1. consulting engineers and scientists
Site-Specific Risk-Targeted Ground Motion Procedures Jorge F. Meneses, PhD, PE, GE, D.GE, F.ASCE Carlsbad, California
AEG Inland Empire Chapter Continuing Education Series
May 31, 2014

2. Outline Overview Site-specific procedures Risk coefficient
NGA Relationships
Deaggregation
Examples
Performance Based EE
Summary

3. Source, Path and Site

4. Evaluating Seismic Hazard and Ground Motions

5. SITE-SPECIFIC STUDY
2103 CBC, , 1616A.1.3
“For buildings assigned to Seismic Design Category E and F, or when required by the building official, a ground motion hazard analysis shall be performed in accordance with ASCE 7 Chapter 21, as modified by Section 1803A.6 of this code.”

6. SITE-SPECIFIC STUDY (cont’d)
Site Response Analysis
Structures on Site Class F sites (Ts > 0.5 seconds)
At least 5 recorded or simulated horizontal ground motion acceleration time histories (MCER spectrum at bedrock)
Seismic Hazard Analysis
Seismically isolated structures (S1  0.6)
Structures with damping systems (S1  0.6)
A time history response analysis of the building is
performed
(ASCE 7-10, Section , p.67)

7. SITE RESPONSE ANALYSIS
Ground
Surface
Rock
base
(ASCE 7-10, Section 21.1, p.207)

8. Probabilistic ground motion
SITE-SPECIFIC GROUND MOTION PROCEDURE
Probabilistic ground motion
Method 1: Uniform-hazard GM * Risk Coefficient
Method 2: Risk-targeted probabilistic GM directly
Deterministic ground motion
84th-%ile GM, but not < 1.5Fa or 0.6*Fv/T
MCER = Min (Prob. GM, Det. GM)
All GMs are max-direction spectral accelerations (Sa)
(Sections 21.2, 21.3, and 21.4)

9. Risk coefficient
Risk coefficient: CR
T ≤ 0.2 s; CR = CRS (Figure 22-17)
T ≥ 1.0 s; CR = CR1 (Figure 22-18)
0.2 s ≤ T ≤ 1.0 s; CR linear interpolation of CRS and CR1

10. Risk Coefficient

11. (direction of max horiz resp)
SITE-SPECIFIC GROUND MOTION PROCEDURE
Prob MCER
Det MCER
MCER
Spectrum
DESIGN
SITE-SPECIFIC
DESIGN SPECTRUM
General DESIGN
General MCE
Site Coord
Site Class
1% Prob. of collapse 50 yr
(direction of max horiz resp)
Lesser of PSHA and DSHA
2/3 MCE Spectrum
> 80% General Design Spectrum
84th percentile
(direction of max horiz resp)
(Sections 21.2, 21.3, and 21.4)

12. SITE-SPECIFIC GROUND MOTION PROCEDURE
Deterministic Lower Limit (DLL) on MCER Spectrum
1.5 Fa
0.6 Fa
Sa = 0.6 Fv/T
0.08 Fv/Fa
0.4 Fv/Fa TL
Period (seconds)
Sa (g)
Sa = 0.6 Fv TL/T2
(Section , p. 209)

13. Attenuation Relationships
Several types of ground motions parameters can be calculated from a recorded EQ time history.
But what do you do if you want to estimate what the ground motion parameters are going to be from an earthquake that hasn’t happened yet?

14. Attenuation Relationships
ANSWER:
Use the data that we’ve collected so far and fit equations to them for predicting future ground motions.
These equations are often called
attenuation relationships.

15. Attenuation Relationships
Ground Motion Parameter
Distance from Source
Initial relationships were just based on Magnitude (M) and Distance (R), but equations become much more complex as researchers looked for ways to
minimize data scatter.

16. Attenuation Relationships
Modern attenuation relationships have terms that deal with such complexities as:
1) Fault type
2) Fault geometry
Pretty complex …. Hard to do by hand!!
3) Hanging wall/Foot wall
4) Site response effects
5) Basin effects
6) Main shock vs. After shock effects

17. Attenuation Relationships
Ideally, every geographic area that experiences EQs would have its own set of attenuation relationships. WHY?
Scatter in the data could be minimized!
…But we can’t really produce site-specific attenuation relationships for places other than those that have a lot of frequent earthquakes. WHY?
Not enough recorded data!
So we start combining earthquake records from geographically different areas with the assumption that the ground motions should be similar despite the differences in location.
Ergodic Assumption

18. NGA=Next Generation “Attenuation” Relations
Three NGA projects:
For active crustal Eqs (California, Middle East, Japan, Taiwan,…): NGA-West
For subduction Eqs (US Pacific Northwest and northern California, Japan, Chile, Peru,…): NGA-Sub
Stable continental regions (Central and Eastern US, portion of Europe, South Africa,…): NGA-East

