LS-DYNA Binout Elout with Erosion

[ricepatti]

🐛 Describe the bug
When reading binout data using binout.py, a bug is encountered when the user is requesting ELOUT data with eroding elements. Typically, data is stored in a rectangular list, which is then converted using np.array to be returned to the user. However, when the ELOUT data contains eroded elements, the data is stored in a list of tuples that are not all the same length. This results in a "inhomogeneous shape" error when calling np.array(data) in binout.py.

🔢 To Reproduce
Steps to reproduce the behavior:

1. I ran DYNA version R15 with Elout *DATABASE_HISTORY_BEAM_SET requested for both eroding and non eroding elements. The eroding elements are deleted during the analysis.
2. Request binout.read(xxx,elout,beam,axial)
3. Error occurs.

💘 Expected behavior
Elout data is returned as an array.

🖥️ Setup

• lasso-python latest version as of 7/18/24
• numpy version 2, latest release as of 7/18/24

ℹ️ Additional context
I feel like this was not a problem in the past, so maybe it's the numpy version or DYNA R15 changed how it stores the binout data. I was able to implement a workaround in binout.py by making a function which checks if the variable "data" is rectangular before calling np.array(data), and if it is not, padding it with zeros. I am not great with python so there is probably a better way to do this.

I am also sorry for the lack of details, this bug occurs on a computer with LS-DYNA that I cannot upload to Github from. If needed I can probably replicate the functions.

[pytunia]

I ran into the same problem and added a small piece of code to handle it.
In the file binout.py insert the following code between line 338 and 340. Make sure to keep the indent aligned (8 spaces):

        # handle eroded elements
        try:
            new_data = np.zeros((len(data), len(data[0])))
            for j in range(len(data)):
                if type(data[j]) == int:
                    new_data[j][0] = data[j]
                else:
                    new_data[j][0:len(data[j])] = data[j]
            new_data.astype(type(data[0][0])) # make sure type does not change
            data = new_data
        except:
            pass

To find the location of your binout.py you can use:

import lasso
print(lasso.__file__)

[codie3611]

Hmmm sounds like we need a patch for that 👀