Investigation On The Shear Behavior Of Normal-Temperature And Post-High-Temperature RC Deep Flexural Members

In recent years,frequent building fires have caused huge losses.Fire not only causes loss of life and property,but also causes some damage to concrete in the buildings.Meanwhile,as deep flexural members are widely used in high-rise buildings and bridges,its mechanical performance has great influences on the whole structure.Therefore,in this paper,seventeen deep flexurnal members were designed and made,eight of which were used for shear test at ambient temperature,and nine of which were for shear test after fire.The effects of high temperature,shear span ratio and span-depth ratio on the shear bearing capacity,deflection,strains of steel,maximum crack width and failure mode of deep flexural members were investigated respectively.Combined with finite analysis results of ABAQUS and particle swarm optimization(PSO)neural network,the prediction model of shear capacity of deep flexural members at ambient temperature and after high temperature were established repectively.Experimental research and ABAQUS finite element simulation on deep flexural members at ambient temperature were carried out,and influence of span-depth ratio and shear span ratio on the shear behavior of deep flexural members was discussed and analyzed respectively.The results showed that when other condition are same,both the initial shear and ultimate shear strength of deep flexural member reduced with an increase in shear span ratio.When the spandepth ratio of specimens were same,under the same load,the mid span deflection of specimens increased with the increase of shear span ratio.Compared with the strain curves of longitudinal tensile steel bars,the strain curves of stirrups had positive and negative values and the changes along the horizontal axis were obvious.For the shear capacity of deep flexural members,the effect of horizontal web reinforcement was greater than that of vertical stirrups.For the specimens with different span-depth ratios,effect of the shear span ratio on the crack width of them were consistent.under the same load,the crack width of the specimen increased with the increase of the shear span ratio.ABAQUS can simulate the stress condition of the deep flexural members and the simulation results were better,which agreed well with the experimental value and strutand-tie model theory.Experimental researches on deep flexural members at ambient temperature were carried out,and influence of high temperature,span-depth ratio and shear span ratio on the shear behavior of deep flexural members was discussed and analyzed respectively.Results showed that when other condition are same,with the increase of high temperature,the shear capacity of specimens increased first and then decreased.Moreover,the effect of high temperature on the shear capacity of specimens weakened with the increase of shear span ratio and span-depth ratio.In general,under the same load,the deflection of specimen increased with an increase in high temperature,and the deflection of specimens increased with the increase of shear span ratio and span-depth ratio after same high temperature.The experimental data of this paper were analyzed based on the basic theory of heat transfer,shear theory of deep flexural members,strut-and-tie model and code for design of concrete structures.The results showed that the strut-and-tie model based on layered method could predict the shear capacity of deep flexural members after fire and had some applicability.In this paper,particle swarm optimization algorithm was used to optimize the weight and threshold of the BP neural network which was easy to fall into the local optimum and poor in robustness and generalization ability.By learing a large number of experimental data,the BP neural network prediction model and PSO-BP neural network prediction model for the shear bearing capacity of deep flexural members were established respectively,and performace of the two prediction model were compared.The results showed that compared with BP neural network prediction model,PSO-BP neural network prediction model had advantages of high prediction accuracy,fine stability and fast convergence speed.The absolute error rate of the test was always less than 15% throughout testing,which can meet the accuracy requirement of actual engineering.