Victoria Brinnel

Determination of global and local cleavage fracture characteristics of high strength bolt steels

Abstract High strength bolts with large diameters are widely used in steel structures from the infrastructure and energy conversion sectors. The brittle fracture behavior of these connections has to be taken into account, especially for structures operating at cold environment, such as the offshore wind energy plant. Generally, two groups of methods are employed for cleavage fracture assessment, i. e., global approaches developed in the fracture mechanics frame and local criteria based on the damage mechanics approach proposed by Beremin. This work aims to determine characteristics of high strength bolt steels in the strength class 10.9 for cleavage fracture assessment. Fracture mechanics tests are conducted to determine the experimental master curves. In addition, numerical approaches are employed to derive the parameters of Beremin model from the master curves.

The Second Blind Sandia Fracture Challenge: improved MBW model predictions for different strain rates

Sandia National Laboratories have carried out the Sandia Fracture Challenge in order to evaluate ductile damage mechanics models under conditions which are similar to those in the industrial practice. In this challenge, the prediction of load-deformation behavior and crack path of a sample that is designed for the competition under two loading rates is required with given data: the material Ti–6Al–4V, and raw data of tensile tests and V-notch tests under two loading rates. Within the stipulated time frame 14 teams from USA and Europe gave their predictions to the organizer. In this work, the approach applied by Team Aachen is presented in detail. The modified Bai–Wierzbicki (MBW) model is used in the framework of the Second Blind Sandia Fracture Challenge (SFC2). The model is made up by a stress-state dependent plasticity core that is extended to cope with strain rate and temperature effects under adiabatic conditions. It belongs to the group of coupled phenomenological ductile damage mechanics models, but it assumes a strain threshold value for the instant of ductile damage initiation. The initial guess of material parameters for the selected material Ti–6Al–4V was taken from an in-house database available at the authors’ institutes, but parameters are optimized in order to meet the validation data provided. This paper reveals that the model predictions can be improved significantly compared to the original submission of results at the end of SFC2 by two simple measures. On the one hand, the function to express the critical damage as well as the amount of energy dissipation between ductile damage initiation and complete ductile fracture were derived more carefully from the data provided by the challenge’s organizer. On the other hand, the experimental set-up of the challenge experiment was better described in the geometrical representation used for the numerical simulations. These two simple modifications allowed for a precise prediction of crack path and estimation of force–displacement behavior. The improved results show the general ability of the MBW model to predict the strain rate sensitivity of ductile fracture at various states of stress.

Improved Limit State Definitions for Pressure Vessels by a Scale-bridging Application of Damage Mechanics

The application of modern high strength low alloy steels (HSLA) in pressure vessel design is currently hindered by the corresponding design standards although it might foster resource-efficient constructions. HSLA steels are penalised by a limited admissible exploitation of their properties which prevents thinner constructions. This limitation is imposed by the design stress definition of the design standards, such as the European code EN 13445. They relate the design stress for high strength steels to the tensile strength, including a high safety factor. This leads to lower design stresses for high strength steels due to their high yield-to-tensile ratio. The admissible design stress is only 46% of the nominal yield strength for a grade P690Q. The current safety factors are based on experiences with common, low strength steels and do not consider the improved toughness properties of modern HSLA steels. Probabilistic safety concepts, as defined in the Eurocode (EN 1990), are a suitable tool to systematically derive more adequate safety factors for modern steels. However, they require a large number of full-scale burst tests. Since these are very costly, probabilistic safety concepts are not directly applicable in pressure vessel design. A possible solution to this problem is to partially replace burst tests by simulations. Ductile failure can hereby be predicted by the methods of damage mechanics. Yet, these models are mainly applied on the small scale or for the failure prediction in components with defects. The aim of this thesis is to develop a comprehensive modelling concept for the prediction of limit states in pressure vessels by damage mechanics simulations and demonstrate its predictive capabilities in comparison to a burst test. Such a modelling concept needs to fulfil three main requirements: It needs to be efficient so that it can be applied in a high number of large-scale pressure vessel simulations, it needs to cover all relevant stress states, and it needs to refer to nominal toughness values. The latter is necessary because pressure vessel design is always based on nominal material values. The modelling concept was developed by joining existing simulation approaches and providing new solutions for the missing links. The failure process in high quality HSLA steels was investigated by stopped experiments to define a suitable limit state. The local onset of softening, homogenised over the typical size of a finite element, was defined to be a suitable limit state. A numerical procedure using cell elements was applied to derive the corresponding limit strains in dependence of the stress state from simulations using Gurson models. This enabled the reference to nominal toughness levels which can be adjusted by an artificial deterioration of the Gurson parameters. This concept was extended to consider Lode angle dependence and adiabatic heating. Additionally, a calibration scheme was defined on the basis of a sensitivity analysis to enable a reliable correlation of Gurson parameters and toughness level. Since the application of strain-based limit criteria may lead to mesh influences in large-scaled simulations with varying mesh sizes, a scaling function for the limit strain criterion was developed and tested.