You are on page 1of 30

Response Spectrum Analysis

Theory, Benefits and Limitations

Emrah Erduran, PhD


NORSAR / International Center of Geohazards (ICG), Kjeller, Norway
Table of contents

1. Introduction
2. Design spectrum for elastic analysis
3. Free Vibration
3.1. SDOF
4.1. MDOF
4. Modal Analysis under Earthquake Forces
5. Response Spectrum Analysis (RSA)
6. RSA in EC-8
7. Tutorial
8. Concluding Remarks
Structural Analysis under Seismic Action

The aim of structural analysis under seismic action is to compute the design actions
(forces and displacements) on the building components and the entire system

G + ΨEi ⋅ Q
General types of analysis methods specified in EC8:

 linear-elastic methods:
(1) lateral force method of analysis
(2) response spectrum analysis
 non-linear (inelastic) methods
(3) non-linear static ('pushover') analysis
(4) non-linear time history analysis
Linear Models

lpl lpl
i j
E, I, A
Non-linear Models

lpl lpl
i j
E, I, A
Moment
plastic hinge region
Moment

lpl = 0.5∙d to d
Curvature with: d - depth of section
Curvature
Design spectrum for elastic analysis
(EN 1998-1:2004 3.2.2.2)

 to avoid explicit inelastic structural analysis in design, the capacity of the structure to
dissipate energy is accounted for by using a reduced response spectrum by q
 q is an approximation of the ratio of seismic forces that the structure would experience
if its response would be completely elastic to the seismic forces used for the design

 2 T 2.5 2 
for 0 ≤ T ≤ TB : Sd (T ) = a g ⋅ S ⋅  + ⋅ ( − )
 3 TB q 3 

Spectral acceleration Sd
q=1
2.5
for TB ≤ T ≤ TC : Sd ( T ) = a g ⋅ S ⋅ q=2
q
q=4
2.5  TC 
 = a g ⋅ S ⋅ ⋅ 
for TC ≤ T ≤ TD : Sd (T ) q T 
TB TC TD
 ≥ β ⋅ag
Period T [sec]

2.5  TC ⋅ TD 
 = a ⋅ S ⋅ ⋅ with: β = 0.20
 T 
g
for TD ≤ T ≤ 4 s : Sd (T ) q 2

 ≥ β ⋅ag (lower bound factor)


Free Vibration - SDOF

 For an SDOF system, the equation of motion:

 For an undamped system, the equation of motion reduces to:

 When an undamped SDOF system is disturbed with an initial


displacement and released, the system osciallates with an
harmonic motion.
Displacement

Time
Free Vibration - MDOF
m3

m2

m1

 Equation of motion for MDOF: [M ]⋅ {u} + [C ]⋅ {u} + [K ]⋅ {u} = 0


 with:
 m1 0 0 0  c 11 .. .. c 1n   k11 .. .. k1n 
     
m2 0 .. c 22 .. ..  .. k22 .. .. 
[M ] =  [C ] =  [K ] = 
0 0
0 0 m3 0 .. .. c 33 ..  .. .. k33 .. 
     
0
 0 0 m i  c
 n1 .. .. c nn  k
 n1 .. .. knn 

Assumption: [C] = zero matrix !


– Undamped system
Free Vibration - MDOF
m3 φn,1 φn,2 φn,3

m2 φj+1,1 φj+1,2 φj+1,3


m1 φj,1 φj,2 φj,3

 Contrary to SDOF systems, for a MDOF system, the motion of each mass (or each floor)
is NOT a simple harmonic motion, when an arbitrary initial deflection is applied to the
system.
 An associated period (or frequency) cannot be defined.
 An undamped MDOF will undergo simple harmonic motion with an associated
frequency only if the initial deformations are arranged in an approprioate distribution

NATURAL MODES OF VIBRATION


or

MODE SHAPES
Free Vibration - MDOF
m3 φn,1 φn,2 φn,3

m2 φj+1,1 φj+1,2 φj+1,3


m1 φj,1 φj,2 φj,3

 At any given instant:

 wn: natural circular frequency of vibration


 Tn: natural period of vibration; time required for one cycle of simple harmonic motion
Determination of Mode Shapes

