Revised Probabilistic Seismic Hazard Map of Turkey and Its Implications in Seismic Design

Sinan Akkar

Boğaziçi University

Kandilli Observatory and Earthquake Research Institute

34684 Çengelköy İstanbul

New Generation Seismic Codes and New Technologies in Earthquake Engineering 26-27 Februray 2015 – Ankara, Turkey

Outline

• Revised Turkish seismic hazard map

• Observations for design ground-motion definition • Comparisons of old and new design spectral ordinates

• Damping scaling factors

• Long-period spectral corner period (TL)

• Vertical-to-horizontal spectral ratios

• Near-fault effects

Sinan Akkar

Revision of Turkish seismic hazard map project

Sinan Akkar

S. Akkar, T. Eroğlu Azak, T. Çan, U. Çeken, M.B. Demircioğlu, T. Duman, S. Ergintav, T.F. Kadirioğlu, D. Kalafat, Ö. Kale, R.F. Kartal, T. Kılıç, S. Özalp, K. Şeşetyan, S. Tekin, A. Yakut, M.T. Yılmaz, M.

Utkucu, Ö. Zülfikar

A multi-institutional project funded by the Disaster and Emergency Management Presidency (AFAD) and Turkish

Catastrophe Insurance Pool (TCIP)

• Revise seismic hazard maps at the national level based on the recent state-of-the-art developments and findings in this field (Turkey and worldwide)

• Provide spectral ordinates (PGA, SA at T = 0.2s and 1.0s) for return periods of 43 years (69%/50 years), 72 years (50%/50 years), 475 years (10%/50 years) and 2475 years (2%/50 years) for their use in the definition of updated code spectra and insurance premiums

• Provide additional information on vertical-to-horizontal spectral ordinate ratios, long-period corner period, damping scaling factors and forward directivity effects

Sinan Akkar

Earthquake Catalog

Sinan Akkar

• Compilation of instrumental catalog (12674 earthquakes) is based on national / international catalogs

• Historical catalog (512 earthquakes) is compiled from the recently finished GEM-Historical Catalog, SHARE and EMME projects

• Minimum magnitude bound is 4 • Homogenized magnitudes (Mw) through empirical

conversion equations developed from the compiled catalog

• Declustering and completness analyses

Seismic Sources

Sinan Akkar

Literature review, already finished national and international projects, earthquake catalogs, GIS maps to determine active fault segments, area sources, maximum magnitudes, slip rates, geometries, style-of-faulting, depth distribution etc.

553 fault segments

Active faults in mainland Turkey

Active faults around mainland Turkey

200 km buffer zone

Area sources in and around mainland Turkey

SOURCE CHARACTERIZATION

Sinan Akkar

AREA SOURCE (AS)

FAULT + BACKGROUND (FS)

GROUND-MOTION CHARACTERIZATION

Sinan Akkar

SHALLOW ACTIVE CRUSTAL SEISMICITY

Akkar et al. (2014) – 0.3 Akkar and Çağnan (2014) – 0.3 Chiou and Youngs (2008) – 0.3 Zhao et al. (2006) – 0.1

INTERFACE AND INSLAB SEISMICITY

Zhao et al. (2006) – 0.4 Lin and Lee (2008) – 0.2 Atkinson and Boore (2003) – 0.2 Youngs et al. (2006) – 0.2

• Compute hazard for each seismic source model considering the GMPEs and seismic source logic-trees.

• Combine the results with alternative source model weights

Sinan Akkar

PGA – 475 yrs (10% in 50 yrs) SA at 0.2s – 475 yrs (10% in 50 yrs) SA at 1.0s – 475 yrs (10% in 50 yrs)

For TR = 475 yrs (10%/50) Increase by 50% for TR = 2475 yrs (2%/50). Reduce by 50% for TR = 72 yrs (50%/50)

Site Classes TA TB

Earthquake Zone

SA = A0S(T)

Current Turkish design spectrum Proposed Turkish design spectrum

Sinan Akkar

Design: 10% probability of exceedance in 50 years (TR = 475 yrs) Immediate Occupancy, Life Safety and Collapse Prevention performances: 50% to 2% probability of exceedance in 50 years

= 8s

Comparisons of design ground motion

Sinan Akkar

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

Spec

tral

Acc

eler

atio

n (

g)

Period (s)

Ataşehir - İstanbul TEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

0

20

40

60

80

100

120

140

0 1 2 3 4 5

Spec

tral

Dis

pla

cem

ent

(cm

)

Period (s)

Ataşehir - İstanbul TEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

Spec

tral

Acc

eler

atio

n (

g)

Period (s)

