抄録
Simple non-linear response prediction method (stochastic RMS Method) is developed, subjected to the building excited by arbitrary shape spectrum input based on statistical equivalent linearization technique. In previous method, hysteretic characteristic is modeled by unit function independently its shape, and it is necessary to introduce nonlinearity term of system like coulomb slider into equation of motion except displacement and velocity. However, mixture of these term causes complication to solve equation, and is reason to use only white noise for input. The equation made with above process can't solve except explicit time history analysis. In this paper, equation of motion is made with only state variable express displacement and velocity to introduce displacement δa at reversal point as parameters. as equation of motion set simple, it is possible to use arbitrary shape spectrum. This method can adopt poly-linear hysteretic characteristic or MDOF system. Decision of this method is examined for 10 story model building compared with MCS (Monte Carlo Simulation).
· For SDOF, difference of stochastic RMS method and MCS is under 20% for ductility factor 1~8.
· For MDOF, when ductility factor is about 2, maximum difference of both method is about 20%. RMS value of story drift angle σ[Rmax] cause larger difference.
·Un-exceedance probability show same tendency.
· Difference of time history response and stationary response is few. With use ordinary envelope function to time like this paper, both method is available.
· When building has stiffness eccentricity, accuracy of stochastic RMS method decline at weak story. bat, response of other story slightly decline.
CPU time to calculate about model building is 10~15 minutes by stochastic MCS, and 1~2 second by stochastic RMS method. RMS method is superior with MCS for time to calculate.