Procedure:

 modal segmentation: [K ] − ω 2 ⋅ [M ] = 0 => ω = km

 derive circular frequencies ωi / periods Ti and mode shapes φi

 or obtain them from finite element software (e.g. ROBOT)


Modal Analysis of MDOF systems under
earthquake forces
m3

m2

m1

φn,1 φn,2 φn,3

φj+1,1 φj+1,2 φj+1,3


m3 φj,1 φj,2 φj,3
m2
=
m1
T1 T2
T3
Modal Analysis of MDOF systems under
earthquake forces
φn,1 φn,2 φn,3

φj+1,1 φj+1,2 φj+1,3


φj,1 φj,2 φj,3
m3

m2 =
m1
T1 T2 T3

qi(t) can be computed fairly easily!


Modal Analysis of MDOF systems under
earthquake forces
 Modal analysis of MDOF systems allows us to conduct a simplified analysis instead of
solving a system of nxn differential equations.
 However, we still need to solve n differential equations!!!
 Modal analysis provides us with the response parameters (e.g. member forces,
displacements, etc...) at every time step of an earthquake record.
 Typically, this time step is between 0.005 seconds and 0.02 seconds.
 Do we really need this information?

ENGINEERS NEED TO KNOW ONLY THE MAXIMUM PROBABLE ACTION FOR DESIGN!

RESPONSE SPECTRUM!
Excursion

Response spectrum: response action

maximum value
 used in earthquake engineering (exclusively)
earthquake impact
 describes the maximum response of a SDOF
system to a particular input motion (i.e. the
respective accelerogram)
 dependent on damping ratio ξ (1–10 %)
 response spectra reflect the maximum
Sa,2
response to simple structures (SDOF) Sa,3
Response

Spectral acceleration Sa
Sa,1 Sa,4 spectrum:

Sa,5

Sa,6

Sa,7

T1 T2 T3 T4 T5 T6 T7
Period T
Response Spectrum Analysis
φn,1
Procedure:  φ j ,1 
  φj+1,1
Given: - circular frequencies ωi / periods Ti {φ1} = φ j +1,1  φj,1
φ 
- mode shapes φi  n ,1 

n T1
∑m
j =1
j ⋅ φ j ,i
 modal participation factors γi : γi = n

∑m
j =1
j ⋅ φ 2j ,i

design spectral accelerations Sa(Ti ) for each mode i :

Spectral acceleration Sa
 Sa,d (T3)

Sa,d (T2)
Sa.d (T1)

T3 T2 T1
Period T [sec]
Response Spectrum Analysis

Procedure:
 lateral story loads Fj ,i : F j ,i = m j ⋅ φ j ,i ⋅ γ i ⋅ S a ,d ( Ti )

Mode shape i: 1 2
3 φn,1 φn,2 φn,3

φj+1,1 φj+1,2 φj+1,3


φj,1 φj,2 φj,3

Fn,1 Fn,2 Fn,3

Fj+1,1 Fj+1,2 Fj+1,3

Fj,1 Fj,2 Fj,3

 resulting shear forces Fb,m, that will be used in the design: ????
Modal Combination Rules

 Absolute Sum(ABSSUM):

 Very Simple (Primitive (?))


 Conservative... REALLY Conservative!!

 Square root of sum of squares (SRSS):

 Simple
 Fairly ‘accurate’ for buildings with well-seperated frequencies.
 Should be avoided if the modal frequencies are close to each other.
Modal Combination Rules

 Complete Quadratic Combination:

 Mathematically complex
 Leads to the most reliable solution for buildings with closely spaced natural
frequencies .
 ... as well as well-seperated frequencies
 ρn,m is the cross-modal coefficient
Response spectrum analysis in EC-8
(EN 1998-1:2004 4.3.3.3)

Criteria:
 shall be applied if the criteria for analysis method (1) are not
fulfilled, this means if: Fb
4 ⋅ TC
T1 >
2.0 sec 1st mode

 response of all modes shall be considered that contribute significantly to the global
building response (i.e., important for buildings of a certain height)
 those modes shall be considered for which:
(1) the sum of the modal masses is at least 90% of the
∑mi ≥ 0.9 ⋅ mtot
total building mass or
(2) the modal mass is larger than 5% of the total building
mi ≥ 0.05 ⋅ mtot
mass
Response spectrum analysis in EC-8

