Simulation of a lightweight battery housing made from Long Fibre Reinforced Thermoplastics (LFT) with local UD-Tape reinforcements

Created by Dr. Marcus Stojek | | Technical Article

Integrative simulation of a battery housing considering anisotropic material behavior of LFT compound and multilayered Tapes.

Introduction

Battery housings in skateboard design represent a relatively new field of application in the automotive sector for the design of plastic components. In addition to short-fiber-reinforced thermoplastics, long-fiber-reinforced and continuous-fiber-reinforced thermoplastics are increasingly attracting the attention of developers, especially due to the high stiffness and strength requirements combined with potentially high unit numbers.

In a publicly funded joint research project LightMat Battery Housing, PART Engineering, together with partners from industry and university, investigated the possible applications of thermoplastic LFT pressed components, which are locally reinforced with UD tapes. Another focus in the project was the functionalization of these tape scrims, e.g. with regard to EMC, fire protection, electrical conductors, etc., in order to achieve a higher degree of integration.

For PART Engineering, the main task was to be able to effectively and reliably calculate such components with a complex material structure even in early development phases and to take into account the properties of the layered tape profiles used. The focus was on a user-oriented conception of the individual work steps. In particular, only solver-internal material models that are available to any simulation specialist were used. The tests required to calibrate the material models were limited to quasi-static tensile and flexural tests. However, local fiber orientation was considered for both the LFT compound and the UD tape scrims. All material models used were formulated anisotropically.

For the pressing compound, the corresponding process simulation provided the necessary fiber data. (A new interface was implemented in Converse for this purpose.) For the tape scrims, the local orientation of the scrims was taken from a drape simulation. The corresponding mapping tool currently exists only as a prototype.The knowledge gained in the project and the workflows developed will be incorporated both in our service portfolio and in our software products (Converse, MatScape).

Test data and material models

For the description of the LFT compound, sheets were produced in 3, 4 and 5 mm thickness with different positioning of the compound which significantly determined the resulting fiber orientation. The comparison of the calculated orientations shows an only moderately pronounced orientation profile over the wall thickness as well as a clear overestimation of the degree of orientation by the simulation (see Figure 2).

Nevertheless, the tensile and bending tests show a clearly pronounced anisotropy of the samples. The material model was created and calibrated in Converse using the orientation profiles from the pressing simulation. Figure 3 shows the comparison of the measured data with the calculated curves. The material model reproduces well all the tests performed (0°, 90° and 45°).

For the description of the different tape structures, solver-internal material models were used (ABAQUS). The reinforcements were modeled as individual UD layers and not as anisotropic, homogeneous shell elements. Depending on the component area, the boundary conditions of the tape deposition process result in quite complex layer structures, which had to be represented in the FE model. Figure 4 shows the different subareas of an X-shaped floor bead with different thicknesses of tape and molding compound. Each of the layers shown has its own main fiber direction in the FE model.

As a prototype, a module for creating material models for layered UD Tapes with thermoplastic matrix has been implemented in our MatScape Software (see Figure 5). This will be available with one of the next release versions.

For the final validation of the individual material models in the hybrid structure, test specimens were manufactured from back-pressed UD fabrics and tested in tensile, flexural and puncture tests.

Figure 6 shows the test and simulation results of some puncture tests. It should be noted that only quasi-static test data were used to calibrate the material models, i.e. no strain rate dependency was considered.

Setup of FE-Models

The technology demonstrator considered in the project consists of the housing and cover components (see Figure 1). Both were manufactured as LFT moldings from PA-GF45. In different variants, semi-finished products and mats made of UD tape layers (glass, aramid) or organic sheets were integrated into the housing in order to achieve a local increase in strength and stiffness. Figure 7 shows the different profiles that were manufactured and used during the project.

In detail, the following steps are required in the developed process flow:

  • Creation of a midplane model
  • Assignment of finite elements to groups of the same structure (only pressing compound, pressing compound + tape, if necessary only tape/organosheet)
  • Mapping of local fiber directions (and degree of orientation for the LFT compound) to the shell model. This step is done automatically in Converse. In addition, the thickness information of the respective layer is transferred in the process.
  • Calibration and validation of an ansiotropic elastic-plastic material model for the LFT molding compound. Input data are the isotropic properties of the matrix as well as the fibers and the fiber content. At least one stress-strain curve with a known degree of orientation is required for the calibration of an elastic-plastic material model. The current release version of Converse has been used for this step.
  • Calibration and validation of an anisotropic material model for the tape fabrics used. Input data are fiber and matrix properties, fiber content and scrim structure. This step is done prototypically in the Laminate Model of MatScape. A damage model currently still has to be defined manually.

Results

In the overall model of the technology demonstrator, both the quasi-static crush test in different variants and the dynamic bottom-impact test were considered. For the crush test, the entire component was tested in a first test series, and the separated lateral crush structure was tested in a second round. The evaluation parameters in all tests were the first occurrence of fractures/cracks and the first contact of the housing wall with one of the battery modules. Figure 8 shows the comparison of test and simulation data for both test series.

The bottom impact test consists of the impact of a steel ball on the bottom of the housing. The evaluation value here is the penetration depth, or again the possible contact between the housing wall and the battery modules. Figure 9 shows a corresponding overview of the results.

Summary

The investigations have shown that using solver internal material models and calibrating them with simple tensile and bending tests, can be sufficient in order to perform meaningful simulations for continuous fiber-reinforced pressed components. In particular, for the molding compound, the procedures available in Converse for creating material models for short-fiber-reinforced thermoplastics could be adopted almost unchanged.

The prediction of the achievable maximum load and the comparison of different design variants can be used for the development process, where they represent a valuable input. In addition to the tolerable max load, the location of first failure was also predicted well in the simulation.

The described approach reached its limits when it came to predict the post-fracture behavior of the UD tapes. Here, both more elaborate tests and more complex material models are required for the description. Likewise, due to the component geometry and the load situation, it was not possible to investigate pronounced shear loading on the tape layers.

However, the investigations have also shown that the processing influence can dominate the component behavior to such an extent that the simulation results deviate significantly from the test. In Figure 8 for example, the load-bearing capacity of variants 2_T_O and 3_T_V_O for the crush structure is clearly overestimated. In the detailed evaluation of the tested components, an insufficient connection between the tape and the LFT compound mass was found to be the cause.

We would like to thank the project sponsor for funding the project and our project partners for the excellent and productive cooperation.

Autor
Dr. Marcus Stojek, Managing Director at PART Engineering GmbH, Bergisch Gladbach

 

Back