I have tried running tutorial 3.11. Hydrogen oxidation in a batch reactor using CHEMKIN mechanism — MFiX 25.2 documentation but every time I try, an error shows up:
Error from read_database.f:590
Error: Solver crash!
The MFiX solver has terminated unexpectedly
Error information:
float overflow in __read_database_mod_MOD_check_cp_vs_t
at read_database.f:590
I successfully ran the “ready-to-run” case on the latest version of MFiX. Did you compare against this one? By the look of the error message, it seems to be something wrong with the checmkin import.
The tutorials run fine at my end with MFiX 25.2.1. I could not reproduce your issue. Please provide more information or the case files you are running. Thanks.
I’m running the standard tutorial as it is comes preinstalled. I have attached the project files. Unfortunately, I’m using Windows 11 Business version 23H2 since my boss doesn’t want me to install Linux.
tutorial_11_hydrogen_oxidation.zip (76.7 KB)
The issue might be related to another issue that I’m facing quite often in some cases. In some cases I get an error message that the file couldn’t be saved. Then if I just try again, it does save the file. Especially in this tutorial it happens every time I try. I have attached another print screen:
MFiX: 25.2.1
Conda: Unavailable
Python: 3.13.5 | packaged by conda-forge | (main, Jun 16 2025, 08:20:19) [MSC v.1943 64 bit (AMD64)]
GFortran: GNU Fortran (Rev5, Built by MSYS2 project) 5.3.0
OpenMPI: Unavailable
Qt: 5.15.15
Qtpy: 2.4.3
Numpy: 2.3.2
Nodeworks: Unavailable
VTK: 9.3.1
OpenGL backend: 2
Solver source: C:\Users\StephanSneijders\miniforge3\envs\mfix-25.2.1\share\mfix\src
System info: Windows-11-10.0.22631-SP0
Unfortunately, I’m using Windows 11 Business version 23H2 since my boss doesn’t want me to install Linux.
Can you get new boss? 
I don’t thikn the float overflow and the file save problem are related. We’ve seen some issues on Windows where a file is open and the OS prevents it from being rewritten because it is in use… not 100% sure how to reproduce or fix this, but I’ll take a look.
I’ll also see if I can reproduce the float overflow running this case on Windows 11.
You might try installing MFiX under WSL… I’m not sure if people have had success with this but it might work.
Thanks for your patience, I’ll see what I can find.
– Charles
I need to get good pre-liminary simulation results to convince my boss for buying a dedicated Linux workstation for simulation purposes. That’s actually why I’m still playing around with the code on my Windows laptop for now. 
I will try running the tutorial on my private Linux machine (after installing MFiX there). Actually, I had already put that on my to-do list.
Anyway, thanks for looking into these issues.
A few other users have reported this [WinError 5] Access is denied but I’ve never been able to reproduce it myself. After doing some research, I think it may be due to antivirus software. We write to a temporary file and then move that to the destination. If the temporary file is being scanned (as all newly written files are), this Access error will occur. I’ve never seen this because I have antivirus scanning disabled on my Windows laptop (don’t tell anyone!) because I’ve found it uses a lot of CPU and slows things down.
If this is the culprit, then an easy fix is to add a little delay/retry before the file move operation. If you’d like to help us test this, download the attached file, open a Miniforge prompt, do conda activate mfix-25.2.1; then navigate to %CONDA_PREFIX%\Lib\site-packages\mfixgui and replace the file gui.py with the file attached below. If this clears up the issue, I’ll add this fix to the next MFiX release (25.3, due out in a few weeks).
Thanks!
gui.py (154.1 KB)
Stephan, as @cgw mentions, and as you have the latest Win11 you can actually run WSL (Linux inside Windows) very easily. This is with a GUI, which didn’t work with WSL 1 before, as I’m aware. Of course, you will have to issue some commands in a terminal, but the internet will help you there. To start you off:
wsl --install --no-distribution
or drop --no-distribution if you’re happy with Ubuntu.
Install miniconda3, wget, git, openmpi or openmpi-devel if they’re not already shipped with the distro. I add thunar file browser when using Fedora distro. Leafpad is a text editor. Install the Linux HPC mfix package.
If you have a paranoid employer or IT department, you might have to allow “man-in-the-middle” certificates to allow them to snoop on your traffic [as they most likely already do]. This can be done inside the certifi package inside the conda installation, look for the .pem file. Not perfect, but works for me.
Launch graphic instances with an & at the end, to continue to have terminal usable. Enable mpi by module load mpi to use all of you cpu’s, then your mfix environment, and mfix & or mfix -n & at the end.
Enjoy!
Regarding the floating-point overflow issue, this is caused by a precision mismatch between Linux and Windows, which leads to incorrect reading of atomic molecular weights on Windows.
Thank you for reporting this. The problem will be fixed in the next release. In the meantime, please copy the attached files into your working directory and rebuild the solver to apply the correction.
read_thermochemical_mod.f (34.3 KB)
atomic.inc (7.2 KB)
I have replaced the gui.py file and didn’t encounter the [WinError 5] Access is denied error anymore. The issue seems to be solved with this patch. In case the error unexpectedly shows up again, I will let you know.
I have rebuild the solver with your files and now the simulation runs perfectly fine. Thanks for that. Unfortunately, I couldn’t mark your answer as the solution since I’ve already marked @cgw 's answer as the solution. Both answers address different issues but at least I wanted to let you know that your solution works!
When viewing the monitor results of the simulation, I actually found out that there is a difference between my results and the screenshots in 3.11. Hydrogen oxidation in a batch reactor using CHEMKIN mechanism — MFiX 25.2 documentation, e.g. in my case H2 and O2 concentrations went negative while in the tutorial screenshots they didn’t. I have noticed that this issue of wrong axis alignment also happens in other tutorials so it is not tutorial specific. Please let me know if you want me to post this in a separate thread.
It’s a good idea to create a new issue once a topic has been marked “solved”. Otherwise it probably will not get any attention.
