@phdthesis{Li2018, author = {Li, Zheng}, title = {Statistical size effect in steel structure and corresponding influence on structural reliability}, issn = {1611-5023}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:co1-opus4-45557}, school = {BTU Cottbus - Senftenberg}, year = {2018}, abstract = {This thesis aims to investigate the statistical size effect in the elasto-plastic material and the corresponding reliability of steel structures. The core idea is that the stochastic material properties are directly embedded in mechanical calculations to develop a more accurate and economical design method for steel structure. Moreover, the results of the experimental investigation with different specimen sizes, whose diameter is limit up to 32 mm, show that the statistical size effect exists in steel structures. This thesis demonstrates finally that the structural reliability is affected by the statistical size effect and the structural safety can be optimized by considering this effect. Because of the uncertainty and non-uniformity of the microscopic imperfection distribution, the material strength in macroscale presents complex randomness. This study described the randomness of material properties through two different ways: developing a stochastic material model for elasto-plastic material and establishing a discrete random field with a general mathematical program. The proposed stochastic material model is extended to analyze the steel structure with multiaxial stress and is integrated into the commercial FEM software for analysis of the complex structures with stress gradient. The stochastic finite element method is implemented to analyze the response of the 3D structures by a general-purpose FEM program when the random field file is imported into the finite element model. The uniaxial tensile tests with different specimen sizes and different material are carried out to demonstrate the statistical size effect in steel structures. The results show that the variations of the yield and tensile strength increase with the decreasing specimen volume. Moreover, according to the bending tests, it is obvious that the structural component strength is not only related to the specimen volume, but also the stress distribution. These two proposed simulation methods, which are an extension and supplement to traditional simulation methods, can effectively simulate the statistical size effect for the tensile and flexural components in steel structures. Finally, it is found by studying the influence of statistical size effect on structural reliability that the strength, which is obtained by small specimens through statistical analysis in the laboratory, is no more accurately applicable to large construction. The reliability theory for the structural safety which exists over the decades can be compared and validated or improved through the embedding the stochastic material properties in the numerical simulation.}, subject = {Statistical size effect; Steel structure; Stochastic material model; Stochastic finite element method; Structural reliability; Statistischer Maßstabseffekt; Stahlbau; Stochastisches Materialmodell; Stochastische Finite-Elemente-Methode; Strukturelle Zuverl{\"a}ssigkeit; Stahlbau; Finite-Elemente-Methode; Stochastik; Stoffeigenschaft; Zuverl{\"a}ssigkeit}, language = {en} }