Help with a simple Xsection model

[cneyens]

I'm trying to create a simple Xsection model. I've read the docs on the subject, which seem to show two ways of doing so: (1) create a standard Model and add 1D infinite elements, or (2) use Xsection models. The examples with Xsection models show the use of multiple inhomogeneities, and if I'm not mistaking, setting up an Xsection model requires the use of at least 1 'inhomogeneity element' through either Xsection3D() or XsectionMaq(). When I try to set-up a simple, homogeneous strip aquifer with the following code, I get an error:

import timflow.steady as tfs

ml = tfs.ModelXsection(naq=1)
tfs.Xsection3D(ml,
               x1=-500,
               x2=500,
               kaq=10.0,
               z=[0, -20.0],
               topboundary='conf',
               N=0.0)
riv = tfs.River1D(ml, xls=0.0, hls=2.0)
rfs = tfs.Constant(ml, xr=-500, yr=0, hr=0.0)
ml.solve()

which complains that 'SimpleAquifer' object has no attribute 'eigvec'. It seems that there are not enough equations (4) for the number of elements (5) (I'm guessing besides the river and the reference head, Xsection3D() adds a bunch of equations, as well as an areasink since N is defined). When I add some additional constants, I get the following error: 'SimpleAquifer' object has no attribute 'T'. For this particular case, I could easily use Model3D() and get results, but I'm wondering why the Xsection approach with a single strip (in)homogeneity does not work.

Perhaps I am misunderstanding the correct use of Xsection models?

timflow v0.1.0

[MattBrst]

Currently, you should define Xsection elements spanning the x-axis from -infinity to infinity. E.g.:

import numpy as np
import timflow.steady as tfs

ml = tfs.ModelXsection(naq=1)
tfs.Xsection3D(ml,
               x1=-np.inf,
               x2=500,
               kaq=10.0,
               z=[0, -20.0],
               topboundary='conf',
               N=0.0)

tfs.Xsection3D(ml,
               x1=500,
               x2=np.inf,
               kaq=10.0,
               z=[0, -20.0],
               topboundary='conf',
               N=0.0)

riv = tfs.River1D(ml, xls=0.0, hls=2.0)
rfs = tfs.Constant(ml, xr=-500, yr=0, hr=0.0)
ml.solve()

This works, although I am not sure whether this is what you want to model.

[cneyens]

Thanks for the reply, @MattBrst. That makes sense. Perhaps this could be added the docstrings somewhere? (maybe I missed it)

Additionally, I see that you specify two Xsection elements. Using only one element spanning from -Inf to +Inf results in a different error ("'SimpleAquifer' object has no attribute 'ltype'"):

import numpy as np
import timflow.steady as tfs

ml = tfs.ModelXsection(naq=1)
tfs.Xsection3D(ml,
               x1=-np.inf,
               x2=np.inf,
               kaq=10.0,
               z=[0, -20.0],
               topboundary='conf',
               N=0.0)

riv = tfs.River1D(ml, xls=0.0, hls=2.0)
rfs = tfs.Constant(ml, xr=-500, yr=0, hr=0.0)
ml.solve()

Again, I can certainly work around this; but I'm trying to understand the behavior and correct usage of the Xsection model.

[MattBrst]

See issue #39. Maybe in future versions it is not necessary anymore to span -inf to inf. But I agree it could be useful to clarify this in the documentation.

[dbrakenhoff]

Sorry, I missed this question, since my alerts weren't set up yet for the new timflow discussions page 😇

Using only one element spanning from -Inf to +Inf results in a different error

This is a bug. I will submit a fix.

Currently, you should define Xsection elements spanning the x-axis from -infinity to infinity.

So yes, this is currently the case, and should be documented more clearly somewhere. I also thought we had code somewhere that at least checked this, but I can't seem to find it anywhere, so apparently not. I will add this check in the PR I will make for the bug above.

As Mattijs already mentioned, there was also a discussion on supporting other boundary conditions in non-finite strips (#39), but we haven't gotten around to implementing that yet.

[dbrakenhoff]

Fixed in #71