strange concentration and mole_fraction with ats-master using multiple processors
Yi Xiao
Coon, Ethan
Hi Yi,
First of all, thanks – this was a great example of creating a simple problem that I can easily reproduce and test, which makes it much easier for me to help get things fixed.
Second, pull your current master, and the parallel runs should be fixed now. Thanks for reporting the issue!
Ethan
From:
ats-...@googlegroups.com <ats-...@googlegroups.com> on behalf of Yi Xiao <yixia...@gmail.com>
Date: Monday, June 2, 2025 at 6:56 PM
To: Amanzi-ATS Users <ats-...@googlegroups.com>
Subject: [EXTERNAL] strange concentration and mole_fraction with ats-master using multiple processors
Hi All, I'm testing the ats-master with a hillslope ATS-PFLOTRAN model, initially driven by the same issue related to loading restart concentration from checkpoint file (https: //groups. google. com/g/ats-users/c/_8hSHpqNFJQ) Specifically, I'm
Hi All,
I'm testing the ats-master with a hillslope ATS-PFLOTRAN model, initially driven by the same issue related to loading restart concentration from checkpoint file (https://groups.google.com/g/ats-users/c/_8hSHpqNFJQ)
Specifically, I'm seeing unexpected values of total_component_concentration and mole_fraction when running with 8 processors, whereas the results seem reasonable with 1 processor.
Here's my case setup:
First, I perform a 10y flow only spin-up, using constant precipitation and water head at left/right boundaries. After this, I add a tracer to the precipitation, and set initial concentration to be the same.
The 1 processor results look reasonable
- the initial mole_fraction equals 1e-3, and the initial total_component_concentration equals 0.055 as expected.
- after 10 days, total_component_concentration near the left boundary decreases due to water flux through
The 8 processor results are strange
- at t=0, the mole_fraction is 0 everywhere, and the initial total_component_concentration is 0.055.
- as time evolves, total_component_concentration remains 0.055 everywhere, and mole_fraction values become very high in top layers.
I've uploaded my case to the google drive. You'll find:
- folders paralleltest-n1-case1tracer and paralleltest-n8-case1tracer contain the ATS transport models
- folders paralleltest-n1-caseflow-run0-spinup_steadystate and paralleltest-n8-caseflow-run0-spinup_steadystate contain the checkpoint files from flow only spin-up simulations.
- the exo file is in folder data-processed.
I don't have this issue with ats-1.5.2, the simulated dynamic of total_component_concentration looks reasonable with 8 processors.
Thanks for your help!
Best,
Yi
--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
ats-users+...@googlegroups.com.
To view this discussion visit
https://groups.google.com/d/msgid/ats-users/0fa36b1b-3be7-404e-838a-16b91714d45cn%40googlegroups.com.