
1 4
5
2
(a) Sway freedoms (b) Rocking freedoms
Figure 4-3-9 Freedoms of ground springs
In case of the structure in the Figure below, by eliminating dependent freedoms, the total number of
freedom becomes 21.
256
0
Node 1-4 :
0
1
2
0
Node 5 3
4
0
0
0
0
0
5
Node 6 6
0
0 6 7
0
0 10
0
0 8 9
7
Node 7 8
0
0 1 2
0
0 5
0
0 3 4
9
Node 8 10
11
0
12
13
0
0
14
15
Node 9 16
0
17
18
19
20
0
0
Node 10 0
21
0
0
Freedom vector
Figure 4-3-10 Example of the freedom vector with ground springs
257
4.4 Transformation matrix of dependent freedom
In case of rigid floor assumption, Equation (4-3-1) expresses the relationship between dependent freedom
and independent freedom, that is;
u xA  1 0 l yA  u xG 
    
u yA  = 0 1 − l xA  u yG 
θ  0 0
 zA   1  θ zG 
Dependent freedom Independent freedom
It can be arranged into the transformation matrix between the freedom vectors of all nodes;
k l m
      
      
    
      
i
    
u xA   1 0 l yA   
u yA   0 1 − l yA   
    
δ zA    u xG  k
 =  u  l
θ xA     yG 
θ yA    δ zG 
    θ 
θ zA   0 0 1
  xG 
     θ yG 
    
     θ zG  m
   
      
Dependent freedom [TI ] Independent freedom
Since the most components of the transformation matrix, [TI ] , are zero, the components of [TI ] are
remembered using two matrices, [ N I ] and [ FI ] .
 
[N I ] = i k m 0

; Matrix for independent freedom numbers
 
 
[FI ] = i 1 l yA 0
 ; Matrix for transformation components from independent freedoms
 
It will reduce the memory size dramatically.
258
In the same way, for the case of including wall elements, Equation (4-3-2) expresses the relationship
between dependent freedom and independent freedom, that is;
 u x1  1 0 0  u x 2 
    
θ y1  = 0 − 1 / w 1 / w δ y1 
θ  0 − 1 / w 1 / w δ 
 y2    y2 
Dependent freedom Independent freedom
It can be arranged into the transformation matrix between the freedom vectors of all nodes;
p q r
      
      
    
      
    
 u x1   0 1 0   
      
    
j θ y1   − 1/ w 0 1/ w  δ y1  p
 =   
     
θ y 2   − 1/ w 0 1/ w  u x 2  q
     
     
     δ y 2  r
    
      
   
      
Dependent freedom [TI ] Independent freedom
The components of two matrices, [ N I ] and [ FI ] will be;
 
[N I ] = j  p r 0

; Matrix for independent freedom numbers
 
 
[FI ] = j − 1 / w 1 / w 0
 ; Matrix for transformation components from independent freedoms
 
259
Initial conditions of [ N I ] and [ FI ] are:
   
[N I ] = i  i 0 0 , [FI ] = i 1 0 0

   
In Figure 4-4-1 (rigid connection), the element node displacement vector of the beam element between
Node 8 and Node 9 is,
{u z8 u z 9 θ y8 θ y9 δ x8 δ x 9 }T (4-4-1)
Those displacements correspond to the location numbers in the freedom vector as;
{u z8 u z 9 θ y8 θ y9 δ x8 δ x 9 }T ⇒ {45 51 47 53 43 49}T (4-4-2)
1 0
Node 1-5 : :
30 0
31 0
32 0 u z8 u z9
Node 6 33 1
34 2
35 3 δ x8 δ x9
36 0 8 9
37 0 θ y8 θ y9
38 0
Node 7 39 4
40 5
41 6
42 0
43 0 6 7
44 0
Node 8 45 7 10
46 8
47 9 8 9
48 0
49 0
50 0
Node 9 51 10 1 Z
52 11
53 12
54 0 5
55 13 3 4 X
56 14
57 0
Node 10 58 0 Y
59 0
60 15 (rigid connection)
Freedom vector
Figure 4-4-1 Example of location matrix for beam element
260
From rigid floor assumption, the components of independent matrices, [ N I ] and [ FI ] will be;
     
43 55 60 0
 43 1 l y 8 0
   
   
45 45 0 0 45 1 0 0
   
   
47 47 0 0 47 1 0 0

[N I ] =  
, [FI ] =   (4-4-3)
49 55 60 0 49 1 l y 9 0
   
   
   
51  51 0 0 51 1 0 0
   
   
53 53 0 0 53 1 0 0
     
From the matrix, [ N I ] , the freedoms of (43) and (49) are replaced to the independent freedoms (55) and
(60). Therefore, the independent location numbers and freedom numbers of the beam element are:
{u z8 u z 9 θ y8 θ y9 δ x8 δ x 9 }T
⇒ {45 51 47 53 43 49}
T
⇒ {45 51 47 53 55 60} ; independent location number

T
⇒ {u z 8 u x10 θ z10 }
T
u z 9 θ y8 θ y9
⇒ {5 8 7 10 11 13} ;
T
freedom number
(4-4-4)
The transformation from independent displacements (= global node displacements) to element node
displacements is obtained from the matrix, [ FI ] , as follows:
u z 8  1 0  u z8   u z8 
u   1    u 
 z9    u z9   z9 
θ y 8   1   θ y 8   θ y 8 
 =   = [TixB ]  (4-4-5)
θ y 9   1  θ y 9  θ y 9 
δ x8   1 l y 8  u x10  u x10 
      
δ x 9  0 1 l y 9  θ z10  θ z10 
261
4.5 Stiffness matrix corresponding to independent degrees of freedom
The constitutive equation of the beam element and formulation of global stiffness matrix from element
stiffness matrix are shown below:
5 8 7 10 11 13
 Pz 8  5 k 5,5 k 5,8 k 5, 7 k 5,10 k 5,11 k 5,13   u z 8 
P  8
 k 8,8 k 8, 7 k 8,10 k 8,11 k 8,13   u z 9 
 z9   
 M y 8  7  k 7,7 k 7 ,10 k 7 ,11 k 7 ,13   θ y 8 
 =   
 M y 9  10  k10,10 k10,11 k10,13  θ y 9 
 Px10  11  sym. k11,11 k11,13  u x10 
    
M z10  13  k12,12  θ z10 
Locate element stiffness
Element stiffness matrix
according to the freedom number
1 2 3 4 5 6 7 8 9 10 11 12 13
1 
 
2 
3 
 
4 
5 k 5, 5 k 5, 7 k 5,8 k 5,10 k 5,11 k 5,13 
 
6 
7 k 7,7 k 7 ,8 k 7 ,10 k 7 ,11 k 7 ,13 
 
8 k 8,8 k 8,10 k 8,11 k 8,13 
9  

10  k10,10 k10,11 k10,13 
 
11  sym. k11,11 k11,13 
12  
 
13  k13,13 
Global stiffness matrix
Figure 4-5-1 Formulation of global stiffness matrix
In general, the transformation from independent displacements (= global node displacements) to element
node displacements for the X-beam is described as Equation (2-1-11).
262
u zA 
u 
 zB   u1 
θ yA  u 
 2
  = [TixB ]  (2-1-11)
θ yB  
δ xA  un 
 
δ xB 
And the constitutive equation of the X-beam is also described in Equation (2-1-16).
 P1   u1 
P  u 
 2  2
  = [K ]
xB   (2-1-17)
   
 Pn  u n 
Using the same procedure in Figure 4-5-1, the element stiffness matrix is added into the global stiffness
matrix.
4.6 Mass matrix corresponding to independent degrees of freedom
Mass is assigned in each node. The inertia force at the node will be also transformed according to the
transformation of the variables. Here, the rotational inertia at each node is ignored.
(1) Rigid floor assumption
MG
PxG
G
mA PxA
A PyG
PyA
The inertia force at the node A is
 PxA   −mAuxA   mA 0 0  uxA 

     
 PyA  =
−mAuyA  =−  0 mA 0  uyA  (4-5-1)
0   0  0  θzA 
     0 0
263
Under the rigid floor assumption, the in-plane freedoms at the nodes in a floor are represented by the
freedoms at the center of gravity of the same floor. Therefore,
uxA  1 0 l yA  uxG  uxG  1 0 l yA 

    

uyA  =0 1 −lxA  uyG  =[TA ] uyG =
, [TA ]
0 1 −l 
 xA  (4-5-2)
θ  0 0 1  θ  θ  0 0 1 
 zA     zG   zG 
G: center of gravity
M zG
G G PxG
l xA
l yA
PyG
A
A PxA
PyA
On the other hand, the inertia force at the center of gravity is calculated as,
 PxG   PxA  1 0 0   PxA   PxA 

        
[TA ]
T
= PyG  = PyG  0 = 1 0   PyG   PyG  (4-5-3)
  l P − l P      0 
 M zG   yA xA xA yG  l yA −lxA 1   0   
Therefore,
 PxG  1 0 0   PxA   mA 0 0  uxA 
     0  
0   PyA  = − [TA ] 0  uyA 
T
 PyG  = 0 1  mA
   1   0  0  θzA 
 M zG  l yA −lxA  0 0
(4-5-4)
 mA 0 0  uxG   m 0 l yA mA  u 
T      A
  xG 
− [TA ]  0 mA 0  [TA ] uyG  =
= − 0 mA −lxA mA  uyG 
    θ 
 0 0 0  
θ zG  2 2
l yA mA −lxA mA lxA + l yA mA   zG  ( )
If we ignore the off-diagonal components,
 PxG  m 0 0  u 
   A   xG 
P
 yG  = − 0 mA 0  uyG  (4-5-5)
      
 M zG   0 0 2
mA lxA ( 2
+ l yA )  θ zG 
Taking the sum of the inertia force from the all nodes at the same floor,
 PxG   mG 0 0  uxG  N N
   0 m  u  , m =
P
 yG  =
−  G 0   yG  G ∑i i G ∑i mi lix2 + liy2
m , I = (4-5-6) ( )
M   0  
0 I G  θzG 
 zG 
where, N is the total number of the nodes at the floor.
264
(2) Including rigid beam
The wall element model has rigid beams at the top and bottom of the wall, and the horizontal displacement
in the wall panel plane, u x 2 , is dependent to the displacement, u x1 .
 u x1  1 0   u x1  u 
δ z1 δ z2 =
  =   [T ] u x1 
u x1 u x2 u x 2  1 0  u x 2   x2 
The inertia force after transformation is
 Px1  m 0  u   m1 + m2 0   ux1 

[T=
]  01 m  [T ] ux1 
T
=   0  
 Px 2   2   x2   0  ux 2 
Therefore, the horizontal mass is
P
=x1 ( m1 + m2 ) ux1
Px1 Px 2 Px1
m1 m2 m1 + m2
1 2 1 2
On the other hand, the vertical mass is the same as before.
Pz1 Pz 2 Pz1 Pz 2
m1 m2 m1 m2
1 2 1 2
265
(3) Series of rigid beams
In case several walls are joined together in series, it is assumed that all walls are connected by rigid beams
at the top and bottom.
δ z1 δ zi δ zN
u x1 u xi u xN
Li
L
The all horizontal displacements at the nodes are dependent to the horizontal displacement of
the first node, u x1 .
u x=
1 u x=
2 = u xN

Also, the vertical displacement at the middle node δ zi is dependent to the vertical displacements of
the nodes at both ends, δ z1 δ zN .
 Li   Li 
δ zi =
1 −  δ z1 +  L  δ zN
 L   
Therefore, the horizontal mass is