19. Attenuation Relationships (GMPEs)
For crustal faults in the Western US and other high- to moderate- seismicity areas, most professionals currently use:
Next Generation Attenuation Relationships (NGAs)
NGA West 1: 5 separate research teams were given the same set of ground motion data and were asked to develop relationships to fit the data. Their results were published in 2008.
-Abrahamson & Silva
-Boore & Atkinson
-Chiou & Youngs
-Idriss
(rock only)
-Campbell & Bozorgnia

20. NGA-West NGA-West 1: 2008 NGA-West 2: 2014 Data set No. EQs No. Rec
Sa Type
Damping
(%)
Periods (sec)
NGA-West 1
173
3,551
AR, GMRotI50 5
NGA-West 2
610
21,331
AR, RotDnn
AR= as-recorded

21. Rotate horizontal components, at each period compute:
RotDnn
Rotate horizontal components, at each period compute:
RotD50 = 50 percentile
RotD100 = max
RotD00 = min
RotD50 is the main intensity measure
PGA, PGV and Sa at 21 periods: 0.01, 0.02,……,5, 7.5, 10 sec
No GMPE for PGD

22. NGA West-2 ranges of applicability
Applicable magnitude range:
M ≤ 8.5 for strike-slip (SS)
M ≤ 8.0 for reverse (RV)
M ≤ 7.5 for normal faults (NM)
Applicable distance range:
0 – 200 km (preferably 300km)

23. Horizontal NGA-West 2 GMPEs parametersAS
BSSA CB CY I
Magnitude Mw
Top of rupture
Ztor
Style of faulting
RV, NM, SS
Dip
Yes
Downdip fault width
Closest distance to rupture
Rrup
Hor. dist. to surface proj.
Rjb
Hor. dist. Perpendicular to strike
Rx, Ry Rx
Hanging wall model
(Rjb)
Vs30
(760m/s)
Vs30, (Sj)
Vs30≥450
Depth to Vs
Z1.0
Z2.5
Hypocentral depth
Hhyp
Vs30 for reference rock (m/s)
1,100
760
1,130

24. NGA West 2 Five models
Abrahamson-Silva-Kamai (ASK)
Boore-Stewart-Seyhan-Atkinson (BSSA)
Campbell-Bozorgnia (CB)
Chiou-Youngs (CY)
Idriss (I)

25. NGA Distance Notations
𝑅 𝑅𝑢𝑝 = Closest distance to rupturing fault plane 𝑅 𝐽𝐵 = Boore−Joyner distance 𝑅 𝑋 = Closest horizontal distance to the top of rupture

26. More on distances
Geotechnical Services Design Manual, Version 1.0, 2009, Caltrans
Development of the Caltrans Deterministic PGA Map and Caltrans ARS Online, 2009, Caltrans

27. NGA Soil vs. Rock
NGA equations don’t have a “trigger” for soil or rock. They just rely on the VS30, which is the average shear wave velocity in the upper 30 meters of the ground.
VS30 (m/s) Type Site Class
>1500
Hard Rock A
Firm Rock B
Soft Rock C
Regular Soil D
<180
Soft Soil E

28. NGA West 2 Excel spreadsheet

29. 2013 CBC, Section 1803A.6 Geohazard Reports
The three Next Generation Attenuation (NGA) relations used for the 2008 USGS seismic hazard maps for Western United States (WUS) shall be utilized to determine the site-specific ground motion. When supported by data and analysis, other NGA relations, that were not used for the 2008 USGS maps, shall be permitted as additions or substitutions. No fewer than three NGA relations shall be utilized
2008 USGS
Boore and Atkinson (2008)
Campbell and Bozorgnia (2008)
Chiou and Youngs (2008)

30. Not an average velocity in upper 30 m
What is Vs30?
Not an average velocity in upper 30 m
Ratio of 30 m to shear wave travel time
(Stewart 2011)

31. Not an average velocity in upper 30 m
What is Vs30?
Not an average velocity in upper 30 m
Ratio of 30 m to shear wave travel time
(Stewart 2011)

32. Not an average velocity in upper 30 m
What is Vs30?
Not an average velocity in upper 30 m
Ratio of 30 m to shear wave travel time
Emphasizes low Vs layers
Travel time is proportional to slowness.
(Stewart 2011)

33. Seismic Source Interpretation from PSHA Results
Deaggregation:
Break the probabilistic “aggregation” back down to individual contributions based on magnitude and distance.
Provides:
- Mean M,R: weighted average
- Modal M,R: Greatest single contribution to hazard

