Discrepancies between fracture-matrix analytical and Slab problem

[Pau Márquez Martín]

Hello!

We are trying to simulate a tracer transport in a dual porosity media with a fracture-matrix process using a Dual Continuum Formulation. 

The conceptual model of the Slab geometry used for the secondary continuum
domain is shown next:

We have obtained satisfactory results with the Tang and Sudicky analytical solutions, as well as with the 2D fracture-matrix system model explicitly discretised. However, when using the DCDM model (both with Slab and Nested Cubes) we get the following curve. 

It is interesting to note that for higher groundwater velocities (i.e. 3m/yr) the PFLOTRAN curve matches the analytical. For this reason, and after many tests, we were wondering whether we are working with the right parameterization and why we are obtaining such results.  

 

The parameters are as follow:

 

Half-aperture    3.8E-4 m
Matrix porosity 6.3E-3
Matrix tortuosity            6.39E-3
Longitudinal dispersivity           0
Free water diffusion coefficient             1E-9 m2/s
Groundwater velocity   3E-2 m/y
Fracture Length             1.4 m
Matrix Length   1 m

 

Attached please find the input file for the slab configuration, as well as the needed .h5 files

Thank you!

[Leone, Rosie]

It appears your .h5 files are corrupted, can you send them another way?

Thanks,

Rosie

---

From: pflotra...@googlegroups.com <pflotra...@googlegroups.com> on behalf of Pau Márquez Martín <pmarq...@gmail.com>
Sent: Tuesday, May 30, 2023 10:35 AM
To: pflotran-users <pflotra...@googlegroups.com>
Subject: [EXTERNAL] [pflotran-users: 7079] Discrepancies between fracture-matrix analytical and Slab problem

 

Some people who received this message don't often get email from pmarq...@gmail.com. Learn why this is important

--
You received this message because you are subscribed to the Google Groups "pflotran-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pflotran-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/a5712ab1-6367-45e1-ada7-36a828d6ea4dn%40googlegroups.com.

[Pau Márquez Martín]

Thanks Rosie for the reply,

Let's try with a .zip

El dia dimarts, 30 de maig de 2023 a les 23:57:34 UTC+2, Leone, Rosie va escriure:

[A Palacios]

Hello, 

We wanted to inform that there was an issue with the input file it was sent a few days ago.
The velocity should have been multiplied by the parameter EPSILON (Fracture volume fraction). This is needed for the good parametrization of the DCDM models, as Darcy flow must be defined by square meter of rock and not square meter of fracture. 

With the correct definition of Darcy flow, we obtain DCDM models with diffusion, and we fit the analytical solutions when Darcy flow is high (unfortunately we still don't find parameters to fit models with very low velocities). 

Plus, in case it is helpful for anybody else working with the Multiple Continuum capability: 
KEYWORD TORTUOSITY does not raise an error but is ignored by the code. The way to define pore diffusivity is through the KEYWORK LIQUID_DIFFUSION_COEFFICIENT = Water diffusion coefficient * Matrix tortuosity. 

Best regards,
Andrea

[paolo trinchero]

Andrea, Pau,

The thing here is that, which such low velocity, transport in the fracture is diffusion controlled. In your PFLOTRAN model you certainly have some numerical dispersion and this most probably explains the earlier arrival. I think that this is an ill-posed benchmark since with such low Peclet numbers the 1D diffusion assumption on which both the DC model and the analytical solution rely breaks down.

Best regards

Paolo