Learning Forum

[Qusai Alomari]

Hello,

I'm trying to model a full RC bridge using LS-DYNA. In this model, one of the pier columns is subjected to the impact of a Single Unit Truck. After that, an explosion charge is detonated. 

I'm using the ALE approach to model the Air, Soil, and Explosive, and all structural elements are modeled using solid elements (except steel reinforcement which is modeled using beam elements). Unfortunately, I'm encountering some issues that I couldn't resolve over the last two weeks. 

All Lagrangian solid elements are deleted at the very beginning (first two to three time steps), and I get error termination even without an explicit error message. I've tried several things to resolve the problem, as summarized below, but nothing changes:
1- reducing time step scaling factor (TSSF) and (DT2MS).
2- Changing solid element type to constant stress (Elform = 1).
3- Using Types 4 and 5 hourglass controls with a coefficient ranging between 0.01 and 0.05.
4- Reconstructing the mesh.
5- Using IMASCL = 2 in the Control ALE option.

Any recommendations?

[Loic Ivaldi]

Hello,

don't use mass scaling when you're at an early stage of your simulation

[Jim Day]

I concur with Loic. If problem persists, de-couple Lagrangian from ALE (comment out the coupling keyword(s)), just as an experiment, to see what happens. Also, use double precision LS-DYNA.

[Qusai Alomari]

Thank you Loic and Day for your suggestions. The model worked after removing the ALE coupling keywords). So, what could the problem be? As indicated in the figures below, the model consists of two ALE coupling ID's. The first (COUP 700) is used to couple the fluid (i.e., air+explosive) with the structural system. The second (COUP 500) is used to couple the soil with the foundation system.

[Loic Ivaldi]

Hello,

Check if the filling of the AMMGs is correct.
Enable the coupling card one by one to check the behavior.

Also, on your coupling cards displayed, normally the master side is the PART representing your ALE domain and then the coupling between AMMG and the structure is handled by MCOUP.
I saw 2 different "master" so maybe this is the issue.

Best Regards,

Loïc

[Qusai Alomari]

Thanks, Loic.

ALE parts are masters in the displayed cards. I'm using two masters as the first one (2000) represents the Soil, and the second one (801) represents the air and the explosive. I have also checked the filling AMMG and it looks correct!

[Loic Ivaldi]

just in case regarding what Jim said : https://www.dynasupport.com/howtos/general/consistent-units

[Jim Day]

Check that system of units is consistent. Are you using explosive constants directly from some reference and are you converting those constants from one unit system to another? Check that conversion.

[Jim Day]

Do you get any messages concerning deletion of Lagrangian elements? What failure/erosion criteria are given in your input that may be triggering the element deletion? Wrtie output every time step so you can hopefully see how the erosion evolves.

[Qusai Alomari]

Thanks, Jim. 

The problem has not been resolved yet! I have checked all the units and they are consistent. Regarding the erosion algorithm, for concrete, I'm using CSC with ERODE = 1.1. Steel is modeled using Piecewise Linear Plasticity and plastic strain at failure has been defined.

The messages I received in most of the previous attempts were negative volume warnings (no error messages) and in some other cases, I got out-of-range moments and NaN!

[Loic Ivaldi]

When you have the time where the instability occurs, output a lot of d3plot just before that time to see what is happening (just be careful to not fill your HDD completely)

[Jim Day]

It's understood that you need to apply pressure from the explosion to the structure. Since it's not clear what's making the solution unstable, I would recommend simplifying the model just temporarily while debugging. Eliminate the ALE portion of the model and just apply a pressure time history to the structure using *LOAD_SEGMENT_SET which very roughly approximates an explosive pressure. Is that simplified model stable?

[Qusai Alomari]

Hello Jim, I got an Error termination with the same element deletion, however, it works for 63 time steps for the first time, and this is just significant progress. I found that the ALE coupling was the source of error. So, I tweaked some parameters such as the number of coupling points. 

I still need to figure out the issue that causes the termination right after the explosive charge is detonated.