Marian Bulla

Data‐driven material modeling techniques have gained significant attention due to their ability to capture the complex constitutive behavior beyond the limitations of classical material models. Physics‐augmented neural networks (PANNs), which incorporate physical restrictions into their architecture, combine the flexibility of machine learning with properties required for reliable engineering simulations. As a result, they have become increasingly relevant in computational material modeling. This work presents an approach to integrate pretrained isotropic, nearly incompressible, invariant‐based hyperelastic PANNs into the explicit finite element solvers Simcenter Radioss and OpenRadioss (Siemens). To this end, a Python framework transfers the trained parameters and the required constitutive expressions of the supported feed‐forward architecture to a standalone Fortran user material routine. The analytically expressed first derivatives of the strain‐energy density are included in the generated code. The finite element simulation therefore does not require an external machine‐learning library or a specialized solver. In addition to the integration strategy, particular emphasis is placed on the computational efficiency of the resulting user material subroutine. The influence of the network architecture on the overall simulation performance is investigated, and approaches for reducing evaluation costs while preserving the quality of the fitted material response are discussed. In this context, replacing the commonly used SoftPlus activation function by the SQuarePlus function reduces the additional constitutive evaluation costs for the investigated networks. To facilitate practical use, a publicly available GitHub repository ( https://github.com/OVGU‐CoMe/PANN_Radioss ) provides the code generator and the implemented material routines. Within the supported model class, the generator accounts for different numbers and widths of fully connected hidden layers. Additional invariant inputs can be incorporated through localized changes to the input and derivative definitions, whereas different constitutive formulations or solver interfaces require more comprehensive adaptations. As a complex example, an impact simulation demonstrates that the generated PANN user materials can be applied to explicit analyses involving large strains and (self‐)contact. The results illustrate the feasibility and computational performance of the investigated hyperelastic PANNs in Simcenter Radioss and OpenRadioss . They also show how a commercial and an open‐source finite element solver can be used to study this specified class of machine‐learning‐based constitutive models.

The mechanical integrity of two commercially available lithium-ion battery separators was investigated under uniaxial and biaxial loading conditions. Two dry-processed microporous films with polypropylene (PP)/polyethylene (PE)/polypropylene (PP) compositions were studied: Celgard H2010 Trilayer and Celgard Q20S1HX Ceramic-Coated Trilayer. The uniaxial tests were carried out along the machine direction (MD), transverse direction (TD), and diagonal direction (DD). In order to generate a state of in-plane biaxial tension, a pneumatic bulge test setup was prioritized over the commonly performed punch test in an attempt to eliminate the effects of contact friction. The biaxial flow stress–strain behavior of the membranes was deduced via the Panknin–Kruglov method coupled with a 3D Digital Image Correlation (DIC) technique. The findings demonstrate a high degree of in-plane anisotropy in both membranes. The ceramic coating was found to negatively affect the mechanical performance of the trilayer microporous separator, compromising its strength and stretchability, while preserving its failure mode. Derived from experimentally calibrated constitutive models, a finite element model was developed using the explicit solver OpenRadioss. The numerical model was capable of predicting the biaxial deformation of the semicrystalline membranes up until failure, showing a fairly good correlation with the experimental observations.

The need for higher capacity battery cells has increased significantly during the past years. Therefore, the subject of this study is to investigate the behavior of high performance 21700 Lithium-Ion cylindric battery cells under several abuse conditions, represented by high mechanical loads with different velocities and states of charge (SoC), and to develop a finite element analysis (FEA) model, using the OpenRadioss’ explicit solver capabilities. The present study is focused on the investigation of the behavior of these cells under high mechanical loads with different loading velocities and different states of charge. The aim of the study is to provide a tool to predict the point of an internal short circuit in FEA, with a very good approximation. Experiments were completed using a hydraulic flat-compression test, set up at four different states of charge, 40%, 60%, 80% and 100%, and three different loading velocities of 10 mms⁻¹, 100 mms⁻¹ and 1000 mms⁻¹. A homogenized FEA model is developed to predict the internal damage of the separator, which can lead to a short circuit with a possible thermal runaway under abusive load conditions. The present model, in combination with well identified material and fracture parameters, succeeded in the prediction of the mechanical behavior at various states of charge and mechanical loading conditions; it can also be used for further crashworthiness analysis within a full-car FEA model. This accurate cell model will be the first building block to optimize the protective structures of batteries in electric vehicles, and reduce their weight through a deeper understanding of their overall behavior during the different crash cases.

