Solution Failure: An input/output error has occurred on the process with MPI Rank ID = 0


I have been having some problems with solving a wheel and hub assembly. Initially I thought it was an issue with having not enough RAM, I only have 8GB available unfortunately.

I reduced the mesh from 2,000,000 Nodes and 800,000 elements to 540,000 nodes and 342,000 elements.

After drastically reducing the mesh and no longer seeing and errors. The solution would fail due to this error:



 **********  CURRENT DISK AND DIRECTORY USAGE  **********


 Volume in drive C is OS

 Volume Serial Number is 40D3-6139


 Directory of C:\Users\andre\Documents\Titomic\Test 1 - 12.05.20\_ProjectScratch\Scr5C95


13/05/2020  03:47 PM    <DIR>          .

13/05/2020  03:47 PM    <DIR>          ..

13/05/2020  11:19 AM                 0 ANSYSFatalError.log

13/05/2020  11:19 AM                 0 ANSYSTraceBack.log

13/05/2020  03:43 PM            83,320 CAERep.xml

13/05/2020  03:43 PM       107,657,511 ds.dat

13/05/2020  03:49 PM                26 ds_file.stat

13/05/2020  03:44 PM               115 ds_file.xml

13/05/2020  03:43 PM                70 dummy.dat

13/05/2020  03:44 PM                78 file.bat

13/05/2020  03:44 PM             7,345 file.cnd

13/05/2020  03:45 PM                 0 file.DSPmatK

13/05/2020  03:45 PM                 0 file.DSPstack

13/05/2020  03:45 PM                 0 file.DSPsymb

13/05/2020  03:45 PM    10,327,709,184 file.DSPtri

13/05/2020  03:45 PM                 0 file.DSPtriU

13/05/2020  03:44 PM                 0 file.emat

13/05/2020  03:45 PM             2,729 file.err

13/05/2020  03:44 PM                 0 file.esav

13/05/2020  03:45 PM       915,275,776 file.full

13/05/2020  03:44 PM           814,970 file.ldhi

13/05/2020  03:44 PM                 0 file.lock

13/05/2020  03:44 PM                10 file.log

13/05/2020  03:44 PM                 2 file.mntr

13/05/2020  03:44 PM                 0 file.osav

13/05/2020  03:44 PM                 0

13/05/2020  03:44 PM       289,210,368 file.rdb

13/05/2020  03:44 PM                 0 file.rst

13/05/2020  03:49 PM               160 file.stat

13/05/2020  03:43 PM            37,130 MatML.xml

13/05/2020  03:43 PM               632 SecInput.txt

13/05/2020  03:49 PM            47,646 solve.out

              30 File(s) 11,640,847,072 bytes

               2 Dir(s)               0 bytes free


 *** ERROR ***                           CP =     500.719   TIME= 15:50:20

 An input/output error has occurred on the process with MPI Rank ID = 0. 

  Please check to see if the disk is full.                               










I'm not sure what the issue is because I have 20GB available on my SSD.





  • AmbarNaik13AmbarNaik13 Dresden, GermanyMember

    Even I am facing the same issue.

    @peteroznewman could you please shed some light on this issue?

  • peteroznewmanpeteroznewman Member
    edited September 27


    The solver ran out of disk space.

    One of the files was 10 GB and it probably needed another 10 GB but it couldn't get that so it terminated with that error.

    FEA jobs, especially transient models, need a lot of free disk space (and RAM). It is best to install a separate SDD of at least 512 GB and only save the current analysis job on that.

    There are ways to reduce the disk usage, such as turning off Output that you don't need, such as Strain. That can be done under Analysis Settings.

  • mrifemrife PHLForum Coordinator

    @AmbarNaik13 & @ADTTT please see the following Help section: Help -> Mechanical APDL -> Performance Guide -> Chapter 4 Memory Usage and Performance.

    There are three sections on the different solvers (sparse, pcg, modal) that discuss the memory and drive space needed in terms of total degrees of freedom. And be sure to read the notes section below the table as it gives scale factors based on element formulation and model type. For example the rules-of-thumb are given for low order solid elements; then scale factors given for high-order solids. Or beam/shell elements.

    These rules assume sparse matrices; some FEM techniques will result in dense regions in the matrices (dense as in the terms are non-zero). This can greatly increase the needed amount of memory or drive space. A usual suspect is a remote point scoped to a region of many nodes.

    I think you will find that 8 GB of memory and 20 GB of disk space is just very, very small amounts. Especially for todays auto-mesh algorithms that can create high node count meshes quickly and efficiently. You may be able to solve the model by implementing some mesh size controls to reduce the node count further.


Sign In or Register to comment.