Maslak - İstanbul TEC07 - 10%/50yrsTEC07 - 2%/50yrsNew Map - 10%/50yrsNew Map - 2%/50yrs

0

20

40

60

80

100

120

140

0 1 2 3 4 5

Spec

tral

Dis

pla

cem

ent

(cm

)

Period (s)

Maslak - İstanbul TEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

Spec

tral

Acc

eler

atio

n (

g)

Period (s)

BursaTEC07 - 10%/50yrsTEC07 - 2%/50yrsNew Map - 10%/50yrsNew Map - 2%/50yrs

0

20

40

60

80

100

120

140

0 1 2 3 4 5

Spec

tral

Dis

pla

cem

ent

(cm

)

Period (s)

BursaTEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

Spec

tral

Acc

eler

atio

n (

g)

Period (s)

EskişehirTEC07 - 10%/50yrsTEC07 - 2%/50yrsNew Map - 10%/50yrsNew Map - 2%/50yrs

0

20

40

60

80

100

120

140

0 1 2 3 4 5

Spec

tral

Dis

pla

cem

ent

(cm

)

Period (s)

EskişehirTEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

Spec

tral

Acc

eler

atio

n (

g)

Period (s)

İzmirTEC07 - 10%/50yrsTEC07 - 2%/50yrsNew Map - 10%/50yrsNew Map - 2%/50yrs

0

20

40

60

80

100

120

140

0 1 2 3 4 5

Spec

tral

Dis

pla

cem

ent

(cm

)

Period (s)

İzmirTEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

0.0

0.5

1.0

1.5

2.0

2.5

0 1 2 3 4 5

Spec

tral

Acc

eler

atio

n (

g)

Period (s)

AntalyaTEC07 - 10%/50yrsTEC07 - 2%/50yrsNew Map - 10%/50yrsNew Map - 2%/50yrs

0

20

40

60

80

100

120

140

0 1 2 3 4 5

Spec

tral

Dis

pla

cem

ent

(cm

)

Period (s)

AntalyaTEC07 - 10%/50yrs

TEC07 - 2%/50yrs

New Map - 10%/50yrs

New Map - 2%/50yrs

• New maps generally yield smaller seismic demands

• Spectral shape provided in TEC 2007 results in large differences in the long-period range

• Ad-hoc 1.5 factor to scale 2475-year spectrum from 475-year spectrum is not realistic. Scaling between 2475-year and 475-year spectra depends on the seismic activity of the region

PGA – TR2475/TR475 SA@T=0.2s – TR2475/TR475 SA@T=1.0s – TR2475/TR475

Spectral ratio distribution between 2475-year and 475-year spectral ordinates

Spectral ratio between 2475-year and 475-year spectral ordinates varies between 1.6 and 3.5. It increases towards

seismically less active regions Sinan Akkar

Dependency of SA2475/SA475 on seismic activity may indicate…

Performance assessment for collapse prevention: [(Seismic Demand)2475/(Seismic Demand)475]high seismicity

< [(Seismic Demand)2475/(Seismic Demand)475]low seismicity

For collapse prevention performance assessment: Buildings located in highly seismic regions will be subjected to lesser seismic demands than those located in low-seismicity regions with respect to their design strength capacities.

Sinan Akkar

Moreover…

The exceedance probabilities of ground motions (spectral accelerations) do not consider the uncertainties in the seismic behavior of structures (structural capacity). Thus, the next step in code-based work should be the definition of design ground motions for a uniform risk according to the performance objective: risk-targeted adjustment of ground motions

Fragility Risk

fcapacity (c) = capacity probability density function

P(SA > c) = annual probability of SA > c Hazard Sinan Akkar

Damping scaling factors (DSFs)

currently expressed as predictive models in terms of magnitude, source-to-site distance, site conditions, faulting style etc)

Requires simplifications for their effective use in the codes

Sinan Akkar

0

0.5

1

1.5

2

0.01 0.1 1 10

DSF

Period (s)

M5.5, VS30 = 255 m/s, SS, x = 1%

5.5, 1, 255 5.5, 5, 255 5.5, 10, 255

5.5, 20, 255 5.5, 30, 255 5.5, 40, 255

5.5, 50, 255 5.5, 75, 255 5.5, 100, 255

0

0.5

1

1.5

2

0.01 0.1 1 10

DSF

Period (s)

M6.5, VS30 = 255 m/s, SS, x = 1%

6.5, 1, 255 6.5, 5, 255 6.5, 10, 255

6.5, 20, 255 6.5, 30, 255 6.5, 40, 255

6.5, 50, 255 6.5, 75, 255 6.5, 100, 255

0

0.5

1

1.5

2

0.01 0.1 1 10

DSF

Period (s)

