Error: Time stepper cannot reduce the time step size further

Usability
1

Dear all, Recently I have built a prj to run ogs, which is about the simulation of THM process in hot dry rock where lies a fracture. when I run the project some errors occur like the following picture. It seems that the max iteration has arrived. Maybe there is something wrong with my numerical method. I just use some default parameters of numerical part in my prj. So what is the problem? Can anyone help me? I attach my prj file here.THM_ogs6.zip (1.3 MB)

微信图片_20210329215323
微信图片_202103292153231229×638 50.9 KB

@Thomas_Nagel @joergbuchwald I saw a similar post in the discourse about the same error that you make a discussion about.the similar topic. I hope you could give me some advice. Thank you !

2

Hi,
I think it is not directly related to the numerical method itself and also not related to the thread you mentioned, but more to the formulation of the problem. However, it’s quite hard to say something without intensive testing what is actually causing the divergence.
It could be related to the time stepping, not sure. Please try to simplify the problem, e.g. use a simpler mesh, test it without source terms or apply them after some time steps… Make it simple until it works and try to make it iteratively complex again. There are some tricks on might use, like printing out every non-linear step to find divergences (I think it requires at least the linear solver to converge), using a direct solver for a small mesh or iteration-based time-stepping.

3

Hi,
I’ve followed your advice to simplify the my model. First I reduce my geometry scale to 100X100m resulting in a simple mesh，and I remove the source term, the project works well. Then I add the source term with a smaller pressure e.g.1e-3, the file also works. As for the numerical method, I enlarge the tolerance to let the calculation go through. Next I gotta adjust the parameters to make it complex and get some more reasonable results. It’s sure there are much for me such a beginner of ogs6 to learn. Wish me good luck.
By the way, you mentioned some tricks such as apply source term later, print out steps… I m not very clear about them. Can you provide me with some useful links or manuals to learn the coding method for an ogs6 project?
Thank you!`

`

4

Concerning printing out every non-linear step, I think it is this parameter:

For switching the the source term later on, you can use this curved-scaled parameter

The usage can be taken from the example files there. You can provide basically time-steps with a corresponding pre-factor for any given parameter.
The doxygen contains basically all input parameters. However, often, a documentation is missing.
But, as it is the intention of the developers to cover the entire code with tests you should find for most things test files in the tests/Data folder.

5

I see. Thank you very much!

6

Hi, Dr. Jörg Buchwald
I want ask whether the CurveScaled can be used to define the properties of media? Or is the CurveScaled only used in the parameter part?
If the CurveScaled can be used to define the properties of media, could you link some example here?
Thank you very much!

7

Hi Rui Feng,
you can define parameters as media properties.
E.g.,

<property>
    <name>decay_rate</name>
    <type>Parameter</type>
    <parameter_name>decay</parameter_name>
</property>
8

Got it! Thank you! :blush: