C.N.R. - Istituto di Ricerca sul Rischio Sismico

The European project SERGISAI developed a computational tool where a methodology for seismic risk assessment at different geographical scales has been implemented. Experts of various disciplines, including seismologists, engineers, planners, geologists, and computer scientists, co-operated in an actual multidisciplinary process to develop this tool. Standard procedural codes, Geographical Information Systems (GIS), and Artificial Intelligence (AI) techniques compose the whole system, that will enable the end user to carry out a complete seismic risk assessment at three geographical scales: regional, sub-regional and local. At present, single codes or models that have been incorporated are not new in general, but the modularity of the prototype, based on a user-friendly front end, offers potential users the possibility of updating or replacing any code or model if desired. The proposed procedure is a first attempt to integrate tools, codes and methods for assessing expected earthquake damage, and it was mainly designed to become a useful support for civil defense and land use planning agencies. Risk factors have been treated in the most suitable way for each one, in terms of level of detail, kind of parameters and units of measure. Identifying various geographical scales is not a mere question of dimension; since entities to be studied correspond to areas defined by administrative and geographical borders. The procedure was applied in the following areas: Toscana in Italy, for the regional scale, the Garfagnana area in Toscana, for the sub-regional scale, and a part of Barcelona city, Spain, for the local scale.

The analysis of damage on buildings affected by an earthquake allow to trace back to the defects of the structural system and the imperfections of the constructive elements with reference to the best or worst behaviour in case of seismic action. As a matter of fact the tendency of buildings to be damaged, also defined as seismic vulnerability, is strictly connected with their geometric-constructive characteristics. The damage appears as "effect" of the phenomenon that has its "cause" in the seismic action. Once the characteristics of the system are defined, it is possible to reconstruct the 'cause-effect' relation between seismic action and damage. These relations can be obtained through survey of damage caused by recent earthquake. However this methodology reveals a series of problems. The various quantities applied must be expressed with indices that allow to synthesize complex and articulated scenarios exhaustively; although the damage is a quantity that can be directly assessed by description, it is difficult to measure, as its quantification should be representative for all aspects connected with it (physical, economic, functional, social, etc.). The level of the seismic action must be expressed by a parameter that is an index of the event's destructive capacity and in direct correlation with mechanical quantities (acceleration, velocity, energy, etc.). The vulnerability is an entity able to characterize the more or less accentuated predisposition of the structures to suffer damages independently from the intensity and direction of the seismic action. In this work, subjects concerning seismic action and vulnerability shall not be dealt with, as they have already been investigated in previous works [see Grimaz S., 1992 - Cella F. 1994 and 1995]; in the following the problem of damage quantification shall therefore be explored.

The studies urban areas directed at the definition of seismic risk, raise the problem of the seismic vulnerability assessment of construction properties that require the estimation of the tendency to damage of a plurality of buildings. Very often one comes upon buildings that have been constructed in former epochs without the use of seismic codes and generally built in masonry. The leads to the search for procedures for vulnerability assessment, based on the rapid acquisition of information on existing buildings, which must furnish a sufficiently reliable assessment of the seismic damageability, generally without the possibility to refer to very sophisticated models. In previous works, assessments of seismic vulnerability have been effected using surveys transferred on National Project for Seismic Prevention (GNDT) sheets (Zonno and Ducarme, eds, 1992). These works have been realized using capes systems in order to treat with the uncertainty of the data. However, the analyses that only refer to GNDT sheets, are limited to single buildings, ideationally understood as isolated. An alternative and maybe complementary attempt is to assess the vulnerability of buildings in a global structural context using Geographic Information Systems to mapping the urban system, integrated with the surveys transferred on COOT sheets. The main characteristics of the building and the structural context, indices of a major or minor damageability, have been individuated, but it is difficult to define a rapid procedure for the assessment of seismic vulnerability. The idea was to use an expert system to codify a basis of the presently acquired knowledge and to apply it automatically on the basis of the results obtained by processes of space analyses calculated by GIS. On the basis of the data obtained with GNDT sheets, the vulnerability of the building can be assessed independently from the structural context (intrinsic vulnerability). The availability of data on the space distribution of the adjoining buildings permits an assessment on the effective vulnerability that takes into account the influence of the structural context. With relation to other works effected on the argument, the proposed system automatically assesses a large quantity of geocoded data either in geometry and in the structure of the components. In particular, in this work the seismic vulnerability assessment of the buildings is effected through the Geographic Information Systems PC Arc-Info connected with the Expert System Shell Nexpert Object, starting from the methods used by the GNDT of the National Council for Research (CNR) (Benedetti and Petrini, 1984; Baldi and Corsanego, 1987) and integrating the effects of anisotropics of the structural behaviour and context (Grimaz, 1992-93).

The purpose of this paper is to emphasize the consequences of the ordinal and qualitative nature of seismic intensity regarding its recording. The classical way of recording by an integer value implies that on many occasions it can be difficult to associate only one intensity degree to an event. Therefore, we propose to record the intensity in a new way so that the expert is no longer restricted to indicating only one value, but can express his belief that the considered event belongs to any one of the intensity classes in the scale. Following this approach, as an example, we study the completeness of the Sannio-Matese catalogue and show how the degree of completeness changes according to the degree of uncertainty in intensity assessment.