GNGTS 2013 - Atti del 32° Convegno Nazionale

that the building is constituted by i th Building Typology BT and j th age of construction AC, where , (1 – p ij ) represents the error made in the disaggregation process respect to the real value identified by the survey: (5) The results presented in Fig. 2 shows the error of mean damage value with the main sources of error. It is to be noticed that an increase of error in the number of storeys yields to an increasing error in the evaluation of mean damage value. Furthermore with an error in the determination of the building typology and the age of construction less than 30% an error less than 10% can be observed, beyond which an increase in the former yields to an increase in the dispersion of the latter while its median value always remains below the 20%. The procedure described in previous Section can also be carried out assuming the ISTAT data instead of the data from the field survey. In this way, the effectiveness of a seismic vulnerability assessment based on a different data source can be investigated, according to a multilevel approach. However, the comparison between results based on different data sources should be more properly carried out considering census cell instead of single buildings, consistent with the scale of the seismic vulnerability assessment. Nevertheless the sources of information are provided aggregated for census cell both for number of storeys and age of construction, and hence it is not possible to know for each census cell the number of RC buildings dating back to a specific age of construction or constituted by a determined number of storeys. Moreover ISTAT data do not allow to obtain information on Plan Irregularity and Soil type, which are not take into consideration in the evaluation of vulnerability assessment of buildings. Hence starting from ISTAT data for each census cell a disaggregation process is made through the use of reference distribution for Masonry and RC Buildings consisting of a sample of 10% of the surveyed building stock. Thus homogeneous classes of buildings as a function of Building Typology, Age of construction and Number of storeys can be defined, and then for each homogeneous class of buildings the resulting mean damage value can be determined corresponding to the macroseimic intensity characteristic of the census cell. Based on this assumption, the mean damage value in each census cell can be evaluated as the weighted average of the mean damage value for each one of the homogeneous classes of buildings defined above, depending on their respective percentage of occurrence within the census cell. In Fig. 3 the results of the procedure from different data sources are compared. In particular the mean damage value for each census cell is shown, in addition to the distribution of age of construction (pie representation). As shown above at single building level a very good correspondence between the results coming from survey data and those from LIDAR can be observed, for less than tolerable scatter in the case of error in the determination of the age of construction of the buildings. For example, in census cells located in the eastern part of the old town can be observed same deviations in the value of the mean damage value primarily due essentially to this reason. It should be noted that LIDAR data, as well as survey data, allow the estimation of the number of storeys of buildings, besides plan irregularity and soil type. It is to be noticed that 57 GNGTS 2013 S essione 2.1