34. Risk-Targeted MCER Probabilistic Response Spectrum
CRS = 0.941
CR1 = 0.906

35. Deterministic MCER Response Spectra

36. Site-Specific MCER Response Spectrum

37. Design Response Spectrum

38. Site-Specific Response Spectra at Ground Surface

39. Site-specific MCE geometric mean (MCEG) PGA
PROB MCEG PGA
The probabilistic geometric mean PGA shall be taken as the geometric mean PGA with a 2% PE in 50 years
DETERMINISTIC MCEG PGA
Calculated as the largest 84th percentile geometric mean PGA for characteristic earthquakes on all known active faults. Minimum value 0.5 FPGA (FPGA at PGA=0.5g)
SITE-SPECIFIC MCEG PGA
Lesser of probabilistic and deterministic
MCEG PGA ≥ 0.80 PGAM
(Section 21.5)

40. Site-specific Probabilistic MCER
SITE-SPECIFIC GROUND MOTION PROCEDURE
Site-specific Probabilistic MCER
(1% probability of collapse in 50 years)
METHOD 1
CR * Sa (2% PE 50 year)
METHOD 2
From iterative integration of a
site-specific hazard curve with a
lognormal probability density function
representing the collapse fragility
CR = risk coefficient (from maps)
T ≤ 0.2s CR = CRS
T ≥ 1.0s CR = CR1
0.2s < T < 1s Linear interp CRS and CR1
(i.e., probability of collapse as a function of Sa)
Collapse fragility with
10% Prob. of collapse;
logarithmic std dev of 0.6
(Section )

41. Performance-Based Earthquake Engineering
Risk is computed using a Do you remember the concept of probabilistic seismic hazard analysis?
Probabilistic framework
All possible distances are considered - contribution of each is weighted by its probability of occurrence
PSHA Review…..
Basic equation:
All possible magnitudes are considered - contribution of each is weighted by its probability of occurrence
All sources and
their rates of
recurrence are considered
All possible effects are considered - each weighted by its conditional probability of occurrence

42. Performance-Based Earthquake Engineering
Pacific Earthquake Engineering Research Center (PEER) developed a probabilistic framework for considering the engineering effects from EQ ground motions:
Engineering demand parameter, EDP
Damage measure, DM
Decision variable, DV
Intensity measure, IM
PGA
PGV IA
CAV
Repair Cost
Lives Lost
Down Time
Pile Deflection
Cracking
Collapse Potential
FSliq
Lateral Spread
Settlement
Story Drift

43. Fragility Curves Example of Fragility curves P[D > di | PGA] PGA
EDP = Displacement = D
IM = PGA
P[D > di | PGA] 1
P[D > 1.0 | PGA=0.3g]
0.3g
1.0cm
2.0cm
P[D > 2.0 | PGA=0.3g]
3.0cm
P[D > 3.0 | PGA=0.3g]
0.0
PGA

44. Fragility Curves and Seismic Hazard Curves
The PEER performance-based framework incorporates seismic hazard curves and fragility curves. Convolving a fragility curve with a seismic hazard curve produces a single point on a new hazard curve!!
Risk curve – lDV vs DV
Fragility curve – DV given DM
Fragility curve – DM given EDP
Fragility curve – EDP given IM
Seismic hazard curve for IM (from PSHA)

45. lD proportional to sum of thick red lines
Fragility Curves and
Seismic Hazard Curves
1.0
P[D>di| PGA]
Fragility curve for D > 2.0cm
lD proportional to sum of thick red lines
0.0
PGA
lPGA
DlPGA
Hazard curve
PGA

46. Seismic hazard curve for Displacement
Fragility Curves and Seismic Hazard Curves
1.0
P[D>di| PGA]
lD proportional to sum of probabilities
Fragility curve lD D IM
Seismic hazard curve for Displacement
0.0
PGA
lPGA
D=2.0cm
DlPGA
Hazard curve IM
PGA

47. Risk-targeted ground motions
(Luco 2009)

48. Risk-targeted ground motions
(Luco 2009)

49. Risk-targeted ground motions - Example

50. Summary of differences
ASCE 7-05
ASCE 7-10
Name
MCE
MCER
Probabilistic GMs (objective)
Uniform hazard
(2%-in-50 yr Pr. of Exc.)
Risk targeted
(1%-in-50 yr Pr. of Collapse)
Deterministic GMs
1.5*median
84%-ile
(approx. 1.8*median)
GM parameter
Geometric mean, Sa
Maximum direction, Sa
USGS web tool
Java ground motion parameter calculator
Seismic design maps web application
Average SDS
0.73g
0.72g
Average SD1
0.38g
0.40g
(Luco 2009)

51. For further information
Contact Jorge F. Meneses, PhD, PE, GE, D.GE, F.ASCE (760)