Criteria (cont'd):
 if the '90%' and the '5%' criteria is not fulfilled (e.g. for buildings prone to torsional
effects), those modes shall be considered for which:

k ≥ 3 ⋅ √n
and
Tk ≤ 0.20 s

with: k - number of modes taken into account


n - story number (from above foundation to top)
Tk - period of vibration of mode k
n=4 Mode shape: 1 2 3 4
Example: 16-story building
k ≥ 3 ⋅ √16 = 12
 and
T12 = 0.002 s ≤ 0.20 s
→ twelve modes shall be
considered !!

Period Tk : 0.85 s 0.29 s 0.16 s 0.07 s


Methods of analysis
(EN 1998-1:2004 4.2.3)

General types of analysis methods specified in EC8:

Regularity Allowed simplification Behavior factor


Plan Elevation Model Linear-elastic analysis (for linear analysis)

● ● lateral force reference value


planar
● ○ modal decreased value (⋅0.8)

○ ● lateral force reference value


spatial
○ ○ modal decreased value (⋅0.8)
Tutorial – VIII – 'Modal RS method'
cf. Tutorial 4.2
3-story RC frame building (residential use)
behavior factor q = 4
ground motion: agR = 0.3 g m3 = m
residential use: γI = 1.0 h k3 = k
structural parameters: m2 = 1.5m
k2 = 2k
E = 2.1 ⋅ 108 kN/m2 I = 2.679 ⋅ 10-5 m4 h
m1 = 2m
h = 3.0 m k = 12 ⋅ EI/h3 k1 = 3k
h
m = 50 tons = 50 kNs2/m

1. Setting up the differential equation of motion:

[M ]⋅ {u} + [C ]⋅ {u} + [K ]⋅ {u} = 0 if [C] = 0 : [M ]⋅ {u} + [K ]⋅ {u} = 0

 m1 0 0  2 0 0  k1 + k2 − k2 0   5 −2 0 
       
[M ] =  0 m2 0  = m ⋅  0 1.5 0  [K ] =  − k2 k 2 + k3 − k3  = k ⋅  − 2 3 − 1
0 m 3  0 0 1  0 − k3 k3   0 −1 1 
 0     
Tutorial – VIII – 'Modal RS method'
cf. Tutorial 4.2
2. Modal segmentation:

[K ] − ω 2 ⋅ [M ] = 0 ⇒ 5k − 2mω 2 − 2k 0
− 2k 3k − 1.5mω 2 −k = 0
0 −k k − mω 2

3. Modal circular frequencies ωi and periods Ti :


ω1 = 4.19 s-1 → T1 = 1.50 sec
ω2 = 8.97 s-1 → T2 = 0.70 sec
ω3 = 13.3 s-1 → T3 = 0.47 sec

4. Eigenmodes:

 0.30   − 0.676   2.47 


     
{φ1} =  0.644  {φ2 } =  − 0.601  {φ3} =  − 2.57 
 1.00   1.00   1.00 
     
Tutorial – VIII – 'Modal RS method'
cf. Tutorial 4.2
5. Modal participation factors γi :

∑m
j =1
j ⋅ φ j ,i
αi
γi = n
=
Mi *
∑m
j =1
j ⋅φ 2
j ,i

α1 = 100 ⋅ 0.3 + 75 ⋅ 0.644 + 50 ⋅ 1.0 = 128.3 kNs2/m


α2 = –100 ⋅ 0.676 – 75 ⋅ 0.601 + 50 ⋅ 1.0 = -62.7 kNs2/m
α3 = 100 ⋅ 2.47 – 75 ⋅ 2.57 + 50 ⋅ 1.0 = 104.3 kNs2/m

M1* = 100 ⋅ 0.32 + 75 ⋅ 0.6442 + 50 ⋅ 1.02 = 90.0 kNs2/m


M2* = 100 ⋅ 0.6762 + 75 ⋅ 0.6012 + 50 ⋅ 1.02 = 122.8 kNs2/m
M3* = 100 ⋅ 2.472 + 75 ⋅ 2.572 + 50 ⋅ 1.02 = 1155.0 kNs2/m