The present study is focused on the development of a material model where the orthotropic-visco-elastic and orthotropic-visco-plastic mechanical behavior of a polymeric material is considered. The increasing need to reduce the climate-damaging exhaust gases in the automotive industry leads to an increasing usage of electric powered drive systems using Lithium-ion (Li-ion) batteries. For the safety and crashworthiness investigations, a deeper understanding of the mechanical behavior under high and dynamic loads is needed. In order to prevent internal short circuits and thermal runaways within a Li-ion battery, the separator plays a crucial role. Based on results of material tests, a novel material model for finite element analysis (FEA) is developed using the explicit solver Altair Radioss. Based on this model, the visco-elastic-orthotropic, as well as the visco-plastic-orthotropic, behavior until failure can be modeled. Finally, a FE simulation model of the separator material is performed, using the results of different tensile tests conducted at three different velocities, 0.1 mm·s⁻¹, 1.0 mm·s⁻¹ and 10.0 mm·s⁻¹ and different orientations of the specimen. The purpose is to predict the anisotropic, rate-dependent stiffness behavior of separator materials in order to improve FE simulations of the mechanical behavior of batteries and therefore reduce the development time of electrically powered vehicles and consumer goods. The present novel material model in combination with a well-suited failure criterion, which considers the different states of stress and anisotropic-visco-dependent failure limits, can be applied for crashworthiness FE analysis. The model succeeded in predicting anisotropic, visco-elastic orthotropic and visco-plastic orthotropic stiffness behavior up to failure.

In the present study, the mechanical properties of a dry-processed polyethylene (PE) separator are investigated in terms of deformation and failure limits. The focus is set on the anisotropic mechanical behavior of this material. A deeper understanding of the damage mechanism is important for further safety and crashworthiness investigations and predictions of damage before failure. It has been found that separator integrity is one of the crucial parts in preventing internal short circuit and thermal runaway in lithium-ion (Li-ion) batteries. Based on uniaxial tensile tests with local strain measurement, a novel failure criterion for finite element analysis (FEA), using the explicit FEA solver Altair Radioss, has been developed to predict the effect of high mechanical loads with respect to triaxiality, large plastic strain and orthotropy. Finally, a simulation model of a PE separator was developed combining the novel failure criterion with Hill’s yield surface and a Swift–Voce hardening rule. The model succeeded in predicting the anisotropic response of the PE separator due to deformation and failure. The proposed failure model can also be combined with other constitutive material laws.

Lithium-ion (or Li-ion) battery systems have been widely used as the main power source in new generation electric vehicles. Their mechanical performance is very important for vehicle crashworthiness design. In this research, a combined experimental and computational study has been performed to investigate the failure behavior of vehicle battery modules subject to high force wedge cutting. The tests were conducted on a commercial Li-ion battery module at very low speed using a high capacity material testing system. The force and voltage change as a function of cutting depth were measured, and the failure modes were analyzed. A finite element (FE) model based on an explicit FEA solver, was developed to simulate the wedge cutting process. The model predictions demonstrated reasonable agreement. In addition, the influence of several material constants in the failure model on the structural response was also discussed.

Rivets, especially self-piercing rivets (SPR), are a primary joining technology used in aluminum bodied vehicles. SPR are mechanical joining elements used to connect sheets to create a body in white (BiW) structure. To ensure the structural performance of a vehicle in crash load cases it is necessary to describe physical occurring failure modes under overloading conditions in simulations. One failure mode which needs to be predicted precisely by a crash simulation is joint separation. Within crash simulations a detailed analysis of a SPR joint would require a very high computational effort. The conflict between a detailed SPR joint and a macroscopic vehicle model needs to be solved by developing an approach that can handle an accurate macroscopic prediction of SPR behavior with a defined strength level with less computational effort. One approach is using a cohesive material model for a SPR connection. The paper describes cohesive element characteristics and calibration effort. Investigated element characteristics are an updated momentum calculation resulting from shear loads. It allows the adjustment of the width-height ratio for a constant meshing approach; handling of lateral distortion of cohesive elements based on sheet deformation or failure of sheet elements and an elemental stabilization if the cohesive element becomes unconnected that comes along with advanced material models and deletion of sheet elements. The updated element characteristics are shown on principle models and the calibration is shown based on coupon level samples. Outlook is given on a vehicle level and application for other joining technologies.