Sensitivity of Design Spectrum on Near- Fault Directivity ......Sensitivity of Design Spectrum on Near-Fault Directivity Effects and its Implications on the Design of Common Structural

Sensitivity of Design Spectrum on Near-Fault Directivity Effects and its

Implications on the Design of Common Structural Systems

Sinan Akkar1, Aydan Seskir Özmen2 and Bahadır Şadan2

1 Earthquake Engineering Research Center, Middle East Technical

University, 06800 Ankara 2 PROTA Engineering Inc., 06550 Ankara, Turkey

EARTHQUAKE ENGINEERING

RESEARCH CENTER

MIDDLE EAST TECHNICAL

UNIVERSITY

ANKARA, TURKEY

15TH WORLD CONFERENCE ON EARTHQUAKE ENGINEERING

24-28 SEPTEMBER 2012 LISBON – PORTUGAL

Special Session: Toward Improvement of GMPEs Incorporating Physic-Based Ground Motion Simulations

Objective

o To observe near-fault directivity effects through PSHA for a set of sites that are located in the vicinity of a fictitious fault

o To compare the computed spectral amplitude factors (due to directivity) with those of UBC97

o To understand how the structural member dimensions change for common (residential/office) buildings under near-fault directivity effects

Sinan Akkar

Tools

o Earthquake Scenario

o Ground-motion model: Boore and Atkinson (2008) and Sommerville et al. (1997) + Abrahamson (2000)

o EZ-FRISK (PSHA)

o Probina – Orion (design and structural analysis)

Sinan Akkar

Earthquake Scenario o Strike-slip fault of 200 km length.

Dip angle is 90 degrees. Width is 20 km.

o Slip: 25mm/year

o Mmin: 5.0, Mmax: 7.5

o Characteristic recurrence model

o Sites are grey diamonds (12 sites: 5km difference along latitude, 50km difference along latitude)

o All sites are located on a generic rock site (VS30 = 760 m/s)

L = 200km

5km

5km

5km

50km 50km

1

2

3

4

5

6

7

8

9

10

11

12

Fictitious Strike-Slip

Fault

Sinan Akkar

Ground-motion models o Conventional GMPE is Boore and Atkinson

(2008) [BA08] o Somerville et al. (1997) + Abrahamson (2000) is

to modify BA08 for directivity effects o Frequently implemented in the current PSHA

applications o Applicable to all generic GMPEs o Directivity is consider by (a) smaller angle between

the directions of rupture propagation and waves traveling from the fault to the site, θ, (b) fraction of the fault rupture to the between the hypocenter and the site, X

o Model is valid for Mw ≥ 6.5 for absolute spectral amplitude estimations and Mw ≥ 6.0 for fault-normal to average spectral ratio.

o Model does not consider near-fault directivity effects for Rrup > 20km

o Forward directivity effects are observed for T 0.6s

Somerville et al. (1997)

Sinan Akkar

Evaluation of UBC97 Near-Source Factors

o For a given site:

o PSAforwarddirectivity / PSAnoforwarddirectivty from PSHA at

different return periods (TR) are compared with the

UBC97 near-fault factors

o PSAforwarddirectivity refers to fault-normal component

o UBC97 near-source factors are computed for

Seismic Source A (slip-rate 5mm/year). Such

faults can generate earthquakes of Mw 7.0

Sinan Akkar

1

100 1000

Ra

tio

Return Period (year)

