Diffusive transport under grid rotation

21 views
Skip to first unread message

Jens Eckel

unread,
Nov 21, 2023, 7:14:16 AM11/21/23
to pflotran-dev
Hi all,

recently, I included anisotropic tortuosities in PFLOTRAN and at the moment I try to verify my changes.

In doing so, I observe a strange behavior on the master branch (dev version, as of 25/09/2023, without my changes). The system is a 1d column in x-direction with transport by diffusion only (input attached Fall_1a_x_direction.in). If the underlying grid is rotated to have the transport in y- and z-direction respectively, the transport coefficients in transport.F90  differ from the transport coefficients for the x-direction (Fall_1a_y_direction.in, Fall_1a_z_direction.in).

Putting some output at the end of the nphase slope in TDispersion

write(*,*)'dist_dn: ',dist_dn,'eps: ',epsilon_dn,'por: ',material_auxvar_dn%porosity
write(*,*)'tort: ',material_auxvar_dn%tortuosity,'diffusion: ',molecular_diffusion_dn(:)
write(*,*)'mechanical_disp: ',mechanical_dispersion_dn
write(*,*)'harm_transp_coeff: ', harmonic_tran_coefs_over_dist(:,iphase)

and correspondingly in TDispersionBC

write(*,*)'dist_dn: ',dist_dn(0),'eps: ',epsilon_dn,'por: ',material_auxvar_dn%porosity
write(*,*)'tort: ',material_auxvar_dn%tortuosity,'diffusion: ',molecular_diffusion(:)
write(*,*)'mechanical_disp: ',mechanical_dispersion
write(*,*)'transp_coeff: ', tran_coefs_over_dist(:,iphase)

shows that the transport coefficients for the boundary condition between x- and y/z-direction differ by a factor of ten, due to the difference between dist_dn(0) in TDispersionBC and dist_dn/dist_up in TDispersion for the y/z-direction. I cannot find a reason for this with the debugger and any help would be appreciated.

Best, Jens
Fall_1a_z_direction.in
Fall_1a_y_direction.in
Fall_1a_x_direction.in

Hammond, Glenn E

unread,
Nov 21, 2023, 3:29:37 PM11/21/23
to pflotr...@googlegroups.com

Jens,

 

When you change the orientation of the column from X to Y or Z, you must also change the face on which the boundary condition is applied. You are currently using the EAST and WEST faces for X, Y and Z.

 

X: WEST – EAST

Y: SOUTH – NORTH

Z: BOTTOM – TOP

 

When I make those modifications to the input deck, the simulation results match regardless of orientation.

 

Glenn

 

From: pflotr...@googlegroups.com <pflotr...@googlegroups.com> On Behalf Of Jens Eckel
Sent: Tuesday, November 21, 2023 4:14 AM
To: pflotran-dev <pflotr...@googlegroups.com>
Subject: [pflotran-dev: 6217] Diffusive transport under grid rotation

 

Check twice before you click! This email originated from outside PNNL.

 

--
You received this message because you are subscribed to the Google Groups "pflotran-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pflotran-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-dev/71442b3f-fcf4-4e58-a4e1-7be9b846b004n%40googlegroups.com.

Jens Eckel

unread,
Nov 23, 2023, 6:08:51 AM11/23/23
to pflotran-dev
Thanks, Glenn.

Just to better understand PFLOTRAN: What is the purpose of FACE for structured grids? If the delta of the coordinate range is zero, e.g. for the x-direction like
COORDINATES
20.d0 0.d0 0.d0
20.d0 1.d0 1.d0
then the region is a 2d object, i.e. a surface. So the input parameter FACE should not be needed, or are there reasons from code history?

Best, Jens 

Hammond, Glenn E

unread,
Nov 27, 2023, 11:49:37 AM11/27/23
to pflotr...@googlegroups.com

Jens,

 

For structured grids, the COORDINATE (whether a point, plane or box) designates the cell id and the FACE specifies the actual face. Yes, given a plane, one could traverse face center coordinates and determine the face, but the code is not set up for such.

 

Glenn

 

From: pflotr...@googlegroups.com <pflotr...@googlegroups.com> On Behalf Of Jens Eckel
Sent: Thursday, November 23, 2023 3:09 AM
To: pflotran-dev <pflotr...@googlegroups.com>
Subject: [pflotran-dev: 6219] Re: Diffusive transport under grid rotation

 

Check twice before you click! This email originated from outside PNNL.

 

Thanks, Glenn.

--

You received this message because you are subscribed to the Google Groups "pflotran-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pflotran-dev...@googlegroups.com.

Reply all
Reply to author
Forward
0 new messages