Contribution to the application of reliability-theoretical calculation methods to problems typical for steel structures

Author: Peter Lieberwirth

Abstract
The subject of this thesis is the application of reliability-theoretical methods to typical steel structure problems based on a probabilistic safety concept. This should enable the realistic verification of ultimate and serviceability limit states for selected types of structures under time-variant actions. To fulfil this objective, proven probabilistic methods are combined with an efficient deterministic basic solution in the form of finite beam elements. This is realised on the one hand by an iterative response surface algorithm and alternatively by means of Latin hypercube sampling as a variance-reducing Monte Carlo method. Special features of typical steel construction checks for stress and stability problems are taken into account in the choice of the approach functions and the limit state definition. The input variables include stochastic uncertainty and are quantified in the form of suitable distribution functions. Populations composed of defined subsets can be captured by multimodal approaches. Dependencies between individual random variables can be described by correlations.
One focus is the development of stochastic models for time-variant climatic effects using the example of the basic snow load and the wind peak velocity, which represent decisive safety-relevant variables for common frame and girder systems of structural steel engineering. Particular importance is attached to the systematics for deriving suitable distribution models and their parameter quantification, taking into account sampling uncertainties of long-term observation time series. The pronounced location dependency of the models enables a differentiated structural analysis. Here, even short reference periods, which are relevant for the safety assessment of temporary construction conditions, for example, can be evaluated computationally.
In selected examples, the calculation of practical problems is carried out in addition to a plausibility check of the solution procedure. The stochastic models provided are used for the verification of steel frame constructions at locations with significant snow or wind loads, whereby internal force interactions are taken into account and the use of plastic system load reserves is possible.