Particle Error

Andrew Greeley

I have just encountered a new-to-me error when attempting to modify physics parameters for a EEPA model and need help getting back up and running. A screenshot of the error message telling me there is an issue with the particle size distribution and screenshot of the particle size distribution as well as the simulation file are attached for reference.

A couple notes:

-This is the first time I have encountered this message, although I have modified the size distribution and physics parameters before.

-This may be the first time I have tried to change the physics parameters on 2021.1, but I'm not sure.

-The application completely exits while loading simulations fairly often. It always loads the simulation after restarting.

-The custom factory points to the "OutputData_Run6.txt" file with an absolute path listed in the "options.txt" file. Either change the path or delete the custom factory in order to run the simulation.

Stephen Cole

Hi Andrew,

Can you also include the setup1_data folder, or specifically the 0.h5 file contain within this folder.  The 0.h5 file contains all the simulation settings/Creator data that we would need to look into this.

The screenshot shows that the total %mass adds up to 100 but could you verify this yourself?  It could be that the GUI is showing 100% but the values only add up to 99.9999 % or similar which could possibly be causing it to return an error.  It needs to be exactly 100%.

 

Regards

Stephen

Andrew Greeley

Hi Stephen,

Thanks for the reply. I have attached the simulation files with the 0.h5 file in the appropriate folder.

I am confident that the mass distribution adds up to 100%, I've been using it in factories for some time now.

Best,

Andrew

Stephen Cole

Andrew Greeley said:

Hi Stephen,

Thanks for the reply. I have attached the simulation files with the 0.h5 file in the appropriate folder.

I am confident that the mass distribution adds up to 100%, I've been using it in factories for some time now.

Best,

Andrew

I received the files and can see the error.  If I create a slightly simplified version of the size distribution it works OK (using less decimal places and a few edits to still ensure it adds to 100%), however i'm unsure why the distribution set is producing an error.

I'll report it to the development team for review.

 

Regards

Stephen

RWood

Stephen is right. Both your distributions add up to 100.00000000000001 %, which, no matter how small, is interpreted as being over 100%. You can see this if you look in the 0.h5 file under the size distributions. 

If I change the cumulative % to be 100.0 manually in the h5, I can add the EEPA parameter data without that error. 

I'm guessing you've copy pasted the values in from Excel (or similar?) and they got rounded along the way. I'd say you're better off typing in the values in manually so that they won't suffer from any rounding issues when copying from elsewhere.

Richard

Andrew Greeley

Hi Richard and Stephen,

Thanks for the insight and help on this. I have confirmed as you did that somehow the distribution does not add up perfectly in the 0.h5 file. However, this is very strange because I have run many simulations including factories with these size distributions and many different EEPA parameter changes before.

To your point Richard, manually reentering the set of numbers does not fix the problem (and I never imported them to begin with). This indicates to me that there is a bug in EDEM, but I am not a developer and I don't need to know the fine details. I do need to get simulations running though.

I will proceed by manually editing the 0.h5 file as necessary, and please let me know if there is a better fix.

Best,

Andrew

 

Andrew Greeley

Richard, 

Could you point me to some information about editing a .h5 file? I have been using HDFView to examine the files but don't see the ability to edit the files.

Best,

Andrew

RWood

You can edit within HDFView by double clicking values and then saving. The danger of doing this though is that you might be changing a particular format, without knowing it. For those size distribution values, I think you're safe, but you might accidentally put a double or float in a place where an integer is expected and corrupt the deck.

For editing the occasional value HDFView is fine. If you're going to be doing more systematic changes then I'd recommend the Python library h5py. There is a MATLAB h5 module as well, though I'm not sure on the name of it and I've not used that one. The Python and MATLAB approaches allow you to check the data types you're working with and know that you aren't going to corrupt a deck as well.

Richard