Issues on Gas-Fabric Interaction in Airbag Simulation Using LS-DYNA ALE

Computer Aided Engineering (CAE) has been deployed to help developing effective occupant restraint systems, such as airbags, in automotive industries for decades. Until recently, control volume method, which assumes a uniform pressure and density inside airbag, is still widely adapted in most airbag applications. Control volume method allows use of simple thermodynamic equations to efficiently model airbag. Using control volume method to simulate fully deployed airbags interacting with crash dummies, such as In-Position (IP) simulation, is appropriate. However, with the stringent safety regulations for protecting occupants with widely distributed sizes and sitting positions, as well as the implementation of side and knee airbags, Out-Of-Position (OOP) simulation becomes more and more important for airbag suppliers and OEMs. In OOP simulation, airbag starts to interact with occupants long before it is fully deployed. The non-uniform distribution of gas pressure inside airbag and the highly dynamic characteristics of airbag cushion invalidate the control volume method. To address this issue, fluid-solid interaction (FSI) is implemented in various codes in different forms. The very high speed gas interacts with soft fabric in airbag simulation is quite a challenge for conventional Computational Fluid Dynamic (CFD) codes, and special treatment to deal with this FSI problem should be carefully planned and developed. Arbitrary Lagrangian Eulerian (ALE) approach from LS-DYNA provides a possibility to model this multi-phrase highly dynamic problem. For OOP simulation, the ALE should be computationally efficient with acceptable accuracy. Normally, gap between layers of airbag is about the same order of magnitude of fabric thickness for flat or folded airbag model. To be computationally efficient, the size of Eulerian elements should be much larger than the fabric gap. This introduces the difficulty for code to handle gas-fabric interaction properly. Larger Eulerian element size slows down the gas propagating speed and causes discrepancy between simulations and testing for airbag deployment. Properly use of initial volume fraction definition from LS-DYNA to introduce gas into cushion fabric gap at time zero, can improve the results without using ultra small Eulerian elements for flat airbag model. However, for folded airbag, the application of initial volume fraction is not so successful. In present study, issues using ALE for airbag simulation will be investigated using several simple test cases. Recommendations for further improving the ALE code for airbag OOP simulation are presented.