Time step

A discussion forum for users of the new HYDRUS 2D/3D. HYDRUS is a software package for simulating water, heat and solute movement in two- and three-dimensional variably saturated media. Happy Posting!
Post Reply
lpom
Posts: 7
Joined: Wed Mar 14, 2018 10:27 am
Location: United Kingdom

Time step

Post by lpom » Mon Jun 11, 2018 1:21 pm

Dear all,

I am still learning how to use HYDRUS and I have some doubts about how the time is discretised. More specifically about how does the T-level information option affects the discretisation. From the manual I understood that the numerical time step, the time step indicated in the BC, the time step at the print-out, and the measured data for the calibration of the model. However, if I don't include an interval output and I repeat the same simulation, the printed times between both simulations vary, and so they do my results. In other words, Why do I get different printed time steps and results in the same simulation if I select the t-level information option and I don't include an interval output?

Any comment on advice on how to make the temporal discretisation would be appreciated,

Thank you,

LP

Jirka
Posts: 4651
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: Time step

Post by Jirka » Mon Jun 11, 2018 6:22 pm

In general, when you select to print the results (e.g., each time step, regular intervals) should not affect the results of the simulations, unless you choose something that dramatically (and unreasonably) alters the temporal discretization of the numerical solution.

The description and recommended rules about temporal discretization (FAQ 8 ) can be found in this text:
http://www.pc-progress.com/Documents/No ... zation.pdf

J.

lpom
Posts: 7
Joined: Wed Mar 14, 2018 10:27 am
Location: United Kingdom

Re: Time step

Post by lpom » Tue Jun 12, 2018 9:35 am

Thank you for your help Jirka,

LP

lpom
Posts: 7
Joined: Wed Mar 14, 2018 10:27 am
Location: United Kingdom

Re: Time step

Post by lpom » Thu Jun 21, 2018 9:35 am

Professor Jirka,

I have change the time settings as suggested but the cumQ results still change depending on the print options, more specifically if the t-level information is selected.

Would you mind to have a look at the case? I am not sure what can be the mistake

Thanks,

LP


https://www.dropbox.com/s/tsr6utqysu98x ... m.zip?dl=0

Jirka
Posts: 4651
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: Time step

Post by Jirka » Thu Jun 21, 2018 5:08 pm

Your setup is very unreasonable and you place a lot of constraints on the time-stepping scheme in HYDRUS. Why would you specify boundary conditions at each 0.01 h (i.e., 200 values) when you have only 2 BCs (precipitation until 1.18 h, and no precipitation after that). Similarly, you fix the time step at 0.01 h, printing at 0.01 h, etc. HYRUS runs best when you allow it to chose its own time stepping scheme (which is optimized at each time step), rather than completely constrain it. If you want to constrain time steps (which is unreasonable in the first place) you would likely need to use much stricter iteration criteria. J.

Post Reply