Nodes discretization for infiltration in mining dumps

A discussion forum for Hydrus-1D users.
Post Reply
Tiaren García-Perez
Posts: 2
Joined: Wed Mar 31, 2021 1:58 pm
Location: Chile

Nodes discretization for infiltration in mining dumps

Post by Tiaren García-Perez » Wed Mar 31, 2021 2:26 pm

Hello,

I've been using Hydrus 1D to estimate infiltration in dumps, with long columns (40 meters) for 500 years. My current problem is that using different nodes (800, 900, 1000) the results in terms of the infiltration are different. Could you help me, please?

Cheers,

Jirka
Posts: 5216
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: Nodes discretization for infiltration in mining dumps

Post by Jirka » Wed Mar 31, 2021 6:24 pm

There is rarely the need to have so many nodes. It is important to have finer discretization close to the surface where the larger gradients are. Check the mass balance errors.

Also, read the rules about spatial and temporal discretization (FAQ 8):
http://www.pc-progress.com/Documents/No ... zation.pdf

J.

Tiaren García-Perez
Posts: 2
Joined: Wed Mar 31, 2021 1:58 pm
Location: Chile

Re: Nodes discretization for infiltration in mining dumps

Post by Tiaren García-Perez » Fri Apr 09, 2021 3:17 pm

Dear Jirek,

Thank you very much for your answer! yes, I refined the mesh in the surface, but the results changed depending on the number of nodes.
In this case, I'm using a suction curve representing a coarse material, I guess that is the problem together with the large size of the column.

Does the 2D software have a limitation in the number of nodes?

We are thinking in getting a license.

Cheers!

Tiaren

Jirka
Posts: 5216
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: Nodes discretization for infiltration in mining dumps

Post by Jirka » Fri Apr 09, 2021 4:38 pm

The memory in HYDRUS (2D/3D) is allocated dynamically depending on the need. I.e., there is a limit only with respect to the memory of the computer and memory addressing. I have run simulations with over one million nodes (which are admittedly quite slow). Typical problems would be below 100k nodes, and that is quite doable. J.

Post Reply