14th European Conference on Earthquake Engineering 2010, Ohrid, Macedonia, 30 August - 03 September 2010, vol.1, pp.6213-6220
In the current practice of assessment of existing buildings it is common to use the static pushover methods in order
to represent the capacity of the structure. There are several procedures in the existing literature some of which are
included in the assessment codes in combination with the Nonlinear Static Procedures (NSPs). It is always a
question however, whether these static pushover procedures would be able to accurately predict the real damage
state of an existing building or not. Almost all, if not all, of the NSPs are continuation of the well known Capacity
Spectrum Method (CSM) which means that they are in intensive use not more than for 15 years. The main dilemma
in this point is that none of these methods was tested against the real damage states caused by real earthquakes.
Existing calibrations have been done by rather limited experimental results and mostly by more complicated
nonlinear time-history analyses. It is comprehensible that there are several inherent uncertainties which are not and
cannot be considered in these methods; however, the main goal of these methods is still to predict efficiently the
expected damage level. Consequently, the aim of this paper is to investigate the assessment procedure that is given
in the latest Turkish Earthquake Code of 2007 (TEC’07) and to compare the outcomes both with the demand level
predicted by the most updated CSM of ATC 55 and the damage state of the case study real building that was found
heavily damaged in Sakarya, Turkey, following the 1999 Earthquakes. Multi-mode adaptive pushover analyses
have been run on this structure to come up with strain levels of the RC sections to be compared with the strain
limits given in TEC’07. Uncertainties, such as prediction of the ground shaking level, the dominant direction of the
earthquake, inabilities in mathematical modelling etc., have also been discussed and the point of view of the
authors has been presented for each case. Errors between the prediction and the real damage state have been
identified and possible improvements have been listed.