Vestnik MGSU (Aug 2024)
Structural reliability analysis of steel structural covering based on the theory of evidence
Abstract
Introduction. Quantitative assessment of the structural reliability and safety level for structural solutions is an actual scientific and technical problem. The measure of reliability in this case can be the failure probability of a structural element. In practical problems of assessment and analysis of structural reliability, data on random variables can be obtained in the interval form (qualitative uncertainty), while the classic methods of reliability analysis do not allow to estimate reliability in the presence of such data. Lack of statistical data (quantitative uncertainty) is also present in practical tasks of reliability analysis.Materials and methods. The paper considers the use of the theory of evidence as an effective tool for reliability analysis of steel span structures in problems with interval uncertainty of statistical data.Results. The graphical interpretation of the reliability analysis algorithm is given. It allows to obtain clearly and operatively an estimate of the failure probability of the structural element, as well as to reduce the permissible load on the element to the required reliability level. Two-dimensional and three-dimensional models are considered for analyzing the reliability of a steel truss bar according to the buckling criterion.Conclusions. The theory of evidence allows effective modelling of various sources of uncertainties in practical engineering problems. Thus, with limited data, it is possible to get an idea of the quantitative expression of the reliability level by its lower bound, which can be increased by strengthening the element, more detailed probabilistic analysis or limiting the operational load on the structural element. By considering more non-deterministic quantities in the design, the engineer obtains a more cautious decision. When considering the cross-sectional area as a random variable, the reliability of the truss bar on the buckling was 23 % lower (lower reliability bound) than in a similar calculation with a deterministic value.
Keywords