HYDRUS 1D Model Fails to Converge

A discussion forum for Hydrus-1D users.
sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

HYDRUS 1D Model Fails to Converge

Post by sar29 » Thu Jul 28, 2016 10:36 pm

Hi all,

I am attempting to simulate groundwater recharge to a shallow water table (initial water table depth = 75 cm below ground surface) in a 300 cm deep sandy loam soil profile. I have my simulation set to run for 1 year (8760 hours) with my upper boundary condition set as "Atmospheric BC with Surface Run Off" and my lower boundary condition set as "Variable Pressure Head". My profile is discretized such that FE nodes occur every 0.5 cm. As I do not know the initial pressure head, I set the pressure head at the ground surface to -0.001 (saturated conditions), 0 at the water table (75 cm), and ~225 at the lower boundary (300 cm). My goal is to compare observed versus simulated groundwater level fluctuations. However, after 3661 hours my model fails to converge.

I am wondering if anyone could help me see why my model will not converge. I have tried changing almost every parameters without success. Currently, I have my model set to no rootwater uptake, but have accounted for transpiration in my evaporation (evapotranspiration) value under variable boundary conditions. I have done this because I do not have enough information to estimate transpiration alone or the meteorological conditions, but am aware that it may not be an accurate representation of natural conditions.

Any suggestions would be greatly appreciated. I have posted the model file on my dropbox account (link below).

Thanks so much,
Sarah

https://www.dropbox.com/s/wujowgj7uvfxtmc/B1.h1d?dl=0

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Fri Jul 29, 2016 1:49 am


sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

Re: HYDRUS 1D Model Fails to Converge

Post by sar29 » Fri Jul 29, 2016 3:11 pm

Hi Jirka,

I apologize for that. If you would please follow the link below, all of the input and output files are now available.

Thanks again for your time,
Sarah

https://www.dropbox.com/sh/d21wf2qjxo04 ... R_7Pa?dl=0

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Fri Jul 29, 2016 9:26 pm

The program crashes (does not converge) on day 3655. There is a simple reason for that. Until day 3655, your bottom pressure head is less than 300 cm, i.e., the groundwater level is below the soil surface. On day 3655 it is 21.2 cm below the surface. Starting on day 3653, you give bottom pressure head of 995 cm, which means that groundwater table is 695 cm above the soil surface. That does not seem very reasonable to me. J.
VariabBC.png
VariabBC.png (44.41 KiB) Viewed 11431 times

sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

Re: HYDRUS 1D Model Fails to Converge

Post by sar29 » Thu Aug 04, 2016 7:51 pm

Hi Jirka,

Thanks very much for your help.

First, I fixed the issue with the unrealistic change in pressure head. (I changed the depth of my soil profile at one point and must not have updated all of the GWL values.) Now that the pressure head is consistent, my model is actually crashing earlier, at hour 686. I've noticed in the T_Level.out file that the value for htop makes a large jump to a positive value where the model crashes. Does this signify that the influx of water to my model is too large relative to the outflux? As I mentioned previously, I have currently lumped evapotranspiration into the "evaporation" boundary condition and not specified rootwater uptake as I believe that I do not have enough information to correctly specify this parameter.

Second, would you recommend possibly changing my lower boundary condition to deep drainage, although I have hourly groundwater level observations for my period of record?

Third, I'm wondering if you would recommend using vBot or the pressure head of a node placed at the bottom of my profile to determine the height of the water table throughout the simulation?

Last, my ultimate goal is to compare recharge for scenarios of varying water table depth and soil zone thickness in fractured bedrock. For this reason, I am wondering if it is feasible to create a model with both single and dual-porosity layers (i.e. soil zone with underlying fractured bedrock).

If you wouldn't mind taking a look at my model, which is still not converging, I would greatly appreciate it. The link can be found below. Like many users who post on this forum, I am new to HYDRUS and I greatly appreciate your time and patience.

Thanks in advance,
Sarah

https://www.dropbox.com/sh/70dlrwbrac3i ... bHi4a?dl=0

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Sat Aug 06, 2016 10:44 am

I’m on vacations in Europe and thus will not have time to look at your files until I get back to US on the Labor Day weekend. J.

sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

Re: HYDRUS 1D Model Fails to Converge

Post by sar29 » Mon Aug 08, 2016 12:15 am

Hi Jirka,

Thanks for your reply. If you won't have time until September, then don't worry about it. I have to finish this project by the end of August. I will try to figure out the issue or just use a different code.

Thanks,
Sarah

sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

Re: HYDRUS 1D Model Fails to Converge

Post by sar29 » Sun Sep 04, 2016 5:01 pm

Hi Jirka,

I ended up putting this project on hold for a few weeks. When you get back from your holiday and have a moment, I would be pleased if you would still take a look at my files and answer my questions from the previous post.

Thanks so much,
Sarah

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Wed Sep 07, 2016 1:46 am

You have 3 m deep soil profile and then after day 725 you specify the bottom BC with values larger than 3 m. That would obviously meant that the entire profile is saturated and that there is even some water standing at the soil surface, while additionally adding water by precipitation. HYDRUS is not intended for problems like that. J.

sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

Re: HYDRUS 1D Model Fails to Converge

Post by sar29 » Wed Sep 07, 2016 8:27 pm

Hi Jirka,

Ok thanks for the information. There were other periods in the time series where the water table surpassed ground surface and the model did not crash, so the precipitation rate or duration must be higher in this instance.

I was also wondering if it is possible to create a two-layer model with both single and dual-porosity layers (i.e. soil zone with underlying fractured bedrock).

Thanks,
Sarah

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Wed Sep 07, 2016 9:50 pm

Yes. You can set up dual-porosity model for all layers and then simply make the mass transfer equal to zero, to convert it (in any layer) to a single-porosity model. J.

sar29
Posts: 13
Joined: Wed Jul 27, 2016 7:22 pm
Location: Canada

Re: HYDRUS 1D Model Fails to Converge

Post by sar29 » Thu Sep 08, 2016 6:10 pm

Ok, thank you very much!

Here is my final question - would recommend using vBot or the pressure head of a node placed at the bottom of my profile to determine the height of the water table throughout the simulation?

Regards,
Sarah

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Fri Sep 09, 2016 12:26 pm

Certainly the pressure head. Flux tells you nothing about the position of the water table. J.

hhfang
Posts: 7
Joined: Thu Mar 14, 2019 1:51 pm
Location: Germany

Re: HYDRUS 1D Model Fails to Converge

Post by hhfang » Wed Mar 25, 2020 2:45 pm

Jirka wrote:
Wed Sep 07, 2016 9:50 pm
Yes. You can set up dual-porosity model for all layers and then simply make the mass transfer equal to zero, to convert it (in any layer) to a single-porosity model. J.
I have a question similar to this..
Will a dual-porosity model work like a single porosity model if I set w2 for the second soil type to be zero? Thank you very much in advance!

Julia
Attachments
SELECTOR-ref.txt
(1.77 KiB) Downloaded 2 times
SELECTOR.txt
(1.85 KiB) Downloaded 3 times

Jirka
Posts: 4861
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: HYDRUS 1D Model Fails to Converge

Post by Jirka » Wed Mar 25, 2020 4:24 pm

It very likely will. Why wouldn't you simply test it yourself? J.

Post Reply