 N

Px1 = ( m1 + m2 +  + mN ) ux1 =  ∑ mi  ux1
 i =1 
m1 +  + mN
Px1 Pxi PxN Px1
m1 mi mN
1 i N 1 N N
266
The vertical mass is
 N  L   
Pz1  ∑ 1 − i
=  mi  δ z1
 i =1  L  
 N L   
PzN =  ∑  i  mi  δ zN
 i =1  L  
Pz1 Pzi PzN Pz1 PzN
m1 mi mN
1 i N 1 N N
Li
L
267
5. Equation of motion
5.1 Mass matrix
In the default setting, the mass at each node is identical and equally distributed as
1
Mi = M floor (5-1-1)
N floor
where, M i : mass at the node i, M floor : total mass of the floor, N floor : total number of nodes in the
floor.
However, you can change the mass at each node depending on the place of the node by setting “proportion
to influence area” in Option Menu. In this case, the mass at each node is determined from the following
equation:
Ai
Mi = M floor (5-1-1)
A floor
where, Ai : influence area of node i, A floor : total area of the floor. Influence area of the node is different
depending on the place of the node as shown in Figure 5-1-1.
i j
X X
Mi
Ai, Mi Aj, Mj liy
G
IG
k lix MG
A k, M k
G : center of gravity of the floor

Y Y
(1) Influence area of the node (2) Mass and rotational inertia at G
Figure 5-1-1 Mass and rotational inertia at the node
The process to determine the mass based on influence area is as follows:
Step 1. Calculate the slab area (block with cross mark)

Step 2. The area of the block is divided equally to the corner nodes. (Figure 5-1-2)
Step 3. If there is no corner node, the area is divided equally to the all nodes in a floor. (Figure 5-1-3)
268
Figure 5-1-2. Influence area of the node (red)
Figure 5-1-3. Distribution of the rest area
Example) Floor weight = 700kN
700kN/8 50kN+12.5
= 87.5kN 87.5kN 112.5kN
= 62.5kN
87.5kN 112.5kN
87.5kN 62.5kN
87.5kN 62.5kN
87.5kN 112.5kN
87.5kN 87.5kN 62.5kN 112.5kN

(a) Same for all nodes (b) Proportional to influence area
Figure 5-1-4 Example of mass distribution
269
In case of rigid floor assumption, in-plane freedoms at the nodes are dependent to the freedoms at the
center of gravity of the floor. Therefore, the mass at the center of gravity, M G , is,
M G = M floor (5-1-2)
The rotational inertia at the center of gravity, IG, along the z-axis is obtained from the following equation:
N
(
I G = ∑ M i lix2 + liy2 ) (5-1-3)
i
where, N is the total number of the nodes at the floor. The rotational inertia at other nodes are,
I i = 0, i = 1, , N (5-1-4)
The mass matrix is obtained as,
 0 0   
0     
   
u xi  Mi  M i 
   
u yi  Mi  M i 
δ zi  Mi  M i 
[M ] =  ⇒  (5-1-5)
θ xi  Ii   Ii 
θ yi  Ii  I 
   i 
θ zi  Ii   Ii 
  0    
   
 0 0    
Since the mass matrix has only diagonal components, those components are saved in one-dimension vector.
For example, the mass vector of the structure in Figure 5-1-5 will be as follows:
M6 
Node 6  0 
 
M7 
 
Node 7  0 
M8 
  6 7
 0 
 
Node 8  0  10
M  8 9
 9
 0 
Node 9  0 
  1 2
M 10 
  5
M 10  3 4
Node 10  I 10 
Figure 5-1-5 Example of mass vector
270
In case a complete rigid floor such as a foundation slab for the ground springs, we need to calculate the
rotational inertia at the center of gravity along each axis.
b
X
Y a
The rotational inertia along Z-axis is

a /2 b /2 c /2
 M  2
a =IZ ∫=
ρ r dV ∫ ∫ ∫
2

abc
( 2
 x + y dxdydz )
− a /2 − b /2 − c /2  
M  
a /2 b /2 c /2 a /2 b /2 c /2
b
 ∫ x dx ∫ dy ∫ dz + ∫ dx ∫ y dy ∫ dz 
2 2
=X
abc  − a /2 − b /2 − c /2 − a /2 − b /2 − c /2 
M 2
=
12
(
a + b2 )
Y (5-1-6)
In the same way, the rotational inertia along X-axis is
M 2 2
IX
=
12
(
b +c ) (5-1-7)
The rotational inertia along Y-axis is

M 2 2
=IY
12
(
a +c ) (5-1-8)
If the mass is located at each node, as already mentioned, the rotational inertia at the center of gravity, IG,
along the Z-axis is obtained as
X
=IZ ρ r dV ∑ M ( l
∫=
2
i ix
2
+ liy 2 ) (5-1-9) Mi
liy
G
IG
lix MG
G : center of gravity of the floor

Y
271
5.2 Stiffness matrix
As shown in Figure 4-4-2, the global stiffness matrix [K ] is formulated from element stiffness matrices.
5 8 7 10 11 13
 Pz 8  5 k 5,5 k 5,8 k 5, 7 k 5,10 k 5,11 k 5,13   u z 8 
P  8
 k 8,8 k 8, 7 k 8,10 k 8,11 k 8,13   u z 9 
 z9   
 M y 8  7  k 7,7 k 7 ,10 k 7 ,11 k 7 ,13   θ y 8  Example of beam element
 =   
 M y 9  10  k10,10 k10,11 k10,13  θ y 9 
 Px10  11  sym. k11,11 k11,13  u x10 
    
M z10  13  k12,12  θ z10 
Locate element stiffness
Element stiffness matrix
according to the freedom number
1 2 3 4 5 6 7 8 9 10 11 12 13
1 
 
2 
3 
 
4 
5 k 5, 5 k 5, 7 k 5,8 k 5,10 k 5,11 k 5,13 
 
6 
7 k 7,7 k 7 ,8 k 7 ,10 k 7 ,11 k 7 ,13 
 
8 k 8,8 k 8,10 k 8,11 k 8,13 
9  

10  k10,10 k10,11 k10,13 
 
11  sym. k11,11 k11,13 
12  
 
13  k13,13 
Global stiffness matrix
Figure 5-2-1 Formulation of global stiffness matrix
272
5.3 Modal analysis
1) Eigen value problem

The free vibration equilibrium equation without damping is
[M ]{u} + [K ]{u} = {0} (5-3-1)
where [K ] is the stiffness matrix and [M ] is the mass matrix in the form;
m1 0  0
0 m2  0 
[M ] =  (5-3-2)
    
 
0 0  mn 
The solution can be postulated to be in the form
{u} = {φ } e iωt (5-3-3)
where {φ } is a vector of order n, ω is a frequency of vibration of the vector {φ } .
Substituting into the equilibrium equation, the generalized eigen problem is obtained as,
[K ]{φ } = ω 2 [M ]{φ } (5-3-4)
This eigen problem yields the n eigen solutions (ω , {φ }), (ω , {φ }),, (ω , {φ }) where the
2
1 1
2
2 2
2
n n
eigen vectors are M-orthonormalized as,
n
]{φ j } ∑ mkφ=
{φi } [ M=
T
i ,kφ j ,k 0 ; i≠ j (5-3-5)
k =1
)
Let’s assume two different set of eigen solutions (ω ,{φ }) , (ω ,{φ }) .
i
2
i
2
j j
Form Equation (5-3-4),
{φi } [ K ]{φ j } φi } ([ K ]{φ j })

{= ωi 2 {φi } [ M ]{φ j }
T T T
= (5-3-6)
Since [K ] and [M ] are the symmetric matrices,
{φi } [ K ]{φ j } {=
φ } ([ K ]{φ }) {φ j } [ M ]{φi } ω j 2 {φi } [ M ]{φ j }
T T T T
= j i ω j 2= (5-3-7)
Subtracting Equation (5-3-7) from Equation (5-3-6),
(ω i
2
)
− ω j 2 {φi } [ M ]{φ j } =
T
0 (5-3-8)
Since ωi ≠ ω j , we obtain Equation (5-3-5).

The vector {φi } is called the i-th mode shape vector, and ω i is the corresponding frequency
of vibration.
273
2) Modal decomposition of equilibrium equation
Defining a matrix [Φ ] whose columns are the eigenvectors and a diagonal matrix Ω [ ]
2
which stores the eigenvalues on its diagonal as,
ω12 
 
ω 2
[Φ ] = [{φ1 } {φ 2 }  {φ n }] , [ ]
Ω =
2

2



(5-3-9)
 2
 ω n 
We introduce the following transformation on the displacement vector of the equilibrium

equation (5-4-2):
{u (t )} = [Φ ]{q(t )} (5-3-10)
Then,
[ M ][Φ ]{q} + [C ][Φ ]{q} + [ K ][Φ ]{q} =

{P} (5-3-11)
Multiplying [Φ ] ,
T
[Φ ] [ M ][Φ ]{q} + [Φ ] [C ][Φ ]{q} + [Φ ] [ K ][Φ ]{q} = [Φ ] {P}

T T T T
(5-3-12)
where
 m1 
 m2 
[Φ ] [ M ][Φ=]  M=  , = {φi } [ M ]{φi }
T T
mi (5-3-13)
 
 
 mn 
 k1 
 
k2
[ ] [ ][ ] [ ][ ] [ ][ ] [ ]  M    =  K  , k = ω 2m
T T
Φ K Φ = Ω Φ M Φ = Ω   =
    i i i
 
 kn 
(5-3-14)
A damping matrix that is diagonalized by [Φ ] is called a classical damping matrix.
c1 
 c2 
[Φ ] [C ][Φ=] C= 
 
T
(5-3-15)
 
 
 cn 
where,  M  , C  and  K  are called generalized modal mass, modal damping and modal
stiffness matrix, respectively.
274
Therefore,
[Φ ] {P}
 M  {q} + C  {q} +  K  {q} =
T
(5-3-16)
It can be reduced to n- equations of the form
mi qi (t ) + ci qi (t ) + ki qi (t ) =
ri (t ) (5-3-17)
 X0 (t ) 
 
where ri (t ) = {φi } { P (t )} = − {φi } [ M ][U ]  Y0 (t ) 
T T
(5-3-18)
 Z (t ) 
 0 
By setting ci / mi = 2hiωi and ki / mi = ωi

2
 X0 (t ) 
T  
− {βi }  Y0 (t )  =
qi (t ) + 2hiωi qi (t ) + ωi 2 qi (t ) = − {βi , x X0 (t ) + βi ,yY0 (t ) + βi , x Z0 (t )}
 Z (t ) 
 0 
(5-3-19)
where
φi } [ M ][U ] {φi } [ M ] {U x } {U y } {U z }

T
{=
T
{β βi ,y βi ,z }
T
{β i }
T
= = (5-3-20)
{φi } [ M ]{φi } {φi } [ M ]{φi }
T T i,x
{φi } [ M ]{U x,y,z }

T
βi , x , y , z = (5-3-21)
{φi } [ M ]{φi }
T
βi , x , y , z is called “participation factor” of i-th mode.
βi , x , y , z is the coefficient when you decompose the vector {U x ,y,z } into mode vectors as,
n
{U } =
[ Φ ]{β } =
x ,y,z ∑ β {φ } x, y, z i, x, y, z i (5-3-22)
i =1
)
[Φ ] [ M ] ,
T
Multiplying
[Φ ] [ M ]{U x,y,z } =
[Φ ] [ M ][Φ ]{β x, y , z } =
 M  {β x , y , z }
T T
(5-3-23)
Therefore,
{=
β } [Φ ] [ M ]{U x,y,z }
−1 T
x, y, z  M  (5-3-24)
It is equivalent to Equation (5-3-21).
275
Equation (5-3-17) can be decomposed again as,
− X0 (t )
xi (t ) + 2hiωi xi (t ) + ωi 2 xi (t ) =

y (t ) + 2h ω y (t ) + ω 2 y (t ) =
i i i i i −Y (t )
i 0 (5-3-25)
 2
− Z0 (t )
zi (t ) + 2hiωi zi (t ) + ωi zi (t ) =
and
qi (t ) = βi , x xi (t ) + βi ,y yi (t ) + βi ,z zi (t ) (5-3-26)
Therefore, the displacement vector is obtained by superposing displacement responses of

single-degree-of-freedom (SDOF) systems in each mode and each direction as,
n n n n
[Φ ]{q(t )} =
{u (t )} = ∑ {φi } qi (t ) =
∑ βi, x {φi } xi (t ) + ∑ βi,y {φi } yi (t ) + ∑ βi,z {φi } zi (t )
=i 1 =i 1 =i 1 =i 1
(5-3-27)
βi , x {φi } is called “participation vector” of i-th mode in x-direction.
3) Effective modal mass

