Changing MODFLOW version

300 views
Skip to first unread message

Rezvane Ghorbani

unread,
Aug 30, 2021, 12:03:53 PM8/30/21
to MODFLOW Users Group
Hi everybody,

I have a MODFLOW model built in the GMS using MODFLOW-2000 (with GMG solver). I want to change the version to MODFLOW-NWT, which is possible in MODFLOW global options menu.

My question is that as changing the MODFLOW version to NWT would change the solver, should I recalibrate my model?

Thanks in advance,
Rezvane

Oscar Fabian Sanchez Ortiz

unread,
Aug 30, 2021, 11:37:49 PM8/30/21
to MODFLOW Users Group
it will change the solver, but if the modflow-2000 model works fine, the solution will be almost the same, so you don't have to recalibrate your model.

Saul Montoya

unread,
Aug 30, 2021, 11:38:00 PM8/30/21
to mod...@googlegroups.com
Hi Rezvane

Changing your solver to NWT will actually have some effect on your calibration and most probably you will have to affine your calibration for new calculated heads. If you model geometry si uniform and your hydraulica parameters don´t change in many orders of magnitude you shouldn't expect high differences.

Best regards

Saul

Saul Montoya M.Sc.
Director

Dir: Caminos del Inca 288 dpto 302, Surco, Lima 33  |  Tel.: (511) 4491922  | Cel: Nuevo! (51) 984 115 861  |  Web: gidahatari.com



--
You received this message because you are subscribed to the Google Groups "MODFLOW Users Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to modflow+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modflow/25c04a59-6d65-4df3-922d-1a1de75c0950n%40googlegroups.com.

Richard B. Winston

unread,
Aug 31, 2021, 12:43:01 AM8/31/21
to MODFLOW Google Group
The flow package in MODFLOW-NWT uses a reformulated flow equation for unconfined layers so you could have a somewhat different solution.

Scott Elliott Boyce

unread,
Sep 1, 2021, 12:38:21 AM9/1/21
to MODFLOW Users Group
Yes there will be some differences.

The NWT solver comes with a modified LPF flow package called UPW (Upstream Weighting) which is based off of the following publication:
Painter, S., Başağaoğlu, H., & Liu, A. (2008). Robust representation of dry cells in MODFLOW models. Groundwater, 46(6), 873-881.

There are two key differences between UPW and LPF.

  • The first is that UPW horizontal conductance is calculated between two model cells by using the head from the with the largest saturated thickness (cell with largest transmissivity),
    while LPF harmonically averages the transmissivity between the two cells to get a conductance (this is problematic cause a "dry cell" would cause a div/0 error when dividing by a zero transmissivit).
    Vertical conductance does not vary with saturation and is held constant at the saturated level.

  • The second is that a cell never goes "dry", this is a consequence cause conductance will have a non-zero value if adjacent cells contain water and that the vertical conductance is held constant at the saturated level.
The second difference is what causes NWT/UPW to have different solutions because water continues to be released from storage for dry cells, while in LPF the storage release stops.

To illustrate the difference, I made a 1D simple model that illustrates storage differences between LPF and UPW.

Its a 1 row, 1 col, 5 layer model with a constant head at lay 5.

The initial head is set to the top of lay 1 and the constant head is set to the top of lay 5 (forcing the model to drain downward drying out lay1, then 2, then 3, then 4).
The additional head is because dry layers in UPW still release water while LPF sets them to DRY status.

I attached the model for your reference, but it only runs in MF-OWHM2 ( https://code.usgs.gov/modflow/mf-owhm ) because it uses a lot of the LPF/UPW alternative inputs,
ie its keyword based, such as CONVERTIBLENO_LAYCBD,  and "NO WETTING",  and OWHM accepts the LPF input as a surrogate for UPW.

Also, MF-OWHM version 2.1.1 should be released next month and you can now include the UPW option LPF_STORAGE, which forces NWT to match exactly the LPF formulation while retaining the robust dry cell representation.

PCG vs NWT.png

Also note that anything that runs on MF-2005 and MF-NWT, should run on MF-OWHM.
I am not sure if GMS uses a modified version of the code, but you could just swap out the exe if you choose to use OWHM.

Hope that is some helpful information.

Scott
vertical_validation.zip

Rezvane Ghorbani

unread,
Sep 7, 2021, 1:37:10 AM9/7/21
to MODFLOW Users Group
Hi everyone,

Thank you so much for your help and thorough explanations; I really appreciate your kindness and consideration.

Best regards,
Rezvane

Reply all
Reply to author
Forward
0 new messages