hysteretic model

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
Lili Yao
Posts: 17
Joined: Fri Sep 29, 2017 6:31 pm
Location: USA

hysteretic model

Post by Lili Yao » Fri Apr 16, 2021 2:13 am

Hi,

I was running a 2D soil column to test the hysteresis effect of a sandy soil. The bottom boundary is a time-variable head boundary, the upper boundary is an atmospheric boundary, and other boundaries are no flow boundaries. If the Lenhard hysteretic model was used (Hysteresis in retention curve; initially drying curve), the hysteretic loop would be observed in the water content vs pressure relation in the unsaturated zone, as shown in Figure 1. The simulation result actually is quite similar with observation. However, if the Scott et al. hysteretic model was selected during modelling (Hysteresis in retention curve; initially drying curve), no hysteretic loop was observed. I am curious that why there is no hysteretic loop when I use the Scott et al. hysteretic model, any help would be greatly appreciated. Thank you very much.

Lili
Attachments
Model_LLY.zip
(160.49 KiB) Downloaded 40 times
Figure1.png
Figure1.png (84.28 KiB) Viewed 630 times

Jirka
Posts: 5268
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: hysteretic model

Post by Jirka » Fri Apr 16, 2021 4:16 pm

Lili,

You will need to use a smaller water content tolerance, since that is the value, which is used not only to determine if the numerical solution of the Richards equation converged, but also in the Kool and Parker model to determine if the reversal from drying to wetting and vice versa occurred (the Lenhard model uses a different criterion). See that in the 2Hyster example (which comes with H1D, and which uses the Kool and Parker model), we use the water content tolerance of 0.0001. Since you use 0.01, you can never have a reversal.
J.

Lili Yao
Posts: 17
Joined: Fri Sep 29, 2017 6:31 pm
Location: USA

Re: hysteretic model

Post by Lili Yao » Sat Apr 17, 2021 2:15 am

Thank you very much, Jirka. However, if the water content tolerance was decreased to 0.0001 (and I have increased the pressure head tolerance from 1cm to 2cm), the simulation had a convergence problem. If I increased the water content tolerance from 0.0001 to a relatively larger number, it may converge, but no hysteretic loop then. Do you have any suggestion to solve this dilemma?

Jirka
Posts: 5268
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: hysteretic model

Post by Jirka » Sat Apr 17, 2021 4:15 am

You should use the Lenhard model since it is a better model that we implemented to replace the Kool & Parker model, which suffers from multiple problems (we kept it in the code as a legacy option). If you want to use the Kool & Parker model, you need to figure out the combination of parameters that will be stable yourself. J.

Lili Yao
Posts: 17
Joined: Fri Sep 29, 2017 6:31 pm
Location: USA

Re: hysteretic model

Post by Lili Yao » Sun Apr 18, 2021 1:26 am

Thanks, Dr. Simunek. I would like to have hysteresis in both retention curve and conductivity, but no option in the Soil Hydraulic Model dialog window that allows me to select the Lenhard et al. model and considering the hysteresis in conductivity at the same time. So, to solve this problem, I selected the “Hysteresis in retention curve (no pupping, Bob Lenhard)”, and then in the Water Flow Parameters dialog window I assigned a value to KsW which is different from the Ks value. Is that the correct way to consider hysteresis in conductivity when using the Lenhard et al model for hysteresis in retention curve? (Since my simulation results did not show any different after I changed the KsW value, so I was not sure if that was because the hysteresis in conductivity did have any effect on my result or maybe just because what I did was wrong). Thank you!

Jirka
Posts: 5268
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: hysteretic model

Post by Jirka » Sun Apr 18, 2021 7:48 pm

The Lenhard model considers hysteresis only in the retention curve, h(theta). Obviously, as a result, there should be hysteresis in the K(h) function, but not in the k(theta) function. J.

Lili Yao
Posts: 17
Joined: Fri Sep 29, 2017 6:31 pm
Location: USA

Re: hysteretic model

Post by Lili Yao » Mon Apr 19, 2021 10:16 pm

Thank you very much for your explanation!

Post Reply