Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Filling in for “Record Selection Insights from the Study of
Epsilon and Inelastic Displacements: Use with Deterministic Code Applications”
(C. Allin Cornell)
Nicolas LucoResearch Structural Engineer
U.S. Geological Survey (Golden, CO)
COSMOS Technical SessionMillbrae, CA
18 November 2005
Preface
Most appropriate definition of target?1) Building code design response spectrum (+ M, R, etc.) ?
2) Uniform Hazard Response Spectrum (UHRS) ?
3) Sa(T1), deaggregated M, R, etc. ?
4) Sa(T1), deaggregated M, R, and ε, etc.?
5) Inelastic Spectral Displacement ?
Different targets can lead to different recommendations for selection/scaling, some easier to apply than others.
Last year the focus was bias associated with scaling for Option #3, this year it’s #4, next year #5?
Same experiments need to be repeated for Option #1 (i.e., bias in nonlinear dynamic structural analysis induced by code procedures?)
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
SF = 2
Bias = 1.3
SF = 0.1
Bias = 0.4
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
Intra-bin scaling results in term of θmax for a Three-Story Frame Model(from Tothong & Cornell)
10-1
100
101
100
Sde: T=0.60
a= 1.04 (0.00)
b= 0.10 (0.00)
]B
ias θ m
ax [s
cale
d/ u
nsca
led]
10-1
100
101
100
Sdi: T=0.60,dy=0.65
[ a= 0.99 (0.06)
b= 0.02 (0.00)
Scale Factor
Does Using Only Higher Epsilon Records Reduce Inter-bin Scaling Bias?
(from Baker & Cornell)
Magnitude 7+/- Random Set Wide M, R Range Set; Epsilon > 1.5 Only
1 Sec Bilinear Oscillator, R factor = 4
10-1
100
101
10-1
100
101
Scale Factor
Sca
led/
Uns
cale
d D
ispl
acem
ent
10-1
100
101
10-1
100
101
Scale Factor
Sca
led/
Uns
cale
d D
ispl
acem
ent
Consideration of Spectral Shape
Considering not only the target Sa(T1) but also the “target spectral shape” …
Target Sa = 2.0g
Consideration of Spectral Shape
… and distinguishing the scaled earthquake records with spectral shapes “most similar” to the target …
Consideration of Spectral Shape
Little bias is induced, even at relatively large scale factors:
Consideration of Spectral Shape
Little bias is induced, even at relatively large scale factors:
2) The Mw and Rclose that contribute most to the occurrence probability of the target Sa(T1) can then be found via “banded deaggregation” (a new USGS website):
Mw=7.9, Rclose=42km
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
3) With the “target” Mw and Rclose, the median spectral shape can be estimated via an empirical ground motion prediction equation:
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
10-2
10-1
100
10-1
100
San Andreas Scenario: Mw = 7.9 , Rclose= 42km
Period, T [sec]
Spec
tral A
ccel
erat
ion,
Sa(
T )
[g]
Median Response SpectrumTarget Sa(T1)
ε(T1)
2) The Mw and Rclose that contribute most to the occurrence probability of the target Sa(T1) can then be found via “banded deaggregation” (a new USGS website):
Mw=7.9, Rclose=42kmMw=7.9, Rclose=42km, ε = 1.1
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
4) With Mw , Rclose , and ε(T1), the expected spectral shape that passes through Sa(T1) can be calculated:
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
)()]([)](),([)]([)](|)([ :,Given 111 TTSTTTSETTSERM aaaclosew εσεερε ⋅⋅+=
10-2
10-1
100
10-1
100
San Andreas Scenario: Mw = 7.9 , Rclose= 42km
Period, T [sec]
Spec
tral A
ccel
erat
ion,
Sa(
T )
[g]
Median Response SpectrumTarget Sa(T1)
Expected Response Spectrum
From, for example,Baker & Cornell (2005)
In short, it can be important to consider ε(T1) in addition to Mw and Rclose in quantifying the expected (or target) spectral shape:
10-1
100
10-1
100 San Andreas and San Gregorio Scenarios
Period, T [sec]
Spec
tral A
ccel
erat
ion,
Sa(
T )
[g]
San Andreas (Median)San Andreas (Expected)
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
2) The Mw and Rclose that contribute most to the occurrence probability of the target Sa(T1) can then be found via “banded deaggregation” (a new USGS website):
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
In short, it can be important to consider ε(T1) in addition to Mw and Rclose in quantifying the expected (or target) spectral shape:
10-1
100
10-1
100 San Andreas and San Gregorio Scenarios
Period, T [sec]
Spec
tral A
ccel
erat
ion,
Sa(
T )
[g]
San Andreas (Median)San Andreas (Expected)
10-1
100
10-1
100 San Andreas and San Gregorio Scenarios
Period, T [sec]
Spec
tral A
ccel
erat
ion,
Sa(
T )
[g]
San Andreas (Median)San Andreas (Expected)San Gregorio (Median)San Gregorio (Expected)
Setting Mw , Rclose , Sa(T1) , and Spectral Shape
10-1
100
10-1
100 San Andreas and San Gregorio Scenarios
Period, T [sec]
Spec
tral A
ccel
erat
ion,
Sa(
T )
[g]
San Andreas (Median)San Andreas (Expected)San Gregorio (Median)San Gregorio (Expected)Uniform Hazard
More questions (without answers) …
What about MDOF structures, and real multi-component ground motions? … vector hazard?
Can we use “risk deaggregation” to defined target ground motion scenarios, even for building code applications?
Additional Slides …
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
Example …
Ground motion scenarioMw = 6.5 - 6.9, Rclose = 0 - 16 kmForward-directivity conditionsStrike-normal componentNEHRP C or D site classificationShallow crustal eventTarget Sa = 2.0g
StructureSDOF oscillator (e.g., bridge bent, 1-story building)T = 1 sec, ζ = 5%R = 4 ( i.e., Fy = Target Sa * m / 4 )Bilinear hysteretic behavior with α = 2%
"Near-Source" Scenario
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
Target Sa = 2.0g
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
Target Sa = 0.07g
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
2004 AIR Worldwide Corporation
Intra-Bin Scaling for Nonlinear Oscillator
SF = 2
Bias = 1.3
SF = 0.1
Bias = 0.4