We consider the vibration of the structure with i-th mode participation vector βi {φi } and the
frequency ωi . The amplitude of the vibration can be expressed as
{ηi (t )} = sβi {φi } (5-3-28)
where s is a scalar value. The amplitude of the velocity is
{ηi (t )} = sωi βi {φi } (5-3-29)
The kinematic energy of the vibration is then calculated as,

1 1 1
{ηi } [ M ]{ηi } ( sωi βi )= {φi } [ M ]{φi } ( sωi βi ) mi
T 2 T 2
=Ei = (5-3-30)
2 2 2
The kinematic energy of SDOF system with an effective mass, M e ,i , under the same
amplitude and the same frequency is
1
( sωi ) M e,i
2
Ee ,i = (5-3-31)
2
Therefore, to be the same kinematic energy, the effective modal mass is defined as,
M e ,i = βi 2 mi (5-3-32)
It can be also expressed as,
({φi } [ M ]{U })
T 2
{φi } [ M ]{U= }m
T
βi {φi } [ M ]=
{U }
T
M e ,i β=
= 2
i mi βi (5-3-33)
{φi } [ M ]{φi } {φi } [ M ]{φi }
T i T
276
The sum of effective modal mass is,
n n
∑ M e ,i βi {φi } [ M ]{U }
∑= {U } [ M ]{U }
T T
= (5-3-34)
=i 1 =i 1
Therefore the ratio of effective modal mass to the total mass is used to judge the number of
significant modes that should be included in the analysis.
4) Initial condition
The initial conditions are obtained from Equation (5-3-10) as,
[Φ ] [ M ][Φ ]{q(t )} =
[Φ ] [ M ]{u (t )} =  M  {q (t )}
T T
(5-3-35)
Therefore,
−1 −1
{q}t  M  [ Φ ] [ M ]{u} {q}  M  [ Φ ] [ M ]{u}
T T
=
0=
=t 0 t 0=t 0
, = (5-3-36)
277
5.4 Damping matrix
In STERA 3D program, the damping matrix is formulated in the following way:
1) Proportional damping
The mass-proportional damping and the stiffness-proportional damping are defined as,
[C ] = a0 [M ] and [C ] = a1 [K ] (5-4-1)
where the constants a0 , a1 have units of sec-1 and sec, respectively.
For a system with a mass-proportional damping, the generalized damping for the i-th mode in
Equation (5-4-1) is obtained as,
ci = a0 mi , ci / mi = 2hi ω i (5-4-2)
Therefore,
a0 1
a0 = 2hiωi , hi = (5-4-3)
2 ωi
Similarly, for a system with a stiffness-proportional damping, the generalized damping for the
i-th mode is,
2
ci = a1ωi mi , ci / mi = 2hiωi (5-4-4)
Therefore,
2hi a1
a1 = , hi = ωi (5-4-5)
ωi 2
hi
C = a0 M C = a1 K
a0 1 a1
hi = hi = ωi
2 ωi 2
In STERA_3D, you can select from the two types of stiffness-proportional damping.
One is the proportional damping using the initial stiffness matrix:
[C ] = 2h [K 0 ] (5-4-6)
ω1
278
where, h: damping factor, ω1 : circular frequency of the first natural mode, [K 0 ] : the initial stiffness.
Another is the proportional damping using the spontaneous stiffness matrix
[C ] = 2h [K p ] (5-4-7)
ω
where, h: damping factor, ω1 : circular frequency of the first natural mode, [K p ] : the spontaneous
stiffness changing according to the nonlinearity of structural elements.
In the scene of the practical design of Japan, it is common to use the proportional damping using the
spontaneous stiffness matrix.
2) Rayleigh damping
A Rayleigh damping matrix is defined proportional to the mass and the initial stiffness
matrices as,
[C ] = a0 [M ] + a1 [K 0 ] (5-4-8)
The modal damping ratio for the i-th mode is,

a0 1 a1
hi = + ωi (5-4-9)
2 ωi 2
The coefficients a0 , a1 can be determined from specified damping ratios h1 , h2 modes,
respectively. Expressing Equation (5-4-9) for these two modes in matrix form leads to:
1 1 / ω1 ω1  a0   h1 
 =  (5-4-10)
2 1 / ω 2 ω 2   a1  h2 
Solving the above system, we obtain the coefficients a0 , a1 :

2ω1ω 2 (ω1h2 − ω 2 h1 )
a0 =
(ω 1
2
− ω2
2
) (5-4-11)
2(ω1h2 − ω 2 h1 )
a1 =
(ω 1
2
− ω2
2
)
hi
C = a0 M + a1 K
a0 1 a1
hi = + ωi
2 ωi 2
279
3) Damping matrix with a base isolation building
In an actual design practice for the base isolation buildings, it is common to assume zero viscous damping
for horizontal components of the base isolation floor. For example, in case of the stiffness-proportional
damping, the damping matrix is defined as:
2h
=[C ] K (  + K 
ω  upper   BI ,V 
) (5-4-12)
where,
[K upper ] : the stiffness matrix consisted with upper structures without base isolation elements,
 K BI ,V  : the stiffness matrix of base isolation elements for vertical components.
4) Damping matrix with viscous damper devices

If there are some viscous damper devices in a structure, in addition to the proportional damping matrix, the
global damping matrix formulated from element damping matrices are considered as:
[C ] = [C pro ] + [C v ] (5-4-13)
[ ]
where, C pro : the proportional damping matrix, [C v ] : the global damping matrix formulated from
element damping matrices in the same manner of the global stiffness matrix.
280
5.5 Input ground acceleration
 , Y and Z , in X, Y and Z

Earthquake ground motions are defined as three components acceleration; X 0 0 0
directions. The inertia forces at node i are defined as,
           
           
       
− M i (uxi + X 0 ) uxi  1 0 0 uxi 
 − M i (uyi + Y0 ) 
        
uyi  0 1 0 
X 0  uyi   X 0 
( )
 − M i δzi + Z0 
= −[M ]
δzi 
− [M ]
0

0 
1     
δ zi   
  Y0  = −[M ]   − [M ][U ] Y0  (5-5-1)
   
 − I iθxi 

θ xi  0 0 0    θ xi   Z 
0   Z   0
 − I iθyi  θyi     
0
0 0 θ yi
        
 − I iθzi 

θ zi  0 0 0 θ zi 
           
       
           
For example, the components of the matrix [U ] of the structure in Figure 5-5-1 will be as follows:
X 0 Y0 Z0

0 0 1
Node 6 0 0 0

0 0 1
 
Node 7 0 0 0
0 0 1
  6 7
0 0 0
Node 8 0 0 0 10
 
0 0 1 8 9
0 0 0
 
Node 9 0 0 0 1 2
 
1 0 0
5
0 1 0 3 4
 
Node 10 0 0 0
Figure 5-5-1 Components of the matrix [U ]
281
Equilibrium condition of the structure under earthquake ground motion is:
 X 0 
[C ]{u} + [K ]{u} = −[M ]{u} − [M ][U ] Y0  (5-5-2)
 Z 
 0
Damping force
Restoring force Inertia force
Finally the equation of motion is obtained as:
 X0 
[M ]{u} + [C ]{u} + [K ]{u} = −[M ][U ] Y0  = {P} (5-5-3)
 Z 
 0
282
5.6 External force by vibrator
A vibrator is assumed to be located at the center of gravity at a certain floor. The external forces from the
vibrator are denoted as Fx , Fy in X and Y directions.
    
    
   
 Fx  1 0 
   
 Fy  0 1 
 0  0 0   Fx  F 
=   =   [V ]  Fx  (5-6-1)
0 0 0   Fy   y 
0 0 0 
   
0 0 0 
    
   
      
For example, the components of the matrix [V ] of the structure in Figure 5-6-1 will be as follows:
Fx Fy
0 0
Node 6 0 0 

0 0
 
Node 7 0 0
0 0
  6 7
0 0
Node 8 0 0 10
 
0 0 8 9
0 0
 
Node 9 0 0 1 2
 
1 0
5
0 1 3 4
 
Node 10  0 0
Figure 5-6-1 Components of the matrix [V ]
283
Equilibrium condition of the structure under vibrator force is:
F 
−[ M ]{u} + [U ]  
[C ]{u} + [ K ]{u} = F
x
(5-6-2)
 y 
Damping force
Inertia force
Restoring force
External force
F 
[ M ]{u} + [C ]{u} + [ K ]{u}= [U ]  Fx = {P} (5-6-3)
 y 
284
5.7 External force by wind
A wind force is assumed to be applied at the center of gravity at each floor with the constant distribution
along the height of the building. The external forces at i-th floor from the wind are denoted as
hi , x Fx ( t ) , hi , y Fy ( t ) , hr , y M z ( t ) in X, Y horizontal directions and Z rotational direction.
Fy ( t )
Fx ( t ) M z (t )
hi F ( t )
Figure 5-7-1 Wind force distribution
       
       
   
 hx ,i Fx ( t )   hx ,i 0 0 
   
 hy ,i Fy ( t )   Fx ( t )  0 hy , i 0 
 0  0 0 0 
=
0
 [W
= ]  Fy ( t )  , [W ]  0 0 0
 (5-7-1)
     
 0  M z ( t ) 0 0 0 
   
hr ,i M z ( t )  0 0 hr ,i 
       
   
       
285
For example, the components of the matrix [W ] of the structure in Figure 5-7-1 will be as follows:
Fx Fy Mz
 0 0 0
Node 6  0
 0 0 
 0 0 0
Node 7  
 0 0 0
 0 0 0 7
  6
 0 0 0
Node 8 10
 0 0 0
  8 9
 0 0 0
 0 0 0
Node 9   1 2
 0 0 0
 
 hx ,1 0 0 5
 0 3 4
hy ,1 0
Node 10  
 0 0 hr ,1 
Figure 5-7-2 Components of the matrix [W ]
286
Equilibrium condition of the structure under wind force is:
 Fx 
 
[C ]{u} + [ K ]{u} =
−[ M ]{u} + [W ]  Fy  (5-7-2)
W 
 z
Damping force
Inertia force
Restoring force
External force
 Fx 
[ M ]{u} + [C ]{u} + [ K ]{u}= [W ]  Fy = {P} (5-7-3)
W 
 z
287
5.8 Numerical integration method
Two numerical integration methods are prepared; one is the Newmark-β method with incremental
formulation using a step-by-step stiffness matrix, and another one is the Force correction method using a
step-by-step stiffness and a force vector together. In case it is difficult to define the step-by-step stiffness of
the element such as the case of using a viscous damper element, the Operator Splitting method is selected.
a) Equation of motion and its incremental form
The equation of motion of a structural system is written as,
[ M ]{a} + [C ]{v} + [ K ]{d } =

{ p} (5-8-1)
where, [M ] , [C ] and [K ] are the mass, damping and stiffness matrices. {d } , {v} , {a} and { p}
are the displacement, velocity, acceleration, and external force vectors.
The incremental formulation for the equation of motion is,
[ M ]{∆ai } + [C ]{∆vi } + [ K ]{∆di } =∆

{ pi } (5-8-2)
where, {∆d i }, {∆vi }, {∆ai } and {∆pi } are the increments of the displacement, velocity, acceleration,
and external force vectors, that is,
{∆d i } ≡ {d i +1 } − {d i }, {∆vi } ≡ {vi +1 } − {vi }, {∆ai } ≡ {ai +1 } − {ai }, {∆pi } ≡ {pi +1 } − {pi } (5-8-3)
In case of a system with hysteresis nonlinearity, the equation of motion can be described as,
[ M ]{a} + [C ]{v} + { f ( d )} =
{ p} (5-8-4)
where f ( d ) is the force as a nonlinear function of the displacement {d } . The incremental form can be,
[ M ]{∆ai } + [C ]{∆vi } + {∆fi ( d )} = {∆pi } (5-8-5)
In a small time-increment, it can be assumed as a linear relationship in force-deformation as shown in

