Dear all,
I found Heister’s model and change it with ASPECT-version2.4.0.
Here is heister’s paper High accuracy mantle convection simulation through modern numerical methods – II: realistic models and problems | Geophysical Journal International | Oxford Academic, and it’s input file paper-aspect-methods-2-data/application/global_convection_3d.prm at master · tjhei/paper-aspect-methods-2-data · GitHub
I found it’s hard to slove after 30+Myrs, the time step become about 3000 yrs after 30+Myrs, it seems that this model wouldn’t end after 37600 CPU hours which is described in his paper(even won’t be end after 3700 time steps, because I run it about 1800 time steps only 39Myrs). I even used the “project q1 only viscosity” compared to his model, it should be more faster. it seems hard to solve after the plume were generated.
The addition files are my input file and logfile.
heister3d.prm (6.3 KB)
log.txt (2.8 MB)
Dear Prof. Heister @tjhei
I also found your model videos on YouTube: https://youtu.be/LRs1Qdm-FSI
(Videos and Interactive Visualization Tests | Integrated Geodynamic Earth Models)
It’s so beautiful, do you have it’s input files for reference? I really want to learn ASPECT from your past .prm files. It will be very helpful for me.
I’ll also understand if you couldn’t share the .prm file with me. But I really want to know how to make the model converge more effectively, could you please give me some suggestion or informantion based on your model?(I always met the wrong cpu time even when I use the .prm file with your or Dr. Philip’s original input file). After all, It’s expensive for me to test these 3d global convection models.
Dear Shaohui,
You could use the prm files we use for a more recent global modeling study (Changes in core–mantle boundary heat flux patterns throughout the supercontinent cycle | Geophysical Journal International | Oxford Academic) with ASPECT from this repository: Changes in core-mantle boundary heat flux patterns throughout the supercontinent cycle: Data
These were run on a modified version of ASPECT that’s in the zenodo repo as well, and I do not recall if the same file now also works with the main version of ASPECT. But at the very least, this would help to see any changes we made in the input file to run these global model with more recent ASPECT versions.
For reference: This is how the model looks like: https://www.youtube.com/watch?v=0ZxLYyc_7tM
Best,
Juliane