Due to technical progress, cars of the future will consist of even more different materials than they already do today. Especially plastic materials will experience a further increase of importance, as they provide advantages such as a low density and the freedom to shape them unconventionally. In view of this trend, it is essential to improve the quality of predictions derived from corresponding simulations. Modelling the material in an appropriate way is crucial when simulating a component. While in case of metals plastic deformation happens at a constant volume and therefore is easy to describe, this kind of incompressibility does not apply to plastics. Furthermore, the hardening behavior of these materials is usually significantly more complicated. Therefore, complex mechanical descriptions have to be used for the simulation of plastics, which describe hardening and failure in a multiaxial state of stress. Although those models have been available for some time, it is still cumbersome to calibrate their parameters. In particular, the correct prediction of the strain field, which is the key to characterize material failure e.g. with GISSMO [5], is challenging, as a large number of degrees of freedom have to be adjusted simultaneously.
According to WHO’s report, there are over 270,000 people who are involved in traffic fatal accidents [1]. Based on this accident data, the third-party assessment organization performs pedestrian protection test to evaluate a vehicle safety performance [2]. The pedestrian protection tests are evaluated for the protective performance of a head and legs of pedestrian. In particular, plastic parts such as a bumper face, a grille and head lights are evaluated by the leg pedestrian test. On the other hand, low speed crash test regulated by the United Nations evaluates a bumper protection performance (ECE42). In general, the pedestrian and bumper protection performances are in a trade-off relationship. Therefore, it has become important to balance these performances because the country which does the pedestrian protection test is increasing in recent years. In order to design these performances, it is essential to use the plastic CAE model with high accuracy. However, there are many types of characteristics for the plastic parts compared to the steel parts. It is an issue to collect the material properties for the many plastic parts in author’s development environment. Investigating the past literature to solve this issue, we found that Reithofer et al.[3] developed the machine and method to create the material property for CAE in a short time. So this study is to validate that the machine can be used efficiently to identify the material property for the pedestrian protection and low speed crash.
Current and future automotive development cycles are driven by the needs for lightweight designs, cost reductions, comfort- and safety improvements and the reduction of time-to-market. One way to cope with the listed challenges is the usage of thermoplastic materials for integrative designs of components. Among the challenges for passive safety supplier Autoliv to design thermoplastic components, which are placed in the load path of seatbelt components, is the strong dependency on loading velocity of the components. As crash situations are the most dominant load cases for design and functionality, a strong demand for predictive strain rate dependent material models is given. Strain rate effects are next to temperature- and humidity effects the major challenge concerning thermoplastics. As an industrial demand for a comprehensive material database, it needs to be fast, efficient, economical and accurate. Also, the need for a fully automatized material model calibration process is expressed. To fulfill these demands a two-stage reverse engineering process fits test results to analytical approaches for a quasi-static and a strain rate dependent stress-strain response along with an analytic approach for modelling of visco-elasticity and strain rate dependent damage. The needed test results, to which the analytical parameters are fitted, consist of force-displacement as well as strainfield characteristics and were measured using a newly developed high-speed tensile testing device. This device is designed to get close to constant loading velocity of specimen resulting in strain rates up to 𝜀𝜀̇ = 320 𝑠𝑠−1. The accuracy of the test results is ensured by a wedge-to-wedge, self-locking coupling mechanism, a start-up length for acceleration travel of the tensile testing machine as well as a local force gauge. Especially by the local force gauge, consisting of strain gauges arranged as Wheatstone bridge, it is realized that oscillations in force signals of dynamic testing are minimized. The automatized material model calibration routine fed with accurate test results from the high-speed tensile testing device shows promising results to further enhance simulation quality and predictability for the design of thermoplastic components in crash load cases.
LS-DYNA MAT224 is a tabulated plasticity and failure model. The plasticity part of the model can include strain rate, strain hardening and temperature effects, and the failure part is based on a failure surface of the equivalent plastic strain to failure as a function of triaxiality and the Lode parameter. The present paper presents two new experiments that have been developed recently in order to support the model. The first experiment adds new points to the failure surface in a region that is important in simulations of projectile impact and penetration. The second experiment is used for determining the Taylor-Quinney coefficient (β), which controls the magnitude of the temperature increase due to plastic deformation. Simulations of impact and penetration events show that failure occurs under a stress state of biaxial tension and out-of-plane compression. This state of stress on the failure surface is not in the region that is populated with data points obtained from typical experiments (tensile tests of flat and round, parallel and notched specimens, tensile tests of wide parallel and notched specimens, pure shear tests, combined tension-compression/shear tests, and compression tests.) In order to obtain an independent measurement of the equivalent strain to failure under a state of stress of biaxial in-plane tension and out-of-plane compression a new experiment was developed. In this experiment a small diameter punch penetrates a thin specimen plate that is backed by another plate. The deformation of the back surface of the plate is measured with DIC. The value of the equivalent strain to failure is determined from measuring the force and matching the LS-DYNA simulation with the measured deformation and force.