Figure 5-8-1,
{∆f ( d )} =[ K ]{∆d }
i i i (5-8-6)
Finally, the equation of motion in incremental form is the same as Equation (5-8-2), that is
[ M ]{∆ai } + [C ]{∆vi } + [ Ki ]{∆di } =∆

{ pi } (5-8-7)
288
f
∆f
f k ( t ) ∆d
∆=
d
∆d
Figure 5-8-1 Nonlinear force-deformation relationship
In the initial condition, the building will deform under the gravity load, i.e., the dead and live loads. It can
be analyzed by solving the following equation,
 X0 
 
− [ M ][U ]  Y0 
[ M ]{a} + [C ]{v} + [ K ]{d } = (5-8-26)
 Z − g 
 0 
where g is the gravity acceleration. When the gravitational acceleration is initially applied, the response
may fluctuate in the beginning. Therefore, it is better to apply the static gravity force { f0 } instead of
acceleration as,
 X0  0
 
[ M ]{a} + [C ]{v} + [ K ]{d } =
− [ M ][U ]  Y0  + { f 0 } , { f 0 } = − [ M ][U ]  0  (5-8-27)
 Z  g 
 0  
and set the initial displacement as {d } = {d0 } , where {d0 } is the solution of
[ K ] {d 0 } = { f 0 }
The incremental form will be
∆X0 
 
[ M ]{∆ai } + [C ]{∆vi } + [ Ki ]{∆di } =
− [ M ][U ]  ∆Y0 
 ∆Z 
 0
289
b) Newmark-β method
The incremental formulation for the equation of motion of a structural system is,
[M ]{∆ai } + [C ]{∆vi } + [K ]{∆d i } − {∆f } = {∆pi } (5-8-1)
where, [M ] , [C ] and [K ] are the mass, damping and stiffness matrices. {∆d i }, {∆vi }, {∆ai } and
{∆pi } are the increments of the displacement, velocity, acceleration and external force vectors, that is,
{∆d i } ≡ {d i +1 } − {d i }, {∆vi } ≡ {vi +1 } − {vi } , {∆ai } ≡ {ai +1 } − {ai }, {∆pi } ≡ {pi +1 } − {pi } (5-8-2)
{∆f } is the unbalanced force vector in the previous step.
Using the Newmark-β method,

1
{di +1}= {di } + {vi }( ∆t ) +  
− β  {ai }( ∆t ) + β {ai +1}( ∆t )
2 2
(5-8-3)
2 
1
{vi } + ({ai } + {ai +1}) ( ∆t )
{vi +1} = (5-8-4)
2
The incremental form is
{∆d i } = {vi }(∆t ) + 1 {ai }(∆t )2 + β {∆ai }(∆t )2 (5-8-5)

2
{∆vi } = {ai }(∆t ) + 1 {∆ai }(∆t ) (5-8-6)

2
From Equation (5-8-5), we obtain
1 1
{∆ai } = {∆d i } − {vi } − 1 {ai } (5-8-7)
β (∆t ) 2
β (∆t ) 2β
Substituting Equation (5-8-7) into Equation (5-8-6) gives
1
{∆vi } = {∆d i } − 1 {vi } + 1 − 1 
{ai }(∆t ) (5-8-8)
2 β (∆t ) 2β  4β 
Equations (5-8-7) and (5-8-8) are substituted into the equation of motion, Equation (5-8-1), and we obtain
 1 1 
{∆d i } [M ] + [C ] + [K ]
 β (∆t )
2
2β (∆t ) 
(5-8-9)
 1  
= {∆p i }+ [M ] {vi }+ 1 {a i } + [C ] 1 {vi }+  1 − 1{a i }(∆t ) + {∆f }
 β (∆t ) 2β   2β  4β  
290
The equation can be rewritten as,
[Kˆ ]⋅ {∆d } = {∆pˆ }

i i (5-8-10)
where,
[Kˆ ] = [K ] + 2β1(∆t ) [C ] + β (∆1t ) 2

[M ] (5-8-11)
1  
{∆pˆ i } = {∆pi } + [M ] {vi } + 1 {ai } + [C ] 1 {vi } +  1 − 1{ai }(∆t ) + {∆f }
 β (∆t ) 2β   2β  4β  
(5-8-12)
291
c) Operator Splitting method
The Operator Splitting (OS) method is a type of mixed integration method in which stiffness is divided into
linear and nonlinear (Nakashima, 1990). The explicit predictor-corrector method is employed for the
integration associated with the nonlinear stiffness, whereas the unconditionally stable Newmark-β method
is used for the integration associated with linear stiffness. The formulations are described as follows:
Using the Newmark-β method,

1
{di +1}= {di } + {vi }( ∆t ) +  
− β  {ai }( ∆t ) + β {ai +1}( ∆t )
2 2
(5-8-13)
2 
1
{vi } + ({ai } + {ai +1}) ( ∆t )
{vi +1} = (5-8-14)
2
Introducing the predictor displacement {d } as,

i +1
1
{d }= {di } + {vi }( ∆t ) +  
− β  {ai }( ∆t )
2
i +1 (5-8-15)
2 
Equation (5-8-13) can be written as
{di+1} + β {ai+1}( ∆t )
{di +1} =
2
(5-8-16)
Therefore
{ai +1}=
1
β ( ∆t )
2 ({d i +1 } − {di +1})=
1
β ( ∆t )
2 {∆di +1} (5-8-17)
where
{∆di +1} = {di +1} − {di +1} (5-8-18)
Substituting Equation (5-8-17) into Equation (5-8-14),

1 1
{v=
i +1} {∆di +1} + {vn } + {an }( ∆t ) (5-8-19)
2 β ( ∆t ) 2
In the equation of motion,
[ M ]{ai +1} + [C ]{vi +1} + { f ( di +1 )} =

{ pi +1} (5-8-20)
The nonlinear internal resisting forces are approximated as follows:
{ f=
( d )} [ K ]{d } − {∆f }
i +1 n +1 (5-8-21)
where
{∆f }
= [ K ]{dn+1} − { f ( dn+1 )} (5-8-22)
292
In this formulation, [K ] is the initial stiffness matrix.
Force
[ K ]{di +1}
∆f
{ f ( d )}
i +1 Corrector
[ K ]{di +1}
∆f
{ f ( d )}
i +1 Predictor
[K ] Displacement
{ }
di +1 {di +1}
The nonlinear internal resisting forces can be written as,
{ f ( d=)}
i +1 ( ) { ( )}
 K di +1  {∆di +1} + f di +1
 
(5-8-23)
( )
where  K di +1  is a predictor stiffness.

The predictor stiffness is not necessary to be the initial stiffness and if the predictor stiffness is close to the
tangent stiffness, the corrector force is more accurate. It is known that if the predictor stiffness is larger than
the tangent stiffness, the OS method is unconditionally stable.
Force
Corrector
{ f ( d )}
i +1
 ( )
 K di +1 

{ f ( d )}
i +1 Predictor
Displacement
{ }
di +1 {di +1}
In STERA_3D, the predictor stiffness is created from the initial stiffness or tangent stiffness
if available.
293
Substituting the above equations into the equation of motion,
[ M ]{ai +1} + [C ]{vi +1} + { f ( di +1 )} =

{ pi +1} (5-8-24)
 1   1 1 
[ ]
M  {∆d i +1  + [ C ] 
}   2 β ( ∆t ) {∆di +1} + {vn } + {an }( ∆t ) 

 β ( ∆t )
2
 2 

 ( )  { ( )} =
+  K dn +1  {∆di +1} + f di +1 {p } i +1
Solving for {∆di +1} ,
 Kˆ  {∆di +1} =
{ pˆ } (5-8-25)
 
where
1 1
   ( )
 Kˆ  = K dn +1  +
 2 β ( ∆t ) [ ] β ( ∆t )2 [ ]
C + M (5-8-26)
− [C ]  {vn } + {an }( ∆t )  − { f ( di +1 )} + { Pn +1}

 1 
{ pˆ } = (5-8-27)
 2 
The procedure for solving the equation of motion is as follows:

Step 1. {~ } by Equation (5-8-15).
Calculate the predictor displacement vector d n +1
Step 2. Obtain the restoring force { f ( d )}
i +1 in reference to the constitutive model.
Step 3. Substitute { f ( d )}
i +1 to Equation (5-8-27) and solve the displacement increment {∆di +1}
from Equation (5-8-25) and obtain the corrector displacement {d n+1 } from Equation (5-8-18).
Under seismic excitation and gravity load, the equation of motion will be,
 X0,n +1 
 
[ M ]{an+1} + [C ]{vn+1} + { f ( di +1 )} =
− [ M ][U ]  Y0,n +1  (5-8-27)
 Z 
 0,n +1 − g 
The initial displacement as {d } = {d0 } , where {d0 } is the solution of
0
[ K ]{d0 } = − [ M ][U ]  0  (5-8-29)
g 
 
294
5.9 Energy
a) Equation of energy
As it was mentioned in Equation (5-5-2), the equation of motion is obtained as:
 X 0 
[M ]{u} + [C ]{u} + [K ]{u} = −[M ][U ] Y0  = {P} (5-9-1)
 Z 
 0
For example, in case of a structure with a rigid floor in Figure 5-9-1, the displacement vector, {u}, consists
of 15 components (see RED numbers in Figure 5-9-1.)
 u1 
u 
{u} =  2  (5-9-2)
  
u15 
1 4
6 7
2 5
7 15 10
10 3 6
13
8 9
8 11
9 14
12
1 2
5
3 4
Node number Freedom number
Figure 5-9-1 Example of the freedom vector of a structure with a rigid floor
The equation of energy is derived by multiplying the velocity vector, {u} , and integrating by the time
T
range [0-t]:
t t t t
∫ {u} [M ]{u}dt + ∫ {u} [C ]{u}dt + ∫ {u} [K ]{u}dt = −∫ {u} {P}dt

T T T T
(5-9-3)
0 0 0 0
295
{u}T [M ]{u} + t {u}T [C ]{u}dt + {u}T [K ]{u} = − t {u}T {P}dt
2 ∫
0
2 ∫
0
(5-9-4)
WK + WD + WP = WI (5-9-5)
where,
WK =
{u}T [M ]{u} : Kinematic energy
2
t
WD = ∫ {u} [C ]{u}dt : Damping energy
T
{u} [K ]{u}
T
WP = : Potential energy
2
t
WI = − ∫ {u} {P}dt : Input energy
T
If the system is nonlinear, the equation of motion can be expressed as:
 X 0 
[M ]{u} + [C ]{u} + Q(u, u ) = −[M ][U ] Y0  = {P} (5-9-6)
 Z 
 0
where, Q(u , u ) is the nonlinear restoring force vector. Then, the equation of energy can be derived as;
WK + WD + WP = WI (5-9-7)
where,
WK =
{u}T [M ]{u} : Kinematic energy
2
t
W D = ∫ {u} [C ]{u}dt
T
: Damping energy
0
t (5-9-8)
WP = ∫ {u} Q(u , u )dt : Potential energy
T
0
t
WI = − ∫ {u} {P}dt
T : Input energy
0
296
b) Decomposition of potential energy
We can decompose the restoring force vector into the restoring force of each member as,
Q(u , u ) = q1 (u , u ) + q 2 (u , u ) +  + q n (u , u ); n : number of members (5-9-9)
Therefore, the potential energy can be decomposed as,
t t n t T
n  n
WP = ∫ {u} Q(u , u )dt = ∫ {u} ∑ qi (u , u )dt = ∑  ∫ {u} qi (u , u )dt  =∑ WP ,i
T T
(5-9-10)
0 0 i =1 i =1  0  i =1
where
t
WP ,i = ∫ {u} qi (u , u )dt ;
T
potential energy of i-th member (5-9-11)
0
297
6. Nonlinear Static Push-Over Analysis
6. 1 Lateral distribution of earthquake force
The static lateral load representing the earthquake force is applied at the center of gravity in each floor.
There are several formulas to define the load distribution along the height of the building. In “STERA 3D”
program, the following distributions are prepared:
1. Ai 2. Triangular 3. Uniform 4. UBC 5. ASCE 6. Mode
(1) Ai distribution
In the “Building Standard Law” in Japan, the design shear force of i-th story, Qi, is defined as,
n
Qi = C i ∑ w j , C i = ZRt Ai C 0 (6-1-1)
j =i
where,
Ci : design shear coefficient of i-th story,
wi : weight of i-th story,
Z: seismic zone factor,
Rt: vibration characteristic factor taking into consideration of soil condition,
Ai : lateral distribution of shear force coefficient,
C0: design base shear coefficient (C0 =0.2 for serviceability limit, C0 =1.0 for safety limit)
If we set, Z=1.0 (Tokyo), Rt=1.0 (stiff soil, a short story building), C0=1.0 (safety design), the design shear
force distribution is simplified as,
n
Qi = Ai ∑ w j (6-1-2)
j =i
“Ai” distribution is defined as,
 1  2T
Ai = 1 +  − αi  (6-1-3)
 α  1 + 3T
 i 
where,
n n
α i = ∑ w j W , W = ∑ w j : the ratio of weight upper than i-th story,
j =i j =1
T: the first natural period of a building (=0.02h, h : the building height)
As shown in Figure 6-1-1, the static lateral load is obtained as,
Fn = Qn , Fi = Qi − Qi +1 (i = 1, , n − 1) (6-1-4)
298
w6
F6 = Q6
Q6 = C6 w6
F5 = Q5 – Q6 w5
Q5 = C5 (w5 + w6)
w4
.
. n
. w3 Qi = C i ∑ w j
. i =i
. w2 C i = ZRt Ai C 0
.
w1
F1 = Q1 – Q2
Q1 = C1 (w1 + w2 + ・・・+ w6)
Figure 6-1-1 Ai distribution
(2) Triangular distribution

