Overflow-Error

Discussion of numerical issues which are not directly linked to specific applications
Post Reply
Moritz
Posts: 40
Joined: Fri Aug 06, 2021 11:13 am
anti_bot: 333

Overflow-Error

Post by Moritz » Wed Mar 09, 2022 5:30 pm

Hello everyone!


I am currently experimenting with very large values for seed densities. Naturally, this leads to very high numbers of grains during solidification. Unfortunately, this seems to cause problems, possibly with the hardware I am using.

The early stages of the simulations look like this:

Image

The width of the simulation domain is approximately 6 µm.

This is the displayed error in the .log file:

Image

The simulation stopps after 22 steps and before it should stop. Has anyone a suggestion on how to prevent this from happening?

As always, I am very thankful for all your replies.

Best regards,
Moritz

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Overflow-Error

Post by Bernd » Wed Mar 09, 2022 10:36 pm

Hi Moritz,

Difficult to say without more information. It could be either a memory overflow, e.g. because of too many grains, or a writing problem. Unfortunately, in your screen snapshot I cannot see the file name at the right side, where MICRESS obviously tried to write unsuccessfully: C:\Temp...

Bernd

Moritz
Posts: 40
Joined: Fri Aug 06, 2021 11:13 am
anti_bot: 333

Re: Overflow-Error

Post by Moritz » Thu Mar 10, 2022 12:51 pm

Hi Bernd,

this error occurs for different writing locations.

In the screenshot it is a file on the C:\ drive and when I tried it on another machine, it is a file on a network drive.

In both cases the TabGD.txt is affected.

Best regards,

Moritz

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Overflow-Error

Post by Bernd » Thu Mar 10, 2022 2:08 pm

Hi Moritz,

The .TabGD-output file can easily get very large, because it gives information about every grain at every output time step. So, the simplest solution would be to just deactivate the .TabGD-output (remove "tab_grain_data" in terse input mode, or replace it by "no_tab_grain_data" otherwise) .

If this removes the problem, then it is just fine. Otherwise, you should try to reduce the number of grains, e.g. by using "categorize" in phase data and nucleation input, so that grains with identical properties can be assigned to the same grain number. Please note that a high number of grains leads to quadratically increasing memory usage: depending on your computer you probably will get a memory overflow somewhere between 10000-20000 grains...

Bernd

Moritz
Posts: 40
Joined: Fri Aug 06, 2021 11:13 am
anti_bot: 333

Re: Overflow-Error

Post by Moritz » Tue Apr 12, 2022 6:56 pm

Hi Bernd,

thanks for the suggestion. Deactivating the TabGD output solved the problem.

Is it correct, that if I want an putput file for every specific grain, because I want to analyse the microstructure, that I can solve this problem by using a more capable computer?

Best regards

Moritz

ralph
Posts: 167
Joined: Wed Apr 27, 2011 4:42 pm
anti_bot: 333

Re: Overflow-Error

Post by ralph » Wed Apr 13, 2022 9:28 am

Hi Moritz,

I guess Bernd meant that reducing the number of grains by categorizing saves time and avoids this 'record overflow' in TabGD output because of less data.

A more capable computer (more main memory) allows you to simulate more grains. It will not solve this output problem.

Best,
Ralph

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Overflow-Error

Post by Bernd » Wed Apr 13, 2022 3:49 pm

Hi,

For the size of the .TabGD-file the number of grains and the number of output time-steps is relevant!

Bernd

Post Reply