UNIT 4
PROBABILISTIC SEISMIC HAZARD ANALYSIS
• Deterministic and probabilistic seismic hazard methods
• Types of earthquake sources-point, line and areal sources
• Epistemic and aleatory uncertainty estimation
• De-aggregation, logic tree
• Hazard estimation at the bedrock level
• Design specifications for structures and foundations under seismic loading.
EARTHQUAKE CAN OCCUR ON
ANY FAULT IN FUTURE
DESIGN FOR MAXIMUM VALUE
– Too Costly
DESIGN FOR MINIMUM VALUE
– Unsafe
How to Take a decision ?
Nuclear reactors, Historical
monuments - life time is high
Apartments – Life time (50 Yrs)
Design Value – Probability or
Frequency
Cover the Risk - Insurance
Seismic Hazard Curve for Mumbai city
(Requirement for Engineers)
SEISMIC HAZARD LEVES AND
PROBABILITY OF OCCURANCE
BUILDING PERFORMANCE
Hazard mapping
Based on statistical analysis of the seismicity in an area
Hazard mapping
Based on scenarios expected for the future
SEISMIC HAZARD CURVE
Magnitude
GROUND MOTION
Distance from the seismic source
What is the Frequency of Occurrence of earthquakes ?
Frequency of 6 magnitude earthquake on a particular seismic source?
Frequency of 7 magnitude earthquake ?
Frequency of 8 ?
Activity rate of ith fault
Magnitude
Ground motion Location of fault
K = Number of faults
Strain build up
CAUSES OF EARTHQUAKES
Earthquakes occur naturally
as a consequence of the
cooling of the Earth.
MANTLE CONVECTION
A single convection cycle
takes on the order of 50
Million Years to 200 Million
Years
PLATE TECTONICS
2 - 5 Cm / Yr
Mw 7.8 Indonesia earthquake Mw 7.8 – 90 cm in few
seconds
Mw 8.8 – 400 cm in few
Mw 8.8 Chile earthquake seconds
Individual Fault Slip Rate ?
MULTIPLE TIME AND SPATIAL SCALES ARE INVOLVED
NO MECHANISTIC MODEL TILL DATE HAS BEEN ABLE TO SIMULATE SEISMIC ACTIVITY
EMPIRICAL MODELS – FREQUENCY OF OCCURENCE OF EARTHQUAKES
[Link]
Assembled
catalogue
2474 BC -2008
38,860 events
The uncertainty in the
reported magnitudes is
taken as 0.5 in the
extreme part. For the
complete part the
magnitude uncertainty is
assumed to be 0.3.
PSHA METHODOLOGY
1. Identification of seismic sources and geological
zones
2. Earthquake catalogue
3. Recurrence relationships
4. Strong motion attenuation relationships
5. Seismic hazard analysis
Epistemic Uncertainty and Aleatory Variability
• Epistemic uncertainty refers to (potentially reducible) lack of knowledge about models
for the state of nature. Aleatory variability refers to random outcomes from natural
variability in a process (for a given model formulation of the process).
• If a parameter sometimes has one value and sometimes has another (e.g., the
outcome from a roll of a die, or the IM value observed from earthquakes described by
a set of rupture parameters), it has aleatory variability.
• If a parameter either has one value or another, and we do not know which value is
correct (e.g., the probability that a die roll will be a 3 in the previous example, or the
mmax for a seismic source), then the parameter has epistemic uncertainty.
• While the distinction between these two is straightforward in many cases, in hazard
calculations the distinction can be less clear (or is dependent upon model
formulation). For example, much of GMM uncertainty, which we typically treat as
aleatory variability, can be viewed as epistemic under some models.
• Further, some conceptually epistemic uncertainties (e.g., the influence of an
underlying sedimentary basin on ground motions at a given location) may not be
reducible within a reasonable amount of time and are hence treated as aleatory.
Finally, when we compute mean hazard curves, the classification of a given
uncertainty often has no impact on numerical results or decisions.
• The distinction of epistemic versus aleatory is not always clear-cut and proper
classification may not be necessary. The necessary issues to address are (1) to
consider all uncertainties, (2) to not double-count uncertainties, and (3) to understand
what sources of uncertainty are potentially reducible.
Circular Source
Point Source
Histogram
Disaggregation in hazard analysis provides insights into the earthquake scenarios that drive the hazard at a
given ground-motion level. The disaggregation process differentiates the mean annual rate of exceedance of
specific ground motion at a site with respect to scenarios of given magnitude M, distance R, and the ground-
motion error term ɛ.
It discloses which earthquake scenario (defined using M, R , and ɛ) controls the hazard at a site, which is
very helpful in comparison with hazard curves. The dominant M R ɛ group calculation using the
disaggregation process helps to select the appropriate ground motion for nonlinear analysis for dynamic
response of soil and structures. ɛ along with M R pair provide a better understanding of probabilistic
ground-motion estimates and ground motions from deterministic scenarios