Triangular distribution is defined as:
 n 
Fi = QB  hi ∑h 
j  (6-1-5)
 j =1 
where,
QB : base shear force
hi : the height of the i-th story from the ground
Fi
h6
h2
h1
Figure 6-1-2 Triangular distribution
299
(3) Uniform distribution
Uniform distribution is defined as:
Fi = QB (1 n ) (6-1-6)
Fi
Figure 6-1-3 Uniform distribution
(4) UBC distribution

The UBC (Uniform Building Code, 1997) gives the following formula for the calculation of lateral force
distribution:
 n 
Fi = (QB − Ft ) wi hi ∑w h j

j  (6-1-7)
 j =1 
 0 , if T ≤ 0.7 sec
Ft =  (6-1-8)
0.07 TQB , if T > 0.7 sec
Ft
Fi
h6
h2
h1
Figure 6-1-4 UBC distribution
300
(5) ASCE distribution
The IBC (International Building Code) in the U.S. refers to the ASCE 7 “Seismic Design Requirements for
Building Structures” which gives the following formula for the calculation of lateral force distribution:
n
Fi = wi hik ∑w h
j =1
j
k
j (6-1-9)
where k is an exponent related to the structural period as follows:
 1 , if T < 0.5sec

k= (T − 0.5 ) / 2 , if 0.5sec < T < 2.5sec (6-1-10)
 2 , if T > 2.5sec

k=2 k=1
Fi
h6
h2
h1
Figure 6-1-5 ASCE distribution
301
(6) Mode distribution
Mode distribution is defined as:
 n 
Fi = QB  wiφ1,i ∑w φ j

1, j  (6-1-11)
 j =1 
where,
φ1,i : component of the first mode distribution in the i-th story
Fi
φ1,i
Figure 6-1-6 Mode distribution
302
6. 2 Capacity Curve
The Capacity Spectrum Method was proposed by Freeman [1978] as an approximate way to estimate the
maximum response of a structure under an earthquake ground motion. The concept was modified by
Kuramoto et.al [2000] to adopt the distribution of nonlinear story displacement as the first mode shape in
each calculation step. The method was adopted as one of the evaluation procedures in the Building
Standard Law, Japan.
The key concept of the Capacity Spectrum Method is to find out the intersection between the Demand
Spectra (= relationship between Sd (displacement spectra) and Sa (acceleration spectra)) and the Capacity
Curve (= nonlinear push-over curve of an equivalent single-degree-of-freedom system).
1400
1200
Demand Spectra
SA-SD（h=0.05)
1000
Demand Spectra
Sa (cm/sec 2)
800 SA-SD（h=0.14)
Capacity Curve
600
400
Performance Point
200
0
0 10 20
Sd (cm)
Figure 6-2-1 Schematic example of the concept of Capacity Spectrum Method
“STERA 3D” provides the menu in the static analysis to show the Capacity Curve based on the following
formula (Kuramoto et.al [2000]):
n n
∑ miδ i2 ∑m δ i i
2
S a = QB i =1
2 , Sd = i =1
n
(6-2-1)
 n

 ∑ mi δ i  ∑m δ i i
 i =1  i =1
where,
mi : lumped mass in the i-th story
δi : component of the distribution of nonlinear story displacement in the i-th story
303
Sd
Fi δi
≈ M Sa
M : Equivalent SDOF mass
Nonlinear static push-over analysis Capacity Curve of SDOF system
Figure 8-2-2 Capacity Curve of the equivalent SDOF system
As schematically shown in Figure 8-2-2, the step-by-step results of nonlinear push-over analysis is used to
obtain the Capacity Curve of the equivalent SDOF system using Equation (8-2-1).
References
Freeman S. A. (1978), ”Prediction of Response of Concrete Buildings to Severe Earthquake Motion”,
Douglas McHenry International Symposium on Concrete and Concrete Structures, SP-55, American
Concrete Institute, Detroit, Michigan, pp. 589-605.
Kuramoto H., et.al. (200), “Predicting the Earthquake Response of Buildings using Equivalent Single
Degree of Freedom System”, 12th World Conference on Earthquake Engineering (12WCEE), Auckland
New Zealand，2000.2.
304
7. Lumped Mass Model
7.1 Decomposition of shear and flexural deformation
a-1) Equivalent plane for each floor from displacement
The equivalent plane ( z = ax + by + c ) is obtained from the vertical displacement distribution by the least
square method:
z = ax + by + c
zi x
Figure 7-1-1 Equivalent plane
L = ∑ ( z i − (axi + by i + c ))
2
Minimize (7-1-1)
where, i : node number in the floor

a, b, c : parameters of equivalent plane
∂L ∂L ∂L
Thus, = 0, = 0, =0 (7-1-2)
∂a ∂b ∂c
Parameters, a, b, c are obtained by solving the following linear equation:
 ∑ z i x i  ∑ x i2 ∑ x y ∑ x  a
i i i
  
∑ z i y i  =  ∑ y ∑ y  b 
2
i i (7-1-3)
 ∑ z i   sym. n   c 
  
where,
n : the number of nodes in a floor
305
a-2) Equivalent plane for each floor from potential energy
The equivalent plane ( z = ax + by + c )is obtained from the vertical potential energy distribution by the
least square method:
= N ( ax + by + c )
Nz
N i zi x
Figure 7-1-2 Equivalent plane
∑ ( N z − N ( ax + by + c ) )
2
Minimize L
= i i i i i (7-1-4)
where, i : node number in the floor

N i : axial load at node i
a, b, c : parameters of equivalent plane
∂L ∂L ∂L
Thus, = 0, = 0, =0 (7-1-5)
∂a ∂b ∂c
Parameters, a, b, c are obtained by solving the following linear equation:
 ∑ N i zi xi   ∑ N i xi2 ∑ N x y ∑ N x  a 
i i i i i
  
 ∑ N i zi yi  =  ∑ N y ∑ N y  b 
i
2
i i i (7-1-6)
 ∑ N i zi   sym. ∑ N   c 
   i
where,
n : the number of nodes in a floor
At this moment, STERA_3D adopts the formulation a-1), since it is easier to implement.
306
b) Decomposition of shear and flexural deformation
A story drift, D, can be divided into shear and flexural components as,
=D DS ( shear ) + DF ( flexure) (7-1-7)
Assuming the distribution of floor deformation is expressed by an equivalent plane ( z = ax + by + c ), the

flexural deformation, DF , can be expressed as,
DF = −aH : x-direction (7-1-8)

DF = bH : y-direction (7-1-9)
Note that the coefficient ‘a’ is the negative angle in x-direction.

Then, the shear deformation can be obtained as,
DS= D − DF (7-1-10)
D
DF DS
in x-direction
−a
H
z
−a z = ax + by + c
DS DF in y-direction
b z
H
z = ax + by + c b
y
Figure 7-1-3 Decomposition of shear and flexural deformation
307
7.2 Lumped mass model with shear and flexural stiffness
a) Linear flexural model

The frame model can be idealized as a lumped mass model with a concentrated mass at each floor and
shear and flexural springs in each story.
Figure 7-1-4 Idealization to lumped mass model
Under the external lateral forces, Fi (i = 1, 2,3) , the shear force and moment of each story are expressed
as below.
F3
θ3
EI 3
h3 Q3 = F3
F2 M 3 F=
= 3 h3 Q3 h3 , θ 2
EI 2
h2 Q=
2 F3 + F2
F1 M 2 = F2 h2 + F3 ( h2 + h3 ) = Q2 h2 + Q3 h3 , θ1
h1 EI1
Q1 = F3 + F2 + F1
M 1 = F1h1 + F2 ( h1 + h2 ) + F3 ( h1 + h2 + h3 ) = Q1h1 + Q2 h2 + Q3h3
Figure 7-1-5 Moment and shear force of lumped mass model
In general
N
M i = ∑ Qj hj (7-1-11)
j =i
Note that if we consider the sign of coordinate

N N
−∑ Qxj h j , M xi =
M yi = ∑ Qyj h j (7-1-12)
=j i =j i
308
From the beam theory
MB
δ
B
τB θB = τ B + R
R
R
τA
θA =τ A + R  M A  2 EI  2 1  τ A  2 EI  2 1  θ A − R 
=  =    
M B  h 1 2  τ B  h 1 2  θ B − R 
A
(7-1-13)
MA
M i +1 M i +1
hi EI i hi EI i
Mi −M i
Figure 7-1-6 Moment and rotational deformation

Substituting
M A = − M i , M B = M i +1 , τ A = θi −1 , τ B = θi −1 + ∆θi , EI = EI i , h = hi
− M i  2 EI i  2 1   θi −1  2 EI i  3θi −1 + ∆θi 
  =     (7-1-14)
 M i +1  hi 1 2  θi −1 + ∆θi  hi 3θi −1 + 2∆θi 
Therefore, the equivalent flexural stiffness can be obtained as

hi
EI=
i ( M i +1 + M i ) , =i 1, , n
2∆θi
(7-1-15)
h
EI n +1 = n +1 ( M n +1 )
2∆θ n +1
309
The increment of rotational deformation ∆θi is the difference of floor angle. Therefore,
∆θ1 =
θ1
(7-1-16)
∆θi = θi − θi −1 , i = 2, , n + 1
From the beam theory, the flexural deformation is
h3 h
δ= Q + (θ A + θ B ) (7-1-17)
12 EI 2
M + MB 6 EI δ
) Q =
− A − 2 {(θ A + θ B ) − 2 R} , R =
=
h h h
Therefore, the flexural deformation of i-th story is obtained as
h13 h
DF 1
= Q1 + 1 θ1
12 EI1 2
(7-1-18)
hi 3 h
DFi = Qi + i (θi −1 + θi ) , i = 2, , n + 1
12 EI i 2
The shear deformation is then calculated substituting the flexural deformation from the story drift as
D=
Si Di − DFi (7-1-19)
Under the nonlinear push over analysis, it is generally assumed that the flexural component is elastic and
only the shear component is considered as nonlinear.
Figure 7-1-7 Decomposition of shear and flexural deformation
310
Therefore, the lumped mass model is obtained from the following procedure:
In the first step of push-over analysis (in elastic stage)

