Simulation of Rollover crash events

Transcription

1 Simulation of Rollover crash events Use of DEFORMABLE_TO_RIGID 1 Anders Jonsson, DYNAmore Nordic Christofer Karlsson, Scania

2 Background Outline Scania s work with active and passive safety Heavy trucks rollover accidents Simulation challange Set up using *DEFORMABLE_TO_RIGID Simulation results Comparisons to testing Dummy injury values Cab acceleration signals Conclusions 2

3 Scania Premium products and services 3 Premium products and services Haulage Construction Distribution Special purpose Network and services City and suburban Intercity and coach Used vehicles Engines

4 Heavy truck rollover accidents 4

5 Accident statistics Fatalities Note: Compiled by ACEA WG-TS l Department/Name/Subject Your date here

6 Heavy truck rollover accidents in the EU Roll over accidents from Volvo ART Report

7 Vehicle safety systems Cab strength Electronic Stability Program Lane Departure Warning and Drowsiness Detection Adaptive Cruise Control Blind Spot Aid Vulnerable Road user Detection Visibility Design to minimise injuries on other road users l Department/Name/Subject Your date here

8 Simulation challenge Traditional frontal crash typical duration 150 ms 8

9 A roll over event takes long time... Simulation challenge 9 Time of impact

10 *DEFORMABLE_TO_RIGID in LS-DYNA A large part of the first phase of the rollover (before impact) consists mainly of rigid body motions, but it still might be of interest to study the side-to-side rocking motion of the seats and dummies. In order to save time, most of the model is turned into rigid material during this phase. This both reduces the number of independent DOFs and allows for an increased timestep. A brief guideline how to use deformable to rigid follows: Use LS-DYNA R7.1.1 Use *DEFORMABLE_TO_RIGID_AUTOMATIC for switching parts. New rigid parts with common nodes will be merged automatically. As a consequence of this, do not use permanently rigid part as master rigid bodies. If there are common nodes, parts may remain merged also after switch back to deformable, since permanently rigid parts cannot be switched to deformable. CNRBs are handled correctly and will be re-born after switch back to deformable. 10

11 *DEFORMABLE_TO_RIGID in LS-DYNA 11 Constraint-based tied contacts will not work between parts that are switched to rigid. Use birth time to activate them right after the switch back to deformable, or alternatively use penalty based tied contacts. Restart can be used for modifying the time step for complex configurations.

12 *DEFORMABLE_TO_RIGID in LS-DYNA 12 The simulation is set-up in three stages: 1 st run, initialization in deformable mode 2 nd run (restart), up to just before impact. Switch to rigid mode and increase time step. 682 parts are switched. 3 rd run (restart), until end. Switch to deformable mode and reduce to original time step Model size is deformable elements Total analysis run-time (for all three stages) on 32 cores is 64 h 37 min for 1500 ms simulation time. Estimated time savings due to use of deformable-to-rigid: 71 h.

13 Model partitioning for def2rigid 13 The model was partitioned in three main groups: Cab BiW, IP/interior and rest of the truck One rigid body Seats Dummies and seatbelts Deformable throughout Each seat, in turn, was partitioned into three parts: Seat upper Suspension arm 1 and arm 2

14 Results Deformation results 14

15 Comparison Passenger side seat backrest Y-motion 15

16 Results 16 Increased time step during the rigid phase

17 Results overview Injury values relative to test results 18

18 Comparison Driver resultant head acceleration 19 About 10 % lower acceleration in the simulation. Also the timing is a bit off

19 Comparison Driver side chest compression 20 Fair agreement when it comes to the peak chest compression.

20 Comparison Driver neck forces 21 Test results = red, green Simulation = blue, black

21 Comparison Driver belt forces 22 Test results = red, blue Simulation = black, green

22 Comparison Passenger resultant head acceleration 23 The conspicuous second peak from the test is not clearly reproduced in the simulation.

23 Comparison Passenger neck forces 24 Test results = red, green Simulation = blue, black

24 Comparison Passenger belt forces 25 Test results = red, blue Simulation = black, green

25 Comparison Passenger neck moment 26

26 Comparison Cab acceleration signal, peak magnitude values 27 The biggest differences is found for LH B pill mid, due to that, in the simulation, the driver seat impacts the B-pillar close to the accelerometer

27 Conclusions and summary 28 *DEFORMABLE_TO_RIGID can save substantial amounts of time, but set-up requires some considerations. In this case, restart was required for controling the time step during the rigid phase. The simulation results show at least some kind of resemblance to the testing results

28 29 N o r m a l t e r m i n a t i o n Troubleshooting convergence problems in LS-DYNA

29 Heavy truck rollover accidents in the EU Roll over accidents from Volvo ART Report