Abstract

Consistent materials information is an important input for many CAE-simulations. A big variety of material card formats for different applications is available. In this paper we are addressing the following practical issues: (a) Material designations are usually related to standards which allow a scatter of properties depending on variations in chemical composition and process history. This scatter is frequently not considered in simulations. (b) Materials testing to describe complex material behaviour is frequently expensive and time consuming. Budgetary and time constraints lead to a situation, that such expensive tasks are not statistically validated so that material cards relying on few measurements may be inconsistent. In many cases material properties can be calculated using materials simulations which is state- of-the-art for many data of structural metals using CalPhad, JMAK and additional physically based models. We are extending this approach by performing high throughput calculations with variation of chemical compositions and variation of processing like heat treatment. As a result of several thousand calculations which are consolidated in a NoSQL database we can derive multiple property maps for a single material designation. This allows a selection of worst-case and best-case materials from which we can derive material cards for different solvers. Additionally those calculations can help to assess the calculated influence of variations on the resulting material models. Frequently material models, e.g. for plasticity are taken from different sources which show significant deviations, like text books, material catalogues and different tests. We can now consolidate these data into a common data structure which allows (a) a direct comparison of curves and (b) the generation of hybrid models by curve fitting to different constitutive equations. This involves methods for thinning/resampling/averaging of measurements represented as parametrized time-series. Data from materials simulation are usually consistent with respect to the influence of composition, temperature and strain rate. Combining this information with selected measurements can be used to calibrate and extrapolate the data. We will present the influence of different material models for plasticity, like Johnson-Cook and Hensel-Spittel as well as the influence of different minimizers, like Nelder-Mead and BFGS on the results. The modelling results are subsequently consolidated in a materials master model, which is designed to be a homogeneous source of information for different solvers of different vendors. Configurable mapping tables allow the export of different material cards from such single source of information. Existing material cards can in turn be imported and data of them can be compared to the data of the master model.