1) Calculate equivalent plane ( z = ax + by + c ) of each floor to obtain the flexural angles ai or bi
2) Calculate increment of angle ∆θi = ai − ai −1 or ∆θi = bi − bi −1
3) Calculate the flexural stiffness
hi
=EI i ( M i +1 + M i ) (7-1-20)
2∆θi
4) Calculate the flexural deformation
hi 3 hi hi 3 h
DxFi = Qi − ( ai −1 + ai ) or DyFi = Qi + i ( bi −1 + bi ) (7-1-22)
12 EI i 2 12 EI i 2
5) Calculate the shear deformation
D=
Si Di − DFi (7-1-23)
From the next step, we use the same flexural stiffness obtained previously.
6) Calculate increment of angle
hi
∆θi
= ( M i +1 + M i ) (7-1-24)
2 EI i
7) Calculate flexural angle of each floor
i i
a=
i ∑ ∆θk =1
k or b=
i ∑ ∆θ
k =1
k (7-1-25)
8) Calculate the flexural deformation

hi 3 h hi 3 h
DxFi = Qi − i ( ai −1 + ai ) or DyFi = Qi + i ( bi −1 + bi ) (7-1-26)
12 EI i 2 12 EI i 2
9) Calculate the shear deformation
D=
Si Di − DFi (7-1-27)
10) The relationship between the shear deformation and the shear force is idealized as a nonlinear
hysteresis model of the shear spring of each story.
311
b) Nonlinear flexural model
To consider nonlinear flexural component, the model to separate shear deformation and bending
deformation is used.
Reference) Akira Wada, et. Al. “Response Control Design of Buildings”, Maruzen (in Japanese), 1998
δs
uB B
δB
ks θB
θA
h φ
uA kb
δA A
The shear and moment force acting at the center springs are
Qs = k sδ s = k s ( u B − u A − hθ B ) for nonlinear shear spring (7-1-28)
M
= b bφ
k= kb (θ B − θ A ) for nonlinear bending spring (7-1-29)
nε n
N n k=
= kn (δ B − δ A ) for linear axial spring (7-1-30)
In a matrix form,
 Qs   k s 0 0  δ s  δ s 
  =    
=M b   0 kb 0  φb  [ kW ] φb  (7-1-31)
N   0 0 kn    ε 
 n ε n   n
312
F3
θ3
N 3 = W3
h3 Q3 = F3
F2 M 3 F=
= 3 h3 Q3 h3 , θ 2
N=
2 W3 + W2
h2 Q=
2 F3 + F2
F1 M 2 = F2 h2 + F3 ( h2 + h3 ) = Q2 h2 + Q3 h3 , θ1
N1 = W3 + W2 + W1
h1 Q1 = F3 + F2 + F1
M 1 = F1h1 + F2 ( h1 + h2 ) + F3 ( h1 + h2 + h3 ) = Q1h1 + Q2 h2 + Q3h3
Figure 7-1-5 Moment and shear force of lumped mass model
By substituting
−M i , M B =
MA = M i +1 , QA ==
Qi , QB Qi +1 , N A ==
N i , N B N i +1 ,
=θ A θi −=
1, θB θ=
i, 1,
u A ui −= i, δA
uB u= δ=
i , δB δ=
i +1 , h hi
the lumped mass model is obtained from the following procedure from the push-over analysis.
1) Calculate equivalent plane ( z = ax + by + c ) of each floor to obtain the flexural angles ai or bi
and the vertical location at the center of gravity ( xci , yci ) as zci = axci + byci + c .
2) Calculate shear deformation

δ s=
1 u1 − η h1θ1 (7-1-32)
δ si =ui − ui −1 − η hiθi , i =2, , n + 1 (7-1-33)
Note that η = −1 for θ i = ai (x-direction) and η = 1 for θ i = bi (y-direction)
3) Calculate the shear stiffness
k si = Qsi δ si (7-1-34)
4) Calculate axial deformation
ε n1 = δ1 (7-1-35)
ε ni = δ i − δ i −1 , i =
2, , n + 1 (7-1-36)
i
Note that δ=
i zic − ∑ hi
k =1
5) Calculate the axial stiffness
313
kni = N i ε ni (7-1-37)
6) Calculate moment at each floor
N
M i = η ∑ Qj hj (7-1-38)
j =i
Note that η = −1 for (x-direction) and η = 1 for (y-direction)

7) Moment of the bending spring is M i +1
M bi M
= = i, i 1, , n (7-1-39)
The rotational deformation of the bending spring is
M bi
φii= θi − θi −1 (7-1-40)
Mi
To make the number of digits equivalent to a shear force, the nonlinear bending moment is divided by the
height of the story, on the other hand, the angle is multiplied by the height of the story to obtain the
deformation. Therefore, the force-deformation of the nonlinear bending spring is transformed as
Mb kb
Equivalent shear force Qb = Qb
= = δ b kbsδ b
h h2
Equivalent story drift δ b = hφ
kb
=
Equivalent stiffness Qb bsδ b ,
k= kbs
h2
After finding the tri-linear model for Qb − δ b relationship, it is returned to M b − φ relationship as,
δb
= b h,
M b Q= φ , kb = kbs h
2
In dynamic analysis, the rotational inertia at each floor is neglected.
314
c) Trilinear modeling of push-over curve
From the push over results up to the ultimate deformation (for example, up to 1/50 drift ratio), the
relationship between the story drift (shear δ s , bending δ b ) and the shear force (shear Qs , bending
Qb ) of each story is transformed into a tri-linear skeleton.
P3 (D3, Q3)
P2 (D2, Q2)
K3
K2
P1 (D1, Q1)
K1
< Case 1 >

When the drift ratio (drift divided by the story height) of the last point is less than the minimum value (for
example, 1/1000)
P3 (D3, Q3)
The skeleton is assumed to be linear.
P1 (D1, Q1)
The last point is P1 K3
K1 = Q1/D1
P2(D2, Q2) P2 (D2, Q2)
D2 = 2×D1
K2
K2 = K1
P1 (D1, Q1)
P3(D3, Q3)
D3 = 4×D1 K1
K3 = K1
315
< Case 2 >
When the last stiffness is large (for example, tangent stiffness > 0.1 K1 (initial stiffness))
P1(D1, Q1)
Find initial stiffness K1
Find Q1 that is the force when the tangent stiffness becomes 0.8K1 and determine D1 = Q1/K1
P2(D2, Q2)
The last point is P2.
K2 is the stiffness between P1 and P2
P3(D3, Q3)
D3 = 2×D2
K3 = K2
P3(D3, Q3)
P2(D2, Q2)
K2
P1(D1, Q1) 0.8K1
K1
316
< Case 3 >
When the last stiffness is small (for example, tangent stiffness < 0.1 K1 (initial stiffness))
P1(D1, Q1)
Find initial stiffness K1
Find Q1 that is the force when the tangent stiffness becomes 0.8K1 and determine D1 = Q1/K1
P2(D2, Q2)
P2 is decided to be the same energy between push-over analysis and the model up to P3.
P3(D3, Q3)
P3 is the last point of push-over analysis
K3 is the tangent stiffness at P3
P3(Q3,D3)
P2(Q2,D2)
K3 = tangent K
K2
0.8K0
P1(Q1,D1)
K1
317
8. P-D effect
Following formulation is suggested in the following book:

James F. Doyle, “Static and Dynamic Analysis of Structures”, Kluwer Academic Publishers, 1991
a) Equilibrium of the beam with an axial load
We consider equilibrium of the beam with a slight displacement with an axial load.
M + ΔM
EI v(x) M
F0
Δv
F0
x Δx V
V
Δx
Figure 7-2-1 Equilibrium of small beam segment slightly deformed
Assuming small deflection, the balance of moment on the small segment “Δx” gives
∆M + V (∆x ) − F0 (∆v ) = 0 (7-2-1)

Therefore
dM dv
+ V − F0 =0 (7-2-2)
dx dx
d 2v
From the relationship, M = EI , the governing differential equation for the deflection shape is
dx 2
d 4v d 2v
EI − F0 =0 (7-2-3)
dx 4 dx 2
The general solutions are,

for compression loading ( F0 < 0 ):
v( x) = c1 cos kx + c 2 sin kx + c3 x + c 4 , k 2 = − F0 / EI , (7-2-4)
for tensile loading ( F0 > 0 ):
v( x) = c1 cosh kx + c 2 sinh kx + c3 x + c 4 , k 2 = F0 / EI (7-2-5)
318
b) Geometric stiffness matrix of the beam with an axial load
We assume that the axial force is constant and compressive. From the general solution, Eq. (7-2-4),
at x = 0
dv(0)
v(0) = v1 = c1 + c 4 , = φ1 = kc 2 + c3 (7-2-6)
dx
Consequently, the deflected shape is
v( x) = c1 (cos kx − 1) + c 2 (sin kx − kx ) + v1 + φ1 x (7-2-7)
Similarly at the end of other node,
v( L) = v 2 = c1 (cos kL − 1) + c 2 (sin kL − kL) + v1 + φ1 L (7-2-8)

dv( L)
= φ 2 = −kc1 sin kL + kc 2 cos kL + φ1 (7-2-9)
dx
Then, the coefficients, c1 , c 2 , can be arranged as,
(1 − C ) (ξ − S )   c1  v1 + φ1 L − v 2 
= (7-2-10)
 ξS
 ξ (1 − C ) c 2   φ1 L − φ 2 L 
where,
C = cos kL, S = sin kL, ξ = kL (7-2-11)
Solving this equation by Cramer’s rule gives

c1 = [v1ξ (1 − C ) + φ1 L( S − ξC ) − v 2ξ (1 − C ) + φ 2 L(ξ − S )] / ∆ (7-2-12)
c 2 = [− v1ξS + φ1 L(1 − C − ξS ) + v 2ξS + φ 2 L(C − 1)] / ∆ (7-2-13)
where
∆ = ξ (2 − 2C − ξS ) (7-2-14)
Now we can rewrite the deflection function in terms of the nodal degrees of freedom. The moment and
shear force distributions can be obtained as
d 2v
M ( x) = EI
dx 2
[
= EI − k 2 c1 cos kx − k 2 c 2 sin kx ] (7-2-15)
d 3v dv
V ( x) = − EI 3
+ F0 = − EIk 2 [φ1 − kc 2 ] (7-2-16)
dx dx
319
Calculating nodal loads, V (0) = −V1 , M (0) = − M 1 , V ( L) = V1 , M ( L) = M 1 , the stiffness matrix is
 V1  ξ 2 S ξL(1 − C ) − ξ 2S ξL(1 − C )   v1 
M    
 1  = EI ξ 
2
− L (ξC − S ) − ξL(1 − C )
2
L2 (ξ − S )  φ1 
(7-2-17)
 V2  L3 ∆  ξ 2S − ξL(1 − C )  v 2 
    
M 2   sym. − L2 (ξC − S ) φ 2 
c) Approximation of geometric stiffness matrix
We simplify the geometric stiffness matrix to be linear in the loading F0.

