Evaluation of a reliability index for steel trusses to the deflection criterion with interval uncertainty of data

The authors describe a new approach to evaluation the reliability index of steel trusses by the criterion of deflection considering the uncertainty of random variables expressed in the interval form. Classical probabilistic-statistical methods of structural reliability analysis require the choice and justification of the cumulative distribution functions for random variables and its parameters. Subjective acceptance of statistical hypotheses can lead to large errors in the structural reliability analysis. In this study, it is proposed to represent random variables in the interval form that characterize the boundaries of their variability. Such intervals can be obtained as tolerances by the technical documentation, can be based on the construction experience or can be got by data analyzing. The Vysochansky - Petunin inequality is used to obtain the limits of variability of a random variable without a hypothesis about a specific probability distribution function. The reliability analysis of bar-systems is complicated due to the uncertainty of the data in each element of the system. For the engineering solution of this problem, an analytical approach to the optimization problem is offered. The truss reliability index can be used to compare several design solutions in a quantitative form according to the criterion of operational safety.