Sobol sensitivity generalization for engineering and science applications

Abstract

Uncertainty is the inseparable part of the engineering analysis. These uncertainties have different Distributions. Sobol decomposition method is one of the most well-known methods of sensitivity analysis. Sobol decomposition as a robust sensitivity analysis method can only be used for uniform distribution. In the present paper, we generalize Sobol sensitivity method for all continuous and discrete distributions. Hence it can be used for wide Varity models in engineering and science of different distributions. The capability of the generalized method for precise numerical values presentation, the results of the sensitivity analysis of the Tehran seismic hazard is given. Due to the variability of materials and earthquakes, Structural and geosciences are full of uncertainties. Calculating the role of the uncertainty of each parameter in the overall uncertainty can be to optimize the efforts and costs necessary for outputs precision, so, it is possible to provide a precise values of sensitivities. The results are satisfactory.

Keywords

• input model
• staistical distribution
• variance
• uncertainty
• hazard analysis