Using the series expansion for the sine and cosine terms, the determinant is,
∆ = ξ (2 − 2C − ξS )
[ ( ) (
≈ ξ 2 − 2 1 − ξ 2 / 2 + ξ 4 / 24 − ξ 6 / 720 +  C − ξ ξ − ξ 3 / 6 + ξ 5 / 120 −  )] (7-2-18)
[
≈ ξ 1 − ξ / 15 +  / 12
5 2
]
also
1 12
[
= 5 1 + ξ 2 / 15 + 
∆ ξ
] (7-2-19)
We now do the expansion on the stiffness terms. For example,

EI ξ 2 2
k11 = 3
L ∆
EI
L
( ) [ (
12
)] [ EI
] [
ξ S = 3 ξ 4 ξ − ξ 3 / 6 +  5 1 + ξ 2 / 15 +  = 3 12 1 − ξ 2 / 10 + 
ξ L
]
(7-2-20)
Substituting ξ
2
= k 2 L2 = − F0 L / EI ,
EI F0 12 
k11 = [12 ] + (7-2-21)
L3 L 10 
In the same manner, we can expand for all the stiffness terms to get the stiffness matrix as
 12 6L − 12 6 L   36 3L − 36 3L 
 4 L2 2 
− 6 L 2 L  F0   4 L2 − 3L − L2 
[k ] = EI3  + (7-2-22)
L  12 − 6 L  30 L  36 − 3L 
   
 sym. 4 L2   sym. 4 L2 
We can write as
[k ] = [k E ] + [k G ] (7-2-23)
where, [k E ] : the element elastic stiffness, [k G ] : the element geometric stiffness
320
d) Implementation for beam element
θA z
τA τB
θB
uB
x
uA
A B y
Figure 7-2-2 Including node movement
For beam element,
 M A  2 EI 2 1  τ A  EI 4 L2 2 L2  τ A 
 M  = L 1 2 τ  = 3  2   (7-2-24)
 B    B  L 2 L 4 L2  τ B 
Including node movement,
u 
1 1  A
τ A   L 1 − 0 θ 
L  A
τ  =  1 1   (7-2-25)
 B  0 − 1 u B 
L L  θB 

 1 1 
 QA  1 1 u 
M 
 L L  1 −
 A
0 θ 
 0  4 L
2 2
 A  = EI  1 2L   L L  A

 QB  L3 − 1 − 1  2 L2 
4 L2   1 1  u 
  0 − 1  B 
M B   L L L L  θB 

 0 1 
 6L 6L  1 1 u 
 A  12 6L − 12 6 L  u A 
 4 L2 2 L2   1 − 0 θ  EI  4 L2 − 6 L 2 L2  θ A 
EI   L L  A = 
= 3   
L − 6 L − 6 L   1 0 − 1
1 u B  L 
3
 12 − 6 L  u B 
 2   
 2L 4 L2   L L  θB 
  sym. 4 L2  θ B 
From (7-2-22), the geometric stiffness matrix will be

 36 3L − 36 3L 
 4 L2 − 3L − L2 
F0 
[k G ] = (7-2-26)
30 L  36 − 3L 
 
 sym. 4 L2 
321
Therefore, the stiffness equation will be
 QA    12 6L − 12 6 L   36 3L − 36 3L   u A 
M   
 A  =  EI

 4 L2 2 
− 6 L 2 L  F0   4 L2 − 3L − L2   θ A 
+
 Q B   L3  12 − 6 L  30 L  36 − 3L   u B 
        
M B    sym. 4 L2   sym. 4 L2   θ B 
322
e) Implementation for column element
B u xB
u yB
λB l ' τ yB
τ xB θ yB
θ xB
l'
θ yA
τ yA
λ Al ' X
θ xA
A u xA
τ xA
u yA Y
Figure 7-2-3 Including node movement
 M yA  2 EI 2 1  τ yA  EI 4 L2 2 L2  τ yA 
M  =  =     in X-Z plane (7-2-27)
 yB  L 1 2 τ yB  L3 2 L2 4 L2  τ yB 
 M xA  2 EI 2 1  τ xA  EI 4 L2 2 L2  τ xA 
M  =  =     in Y-Z plane (7-2-28)
 xB  L 1 2 τ xB  L3 2 L2 4 L2  τ xB 
Including node movement,
1 1 u 
   xA 
τ yA  − 1 0 θ
L L  yA 
τ  =  1 1  u  in X-Z plane (7-2-29)
 yB  − 0 1  xB 
 L L  θ yB 
 
u
1 1   yA 
τ xA   L 1 − 0 θ 
L
  
xA
τ  =  1 1
 xB   0 − 1 u yB 
L L  θ xB 
Note that the matrix for node movement in X-Z plane is different from that of beam element. The
force-deformation relationship in X-Z plane is then,
323
 Q xA   1 1
u 
M  − L −
L  1 1   xA 
2 L2  − 1 0 θ
 yA  = EI  1 0  4 L
2
L L  yA 
 Q xB  L3  1 1  2 L2 
4 L2  − 1 1  u 
   0 1  xB 
 M yB   L L   L L  θ yB 
 0 1   
− 6 L − 6 L   1 1 u   12 − 6 L − 12 − 6 L  u xA 
 xA  
 2 2 L2  −
EI  4 L 1 0 θ  EI  4 L2 6 L 2 L2  θ yA 
= 3  L L  yA  =
 u  
L  6L 6 L  − 1 0
1 3
1  xB  L 
 12 6 L  u xB 
 2   
 2L 4 L2   L L  θ yB 
   sym. 4 L2  θ yB 
(7-2-31)
Considering the difference of sign of stiffness matrix in X-Z plane, the geometric stiffness matrix will be
 36 − 3L − 36 − 3L 
 4 L2 3L − L2 
[k xG ] = F0  in X-Z plane (7-2-32)
30 L  36 3L 
 
 sym. 4 L2 
 36 3L − 36 3L 
 4 L2 − 3L − L2 
F
[ ]
k yG = 0 
30 L  36 − 3L 
 
 sym. 4 L2 
Therefore, changing the order of vector component, the force-deformation relationship of column will be
 Q xA  u xA   36 − 36 − 3L − 3L 0 0 0 0 0 0 0 0 u xA 
Q  u   − 36 36 3L 3L 0 0 0 0 0 0 0 0 u xB 
 xB   xB   
 M yA  θ yA  − 3L 3L 4 L 2
−L 2
0 0 0 0 0 0 0 0 θ yA 
      
θ yB  0 θ yB 
2 2
M yB  − 3L 3L − L 4L 0 0 0 0 0 0 0
 Q yA  u yA   0 0 0 0 36 − 36 3L 3L 0 0 0 0 u yA 
      
 Q yB  u yB  F0  0 0 0 0 − 36 36 − 3L − 3L 0 0 0 0 u yB 
  = [K ]  +  
 M xA 

θ xA  30 L  0 0 0 0 3L − 3L 4 L2 − L2 0 0 0 0 θ xA 

 M xB  θ xB   0 0 0 0 3L − 3L − L2 4 L2 0 0 0 0 θ xB 
     
 N zA  δ zA 
 0 0 0 0 0 0 0 0 0 0 0 0 δ zA 
 
N  δ   0 0 0 0 0 0 0 0 0 0 0 0 δ zB 
 zB   zB    
 M zA  θ zA   0 0 0 0 0 0 0 0 0 0 0 0 θ zA 
M 
 zB 
θ 
 zB   0 0 0 0 0 0 0 0 0 0 0 0 θ zB 
324
u xA 
u 
 xB 
θ yA 
 
θ yB 
u yA 
 
u yB 
= [[K ] + [K G ]]  (7-2-34)
θ xA 
θ xB 
 
δ zA 
δ 
 zB 
θ zA 
θ 
 zB 
where,
 36 − 36 − 3L − 3L 0 0 0 0 0 0 00
 − 36 36 3L 3L 0 0 0 0 0 0 0
0

2 2
− 3L 3L 4 L −L 0 0 0 0 0 0 00
 2 2 
− 3L 3L − L 4L 0 0 0 0 0 0 00
 0 0 0 0 36 − 36 3L 3L 0 0 00
 
0 0 0 0 − 36 36 − 3L − 3L 0 0 00
[K G ] = F0  (7-2-35)
30 L 0 0 0 0 3L − 3L 4 L2 − L2 0 0 00
 
 0 0 0 0 3L − 3L − L2 4 L2 0 0 00
 0 0 0 0 0 0 0 0 0 0 00
 
 0 0 0 0 0 0 0 0 0 0 00
 
 0 0 0 0 0 0 0 0 0 0 00
 0 0 0 0 0 0 0 0 0 0 0 0
Then, applying translation of Equation (2-2-17), the constitutive equation of the column is;
 P1   u1 
P  u 
 2  2
  = [K ]
C   (7-2-36)
 
 Pn  un 
where,
[K C ] = [TC ]T [k C ][TC ] + [TiC ]T [K G ][TiC ] (7-2-37)
325
9. Unbalance force correction
a) Procedure to correct unbalance force

In nonlinear analysis, sudden change of spring stiffness sometimes causes severe error for estimating
element force. For example, estimation of spring force f i +1 is overestimated in Figure 7-2-1 and
“unbalance force” is defined as,
∆f = f i +1 − f i′+1 (7-3-1)
where, f i′+1 is the force on the nonlinear skeleton curve
The most preferable way to minimize the error is to adopt iterative calculations such as
Newton-Raphson method. However, this iteration may consume calculation time significantly.
Therefore, the following simple way is adopted to correct unbalance force:
1) Calculate unbalance displacement ∆d from the unbalance force ∆f
∆d = ∆f / k (7-3-2)
where, k is the spring stiffness
2) Subtract unbalance displacement ∆d from the increment displacement in the next step
calculation
i +1
f i +1
∆f
f i′+1
i
fi ∆d
k
d
di d i +1
Figure 7-3-1 Unbalance force
326
b) Unbalance force correction of MS model
For the Multi-spring model (MS model) of Column element, the sum of the unbalance forces of
nonlinear vertical springs in the member section is calculated as:
5 5
∆N = ∑ ∆ f i = ∑ (∆f c ,i + ∆f s ,i ) (7-3-3)
i =1 i =1
where ∆f c ,i : unbalance force of concrete spring,

∆f s ,i : unbalance force of steel spring
The unbalance displacement is then calculated as:
5 5
∆D = ∆N ∑ k i = ∆N
i =1
∑ (k
i =1
c ,i + k s ,i ) (7-3-4)
where k c ,i : stiffness of concrete spring,

k s ,i : stiffness of steel spring
In the next step calculation, the increment displscement of each spring is ajusted as follows:
∆d i′ = ∆d i − ∆D (7-3-5)
where ∆d i : increment displacement of i-th spring
∆d i′ : adjusted increment displacement of i-th spring
∆f 1
∆f 2 ∆f 5
xs x
∆f 4
∆f 3
Figure 7-3-2 Unbalance force in MS-model
The same procedure is adopted for the MS model of Wall element.
327
10. Calculation of ground displacement
In STERA_3D, the ground displacement is calculated from the ground acceleration data using FFT method
and filtering techniques based on the description in the following reference:
Reference: Yorihiko Osaki, “Introduction of Spectral Analysis of Earthquake Ground Motion”, Kajima
publishing corporation, 1981 (in Japanese)
a) Discrete Fourier Transform

Assume that the acceleration data is collected at an interval, ∆t = T / N and consists of the
N measurement data xm (m = 0, 1, 2,, N − 1) , where T is the period of the data that
corresponds to the duration time of data. The coefficient of a Fourier series is obtained as:
1 N −1
Ck =
N
∑x
m =0
m e −i ( 2πkm / N ) k = 0, 1, 2,, N − 1 (7-4-1)
The inverse discrete Fourier transform is

N −1
xm = ∑ C k e i ( 2πkm / N ) m = 0, 1, 2,, N − 1 (7-4-2)
k =0
b) Integration of the data in time domain
Assume y m (m = 0, 1, 2,  , N − 1) is the integration of the discrete data x m in time domain.
The data y m is obtained by the following inverse discrete Fourier transform:
N∆t N −1
y m =  ∫ x m dt  =
t = m∆t
 0
∑
 2π k = 0
S k e i ( 2πkm / N ) m = 0, 1, 2,  , N − 1 (7-4-3)
where, the coefficients S k are obtained from the coefficients C k as,
2πv 0 N / 2 −1
Im(C k ) π ( N − 1)C 0
S0 = −2 ∑ +
N∆t k =0 k N
πC 0
Sk = [− 1 + i cos(πk N )] − i C k *
, S N −k = S N −k k = 1, 2,  , N / 2 − 1 (7-4-4)
N k
πC 0
SN /2 = −
N
328
The following band pass filter (Butterworth filter) in frequency domain is applied to the
coefficient S k .
GB ( f ) = GL ( f )GH ( f ) (7-4-5)
( f fL )
2n
GL ( f ) = (7-4-6)
1+ ( f fL )
2n
1
GH ( f ) = (7-4-7)
1+ ( f fH )
2n
1.2 1.2
N=3
1 1
N=5
0.8 0.8
N = 10
0.6 0.6
N=3
0.4 N=5 0.4
0.2 N = 10 0.2
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
GL ( f ) GH ( f )
STERA_3D adopts the following frequency parameters:

f L = 0.1 (Hz)
f H = 20 (Hz)
329
c) Calculation flow
The ground acceleration data is integrated twice to obtain displacement data. Band pass filter is applied
each time of the integration. The flow of calculation is summarized below:
[1] From acceleration data to velocity data
x m (m = 0, 1, 2,  , N − 1)
FFT Calculate Fourier coefficients of the data
Ck k = 0, 1, 2,  , N − 1
Eq. (7-4-4) Calculate Fourier coefficients of the data of the integration
Sk k = 0, 1, 2,  , N − 1
Eq. (7-4-5) Apply band pass filter
hk S k k = 0, 1, 2,  , N − 1
IFFT Calculate the data of integration by Inverse Fourier transform
y m (m = 0, 1, 2,  , N − 1)
[2] From velocity data to displacement data

