U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-04-096
Date: August 2005
Evaluation of LS-DYNA Wood Material Model 143
PDF Version (8.16 MB)
PDF files can be viewed with the Acrobat® Reader®
14 - Dynamic Wood Post Test Simulations
The model used for dynamic post test simulations is shown in figure 86. The bogie impact height is 610 mm (24 inches) above ground level. The impactor on the bogie is modeled with a rigid cylinder wrapped with a deformable neoprene model. The ground is modeled with rigid, fixed solid elements. Deformable neoprene is placed between the ground and the post on both the back and front sides of the post. On the impact side of the post, the neoprene sticks out of the ground approximately 200 mm, while on the other side, the neoprene was slightly below ground level.
The dynamic model went through a few modifications. The two primary changes were: (1) the neoprene cover on the cylinder was remeshed to a much finer mesh, and (2) the neoprene used between the ground and the post was also remeshed and sized to match the majority of the testing conditions. Finally, unusual energy problems exhibited on certain hardware platforms with the bogie wheels were fixed by making the wheels rigid (this had no effect on the simulation results).
A baseline model was established to compare the various dynamic post test simulations. This baseline model used the default wood material properties for southern yellow pine that had a grade of 1, a moisture content of 30 percent, and a temperature of 20°C (*MAT_WOOD_PINE). Note that the different colors on the post shown in figure 86 represent the different parts being used to store those post elements. This was done to monitor the behavior and the energy distribution more precisely. The material properties for all of the post parts were identical.
Figure 86. Dynamic wood post test model.
The initial simulation of the dynamic post test resulted in the post vaporizing at around 10.5 ms into the event (as shown in figure 87).
Figure 87. Post vaporization.
After many iterations and model changes, the problem was isolated and proven to be a time-step problem. The LS-DYNA calculated time step for the 25.4-mm hexagonal element using the wood material is 0.005 ms. This time step is too large and can cause model instabilities as shown in figure 87. It was found that by limiting the time step to 0.001 ms for the 25.4-mm hexagonal, the material remained stable for this specific simulation case. The maximum time step allowable can be set in LS-DYNA using the *CONTROL_TIMESTEP command.
With a reduced time step, the dynamic simulation proceeded farther into the impact, but resulted in contact difficulties at the sharp edge between the wood post and the neoprene-lined concrete base (as shown in figure 88).
Figure 88. Contact at sharp corner.
This type of contact penetration is typical when sharp edges are involved. The sharp edge contact was originally analyzed for static post testing simulations. The correction was to round off the sharp edge. For the dynamic wood post model, the contact correction will be to add a neoprene flap to that side of the foundation, identical to the flap on the other side. Note that, in testing, both cases (with and without double flaps) were tested with the results showing no identifiable differences.
With the double flap in place, simulation of the dynamic post test continued until completion. However, the post did not snap off as it did in physical testing. Instead, the post bent about the last remaining elements (as shown in figure 89). Several runs were made with various contact friction properties to see if this bending (instead of snapping off) was the result of something other than the material itself. However, the results were similar between all runs, indicating that something was not quite right in the material model or its parameters for pine.
Figure 89. Post bending.
The developer recommended an investigation into the cause of the bending elements, possibly related to the neoprene. To start, the neoprene was removed from the model and a rigid sleeve was used to constrain the post. To avoid potential contact troubles, the rigid sleeve was rounded at the edge. However, penetrations were observed as shown in figure 90. Even though the soft option for contact was being used in this model, the contact needed additional study to eliminate significant penetrations. To fix the trouble, an additional node-to-surface contact was added between the rigid sleeve and the wood post. This addition seemed to work appropriately (as shown in figure 91). Note, however, that this was not the only contact difficulty experienced during this investigation. For some reason, it appears that the new wood material model was overly sensitive to contact behavior. Further study is recommended.
Figure 90. Contact penetrations.
Figure 91. Improved contact.
With this latest model (i.e., rigid sleeve with no observable contact penetrations), the results were undesirable. Specifically, even with the previously discussed smaller time step of 0.001 ms being used, the lower post began to vaporize (as shown in figure 92). Also, the volume of the elements at the bottom of the post began to significantly expand (as shown in figure 93). This was discussed previously for a single-element model and it was noted that the user did not believe that this volume expansion was realistic.
Figure 92. Element vaporization at bottom of post.
Figure 93. Volume expansion of 75 percent in an element near the bottom of post.
Next, because vaporization had returned to the model, it was thought that the cause might be a result of the time step (as discussed in section 6.2). However, when the time step was reduced again to 0.0005 ms and then even further down to 0.0001 ms, the results did not improve. In fact, using the very impractical time step of 0.0001 ms, vaporization occurred much more quickly in the model and caused the model to become unstable much earlier than with the larger time step. The deformed geometry of the post just before becoming unstable is shown in figure 94.
Figure 94. Vaporization with a time step of 0.0001 ms.
In a final attempt to determine more about the difficulties occurring with the dynamic post model (i.e., contacts, vaporization, bending instead of snapping off, etc.), the rigid-sleeve model with some contact penetrations (which alleviated the vaporization) was investigated near the damage area. It was believed that the failure of the wood elements away from the contact area would be unaffected by the contact penetrations. Unfortunately, the damaged area appeared to be incorrect (as shown in figure 95) (i.e., the elements in the center of the damage area became highly distorted, but did not erode). Plotting the damage of one of those elements showed that the element should have eroded since it had reached the critical value of 0.99 (the preset damage level for erosion) (see figure 96).
Figure 95. Highly distorted elements sometimes do not erode.
Figure 96. Damage of highly distorted element.