RESEARCH ON RESISTANCE MODEL UNCERTAINTY OF HIGH STRENGTH STEEL AXIAL COMPRESSION MEMBERS

摘要

High strength steel has been used successfully in many steel structures all over the world.But there hasn't been so much research about high strength steel structures to support the design codes.The design resistance of one particular structural member is an important part in steel structure design codes.The structure resistance is mainly influenced by material property uncertainty,geometric uncertainty and resistance model uncertainty.In order to investigate the resistance model uncertainty of high strength steel axial compression members,experimental data and numerical results from finite element method(FEM)in past researches were summarized.The design formulae of axial compression members in high strength steel design code of China(draft)were then introduced,which showed some differences from the ordinary strength ones.Code values by applying these design formulas were calculated.The differences between experimental values and theoretical values(Kp1),theoretical values and code values(Kp2)have be analysed utilizing the method of mathematical statistics.The probability distribution was examined,which proved that both Kp1 and Kp2 obey normal distribution,and the statistical parameters of Kp1 and Kp2 were then obtained.On that basis,the statistical parameters of resistance model uncertainty(Kp)of high strength steel axial compression members have been calculated.The mean value of Kp is 1.068 while the variable coefficient is 0.104.These statistical parameters will be used for calculating the design strength of high strength steel.Furthermore,the parameters which influence the resistance model were studied.The parameters contain types of section(welded I and box),steel grades(Q500,Q550,Q620 and Q690),buckling modes,design curves(column curve a and b)and different slenderness ratios.It is shown that steel grades,design curves and slenderness ratios influence the resistance model uncertainty a lot.Moreover,when members are designed by the column curve a or the slenderness ratio is small,the mean values of uncertainty seem less than that of the population.