Repeat the above process again
330
11. Damage Index
11.1 Damage Index of RC Members

Reference:
- Young-Ji Park, A. H. Ang (1985) “Mechanistic Seismic Damage Model for Reinforced Concrete”,
Journal of Structural Engineering, ASCE
1) Park and Ang Damage Index
STERA_3D adopts the following damage index, so called Park and Ang damage index, to evaluate the
structural damage under earthquake.
δ Eh µm E
D =+
m
β =+ β h (11-1)
δu Q yδ u µu Q yδ u
where
δ m = µmδ y : maximum deformation under an earthquake,

δ u = µu δ y : ultimate deformation under a monotonic loading,
µm : maximum ductility factor under an earthquake,
µu : ultimate ductility factor under a monotonic loading,
δy : yield deformation,
Qy : yield strength,
β : parameter related to the cumulative loading effect,
Eh = ∫ dE : dissipated hysteretic energy.
Qy
δy δM δ u = µu δ y
Eh = ∫ dE
Figure 11-1 Force-deformation relationship of the member
331
The cumulative ductility factor is the ratio of the cumulative dissipated energy defined as
Eh
η= (11-2)
Q yδ y
The damage index can be rewritten as
µm η
D
= +β (11-3)
µu µu
2) RC Beam and Column
Ultimate ductility factor

According to Park and Ang (1985), the ultimate ductility factor, µu , for reinforced concrete beams and
columns is highly variable and depends on the failure mode of the member as shown in Figure 11-2.
Figure 11-2 Ultimate ductility factor and failure mode (Park and Ang (1985))
In case of the flexural failure, the value is greater than 10. Therefore, in STERA_3D,
µu = 15
is adopted for the nonlinear flexural springs at both ends of the reinforced concrete beams and columns.
332
Parameter β
The parameter β represents the effect of cyclic loading on damage. According to Park and Ang (1985), β is
calculated as,
 l ρ 
β=
 −0.447 + 0.073 + 0.24n0 + 0.314 pt  × 0.7
w
(11-4)
 d 
where
l/d : shear span ratio (replaced by 1.7 if l / d <1.7),

n0 : normalized axial stress (replaced by 0.2 if n0 <0.2),
pt : longitudinal steel ratio as a percentage (replaced by 0.75% if pt <0.75%),
ρw : confinement ratio.
Figure 11-3 shows the comparison between the calculated and experimental results of β. The applicable
range of the above equation is
1.0 < l / d < 6.6
0 ≤ n0 < 0.52
0.2 < ρ w < 2.0
15.9 MPa < f c ' (concrete strength) < 41.4 MPa (11-5)
Figure 11-3 Parameter β (Park and Ang (1985))
The default values in STERA_3D are

µu = 15
β = 0.2
333
3) RC Wall
There is not much study about the damage index of RC shear walls. Therefore, STERA_3D adopts
arbitrary values for µu and β .
STERA_3D adopts
µu = 15
β = 0.05
for the nonlinear flexural springs at both ends of the reinforced concrete wall.
Also
µu = 8
β = 0.1
is adopted for the shear spring of the reinforced concrete wall.
4) Damage Index of group of members
The damage index for a part of a structure, such as individual story and for the entire structure, can be
evaluated as the weighting average of damage indices of structural elements in the part.
n
D part = ∑ wi Di (11-6)
i
where
D part : damage index of the part of the structure
n : number of elements in the part of the structure
wi : weighting factor of the i-th element.
Di : damage index of the i-th element
The weighting factor wi can be based on the dissipated hysteretic energy of each element as,
Eh ,i
wi = n
(11-7)
∑E
i
h ,i
334
11.2 Damage Index of Steel Members
Reference:
- Study on Seismic Performance for Super-High-Rise Steel Buildings against Long-Period Earthquake
Ground Motions, Building Research Institute, Building Research Data, No. 160, 2014,7 (in Japanese)
1) Steel Beam Connection
a) Damage index based on fatigue curve

The linear cumulative damage model known as the Miner rule is one of the frequently applied procedure
to estimate the cumulative damage index (CDI) of element with random cyclic loadings. It is described as,
ni
CDI
= ∑N
i
≤1 (11-8)
i
where
CDI : cumulative damage index
ni : number of cycles accumulated at strain level ∆ε i
Ni : number of cycles to fracture
For the low cycle fatigue with the cyclic plastic deformation, the relationship between the strain amplitude
∆ε i and the number of cycles to fracture N i is expressed by the Mason-Coffin equation as,
∆ε i ( % ) =
C ⋅ Ni − β (11-9)
or
1 1
−
 ∆ε i  β  C  β
=N i =    (11-10)
 C   ∆ε i 
It can be written as follows using the ductility factor µi instead of ∆ε i ,
1 1
−
 µi   C β β
=N i =    (11-11)
C  µi 
According to Figure 11-4 in the report “Study on Seismic Performance for Super-High-Rise Steel Buildings
against Long-Period Earthquake Ground Motions” (BRI, 2014),
C = 4  10
β =1 3
335
Figure 11-4 Fatigue curve for different connection types of steel beams (BRI, 2014)
In this method, the number of cycles ni accumulated at strain level ∆ε i (or ductility factor µi ) must be
calculated using the Rain-flow method.
336
Appendix) Rain-flow method
Reference:
- RAINFLOW CYCLE COUNTING IN FATIGUE ANALYSIS, Tom Irvine, 2018
The Rain-flow algorithm is the method for counting fatigue cycles from a time history.
337
b) Damage index based on the maximum response
Since it is an intensive work to record the time history of strain (or ductility factor) for all beams and
calculate the damage index using the rain-flow method, a practical method is proposed using the maximum
ductility factor and the cumulative ductility factor (BRI, 2014).
The cumulative ductility factor is defined as

Eh
η= (11-12)
Q yδ y
( µm − 1) δ y
Qy
δy δ m = µmδ y
E0 ≈ 4 ( µm − 1) Qyδ y
The energy dissipation per cycle with the deformation of the maximum ductility µm is
E0 ≈ 4 ( µm − 1) Qyδ y (11-13)
Therefore, the equivalent number of cycles is
Eh η Q yδ y η
N= = = (11-14)
E0 4 ( µm − 1) Qyδ y 4 ( µm − 1)
e
The number of cycles to fracture with the maximum ductility µm is

1
−
µ  β
Nf = m  (11-15)
 C 
Therefore, the damage index is evaluated as
1
Ne η  µm  β
CDI
= = (11-16)
N f 4 ( µm − 1)  C 
338
2) BRB (Buckling Restrained Brace)
Reference:
- Bucking-Restrained Braces and Applications, Edited by T. Takeuchi and A. Wada, JSSI, 2017
a) Damage index based on fatigue curve

The Miner rule is described as,
ni
CDI
= ∑N
i
≤1 (11-17)
i
The Mason-Coffin equation for the relationship between the strain amplitude ∆ε i and the number of
cycles to fracture N i is expressed as,
∆ε i ( % ) =
C ⋅ Ni − β (11-18)
For the BRB (buckling restrained brace) damper, Takeuchi et al. (2008), proposed the following formulas,
∆ε i ( % ) =
0.5 ⋅ N i −0.14 ( ∆ε ( % ) < 0.1% )
i
∆ε i ( % ) = 20.48 ⋅ N i −0.49 ( 0.1% < ∆ε ( % ) < 2.2% )

i (11-19)
∆ε i ( % ) =54.0 ⋅ N i −0.71 ( 2.2% < ∆ε ( % ) )

i
Figure 11-3 Relationship between strain and number of cycles to facture (Takeuchi et al. (1985))
339
By combining with Coffin-Manson equations,
( ∆ε i < 0.1% ) ( 0.1% ≤ ∆ε i < 2.2% ) ( 2.2% ≤ ∆ε i )

 
ni ni ni
CDI = ∑  , , 
(11-20)
 −
1
−
1
−
1

  ∆ε i   ∆ε i   ∆ε i 
0.14 0.49 0.71


  0.5   20.48   54.0  
   
b) Damage index based on the maximum response

Using the same concept as in the case of steel beams, the damage index is evaluated as
1
Ne η  µm  β
CDI
= = (11-21)
N f 4 ( µm − 1)  C 
340

STERA3D Technical Manual

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STERA3D Technical Manual

Uploaded by

Copyright:

Available Formats

STERA 3D

STructural Earthquake Response Analysis 3D

Dr. Taiki SAITO

TOYOHOSHI UNIVERSITY OF TECHNOLOGY (TUT), JAPAN

2008/07/08 STERA_3D Technical Manual Ver.2.0 is uploaded.

(1) Global Coordinate

Figure 1-1-1 Global coordinate

Figure 1-1-1 Global coordinate

Figure 1-1-2 Local coordinate of a beam element

Figure 2-1-1 Element model for beam

Force-displacement relationship for elastic element

From the relationship between shear force and moment,

θ ' yA  τ ' yA  φ yA  η yA   M ' yA 

 M ' yA  θ ' yA  θ ' yA 

where, [k B ] is the stiffness matrix of the beam element.

θ ' yA  θ yA − τ  (u zB − λ B l 'θ yB ) − (u zA + λ Al 'θ yA )

Figure 2-1-2 Including rigid parts and node movement

From node axial displacements, relative axial displacement is,

Combining Equations (2-1-7) and (2-1-9),

θ ' zA  θ zA − τ  (u yA − λ Al 'θ zA ) − (u yB + λ B l 'θ zB )

λ Al ' l' λB l '

Figure 2-1-3 Beam displacement with rigid connection (X-Y plane)

We can summarize for both ends as

From global node displacement to element face displacement

Transformation matrix for nonlinear spring displacement

Local coordinate of Y-beam Global coordinate

Figure 2-1-4 Relation between local coordinate and global coordinate

X-Z plane Y-Z plane

Figure 2-2-1 Element model for column

Force-displacement relationship for elastic element

X-Z plane Y-Z plane

Figure 2-2-2 Element model for column

φ yA = f yA M' yA , φxA = f xA M'xA at end A (2-2-5)

φ yB = f yB M' yB , φxB = f xB M'xB at end B (2-2-6)

Nonlinear interaction M x − M y − N z is considered in the nonlinear bending springs,

where, [ f pA ] and [ f pB ] are the flexural

φ xB   M ' xB  Figure 2-2-3 Nonlinear bending springs

The force-displacement relationship is then expressed as,

The force-deformation relationship of shear spring is

Q'x   ksx 0   sx   sx  1 ksx 0  Q'x 

From the relationship between shear force and moment,

η yA  1 l ' 0  1 l ' 0  1 ( ksx l') 0 

Figure 2-2-5 Nonlinear shear springs

The force-deformation relationship of shear spring is

=  f 21 l ' f 21 l ' f 22 l' f 22 l' f 23 

Both cases can be written in

The flexural matrix [ f C ] is;

 f p11 f p12 f p13 f p14 f p15 0

 M ' yA  θ ' yA  θ ' yA 

Figure 2-2-6 Including rigid parts and node movement

From global node displacement to element face displacement

[K C ] = [TC ]T [kC ][TC ] (2-2-32)

ϕ yA   M ' yA   M' yA   M' yA 

M ' xB1 M ' yBc M ' xB 2

M ' xA1 M ' yAc M ' xA2

Figure 2-3-1 Element model for wall

Force-displacement relationship for elastic element

N ' zBc , ε zBc

Figure 2-3-2 Nonlinear bending springs

φ yAc   M ' yAc 

φ yBc   M ' yBc 

φ xB1   M ' xB1 