I62 wstat

[olwijn]

Closes #62.

The default type for production wells is set to SHUT instead of None to ensure that schedule entries rae created also after shutting in the well.

[olwijn]

The PR does not pass the Norne full model test.

In a different experiment with 4 realizations and 2 iterations some (not all) Norne runs are failing due to convergence problems (4 realizations, 2 iterations). The problems arise at different times and for different wells in different FlowNet realizations. The prior distributions were different from those used in the test however. Also, failure may be due to the small ensemble size resulting in unrealistic updates.

Running a failed FlowNet run with ECLIPSE instead triggers a different error :

117 READING INCLUDE .\include\SOLUTION_PARAMETERS.inc
118 READING EQUIL
119 READING RSVD
120 READING THPRES

@-- ERROR AT TIME 0.0 DAYS ( 6-NOV-1997):
@ THPRES KEYWORD SHOULD ONLY BE USED IF THE
@ THRESHOLD PRESSURE OPTION IS SET BY EQLOPTS
@ IN THE RUNSPEC SECTION
END OF INCLUDE FILE

Running the Norne HM test does not produce convergence errors on an Azure VM but the test runs only 2 realizations and performs only one update.

Some differences are induced in the schedule file for the Norne case with the changes made in this PR. In the original schedule wells are activated with a zero rate while in from_flow.py they only become active if the rate > zero.
I also notice that in Flownet well nodes that are generated with "bottom_node" option appear with multiple connections in the COMPDAT list. Is that intended?