Week - 4 - Crash Box Simulation

OBJECTIVE

To simulate a crash test of a meshed crash box on a rigid wall. The crash box will be of two differing thicknesses (1.2mm and 1.5mm) and in addition to generating their animations, their results need to be compared based on the following outputs:

MODEL IMAGE

PROCEDURE

1. Firstly, on LS PrePost, the .k file of the model is imported via File > Import > LS-DYNA keyword file.

2. After importing, we can rename the parts of the model. This is done by accessing the keyword manager on the right toolbar and going to PART On selecting it, we are met with the PART keyword input form, where we can edit properties of the parts in the model. For now, we can assign the name and the PID (Part ID). We shall look into SECID (Section ID) and MID (Material ID) separately.

3. The part contains shell elements and we are to assign thickness to it. So, we can access the keyword manager again, select the 'all' option, scroll to SECTION and select SHELL under it. The SHELL section will be used to define the crash box. We can assign a section ID with ELFORM 2 and thickness 1.2mm. This thickness value is to be changed (to 1.5mm) later when we carry out case 2.

4. Similarly, to create a material, we can go to the keyword manager again, select 'all', and scroll down to the MAT card, under which we can select the 024-PIECEWISE_LINEAR_PLASTICITY. This card is used to define the crash box's material. We can assign the material ID (MID), density (RO), Young's Modulus (E), and Poisson's Ratio (PR) here (as of steel in this case). Once we are done, we can click 'accept'. (Here, I have also defined SIGY (Yield stress) and ETAN (Tangent Modulus))

Now, we can go back to the part card and assign the MID and SECID of the part as discussed before.

5. Moving on, we can now define the boundary conditions. The aspects we need to define are the moving body and a rigid wall.

Firstly, we can assign a velocity condition to the crash box part. To do so, we need to select the 'Create Entity' tool from the right toolbar as shown. In the entity creation window, select 'Initial' and 'Velocity' under it. Then, using the same node selection process, we shall select the crash box's nodes. Then we can assign a value for Vx. Since it will be moving in the negative x-direction, we can give it a value of -13.88 mm/ms (-50 kmph). We can then click 'Apply' and then 'Done'.

The following screenshot shows the selected nodes with the initial velocity keyword applied:

For rigid wall creation, the planar wall method was used from the same create entity menu.

The 1n+NL option creates a rigid wall with the help of one node and a normal from the node. Selecting a node at the edge of the crash box as a reference and manipulating the X coordinate values for the tail (T) and head (H) of the normal helps us generate the required rigid wall. Since we had already defined the velocity in the negative X-direction, care must be taken to ensure the rigid wall is generated in that same direction.

6. Moving on, we need to define the self-contact for the crash box. To do so, we can go back to the keyword manager, choose to show all keywords, go to the CONTACT keyword and select AUTOMATIC_SINGLE_SURFACE. Here, we just need to define the slave of this contact property using the part ID. To enable this option, we need to select '3' for SSTYP (Slave segment set or node set type). Once we do that, we can pick the slave, which would be the entirety of the crash box. In addition to naming the contact and assigning a CID (contact ID), we needn't worry about anything else here. We can click 'accept' and then 'done'.

7. Next, we can create a CONTROL card to specify the end time of the simulation. This can be assigned via the TERMINATION card option under the CONTROL keyword in the complete list of keywords in the keyword manager. We can assign a value of 2ms to capture the entire simulation.

Finally, we can assign a couple of DATABASE keyword cards for the outputs - namely ASCII_option and BINARY_D3PLOT with a DT (Time interval between outputs) of 0.02ms. The GLSTAT, MATSUM, SLEOUT, NODOUT, RWFORC, SECFORC attributes are selected under ASCII_option.

The following screenshot shows the GLSTAT and MATSUM options selected. Energy values are written on a part-by-part basis in MATSUM and energy plots overall are evaluated using GLSTAT.

We are also required to query specific output requests. Firstly, we are to generate the strain outputs. To do that, we can create a keyword by going to Keyword manager and selecting DATABASE>EXTENT_BINARY. The DATABASE_EXTENT_BINARY card with STRFLG =1, would be used to compute the elastic strain in the model.

