Scenario Damage Simulation–Fragility Function Customization Query
Benjamin Tarver
I am attempting to simulate damage from the 2017 Puebla-Morelos earthquake using the OpenQuake Engine, and one issue I am encountering is accurately capturing the effect of shaking on buildings built on soft lakebed soil. GEM makes available a number of fragility functions for buildings in Mexico which rely on the 0.3s and 0.6s-period spectral accelerations. I understand that the soil conditions in Mexico City's historic lakebed area mean that the dominant site periods are more commonly on the order of a few seconds. Earthquake engineering (and earth sciences more broadly) is not my domain of expertise, so I am unsure of the degree to which this mismatch needs to be accounted for (e.g., via implementing a function to multiply the damage values for buildings constructed in areas with high mismatches between their corresponding fragility function's spectral acceleration period and their site period) in my fragility functions. (In case it's helpful context, I have already captured the effect of the softer soil on time-average shear-wave velocities in the study area.) Any and all help is greatly appreciated, thank you (and thank you to the folks at GEM for making this wonderful tool available to the public for free)!
Vitor Silva
Dear Benjamin,
Thank you for your kind message.
I understand that the Mexico City lakebed basically modifies the response spectrum, often producing strong amplification and long durations at site-dependent periods. If I remember correctly, in the 2017 Puebla earthquake the strongest amplification was for periods of vibration above 0.5 seconds. So, in my opinion the key question is not whether the IM in the fragility/vulnerability should be changed because of the soft soils, but rather whether the hazard/site-response model produces the correct spectral ordinates at the periods used by the fragility functions.
In OpenQuake, each fragility function is explicitly defined for a particular IMT, such as PGA or SA(0.3), SA(0.6), and that IMT is part of the fragility model itself, and it was selected based on the dynamic properties of the buildings (low-rise buildings are affected by PGA or short-period SA, while taller, more flexible buildings respond to long period SA)
That means you generally should not add an ad hoc multiplier to damage just because the site period differs from the fragility period. A better approach is to keep the fragility IMTs tied to the structural classes they were calibrated for, and make sure the ground-motion input properly captures the lakebed amplification at those periods. If the buildings you are modelling are genuinely most sensitive around 0.3 s and 0.6 s, then those fragility functions can still be appropriate.
I would recommend investigating instead whether simplified site proxies such as Vs30 may miss important basin and resonance effects, in which case the resulting SA(0.3) and SA(0.6) may be biased. Please note that with the OpenQuake-engine, you can define your own site amplification model (which can be a nonlinear functions).
Hope this helps,
Vitor Silva
--
You received this message because you are subscribed to the Google Groups "OpenQuake Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
openquake-use...@googlegroups.com.
To view this discussion visit
https://groups.google.com/d/msgid/openquake-users/19dc50a1-9531-4c92-b3eb-a8d3414753fbn%40googlegroups.com.