Forward Directivity / No Forward Directivity

Site 1 Site 5

Site 9 UBC97

T = 0.6sd = 0km

1

100 1000

Ra

tio



Site 2 Site 6

Site 10 UBC97

T = 0.6sd = 5km

1

100 1000

Ra

tio



Site 3 Site 7

Site 11 UBC97

T = 0.6sd = 10km

1

100 1000

Ra

tio



Site 4 Site 8

Site 12 UBC97

T = 0.6sd = 15km

1

100 1000

Ratio



Site 1 Site 5

Site 9 UBC97

T = 0.7sd = 0km

1

100 1000

Ratio



Site 2 Site 6

Site 10 UBC97

T = 0.7sd = 5km

1

100 1000

Ratio



Site 3 Site 7

Site 11 UBC97

T = 0.7sd = 10km

1

100 1000

Ratio



Site 4 Site 8

Site 12 UBC97

T = 0.7sd = 15km

1

100 1000

Ratio



Site 1 Site 5

Site 9 UBC97

T = 0.8sd = 0km

1

100 1000

Ratio



Site 2 Site 6

Site 10 UBC97

T = 0.8sd = 5km

1

100 1000

Ratio



Site 3 Site 7

Site 11 UBC97

T = 0.8sd = 10km

1

100 1000

Ratio



Site 4 Site 8

Site 12 UBC97

T = 0.8sd = 15km

1

100 1000

Ra

tio



Site 1 Site 5

Site 9 UBC97

T = 0.9sd = 0km

1

100 1000

Ratio



Site 2 Site 6

Site 10 UBC97

T = 0.9sd = 5km

1

100 1000

Ra

tio



Site 3 Site 7

Site 11 UBC97

T = 0.9sd = 10km

1

100 1000

Ra

tio



Site 4 Site 8

Site 12 UBC97

T = 0.9sd = 15km

1

100 1000

Ra

tio



Site 1 Site 5

Site 9 UBC97

T = 1.0sd = 0km

1

100 1000

Ra

tio



Site 2 Site 6

Site 10 UBC97

T = 1.0sd = 5km

1

100 1000

Ra

tio



Site 3 Site 7

Site 11 UBC97

T = 1.0sd = 10km

1

100 1000

Ra

tio



Site 4 Site 8

Site 12 UBC97

T = 1.0sd = 15km

1

100 1000

Ratio



Site 1 Site 5

Site 9 UBC97

T = 1.2sd = 0km

1

100 1000

Ratio



Site 2 Site 6

Site 10 UBC97

T = 1.2sd = 5km

1

100 1000

Ratio



Site 3 Site 7

Site 11 UBC97

T = 1.2sd = 10km

1

100 1000

Ratio



Site 4 Site 8

Site 12 UBC97

T = 1.2sd = 15km

1

100 1000

Ratio



Site 1 Site 5

Site 9 UBC97

T = 1.4sd = 0km

1

100 1000

Ratio



Site 2 Site 6

Site 10 UBC97

T = 1.4sd = 5km

1

100 1000

Ratio



Site 3 Site 7

Site 11 UBC97

T = 1.4sd = 10km

1

100 1000

Ratio



Site 4 Site 8

Site 12 UBC97

T = 1.4sd = 15km

1

100 1000

Ratio



Site 1 Site 5

Site 9 UBC97

T = 1.8sd = 0km

1

100 1000

Ratio



Site 2 Site 6

Site 10 UBC97

T = 1.8sd = 5km

1

100 1000

Ratio



Site 3 Site 7

Site 11 UBC97

T = 1.8sd = 10km

1

100 1000

Ratio



Site 4 Site 8

Site 12 UBC97

T = 1.8sd = 15km

1

100 1000

Ra

tio



Site 1 Site 5

Site 9 UBC97

T = 2.0sd = 0km

1

100 1000

Ra

tio



Site 2 Site 6

Site 10 UBC97

T = 2.0sd = 5km

1

100 1000

Ra

tio



Site 3 Site 7

Site 11 UBC97

T = 2.0sd = 10km

1

100 1000

Ra

tio



Site 4 Site 8

Site 12 UBC97

T = 2.0sd = 15km

Sinan Akkar

Major Remarks (limited to the case study)

o UBC97 near-source factors are conservative at short distances and at short periods

o Discrepancy between the case study results and UBC97 factors decreases towards larger distances and longer periods. However, UBC97 near-source factors are still larger than those computed in the case study

Sinan Akkar

Major Conclusion

• Simplified UBC97 approach to account for near-fault effects is practical in many aspects. However, it may fail to reflect actual seismic demands

• Models adjusting non-directivity GMPEs (such as the one presented here or many others that were developed or under development) can be used to come up with simplified directivity-sensitive code spectra.

Sinan Akkar

Code-based design for buildings located in the vicinity of faults

o PSHA results at T = 0.2s and 1.0s are used to construct the design spectrum for designing a 20-story RC office building located at Site 9 (the most critical site in terms of directivity effects in this study).

o Design spectrum is 2/3 of TR = 2475 years spectrum

o Provisions of Turkish Earthquake Design code are considered in design.

o Building is designed twice by considering directivity and no-directivity effect separately.

Sinan Akkar

• Target ductility is chosen as normal ductile behavior (R, strength reduction factor, is 4 in design). Fundamental period (T1) of the building is ~ 1.9s.

• When direct code approach is used to establish the design spectrum (i.e., PSA at T = 0.2s and T = 1.0s), directivity effects cannot be identified very well (blue and red curves with respect to red-dashed curve derived from UHS)

• Thus, no change in the structural dimensions when direct code approach is used to for design spectrum

0

20

40

60

80

100

120

0.00 0.50 1.00 1.50 2.00

Spectr

al D

ispla

cem

ent

(cm

)

Period (s)

TR = 2475 yrs, Site 9

Code-NoFD

Code-FD

UHS-FD

Sinan Akkar

Performance under Directivity (FD) and No-directivity (NoFD) spectra of TR = 2475 yrs

≈ %

1 u

nd

er C

od

e-N

oFD

(T R

= 2

47

5 y

rs)

Sinan Akkar

≈ %

1.1

un

der

Co

de

-FD

(T R

= 2

47

5 y

rs)

≈ %

1.5

un

der

UH

S-FD

(T R

= 2

47

5 y

rs)

Roof drift is ≈ 1% under code-based spectra with and without directivity

Roof drift is ≈ 1.5% under UHS with directivity

Even if the building is not designed for the demands described by UHS, the building does not collapse due to its reserved capacity

Final Remarks

o Definition of near-fault directivity effects for code-based spectrum requires further studies.

o A close collaboration between engineers and seismologists is needed for these studies to consider the complexity of the problem from both engineering and seismological points

Sinan Akkar

Thank you

Documents

Sensitivity of Design Spectrum on Near- Fault Directivity ......Sensitivity of Design Spectrum on Near-Fault Directivity Effects and its Implications on the Design of Common Structural