FOSTA
P 1035 - Increasing the load-bearing capacity of welded I-beams made of high-strength structural steel through improved approaches
P 1035 - Increasing the load-bearing capacity of welded I-beams made of high-strength structural steel through improved approaches Welded solid wall beams, such as I-beams, are used in many areas of modern steel construction. In contrast to rolled sections, the overall height, cross-sectional shape and sheet thicknesses can be adapted to the load, which results in economic advantages. Nowadays, high-strength fine-grain structural steels with (minimum) yield strengths of up to 690 MPa are increasingly being used. To improve the utilisation of the load-bearing potential of welded profiles made of these steels, there are no suitable models for determining the residual stresses caused by welding on real components. These data are approximated using conservative assumptions. The potential load-bearing capacity of welded structures is therefore generally underestimated. The residual stress assumptions in DIN EN 1993 are based on strong simplifications, e.g. in the context of so-called "equivalent member verifications". These assume the proportionality of the residual stress amplitudes to the yield point. Comparative calculations with a parameterised residual stress model showed possible deviations in the calculated load-bearing capacities over several buckling lines. However, the specifications regarding the buckling line to be applied influence the sheet thicknesses and therefore also the production costs to a considerable extent. As an optimisation strategy, various application-oriented model approaches were therefore developed in this research project, which should enable higher utilisation of load-bearing capacity reserves with the aid of non-linear load-bearing capacity calculations. The research objective was therefore primarily the development and provision of corresponding calculation approaches. To this end, the residual stress state of conventionally welded structures was determined experimentally using the example of I-beams made of two common structural steels, a non-alloy structural steel S355J2+N and a fine-grained structural steel S690QL, which are particularly frequently used in steel construction, and supplemented by numerical welded structure simulations. A simplified calculation model was then derived for further consideration in non-linear load-bearing capacity calculations. The findings are to be incorporated into a corresponding user guide, published by Technical Committee 8 of the European Convention for Steel Structures, and in this respect serve as orientation for a possible future adaptation of the buckling lines in DIN EN 1993-1-1. In the project, the above-mentioned steels were joined in two typical thickness combinations of 25/15 mm and 20/10 mm (chord and web plates) to form large "I-beams" specimens under near-industrial production conditions using metal active gas welding. These produced the typical residual stresses that are present in such beams after production. During welding, the temperature fields were recorded using thermocouples. The residual stresses were determined experimentally after welding by a combination of mechanical disassembly (disassembly method) and subsequent mobile X-ray diffraction (XRD). The influence of the standard industrial flame-cut edge preparation of the sheets on the final residual stress states was also investigated experimentally on small-scale samples. Metallographic analyses (micrographs and hardness measurements) completed the experimental programme of FE 2 (BAM, Berlin). The results showed that a large proportion of the residual stresses had already been removed during mechanical dismantling. However, it was only possible to make a statement about the residual tensile stresses close to the seam by superimposing them with the XRD. The S690QL predominantly showed better behaviour than the S355J2+N, which was expressed, among other things, in lower residual tensile stresses in relation to the respective yield point. The maximum residual stresses were mostly measured in the heat-affected zone (HAZ) of the weld seam, irrespective of the applied line energy or the strength of the base material. The practical construction assumption that the maximum residual stresses can be present at the level of the yield point was only confirmed in individual cases and only for S355J2+N. The experimental results also showed reduced relative residual stress values of the S690QL for the compressive residual stresses. This shows the need for further optimisation when selecting the evaluation approaches for residual stresses on the load-bearing capacity. The measured values generated were subsequently used as input values for the model development and the numerical simulations on I-beams by FE 1 (LSH, Cottbus). The numerical welding simulations were carried out in Simufact.Welding® for the simulation models with a "classic" welding simulation approach and in Ansys® for simplified simulations and approximation approaches. Overall, the models delivered comparable results for the longitudinal residual stresses, although there were some considerable differences in the calculation times. The calculation time could be significantly reduced in all cases by adapting the meshing. However, due to the component sizes used, the application of modelling with solid elements is usually not practicable. For typical beam welds, cross-section models with the assumption of a generalised plane distortion state offer a possible alternative instead. The comparison of the results with the experiments provided predominantly good to satisfactory agreement. The experimentally determined tendencies were correctly predicted in all cases. A comprehensive evaluation of the achievable accuracies has not yet been possible. On the one hand, this was due to the fact that the data density of the results obtained with the decomposition was low. On the other hand, the superimposed results with the XRD provided contradictory statements on the influence of residual stress in parts, particularly in the pressure ranges of interest, which could not be conclusively clarified to date. In addition, any deviations in the simulations were also caused by the underlying material data itself. These corresponded to standard data sets from material libraries. These were scaled uniformly over the entire temperature range. For the S690QL in particular, too few reliable material parameters were available. Corresponding tests could not be carried out as part of this project. Nevertheless, it was possible to gain some important insights into a simplified calculation procedure for taking residual welding stresses into account in load-bearing capacity calculations. Based on the plastic strain patterns determined in the numerical welding simulation models, also known as "residual stress sources", a simplified simulation procedure was proposed and verified. The simplifications introduced were selected with regard to the calculation of only the plastic longitudinal strains or longitudinal residual stresses. The applicability is therefore limited to sufficiently long welded beams with longitudinal seams. Some model modifications for further simplification or model implementation in the structural-mechanical calculation of large components were presented. These investigations are ongoing. Finally, the classification into a possible buckling line is based on a scatter band of load capacities simulated by GMNIA (geometrically and materially non-linear analysis with imperfections) calculations using Abaqus® and Ansys®. The initial hypothesis of a more favourable buckling line of a S690QL compared to the S355J2+N was confirmed by calculation. However, experimental proof by means of corresponding load-bearing capacity tests is still pending.
Publication: 2018
Authors: H. Pasternak, B. Launert, T. Kannengießer, M. Rhode
View details →