M7.5, VS30 = 255 m/s, SS, x = 1%

7.5, 1, 255 7.5, 5, 255 7.5, 10, 255

7.5, 20, 255 7.5, 30, 255 7.5, 40, 255

7.5, 50, 255 7.5, 75, 255 7.5, 100, 255

DSFs for x = 1% for different magnitudes and distances

No significant differences in DSF variation for different distances

Sinan Akkar

Magnitude dependency of DSFs can also be averaged out for their implementation to the codes

0

0.5

1

1.5

2

0.01 0.1 1 10

DSF

Period (s)

x = 1%

M5.5 M6.5

M7.5 Average

0

0.5

1

1.5

2

0.01 0.1 1 10

DSF

Period (s)

x = 1% x = 3% x = 7% x = 10%

x = 20% x = 30% x = 50%

Sinan Akkar

0

0.5

1

1.5

2

0.01 0.1 1 10

DSF

Period (s)

x = 1% x = 3% x = 7% x = 10%

x = 20% x = 30% x = 50%

1%

3%

7%

10%

20% 30%

50%

Horizontal-to-vertical spectrum ratio

Vertical spectrum should be consistent with the horizontal spectrum

Develop the vertical spectrum from the already defined horizontal spectrum

Sinan Akkar

Behavior of vertical spectrum is different than the horizontal spectrum: The vertical constant

acceleration plateau is shorter than its horizontal counterpart

Short-period corner period as well as the decaying branch is sensitive to the variations in the short-period vertical spectral ordinates and long-period horizontal spectral ordinates

• Compute Sav,0.1s and Sav,1.0s from Sah,0.2s and Sah,1.0s:

SAhorizontal,max

(g)

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

SA

ver

tical

,max

(g)

0.0

0.2

0.4

0.6

0.8

1.0

SAhorizontal,T1s

(g)

0.00 0.10 0.20 0.30 0.40 0.50

SA

ver

tical

,T1s (

g)

0.00

0.05

0.10

0.15

0.20

SAv=a*SAhb T=0.2s T=1s

a 0.6682 0.4681

b 1.0897 0.9609

Empirical expressions developed from many earthquake scenarios

• Define Cv and CL from computed Sav,0.1s and Sav,1.0s as well as from Sah,0.2s: Vertical DS T0V TSV n

CL ≥ 0.80 0.05 0.15 0.90

0.75 ≤ CL < 0.80 0.05 0.17 0.85

0.70 ≤ CL < 0.75 0.05 0.20 0.80

0.60 ≤ CL < 0.70 0.05 0.25 0.75

0.50 ≤ CL < 0.60 0.05 0.30 0.70

CL < 0.50 0.05 0.30 0.60

CL describes the width of const. acc. plateau and slope of decaying branch in vertical spectrum

Sah,0.2s (g) Sah,1.0s (g)

Sav,

1.0

s (g)

Sav,

0.1

s (g)

Sinan Akkar

-1

-1

-1

Sp

ektr

al

ivm

e

T0

TS T

L

Spe

ctra

l Acc

eler

atio

n

TA TB TL

Acc. Vel. Disp.

Important for long-period structures that are more sensitive to deformation demands

Long-period spectral corner period

Important remark: Large magnitude events are richer in long-period ground-motion components. Thus, TL is sensitive to magnitude:

TL Magnitude

Sinan Akkar

Near-fault (forward directivity)

Rupture velocity Particle shear wave velocity

Most of the seismic energy arrives in a short-duration waveform with a large

amplitude single pulse

Long-duration waveform with low amplitudes

Sinan Akkar

Implementation to design-based spectra – 475 years

Prob. dist. of Sa considering the occurrence and non-occurrence of pulse-like ground motion

Prob. dist. of m, r, tp

Pattern used to investigate the influence of forward directivity on code-based design spectrum

Fault with different rupture lengths and slip rates to mimic seismic activity

30

km

@ 5

km

Sinan Akkar

Sinan Akkar

Implementation to design-based spectra – 2475 years

Rx/L

Ry/

L

Amp2475/Amp475 regions

In terms of period, slip rate and Mchar

Factor 1

Factor 2

Factor 3

Determine the regions where simplified factors apply

Simplification

Spectra with NF effects

Sinan Akkar

Success of the proposed model for NF effects

Conclusive remarks

• Revision of Turkish seismic hazard map project will be the basis of new design spectrum

• The project helped to do many important observations and findings for engineering use

• Many more can be done…

• These observations can be elaborated further for their use in the definition of design ground motion in the new Turkish code

Sinan Akkar

Thank you

Sinan Akkar