→ γ1 = 128.3 / 90.0 = 1.426


→ γ2 = -62.7 / 122.8 = –0.511
→ γ3 = 104.3 / 1155.0 = 0.090
Tutorial – VIII – 'Modal RS method'
cf. Tutorial 4.2
6. Design spectral accelerations Sa(Ti ) for each mode i :

2.5  TC  2.5  0.6 


T1 = 1.50 sec : S a ,d ( T ) = a g ⋅ S ⋅ ⋅   = ( 2.943 ⋅1.0 ) ⋅1.15 ⋅ ⋅ = 0 .846 m / s 2

q  T1  4.0 1.50 

Check: Sa,d (T) = 0.846 m/s2 ≥ β ∙ ag = 0.20 ∙ 2.943 = 0.5886 m/s2

2.5  TC  2.5  0.6 


T2 = 0.70 sec : S a ,d ( T ) = a g ⋅ S ⋅ ⋅   = ( 2.943 ⋅1.0 ) ⋅1.15 ⋅ ⋅   = 1.813 m / s 2
q  T1  4.0  0.7 

Check: Sa,d (T) = 1.813 m/s2 ≥ β ∙ ag = 0.20 ∙ 2.943 = 0.5886 m/s2

2.5 2.5
T3 = 0.47 sec : S a ,d ( T ) = a g ⋅ S ⋅ = ( 2.943 ⋅1.0 ) ⋅1.15 ⋅ = 2.115 m / s 2
q 4.0
Tutorial – VIII – 'Modal RS method'
cf. Tutorial 4.2
5. Lateral story loads Fj,i :

F j ,i = m j ⋅ φ j ,i ⋅ γ i ⋅ S a ,d ( Ti )
F3,1= 60.3

F1,1 = 100 ⋅ 0.30 ⋅ 1.426 ⋅ 0.846 = 36.2 kN F2,1 = 58.3


F2,1 = 75 ⋅ 0.644 ⋅ 1.426 ⋅ 0.846 = 58.3 kN
F1,1 = 36.2
F3,1 = 50 ⋅ 1.00 ⋅ 1.426 ⋅ 0.846 = 60.3 kN

F3,2 = –46.3

F1,2 = 100 ⋅ (–0.676) ⋅ (–0.511) ⋅ 1.813 = 62.6 kN F2,2 = 41.8


F2,2 = 75 ⋅ (–0.601) ⋅ (–0.511) ⋅ 1.813 = 41.8 kN
F1,2 = 62.6
F3,2 = 50 ⋅ 1.00 ⋅ (–0.511) ⋅ 1.813 = –46.3 kN

F3,3 = 9.5

F1,3 = 100 ⋅ 2.47 ⋅ 0.090 ⋅ 2.115 = 47.0 kN F2,3 = –36.7

F2,3 = 75 ⋅ (–2.57) ⋅ 0.090 ⋅ 2.115 = –36.7 kN


F1,3 = 47.0
F3,3 = 50 ⋅ 1.00 ⋅ 0.090 ⋅ 2.115 = 9.5 kN
Tutorial – VIII – 'Modal RS method'
cf. Tutorial 4.2
5. Maximum shear forces Fb :

60.3 -46.3 9.5 76.6

118.6 -4.5 -27.2 121.7


154.8 58.1 19.8 166.5

n
Fb , m = ∑
i =1
Fb2, m ,i
Conclusion

• Response Spectrum Analysis (RSA) is an elastic method of analysis and


lies in between equivalent force method of analysis and nonlinear
analysis methods in terms of complexity.
• RSA is based on the structural dynamics theory and can be derived from
the basic principles (e.g. Equation of motion).
• RSA, unlike equivalent force method, considers the influence of several
modes on the seismic behaviour of the building.
• Damping of the structures is inherently taken into account by using a
design (or response) spectrum with a predefined damping level.
• The maximum response of each mode is an exact solution.
• The sole approximation used in RSA is the combination of modal
responses.
Conclusion (cont.d)

• ABSSUM is the most conservative modal combination rul. Too


conservative?
• SRSS is the most popular modal combination rule due to its simplicity and
‘accuracy’ for buildings with well-seperated frequencies.
• CQC is regarded as the most viable option for any structure.

You might also like