Example: Automotive Application

For automotive applications, a target time step set to a value of 10 times the reference (nodal based constant) time step is a reasonable value to start with. However, greater than 20 times higher than the reference time step have been successfully utilized.

Table 1. Summary of Full Vehicle Tests Tried with AMS
Load Case Vehicle Program Radioss Version Number of Cores Model Zize Original Nodal Time Step Target Time Step Speed-up v. Original Nodal Time Step Comments ODB

Offset Deformable Barrier

FFB

Full Frontal Barrier (rigid)

Front crash Neon public 10.0.2   1M 1 μ s 10 μ s 4 x  
Taurus public FFB 11.0.201   0.8 M 0.5 μ s 5 μ s 2.83 x Rigid wall replaced by rigid mesh and contact interface
OEM1 FFB 11. 24 x 1 1.3 M 0.5 μ s 5 μ s 3 x Figure 1
OEM2 ODB 11.0.230 32 x 1 2 M 0.5 μ s 5 μ s 1.5 x Figure 2
OEM2 ODB 11.0.230 6 x 6 2 M 0.5 μ s 5 μ s 2 x  
OEM4 FFB 11.0 32 x 1 1.5 M 0.5 μ s 4 μ s 1.9 x  
OEM4 FFB 11.0.230 8 x 6 6 M 0.5 μ s 5 μ s 1.3 x Rigid wall replaced by rigid mesh and contact interface
Rear crash OEM1 11.0_sp 16 x 1 1.1 M 0.1 μ s 2 μ s 5.8 x Figure 3
OEM1 11.0 16 x 1 1.1 M 0.1 μ s 2 μ s

6.5 x

Side impact OEM4 11.0 32 x 1 1 M 0.5 μ s 10 μ s 2.1 x Figure 4
Roof Crush OEM4 11.0 16 x 1 1.2 M 0.5 μ s 10 μ s 2.9 x Figure 5
OEM4 11.0 32 x 1 1 M 0.5 μ s 10 μ s 3.4 x  
Taurus public 11.0 16 x 1 0.8 M 0.5 μ s 10 μ s 4.6 x Figure 6

All speedup's compare original nodal time step double precision with AMS in single precision.

Automotive Related Results



Figure 1. OEM1 Full Frontal


Figure 2. OEM2 Frontal Offset




Figure 3. Front Rear Impact


Figure 4. OEM4 Side Impact


Figure 5. OEM4 Roof Crush


Figure 6. Ford Taurus Public Roof Crush


Figure 7.
Table 2. Summary of Full Vehicle Tests tried with AMS in Version 13.0
Load Case Vehicle Program Radioss Version Number of Cores Model Size Original Nodal Time Step Target Time Step Speed-up v. Original Nodal Time Step Comments ODB Offset Deformable Barrier

FFB Full Frontal

Front crash Neon public 10.0.2   1M 1 ms 10 ms 4   Neon public
Taurus public

FFB

13.0 3 x 6 x 2 0.8 M 0.5 ms

/dt/inter/del 0.4 ms

5 ms tol 0.001

/dt/inter/del 0.4 ms

1.87 Rigid wall replaced by rigid mesh and contact interface
OEM1 FFB 11. 24 x 1 1.3 M 0.5 ms 5 ms 3  
OEM2 ODB 11.0.230 32 x 1 2 M 0.5 ms 5 ms 1.5  
OEM2 ODB 13.0 3 x 6 x 2 2 M 0.5 ms

/dt/inter/del 0.1 ms

5 ms tol 0.001

/dt/inter/del 0.1 ms

1.51  
OEM4 FFB 13.0 3 x 6 x 2 1.5 M 0.5 ms

/dt/inter/del 0.001 ms

4 ms tol 0.001

/dt/inter/del 0.001 ms

0.96  
OEM4 FFB 11.0.230 3 x 6 x 2 6 M 0.5 ms

/dt/inter/del 0.1 ms

5 ms

/dt/inter/del 0.1 ms

1.3 Rigid wall replaced by rigid mesh and contact interface
Rear crash OEM1 13.0 3 x 6 x 2 1.1 M 0.1 ms

/dt/inter/del 0.04 ms

/dt/brick/cst 2.0 ms

2 ms tol 0.001

/dt/inter/del 0.04 ms

/dt/brick/cst 2.0 ms

4.03  
Side impact OEM4 13.0 3 x 6 x 2 1 M 0.5 ms 10 ms tol 0.001 2.11  
Roof Crush OEM4 meshed 13.0 3 x 6 x 2 1.2 M 0.5 ms 10 ms tol 0.001 2.83 Rigid wall replaced by rigid mesh and contact interface
OEM4 meshed 13.0 3 x 6 x 2 1 M 0.5 ms 10 ms tol 0.001 1.13 Rigid wall replaced by rigid mesh and contact interface
Taurus public

meshed rigid ram

13.0 3 x 6 x 2 0.8 M 0.5 ms 10 ms tol 0.001 3.80 Rigid wall replaced by rigid mesh and contact interface
Taurus public

rigid wall ram

13.0 3 x 6 x 2 0.8 M 0.5 ms 10 ms tol 0.001 2.18 Rigid wall parallelepiped

All speedup's compare original nodal time step with AMS both in double precision. The greyed rows relate to models that were not rerun in 13.0 and speedups belong to Table 2.





Figure 8. Ford Taurus Public Roof Crush