Fragility Model

A uniform fragility model was developed for the most common building classes in the Andean region, considering the typical local structural properties and the regional seismic hazard. To this end, sets of single-degree-of-freedom (SDOF) oscillators were created and subjected to a series of ground motion records using non-linear time history analyses (NLTHA). The resulting damage distributions were employed to derive sets of fragility functions.


1. Derivation of capacity curves

A series of parameters were defined to describe the structural characteristics of each building class. The values for each of these parameters were based on an extensive literature review, comprising the most relevant construction types. In addition, two workshops (developed in Medellin, Colombia and Lima, Peru between 2014 and 2015) allowed considering the opinion of several local experts. Using the gathered information, a capacity curve was derived for each building class in terms of spectral acceleration vs. spectral displacement. An elastic-perfectly plastic behaviour was assumed for most building classes.

In the case of concrete frames with infills (LFINF) and confined masonry (MCF), a trilinear model was assumed instead (see figure below). This model allows taking into consideration the loss of strength and stiffness due to the degradation in the masonry panels.

For each building class, the resulting capacity curve was considered to be the median in a set of 150 normally distributed capacity curves which allowed to consider the building-to-building variability. This synthetic building portfolio was generated by means of a Monte Carlo simulation. The figure below shows two examples of the generated sets of curves.

2. Selection of ground motion records

The selection of ground motion records was performed considering the tectonic environment and seismicity in South America. Most of the seismic activity in this region is due to the subduction of the Nazca and Antarctica plates beneath the South American plate. In addition, there are also significant events due to shallow crustal faults. For distances greater than 50 km, ground motion records with moment magnitudes between 7.0 and 9.0 were selected, while for shorter distances, records with moment magnitudes between 5.0 and 7.0 were considered. These records were selected from the PEER database. A rock soil type was assumed.

The selected records were scaled considering a maximum scaling factor of 2 and three intensity measure types (IMTs) (PGA, Sa at 0.4 s and Sa at 1.0 s). Ten levels of ground shaking were defined for each IMTs and 30 records were scaled at each level of ground shaking, leading to 300 records.

3. Definition of damage criterion

Four damage states were defined for this study: slight, moderate and extensive damage and collapse. The definition of each of these is summarised in the following table.

4. Development of fragility functions

Each set of 150 SDOF systems was subject to the associated set of ground motion records using the NLTHA on SDOF module from the GEM's Risk Modeller's Toolkit (RMTK). The behaviour of each SDOF was described by its maximum spectral acceleration, which was then compared against the damage thresholds described above, allowing to allocate each structure within a damage state. This way, it was possible to obtain a damage matrix with the amount of buildings in each damage state for each ground motion record, for each building class. In order to increase the usefulness of the curves, it was decided to select the three intensity measure types (IMTs) used by the ShakeMap system of the United States Geological Survey (i.e. PGA, Sa at 0.3 s and Sa at 1.0 s). Considering these three IMTs, a regression analysis was performed, which led to the fragility functions. These functions were modelled following a cumulative lognormal distribution and their parameters (logarithmic mean and logarithmic standard deviation) were derived following the least squares method. The figure below presents the resulting sets of fragility functions for two building classes.

<== Back to Exposure and Physical Vulnerability

<== Back to the SARA Project main page

  • risk/vulnerability.txt
  • Last modified: 2016/08/30 22:18
  • by MabĂ© Villar