How to get uniform tension in a rail using dynamic/explicit analysis?

[Kamyar_K]

Hello everyone,

I’m working on a 3D simulation of a railroad track in ABAQUS using a dynamic/explicit analysis to investigate the longitudinal behavior of the rail. My model consists of multiple steps. In the first step, I apply a tensile force at one end of the rail, while the opposite end is fixed in the axial direction (see Figure 1). The force is applied using a linear amplitude over 0.1 seconds, followed by an additional 0.1 seconds to allow the system to stabilize.

The objective of this step is to generate uniform axial tension along the rail equal to 170 kips. This approach works well for short models (i.e., with a small number of crossties). For example, in a model with 10 crossties, cross-sectional force measurements show consistent axial force along the rail length (see Figure 2).

However, when I extend the model to include 200 or 680 crossties, the axial forces are no longer uniform along the rail (see Figures 3 and 4). The axial force in the middle cross sections continues to increase and then fluctuates around values higher than the intended tension. Since my ultimate goal is to simulate a long section of track, I would appreciate any guidance on how to achieve and maintain uniform axial tension in longer models.

Any suggestions or insights would be greatly appreciated.

Figure 1. Railroad track model


Figure 2. Model with 10 crossties



Figure 3. Model with 200 crossties (only few number of cross-sectional forces have been shown)


Figure 4. Model with 680 crossties (only few number of cross-sectional forces have been shown)

 

[FEA way]

Why are you using explicit solver for that ? Even if later steps require it, it's better to solve midly nonlinear quasi-static stages in Abaqus/Standard and use the import technique to continue the analysis in Abaqus/Explicit.

 

[rb1957]

the crossties are reacting some of the rail axial load, yes? Release this freedom.

the panels too are going to be picking up some of the load too, no?

You want to release this loadpath as it is causing bending in the rail, as these reactions are offset from the applied load.

 

[Kamyar_K]

Why are you using explicit solver for that ? Even if later steps require it, it's better to solve midly nonlinear quasi-static stages in Abaqus/Standard and use the import technique to continue the analysis in Abaqus/Explicit.

I see, thanks for the tip. I tried to do the import technique, however, the second model gets stuck in "submitted" phase and doesn't continue running. Do you have any idea or any reference that I could use to overvoem the issue?

 

[Kamyar_K]

the crossties are reacting some of the rail axial load, yes? Release this freedom.

the panels too are going to be picking up some of the load too, no?

You want to release this loadpath as it is causing bending in the rail, as these reactions are offset from the applied load.

Thank you for your reply.
There is no interaction between rails and crossties in this step, the rail is supposed to get pulled individually without transferring load to other components.

 

[FEA way]

I tried to do the import technique, however, the second model gets stuck in "submitted" phase and doesn't continue running.

Export the input file and submit it outside of Abaqus/CAE, using command line. Then monitor the log files.

 

[rb1957]

"There is no interaction between rails and crossties in this step, the rail is supposed to get pulled individually without transferring load to other components."

I read that to mean there is no intended interaction, but what does the model tell you ? Has the axial load (as well as the stress distribution) changed along the length of the rail ?