This paper discusses the relationship between the probability distribution functions (PDFs) of residual deformation and total plastic deformation of a perfectly elastic-plastic single-degree-of-freedom oscillator, from a viewpoint that, if this relationship is known, it is possible to use the residual deformation after an earthquake as an index of the total plastic deformation, that is, the accumulated structural damage.
In the previous study, the author showed, using the “random walk hypothesis”, that the conditional probability of the residual deformation given the number of plastic excursions,
n, and the total plastic deformation,
dt, is almost a normal distribution with an average of zero and a standard deviation of
dt over the square root of
n. By calculating the inverse probability, in this paper, the conditional PDF of
dt under the residual deformation,
dr, was formulated. This formula enables one to estimate the structural damage the structure has received during an earthquake, by measuring the residual deformation after the earthquake.
In order to validate the derived formula for the conditional PDF of
dt given
dr, dynamic response analysis was performed using a large number of pseudo ground motions simulated by the inverse Fourier transform and actual ground motions recorded at two locations in Japan over the period of 10 years.
To calculate the conditional PDF of
dt given
dr with the formula, it is necessary to obtain the joint PDF of
n and
dt. In order to do this, first, the log-normal PDFs for
n and
dt were obtained by regression analysis of dynamic response analysis results. The joint PDF of
n and
dt can be assumed as a product of each log-normal PDF of
n and
dt, if
n and
dt are statistically independent. This assumption is used although it is not always justified and therefore this may affect the accuracy of the estimation.
Once the joint PDF of
n and
dt are obtained, it is possible to calculate the conditional PDF of
dt given
dr. This calculated PDF shows reasonable agreement with the histogram obtained from the dynamic response analysis, for both cases of simulated ground motions and actual ground motions, although the accuracy for the actual ground motions is worse than that for the simulated ground motions. It is presumed that this error might come from the rough assumption of statistical independence in the joint PDF of
n and
dt for the case of actual ground motions. There may be a possibility to improve the accuracy by obtaining more accurate joint PDF of
n and
dt, which is considered as an index of earthquake characteristics.
抄録全体を表示