University of GuilanConcrete Research2008-424214220210622Sensitivity and Reliability Analysis of Slab-to-Column Concrete Joints Using Monte Carlo Method Based on Neural NetworksSensitivity and Reliability Analysis of Slab-to-Column Concrete Joints Using Monte Carlo Method Based on Neural Networks119127467210.22124/jcr.2021.15890.1428FAAliGhorbaniAsistant Professor, Department of Engineering, Payame Noor University,Tehran, Iran.Journal Article20200302According to recent research reports on the parameters involved in the retrofitting and reliability analysis of the concrete slabs connected to columns, Monte Carlo simulation has been introduced as a suitable method. However, a large number of analyzes are required to simulate the probability of failure. This, increases the volume of computations required. In the present study, a useful method is introduced to determine the effect of statistical parameters of random design variables on the probability of failure of slab connection. All the proposed relations are proved analytically and while providing a numerical example, the efficiency of these relations is clarified. The proposed method is based on radial basis function neural networks on the way to replace the analysis in the Monte Carlo method and reduce the computational volume. Therefore, with a combination of neural networks and Monte Carlo method, a new method is presented in the analysis of the reliability of slabs without shear steel. Sensitivity analysis is also presented to assess the effective parameters in the reliability analysis of punch design. Column dimensions, concrete compressive strength, live load, slab flexural reinforcement and slab thickness are evaluated as probabilistic design variables using both European Code (EC 2004) and American Concrete Institute relations (ACI 318_11). The results of validation in the proposed method indicate that the combined method of Monte Carlo and neural network with radial basis functions has an appropriate speed and sufficient accuracy. Also, a comparison of the results of the two codes, shows that EC2 (2004) offers a more economical design than the ACI. The reason for this is to consider the effect of flexural reinforcement on the punch capacity. Sensitivity analysis results show that although in both codes, the thickness of the slab has the greatest effect on the probability of connection failure, but in European code, the percentage of flexural steel has as similar effect as column dimensions and compressive strength of concrete effect. In the relations of ACI, after the thickness of the slab, only the compressive strength of concrete will have a significant effect on the probability of failure. Therefore, it can be concluded that although the flexural reinforcement provides the flexural strength of the slab, but also increases the punch shear strength of the slab by reducing the slab rotation. In general, the proposed method can be used in reviewing other design criteria, especially in credibility analysis and proper retrofitting of structures, due to the reduction of computational volume and appropriate accuracy.According to recent research reports on the parameters involved in the retrofitting and reliability analysis of the concrete slabs connected to columns, Monte Carlo simulation has been introduced as a suitable method. However, a large number of analyzes are required to simulate the probability of failure. This, increases the volume of computations required. In the present study, a useful method is introduced to determine the effect of statistical parameters of random design variables on the probability of failure of slab connection. All the proposed relations are proved analytically and while providing a numerical example, the efficiency of these relations is clarified. The proposed method is based on radial basis function neural networks on the way to replace the analysis in the Monte Carlo method and reduce the computational volume. Therefore, with a combination of neural networks and Monte Carlo method, a new method is presented in the analysis of the reliability of slabs without shear steel. Sensitivity analysis is also presented to assess the effective parameters in the reliability analysis of punch design. Column dimensions, concrete compressive strength, live load, slab flexural reinforcement and slab thickness are evaluated as probabilistic design variables using both European Code (EC 2004) and American Concrete Institute relations (ACI 318_11). The results of validation in the proposed method indicate that the combined method of Monte Carlo and neural network with radial basis functions has an appropriate speed and sufficient accuracy. Also, a comparison of the results of the two codes, shows that EC2 (2004) offers a more economical design than the ACI. The reason for this is to consider the effect of flexural reinforcement on the punch capacity. Sensitivity analysis results show that although in both codes, the thickness of the slab has the greatest effect on the probability of connection failure, but in European code, the percentage of flexural steel has as similar effect as column dimensions and compressive strength of concrete effect. In the relations of ACI, after the thickness of the slab, only the compressive strength of concrete will have a significant effect on the probability of failure. Therefore, it can be concluded that although the flexural reinforcement provides the flexural strength of the slab, but also increases the punch shear strength of the slab by reducing the slab rotation. In general, the proposed method can be used in reviewing other design criteria, especially in credibility analysis and proper retrofitting of structures, due to the reduction of computational volume and appropriate accuracy.https://jcr.guilan.ac.ir/article_4672_a445b52fbd7f15eded8637acdc394ed3.pdf