equinor
Error message when determining the convex hull
The following error message (1) has something to do with flownet not being able to determine the convex hull in a particular case. Flownet needs at least 5 wells for the algorithm to be able to work. The connections need to define a 3D volume.
The model I'm working with has exactly the same depth for all wells.
A temporal solution to do a test is to change the perforation strategy from bottom_point to multiple that would mean each connected block in the Eclipse model with getting a connection in FlowNet.
This may be not what you want in the long run, but it is a simple test to see if you get further.
(1) Error message:
(venv) manuel@manuel:~/repos/projects/DataDrivenModels/flownet$ flownet ahm tests/configs/egg_parameters.yml egg_output_folder
Adding flow nodes: 0%Traceback (most recent call last):
File "/home/manuel/repos/projects/DataDrivenModels/flownet/venv/bin/flownet", line 11, in <module>
load_entry_point('flownet', 'console_scripts', 'flownet')()
File "/home/manuel/repos/projects/DataDrivenModels/flownet/src/flownet/_command_line.py", line 206, in main
args.func(args)
File "/home/manuel/repos/projects/DataDrivenModels/flownet/src/flownet/_command_line.py", line 67, in flownet_ahm
run_flownet_history_mathing(config, args)
File "/home/manuel/repos/projects/DataDrivenModels/flownet/src/flownet/ahm/_run_ahm.py", line 162, in run_flownet_history_mathing
df_connections: pd.DataFrame = create_connections(df_coordinates, config)
File "/home/manuel/repos/projects/DataDrivenModels/flownet/src/flownet/network_model/_generate_connections.py", line 451, in create_connections
starts, ends = _generate_connections(df_coordinates, configuration)
File "/home/manuel/repos/projects/DataDrivenModels/flownet/src/flownet/network_model/_generate_connections.py", line 125, in _generate_connections
random_seed=configuration.flownet.random_seed,
File "/home/manuel/repos/projects/DataDrivenModels/flownet/src/flownet/network_model/_mitchell.py", line 88, in mitchell_best_candidate_modified_3d
hull = Delaunay(np.column_stack([x, y, z]))
File "qhull.pyx", line 1839, in scipy.spatial.qhull.Delaunay.__init__
File "qhull.pyx", line 357, in scipy.spatial.qhull._Qhull.__init__
scipy.spatial.qhull.QhullError: QH6154 Qhull precision error: Initial simplex is flat (facet 1 is coplanar with the interior point)
While executing: | qhull d Qc Qbb Q12 Qt Qz
Options selected for Qhull 2019.1.r 2019/06/21:
run-id 1237336110 delaunay Qcoplanar-keep Qbbound-last Q12-allow-wide
Qtriangulate Qz-infinity-point _pre-merge _zero-centrum Qinterior-keep
Pgood _max-width 4.4e+02 Error-roundoff 8.1e-12 _one-merge 7.3e-11
Visible-distance 4.9e-11 U-max-coplanar 4.9e-11 Width-outside 9.7e-11
_wide-facet 2.9e-10 _maxoutside 9.7e-11
precision problems (corrected unless 'Q0' or an error)
9 zero divisors during gaussian elimination
The input to qhull appears to be less than 4 dimensional, or a
computation has overflowed.
Qhull could not construct a clearly convex simplex from points:
- p5(v5): 60 68 4e+03 0
- p6(v4): 2.5e+02 12 4e+03 1.2e+02
- p0(v3): 36 4.5e+02 4e+03 4.3e+02
- p7(v2): 4.5e+02 44 4e+03 4.3e+02
- p2(v1): 12 2.8e+02 4e+03 1.5e+02
The center point is coplanar with a facet, or a vertex is coplanar
with a neighboring facet. The maximum round off error for
computing distances is 8.1e-12. The center point, facets and distances
to the center point are as follows:
center point 162.4 170.4 4026 223.6
facet p6 p0 p7 p2 distance= 0
facet p5 p0 p7 p2 distance= 0
facet p5 p6 p7 p2 distance= 0
facet p5 p6 p0 p2 distance= 0
facet p5 p6 p0 p7 distance= 0
These points either have a maximum or minimum x-coordinate, or
they maximize the determinant for k coordinates. Trial points
are first selected from points that maximize a coordinate.
The min and max coordinates for each dimension are:
0: 12 452 difference= 440
1: 12 452 difference= 440
2: 4026 4026 difference= 0
3: 0 4026 difference= 4026
If the input should be full dimensional, you have several options that
may determine an initial simplex:
- use 'QJ' to joggle the input and make it full dimensional
- use 'QbB' to scale the points to the unit cube
- use 'QR0' to randomly rotate the input for different maximum points
- use 'Qs' to search all points for the initial simplex
- use 'En' to specify a maximum roundoff error less than 8.1e-12.
- trace execution with 'T3' to see the determinant for each point.
If the input is lower dimensional:
- use 'QJ' to joggle the input and make it full dimensional
- use 'Qbk:0Bk:0' to delete coordinate k from the input. You should
pick the coordinate with the least range. The hull will have the
correct topology.
- determine the flat containing the points, rotate the points
into a coordinate plane, and delete the other coordinates.
- add one or more points to make the input full dimensional.
I think Delauney should be called with qhull_option='QJ Pp' inside the mitchells best candidate algorithm. That will joggle the input points slightly so you get to 3D.
I think Delauney should be called with qhull_option='QJ Pp' inside the mitchells best candidate algorithm. That will joggle the input points slightly so you get to 3D.
That could definitely be something to try!
I change the depths of the wells to avoid co-planar triangulations. The option of qhull_option='QJ Pp' solve the problem, but you get further errors due to the co-planar triangulation that I haven't solved properly. So the easiest way to get egg model working was changing the depths.
Are we now talking about minimal changes? If we would simply add a minimal random depth differences (centimetres) to wells with the exact same depth, would that solve the problem? That should be simple to add.