Then, selecting DATABASE>HISTORY_NODE lets us request outputs at specific nodes. Here, we are to compute the acceleration at node 5094 in the crash box.

Next, we can select DATABASE>CROSS_SECTION_SET to request output. This would be used to compute the cross-sectional forces in the model. We are to define the node set (NSID) and shell set (SSID) of the location of the cross-section slice as shown.

8. Finally, we can check the created deck for errors using the option keyword manager>model check.

Once done, we can save it as a .k file and solve it using LS-RUN. The keyword file is inserted and we can click the play button to run the simulation. The number of cores to be utilised can be changed if needed.

Once it is finished without errors (aka Normal Termination), we can reopen LS-Prepost to view the results.

We can go back to the section card and change the thickness to 1.5mm and save the .k file as a separate one and run a simulation for that as well.

RESULTS

Directional Stress
We can input the binary d3 plot keyword file and plot the contour animation for x and y directional stress. To do that, we need to go to the post-processing deck and click on the fringe component tool. We can select the particular contour characteristics to plot from the given list here. The following are the contour animations of X & Y directional stresses for each case:

1.2mm

X-Direction

Y- Direction

1.5mm

X-Direction

Y-Direction

Directional Strain

Just as we did for directional stress, we can go to the same fringe component tool to access strain rate options, where we can select the required directional strain options to generate the contour animations.

1.2mm

X-Direction

Y-Direction

1.5mm

X-Direction

Y-Direction

Energy Plots

To generate plots, we need to access the ASCII tool in the post-processing deck. Here, we can select the type of plot data (glstat, matsum, etc) and pick the plots to be loaded accordingly. For energy plots, we shall be selecting glstat and the energies to plot would be internal, kinetic, total, hourglass, and sliding energies.

1.2mm

1.5mm

CROSS-SECTIONAL FORCE GENERATED IN THE MIDDLE OF THE COMPONENT

For this plot, we need to select 'secforc' from the list of plot data in the same ASCII tool window (instead of glstat last time). When we do that, we have the option of selecting the particular section we created, of which we can select the output we are looking for. Here, I went for X-directional force and resultant force.

1.2mm

1.5mm

ACCELERATION PLOT OF NODE 4853

As in the previous plot, we can select 'nodout' this time (instead of 'secforc'). Loading it lets us select node 4853 in the next section, in addition to selecting the required plots. Here, we are going for x-acceleration and resultant acceleration.

1.2mm

1.5mm

OBSERVATIONS

NB: Case 1 = 1.2mm thickness; Case 2 = 1.5mm thickness

Comparing the X and Y directional stress and strain contours for both cases:

Case 2 generates lesser stress in the direction of the impact, probably due to the higher cross-sectional area of the component (Stress is inversely proportional to the cross-sectional area). Furthermore, there is lesser deformation in case 2 as evident from the strain values, probably due to its higher energy absorbing capacity as a result of the increased thickness.

With more material, the higher the impact force is. Which explains why the cross-section in the 1.5mm thick crash box experiences a higher peak force of 70N (compared to 1.2mm's 40N).

Coming to the node acceleration, both cases have similar peak resultant accelerations of 4E+3 mm/ms^2. But case 2 with 1.5mm thickness experiences higher x-direction deceleration with a value of 4E+3 mm/ms^2, compared to the 3E+3 mm/ms^2 in case 1.

Theoretically, the crash box with the higher thickness should be able to absorb more of the impact energy. Case 1 absorbs 0.1E+6 N-mm of energy on impact (internal energy) whereas 0.125E+6 N-mm of internal energy is generated in case 2. Understandably, this is due to more material in case 2 as a result of the increased thickness. Due to the same reason, case 2 has higher kinetic energy (and total energy as a result) as well.

CONCLUSION

The given crash box model was simulated in two cases and the required outputs were generated. The increased thickness in case 2 definitely had an impact on the values and from a safety viewpoint, the 1.5mm thick crash box would be preferred since it was capable of absorbing more energy.