Abstract
Research and/or Engineering Questions/Objective
When and how to deploy an airbag (or not) during an “acceleration event”? The goals are: - Tuning the airbag triggering algorithm, based on crash simulations - Using a smart triggering criterion in crash simulation and no longer a fixed deployment time The algorithm is represented in a System Modeling language (Modelica) and the crash simulation is performed with Virtual Performance Solution solver (PAM-CRASH). A coupling is made between VPS Solver and CyVertex, ESI’s Modelica implementation.
Methodology
A demonstrator is made using a simple vehicle model, with realistic dummy and restrain system. Based on this model, several simulations are made: - At relatively low velocities, when the airbag should or should not deploy o According to NHTSA, the airbags should never deploy below 8 PMH and always over 14 MPH - At normal crash velocities (35 and 40 MPH) Simulations with low velocities allows designing a criterion that discriminates the acceleration levels at 8MPH and at 14MPH with various obstacles. This criterion will be then used in the next part of the study and is represented by a Modelica model. Thereafter a coupling between VPS (Virtual Performance Solution) solver and CyVertex is made. During the simulation, each time step (0.1 ms): - VPS delivers accelerometers signals CyVertex delivers the airbag firing order when necessary.
Results
The airbag triggering criterion was based on 24 simulations: 12 simulations at 8MPH and 12 other at 14MPH. For all the 14MPH conditions, the criterion triggers the airbag and at 8MPH, the criterion does not trigger 11 times over 12. The criterion is more than sufficient for a demonstrator. A next step proved that this criterion gives better results at high velocities (35 and 40 MPH) than fixed times (18, 22 and 26 ms after T0). Finally, the airbag triggering criterion is implemented in Modelica to be used in combination with VPS solver. The crash simulation introduces a higher computation time (+50%), that should be reduced in next versions of the coupling framework. A full view of the model (i.e. Crash model + Modelica system model) can be obtained using the application Visual-Systems in ESI’s Visual-Environment software.
Limitations of this study
During the criterion design phase, we assumed that the vehicle data were coming from accelerometers. In this first version of ESI’s coupling framework (so called XMX), communication is based only on nodal displacement, rather than accelerations, to avoid introducing an anti-aliasing filter. At the time FISITA congress will occur, this limitation will have been removed. The overshoot of simulation computation time should be reduced as well (target: + 15%).
What does the paper offer that is new in the field including in comparison to other work by the authors?
This study is new and was never published.
Conclusions
Introducing airbag triggering criterion in crash simulations improves the reliability of the models. The airbags are not deployed at a fixed time but at the time they would be deployed in a real crash. It becomes easier to implement triggering criteria based on a large variety of acceleration events (crashes and all events generating abnormal accelerations). The set-up of this coupled model can be fully achieved using the software application Visuals Systems. This will lead in the future to Hardware in Loop applications, where Electronic Control Units will be connected to the crash model to directly test ECU algorithms.
Key Words : Airbag triggering; System modeling coupling; Crash simulation