Nowadays the possibility to accurately simulate steel alloys is crucial to expect accurate results in crash analysis. Just as much as welding spots, innovative materials like polymers or composites more predictive steel material models are continually sought after throughout the industry. Complex models require a significant effort to calibrate them to the physical behaviour of the materials, but they can perform better. This work will evaluate and compare different techniques to characterize materials for finite element simulations.
Nowadays, strain fields can be experimentally measured with high accuracy through digital image correlation (DIC). This kind of measurement is becoming standard when it comes to physical testing of materials. The information from such measurements is then often used in the calibration and validation of material cards to be later used in LS-DYNA. Especially regarding the prediction of failure, the experimentally measured strain fields can be quite helpful. Among several methods for the calibration of material cards, one method relies on the direct use of such strains in the definition of the failure curve as a function of the stress triaxiality ratio. However, in such method, the triaxiality is usually estimated from the simulation of the specimens adopted in the physical tests or, sometimes, it is estimated from analytical calculations based on the loading type and on the geometry of the specimen. It is however widespread known that the triaxiality typically varies during experiments. Therefore, it would be interesting to observe the evolution of the triaxiality throughout the physical test. As mentioned before, the typical way of doing this is through the use of numerical simulation to perform this task. In this paper, we concentrate efforts in developing a method to estimate the triaxiality distribution and evolution using information directly from the DIC measurement. To that end, a plane stress state is assumed and the strain ratio is calculated from the measured strains. The stress triaxiality ratio is, in turn, a relation between the hydrostatic and the equivalent stress. Therefore, in order to calculate the triaxiality from the strain field, a relationship between the strain ratio and the triaxiality has to be defined. This is only possible through the consideration of a constitutive (i.e., material) model. Typically, the J2-based plasticity model (commonly known as the von Mises model, e.g., *MAT_024 in LS-DYNA [1]) is used for this kind of task. However, our research on the topic has shown that this assumption may lead to wrong triaxialities even in cases when the triaxiality is known beforehand, for instance, in a uniaxial tensile test before necking. This error can be significantly reduced if the anisotropy of the material is also taken into account. To that end, we use a Hill-based transversely anisotropic material law in order to consider the effect of the anisotropy. After some mathematical derivations under the assumption of plane stress, negligible elastic strains and proportional loading, it is possible to find a closed-form relation between the strain ratio and the triaxiality including the effect of the R value. The results for an aluminum sheet show that the triaxiality is much better predicted using the new formula. Using a software dedicated to the evaluation and visualization of optically measured strain fields, it should be possible to plot triaxiality fields from experimental data that can be later used either for the calibration or validation of a material card. Furthermore, this novel technique can also be employed on the development of new specimen geometries in order to better assess the stress triaxiality ratios obtained with the new geometry without having to first calibrate a material card for that.
The determination of material properties under high-speed loading has been a challenge for many years. Structural vibrations, also called system ringing, in conventional testing machines deteriorate the quality of force measurement, which makes a precise determination of stress-strain curves and corresponding mechanical properties impossible. In this work, a new specimen geometry with its basic mechanical principle and the corresponding measurement technique are presented and discussed. The new method allows the determination of true stress-strain curves at high strain rates free from oscillation. Due to the additionally minor plastic deformation area in the new Generation III specimen, a quasi-movable bearing condition for the specimen fixation was created. Forces, based on the natural frequency of the specimen, deform the cross-section and create a displacement, which keeps the kinetic energy of the measurement area high. In this way, the elastic ringing effect has been reduced significantly. Any kind of filtering, smoothing or similar manipulations of the result are no longer needed. This new method has been validated on three steels types and one type of aluminium alloy with different strain hardening behavior through measurements and numerical analysis. The Generation III specimen can also be used for the quasi-static test and cover the strain rate range from 4.4·10-4 - 103 /s accordingly.
Development of a New Method for Strain Field Optimized Material Characterization